Limes XVIII - Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000): A conference held under the auspices of the Department of Antiquities of the Hashemite Kingdom of Jordan, The Council for British Research in the Levant and the Department of Archaeology at the University of Liverpool 9781841714639, 9781841714646, 9781841714653, 9781407324784

Table of contents :
Title Page: Volume 1
Copyright
Contents
Introduction
Address by His Royal Highness Prince El Hassan bin Talal of the Hashemite Kingdom of Jordan at the 18th International Congress of Roman Frontier Studies
Opening ceremony in the Royal Cultural Centre, Amman, 2nd September 2000, in the presence of HRH Prince Hassan bin Talal and HRH Princess Sumaya a bintHassan
Jan Kees Haalebos
18th Limes Congress, Amman 9th September, 2000
Contributors addresses
Fifty years of Roman frontier studies
The onset of crisis: the reign of the emperor Severus Alexander (AD 222- 235)
Bezüglich der Europäischen Wanderschaft der legio III Gallica zu Beginn des 3. Jh.
Why did the Roman empire cease to expand?
Mints and the Roman army from Augustus to Diocletian
Problems in the study of Roman and native
Die Legionen des Augustus. Probleme der römischen Heeresgeschichte nach dem Ende des Bürgerkrieges: Die Truppengeschichte Galatiens und Moesiens bis in Tiberische Zeit und das Problem der Legiones Quintae
The northern, southern and eastern frontiers and the climate c.AD200
Similarities and differences in the late Roman defense system on the European andeastern frontiers
The Roman frontier in Jordan: An overview
Galerius and the eastern frontier
Diocletian: Politics and limites in the Near East
The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and theMoabite plateau
Nabataean or Roman? Reconsidering the date of the camp at Avdatin light of recent excavations
The military presence in the countryside of Petra in the C6th
Limes Arabicus, via militaris and resource control in southern Jordan
Nomads and the Arabian frontier: the epigraphic perspective
A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus in the second century AD
Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes
45 years of research on the limes Palaestinae – the findings and their assessment in the light of the criticisms raised (C1st – C4th)
Excavations at the Roman siege complex at Masada – 1995
Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia
Lajjun – Legio in Israel: Results of a survey* in and around the military camp area
Road use in late antique Palestine
Roman military forts at Zeugma. A preliminary report
Preliminary report on the Roman period at Kamid el-Loz (Lebanon)
“. . . ostentasse Romana arma satis . . .” The military factor in Roman-Parthian relations under Augustus and Tiberius
The circumvallation at Hatra
Roman treaties with Parthia: Völkerrecht or power politics?
Die ‘Sklaven des Kaisers’ und der Kopf des Crassus. Römische Bilder des Ostens und parthische Bilder des Westens in augusteischer Zeit
The Cappadocian frontier: from the Julio-Claudians to Hadrian
Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey):second preliminary report on the inscriptions
Gregorios Thaumaturgos und die pontischen Beutezüge der Boranen und Gotenim 3. Jh. n. Chr.
Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)
Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis
The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs (Eastern Desert of Egypt) 1998-2000
Merchants, pilgrims and soldiers on the Red Sea route
The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL)
Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes: Topographie, Siedlungsstruktur, Nutzungszonen, Grundrisstypen sowie belegbare Nutzungen
Die Früheste Belegung des Hunerberges in Nijmegen
Reduktion von numerus- und Kleinkastellen des obergermanischen Limes im 3. Jahrhundert
Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr.
Die spätaugusteische Siedlung in Lahnau-Waldgirmes - Zusammenfassende Bemerkungen zum Stand der Fundauswertung
Recent developments in south-west Germany (eastern Germania Superior - western Raetia)
The late Roman fort at the Valkhof in Nijmegen. Nijmegen at the transition from the Roman period to the middle ages
Die Ausgrabungen in Lahnau-Waldgirmes
Timber baths in Augustan and Tiberian fortresses
Early Roman military activity on the Frisian coast
Zur Datierung der frühkaiserzeitlichen Militärlager von Novaesium (Neuss)
Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit
Kalkriese und die Varusschlacht – Archäologische Nachweise einer militärischen Auseinandersetzung zwischen Römern und Germanen
Neue Grabungen in Haltern, Oberaden und Anreppen
Title Page: Volume II
Copyright
List of Contents
Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen an der Donau und im raetischen Limesgebiet
The Flavian timber fort at Eining (Abusina) on the Upper Danube
Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg
Neue Grabungen im Vicus des Auxiliarkastells Favianis (Noricum) – ein Handwerksbezirk im östlichen Kastellvicus
Vindobona – legionary fortress, canabae legionis and necropolis
Vindobona – the Roman civil settlement at Vienna (poster-presentation)
Die Ausgrabungen am Michaelerplatz – ein Bereich dercanabae legionis von Vindobona
Legionslager an der Wende zur Spätantike – Ein Überblick zu Carnuntum und vergleichbaren kaiserzeitlichen Standlagern des Rhein-Donau-Raumes in einer Periode des Umbruchs
The history and perspectives of the research of the Csörsz Ditch (‘Limes Sarmatiae’)
Decorated lead sarcophagi in Moesia Superior
Artillery troops detached north of the lower Danube in the late Roman period
The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)
Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes?
Archaeological survey on the Roman frontier on the lower Danube between Novae and Sexaginta Prista. Preliminary report (1997-2000)
Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae (Moesia Inferior)
Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior)
Was there a militarisation of the southern Balkans during late antiquity?
Tilurium, Burnum and Bigeste. A new contribution to the dating of the Delmataen limes
Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia
Deserted forts – the Moesian limes after the conquest of Dacia
The military equipment of oriental archers in Roman Dacia
Die römischen Lager bei Renieblas, Prov. Soria (Spanien). Ergebnisse der Vermessungskampagnen 1997-2000
Twenty years of Roman military archaeology in Spain
The legionary fortress of VI Victrix at León (Spain). The new evidence (1995-2000)
Les castra et les castella aux extrémités de l’Empire après la fin de la dominationromaine: le Nord-ouest Ibérique et le Haut Rhin au Ve. siècle
New intervention in the surroundings of the gate of the Roman wall of Gijón (10 Recoletas Street)
The material culture of Hadrian’s Wall
A revised programme and chronology for the building of Hadrian’s Wall
Recent research on Roman camps in Wales
Supplying the frontier zones: the rôle of the East Anglian Fens
Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999
Cawthorn Camps – 70 years after Richmond
The Roman Gask frontier: 1997-2000
The timing of marriage in the Roman army
The Roman auxilia in the east – different from the west ?1
Ditches for defence: towards a study of multiple fort ditches
‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts at Wallsend and South Shields on Hadrian’s Wall
The rampart buildings of Roman legionary fortresses
Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica)
Neue Grabungen an der Westseite des römischen Flottenlagers Köln - Alteburg
Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg (Alteburg)
Now you see it, now you don’t. The British fleet in Vegetius IV.37
The Roman army in the Vindolanda Tablets
Food as substance and symbol in the Roman army: a case study from Vindolanda
Vespasianvs Velageno

Citation preview

Limes XVIII Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000) A conference held under the auspices of the Department of Antiquities of the Hashemite Kingdom of Jordan, The Council for British Research in the Levant and the Depart1nent of Archaeology at the University of Liverpool

Volume I Edited by

Philip Freeman, Julian Bennett, Zbigniew T. Fiema and Birgitta Hoffmann

BAR International Series 1084 (I) 2002

ISBN 9781841714639 (Volume I) paperback ISBN 9781841714646 (Volume II) paperback ISBN 9781841714653 (Volume set) paperback ISBN 9781407324784 (Volume set) e-format DOI https://doi.org/10.30861/9781841714653 A catalogue record for this book is available from the British Library

BAR

PUBLISHING

List of Contents Volume 1 Introduction

viii

Timetable of sessions, presentations and events

xvii

Formal Addresses Address by His Royal Highness Prince El Hassan bin Talal of the Hashemite Kingdom of Jordan

xxiii

Professor David Breeze, Chairman of the Congress Welcoming address

xxvii

David J. Breeze Jan Kees Haalebos†

xxix

Closing ceremony Speech given by Professor Zsolt Visy

xxx

Contributors’ addresses

xxxii

Sessions General Themes A.R.Birley Fifty years of Roman frontier studies

1

Lukas de Blois The onset of crisis: the reign of the emperor Severus Alexander (AD 222- 235)

13

Nicolae Gudea Bezüglich der Europäischen Wanderschaft der legio III Gallica zu Beginn des 3. Jh.

19

William S. Hanson Why did the Roman empire cease to expand?

25

Johann van Heesch Mints and the Roman army from Augustus to Diocletian

35

Fraser Hunter Problems in the study of Roman and native

50

Karl Strobel Die Legionen des Augustus. Probleme der römischen Heeresgeschichte nach dem Ende des Bürgerkrieges: Die Truppengeschichte Galatiens und Moesiens bis in Tiberische Zeit und das Problem der Legiones Quintae

51

Wolfgang Vetters & Heinrich Zabehlicky The northern, southern and eastern frontiers and the climate c.AD200

67

Zsolt Visy Similarities and differences in the late Roman defense system on the European and eastern frontiers

71

The Eastern Frontier S. Thomas Parker The Roman frontier in Jordan: An overview

77

i

Bill Leadbetter Galerius and the eastern frontier

85

Ariel Lewin Diocletian: Politics and limites in the Near East

91

Chaim Ben-David The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau

103

Tali Erickson-Gini Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

113

Zbigniew T. Fiema The military presence in the countryside of Petra in the C6th

131

George Findlater Limes Arabicus, via militaris and resource control in southern Jordan

137

David F. Graf Nomads and the Arabian frontier: the epigraphic perspective

153

Theodor Kissel A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus in the second century AD

161

Maurice Lenoir Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes

175

Mordechai Gichon 45 years of research on the limes Palaestinae – the findings and their assessment in the light of the criticisms raised (C1st – C4th)

185

Haim Goldfus & Benny Arubas Excavations at the Roman siege complex at Masada – 1995

207

Israel Roll Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

215

Yotam Tepper Lajjun – Legio in Israel: Results of a survey in and around the military camp area

231

Susan Weingarten Road use in late antique Palestine

243

Martin Hartmann & Michael A. Speidel Roman military forts at Zeugma. A preliminary report

259

Lars Petersen Preliminary report on the Roman period at Kamid el-Loz (Lebanon)

269

Rome and Parthia Edward Dąbrowa “. . . ostentasse Romana arma satis . . .” The military factor in Roman-Parthian relations under Augustus and Tiberius

ii

275

Gwyn Davies The circumvallation at Hatra

281

Everett L. Wheeler Roman treaties with Parthia: Völkerrecht or power politics?

287

Josef Wiesehöfer Die ‘Sklaven des Kaisers’ und der Kopf des Crassus. Römische Bilder des Ostens und parthische Bilder des Westens in augusteischer Zeit

293

The Anatolian Provinces and the Black Sea Region Julian Bennett The Cappadocian frontier: from the Julio-Claudians to Hadrian

301

Ergün Lafli Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey): Second preliminary report on the inscriptions

313

Jeorgios Martin Beyer ίνα ... αύτοί άλλοις Βοράδοι καί Γότθοι γίνωνται. Gregorios Thaumaturgos und die pontischen Beutezüge der Boranen und Goten im 3. Jh. n. Chr.

327

Oleg Savelya Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)

339

North Africa Alan Rushworth Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

349

Steven E. Sidebotham The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs (Eastern Desert of Egypt) 1998-2000

361

Gabriele Ziethen & Erik Klingenberg Merchants, pilgrims and soldiers on the Red Sea route

379

The Germanies Harry van Enckevort The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL)

387

Meingad Filgis Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes: Topographie, Siedlungsstruktur, Nutzungszonen, Grundrisstypen sowie belegbare Nutzungen

395

J.K. Haalebos† Die früheste Belegung des Hunerberges in Nijmegen

403

Marcus Jae & Markus Scholz Reduktion von numerus- und Kleinkastellen des obergermanischen Limes im 3. Jahrhundert

415

Stefan Neu & Matthias Riedel Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr.

425

Gabriele Rasbach Die spätaugusteische Siedlung in Lahnau-Waldgirmes - Zusammenfassende Bemerkungen zum Stand der Fundauswertung

433

iii

C. Sebastian Sommer Recent developments in south-west Germany (eastern Germania Superior - western Raetia)

441

J. Thyssen The late Roman fort at the Valkhof in Nijmegen. Nijmegen at the transition from the Roman period to the middle ages

453

Early Roman Germany Armin Becker Die Ausgrabungen in Lahnau-Waldgirmes

461

Paul Bidwell Timber baths in Augustan and Tiberian fortresses

467

Marjan C. Galestin Early Roman military activity on the Frisian coast

483

Thomas Grane Problems concerning the early Roman fortifications at Marktbreit (D) and Valkenburg ZH (NL)

489

Norbert Hanel Zur Datierung der frühkaiserzeitlichen Militärlager von Novaesium (Neuss)

497

Andrea Hagendorn & Christine Meyer-Freuler Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit

501

Katrin Roth-Rubi Why Dangstetten?

509

Susanne Wilbers-Rost Kalkriese und die Varusschlacht – Archäologische Nachweise einer militärischen Auseinandersetzung zwischen Römern und Germanen

515

Siegmar von Schnurbein Neue Grabungen in Haltern, Oberaden und Anreppen

527

Volume 2 The Danubian and Balkan Provinces Claus-Michael Hüssen Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen an der Donau und im raetischen Limesgebiet

535

Markus Gschwind The Flavian timber fort at Eining (Abusina) on the Upper Danube

549

Andreas Schaub Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg

559

Stefan Groh & Helga Sedlmayer Neue Grabungen im Vicus des Auxiliarkastells Favianis (Noricum) – ein Handwerksbezirk im östlichen Kastellvicus

567

M. Kronberger & M. Mosser Vindobona – legionary fortress, canabae legionis and necropolis

573

iv

Ingrid Mader Vindobona – the Roman civil settlement at Vienna

585

Sylvia Sakl-Oberthaler & Patrizia Donat et al Die Ausgrabungen am Michaelerplatz – ein Bereich der canabae legionis von Vindobona

591

Raimund Kastler Legionslager an der Wende zur Spätantike – Ein Überblick zu Carnuntum und vergleichbaren kaiserzeitlichen Standlagern des Rhein-Donau-Raumes in einer Periode des Umbruchs

605

Eszter Istvánovits & Valéria Kulcsár The history and perspectives of the research of the Csörsz Ditch (‘Limes Sarmatiae’)

625

Snežana Golubović Decorated lead sarcophagi in Moesia Superior

629

Dorel Bondoc Artillery troops detached north of the lower Danube in the late Roman period

641

Andrzej B. Biernacki The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

649

Gerda v. Bülow Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes?

663

Sven Conrad & Dimităr Stančev Archaeological survey on the Roman frontier on the lower Danube between Novae and Sexaginta Prista. Preliminary report (1997-2000)

673

Piotr Dyczek Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae (Moesia Inferior)

685

Elena Klenina Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior)

695

Archibald Dunn Was there a militarisation of the southern Balkans during late antiquity?

705

Mirjana Sanader Tilurium, Burnum and Bigeste. A new contribution to the dating of the Delmataen limes

713

Dacia Ioana Bogdan Cătăniciu Géographie antique (Ptolémée, Tabula Peutingeriana, Ravennatus) et la stratégie impériale en Dacie

719

Cristian Gazdac Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia

737

Miroslava Mirković Deserted forts – the Moesian limes after the conquest of Dacia

757

Liviu Petculescu The military equipment of oriental archers in Roman Dacia

765

v

The Spains Martin Luik Die römischen Lager bei Renieblas, Prov. Soria (Spanien). Ergebnisse der Vermessungskampagnen 1997-2000 771 Angel Morillo & Victorino García-Marcos Twenty years of Roman military archaeology in Spain

779

Victorino García-Marcos & Angel Morillo The legionary fortress of VI Victrix at León (Spain). The new evidence (1995-2000)

791

Jorge López Quiroga & F.Germán Rodríguez Martín Les castra et les castella aux extrémités de l’Empire après la fin de la domination romaine: le Nord-ouest Ibérique et le Haut Rhin au Ve. siécle

801

C. Fernández. Ochoa, V. García Entero, F. Gil Sendino & C. Valenciano Prieto New intervention in the surroundings of the gate of the Roman wall of Gijón (10 Recoletas Street) 813

Britain Lindsay Allason-Jones The material culture of Hadrian’s Wall

821

Julian Bennett A revised programme and chronology for the building of Hadrian’s Wall

825

J.L. Davies & R.H. Jones Recent research on Roman camps in Wales

835

Alison Taylor Supplying the frontier zones: the rôle of the East Anglian Fens

843

Tony Wilmott Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999

851

Pete Wilson Cawthorn Camps – 70 years after Richmond

859

D.J.Woolliscroft The Roman Gask frontier: 1997-2000

867

The Roman Army Sara E. Phang The timing of marriage in the Roman army

873

D.B. Saddington The Roman auxilia in the east – different from the west?

879

Roman Fortifications David J. Breeze Ditches for defence: towards a study of multiple fort ditches

883

Nicholas Hodgson ‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts at Wallsend and South Shields on Hadrian’s Wall 887

vi

Birgitta Hoffmann The rampart buildings of Roman legionary fortresses

895

Fleets and Frontiers Maureen Carroll Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica)

901

Thomas Fischer Neue Grabungen an der Westseite des römischen Flottenlagers Köln – Alteburg

904

Norbert Hanel Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg (Alteburg)

913

Boris Rankov Now you see it, now you don’t. The British Fleet in Vegetius IV.37

921

Documents and Archives A.R. Birley The Roman army in the Vindolanda Tablets

925

John Pearce Food as substance and symbol in the Roman army: a case study from Vindolanda

931

Margaret Roxan Vespasianvs Velageno

945

vii

Introduction The papers published here represent approximately two-thirds of the 150+ presentations made to the XVIIIth International Congress of Roman Frontier Studies in September 2000 in Amman Jordan and held under the auspices of the Department of Antiquities of the Hashemite Kingdom of Jordan, The Council for British Research in the Levant (CBRL) and the School of Archaeology, Classics & Oriental Studies (SACOS) in the University of Liverpool. As was noted at the time this was the first time in over 50 years and 17 previous occasions that the Congress had visited an Arab country and with one exception, the first time it had left Europe. Nearly eighteen months since the Congress, I think the event can still be deemed to have been a success. That it was so successful was due to the coming together of the work and enthusiasm of a large number of people and organisations. The following paragraphs explain the progress of the Congress and to record for posterity the work of those individuals and institutions. Whilst they may read as a list, still the Organising Committee is extremely grateful for the help of so many. The background to the Amman Congress In the light of Professor Birley’s comments (see below) about earlier congresses and the loss of information concerning who was there, who did what and what happened at them, the following paragraphs explain the origins of the Amman Congress and the course of its organisation. It had frequently been lamented that the Limes Congress, with one exception in the late 1960s, had never managed to visit any of the non-European provinces of the Roman empire. Indeed such was this failing, with the concomitant effect that the number of ‘oriental’ specialists who attended Congresses was relatively small, what looked to be a ‘breakaway’ conference was initiated in the early 1980s, complemented not just by the publication of parallel conference proceedings which are still widely cited1 but had the additional effect of stimulating a fresh impetus to research in those areas. Throughout this period on a number of occasions it was suggested that the Limes Congress might visit the Middle East, with either Turkey, Syria or Jordan as suitable locations. This proposal was discussed as a serious proposition as far back as the excursion which complemented the 1988 Ankara conference. As the political and economic climate developed in the 1990s along with the current burst of research on the Roman military presence there coming to a climax, Jordan was perceived as the more attractive venue. In the early 1990s Philip Freeman talked with Ms. Alison McQuitty, the then in-coming Director of the British Institute in Amman for Archaeology and History (BIAAH), about the prospect of holding a Congress there in the late 1990s. The research being undertaken by a number of scholars, in particular to name but a few, that by the Professors Tom Parker, David Kennedy and David Graf, coupled with the enthusiasm of Prof. Parker, made the country a natural venue for a Congress. In the meanwhile, Mr. Tim Strickland, who had long been associated with the archaeology of Jordan, was thinking along the same lines. It was the bringing together of these suggestions which led to what became the Amman Congress. Accordingly, at the meeting of the International Committee of the International Congress of Roman Frontier Studies in Rolduc, Prof. David Breeze was invited to seek a formal invitation from colleagues in Jordan to hold a Congress in that country. Through the good offices of Freeman and Mr Strickland, contact was made with Ms. McQuitty of the BIAAH and on 26th August 1997 she and Dr. Ghazi Bisheh, Director-General of Antiquities in the Hashemite Kingdom of Jordan wrote to Prof. Breeze inviting the Congress to be held in Jordan in 2000. This invitation was accepted with acclamation at a plenary session the following month during the Zalau Congress in Romania. The Congress The organisation of the Congress was co-ordinated by a small committee which met in both Edinburgh and Amman. Professor Breeze acted as its Chair, whilst Freeman representing the University of Liverpool, one of the sponsors of the Congress, was its secretary. In time Mr. Strickland and Dr. Mike Bishop, original members of the committee, had to stand down because of other commitments. Following the end of her tenure as Director of the BIAAH, Ms. McQuitty’s place on the Committee was taken by her successor, Dr. Bill Finlayson, Director of the newly created Council for British Research in the Levant (CBRL). In their contributions, Ms. McQuitty and Dr. Finlayson were ably supported by Mr. George Findlater, Depute Director of the BIAAH/CBRL. Much of the day-to-day administration in Liverpool fell to Mrs. Pat Sweetingham and Mrs. Jennifer Mirdamadi.

1

Swansea 1981 - cf. S. Mitchell (ed.) Armies and frontiers in Roman and Byzantine Anatolia (Oxford: 1983): Sheffield 1986 - cf. P. Freeman & D. Kennedy The defence of the Roman and Byzantine east (Oxford: 1986); Ankara 1988 - cf. D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman army (Oxford: 1989): Krakow 1992 - cf. E. Dabrowa (ed.) The Roman and Byzantine army in the east (Krakow 1994)

viii

His Royal Highness Prince Hassan graciously agreed to act as Patron of the Congress. Prince Hassan and his daughter Princess Summaya took a keen interest in the organisation and execution of the Congress. Both were of considerable assistance throughout the planning of it. In Jordan, the Minister of Tourism, His Excellency Mr. Akel Biltaji was an enthusiastic supporter of the conference. Dr. Ghazi Bisheh and his successor as the DirectorGeneral of the Department of Antiquities, Mr. Fawwaz al Khraysheh made available the support and resources of their department. The British Embassy also offered advice and made introductions to contacts. The Chairman of the CBRL, Mr. Adrian Sindall, also offered support and experience in the preparation of the Congress. The University of Liverpool provided seed funding for the Congress and in the persons of Mrs. Sweetingham and Mrs. Mirdamadi, co-ordinated the publicity and original accommodation bookings etc. Modern technology considerably aided the organisation of the Congress. Indeed over 60% of the Congress’ business, including enquiries, reservations and bookings, the despatch of paper abstracts and other issues was conducted via e-mail and the Internet. In Jordan the logistics were taken up by Nebo Tours. Mr. Najeeb Ayoub was the company representative with responsibility for managing the conference and its accommodation in Amman and Petra. We were fortunate that all these individuals or their representatives were able to attend most of the conference. A particular concern of the Organising Committee was to ensure that our Jordanian colleagues were represented at the Congress. Another concern for the Committee at an early stage in discussions about the format of the Congress was ensuring that the conference attracted as many persons from as many different parts of the orbis Romanum as possible. Toward this end we wished to offer discounts on the Congress fee for ‘students’ and those from the Middle East and Eastern Europe. Whilst trying to set an affordable conference fee we also wanted to create a small surplus which might be used to meet in part the costs of attending for a number of individuals. Towards this end we were also fortunate to receive a generous award from the British Academy for the express purpose of supporting the attendance of a number of delegates. In the end we were able to offer ‘sponsorship’ in one way or another to upwards of 40 persons from Eastern Europe, Africa and the Middle East, with a mix of ‘established’ scholars, other researchers and students. The Congress was held in Amman between Saturday 2nd September and Monday 11th September 2000. Participants were housed in the Regency, Grand Palace, Arwad and San Rock hotels, with the lectures held in the Royal Cultural Centre (Figs. 2 and 3). The Congress programme is appended at the end of this Introduction. The Sessions As in recent Congresses special sessions were arranged, including one on the archaeology of Jordan and others on aspects of the Eastern frontier; these are detailed in the programme below. In putting together the Plenary Session, the Organising Committee wanted to provide a combination of an introduction to the history and geography of Jordan as well as a broad resume of recent progress on the (military) archaeology of the Roman province. Given the introductory and/or summary nature of these presentations, most of these papers have not made it into these proceedings. In some instances time constraints meant that speakers could offer little more than the broadest introduction. Elsewhere Prof. Kennedy’s paper on the frontier in Jordan from the air is effectively published as the Congress handbook. Two papers included in these Proceedings represent more substantial presentations. Prof. AR Birley was asked to provide a retrospective on the progress of Roman frontier studies since the first Congress in 1949. Prof. Parker’s paper is a summation of work on the frontier in Jordan, work which was in large part inspired by his pioneering Limes Arabicus survey and excavations of the 1980s. We are pleased too to be able to include HRH Prince Hassan’s address made at the Opening Ceremony (Fig. 1). The Receptions The Congress was fortunate to receive the generous hospitality of a number of organisations in Jordan. On the first night (September 2nd) a reception, hosted by the CBRL and the British Ambassador to Jordan, was held in the grounds of the Ambassador’s Residence in Amman (Figs. 4, 5, 6, and 7). To complement our visit to Jerash (September 5th) the Dept. of Antiquities provided a lavish feast in the restaurant at the Visitors Centre to the ancient city. A completely unexpected offer was HRH Prince Hassan’s wish to host a formal dinner for the delegates. This was held in the Meridien Hotel on September 6th. At an early stage in the discussions concerning the organisation of the Congress the American Center for Oriental Research (ACOR) in Amman offered to host a reception. The reception was held on September 8th. This was also an opportunity for the Center to explain its work and to display the progress on the preservation and deciphering of the recently found ‘Petra Papyri’. The Center certainly ‘pushed out the boat’ in welcoming everybody to its premises. We are grateful to Prof. Parker for co-ordinating the visit and especially to the Director of ACOR, Dr. Pierre Bikai and his wife Patricia for their hospitality. In the evening of the day when formal sessions were concluded (September 9th) the Darat al-Founoun – Abdel Hameed Shoman Foundation graciously hosted a reception and tour of its premises on Jebel Amman. The Foundation is devoted to fostering and publicising the arts and crafts of Jordan. Last but by no means least,

ix

on the evening of our arrival in Wadi Musa, the Petra Regional Council hosted another dinner which prepared us for the tour of the ancient city the next day. The Tours In determining the format of the tours that complemented the Congress, we were keen that in addition to visiting the relevant parts of the frontier, appropriate Nabatean and Roman civilian sites were seen as well as exploring a representative sample of the topographic, climatic and floral environment of Jordan. It was also decided that trips to the major urban sites of Jerash and Petra could not be avoided. Naturally within these parameters certain sites we would have liked to have visited had to be omitted. These included forts such as Qasr al-Hallabat and alQihati in the north and the excavations at Humeima as well as forts further south such as Khalde, and Quweira and ultimately Aqaba which were deemed to be too far distant for us to visit comfortably. To facilitate the excursions as well as to create a lasting memorial of the Congress, Prof. Kennedy was asked to prepare the handbook which traditionally accompanies the Congress. The resulting volume, The Roman Army in Jordan, was more than the Organising Committee could have reasonably expected. Our plans to organise Vor- and Nachexkursions to Syria and Israel had to be abandoned because of the insurmountable logistical and political problems. Two preliminary visits aided the planning of the tours. However, again, despite our original plans, all the tours had to be redevised during the Congress. The revised itinerary of sites visited in recorded in the Congress timetable below. The intention to visit the spectacular site at Qasr Asheikhin in north Jordan had to be abandoned after the arrangement for army lorries we were hoping to procure fell through. Whilst we were able to re-jig the rest of the itinerary for that day, the lack of lorries had potentially fatal repercussions for our other trips. Fortunately the resourcefulness of Mrs. Nazmir Tawfiq of the Dept. of Antiquities led her to devise an innovative mode of transport which permitted us access to the other difficult sites of Qasr Bshir and Da’janiya (Fig. 9). Throughout the excursions Mrs. Tawfiq was a godsend in the way she facilitated access to sites and generally smoothed our progress (Fig. 10). The Congress tours were accommodated on 5 buses each with a dedicated guide and assistant. Guiding at sites was undertaken by Dr. Ignacio Arce at Amman, Ms. Ina Kehrberg at Jerash, Freeman at Azraq (Fig. 8) and Tom Parker and Umm il-Jimal, Qasr Bshir, Da’janiya and Udruh. Dr. Fawzi Zayadine of the Dept. of Antiquities also provided in-field introductions and summaries at a number of sites. We thank Zbig Fiema and Ina Kehrberg for agreeing to act as bus guides and the various sites guides. Isabelle Ruben and George Findlater were also stalwarts throughout the tours. The staff of the CBRL prepared the picnic lunches for the excursions. Our ability to run so many excursions was due largely to the generosity of the British Embassy which made available a very generous sum to pay for the cost of the buses. We are grateful to the Director of Antiquities and ultimately the Minister of Tourism for waiving the entry fees for our visits to Jerash and Petra. The good offices of the Dept. of Antiquities also negotiated a discount for our extended use of the Royal Cultural Centre. In addition to the official excursions, Mr. Najeeb of Nebo Tours also facilitated supplementary tours for those who wanted them and in particular a half-day visit to Umm Qais (Gadara) as well as Mt. Nebo and the Dead Sea and for those who wanted to extend their stay in Jordan. Squadron Leader David Kennedy and Flying Sergeant Bob Bewley organised a helicopter flight over Amman and Jerash for those brave enough to place their faith in an antiquated Soviet-era flying machine. The Proceedings Papers published in these Proceedings are a mix of those papers presented at the Congress, those delivered in absentia and those offered but where the authors for a variety of reasons were unable to come to Jordan. We also decided to invite those who made Poster Presentations to submit lengthier versions. We are pleased to say that a number of individuals have availed themselves of this offer. The original plan was to publish the Proceedings towards the end of 2001. Contributors were asked to submit their papers to the Liverpool department within 6 months of the Congress. Over 75% of authors met this deadline. After preliminary editing and formatting, the papers were circulated among the three other editors who made recommendations with respect to content as well as presentation. At least one set of proofs was sent out to authors. Where necessary, authors were asked to consider making such alterations. On rare occasions an additional external reader was asked for comments. We are grateful for the prompt manner in which authors returned their contributions. In general the editors were particularly concerned not to interfere with contributions x

unless it was absolutely necessary. We originally planned to offer the barest minimum of corrections with respect to the language in which the paper was written. But in truth, there has been a degree of re-writing. The editors were also concerned, based on previous experience, to ensure that the standard of presentation and bibliographic referencing was at least consistent. In another volume of conference papers that I once helped in a minor way to edit, my co-editor wrote “(Some) papers caused ...problems and delay from the way they were presented; one in particular, which had better remain anonymous, ignored almost all our requirements, and required some 12 hours of work to recast and complete several dozen (sic) incomplete or erroneous references. In the sort of world envisaged in classical mythology, the author would be condemned to an eternity of editing papers such as his own”. So it was 15 years ago and despite the advances in word-processing technology, the same applies today. Two or three of the papers published here required the same amount, if not more, attention. Papers were received in a bewildering variety of formats and word processing packages. Most of these could be dealt with. However some delays in the processing of papers were caused by some authors not being able to provide full bibliographic details for citations. If the processing of the papers was a relatively straightforward matter, where there were problems was with respect to the quality – or suitability – of some the illustrative material that was submitted. For the sake of clarity and to maintain a minimum standard of quality, approximately 30% of the original artwork has had to be redrawn. It this reason above all others which has delayed the appearance of these proceedings. The resulting quality of the artwork is a reflection of the skill of the illustrator who has overseen the preparation of these proceedings, Ms. Lorraine McEwan of the Dept. of Archaeology at the University of Glasgow. Mr. Jonathan Trigg, Ms. Carmel Malone and Ms. Estelle Bougard, all of the Liverpool department, provided valuable assistance in the processing and proofing of some contributions. Some papers originally delivered at the Congress have been published elsewhere. Fraser Hunter’s contribution on Roman and natives has been expanded into a longer article now published in the Journal of Roman Archaeology (14 (2001): 289-309). Denis Saddington spoke on ‘The Classis Syriaca and Roman operations on the Euphrates’, now published as ‘The Roman naval presence in the east, the Classis Syriaca and the Roman approach to the Euphrates’ in the Archäologisches Korrespondenzblatt (31 (2001): 581-586). Martin Hartman and Michael Speidel’s discussion of the fort(s) at Zeugma (Turkey) will be enlarged in paper to be published in a forthcoming issue of again, the Journal of Roman Archaeology. Shortly after the Congress ended we received the sad news that a long-time friend of the Congress Professor Haalebos had died suddenly. In the aftermath, Professor Haalebos’ widow kindly agreed that we might publish the paper Jan Kees had already provided. I am grateful to Rien Polak and Dr. Katrin Roth-Rubi for seeing Professor Haalebos’ paper through the editing process and for seeking out the illustrations he planned to include with his text. Prof. Breeze has also kindly provided a memoir of a friend and colleague (Fig. 11). Acknowledgments We are grateful to the many people who made the Congress such a success: to Ms. Alison McQuitty of the British Institute at Amman for Archaeology and History/Council for British Research in the Levant; to Mrs. Nazmir Tawfiq of the Dept. of Antiquities who facilitated the smooth running of the Congress tours; to those who led the site tours (Ms. Ina Kehrberg, Prof. Tom Parker, Dr. Fawzi Zayadine, Dr. Ignacio Arce); to those who acted as assistants on the tours (Dr. Zbig Fiema, Mrs. I. Ruben and George Findlater); and to those who oversaw the lecture theatres and projection facilities in the Royal Cultural Centre, Ms. Samantha (‘Sam’) Dennis, Ms. Charlotte (‘Charlie’) Schriwer and Ms. Suzanne Schaubel. The photographs of the Opening Ceremony and the Reception at the Residence of the British Ambassador were taken by Mr. Omar Mohammad el-Fahmawi. Afterword Finally, if I might be now indulged with a few observations about the Congress. During the Closing Ceremony I was invited to make a few comments about the event. As I was not expecting to have to say anything, I felt that my mutterings missed the opportunity. As I said at the time, and think even more so now, it was a remarkable achievement that we managed to organise a conference from Britain to be held in another country, bring together ‘friends’ from more than 25 countries and manage to feed, accommodate, stimulate and amuse them for nearly two weeks. As far as we know, during the course of the Congress nobody was lost, arrested or died. There was at least one dented skull, a twisted ankle and a few cases of what might be called ‘upset stomachs’, but nothing too serious. There were also a few hangovers. Seeing how political conditions have developed in the Middle East since late 2000, we were fortunate to hold the Congress when we did. We even managed to make a small profit which has since been re-invested in the preparation of these Proceedings. It bears repetition that this was all possible because of the unpaid assistance of others and the co-operation of any number of institutions and organisations. As proof of this, much of what we heard of and saw of the archaeology of Jordan was due in large xi

measure to the work of the Jordanian Dept. of Antiquities working in tandem with the various foreign archaeological schools based in Jordan. Those of you with any doubts about the value of the schools should remember that much of the work we saw, whether it be at Petra and Jerash or the less impressive but still as important via nova Traiana (and indeed the archaeology of Jordan in general), is due to the work of teams coordinated through the American British, French, German, Spanish etc. institutes in Jordan. That work is also the result of the happy marriage of the schools and the Dept. of Antiquities and its previous and current directors. Long may it continue. Undeniably there were hitches and problems during the Congress. These were unavoidable for a conference of this size and length. The majority of them were beyond our control. I think virtually every decision the Organising Committee made in the two years up to the Congress was changed in some way or other in the days during the Congress. But these and other ‘glitches’, we managed to rectify. That this was possible was due in large part to the forbearance and overall exceptionally accommodating nature of those attending the Congress. More than anything, this is the abiding memory of the Congress, it good humour and friendly atmosphere, undeniably fuelled by the copious quantities of home made ‘hospitality’ our Romanian colleagues somehow managed to bring into the country. I hope that as a group, the Organising Committee has acknowledged its debts to those who made the Amman Congress such a success. Any omissions are a measure of the size of our debt. Finally, I have my own thanks to make. I owe a particular debt to Prof. Liz Slater of the University of Liverpool. Not only was she prepared to let the Liverpool department act as the vehicle for arranging the Congress and so bear some of the initial costs of setting it up, I can not think of many other senior academics, officially there as the representative of her University, who would be prepared ‘to muck in’ as much as she did with the actual running of the Congress in Amman. Pat Sweetingham’s – ‘dear ‘ole Pat’ – ability to keep control of the Congress paperwork and figures and to remember names and faces was remarkable, as was her ability to remain a picture of tranquillity when all seemed to be going wrong. David Breeze was a tower of strength throughout the three years up to the Congress. His ability to process an enormous volume of work, smooth away problems and still manage to be the life and soul of the Congress never ceased to surprise. Bill Finlayson of CBRL could never have anticipated what he was inheriting from Alison McQuitty when he took on the directorship of the Amman institute. Throughout the Congress I don’t think we had a crossed word (well, may be one in the bar at Petra), and he developed a remarkable ability for being able to exist on 2 or 3 hours sleep a night. The decision was announced in the Closing Session at Amman that the next Congress would assemble in Pecs, Hungary in 2004. As I write the arrangements for that conference are already well-advanced and soon to be circulated. Following the end of the Amman Congress and the relief thereof (this was the second conference I was involved in organising in 2000 and this the third publication associated with a conference in the same period), I did not envy the task Prof. Visy and his team agreed to take on. Now, however, with the completion of the final part of the commitment made by the Amman organisers, I think we are in the position to repeat the offer made in Amman and to help organise the next Congress there. We propose the year 2100. Phil Freeman Liverpool, June 2002

xii

Fig. 1. HRH Prince Hassan opening the Congress. Platform party, from left to right: Mr Adrian Sindall, Mr Akel Biltaji, Professor David Breeze and Professor Siegmar von Schnurbein.

Fig. 2. Some members of the Congress at the Opening Ceremony, from right to left: Dr Arend Hubrecht, Mrs Chava Gichon, Professor Mordechai Gichon, Professor Tom Parker and Dr Renate Miglbauer.

Fig. 3. The audience at the Opening Ceremony: Dr Christine Meyer-Freuler, Professor Colin Wells, Dr Susanne and Dr Heinrich Zabehlicky; Dr Vivien Swan, Dr Michaela Konrad, Dr Sonja Jilek, Dr Andrea Faber and Dr Markus Gschwind; Dr Martin Hartmann.

Fig. 4. The reception at the British Ambassador’s Residence. From left to right: Dr Pia Eschbaumer, Dr Sebastian Sommer, Dr Michaela Konrad, Dr Erika Fischer and Professor Thomas Fischer.

xiii

Fig. 5. The reception at the British Ambassador’s Residence: Dr Herma Stiglitz and Dr Teodora Tomasevic-Buck.

Fig. 6. Professor David Breeze, Professor David Kennedy and Dr Zsolt Visy at the Reception.

Fig. 7. Dr Eszter Istvánovits, Dr Valeria Kulcsar, Dr Ioanna Bogdan-Cataniciu and Dr Dan Isac at the Reception.

Fig. 8. Dr Phil Freeman explains Azraq.

xiv

Fig. 9. Visiting Qasr Bshir: transferring from the coaches to pick-up trucks.

xv

xvi Fig. 10. The Congress at Lejjun.

Timetable of sessions, presentations and events In the three years spent organising the Congress the Organisers received innumerable offers of papers. Throughout the same period offers were often withdrawn or titles and themes altered. Up until and indeed during the Congress we continued to receive notices, mainly of withdrawals. In a small number of cases individuals who had offered papers failed to appear and failed to inform us of their decision. One result of these changes was that the timetable for papers and sessions had to be changed daily. It created especial problems with regard to some of the sessions devoted to papers discussing the Roman provinces of eastern Europe. As a consequence, certain sessions had to be cancelled or elsewhere merged with others. In some instances, one or two speakers chose, at their own discretion, to move their presentation to another session. The following timetable is, to the best of our records, an accurate summary of who spoke and approximately when. It omits those who did not appear at the Congress and or who have not contacted us subsequently. Please note titles of papers listed here may not necessarily match those published here. Likewise, the sessions organised at the Congress have been reorganised for the Proceedings. Key. ¶ = a paper presented at the Congress but not included in these Proceedings # = a paper that was offered to the Congress but which for one reason or another could not be presented on the day. It is however included in these Proceedings.

Saturday September 2 10.00 Plenary Session Welcomes and Opening addresses HRH Prince Hassan-bin-Talal, accompanied by HRH Princess Sumaya-a-bint-Hassan Mr Akel Biltaji, Minister of Tourism & Antiquities Mr Adrian Sindall – Chairman CBRL Professor Siegmar v. Schnurbein Professor David Breeze Prof. Antony Birley

Fifty years of Roman frontier studies

LUNCH Dr. Fawzi Zayadine Prof. David Kennedy Dr. Philip Freeman Prof. S. Thomas Parker Dr. Fawwaz al Khraysheh

¶ The Nabateans ¶ The Roman army in Jordan from the air ¶ Research on the Arabian frontier:1800-1983. The Arabian frontier: the current state of research. ¶ The Romans in Safaitic inscriptions

6.00-8.00 RECEPTION AT THE BRITISH AMABASSADOR’S RESIDENCE HOSTED BY THE BRITISH EMBASSY AND THE CBRL

Sunday September 3 Morning Sessions Session A: The Eastern Frontier Prof. David F. Graf Nomads and the Arabian frontier: the epigraphic evidence. Prof. Michael R. Werner ¶ Roman military architecture on the Upper Moesian and Arabian frontiers: a comparison of fort types. Dr. Bill Leadbetter Galerius and the eastern frontiers Prof. Ariel Lewin Diocletian: politics and limites in the east Mr. P.J. Casey ¶ Byzantine coinage and frontiers Session B: Roman Fortifications Prof. Zsolt Visy Similarities and differences in the late Roman defence system in the European and eastern frontiers Prof. David Breeze Ditches for defence: towards a study of multiple fort ditches Dr. Birgitta Hoffman The rampart buildings of Roman legionary fortresses

xvii

Prof. Raymond Brulet

Raetia Dr. Claus-Michael Hüssen Dr. Markus Gschwind Dr. Andreas Schaub

¶ Occupation et abandon au Ve siecle des fortifications tardives de la Gaule

Neue Forsuchungen an des Donau und im raetischen Limesgebiet The Flavian timber fort Abusing at Eining on the upper Danube Ausgewählte neue Befunde zur frühen Kaiserzeit in Augsburg

Session C: Early Roman Germany Dr. Marjan Galestin Early Roman military activities on the Frisian coast Dr. Susanne Wilbers-Rost Kalkriese und die Varusschlacht – Archäologische Nachweise einer militärischen Auseinandersetzung zwischen Römern and Germanen Dr. Armin Becker Der augusteische Stutzpunkt von Waldgirmes an der Lahn Gabriele Rasbach Der spätagusteische Stutzpunkt Lahn-Waldgirmes (D) – Die Funde Dr. Paul Bidwell The earliest Roman military baths Dr. Norbert Hanel Zur Datierung der frühkaiserzeitlichen Militärlager von Novaesium Prof. Siegmar v. Schnurbein Neue Grabungen in Haltern, Oberaden und Anreppen The Germanies Dr. Sebastian Sommer

Recent developments in south-west Germany (Eastern Germania Superior – Western Raetia) Prof. Jan Kees Haalebos† Recent excavations at the Roman fort at Alphen ad den Rijn Afternoon Sessions Session A: The Eastern Frontier Prof. Jodi Magness ¶ The Roman legionary pottery from Binyanei Ha’uma, Jerusalem Ms. Tali Erickson-Gini Nabataean or Roman ? Recent excavations in the army camp in Avdat. Mr. Yotam Tepper # Lejjun – Legio in Israel: results of a survey in and around the military camp area. Mr. Haim Goldfus Excavations of the Roman siege complex at Masada - 1995 Session B: The Roman Army Dr. J. Pamment Salvatore ¶ The emergence of the legionary cohort as an army unit in Roman Spain in the late Republican period: evidence from archaeology and elsewhere. Prof. Denis Saddington The Roman auxilia in the east: different from the west ? Mr. Guy Stiebel ¶ The arms and armour of the Bella Judaica – a retrospect Session C: The Germanies Marcus Reuter ¶ Western troops at the eastern front – auxiliary units from Germania Superior and Raetia in the Parthian wars of Septimius Severus Christine Meyer Freuler & Andrea Hagendorn Neue Forsuchungen zur Frühzeit von Vindonissa. Dr. Piotr Kaczanowski ¶ Zur Übernahme der römischen Bewaffnung und der römischen Kampfweise von der Bevölkerung des mittel-und nord-europäischen Raums Dr. Harry van Enckevort The eastern canabae legionis IX Gemina at Njmegen Dr. Jan Thyssen The late Roman fortress at Njmegen 4.00-700 TOUR OF ROMAN AMMAN. The itinerary included the Citadel, its museum and the city theatre and was led by Dr. Ignacio Arce of the Spanish Archaeological Mission in Jordan.

Monday September 4 VISIT TO NORTH JORDAN. The itinerary included visiting the fort at the oasis at Azraq, a picnic lunch at Safawi Research Station, a brief stop at the milecastle at Qasr Huweinit, a tour of the fort at Deir al-Kahf, passing through the fort at Umm il-Quttein, a long stop at the Nabataean, Roman and Byzantine town at Umm il-Jimal, and finally sightings of the via nova Traiana.

xviii

Tuesday September 5 Morning Sessions Session A: The Eastern frontier Dr. Zbig. T. Fiema The Petra papyri and the military presence in southern Jordan in the C6th Dr. Martin Hartmann & Dr. Michael A. Spiedel Military camps at Zeugma on the Euphrates Prof. Michaela Konrad ¶ Late Roman frontier policy in Syria. New investigations on late Roman fortifications in the province of Syria Euphratensis. Prof. Marcus Lenoir Le camp de la légion IIIe Cyrénaique à Bostra. Recherches récentes Mr. George Findlater Resource control and management in southern Jordan Mr. Lars Petersen Eine römische Befestigungsanlage in Kamid el Loz (Libanon) Dr. Ignacio Arce ¶ Building techniques interchanges in the Middle East: perseverance, change and synthesis. Dr. Theodor Kissel # A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus in the second century AD Session B: Fleet & Frontiers Dr. Boris Rankov Now you see it. Now you don’t. The British fleet in Vegetius. Dr. Maureen Carroll Belgic, Frisian and North Sea supply lines of the Classis Germania Dr. Norbert Hanel Recent research on the fortifications of the headquarters of the Classis Germania: Köln-Marienburg (Alteburg) Prof. Thomas Fischer Neue Forsuchungen zum Flottenlager Köln-Alteburg Prof. Denis Saddington ¶ The Classis Syriaca and Roman occupation on the Euphrates Dr. Perlina Varon ¶ Marines in legio X Fretensis Session C: The Germanies Dr. Meinrad Filgis Streifenhaustypus im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes (2 Jh. N. Chr.) Dr. Stefan Neu & Matthias Riedel Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr. Mr. Thomas Grane Problems concerning the early Roman fortifications at Marktbreit and Valkenburg Dr. Katrin Roth-Rubi Dangstetten Markus Scholz Späte Umbauten und Reduktion in Kastellen des obergermanischen limes in 3. Jahrhundert H.U. Nuber ¶ The late Roman fortifications between Oedenurg (F) and Kaiserstuhl (D) 2.30 – 10.00 JORDAN.

VISIT TO JERASH WITH DINNER HOSTED BY THE DEPT. OF ANTIQUITIES OF

Wednesday September 6 Morning Sessions Session A: The Eastern Frontier Dr. Nabeel Atallah ¶ Civilian and military functionaries at Umm il-Jimal Dr. Chaim Ben David A newly discovered Roman road connecting Zoar – Safi and the King’s Highway Prof. Israel Roll Crossing the Rift Valley: the connecting arteries between the road networks of Judaea-Palaestina and Arabia. Dr. Susan Weingarten Road use in late antique Palestine Prof. Mordechai Gichon Forty-five years of research on the limes Palaestinae – the findings and their assessment in the light of the criticisms raised. Session B: The Spains Dr. Angel Morillo Cerdan & Victorino Garcia-Marcos Twenty years of Roman military archaeology in Spain C. Fernandez Ochoa, V. Garcia Entero, F. Fil Sendino & C. Valenciano The gate of the Roman wall of Gijon (Asturias, Spain). New considerations.

xix

Dr. Martin Luik

Renieblas, Lager V: Ergebnise der Vermessungskampagnen 19972000. Prof. Jose Lopez Quiroga & F. German Rodriguez-Martin Les ‘castra’ et les ‘castella’ aux extrémités de l’Empire après la fin de la domination romains: La ‘Gallecia’ et les Agri Decumates’ au Ve siecle. Victorino Garcia-Marcos & Dr. Angel Morillo Cerdan The legionary fortress of VI Victrix at Léon (Spain). The new evidence (1995-1999). Rome and Parthia Prof. Everett T. Wheeler Prof. Edward Dabrowa

Roman treaties with Parthia: Völkerrecht or power politics ? “…ostentasse Romana arma satis…” The military factor in RomanoParthian relations under Augustus and Tiberius. Prof. Josef Weisehöfer (in absentia) Manicipia Caesaris and the head of Crassus. Roman images of Parthians and Parthian images of Romans in Augustan times. Dr. Menai Davies The circumvallation at Hatra Session C: The Pannonian provinces Ms. Eszter Istvánovits & Valéria Kulcsár The history and perspectives of the research of the so-called limes Sarmaticus in the Great Hungarian Plain Raimund Kastler The legionary fortress at Carnuntum in late antiquity. The classical encampment of the Roman legions in a phase of change. Martin Mosser Vindobona and its fortress – new excavations and topography 7.00DINNER HOSTED BY HIS ROYAL HIGHNESS PRINCE HASSAN AT THE MERIDIEN HOTEL, AMMAN

Thursday September 7 VISIT TO CENTRAL JORDAN. The itinerary included a visit to the fort at Qasr Bshir, lunch at the fort of el-Fityan, tour of the legionary fortress of el-Lejjun and a visit to the Nabataean temple at AlQasr.

Friday September 8 Morning Sessions Session A: Documents & Archives Prof. Antony Birley The Vindolanda writing tablets: a review. Dr. John Pearce Diet and dining at Vindolanda Andreas Thiel ¶ Praefecti et tribuni – Zur Rangordnung der militia equestris Prof. Margaret Roxan Veteran settlement in the Roman auxilia. An overview of progress so far. Dr. Mark Hassall ¶ Technical drawing in the de Rebus Bellicus: the liburna and the ballista fulminaris. Prof. Ioana Bogdan-Cătăniciu Dacian ancient geography (Ptolemy, Tabula Peutingeriana, Ravennatus) and Roman strategy Session B: The Balkans Prof. Mirjana Sanader Militärlager Tilurium – ein wichiger Punkt in der Verteidigung des römischen Provinz Dalmatien Dr. A.W. Dunn # Was there a militarisation of the southern Balkans during late antiquity ? The Danubian provinces Dr. Grzegorz Domaëski ¶ Entwicklung der militärischen Kraft mitteleuropäischer barbarischer Völker bis zu den Markomannenkriegen Session C: Britain II Dr. David Woolliscroft Dr. Fraser Hunter Mr. Alan Whitworth

The Roman Gask frontier (1997-2000) Roman and native in Scotland: recent work and approaches ¶ The archaeological recording of Hadrian’s Wall

xx

Dr. Julian Bennett The building of Hadrian’s Wall: design, logistics and chronology. Dr. Lindsay Allason-Jones Material culture and Hadrian’s Wall Mr. Tony Wilmott The Birdsoswald sector of Hadrian’s Wall 1949-1999 Dr. Nick Hodgson “Where did they put the horses ?” revisited: the recent discovery of cavalry barracks in the Roman forts at Wallsend and South Shields on Hadrian’s Wall. Dr. Peter Wilson Cawthorn Camps: 70 years after Richmond Afternoon Sessions Session A: The Egyptian Desert and North Africa Prof. Steven Sidebotham The Roman empire’s south-eastern-most frontier: recent discoveries at Berenike and environs (Eastern Desert of Egypt 1998-2000). Dr. Anne Haeckl ¶ The Schola of the Palmyrene at Berenike. New evidence for Palmyrenes in Egypt and the Roman military at the turn of the third century. Dr. Gabriele Ziethen (in absentia) Merchants, pilgrims and soldiers on the Red Sea route Dr. Alan Rushworth Defensores Provinciae: the militarisation and frontier zone élites in C3rd Mauretania Caesariansis. Session B: Upper and Lower Moesia Ms. Snezana Golubovic Decorated lead sarcophagi in Moesia Superior Dr. Gerda v. Bülow Iatrus – spätantikes Kastell oder befestige Zivilsiedlung am Unterdonaulimes ? Dr. Andrew Poulter ¶ Gradishte. A new late Roman fortress and supply base on the lower Danube. Dr. Sven Conrad & Dimităr Stančev Prospektionen am Donau-Lies zwischen Novae und Sexaginta Prista. Vorläufige Ergebnisse. Prof. Andrzej Biernacki The Roman legionary bath at C2nd AD Novae (Moesia Inferior) Dr. Elena Klenina Some remarks about Roman and Byzantine pottery from Novae (Moesia Inferior) Dr. Mihail Zahariade ¶ Tetrarchic building activity on the lower Danube frontier Dr. Piotr Dyczek Supply of the valetudinarium at Novae – some considerations Session C: The Black Sea and the provinces of Asia Minor Dr. Julian Bennett Developments in the study of the Cappadocian frontier Mr. Oleg Savelya # Excavation of a new Roman base near Balaklava in the Crimea (1992-1999) Prof. Teodora Tomasevic-Buck ¶ Vier römerzeitliche Siedlungen am Dondelta Jeorjios Martin Beyer “ίνα ...αύτοί άλλοις Βοράδοι καί Γόθοι γίνωνται.” Gregorius Thaumaturgos and die pontischen Beutezüge dr Boranen und Goten in 3 Jh. n. Chr. Mr Ergün Lafli # Seleuceia Sidera in Pisidia (south-western Turkey) in the Roman period 6.30-7.30 VISIT TO AMERICAN CENTER FOR ORIENTAL RESEARCH AND RECEPTION HOSTED BY THE DIRECTOR OF ACOR

Saturday September 9 Morning Sessions Session A: Dacia Dr. Liviu Petculescu Mr. Dorel Bondoc Prof Donea Benea

The military equipment of oriental archers in Roman Dacia # Artillery troops dislocated (sic) on the northern frontier of Dacia ¶ Die palymrenischen Truppen aus der Provinze Dakieu: Organisierung, Struktur, Entwicklung, Kulterelle Integration. Prof. Miroslava Mirković # Deserted forts – the Moesian limes after the conquest of Dacia Mr. Cristian Gazdac Monetary circulation and the abandonment of auxiliary forts in Dacia Dr. Dan Isac ¶ The Roman fort at Capeiu (Samum) as part of the northern line of the limes in Dacia

xxi

Session B: Britain II Dr. J.L. Davies & Dr. R. Jones Recent research on Roman camps in Wales Dr. Alison Taylor Supplying the frontier zone: the rôle of the East Anglia Fens Dr. Bob Bewley ¶ Aerial survey, fieldwork and the limes: examples from northern England and North Africa. Session C: General Themes Prof. Karl Strobel Die Legionen des Augustus: Probleme der römischen Heeregeschichte nach dem Ende des Bürgerkrieges (30 v. – 17 n. Chr.). Prof. William Hanson Why did the empire cease to expand ? Dr. Heinrich Zabehlicky Northern eastern frontiers and climate, c.AD200 Prof. Lukas de Blois The on-set of crisis: The reign of the emperor Septimius Severus (AD222-235). Dr. Marcin Biborski ¶ Neue Angaben über die spätrömische Bewaffnung im Lichte der Funde des 2.-4. Jh. Aus dem barbarischen Europa Dr. Johann van Heesch Mints and the Roman army from Augustus to Diocletian Prof. Nicolae Gudea Legio III Gallica and its European periplus in the Severan period Dr. Sara Phang Roman soldier’s marriages 3.00

CLOSING SESSION Professor David Breeze Profess Zsolt Visy

5.00 RECEPTION HOSTED BY THE DARAT al-FOUNOUN – ABDEL HAMEED SHOMAN FOUNDATION

Sunday September 10 VISIT TO SOUTH JORDAN. This tour involved a visit to the fort at Da’janiya, a drive past the medieval castle at Shobak and a tour of the Roman and Byzantine site at Udruh before arriving in Petra that evening and a reception kindly hosted by the PETRA REGIONAL COUNCIL

Monday September 11 VISIT TO PETRA & RETURN TO AMMAN. The bulk of the day was left free for a tour of the ancient city before departing for Amman late in the afternoon and back to the Congress hotels.

Tuesday September 12 DEPART AMMAN

xxii

Address by His Royal Highness Prince El Hassan bin Talal of the Hashemite Kingdom of Jordan at the 18th International Congress of Roman Frontier Studies Ladies and Gentlemen, Dear Friends It has been said that “…time is like a seal certifying existence, but like a seal it is artificial. Past, present and future are really just verses of the same poem. Our rôle is to trace its theme back to God.” I would like to suggest that rather than speaking of frontiers we speak of limes as we know from limetas, and to recall the Arab saying, “God has compassion on he who knows his limits.” In the context of 50 years to which you refer since the creation of this series of studies which I understand was initiated as a concept before the Second World War much has changed in our world. But I would like to return to Marcus Aurelius and cite his words, “In the vastness of the universe, Asia and Europe are just two small corners, all the seas are a single raindrop and Mt. Athos is a little clod; all of present time is but a pinpoint in eternity”.1 I would like to suggest that as we look at the second congress to be held in the southern provinces of Rome and in the series of 17 and about to become 18, that this conversation (and I believe in promoting what my professor of history and arts Theodore Zeldin used to refer to as the “noble art of conversation”) I would like to suggest that it took a very long time indeed for us to move from nomism, from one set of cultural values in a European context to a broader understanding of a multicultural conversation. As we speak of Rome, I would return to Giambattista Vico and also to Johann Gottfried von Herder and quote that they “…believed in the independence of cultures and the need to preserve each in its uniqueness.” I describe myself as a Berliner of the Isaiah type. It was Isaiah Berlin who suggested that, “…Rome, which crushed native civilisations in Asia Minor in order to produce one uniform Roman culture, committed a crime. The world was a great garden in which different flowers and plants grew, each in its own way, each with its own claims and rights and future. From which it followed that no matter what men had in common – and of course, again, there was a common nature to some degree – there were no universally true answers, as valid for one culture as for another”.2 Speaking about culture today of course is different to speaking about it in the C2nd and C3rd. As the frontiers between countries erode and the impact of new technologies and free trade become more prominent – and I speak here of the spirit of the Barcelona Declaration for co-operation in that Roman pond, the Mediterranean – where I seek to promote not medi terra but terra media (middle ground), I would just like to point out that there are three baskets of co-operation. You mentioned, Professor von Schnurbein, that money rules the world when it comes to paying for the tickets of participants to attend such conferences. It is interesting to note within the Mediterranean Charter and the Barcelona process, the baskets of security and economy precede the afterthought of the basket of culture and I would like to suggest that the following questions should be posed between the lines of our conversations. Why it is that the physical proximity that connected Mediterranean countries in the Hellenistic and Roman period, and in the later medieval centuries that tied Jews. Christians and Muslims in a web of inter-communal relations, they can no longer restore balance and tolerance today between cultures so similar that they revere the same Abrahamic origins? They are committed to the same covenant of Noah. In Amman last November, we hosted in the presence of His Majesty King Abdullah and in my capacity as elected Moderator of the World Conference on Religion and Peace, a conference on scenarios of reconciliation and among those scenarios of reconciliation in the Balkans. We had on the podium of representatives of the Catholics, the Orthodox, the Jewish and the Muslim communities and one of them said, “We revere the prophets – all the prophets. However, I would like to remind you of the Prophet Noah.” He said, and I believe it was the Muslim participant, “It was the Prophet Noah who created an ark for the salvation of humanity. Is it not time that we created an ark for the salvation of our common humanity?” So I asked why is it that some of us have chosen to retreat into inclusionist private space rather than developing the concept of the public realm so well known in the heyday of Rome? Why is it that some of us are determined to resist the ideas and influences of neighbours? I think that the answer is possibly that the concept of culture and the cultural promotion of our rich identity is sadly lacking. Turning to Arabia. I cite St. John Philby who described Arabia as it was seen from a northern or a western perspective. He said, “…Arabia was the lap of luxury from which the courts and epicures of the countries bordering on the Mediterranean drew their supplies of exotic aids to the enjoyment of life. And thence came frankincense and myrrh for the Egyptian mummies; apes and peacocks for the delectation of Solomon; gold for 1 2

Meditations Bk VI.36. The power of ideas. (London).

xxiii

the treasuries of the world; and spices for its kitchens.”3 I sometimes wonder whether the concept of province and providence were not blurred in the minds of the ancients. I would like to remind you that some of your number corresponded with me in 1990 during the tragedy of the Iraq/Kuwait affair – the on-going tragedy that we are still living – and the most moving lines of your correspondence was, “Remember the debt we owe Mesopotamia!” I don’t think we were talking of incense and myrrh. I think we were talking of our shared humanity. There is a difference of course, turning to the importance of money in our world between globalisation or and universalisation and I think you will agree with me that there is a difference between the recognition of the great contribution of Western culture to human civilisation and its impact on other cultures on the one hand, and the claim that modern Western culture, sui generis, is also capable of generating – we agree that it is capable of generating economic progress – but that it is also capable, without visiting the roots of the three concepts of democracy, of simple democracy as Professor Larry Siedentop, has described it recently in his book Democracy in Europe, democracy of government and democracy of society without revisiting the concepts of governance that prevail in this part of the world, the eastern Mediterranean. The sophist Livarnus has written on his grave the following rather touching and surprising words, “Il était habité par la passion du service public” – “he was touched by the love of the public good” as though this came as a revelation. I would like to suggest that this conference, this conversation, is in part about moving from private space to the public realm – from the provincial to the universal – in which we all share. I consulted the Oxford companion to archaeology and I would like to suggest that the disciplines addressed here include as we all know the ‘prehistoric’, the ‘historic’, the ‘classical’ and the ‘industrial’ and I personally, for what it is worth, would suggest that there is a great deal of “…similarity between the historical and classical”, but I would like to make a humble plea for the reference not only for “…the suite of societies in which classical archaeologists are primarily interested”,4 but also in the greater understanding of the peoples and their origins in this very rich eastern Mediterranean region. In his paper in 1986, Professor Parker, who I believe is among your number here today, referred to the “mid 3rd century (w)as the critical turning point in the evolution of many imperial frontiers …Of much greater importance is the nature of relations between Roman and the Arab tribes in this period. There were major tribal migrations, such as the movement of the Tanukh from the north-eastern Arabian peninsula into the Hauran. Werner Caskel”, Professor Parker reminded us, “argue(s) for the ‘bedouinization’ of the Arab tribes in this period. Caskel suggested that the now impoverished nomadic tribes were tempted towards increased raiding by the very weakness of the Empire during the mid 3rd century crisis.”5 I suppose as a descendant of those Bedouin tribes, I have to remind myself that when I visited the museum in Cairo years ago, every time I tried to dwell on the subject of the end of civilisation I was told by the guide that this was the work of Hyksos and finally he said to me, “… and they came from central Arabia.” I said I am happy to speak of the “the last ten years” as Professor Parker has described, “that have witnessed a remarkable advance in our understanding of the Roman frontier in Transjordan.” I would humbly return possibly not to the letter but to the point of Professor J. D. Bernal in his book Black Athena and suggest that possibly the civilisations of the more temperate zones had something to do with the development of the civilisations of the Mediterranean. “Continental states”, as we know, “have often been created from the ‘top’ downwards”. I would just like to say to something before concluding about governance – the public realm – “…that rulers relying on a Roman law tradition which regarded regional and local units of government as mere appendages of central government. The Roman Law tradition did not, that is, give the habit of local self-government a primordial value. In part that was due to the greater difficulty of state-making on the continent, where recalcitrant regions and cities were often subdued finally only by force or ruse”. I am intrigued to find the remains of Roman artists on the walls of Oxford – the remains of that extraordinary Ninth Legion that travelled from Romania to Britain and I ask myself if that was the contribution of people from this part of the world to what must have been a tremendous cultural exchange. You (Professor Breeze) referred to my linkage very kindly, and I, the humble servant of my Creator, as my dear friend would recognise me in Muslim terms, would like to suggest as you speak, of the importance of many that “…Karl Marx distinguished between a ‘class in itself’ and a ‘class for itself’. By that he meant that a class could exist objectively – identified by its occupation, for example – without having any consciousness of itself as a 3

H. St. John Philby The empty quarter. (London, 1933). George Michaels Archaeology as a discipline. In B. Fagan (ed.) The Oxford companion to archaeology. (Oxford, 1996): 42. S.T. Parker The Roman limes in Jordan. In A. Hadidi (ed.) Studies in the History and Archaeology of Jordan III (Amman, 1987): 151-164; esp. 154.

4 5

xxiv

class. A primary example of that in European history was the peasantry in the middle ages, a class that lived dispersed, local and ignorant, a class which lacked even a term for describing itself. In that way the peasantry differed, Marx argued, from the towns-peoples or bourgeoisie. If we apply this distinction to the rôle of Christianity, or to the rôle of Islam for that matter, it can be said that Christianity took humanity as a species in itself and sought to convert it into a species for itself. Thus, the defining characteristic of Christianity was its universalism. It aimed to create a single human society, a society composed, that is of individuals rather than tribes, clans or classes.”6 I would like to mention that travelling the world and speaking of Islam as I do, I find it important to develop what I call a radically tolerant platform or an aggressively moderate platform, if you will forgive the contradiction. (I am also a Marxist, but of the Groucho kind). It has been said that traditional Muslim society was ordered around a central political authority combining temporal and spiritual affairs for public space which was immediately shared by a variety of collective associations which included the merchants, the guilds, the ulama, Muslim and non-Muslim sects and tribes. Some of these groupings enjoyed considerable autonomy from central government in terms of power as well as resources. I would leave it to the distinguished participants to advise me to whether or not, the rise and flourishing of Christianity and Islam was not in a sense a reaction to the centrality of government of the Roman imperative. In the 630s the “…shift from a heavily fortified frontier manned by Roman troops to an essentially unfortified frontier protected by Arab federates was of course to have far-reaching consequences for Western history. It paved the way for the Muslim conquest of Transjordan, Palestine and Syria. But what if the Arabian frontier had been maintained as the strong level of the C4th? Could the empire have slowed or even blocked the expansion of Islam towards the Mediterranean world? How would the empire have been affected by the containment of Islam and continued Byzantine control of Syria, Palestine, Egypt and North Africa? One may only speculate about such questions, but certainly the neglect and decay of the south-eastern frontier proved to have momentous consequences for Western history”.7 May I suggest that in these fateful moments of contemporary Middle Eastern history the neglect and the decay of inter-communal relations in south-eastern Mediterranean, from the Balkans to the Levant, may have major drastic consequences for our shared universal commitment to peace. To know truth is to know our origins and to develop a conversation as we have tried in the Balkans where we took the conversation from Sarajevo to Kosovo and beyond. Small attempts at reconciliation, small attempts at conversation, whether in the Balkans, whether in Soweto where the subject today as we prepare for the year of dialogue for cultures and civilisations8 in which I am actively involved, whether in Irian Jaya, Ambon, Timor, where the archaeological history tells us a story of hundreds of years of civilising trade. Six hundred years of trade from the Atlantic Moroccan coast to Malabar and beyond. One day da Gama arrived off the coast of Malabar and said, “You will expel the traders – Jews, Christians and Muslims – or I will kill you” and the Jain king, a pacificist, answered with one question, “What is kill?” So, we recognise the importance of empire. I remind myself of my visit to the Turkish parliament some years ago as a descendant of a member of the Ottoman parliament. And recognising my ill-concealed British education, one of my Turkish hosts said, “Why is it that the British always speak of their commonwealth and our empire? Why do you not consider the millet system more closely? Why do you not give greater credits to the plurality of peoples and their contributions to the shared vision?” Ladies and Gentlemen, dear friends; I would like to pay tribute today to the work of the Council for British Research in the Levant and to its Chairman, Adrian Sindall, by saying that interest in the cultural and the humanitarian dimension is sometimes less significant than interest in the political and the economic dimension. Nonetheless although there are those of us who believe that the eternal nature of the land is the essential contribution of its peoples, the focus that is most timely and most welcome to the politicians is that they must turn to the thinkers for their inspiration and their ideas. I visited one of my professors who is 96 years old the other day – Professor Hassan Al Karmi. He received me at his door with walking stick in hand and we sat down and I said to him, “What did you think of the proposal of a 6

L. Siedentop Democracy in Europe. (Harmondsworth, 2000). S.T. Parker Roman and Saracens: a history of the Arabian frontier. (Winona Lake, Ind., 1986): 159. 8 2001 – United Nations year of Dialogue among civilizations 7

xxv

Parliament of Cultures?”“ A Parliament of Cultures is a concept I discussed with the late Yehudi Menuhin in the 90s in the Balkans. Professor Al Karmi said to me. “Young man before I answer you, let me just say one thing – I am glad you didn’t ask me about my health. I am 96 years old what I need is the stimulation of my mind. Then he asked why I did not call it a Parliament of Thinkers”. I answered that for some, unfortunately, Isaiah Berlin’s concept of The power of ideas is regarded as seditious. I think the time has come for us to re-focus The power of ideas which has been the salvation of civilisation universally and through the ages. I would like to pay tribute to a dearly departed colleague, Jim Sauer, to thank him for his work towards the legacy of Jordan and throughout the Middle East. I would like to join the chorus of gratitude for the work of past and present Directors of the Department of Antiquities, to Akel Biltaji – I would say that he follows in the spirit of those who like myself, try to take other seriously and not to take themselves so seriously. If I may be allowed to say in public that my daughter Sumaya has done a lot in travelling the length and breadth of archaeology in Jordan – well done! I wish you all well, and Professor Breeze may I say that we sowed a Gaelic seed some two or three years ago. I thought of Rome in terms of the north and I think you detected my confusion as to why you referred to Rome as in the south and south-east until I realised that you were doing it for reasons which we can all share – the reasons of expanding these horizons. You say this is the 18th Congress and the first after your 50th. It is quite extraordinary that in 1946 the Japanese initiated their studies of this part of the world. Their oriental institutes came into being in the Middle East in a country that neighbours the Middle Kingdom. They referred to us as the ‘Middle West’. I am aware of the fact that after the horrors of war and destruction, people pick themselves up and focus once again on what is important: on human dimension, on human communication. I wish you all an enjoyable and profitable stay and hope you will give me the opportunity of sharing with you Jordanian hospitality before you leave.

xxvi

Opening ceremony in the Royal Cultural Centre, Amman, 2nd September 2000, in the presence of HRH Prince Hassan bin Talal and HRH Princess Sumaya a bint Hassan Welcoming address by Professor David Breeze, Chairman of the Congress Your Royal Highness, your Excellencies, distinguished guests, member of the Congress of Roman Frontier Studies, welcome to the opening ceremony of the 18th International Congress of Roman Frontier Studies. The Congress of Roman Frontier Studies was founded in 1949 by Professor Eric Birley, and we will be hearing later from his son Professor Anthony Birley something of the history of the Congress over the last 50 years. Since 1949 the Congress has visited over 12 countries. This is the first time that the Congress has met in an Arab country and we are delighted to be joined here by colleagues from Jordan. The aim of the Congress of Roman Frontier Studies is two-fold: to explore a particular stretch of frontier and to learn about new research and consider new ideas about that frontier and other Roman frontiers. We are indeed most fortunate to be here in Jordan which contains so many impressive Roman military remains and we look forward to visiting many of these. Our interests are, of course, not restricted to Roman military sites and we will be visiting both Jerash and Petra as well as other Nabataean sites. There will be lectures on the Nabataeans and on the Safaitic peoples and their relationship with Rome. We also intend to return home with a much better understanding of Jordan, its people, their history and traditions. For many years the Congress of Roman Frontier Studies returned to Britain every 10 years to coincide with the decennial Pilgrimage of Hadrian’s Wall. We have recently broken with that tradition and thus there was no Congress held last year, the year of the 50th anniversary of the first Congress. However, I do believe that it is a sign of the strength and vitality of the Congress of Roman Frontier Studies that we have realised that it was time to move to a new pattern of meetings and more importantly, to seek new frontiers to explore. Certainly the vitality of the Congress is demonstrated by the attendance here in Jordan. About 250 archaeologists and scholars from nearly 30 countries and all 5 continents are here. We have a full lecture programme and an equally full itinerary during which we will explore Jordan from its northern boundary as far south as Petra. In preparing for this congress we have received considerable help from very many people including His Excellency the Minister of Tourism; Mr Akel Biltaji, Dr Ghazi Bisheh and Dr. Fawwaz el Kraysheh past and present Director-Generals of the Department of Antiquities. Your Royal Highness, you have been a constant source of support and encouragement, and behalf of my colleagues and myself, I should like to thank you most warmly for your contribution to this Congress. I now have great pleasure in asking Your Royal Highness to open the 18th International Congress of Roman Frontier Studies.

xxvii

xxviii

Jan Kees Haalebos

A few months after the close of the Congress, Jan Kees Haalebos died suddenly on 6 March 2001 in Nijmegen at the age of 60. I first recall meeting Jan Kees on the 1972 Congress in Romania. He and I formed part of a select band which embarked on the post-Congress excursion. In a small but sturdy coach we explored Transylvania. It was, quite simply, a magical experience, marred only for the Brits by the impossibly early starts each day. And, naturally, to the Brits, Jan Kees became “Holly bush”. Jan Kees became a regular attendee at the Congress of Roman Frontier Studies, and served on the Committee which organised the 1995 Congress at Rolduc. His other main international conference was the Fautores, of which he was President from 1992 until his death. Jan Kees’ home was Nijmegen, where he taught at the Catholic University, succeeding Jules Bogaers as head of the Department of Provincial Roman Archaeology in 1991. Between 1972 and 1996, often in conjunction with Professor Bogaers, he excavated at many sites along the limes, including Nijmegen, Woerden, Alphen aan den Rijn and Zwammerdam, where the laid-out fort as well as the excavation report is a potent memorial. He formed part of that distinguished band of Dutch archaeologists who have investigated Nijmegen and done so much to make the Roman fortress and town one of the best known sites in the Roman empire – even though so little is visible above ground: a signal achievement indeed. More recently, he and colleagues had commenced nonintrusive survey work on forts in Romania following the Zalău Congress in 1997. Quiet and taciturn he may have been, but Jan Kees was a firm friend, a sold rock for his students and fellow archaeologists, and a great hope for Dutch archaeology. His family and colleagues have suffered a grievous loss. David J Breeze

Fig. 11 Jan Kees Haalebos and his wife (to his right).

xxix

18th Limes Congress, Amman 9th September, 2000 Closing ceremony Speech given by Professor Zs. Visy Fifty years are not long in the flow of a conference series, but quite long in our life. The initiator and the participants of the first congress – it was a clear sign of renewal after the terrible time of WW II – cannot be with us. Eric Birley, András Alföldi, Jean Baradez gave us a two-fold example: that of the power of the hard and consequent scientific work and that of the power of the international scientific co-operation. The next generation took over the task and fulfilled the goals of the initiators. Many of them are missing now, Erich Swoboda, Rudolf Laurt-Belart, Jaro Šašel, Aladár Radnóti, Emil Condurachi and Constantin Daicoviciu, András Mócsy and Sándor Soproni, Petar Petrović and Teofil Ivanov, Anne S. Robertson and Jules Bogaers. The latest lost of this honorable company was István Ferenczi, who died this year. According to my duty I have to express our thanks for being here, for the possibility to take part in this Limes Congress. However, my – our – first thanks are due to all our predecessors, mentioned here by name or not, and let me ask you all to stand for a moment and to commemorate them. – Thank you! We thank of course other scholars who are not present, as Hans Schönberger or Harald von Petrikovitz, Sheppard Frere or Rudolf Fellmann and wish them our best. Last time we decided that we should have our next Congress, the 18th, in Jordan. Now we can say that it was a good decision. It is a country of ‘welcome’, which proved through many ways the high level of its culture, among other supporting historical research, archaeological investigations and conservation of ancient monuments. We could applaud at heart His Royal Highness Prince Hassan, who kindly greeted the Congress, and who gave two splendid lectures about the real values of cultural heritage, about the necessity of studying and knowing cultural traditions of every present and past nation; about humanism. We thank him. We also thank His Royal Highness Prince Hassan, Princess Sumaya and His Excellency the Minister of Tourism Mr. Akel Biltaji for supporting the preparations and organization of the 18th Limes Congress, and the same is to be expressed to the past and present director-generals of antiquities, Dr. Ghazi Bishir and Dr. Fawwaz al Khraysheh, the many officials of the Department including Ms. Nazmir Tawfiq, and all the officers who accompanied our visits and helped to make them successful. We have learned a lot. There have been given almost 150 lectures, and we tried to catch at least a third of them. The offer was enriched by the well-organized poster sessions. The contents were rich and abundant; it was often not easy to choose among them. There were colleagues who learned through listening, through discussion, and there was at least one who opened his head to let information directly into his brain. Many thanks for the lectures, many thanks for the new data and information, because we do know that to create and formulate every lecture you needed many weeks and months of hard fieldwork and intellectual activity. We have had an opportunity to get acquainted with the many historical and archaeological remains of the Roman heritage of Jordan. The excursions into the basalt desert, to Jerash or to Lejjun testified to the building activity of the Roman soldiers and civilians through seven centuries. The splendid handbook by David Kennedy became our guide in the fortresses of Qasr Aseikhin, Deir el Kahf, Qasr Bshir and other places. The sometimes up to two-three storey high surviving fortresses recall the monumentality of the Roman empire and for many of us show for the first time, how the fortifications, known only by their foundations, looked like in the reality. The visit in the ancient towns and surely that to Petra prove the high level of conservation and the growing importance of cultural tourism. All these are to be thanked for the 18th Limes Congress, and first of all its organizers and helpers. The main organizers were Dr. Philip Freeman and Liverpool University, Dr. Bill Finlayson and Ms. Alison MacQuitty of the Council for the British Research in the Levant. They were supported here and in Liverpool by Mrs. Pat Sweetingham, Ms. Susanne Schaubel, Ms. Samantha Dennis, Ms. Charlie Schriver and Professor Liz Slater of Liverpool University. The handbook was written by David Kennedy helped by Neil and Julie Kennedy, and edited by Philip Freeman. It was type set and seen through the press by Caroline Middleton in 5 weeks. We thank all the chairmen of lecture sessions and all the site guides, first of all Philip Freeman and Professor Tom Parker. The Congress was kindly supported by several institutions: the British Academy, the British Embassy, who paid for the buses, the British Council, the Jordanian Department of Antiquities, the University of Liverpool. The hosts for receptions were the British Embassy, ACOR, Ministry of Tourism, Petra Municipality, Abdel Hameed Foundation.

xxx

Last but not least let me mention by name the colleagues, who have taken the spiritual leadership of the Limes Congresses for many years: Siegmar von Schnurbein and here and now especially David Breeze. Many thanks for their kind and thorough work! Coming here I have found a broken inscription. I wrote it out and tried to restore its text. Let me show you with the request of helping me to learn it right.

[PRO] SAL ET INCOLVMITAT [DOMI]NI ET COLLEGAE NOST [DAVID BR]EEZE ARCHAEOL [DISC VNIV D]VRHAM PROF 5

[ERICH BIR]LEY DOCENTE

[DOCTISS PHIL]OSOPH IVVANTE [BRIAN DOBS]ON INVESTIGAT [LIMITI]S IMP ROMANI VALLI [HADR]IANI ANTONIIQVE ITEM 10

[HI]ST RERVM MILIT ROMAN

[AV]CTORIS PERMVLT STVD ORGANIS CONGRES STVDIOS LIMITIS ROMANI AMICI COLLEGAE COMMILIT 15

EX TOTO ORBE ROMANO IN

PHILADELPHIA CONGREGATI DEVOTISSIMI NVMINI EIVS [Pro] sal(ute) et incolumitat(e) / [domi]ni et collegae nost(ri) / [David Br]eeze archaeol(ogi) / [disc(ipuli) univ(ersitatis) D]urham(ensis) prof(essore) /5 [Erich Bir]ley docente / [doctiss(imi) phil]osoph(iae) iuvante / [Brian Dobs]on investigat(oris) / [limiti]s Romani valli / [Hadr]iani Antoninique item /10 [hi]st(oriae) rerum milit(arium) Roman(arum) / [au]ctoris permult(orum) stud(iorum) / organis(atoris) congres(sorum) studios(orum) / limitis Romani / amici collegae commilit(ones) /15 ex toto orbe Romano in / Philadelphia congregati / devotissimi numini eius.

xxxi

Contributors addresses * denotes a paper that was offered for the Congress but which could not be delivered in person. It has, however, been able to include it in these Proceedings. Ms. Lindsay Allason Jones University & Society of Antiquities of Newcastle upon Tyne, Museum of Antiquities, University of Newcastle, Newcastle upon Tyne NE1 7RU, UNITED KINGDOM Benny Arubas Ben-Gurion University of the Negev, POB 653, Beer-Sheva 84105, ISRAEL Dr. Armin Becker Landesamt für Denkmalpflege Hessen, Abteilung Archäologische und Paläontologische Denkmalpflege, Schloss Biebrich/Ostflügel, D-65203 Wiesbaden, GERMANY Dr. Chaim Ben-David Moshav Keshet, 12410, ISRAEL Dr. Julian Bennett Dept. of Archaeology & History of Art, Faculty of Humanities & Letters, Bilkent University, 06533 Bilkent, Ankara, TURKEY. Dr. Jeorjios Martin Beyer Sonderforschungsbereich 295, ‘Kulturelle und Sprachliche Kontake’, Johannnes Gutenberg-Universität Mainz, D55099 Mainz, GERMANY. Dr. Paul Bidwell Arbeia Roman Fort & Museum, Baring Street, South Shields, Tyne & Wear, NE33 2BB, UNITED KINGDOM Professor Andrzej Biernacki Ekspedycja Archaeologiczna w. Novae, Uniwersytet im. Adama Mickiewicza, ul. Św. Marcin 78, 61-809 Posnań, POLAND Professor Antony Birley Ulmenweg 11, D-61169 Friedberg, GERMANY Professor Dr. Ioana Bogdan Cătăniciu Institutul de Arheologie si Istoria Artei, Str. C. Daicoviciu No. 2, 3400 Cluj-Napoca, ROMANIA Dorel Bondoc* Muzeul Olteniei Craiova, Str. Madona Dudu 44, Craiova 1100, ROMANIA Professor David J. Breeze Historic Scotland, Longmore House, Salisbury Place, Edinburgh EH9 1SH, SCOTLAND. Dr. Gerda v. Sommer Bülow Römisch-Germanische Kommission des Deutschen Archäologischen Instituts, Palmengartenstraße 10-12, D-60325 Frankfurt a. M, GERMANY Dr. Maureen Carroll Dept. of Archaeology & Prehistory, University of Sheffield, Northgate House, West Street, Sheffield S1 4ET, UNITED KINGDOM Dr. Angel Morillo Cerdán Area de Arqueología, Depto. Estudias Clásícos, Facultad de Filosofia y Lettras, Universidad de León, Campus de Urgazana s/n, 24071 León, SPAIN. Dr. Sven Conrad Römisch-Germanische Kommission des Deutschen Archäologischen Palmengartenstraße 10-12, D-60325 Frankfurt a. M, GERMANY

Instituts, (Projekt Iatrus/Krivina),

Professor Edward Dąbrowa Uniwersytet Jagielloński, Instytut Historii, ul. Golębia 13, PL 31-007 Kraków, POLAND

xxxii

Professor Dr. Lukas de Blois Katholieke Universiteit Nijmegen, Postbus 9103, 6500 HD, THE NETHERLANDS Dr. Gwyn Davies 75 Coombe Lane, Raynes Park, London SW20 0BD, UNITED KINGDOM Dr. Jeff Davies, Dept. of History & Welsh History, University of Wales – Aberystwyth, Hugh Owen Building, Penglais, Aberystwyth, Ceredigion SY23 3DY, UNITED KINGDOM Patrizia Donat Staatarchäologie Wien, Friedrich-Schmitt-Platz 5/1, A-1082 Wien, AUSTRIA. Dr. Archie Dunn* Centre for Byzantine, Ottoman & Modern Greek Studies, University of Birmingham, Edgbaston, Birmingham B15 2TT, UNITED KINGDOM Dr Piotr Dyczek Instytut Archaeologii, Uniwersytet Warszawski, ul. Źwicki i Wigury 97/99, 02-089 Warszawa, POLAND Dr. Harry van Enckevort Gemeente Nijmegen, Bureau Archeologie, Postbus 9105, NL-6500 HG Nijmegen, THE NETHERLANDS Virginia Garcia Entero Facultad de Filosofia y Letras, Ciudad Universitaria de Cantoblanco, 28049 Madrid, SPAIN Mrs. Tali Erickson-Gini Moshar Kodish Barma, DN Halurtza 84900, ISRAEL Dr. Zbigniew T. Fiema Institutum Classicum, Pl 4 (Vuorikatu 3A, Fin – 00014, University of Helsinki, Helsinki, FINLAND Dr. Meinrad Filgis Landesdenkmalamt Baden-Württemberg, Archäologische Denkmalpflege, Pf. 102937, D-70025 Stuttgart, GERMANY George Findlater Dept. of Archaeology, University of Edinburgh, Old High School, Edinburgh EH1 1LT, SCOTLAND Professor Thomas Fischer Archäologisches Institut der Universität zu Köln, Abt. Archäologie der Römischen Provinzen, Kerpener Straße 30/Eingang Weyertal, D-50923 Köln, GERMANY . Dr. Marjan Galestin Afdeling Archaeologie, Groninger Instituut voor Archeologie, Ryksuniversiteit Groningen, Poststraat 6, 9712 ERGroningen, THE NETHERLANDS Victorino Garcia-Marcos C/San Alvito 2, 1/C 240003, León, SPAIN Cristian Gazdac, C/o Merton College, Oxford, OX1 4JD, UNITED KINGDOM Professor Mordechai Gichon 14 Assael Street, Zahala, Tel Aviv 69083, ISRAEL Dr. Haim Goldfus Ben-Gurion University of the Negev, POB 653, Beer-Sheva 84105, ISRAEL/ 1 (Old) Rahbag Street, Jerusalem 93302, ISRAEL Ms. Snezana Golubovic Archaeological Institute, Knez Mihailova 35/IV, Belgrade, YUGOSLAVIA Professor David F. Graf Dept. of History, PO Box 248107, University of Miami, Coral Gables, Florida 33124-4662, UNITED STATES OF AMERICA

xxxiii

Mr. Thomas Grane Nordre Frihavnsgade 16 l.tv, DK-2100, Copenhagen, DENMARK Stefan Groh Österreiches Archäologisches Institut, Franz Klein-Gaße 1, A-1190 Wien, AUSTRIA Dr. Markus Gschwind C/o Institut für Vor- und Frühgeschichte und Provinzialrömische Archäeologie, Ludwig-Maximilians Universität, Geschwister-Scholl-Platz 1, D-80539, München, GERMANY Professor Nicolae Gudea Facultatea de Teologie Greco-Catolicǎ, Universitatea Babeş-Bolyai Cluj-Napoca, Str. Moţilor nr.26, 3400 ClujNapoca, ROMANIA Andrea Hagendorn Chapfstrasse 4, CH-5210 Windisch, SWITZERLAND Dr. Norbert Hanel Archäologisches Institut – Archäologie der römischen Provinzen, Universität zu Köln, Albertus-Magnus-Platz, D50923 Köln, GERMANY Professor William S. Hanson Dept. of Archaeology, University of Glasgow, Gregory Building, Lilybank Gardens, Glasgow G12 8QQ, SCOTLAND Dr. Martin Hartmann Moselstrasse 4, CH-5417, Untersiggenthal, SWITZERLAND Dr. Johann van Heesch Cabinet des Médailles Bibliothèque Royale de Belgique, Boulevard de l’Empereur 4, B-1000, Bruxelles, BELGIUM Dr. Nick Hodgson Arbeia Roman Fort & Museum, Baring Street, South Shields, Tyne & Wear, NE33 2BB, UNITED KINGDOM Dr. Birgitta Hoffman Dept. of Classics, University College Dublin, Belfield, Dublin 4, IRELAND Dr. Fraser Hunter Dept. of Archaeology, National Museums of Scotland, Chambers Street, Edinburgh EH1 1LF, SCOTLAND Dr. Claus-Michael Hüssen, Römisch-Germanische Kommission des Deutschen Archäologischen Instituts Forschungsstelle, Jesuitenstraße 3, D-85049 Ingolstadt, GERMANY. Eszter Istvánovits Jósa Andras Múzeum, Nyíregyhaza, H-4400 Benczúrtér 21, HUNGARY Marcus Jae Institut für Provinzialröm. Archäologie, Universität Freiburg, Glacisweg 7, D-79098 Freiburg i. Br., GERMANY Ms. Rebecca Jones, RCAHMS, John Sinclair House, 16 Bernard Terrace, Edinburgh EH8 9NX, SCOTLAND Dr. Raimund Kastler Salzburger Landsearchäologie, c/o Salzburger Museum Carolino Augusteum, Alpenstrasse 75, A-5020 Salzburg, AUSTRIA Dr. Thomas Kissel Institut für Alte Geschichte, Sonderforschungsbereich 295, ‘Kulturelle und sprachliche Kontakte’, Johannes Gutenberg-Universität, D-55099 Mainz, GERMANY. Elena Klenina St. Mihaylovskaya – 9, apt. # 153, Sevastopol 99002, UKRAINE Erik Klingenberg* C/o Dr. Gabriele Ziethen, Landesamt für Denkmalpflege, Abt. Archäologische Denkmalpflege, Amt. Mainz, Große Langgasse 29, D-55116 Mainz, GERMANY

xxxiv

Mag. Michaela Kronberger Stadtarchäologie, Magistrat der Stadt Wien, Friedrich-Schmidt-Platz 5, A-1080 Wien, AUSTRIA. Valéria Kulcsár Petifi Muzeum, H-2170 Azod, HUNGARY Ergun Lafli* Theresienstraße 52, Zimmer 03, D-50931 Köln, GERMANY Dr. Bill Leadbetter School of Education, Edith Cowan University, Mount Lawley Campus, 2 Bradford Street, Mount Lawley, Perth, Western Australia 6050, AUSTRALIA Professor Maurice Lenoir AOROC – umr 8546 CNRS-ENS, Ecole normale supérieure, 45 rue d’Ulm, 75230 Paris Cedex 05, FRANCE Professor Ariel Lewin CP 1511, 50121 Firenze, ITALY Dr. Martin Luik C/o FB III Alte Geschichte, SFB Projekt Wirtschaftsstrukturen, Universität Trier, D-54286, Trier, GERMANY Dr. Ingrid Mader Geschäftsgruppe Kultur – Stadtarchäologie, Magistrat der Stadt Wien, Friedrich-Schmidt-Platz 5/1, A-1080 Wien, AUSTRIA. Dr. Christine Meyer-Freuler Oberhaslistrasse 15, CH-6005 Luzern, SWITZERLAND Professor Miroslava Mirković* Filozofski fakulet, Čika-Ljubina 18-20, YU-Beograd, YUGOSLAVIA Martin Mosser Stadtarchäologie, Magistrat der Stadt Wien, Friedrich-Schmidt-Platz 5, A-1080 Wien, AUSTRIA Dr. Stefan Neu, Archäologische Bodendenkmalpflege, Römisch-Germanisches Museum, Roncalliplatz 4, D-50667 Köln, GERMANY Carmen Fernandez Ochoa Isla de Mallorca 28, Urbanización Sierra Real, Soto del Real, Madrid 28791, SPAIN Professor S. Thomas Parker Dept. of History, PO Box 8108, North Carolina State University, Raleigh, NC 27695, UNITED STATES OF AMERICA Dr. John Pearce Centre for the Study of Ancient Documents, 67 St Giles, Oxford OX1 4TD, OXFORD Dr. Liviu Petculescu Muzeul Naţional de Istoire al Romaniei, Calea Victoricei 12, 70012 Bucuresti, ROMANIA Mr. Lars Petersen Hackersteigle 3, D-72076, Tübingen, GERMANY Dr. Sara E. Phang, Dept. of Classics, University of Southern California, THH 224 USC, Los Angeles, CA 90007, UNITED STATES OF AMERICA C. Valenciano Prieto C/o Fernando Gil Sendino, FMC Ayto Gijón c/ Zovellanos 21, 33201 Gijon, SPAIN Dr. López Quiroga Historia y Filosofía, Antiguo Colegio de Malaga, Universidad de Alcalá, C/. Colegios, 2, 28801 Alcalá de Henares (Madrid), SPAIN Dr. Boris Rankov Dept. of Classics, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UNITED KINGDOM

xxxv

Dr. Gabriele Rasbach Römisch-Germanische Kommission des Deutschen Archäologischen Instituts, Palmengartenstraße 10-12, D-60325 Frankfurt a. M, GERMANY Dr. Matthias Riedel Archäologische Bodendenkmalpflege, Römisch-Germanisches Museum, Roncalliplatz 4, D-50667 Köln, GERMANY F.Germān Rodriguez Martin Calle Bella Vistas 9, E-28411 Moralzarzal, (Madrid), SPAIN Professor Israel Roll Dept. of Classics, The Lester & Sally Entin Faculty of Humanities, Tel Aviv University, POB 39040, Ramat Aviv, Tel Aviv 69978, ISRAEL Dr. Katrin Roth-Rubi Lorrainestrasse 32, Postfach, CH-3000 Berne 11, SWITZERLAND Dr. Margaret Roxan C/o Institute of Archaeology, University College London, 31-34 Gordon Square, London WC1 0PY, UNITED KINGDOM Dr. Alan Rushworth The Archaeological Practice, Dept. of Archaeology, University of Newcastle, Newcastle upon Tyne NE1 7RU, UNITED KINGDOM. Professor Denis B. Saddington Dept. of Classics, University of Witswatersrand, Johannesburg, SOUTH AFRICA Sylvia Sakl-Oberthaler Staatarchäologie Wien, Friedrich-Schmitt-Platz 5/1, 1082 Wien, AUSTRIA Professor Mirjana Sanader Dept. of Archaeology, University of Zagreb, I. Lucica 3, 10 000 Zagreb, CROATIA Dr. Oleg Savelya* Str. Yavena – 14, q. 37, Sevastopol 99028, UKRAINE Dr. Andreas Schaub, Römisches Museum/Stadtarchäologie Ausburg, Dominikanergasse 15, D-86150 Augsburg, GERMANY Professor Dr. Siegmar v. Schnurbein Römisch-Germanische Kommission des Deutschen Archäologischen Instituts, Palmengartenstraße 10-12, D-60325 Frankfurt a. M, GERMANY Markus Scholz Institut für Provinzialrömische Archäologie, Universität Freiburg, Glacisweg 7, D-79098 Freiburg i. Br., GERMANY Helga Sedlmayer Österreiches Archäologisches Institut, Franz Klein-Gaße 1, A-1190 Wien, AUSTRIA Fernando Gil Sendino FMC Ayto Gijón c/ Zovellanos 21, 33201 Gijon, SPAIN Professor Steven Sidebotham Dept. of History, College of Arts & Science, University of Delaware, 236 John Munroe Hall, Newark, DE 197162547, UNITED STATES OF AMERICA Dr. C. Sebastian Sommer, Landesdenkmalamt Baden-Württemberg, Archäologische Denkmalpflege, Pf. 102937, D-70025 Stuttgart, GERMANY Dr. Michael A. Speidel Historisches Institut, Abt. für Alte Geschichte und Epigraphik, Universität Bern, Laenggassstrasse 49, CH3000, Bern 9, SWITZERLAND

xxxvi

Dimităr Stančev C/o Dr. Sven Conrad, Römisch-Germanische Kommission des Deutschen Archäologischen Instituts, (Projekt Iatrus/Krivina), Palmengartenstraße 10-12, D-60325 Frankfurt a. M, GERMANY Professor Karl Strobel Institut für Geschichte, Universität Klagenfurt, Universitätstrasse 65-67, A-9020 bKlagenfurt, AUSTRIA Alison Taylor 40 Hertford Street, Cambridge CB4 3AG, UNITED KINGDOM Yotam Tepper* 32 Habrosh St., Binvamina 30500, ISRAEL Dr. Jan Thyssen Gemeente Nijmegen, Bureau Archeologie, Postbus 9105, NL-6500 HG Nijmegen, THE NETHERLANDS Dr. Wolfgang Vetters Österreiches Archäologisches Institut, Franz Klein-Gasse 1, A-1190 Wien, AUSTRIA Professor Zsolt Visy Mádach U5/A, H-7623 Pécs, HUNGARY Dr. Susan Weingarten 9 Rehov Katz, Petah Tiqva, 49553, ISRAEL Professor Josef Wiesehoefer,* Institut für Klassische Altertumskunde, Universtität Kiel, GERMANY Prof. Everett T. Wheeler Dept. of Classical Studies, Duke University, 236 Allen Building, Box 901013, Durham, North Carolina 277080103, UNITED STATES OF AMERICA Dr. Susanne Wilbers-Rost Varusschlacht im Osnabrücker Land, Museum und Park Kalkriese, Venner Strasse 69, D-49565 BramscheKalkriese, GERMANY Tony Wilmott Centre for Archaeology, English Heritage, Fort Cumberland, Fort Cumberland Road, Eastney, Portsmouth PO4 9LD, UNITED KINGDOM Dr. Pete Wilson English Heritage, Fort Cumberland, Fort Cumberland Road, Eastney, Portsmouth PO4 9LD, UNITED KINGDOM Dr. David Woolliscroft The School of Art History & Archaeology, University of Manchester, Oxford Road, Manchester M13 9PL, UNITED KINGDOM Dr. Heinrich Zabehlicky Österreiches Archäologisches Institut, Franz Klein-Gaße 1, A-1190 Wien, AUSTRIA Dr. Gabriele Ziethen* Landesamt für Denkmalpflege, Abt. Archäologische Denkmalpflege, Amt. Mainz, Große Langgasse 29, D-55116 Mainz, GERMANY.

xxxvii

xxxviii

Fifty years of Roman frontier studies A.R.Birley In this paper the following topics and themes are reviewed: 1: The origins of the first congress (Newcastle 1949) and its relationship to the Hadrian’s Wall Pilgrimage will be explained. 2: The location and publication of the subsequent congresses will be reviewed briefly. 3: The effect of contemporary politics on thinking about Roman frontiers (eg. the Iron Curtain) will be mentioned. 4: Other congresses dealing with frontiers, particularly the ones devoted to the eastern frontiers, will be discussed. 5: New thinking on Roman frontier policy and strategy (eg. in books by Luttwak, Isaac and Whittaker) will be reviewed. 6: Important new discoveries (eg. Kalkriese and Walgrimes in Germany) which may throw light on frontiers and frontier policy will be summarised.

Your Committee has done me the honour of inviting me to open the proceedings by a kind of retrospective. I have to admit to being just an (old-fashioned) ancient historian rather than an archaeologist; but at least I have been attending these gatherings since the IVth Congress in 1959, if not every one of them after that; and have contributed papers since 1964. All the same, the principal reason for the Committee’s choice is no doubt that a son of the late founder and Life President, Eric Birley, represents some kind of continuity.

surveying the fort of Carrawburgh on the Wall. Newbold fell in action in 1916, when the 5th Pilgrimage should have been taking place. It was held, in restricted form, in 1920, with Collingwood, who had succeeded to Haverfield’s rôle as the leading light in Romano-British archaeology, now taking a prominent rôle along with Simpson, as again at the 6th, in 1930, when the 24 year old Eric Birley first attended. Eric Birley had already dug on the Wall - sent there by Collingwood - and had made contact with German research; Ernst Fabricius of Freiburg, doyen of Limesforschung in Germany, sent his Assistent, Kurt Stade, to the Birdoswald excavation in 1929; and Gustav Behrens, a special delegate from the German Limeskommission, joined the Pilgrims in 1930.

Fifty years: another anniversary in the bimillennial year. Anniversaries - and contemporary political developments have affected the timing of Roman Frontier Studies from the start. It all began with or from Hadrian’s Wall, as a byproduct of, or a supplement to, the so called Pilgrimages, intended, when they were re-established in 1886, to be decennial, a sequence interrupted by two World Wars. The seed was sown in 1848: John Collingwood Bruce, a Newcastle clergyman, had planned to visit Rome, but was deterred by the revolutions of that year, which affected much of Continental Europe, notably in Spain, France, Germany, the Austrian empire and Italy. Instead, he took a small party along the nearby ‘Roman Wall’, then not yet generally known as Hadrian’s Wall. The success of this little tour prompted a request for him to repeat it in 1849 for a larger group, as a kind of ‘Pilgrimage’.1

Scientific Limesforschung in Germany had by then been under way for several decades. After half a century of debate, the Reichslimeskommission had finally been established in 1892, led by the great Theodor Mommsen. It was a high-powered organisation, with military backing. A contrast to the British arrangements, where in both England and Scotland the local archaeological societies had been in charge: respectively of Newcastle-upon-Tyne and of Cumberland & Westmorland, the Glasgow Archaeological Society and the Society of Antiquaries of Scotland. It is worth noting - because it is sometimes assumed that the British mural studies were inspired by those in Germany - that Mommsen, when arguing in 1890 for the creation of the RLK, had actually stressed the Vorsprung achieved by English and Scottish research on their respective Walls.2 Be that as it may, support and funding from the top in the revived German Kaiserreich was a contrast to the situation in Britain; official funding for Limesforschung was also matched by Kaiser Wilhelm II’s patronage of the reconstruction of the Limeskastell at the Saalburg (celebrated with some pomp in 1900).

Bruce soon began a long series of publications about the Wall; and a great deal of research followed. The Pilgrimage was revived in 1886 as a tribute to the now 81 year old Bruce, and a decennial cycle was initiated: the Pilgrims of 1896 and 1906 could review much new work. Local enthusiasts, notably F.G. Simpson, still took the lead, but Oxford University was now involved, through F.J. Haverfield (who had been closely in touch with Mommsen) and soon R.G. Collingwood - the latter, it is remarkable, was a philosopher, later indeed Professor of Metaphysical Philosophy, as well as an archaeologist - as was also, from 1910, a lecturer at Armstrong College, Newcastle, Philip Newbold. The fateful August 1914 and the Great War followed - Collingwood was briefly arrested as a suspected German spy shortly after its outbreak when

2

This point is worth making, since the reverse process (German inspiration for Wall research) was stressed by M. Browning at the Canterbury Congress (XV, 1989), overlooking eg. E. Birley's comments at Carnuntum (XIV, 1986) - his reference there to Mommsen's statement of 1890 was based, clearly, on the paper by J. Irmscher Die Begründung der Limesforschung in Deutschland. Corolla Memoriae E. Swoboda Dedicata (Graz & Cologne 1966: 137-145); 143 to which one should turn for more detail.

1

A chronological listing of Pilgrimages and Congresses, with some bibliographical details, is offered in the Appendix at the end of this paper.

1

Limes XVIII

So much for the prehistory. To prepare this paper it was necessary to re-read all the proceedings of our Congresses and of the, let us say, parallel - rather than rival - collective enterprises on eastern frontiers, also a selection of some other published colloquia, as well as various significant monographs. The hope that it could all be digested properly proved illusory. There is by now so much of it and, while some has proved ephemeral, so much is excellent. I repeatedly asked myself: why on earth did not I read this paper or that one properly before? Of course, I have to stress, as an excuse, Roman frontier studies cannot be a full time occupation, for most of us at least. There are no Institutes of Limesforschung or Chairs of Roman frontier studies. There is, to be sure, the RömischGermanische Kommission; and there is now a ‘Director of Hadrian’s Wall’, an officer of English Heritage - but his rôle is not to carry out full-time research. Publication of our proceedings has been, anyway, very heterogeneous: not many libraries have the whole series, they never seem to be shelved together, and not many individuals could afford to buy them all.

late antiquity and Byzantine times, that saw a Roman military presence here in the east. A brief word on arguments about the term limes,4 or indeed about the use of the words Grenze, frontière, boundary, and the rest: at least to this speaker they seem slightly misconceived. Surely we know what we are dealing with: the regions on the edge of empire, where the Roman army was stationed to exercise control over those within and those without - clearly to some extent stationed with a view to future expansion if this was held desirable (for whatever reason); to intercept and defeat invasion from without; to crush rebellion within; to ensure the collection of revenues; to keep the peace. The troops had to be put somewhere. Political reasons may initially have played a big part in stationing them as far away from the centre as possible, after the repeated bouts of devastating civil war in the Mediterranean that ended in 30 BC. Hence ‘Roman Frontier Studies’ involves not just investigating the remains of the old frontier works and all aspects of life in the ‘frontier zones’: in a sense it means Roman Army Studies.

Hadrian and his Wall - and the Rhine-Danube limes, especially the ORL - were the starting-point, the springboard. Probably with quite misleading effects for the interpretation of Roman frontiers, some now think. Linear running barriers, with heavy concentrations of garrisoned forts on or close to them, or for that matter, the long European river-lines with regularly disposed legions and auxiliary regiments strung out along them, were not matched in the east. They were not needed there - the desert (the most effective kind of frontier in Napoleon’s view3) was sufficient barrier and there were plenty of towns or ‘nucleated settlements’ to house many of the troops, unlike in wild northern Britain - or along Rhine and Danube and their Vorland (at least when the Romans first arrived there). Besides, many argue, the construction of running barriers meant no significant development: it is claimed Hadrian’s Wall in Britain and his palisade in Germany were not essentially different from the FlavioTrajanic systems in Britain and Germany. Still, the symbolic nature of Hadrian’s initiative remains: he was drawing a line, maybe not least sending messages to his own still expansion-hungry military men: ‘No more expansion!’

By 1939 Eric Birley had been for eight years Lecturer in Romano-British History and Archaeology at Durham University, which by then had its own Excavation Committee, concentrating on the Wall. The 7th Pilgrimage was due in 1940, and he had plans, which he discussed with Kurt Stade, for an international congress to be held after it. Both men must have sensed that the chances of it taking place were rapidly diminishing. By the time Eric Birley went to the international archaeological congress in Germany in August 1939, he must have known the thing was off for the foreseeable future. He only stayed for three days, meeting among others Howard Comfort and Andreas Alföldi. Shortly after he arrived, a coded telegram from the British War Office summoned him to London (he was already an officer of the Special Reserve).5 Within days he was serving in Military Intelligence, in a branch initially called MI 3b, soon re-labelled MI 14; and devoted the next six years to analysing the German Wehrmacht instead of the Roman army. Eric Birley gave his own retrospectives on three occasions, at Cardiff (1969), Aalen (1983) and Carnuntum (1986). I shall not repeat all he said there, only summarise some salient points. It all goes back to 1949 when the 1st Congress finally took place - a small-scale affair (nowadays it would probably be called a colloquium or a workshop). This was the earliest feasible date for both

Aside from these considerations, another reason not to make all this research revolve around Hadrian’s Wall is that the frontier zones represented by the Wall - still more, the forward line in Scotland - were, when all is said and done, relatively short-lived, lasting at most under 300 years. Contrast the many centuries, from the C1st BC to

4

C. Zuckermann Historia 17 (1998): 108-128, esp. 112ff., has presented some compelling ancient evidence (from hitherto neglected texts) which seems to justify the - now much criticised - modern use of the term limes (but it is fair to note that B. Isaac, as Zuckermann himself registers, was in 'désaccord' with his conclusions. No doubt further debate may be awaited). 5 His passport shows that he entered Germany on 20 August and left on the 23rd in haste. He left his pyjamas with his friends the Nesselhaufs, who were able to return them after the war. He did deliver his paper (it was not published).

3

Cf. Correspondance de Napoleon Ier, vol. 30 (Paris 1870) 10: 'De tous les obstacles qui peuvent couvrir les frontières des empires, un desert pareil à celui-ci [the Sinai] est incontestablement le plus grand. Les chaînes de montagnes, comme les Alpes, tiennent le second rang, les fleuves le troisième'. This must be the passage quoted (in English) by the Right Hon. Lord Curzon of Kedleston, Frontiers (The Romanes Lecture, Oxford 1907) 15f., as the verdict of 'the greatest Captain of modern times'.

2

A.R.Birley: Fifty years of Roman frontier studies

Pilgrimage and Congress, only four years after WWII ended; and 1949 marked the centenary of the 1st Pilgrimage. Stade could not attend: he was still a POW in the Soviet Union. Alföldi came, but not from Budapest. He was by then at Bern in Switzerland. The 1st Congress, limited and incomplete though it was, produced some important new information and adumbrated themes that are still very much with us. There were three separate surveys of the Rhine frontier (two by German scholars, one by Rudolf Laur-Belart of Basel) and one, by A.E. van Giffen, on three frontier forts in Holland; one by an Italian archaeologist, Antonio Frova, on the Danubian frontier in Bulgaria; papers on Wales by V.E. Nash-Williams and on the Antonine Wall by Anne Robertson (whose last contribution to a RFSC was to be at Canterbury in 1989) Hadrian’s Wall did not need to be featured, for that would have repeated what had been put across during the Pilgrimage; other papers were by Mortimer Wheeler - soon to become Sir Mortimer and a famous TV personality in the UK - on the Roman frontier in Mesopotamia; by the Danish archaeologist Norling-Christensen on Roman bronze and glass objects found in Denmark; and published only in a brief summary prepared by John Morris - Jean Baradez on what he called the Fossatum Africae. Baradez brought to Newcastle a copy of his great tome, ‘wet from the press’. Finally, as well as Eric Birley’s brief introduction, there was Andreas Alföldi on ‘The moral barrier on Rhine and Danube’. This is the only contribution that is still much quoted (not always with approval). Based almost exclusively on the literary sources, it has a timeless quality which reports on archaeological research generally cannot achieve. Needless to say, it was influenced by recent and current events: Alföldi’s war experience in embattled Budapest had been appalling; and he was personally affected by the emergence of the Iron Curtain, which led him to leave his beloved Pannonia. The papers, all in English, with a few maps but no photographs, were published in a slim volume in 1952.

we would have had at least four full reports on the Inchtuthil legionary fortress by its excavator, who died in 1965, a few days after completing his 14th season there).6 The 1949 gathering was felt to be successful and to deserve repeating: quinquennial congresses were proposed - clearly with the intention to alternate between a congress linked to the decennial Pilgrimages and one on another frontier, according to possibilities. The second one was planned for April 1954, to take place in Algeria. But the death of Louis Leschi, Director-General of Algerian Antiquities, made that impossible. Only a small group was able to visit the Numidian frontier under the guidance of Baradez. Subsequent developments in Algeria have so far ruled out a Numidian RFSC. Still, regular reports on Roman North Africa have been presented at other congresses and colloques. Here one may also recall (to mention only a selection) the work by the late Charles Daniels at Germa in the Sahara, clarissimum caput Garamantum, now being continued; the UNESCOsponsored research programme on the Libyan pre-desert, led by Graeme Barker and the late Barri Jones; the remarkable series of convegni di studio on Africa Romana launched by Attilio Mastino of Sassari; and René Rebuffat’s work at Bu-Ngem. A notable event in 1954 was that Mortimer Wheeler published Rome beyond the Imperial Frontiers; its inspiration was his own work in India, but it covers Germany beyond the Rhine and Africa beyond the limes as well. This will remain a vital theme for us. What E. Kornemann called ‘the invisible frontiers of the Roman empire’ and others ‘client-states’ (some of these were of course enclaves well within the frontiers) were only the first part of the zone beyond the frontier, subject to Roman hegemony in varying degrees. Further out still lay peoples subject to influences at least economic and perhaps cultural too - and in the east and south, with the rise of Christianity, a new kind of Roman hegemony would emerge, for example with the conversion of Ethiopia.

Had participants in the 1949 Pilgrimage and Congress been able to see how the subject was to have developed 50 years on, they would have had some shocks. That Hadrian’s Wall - the whole works, of course, not just the curtain and ditch - has been designated a “World Heritage site” might have been an agreeable surprise (though they would probably have wondered what exactly this meant in practice; the same distinction is now planned for the 550kms of the ORL). They might have been less enthralled by the development of archaeological bureaucracy. I refrain from further comment, except to say that I sometimes think that an ideal combination of subjects for a university degree would be Archaeology and Politics. Some changes in the UK are unquestionably welcome, and ought to be imitated elsewhere, eg. that permission is now required to excavate a scheduled site - and that after three years no further excavation is permitted until the results of previous work have been published. (What a difference that regulation would have made had it applied earlier: eg.

The setback over the Algerian congress might have caused the project to founder. But a new possibility opened. Erich Swoboda invited a number of those involved in the 1st Congress - and others - to a Tagung at Carnuntum in 1955: Austria, astonishingly, had a few months earlier achieved a new status, neutral and free of occupying troops. This was not planned as a Limeskongress, but was retrospectively deemed to have been the IInd RFSC. It included, after all papers by Baradez, on Roman Algeria and Pannonia; D. van Berchem, on the Marcomanni in the Roman army; A. Betz, on the Dislokation of the legions during the principate; Eric Birley, on Hadrianic frontier policy; J.P. Gillam, on Roman pottery in the north of Britain; LaurBelart, on the Vindonissa fortress; Ian Richmond, on Agricola in Scotland; W. Schleiermacher, on urbanisation 6

All the same, it is sad that Ian Richmond's name did not appear on the title-page of Inchtuthil: the Roman legionary fortress when it appeared 20 years later (1985), with those of L. Pitts & J.K.S. St Joseph.

3

Limes XVIII

in Raetia; J. Szilágyi, on excavations at Aquincum; and G. Walser, on the Roman reporting of conditions in barbaria.

Republic of the Southern Slavs to participate, even if the lion’s share in organisation was taken by Grga Novak, President of the Croatian Academy of Sciences. It should be stressed that the pre- and post-congress tours and the site visits during the programme were enormously valuable for the participants, at this congress as at others. There were a lot of ceremonies; the hospitality was lavish (slivovica and Steak Chateaubriand are enduring memories); and the programme was fairly exhausting, often behind schedule. The visit to the remains of Apollodorus’ bridge over the Danube was nearly postponed, but the congressees insisted - the river was exceptionally low and the surviving piers could be made out by clair de lune. On the final excursion from Skopje (Scupi) to Stobi some congressees could not help observing that the bus was being followed by vultures. Nineteen papers were published, with lavish illustrations, including ones on the Roman provinces of Raetia, Numidia, Britannia, Belgica, Pannonia, Germania inferior and superior, Thrace, Moesia superior and inferior - and, by Gichon, on Arabia; and several on multi-provincial or general topics; further, the full list of participants was published, with entire programme, including the material sent out in advance. It is impossible to resist mentioning the instructions to ‘The gentlemen whose reports will be accompanied by diapositives’. Only one lady, to be sure, contributed a paper, the numismatist D. VučkovićTodorović.

Laur-Belart then issued an invitation for 1957, to the IIIrd Congress, at Rheinfelden in Switzerland. Other locations were, it has to be said, still hardly feasible. One need only refer to the events of summer and autumn 1956 in the Near East and in Hungary. Peaceful, prosperous, permanently neutral Switzerland was ideal. Thus, if there was not yet exactly a north-western European bias, to have moved from Britannia to Noricum and Upper Pannonia, and then to Raetia, was a different kettle of fish from what had been hoped for: from Britain to Numidia and back. Of the contributions at Rheinfelden one may note those by Harald von Petrikovits, back from Soviet captivity and able to report on work in the lower Rhineland, and by Baradez, A. Radnoti, Richmond and Hans Schönberger. A friendly reviewer of the proceedings, published in 1959, lamented the lack of attention to the east and the desirability of a uniform format for the publications: neither deficit has really been made good. The next Pilgrimage was due in 1959, and the IVth Congress was duly held after it, at Durham, followed by an excursion to Scotland. Various circumstances combined to prevent publication, not least the unwillingness of Durham University to provide any funding. Eric Birley had slipped a disc, and could only join the last day of the Pilgrimage, leaning on a stick and attended the Congress proceedings lying at full length in the front row of the lecture theatre. (It may be noted that a printers’ strike had made it impossible to issue the Pilgrims with a handbook. Instead, two years later, they received something much more substantial: Research on Hadrian’s Wall, still of interest and value). There was, it should be noticed, a very full turn out: the eminent historian André Piganiol gave the opening lecture; and Baradez was back. Kurt Stade was also there (returning to northern Britain after 30 years), likewise Shimon Applebaum (who had been at the 1st Congress), with a fellow-countryman who was to become a regular attender, Mordechai Gichon; there were also, in spite of political difficulties, representatives from the middle and lower Danube, including T. Ivanov from Bulgaria and several colleagues from Yugoslavia - where an invitation to meet only two years later and thus to be able to see sites in Dalmatia, Pannonia inferior, Moesia superior and Macedonia, was accepted with enthusiasm.

Hans Schönberger, Director of the Saalburg Museum, issued the invitation to the VIth Congress, held at Arnoldshain in the Taunus in 1964. It was fitting that the RFSC should come to the territory of the old RLK and there was already a great deal new to see in southern Germany, not least in the Taunus and the Wetterau. The bias of the programme was still heavily towards Britain, Rhine and Danube (including a paper on the Vorland of the upper Danubian frontier), but there was at least a paper by Gichon on the limes Palestinae and two on Africa, M. Euzennat on Tingitana and Baradez with complements inédits to the ‘Fossatum’. There was also a splendid paper by Petrikovits on Annäherungshindernisse (amazing word). More people, myself included, should have read this paper properly. Schönberger modestly did not call himself editor of the excellently produced volume (published in 1967), which, it is a slight pity, lacks an introduction and a list of participants.

Hungary was still out of the question - so too was Czechoslovakia. The Slovaks had in fact organised their own small Limeskongress in 1957, published in 1959. A second one was announced for 1961, at Trenčin, the ancient Leugarico. Being engaged on a doctoral thesis dealing with the Marcomannic wars, I was eager to attend: an invitation was received, but no visa was forthcoming. The Vth Congress duly took place in Yugoslavia in September 1961: a biennial rhythm had been established. Many participants did not realise that the elaborate peripatetic programme of this congress had been designed to allow each of the constituent parts of the Federal

The invitation from Gichon to go outside Europe for the first time for the next Congress in April 1967 was understandably received with excitement, in spite of political problems. The VIIth RFSC, in Israel, which gave the chance of seeing recently excavated sites, including Masada, in parts of the Roman provinces of Judaea and Arabia, was indeed a memorable experience. All the same, Britain and the Rhine-Danube line still predominated, with some two-thirds of the 38 papers, most of the rest, appropriately enough, being by Israeli contributors (one other, on Augustan architecture at Ephesus, seems slightly out of place). The published result, for reasons which need 4

A.R.Birley: Fifty years of Roman frontier studies

not be spelled out, appeared after a long delay: the volume is dated ‘1971’ but it was some time after that before copies were available (and not all contributors received one).

provinces of the host country. Eric Birley, you will allow me to note, was there elected in absentia Life President, a position he was to hold for 25 years. I also mention that 1974 saw the publication of the first Vindolanda writingtablets, found by Robin Birley the previous year, which, to quote Brigitte Galsterer-Kröll (reviewing the first full publication of 19839), were regarded as a kleine Sensation, with some relevance for frontier studies.

Another development in 1967 which deserves brief mention was the launching of a new series on ‘The provinces of the Roman empire’ by the British publisher Routledge and Kegan Paul. Alas, only four volumes appeared, between 1967 and 1974, and the project was terminated.7 But an important by-product was to be work on Roman Arabia by Glen Bowersock, a member of the editorial committee. His volume on the province would eventually appear elsewhere.8

Then it was back to the Danube, at last to Hungary, for the XIth Congress, in 1976. This was published with record speed, using new methods, in 1977, the same year as the Akten of the Xth Congress. With remarkable self-restraint, no papers from the host country were included. Those who could not afford to attend - payment had to be in US dollars and accommodation was in part at the Budapest Hilton, ruling this out for many (the £ was then exceptionally weak) - rather regretted this omission from the published record, it must be added.

The timing – 1969 - and location - in Britain - of the VIIIth Congress were again predetermined by the 9th Pilgrimage. The Congress went to Cardiff, with a closing session in York. There were more contributions from eastern Europe than before, and four on general themes; but it has to be noted that of the 32 papers - published five years later - 27 dealt with Britain and the Rhine-Danube/Balkan area, with only Gichon on his first season at En-Boqeq, near Masada, going outside Europe. It was five years before the proceedings were published.

1976 saw the appearance of a work still much discussed, Edward N. Luttwak’s The Grand Strategy of the Roman Empire. ‘The outsider sees most of the game’, it is said, and the author, neither archaeologist nor ancient historian, but member of a high-powered US think-tank, had done his homework fairly well - and he had seen a gap in the market, so to speak. None of the Limesforscher, preoccupied with burrowing into the ground or into details of their multifarious sources, had managed to produce a synthesis, so Luttwak did it for them. Perhaps there has been rather too much fuss about this book, but it did get people thinking.

The invitation to go to Rumania for the IXth Congress in 1972 raised hopes that there would be a chance to see another province outside Britain with a different sort of frontier system from Upper Germany and Raetia, or from the river-lines: ie. Dacia. In the event, although there was a post-Congress tour to the limes Dacicus, the Congress was held on the Black Sea, at Mamaia, hence with a lower Danubian, or lower Moesian, Schwerpunkt among the 55 contributions, promptly published two years later, with, further, several dealing with the limes Scythicus, which was to become so important for Constantinople.

It was time for the Pilgrimage again, the 10th, in 1979, so the Congress returned to Britain for the fourth occasion, this time to the remote fringes of Britannia, indeed to Caledonia. The Scots equalled the Hungarian record of publication only a year after the Congress - and the XIIth RFSC needed three volumes, produced by BAR, the first volume, to be sure, entirely devoted to Britain, much of it to the Romans in Scotland. An innovation was the introduction of overviews of work in various provinces. But Africa and the east, if quite respectably represented, were still rather thin on the ground.

There was by now evidently some eagerness to meet as often as possible: the Congress became biennial again, going from the lower Danube to the lower Rhine for the Xth Congress in 1974. The publication, three years later, was of a high standard, including an excellent index. For whatever reason, the biggest group of papers, 20 in all, dealt with Pannonia, Dacia and Moesia, rather than the

At this point Stephen Mitchell justifiably felt that in the 12 RFSCs up to that point ‘the East, from the Caucasus to the Negev, ha[d] been seriously neglected’ and organised a Colloquium at Swansea in April 1981, on ‘Armies and frontiers in Roman and Byzantine Anatolia’. The result, published in 1983, again by BAR, is a superb collection, covering well over a thousand years of Roman and Byzantine history and, it should be added, going well beyond ‘Anatolia’ itself. Not all the contributions were published - my own was rendered out of date during the Colloquium by Jörg Wagner’s presentation of new

7

S.S. Frere Britannia (1967 and several reprints); J.J. Wilkes Dalmatia (1969); G. Alföldy Noricum (1974); A. Mócsy Pannonia and Upper Moesia (1974). I may be allowed to mention that Wilkes' Dalmatia volume was a revised version of a Durham PhD thesis supervised by Eric Birley, who had assigned other pupils a number of comparable subjects, covering eg. Mauretania Tingitana (J.E.H. Spaul); Numidia (M.G. Jarrett); and Cappadocia (R.P. Harper); and Wilkes' own pupil A.G. Poulter wrote a thesis on Moesia Inferior. For various reasons, none of these emerged as a monograph. 8 Roman Arabia (Cambridge, Mass. 1983). This volume, and various Vorarbeiten, were a major stimulus to research in Jordan. It is worth noting that further provinces, which might have been part of the Routledge series, were eventually covered by another excellent London publishing house, B.T. Batsford: Edith M. Wightman Gallia Belgica (1985); A.L.F. Rivet Gallia Narbonensis (1988); D.J. Mattingly Tripolitania (1995). Batsford, alas, was taken over by another firm in the 1990s; and then went bankrupt.

9

For details of publication cf. my contribution on Vindolanda in this volume; the quotation is from B.Galsterer-Kröll Bonner Jahrbücher 188 (1988): 655-8.

5

Limes XVIII

inscriptions from Osrhoene, eg. of the year 195, showing the procurator Julius Pacatianus marking out the boundary, fines, between the new provincia Osrhoene and the residual regnum Abgari. The volume includes a splendid paper by Ronald Syme on Tigranocerta read in his absence by Stephen Mitchell: perhaps Sir Ronald, although still only 78 - he remained active until his death 8 years later felt that conditions in a student hall of residence might be a bit Spartan.

thanks to advances in technology - and to the editors’ astonishing energy and powers of persuasion; again with BAR, and at a much improved standard. As the editors pointed out, ‘it has become a commonplace that in the study of the Roman limes the eastern frontier of the Empire is the poor relation....A striking feature of those attending [over seventy] was how few of them were regular attenders of Frontier Studies Congresses’ - true indeed, and something that should make us think hard. But cost, of course, is a factor: it was easier to go to Sheffield than to Mamaia, Székesfehérvár - or Aalen. (There is a sad footnote: the Thatcher cuts led not long after to the closure of the excellent Sheffield Dept. of Ancient History & Classical Archaeology.)

Be this as it may, a few months later Syme was off to Cantabria and Asturia for a conference held at Santander and Oviedo on ‘Indigenism and Romanisation in marginal areas of the Roman empire’ (‘marginal areas’ meaning ‘areas on the edge’). It was to mark a bimillenary - of the Roman conquest of north-west Spain in 19 BC. A year early, some of us thought, but this was pedantry, as we must accept in the annus bimillensimus. (The Romans themselves, after all, mostly celebrated decennalia etc. a year in advance.) It was an unforgettable experience for those who took part (including some RFS veterans). Alas, the typescripts went missing and most contributions, many of them unintelligible to non-Hispanists among the participants because of the speed of delivery, disappeared without trace or were recycled. Syme’s, on ‘The subjugation of mountainous zones’, came out in his Roman Papers V (1984). ‘Indigenism and Romanisation’ is an interesting concept, though the latter term, properly now regarded as problematic in any case, is especially so when applied to the east: what does ‘Romanisation’ mean there? All the same, Shimon Applebaum gave a splendid paper on ‘Indigenism and Romanisation in Palestine’ (never published, it seems).

Sheffield 1986 was followed by Ankara 1988, with a tour of the Euphrates as well. It was organised by David French and Chris Lightfoot, and published almost as quickly as Sheffield, in 1989, again with BAR. The quality of papers in both volumes is outstanding, a real eye-opener. Lack of time prevents me from dwelling on details, but Shelagh Gregory’s wonderful paper in the Ankara Colloquium, ‘Not “Why not playing cards?” but “Why playing cards in the first place?”‘, is in itself an argument for more interchange between east and west. It is very welcome that further volumes on these lines have appeared, edited by Edward Dabrowa in 1994 and by Kennedy in 1996. One may note here also two further sets of papers from the early 1990s, originating at Lyon and Nemours respectively. Here in Amman it is superfluous to hold forth at length on the great revival in eastern frontier studies, in particular Syria and Arabia, going back nearly a quarter of a century and associated with the names of G.W. Bowersock, D.F. Graf, D. Kennedy, H.I. MacAdam, S.T. Parker and M. Sartre - others could be cited.

The RFSC continued to pursue the traditional path, going for the XIIIth to southern Germany again, this time to Aalen in the province of Raetia in 1983; and in 1986, for a second time, over 30 years on, for the XIVth, back to Carnuntum. The Life President, it may be added, told me before each of these occasions that it would be his last Congress and therefore that he needed filial support. How glad I am that I believed him and took part again, after a gap of 16 years, since the Congresses, now greatly expanded in numbers of participants, are so valuable not just for the papers read and site visits but for the chance to meet friends and colleagues, old and new. (In these as in other respects Aalen was a model of its kind.) Eric Birley gave retrospective papers on both occasions. In the event he was able to go Canterbury in 1989 as well, for the XVth, aged 83. Both XIII and XIV resulted in massive publications, respectively at three and four years intervals; and both were still weighted to the Rhine and Danube frontiers. At Carnuntum, however, there was a major innovation: at David Breeze’s urging, there was more stress on the ‘natives’, with a special thematic session.

The timing and location of the XVth Congress, held at Canterbury in 1989, was again dictated by the Pilgrimage (the 11th). It was felt that this practice had outlived its usefulness; the decision not to hold the congress in Britain in association with the Pilgrimage of 1999 was painlessly accepted at the XVIth Congress at Rolduc - which was, however, not held until six years after Canterbury, in 1995. The plan had been to go for the XVIth to Serbia, in 1993. External events intervened. At first, the signs had been very positive. The collapse of the East German government and the dismantling of the Iron Curtain, especially its most odious subsection, the Berlin Wall, shortly after Canterbury, and two years later the dissolution of the Soviet Union, created an atmosphere of great optimism; and Serbian colleagues began active preparations. But the break up of the Federal Republic of Yugoslavia ensued and the series of hideous wars in that unhappy country - not, it is true, on Serbian soil; but all the same, it became clear that the XVIth Congress must go elsewhere. Our colleagues in the Netherlands and Belgium stepped into the breach and the XVIth RFSC was held at Rolduc Abbey, Kerkrade, in 1995. The location was particularly felicitous, a town on the old border between Germany and the Netherlands, which the creation of the European Union had

Before the XIVth took place, the Mitchell initiative had been followed up by David Kennedy and Philip Freeman at Sheffield in 1986: they actually published their Colloquium, full of excellent material, the same year, 6

A.R.Birley: Fifty years of Roman frontier studies

made into a line that could be stepped across without controls. At least some modern frontiers are disappearing.

geography and topography, law, onomastics (personal and place names), air photography, but - I will just reel off a list - specialised study not just of pottery, glass, leather, textiles, bone, but all kinds of environmental material; dendrochronology (C-14 is seldom much use for Roman sites); palaeobotany and the whole spectrum of archaeometry (ichnology, identifying animal-pawprints, is one branch of science I have only recently heard about); resistivity surveys; metal detectors (mostly used by nonspecialists, resulting however not least in an astonishing flood of military diplomas, of which so many new ones since CIL XVI and supplement have already been made accessible by Margaret Roxan); and of course, computers, e-mail and www. This is our first Internet Congress.

In the meantime two monographs had appeared which generated great interest and debate, by C.R. Whittaker, an ancient historian, first in French in 1989, re-issued in revised form in English in 1992, and by B. Isaac, an archaeologist, in 1990. I have my own views on both, but do not need to air them here; and they have probably been discussed quite enough.10 I note here also the valuable dissertation by A.D. Lee, published in 1993. Canterbury repeated the ‘Romans and Natives’ thematic session started at Carnuntum, and added one on desert frontiers, at which Colin Wells’ address as chairman deserves special mention, not least for his remarks on how modern nationalism and imperialism (with a post-colonial reaction) have affected research and its interpretation. Another valuable innovation was the poster sessions, which allowed more input in conditions more relaxed than in the lecture theatre. Rolduc, which also featured poster sessions, in the abbey cloisters, took these initiatives further, with a variety of thematic sessions, including one on ‘across the frontier’, but also allowing for a substantial set of ‘Miscellanea’, many fascinating; the Rolduc volume also appeared two years after the Congress.

Some essential tools for the all-round RFS-practitioner are not so well honed as they used to be: Latin and Greek, for example - but still, as there are people here, who know Arabic, Hebrew, Aramaic, Old Parthian, Phrygian, Coptic, and other languages which are a closed book to myself, I must not press this. Comparative studies are now ‘in’. Not merely comparing Roman frontiers with others (Lawrence Keppie’s Canterbury paper on the C18th American frontier is a superb example), but bringing in anthropology, economics, sociology and so on. ‘Models’ are favoured what old-fashioned persons prefer to call argument by analogy. A remarkable specimen is the notion of the Roman army as a ‘total institution’, apparently derived from a study by the American sociologist Ervin Goffmann, on psychiatric asylums. ‘Core and periphery’ and the longue durée are other favourites. Well, why not?

While the Balkans were in turmoil, the Near East seemed for a while to be on the verge of real peace. There was even talk of meeting in 2000 jointly in Israel and Jordan. But Rumania, strongly represented at Rolduc, won the vote, and the XVIIth Congress went to Transylvanian or Dacian Zalău in 1997. The Proceedings were published in 1999, but were not yet available outside Rumania in summer 2000. It is not possible to offer detailed comment on this substantial volume (953 pages), which includes traditional regional surveys with very full bibliography and, understandably, a heavy bias towards work in Rumania (about a third of the contents).11 Meanwhile the 1999 Pilgrimage, the 12th, duly took place without a Congress, and now we are here, for the XVIIIth.

One area of special progress is mapping. The Tabula Imperii Romani and Tübingen project, TAVO, must be mentioned, and this month the long heralded Barrington Atlas duly appeared (an advance copy could be inspected at the Congress). Meanwhile, the view that the Romans themselves were incapable of producing maps has gathered strength and has almost become a new orthodoxy. A new discovery, a papyrus so expensive that even the Getty Museum jibbed at the price its anonymous owner asked for, has been just partially unveiled. It is a C1st AD copy of the late C2nd BC Geography by Artemidorus of Ephesus - and it has illustrations - including a map, of Roman Spain. We await more details. But the article gives some support for the view that the Romans were not so helpless at cartography after all.12 There are also significant advances in research into trade, especially the supply of olive-oil by amphorae.13

Some concluding remarks are needed - on developments since 1949. Apart from digging-up Roman and native remains and, ideally, publishing and reporting the results promptly and intelligibly, practitioners of RFS now have to deploy: not just the good old historical/literary sources, epigraphy, numismatics, papyrology (or ability to read cursive, including now Vindolanda tablets and ostraka from Bu-Ngem and Mons Claudianus), historical

12 C. Gallazzi & B. Kramer 'Artemidor im Zeichensaal. Eine Papyrusrolle mit Text, Landkarte und Skizzenbüchern aus späthellenistischer Zeit', Archiv für Papyrusforschung 44 (1998): 189-208. One may note that this papyrus was a private person's copy. No doubt Roman generals campaigning in Spain, such as Cn. Pompeius, could have equipped themselves with maps of higher quality to accompany a text of Artemidorus. (I am grateful to R.W.B. Salway for drawing my attention to this article). 13 Note especially work by J. Remesal Rodríguez, eg. his La annona militaris y la exportación de aceite bético a Germania (Madrid 1986) and Heeresversorgung und die wirtschaftlichen Beziehungen zwischen der Baetica und Germanien (Stuttgart 1997).

10 One may note in particular the lengthy discussion by E.L. Wheeler Methodological limits and the mirage of Roman strategy, Journal of Military History 57 (1993): 7-41 & 215-240, dealing with Luttwak, Whittaker and much else. A revised and expanded version of this contribution in monograph form would be very valuable. 11 Mild dismay must be voiced at the difficulty, indeed impossibility, of procuring the volume from bookshops outside Rumania (I am grateful to V. Lica for presenting me with a copy), and at the price (260 DM) - no doubt justified by the high quality of paper; but most would prefer normal paper and a lower price.

7

Limes XVIII

A few other outstanding discoveries from the past 50 years rate a brief mention (just a personal selection): the inscription of M. Valerius Maximianus found in Algeria in the early 1950s, linking Leugaricio-Trenčin in Slovakia, 120 miles north of the Danube frontier, where Maximianaus had commanded a detachment in winter 179/180. with Diana Veteranorum in Numidia, where he served a few years later; the Tabula Banasitana (only published 14 years after its discovery in 1957), so revealing for Roman relations with ‘native’ peoples in Tingitana; the Hunt Pridianum, showing that Transdanubian territory was already annexed after Trajan’s 1st Dacian War; the Bar Kokhba documents (not fully published for decades), shedding so much light on a remarkable example of resistance to Roman rule; the inscription of Ti. Claudius Maximus, who cut off Decebalus’ head - also revealing something vital about Roman army pay scales; the inscription from Ruwwafa in the Hejaz, establishing ‘beyond doubt’ in Bowersock’s view ‘the incorporation of this region in the province’ of Arabia; the Vindolanda writing-tablets (there are more to come; I note here only that some ‘natives’ were labelled Brittunculi, an instructive, hitherto unrecorded word); the inscription from Jordan which carries the words praetensionem coligare, telling us something very important about how the Romans manned and named part of their frontier here in Arabia - and indeed in Britain, for Michael Speidel patruus, who discussed these words, has now spotted that the road going north from Hadrian’s Wall into the Vorland, with its outpost forts, was also called a praetensio; Marktbreit, a massive Augustan legionary base in southern Germany, first revealed from the air; Kalkriese near Osnabrück, surely at last the true site of the clades Variana; the Augsburg altar with its amazing account of a great battle between Romans and Germanic invaders in the year 260; Waldgirmes, an Augustan new town - not a military base, in the Lahn valley in Transrhenane Germany. We will hear more about some of these last in the next few days.14

by separate bodies, archaeological units or firms of archaeological contractors, who have to tender. As for work on finds, laboratories are expensive and heavy fees are charged, inflated, it seems, in the UK since money was made available from the state lottery. Modern methods seem very advanced, of course, in comparison with those prevailing 50 years ago. But this also involves, so it often seems, publishing everything that is found in full detail, which increases the price and diminishes the digestibility for the non-specialist - and none of us can be specialists right across the field. Another tendency to note is ‘demilitarisation’ of, for example, Roman Britain: is this a kind of pacifism; or has it something to do with the need to use historical sources, in Latin and Greek, when one deals with the army? Those trained as prehistorians find written sources a nuisance, even a kind of contamination, it sometimes appears. Anyway, one notes with relief the continued publication of M.P. Speidel’s valuable series on the Roman army, MAVORS.15 In conclusion, let me mention the latest work on our subject, by S.P. Mattern, thoroughly influenced by modern approaches; but our old friend the Wall of Hadrian appears on her dustjacket - the author concedes that it ‘must have seemed impressive, even terrifying’ to the natives, but goes on to pronounce that the frontier has ‘a nebulous quality’. Academic activity tends to generate argument, as it should; but sometimes, and RFS are no exception, descends into odium academicum. Let us hope that Limesfreunde, to use Eric Birley’s term for RFS Congressees, can do without this, above all here in Amman, Philadelphia.16 Appendix: Congresses etc. and publications on Roman frontier studies (1949-1999) and some prehistory (Publications on frontiers - or frontier-zones - of individual regions or provinces, notably of Arabia are mostly not here listed: the latter were, very properly, prominently discussed in the course of this Congress)

Professionalisation of our subject is now very marked: amateur archaeologists still play a part, but it is diminishing. It depends how you define ‘amateur’, of course. If it just means one who engages in research in his spare time and without payment, let us here note the contributions on the Cappadocian frontier by Timothy B. Mitford. It must be recalled that back in 1949 degrees in archaeology scarcely existed; and the rôle of local societies was still very important. In the meantime archaeology has sprouted up in universities all over. But the universities - in the UK at least - do not much practise Roman provincial archaeology. The subject is very much a stepsister compared with prehistory and classical archaeology. To refrain from other comparisons, one might say that it is, like North Carolina vis-à-vis its neighbours South Carolina and Virginia, a vale of humility between two mountains of pride. A great deal of work in the field is now carried out

1848

15

John Collingwood Bruce leads a small group along the Roman Wall in north England after the

Pietas demands reference here to a recent volume, edited by G. Alföldy, B. Dobson & W. Eck Kaiser, Heer and Gesellschaft in der Römischen Kaiserzeit. Gedenkschrift für Eric Birley (Stuttgart 2000). The 25 contributions, in English, French, German and Italian, cover a wide spectrum of Roman military affairs, including, of course, questions about frontier zones. I note also the initiative of Y. Le Bohec in organising two congresses, of which he edited the proceedings: La hiérarchie (Rangordnung) de l'Armée romaine sous le haut-empire (Paris 1995); and Les légions de Rome sous le haut-empire (2 vols., Lyon & Paris 2000) respectively taking classic works by A. v. Domaszewski and E. Ritterling as their starting-point. 16 David Breeze kindly offered comments on a draft; I am solely responsible for any remaining errors or omissions. Apart from minor verbal corrections and the addition of a few footnotes, this represents the text of the paper given at Amman. Like, I am sure, all participants, I would like to register here my very great appreciation of the work of the Organising Committee and of the warm friendship and hospitality extended to us in Amman and everywhere in Jordan.

14 I must refrain from supplying references for this purely personal choice of remarkable discoveries.

8

A.R.Birley: Fifty years of Roman frontier studies

revolutions on the Continent that year deter him from visiting Rome. 1849 1st Pilgrimage of Hadrian’s Wall 1851 Bruce publishes his first monograph on the Wall 1863 1st edition of Bruce’s Handbook (at first called Wallet Book; 2nd ed. 1884; 14th edition, by David Breeze, now in preparation) 1886 2nd Pilgrimage: it was agreed that there should thenceforth be decennial Pilgrimages 1892 Founding of the Reichslimeskommission in Germany 1896 3rd Pilgrimage 1906 4th Pilgrimage 1920 5th Pilgrimage, deferred from 1916 because of the Great War 1926 E. Fabricius publishes the article ‘limes’ in the Realencyclopädie 1929 Kurt Stade, Assistant of Fabricius, joins excavation at Birdoswald on Hadrian’s Wall 1930 6th Pilgrimage 1939 Eric Birley and Kurt Stade plan Congress of Roman Frontier Studies for 1940 in conjunction with 7th Pilgrimage; plan aborted by WWII 1949 7th Pilgrimage, deferred from 1940, timed to coincide with centenary of 1st one, followed by first Congress of Roman Frontier Studies at Newcastle-uponTyne. I: 1949. England: Britannia (inferior) E. BIRLEY (ed.) 1952: The Congress of Roman Frontier Studies 1949. (Durham). [1954: Second Congress, planned to take place in Algeria, cancelled; instead, J. Baradez conducts small group to Gemellae and stretches of the ‘Fossatum’ in April of that year] R.E.M. WHEELER 1954: Rome beyond the imperial frontiers. (London)

VI:1964. Germany: Germania (superior) H. SCHÖNBERGER (ed.) 1967: Studien zu den Militärgrenzen Roms. (Cologne). VII: 1967. Israel: Judaea/Syria Palaestina, Arabia S. APPLEBAUM (ed.) 1971: Roman Frontier Studies 1967. (Tel Aviv). 1969 9th Pilgrimage, followed by VIIIth Congress VIII: 1969. Wales: Britannia (superior) E.BIRLEY, B. DOBSON & M. JARRETT (edd.) 1974: Roman Frontier Studies 1969. (Cardiff). IX : 1972. Rumania : Moesia (inferior) D.M. PIPPIDI (ed.) 1974: Actes du IXe congrès d’études sur les frontières romaines. (Bucharest). X: 1974. Germany: Germania (inferior) D. HAUPT (ed.) 1977: Studien zu den Militärgrenzen Roms II. (Cologne). XI: 1976. Hungary: Pannonia (inferior) J. FITZ (ed.) 1977: Limes. (Budapest). E. N. LUTTWAK 1976: The grand strategy of the Roman empire. (Baltimore). 1979 10th Pilgrimage, followed by 12th Congress XII: 1979. Scotland: Britannia W.S.HANSON & L.J.F. KEPPIE (edd.) 1980: Roman Frontier Studies XII, 1979. BAR Int. Ser. 71. (Oxford). [J. ARCE, organiser: ‘Indigenismo y romanisación en las áreas marginales del imperio romano’. Conference at Santander and Oviedo, 1981 - not published] S. MITCHELL (ed.) 1983: Armies and frontiers in Roman and Byzantine Anatolia. (Colloquium at Swansea 1981) BAR Int. Ser. 156. (Oxford).

II: 1955. Austria: Noricum/Pannonia (superior) E. SWOBODA (ed.) 1956: Carnuntina. (Vienna).

XIII: 1983. Germany: Raetia C. UNZ (ed.) 1986: Studien zu den Militärgrenzen Roms III. (Stuttgart). P. FREEMAN & D. KENNEDY (eds.) 1986: The defence of the Roman and Byzantine east, (Colloquium at Sheffield 1986). BAR Int. Ser. 297. (Oxford).

III: 1957. Switzerland: Raetia R. LAUR-BELART (ed.) 1959: Limes-Studien. (Basel). 1959 8th Pilgrimage (E. Birley 1961: Research on Hadrian’s Wall, Kendal, replaced the handbook for Pilgrims) followed by IVth Congress

XIV: 1986. Austria: Noricum/Pannonia (superior) H. VETTERS & M. KANDLER (edd.) 1990: Akten des 14. Internationalen Limeskongresses 1986 in Carnuntum. (Vienna). D. H. FRENCH & C.S. LIGHTFOOT (edd.) 1989: The eastern frontier of the Roman empire. (Colloquium at Ankara 1988). BAR Int. Ser. 553. (Oxford). 1989 11th Pilgrimage, followed by XVth Congress C.R. WHITTAKER 1989: Les frontières de l’Empire romain. (Paris).

IV: 1959. England: Britannia (inferior) E. BIRLEY, organiser—not published A. TOČÍK (ed.) 1959: Limes Romanus Konferenz Nitra, Bratislava (held at Nitra, Czechoslovakia, 1957) V: 1961. Yugoslavia: Dalmatia, Pannonia (inferior), Moesia (superior) G. NOVAK (ed.) 1963: Quintus Congressus Internationalis Limitis Romani Studiosorum. (Zagreb).

9

Limes XVIII

XV: 1989. England: Britannia (superior)/Maxima Caesariensis V.A. MAXFIELD & M.J. DOBSON (eds.) 1991: Roman Frontier Studies 1989. (Exeter). B. ISAAC 1990: The limits of empire. (Oxford, repr. with addenda 1992) Y. ROMAN (ed.) 1991: La Frontière, Paris (Séminaire de recherche, Lyon 1988-1991) P. BRUN, S. VAN DER LEEUW & C.R. WHITTAKER (edd.) 1993: Frontières d’empire. Nature et signification des frontières romaines, Nemours (Table ronde, Nemours 1992) A.D. LEE 1993: Information and frontiers. Roman foreign relations in late antiquity (Cambridge). E. DABROWA (ed.) 1994: The Roman and Byzantine army in the east, Kraków (Colloquium at Kraków 1992). C.R.WHITTAKER 1994: Frontiers of the Roman empire. A social and economic study. (Baltimore; rev. version of Whittaker 1989). XVI: 1995. Netherlands: Germania (inferior) /Belgica W. GROENMAN-VAN WAATERINGHE, B.L.VAN BEEK, W.J.H. WILLEMS & S.L. WYNIA (edd.) 1997: Roman Frontier Studies 1995. (Oxford). D. KENNEDY (ed.) 1996: The Roman Army in the east. Journal of Roman Archaeology Supplementary Series 27. XVII: 1997. Rumania: Dacia N. GUDEA (ed.) 1999: Roman Frontier Studies XVII/1997. (Zalău). 1999 12th Pilgrimage—NOT followed by Congress S. P. MATTERN 1999: Rome and the enemy: Imperial strategy in the Principate. (Berkeley). XVIII: 2000. Jordan: Arabia

10

A.R.Birley: Fifty years of Roman frontier studies

Fig. 1. VIIth Congress, Israel 1967

Fig. 2. Plenary session, 1983 Congress, Aalen, Germany

11

Limes XVIII

12

The onset of crisis: the reign of the emperor Severus Alexander (AD 222- 235) Lukas de Blois The main causes of the so-called ‘Third Century Crisis’ in the Roman empire were: increased dependence on the armies in a period of heavy actual or impending warfare combined with a lack of means to keep up with ensuing military demands and with a tenacious endeavour to stick to traditional systems and policies, even if they were no longer suitable. Severus Alexander and his advisers tried to maintain traditional monetary standards and to pay their soldiers in good gold and silver coins, with proper weight and fineness, so that they would not easily follow other pretenders to the throne. Unfortunately, in Severus Alexander’s time the Roman administration did not have at its disposal enough means to mint sufficient quantities of good gold coins. P.Fayum 20, passages in Herodian (Bk.VI) and other sources show that the administration of this emperor had come into financial straits. So, in spite of conservative monetary attitudes, the imperial mints had to widen the range of weight and fineness in the denomination of the aurei. The emperor probably gave the good coins to the military, especially to middle cadre officers, who must have been paid in gold, simultaneously attributing debased aurei to other receivers. Cassius Dio (LXXVII.77.14.3-4) attributes this monetary policy to Caracalla, but coins reveal that this device was used by Severus Alexander. A hazardous monetary policy indeed. In the end aurei no longer counted as coins representing a standard value, but as individual ornaments which people pierced and carried around their necks.

The question I want to answer in this paper is, how did this emperor and his advisers react to the rising problems of his times?

many means devised by Severus and his son for the undermining of military discipline, it was impossible, on the one hand, to give the troops their full pay in addition to the donatives that they were receiving, and impossible, on the other hand, not to give it” (Dio LXXVIII.36).

In a mutilated papyrus text which contains a copy of an imperial letter which Oliver convincingly ascribes to the emperor Severus Alexander, we read, that the emperor, who has sent this letter, will not compel cities to contribute more golden crowns or sums due in place of them, than they are able to give. He would have liked to offer, he says, a more conspicuous proof of his magnanimity and to remit arrears, but the poverty of the government prevents him to do so. He has, however, not failed to observe, that the amounts he has received are all that the cities can afford to pay. So he remits the last contribution, due on the occasion of his accession to the imperial throne.1 Is this the language of an emperor displaying his power? It sounds more like an apology of a hard-pressed ruler in times of need. Apparently this emperor abolished a fiscal trick that Cassius Dio ascribes to the emperor Caracalla, who reigned from 211 until 217. In Book LXXVII.9, he writes that this emperor was fond of spending money upon the soldiers and made it his business to strip, despoil and grind down all the rest of mankind. According to Dio, one of Caracalla’s devices was to demand gold crowns from the communities in the empire, on the constant pretext that he had conquered some enemy or other.

A third text. In the oration which he puts into the mouth of Maecenas Dio discusses fiscal schemes which might produce more money to pay the soldiers, in his view a prerequisite condition of survival of the empire. He says: “We cannot survive without soldiers and men will not serve as soldiers without pay” (LII.28.1). In my view in these three texts we find the main cause of the so-called C3rd crisis in the Roman empire: increased dependence on the armies in a period of heavy actual or impending warfare combined with a lack of means to keep up with ensuing military demands and with a tenacious endeavour to stick to traditional systems and policies, even if they were no longer suitable.2 Many problems were already there during the reign of the emperor Marcus Aurelius. War broke out in the east and new strong enemies appeared at the northern frontiers. The Roman empire had to contend with severe wars within its own borders, from Pannonia and Noricum to northern Italy, and it had to cope with the inevitable consequences, like foraging, plunder, devastation, famine, and plague, which caused people to leave their homesteads and join roving bands of robbers, deserters and barbarians who had stayed behind. In 185, five years after the last major campaigns against the Marcomanni, Quadi and other tribes, war - stricken areas and adjoining regions were infested by wandering bands of deserters and bandits. Their leader, Maternus, even besieged the legio VIII Augusta in its camp at Strasbourg and then invaded and

Another text. In his Roman History Dio mentions a letter written by the emperor Macrinus, who reigned 217-218. Dio tells us: “He (ie. Macrinus) wrote a letter also to Maximus, the prefect of the city of Rome, in which, after mentioning various matters of a routine nature, he stated that even the newly-enlisted soldiers insisted on receiving everything that the others had been getting, and that these other soldiers, who had not been deprived of anything, made common cause with the new recruits in their anger at what was being withheld from them. And he said, of all the

2

On the crisis in the Roman empire in the C3rd AD, see Alföldy 1989b; Strobel 1993; Christol 1997; de Blois 1998: 3394-3404; Cosme 1998: 29174; Witschel 1999). On the preceding decades, see Birley 1987; 1988

1

Pap. Fayum 20 = A.S. Hunt & C.C. Edgar Select Papyri II. (London 1956): No. 216.

13

Limes XVIII plundered Italy.3 Unavoidable depopulation in border areas meant less production, smaller food surpluses, less tax income and a weakened logistical basis in the hinterland of important armies. The plague, which raged all over the empire from 166, must have diminished the population in many parts of the empire. The decline in population must have resulted in a more uneven distribution of people: more soldiers, fewer farmers. Marcus Aurelius had to replenish depleted legions and auxilia and he created two new legiones Italicae. In his monograph on this emperor Anthony Birley describes his difficulties in finding recruits: Marcus had to enlist all kinds of people, even slaves.4 That is another sign of a lack of good men who could be missed in agrarian and other economic activities.

to one aureus (Harl 1996: 133; Bland 1996: 68). One may wonder, however, if the upper layers of the public at large and the military, the main receivers of gold and silver coins, were aware of actual intrinsic value relations between gold and silver coins. People who tested and exchanged coins certainly were, and they may have passed on what they had found out. According to Roger Bland Caracalla introduced a new radiate gold multiple, which weighed a little over 13gms and must therefore have been a double aureus (Bland 1996: 68). Unfortunately there is no evidence which could indicate the receivers of those multipla. I presume that they found their way to higher and middle cadre officers, whose loyalty the emperor could not afford to lose.

The situation slowly improved after 180, but recurring warfare and civil strife kept causing damage and put imperial finances and logistics under a heavy strain. Besides Septimius Severus created three more legions, the legiones Parthicae, he raised the soldiers’ pay from 300 to 450 denarii per annum and he improved their fringe benefits. The soldiers were also paid extra allowances in cash and kind. Those were “the many means devised by Severus and his son for the undermining of military discipline”, which Dio has Macrinus mention in the letter I just mentioned (cf. Cosme 1998: 75ff). It may be that Septimius Severus only compensated the military for a slow and gradual rise in prices which had been going on during many decades, at least from the days of the emperor Trajan (Duncan Jones 1994: 29, 32). Those measures cost Severus lots of money, so he was compelled to use many devices to enlarge his income and to tamper with the silver denarius.5

After a single issue of antoniniani or radiates at the very start of his reign Macrinus quickly discontinued their production and several months later he restored the weight of the aureus to its pre-215 standard of 7.2gms (Bland 1996: 69, 94, fig. 8). He appears to have increased the fineness of the silver coinage, although analyses have produced a wide range of results ranging from 29% to 90% silver (cf. Walker 1978: nrs. 4044-4074). Did Macrinus buy the loyalty of his military cadre by giving them good coins, leaving the debased currency to other people? Again, there is no evidence to prove this. The traditional view is that Macrinus tried to revert to a more traditional monetary system, without having the means to do so. Those are the backgrounds of the letter which Cassius Dio put into his mouth. His successor Elagabalus restored Caracalla’s reforms, debasing the coinage even further, although from 219 he did no longer produce any radiates or antoniniani (Bland 1996: 69).

The son mentioned by Dio was the emperor Caracalla, who reigned from 211 until 217. He bought off the Alamans in order to concentrate his forces in the east, where he lost his life during a rather unsuccessful campaign against the Parthians. If we may interpret remarks and invectives of Dio and Herodian in a more neutral way he continued his father’s policy of making military service more attractive by granting large donativa, thus depleting his treasuries even further, which forced him to find new ways to raise his income. In the passage I just mentioned Dio tells us that Caracalla demanded many gold crowns from the communities in the empire. He also created a new silver coin, the antoninanius, nominally a double denarius which actually contained only 1.6 times the quantity of precious metal, and lowered the weight of the aureus to 6.59, a 9.3% reduction. The traditional view is, that he did so in order to preserve the official rate of exchange: 25 denarii

The succeeding reign of Severus Alexander was characterised by a serious endeavour to maintain traditional Antonine policies in appointments, rescripts, the administration of justice, the presentation of imperial power and monetary policies. This emperor did not produce any radiates or antoniniani and tried to maintain the average weight and fineness of gold and silver coins. Like Trajan, Hadrian and the Antonines he produced good copper money (sesterces and asses). The Historia Augusta depicts his reign as the Indian summer of senatorial government and as a benign interlude between periods of military tyranny. This Vita, however, is not a reliable source. It belongs to the literary genre of mirrors-of-princes more than to serious historiography. Herodian holds that Severus Alexander, giving the senators more latitude and taking their advice seriously, returned to aristokratia, after a period of tyranny. He also mentions Ulpian, a jurist who had published learned works, as one of the most powerful men at the start of the reign. He was a praefectus annonae under Elagabalus and became a praetorian prefect in 222. The influence of the so-called council of senators has been heavily over-rated, though, and jurists like Ulpian were not omnipotent, if only because the military did not accept their dominant position.

3

On Maternus and his army of bandits, see Herodian I.10; Scriptores Historiae Augustae (= SHA), Commodus 16.2; Niger 3.4; AE 1956: 90. See Grosso 1964: 437ff.; Whittaker 1969: 62f. n.1; Alföldy 1989a. 4 On the recruitment of new legions by Marcus Aurelius and the ensuing increase in expenditure, see Birley 1977: 291ff. 5 On the debasement of the denarius and other coins in the period of the Severi (193-235), see Bland 1996: 63, 67-70; Harl 1996: 126-136. In general on monetary problems and policies: Carson 1990; Duncan-Jones 1994; Lo Cascio 1984: 133-201.

14

Lukas de Blois: The onset of crisis: the reign of the emperor Severus Alexander (AD 222- 235)

In 224 Ulpian was murdered by discontented soldiers.6 In practical politics the difference between Severus Alexander’s reign and the reigns of his Severan predecessors was not that big. On the other hand, some tangible differences are perceptible.

already do. For the military, on the other hand, times were not that bad. The most dangerous wars were still to come and in the first decades of the C3rd they reached, as Michael Speidel has convincingly demonstrated, a situation of relative prosperity (Speidel 2000: 92f). The balance of power between local notables and the military had shifted in favour of the latter, and material welfare had followed suit. Furthermore, primipilares, centurions and lower cadre became ever more important in provincial administration (cf. Dobson 1974: 432; 2000: 152; Breeze 1974: 443ff.) and could collect concomitant material bonuses. Petitioners like those of Skaptopare and Aragoe now used military men as intermediaries who should bring their complaints to emperors and governors, not rhetors or local magnates.10 Herodian tells us that people in the empire complained about the greed of the emperor Maximinus Thrax to relatives who served in the armies, he does not mention any local notables or intellectuals in this context (Herodian VII.3.6). In my view military cadre had a chance to become affluent and powerful, sometimes to the detriment of local gentry. The new emperor, Severus Alexander, could not afford to lose their loyalty, but he probably was not really on their side. His connections with senators and jurists were good. Most senators and equestrian jurists, however, did not have military careers. Status set senators fulfilled curatorships in Italian and provincial cities and were appointed to offices like iuridicus and other administrative posts, and accepted only one or two short stays in the armies. Learned jurists gave legal advice, had a private practice, served in the imperial council and were appointed to procuratorships, and prefectures, predominantly in Rome. So status set senators and jurists did not meet many important military men or lower and middle cadre. They did meet, however, local notables and urban aristocrats, who had large domains in Italy and the provinces (de Blois 2001:148). They met, in other words, the victims of military highhandedness, not the perpetrators. This must have coloured the advice they gave to Severus Alexander. Their attitude towards the military is mirrored in Dio’s work, whose main audience they were. In all his 80 books Dio is almost obsessed with the risks of military high-handedness. He regarded armies as the main source of power and he branded soldiers as a naturally rebellious species that is to be classed with robbers and that is difficult to control. And yet controlled it must be, for otherwise a legitimate monarchy or a republican government will degenerate into a military tyranny and the whole social structure will be turned upside down. In his view there is a great risk of this taking place when the government loses its dignity, when the community is weakened by discord or when a despotic or uneducated ruler regards the armies as his personal retinue and spoils them, at the expense of the rest of the population.11

If the papyrus text which I mentioned at the start of this paper contains a copy of a letter by Severus Alexander, which I believe it does, this emperor tried to alleviate the burdens of local communities in the empire. This was really not an unnecessary action. A problem that runs through all the sources of the times is the exceedingly heavy burden of taxation. In a recurring commonplace bad emperors are accused of feasting on money wrung from the poor and of robbing the rich to satisfy the soldiers.7 Complaints of a lack of means, caused by military foraging and plunder, recur with distressing regularity in petitions and literary sources.8 Actual heavy warfare, shifting from one border region to another, resulted in frequent transfers of military units, which must have caused lots of damage. Soldiers travelling over the military highways foraged, looted and demanded transport facilities. Petitions, like those from Skaptopare in the Balkans and from Takina and Aragoe in Asia Minor, reveal the misery caused by military misbehaviour. Louis Robert mentions a text in which people complain that normal everyday affairs, such as using the public baths, could no longer take place due to a lack of means for the upkeep. The soldiers had taken everything.9 The problem may have been more widespread than available evidence indicates. Plundering armies and groups of soldiers were also passing through northern border regions which presumably produced fewer petitions than the Greek-speaking regions in the Balkans and Asia Minor. So we do not see much of their complaints. We should not forget, that local notables had come into financial straits at the end of the C2nd already and that financial pressure continued to harass them in the following decades. In the papyrus text with which I started this paper the emperor who is speaking to us admits that the communities of the empire cannot pay more than they 6 On the reign of Severus Alexander, see Herodian VI and SHA Alex. Sev.. See also Aurelius Victor Caes. 24.1-7; Eutropius Brev. VIII.23; Zosimus I.11-12. On Ulpian’s position of power see Dio LXXX.2 (on his death: LXXX.2.2); Eutropius Brev. VIII.23. The C4th sources describe Severus Alexander’s severity towards the soldiers (probably wishful thinking), the influence of Ulpian and Mamaea, and the glorious war against the Persians (also wishful thinking). Like Herodian Zosimus is more nuanced. They criticise Severus Alexander’s dependence on his mother and his excessive thrift. On Ulpian’s career and death see Honoré 1982: 6-41. On the great jurists and juridically skilled bureaucrats of Severan times, esp. on Papinian, Paul, and Ulpian, see Syme 1979: 790-804; 1980: 97ff; 1991: 216f. On Severus Alexander’s monetary policy, see Harl 1996: 128 7 see Dio LII.28-29; LXXII.3.3f.; LXXIII.16.2f.; LXXV.8.4f.; LXXVII.9 and 13-16; LXXVIII.9-14; Herodian III.8f.; VI.1.8f.; VII.3.1ff.; Philostratus Vita Apollonii V.36; Ps.-Aelius Aristides Eis basilea 16 and 30 ff.; P. Lond. inv. 2565; Pap. Fayum 20 (cited above); see Cosme 1998: 86-95. 8 Petitions and complaints: Robert 1943: 115ff.; 1989: 596ff.; Mitchell 1976: 114f.; Herrmann 1990: nos. 4, 6, 8; Millar 1977/1992: 646 (= SEG 37 (1987): 1186). 9 see Robert 1943: 115ff. On the petitions from Skaptopare and Aragoe see CIL suppl. 12336 = IGR I 674 = Sylloge 888 (from Skaptopare); OGIS 519 = IGR IV 598 (from Aragoe).

10 see CIL suppl. 12336 = IGR I 674 = Sylloge 888, l. 7 (from Skaptopare); OGIS 519 = IGR IV 598, l. 3 (from Aragoe). 11 On Dio’s view of the soldiery, see de Blois 1997: 2650-2676; 19981999; 275ff.

15

Limes XVIII

So it does not come as a surprise that Severus Alexander and his advisers tried to avoid wars that would deliver him up to the whims of the armies and would deplete his coffers. Severus Alexander, Mamaea and their advisers favoured a policy of peace and appeasement, rather buying off wars than fighting the enemies.12 After having chosen an aristocratic type of ruling instead of strong military government, having alleviated the burdens of local communities and having chosen a rather conservative monetary policy (cf. Harl 1996: 128), Severus Alexander could do nothing else. In this way he saved money and did not become too dependent on the military cadre. A secondary effect may have been that he did not have to pay too many donatives. Alexander’s policy greatly annoyed some of the military. According to Herodian soldiers called Severus Alexander names like “sissy emperor” and accused him and his dominant mother Mamaea of meanness (Herodian VI.1.8; VI.2.3ff.; VI.4.4; VI.7.9). Well, in the papyrus text with which I opened this lecture, probably a copy of a letter of Severus Alexander, thrift and a lack of means are ostentatiously brought forward. The emperor who is speaking to us in this text even seems to be proud of his careful financial policies and propagates them as a means to restore the Roman world.

higher purity for distribution to the German tribes, although many Roman aurei certainly found their way to Free Germany, just as there is no evidence that he struck gold coins that were debased with copper. In fact this sounds remarkably like a literary topos” (Bland 1996: 75). This may be perfectly true. Dio considered Caracalla a typical tyrant, to whom he would attribute any kind of tyrannical behaviour. But one question remains to be solved: why did Dio not speak of good and bad denominations of money, but of good and bad coins? This must be an anachronistic way of speaking about earlier monetary policies. Dio knew Severus Alexander quite well, he was his coeval and he served him in several difficult jobs. In 229 he was consul iterum ordinarius with the emperor.14 So Dio must have been acquainted with all kinds of policies of Severus Alexander, his monetary tricks included. In other words, Dio knew about good and bad aurei, and about the widening of the range within this denomination, but he applied this knowledge in an anachronistic way to Caracalla’s gilded diplomacy in Germany. This was a hazardous monetary policy indeed. In the middle of the century already aurei did no longer count as coins representing a standard value, but as individual ornaments which people pierced and carried around their necks.

Nonetheless the administration of Severus Alexander came into financial straits and had to find ways and means to make ends meet with a too small amount of plate. They did not lower the average weight and fineness of gold, silver and copper coins. As Roger Bland has demonstrated, however, they greatly widened the range of weights of individual aurei.13 Like the emperor Gallienus in the period after the calamitous year 260 they may have given good gold coins to the military, particularly to higher and middle cadre officers who could influence the troops in a decisive way and had become important in provincial administration, simultaneously attributing debased aurei to other receivers (de Blois 1976: 95ff.). Doing so they may have followed examples given by Caracalla and Macrinus (see above). Dio may have been aware of this monetary policy, as he shows in a passage on Caracalla’s policies in ‘Free Germany’. In LXXVII.14.3-4 he says, in a passage on the gold with which Caracalla bought off German violence: “The gold that he gave them was of course genuine, whereas the silver and the gold currency that he furnished to the Romans was debased, for he manufactured the one kind out of lead plated with silver and the other out of copper plated with gold”. The variation of this text given by the other summary, that of Xiphilinus, is shorter and less specific: “With Antoninus the coinage as well as everything else was debased, both the silver and the gold that he furnished us” (333.18). Roger Bland comments: “There is no evidence that Caracalla produced coins of a

Severus Alexander’s policy of avoiding war came to an end in 230. Great wars broke out, in 230 in the east, and in 233 on the Rhine. During those last years the administration of Severus Alexander struck heavier gold and silver coins (Bland 1996: 69). Probably the emperor had to do so, in order to buy his soldiers’ loyalty now that he had become more dependent on them. Bibliography Alföldy G. 1989a Bellum desertorum. In G. Alföldy Geschichte, Geschichtsschreibung und Geschichtsbetrachtung. Ausgewählte Beiträge. (Stuttgart): 69-80 (= Bonner Jachbucher 171 (1971): 367-376). Alföldy G. 1989b Die Krise des Römischen Reiches. Geschichte, Geschichtsschreibung und Geschichtsbetrachtung. Ausgewählte Beiträge. (Stuttgart): 319-342 (= The crisis of the third century as seen by contemporaries. Greek, Roman & Byzantine Studies 15 (1974): 89-111). Birley A.R. 1977 Mark Aurel. (Munich). Birley A.R. 1987 Marcus Aurelius. A Biography (London, 2nd ed.). Birley A.R. 1988 The African emperor, Septimius Severus. (London; 2nd ed.). Bland R.F. 1996 The development of gold and silver coin denominations, A.D 193-253. In C.E. King & D.G. Wigg (edd.) Coin finds and coin use in the Roman

12 On the excessive thrift and cautious diplomacy of Severus Alexander, Mamaea, and their advisers see Herodian VI.1.8; VI.2.3ff.; VI.4.4; VI.7.9. cf. Pap. Fayum 20. 13 On the widening of the range of weight and fineness within the denomination of aurei under Severus Alexander see Bland 1996: 69, 96, fig. 10. The traditional view is, that this development set in later on, but Bland has convincingly argued that it took off under Severus Alexander.

14 On Dio’s life and career, cf. Millar 1964: 7-27; Reinhold 1988: 1-4; Leunissen 1989: 163 n. 147; Hose 1994: 356ff.

16

Lukas de Blois: The onset of crisis: the reign of the emperor Severus Alexander (AD 222- 235)

world. The Thirteenth Oxford symposium on coinage and monetary history, 25-27.3.1993 (Berlin): 63-100. de Blois L. 1976 The policy of the emperor Gallienus. (Leiden). de Blois L. 1997 Volk und Soldaten bei Cassius Dio. In W. Haase & H. Temporini (edd.) Aufstieg und Niedergang der Römischen Welt II.34.3. (Berlin): 2650-2676. de Blois L. 1998 Emperor and empire in the works of Greekspeaking authors of the third century A.D. In W. Haase & H. Temporini (edd.) Aufstieg und Niedergang der Römische Welt II.34.4. (Berlin): 3391-3443. de Blois L. 1998-1999 The perception of emperor and empire in Cassius Dio’s Roman History. Ancient Society 29: 267-281. de Blois L. 2001 Roman jurists and the crisis of the third century A.D. in the Roman empire. In L. de Blois (ed.) Administration, prosopography and appointment policies in the Roman empire. (Amsterdam): 136-153. Breeze D.J. 1974 The career structure below the centurionate during the Principate. In W. Haase & H. Temporini (edd.) Aufsteig und Niedergang der Römischen Welt II.1. (Berlin): 435-451. Carson R.A.G. 1990 Coins of the Roman empire. (London ). Christol M. 1997 L’empire romain au IIIe. siele. (Paris). Cosme P. 1998 L’État romain entre éclatement et continuité. L’empire romain de 192 - 325 (Paris). Dobson B. 1974 The significance of the centurion and primipilaris in the Roman army and administration. In W. Haase & H. Temporini (edd.) Aufstieg und Niedergang der Römischen Welt II.1. (Berlin): 392434. Dobson B. 2000 The primipilares in army and society. In G.Alföldy, B. Dobson & W. Eck (edd.) Kaiser, Heer und Gesellschaft in der römischen Kaiserzeit. (Stuttgart): 139-152. Duncan-Jones R.P. 1994 Money and government in the Roman empire. (Cambridge). Grosso F. 1964 La lotta politica al tempo di Commodo. (Turin). Harl K.W. 1996 Coinage in the Roman economy, 300 BC to AD 700. (Baltimore). Herrmann P. 1990 Hilferufe aus den römischen Provinzen. Ein Aspekt der Krise des römischen Reiches im 3. Jh. n.Chr. (Hamburg). Hose M. 1994 Erneuerung der Vergangenheit. Die Historiker im Imperium Romanum von Florus bis Cassius Dio. (Stuttgart). Leunissen P.M.M 1989 Konsuln und Konsulare in der Zeit von Commodus bis Severus Alexander (180-235 n.Chr.). (Amsterdam). Lo Cascio E. 1984 Dall’ antoninianus al ‘laureato grande’: l’evoluzione monetaria del III secolo alla luce della nuova documentazione di età diocleziana. Opus 3: 133201. Millar F. 1964 A study of Cassius Dio. (Oxford). Millar F. 1977 The emperor in the Roman world. (2nd ed. 1992, London). Mitchell S. 1976 Requisitioned transport in the Roman empire. A new inscription from Pisidia. Journal of Roman Studies 66: 106-131. Reinhold M. 1988 From Republic to Principate. An historical commentary on Cassius Dio’s Roman History, Books 49-52, 36-29 BC. (Atlanta). Robert L. 1943 Sur un papyrus de Bruxelles. Revue de Philologie 17: 115ff. Robert L. 1989 Opera Minora Selecta. (Amsterdam).

Speidel M.A. 2000 Sold und Wirtschaftslage der römischen Soldaten. In G.Alföldy, B. Dobson & W. Eck (edd.) Kaiser, Heer und Gesellschaft in der römischen Kaiserzeit. (Stuttgart): 65-94. Strobel K. 1993 Das Imperium Romanum im 3. Jahrhundert. (Stuttgart). Syme R. 1979 Three jurists. In Roman Papers II. (Oxford): 790804. Syme R. 1980 Fiction about Roman jurists. Zeitschrift der Savigny-Stiftung für Rechtsgeschichte 101, Romanistische Abteilung 78: 97ff. Syme R. 1991 The jurists approved by Antoninus Pius. In K. Rosen (ed.) Antiquitas. Reihe 4. Beiträge zur HistoriaAugusta-Forschung. (Bonn): 201-217. Walker D.R. 1978 The metrology of the Roman silver coinage III. From Pertinax to Uranius Antoninus. British Archaeological Reports Suppl. Ser. 40: (Oxford). Whittaker C.R. 1969 Herodian I. (London). Witschel Chr. 1999 Krise-Rezession-Stagnation? Der Westen des Römischen Reiches im 3. Jahrhundert n.Chr. (Frankfurt a.M. ).

17

Limes XVIII

18

Bezüglich der Europäischen Wanderschaft der legio III Gallica zu Beginn des 3. Jh. Nicolae Gudea I: In den Jahren 1939-1940 wurde in Porolissum während der archäologischen Grabungen ein Ziegel mit dem Stempel der l(egio) III G(allica) gefunden ? Die Überraschung war besonders groß und die Lesung des Stempels nicht die beste (Daicoviciu 1951: 257).

beziehen, ist wenig bekannt (Ritterling 1925: Sp. 15251527). Es wurden zwei Hypothesen ausgearbeitet: (a) man glaubt, daß die Einheit vom Kaiser Septimius Severus wegen ihrer Beteiligung am Bürgerkrieg (193-194) auf der Seite des Pescennius Niger bestraft wurde; Einzelheiten fehlen (Ritterling 1925: Sp. 1525); (b) es wird angenommen, daß sie erst unter Elagabalus aufgelöst und der damnatio memoriae unterworfen wurde, infolge eines Aufruhrversuches ihres Legaten (Ritterling 1925: Sp. 1526; Dabrowa 1989: 278, vgl. Dio LXXIX.7.1). Weitere Verfasser, die sich ebenfalls mit der Geschichte und Struktur der römischen Armee befaßt haben: G.R. Watson (1969: 118-121), L. Keppie (1984: 214), Y. Le Bohec (1989: 62-63), B. Campbell (1994: 104-109) erwähnen keine weiteren Einzelheiten bezüglich der Strafmaßnahmen.

Im Jahre 1943, während weiterer archäologischer Grabungen, traten die Stempel der Legion in großer Zahl (36) in der Zerstörungsschicht im Kommandaturgebäude auf, zusammen mit den Stempeln der legio VII Gemina Felix und einer cohors III. E. Tóth bot eine bessere Lesung des Textes und schlug auch eine Datierung der Anwesenheit der Legionsabteilung vor, u. zwar während der Bau- und Umbautätigkeiten zu Beginn des 3. Jh. (Tóth 1978: 45-46). In der Zeit zwischen 1943 bis zum Beginn unserer Grabungen im Kastell (1977) war die Deutung des Vorkommens dieser Stempel unterschiedlich. Überraschend wußte J. Szilagyi nichts über diese Stempel (Szilagyi 1946: 14, Amm. 64). M. Macrea (1969: 194) akzeptierte die Lesung legio III Gallorum von C. Daicoviciu und deutete das Vorkommen der Stempel nicht richtig. Im Jahre 1978, als ich die zufälligen Funde von Porolissum veröffentlichte (Gudea 1978: 66), stellte ich auch zwei Stempel der Legion vor und gelangte zu denselben Schlußfolgerungen wie E. Tóth (dessen Arbeit erst 1979 zu uns gelangte).

E. Ritterling (1925: Sp. 1528) glaubte, daß der neue Sitz der Legion nach ihrer Neugründung unter Alexander Severus Damaskus war. E. Dabrowa (1996: 278) behauptet, daß nach diesem Zeitpunkt ihr Sitz nach Osten, nach Danaba verlegt wurde. III: Bezüglich der Art und Weise der Bestrafung einer Militäreinheit durch damnatio memoriae sind die Dinge nicht sehr klar. H.M.D. Parker (1958: 232-237) erwähnt eine solche Strafe nicht. Von E. Ritterling (1925: Sp. 15261527) erfahren wir, daß ihr Name von den Denkmälern entfernt wurde; sie konnte nicht mehr als eine Einheit auftreten; sie wurde in Abteilungen eingeteilt (deren Größe unbekannt war), die in verschiedene Teile des Reiches verlegt wurden; der Sold der Militärs wurde stark verkleinert (sie verloren die Gewinne außer dem Sold und die Nachdienstrechte); die Soldaten wurden nur zu Arbeiten (Straßenbau usw.) verwendet. Die legio III Gallica wird zu wiederholten Malen unter den bestraften Militäreinheiten erwähnt, jedoch nicht für die Zeit des Septimius Severus.

Zwischen 1977-1980 stieg die Anzahl der Funde während der archäologischen Grabungen im Kastell auf dem Gipfel Pomet auf 28. Inzwischen hatte auch die Veröffentlichung aller alten Angaben (Gudea 1989: 518-519, Taf. CXVI. 13) die archäologische Welt von der Anwesenheit der Einheit in Porolissum überzeugt. Jedwelche Zweifel bezüglich der Lesung der Stempel verschwanden; es war bereits offensichtlich, daß eine Abteilung dieser Legion zu Beginn des 3. Jh. n. Chr. in Porolissum stand (Gudea 1978: 69; 1989: 59); die Stempel der legio III Gallica wurden stets zusammen mit jenen der legio VII Gemina Felix und der cohors III (Dacorum) gefunden. E.L. Wheeler, der noch 1996 lediglich von der Existenz eines einzigen Ziegelstempels der cohors III Gallica in Porolissum wußte (Wheeler 1996: 234, Anm. 20), schrieb dies der Anwesenheit einer Einheit dieser Legion zu. Er räumte aber ein, daß es sich um einen Lesungsfehler handeln könnte: “…more probably the stamps reflect an error for XIII Gemina, the Dacian legion of a long standing”. In dem von Wheeler zitierten Werk (AE 1979: 501a) werden aber, wie ich feststellen konnte, 35 Ziegelstempel mit diesem Text angeführt, so daß eine falsche Lesung ausgeschlossen werden kann.

J. Hasebroek schrieb: “Von den Truppen Nigers, die sich nach den Kämpfen der Jahre 193/194 retten können und über den Tigris zu den Barbaren geflohen waren, bringt der Kaiser durch Zusicherung einer Amnestie die Mehrzahl auf seine zurück” (Hasebroek 1921: 73 vgl. Herodian III.4.7). Entsprechend schrieb A. Birley; “No steps were taken to degrade or punish the army that fought for Niger” (Birley 1971: 181). Er erwähnt jedoch einen Ex-centurio, einen Mitglied der speculatores (der Geheimpolizei), der zum primus pilus der legio III Gallica ernannt wurde, um Disziplinärmaßnahmen zu treffen.

II: Die Geschichte der Legion in der Zeit, auf die wir uns

E. Dabrowa (1996: 280-283) glaubt immer noch, daß als

IV: Was über die Legio III Gallica zu beginn. des 3. Jh. bekannt ist, ist sehr wenig:

19

Limes XVIII

Legionslegaten weiterhin die Statthalter der Provinz Syrien fungierten. Leider erwähnen die benutzten Inschriften diese Beziehung nicht und in den von den Statthaltern angeführten Tätigkeiten (Straßenbauten und –reparaturen, Bau von Kastellen usw.) wird der Name der Legion nicht erwähnt. In dieser mit vielen Unsicherheiten “aufgebauten” Reihe von Statthaltern gibt es zwei Unterbrechungen, die auf die Maßnahmen gegen die Legion zurückgeführt werden könnten: von 207/209 bis 213 bzw. von 217/218 bis 226.

193-194 bekannt. Dabei handelte es sich um die Legionen: I Minervia und XXX Ulpia Victrix (Untergermanien), VIII Augusta und XXII Primigenia (Obergermanien), III Italica (Raetien), II Italica (Noricum), XIII Martia Victrix und I Adiutrix (Oberpannonien), II Adiutrix (Unterpannonien), IIII Flavia und VII Claudia (Obermoesien), XIII Gemina (Dacia Apulensis), V Macedonica (Dacia Porolissensis), I Italica und XI Claudia (Untermoesien). Aus der Münzprägung des Septimius Severus fehlen die Legionen des Ostens: III Gallica, IV Scythica und XVI Flavia (Syrien), VII Ferrata und X Fretensis (Iudaea), III Cyrenaica (Arabien), II Traiana (Ägypten), XX Fulminata und XV Apollinaris (Kappadokien).

R. Saxer (1967: 313) glaubte, daß eine Legionsabteilung unter Septimius Severus in die afrikanische Provinz Numidien geschickt wurde. Diese Ansicht könnte mit einer älteren von E. Ritterling (1925: Sp. 1527) übereinstimmen, der behauptete, daß “zahlreiche Mannschaften der III Gallica in die legio III Augusta versetzt wurden.” Drei Inschriften aus der Regierungszeit des Septimius Severus erwähnen eine (oder einige) Abteilungen der legio III Gallica, die zusammen mit Vexillationen der legio III Augusta und einer ala Pannaniorum ein Kastell in Dimmidi in Numidien bauten (Picard 1944: 188-190, Nr. 12-14).

Es ist aber bekannt, daß sich der Statthalter von Ägypten, L. Mantennius Sabinus (Ende 193), die Legionslegaten von VI Ferrata und X Fretensis (nach der Schlacht von Kyzikos am 1. Januar 194), der Statthalter Arabiens und der Legat der III Cyrenaica (gleichzeitig mit den oben erwähnten) von Niger losgesagt haben.

Eine Inschrift von Viminacium (Obermoesien) suggeriert die Anwesenheit einer Legionsabteilung in der Nähe des Lagers der legio VII Claudia (Vulic 1905: Bbl. 6, Nr. 14). Wahrscheinlich infolge der Neugründung der Einheit übernimmt diese Abteilung den Beinamen Severiana Alexandriana.

Nicht erwähnt werden die Legionen: X Gemina (Oberpannonien) und III Augusta (Numidien). Es ist anzunehmen, daß die beiden nicht sofort dem Severus ihre Anerkennung leisteten. Daß sie sich letztlich doch für Severus entschieden, gilt als sicher, da sie auch später nachweislich in ihren Hauptlagern blieben. Die legio X Gemina erhielt später sogar den Ehrentitel pia Severiana (AE 1913: 56).

Eine weitere in Sirmium (Unterpannonien) gefundene Inschrift zeigt das Vorkommen eines Soldaten der legio III Gallica an, der, dem Namen nach zu urteilen, aus den kleinasiatischen oder syrischen Provinzen stammte. Die Inschrift wurde aufgrund des Schrifttyps an das Ende des 2. oder den Anfang des 3. Jh. datiert ! M. Mirkovic schlug mehrere Deutungen vor, darunter auch “des soldates colonisés dans les pays au bord du Danube ou des détachements qui y étaient installés pour un temps au cours des événements sur les frontières” (Mirkovic 1990: 635636).

Es fehlen auch (aus den Münzemissionen) die drei britannischen Legionen: II Augusta, VI Victrix und XX Valeria Victrix, wie auch die hispanische legio VII Gemina. Die Spannungen zwischen Severus und dem Statthalter Britanniens, D. Clodius Albinus und dessen Anhänger, dem legatus Augusti pro praetore der Provinz Hispania Tarraconensis, L. Novius Rufus (der Geld, Nahrungsmittel und Soldaten für Albinus beschaffte), werden durch das Fehlen der unter ihrem Kommando stehenden Legionen auf den Münzen des Severus verdeutlicht (vgl. SHA Severus XIII.7).

Die über 50 Stempel auf Ziegeln und Wandziegeln von Porolissum (Dacia Porolissensis) vom Typ L III G bezeugen mit Sicherheit die Anwesenheit einer Abteilung.

Treu blieben dem Niger im Orient zu Anfang des Jahres 194 nur die drei syrischen (III Gallica, IV Scythica und XVI Flavia) und die beiden kappadokischen Legionen (XX Fulminata und XV Apollonaris).

V. Die Datierungen der Ziegel von Porolissum, der Inschrift von Sirmium, des Vorkommens der Abteilung in Numidien passen zu einer eventuellen Bestrafung der Einheit durch Auflösung unter Septimius Severus und nicht zu jener, die für die Herrschaft des Elagabalus vorgeschlagen wurde. Dies ist aber nicht sicher.

2. Das Drei-Legionen-Kommando in Syrien wurde abgeschafft (Hasebroek 1921: 66-70). 3. Die Befehlshaber Nigers wurden von Severus erpreßt (ihre Kinder wurden von Severus als Geiseln genommen: Herodian III.2.3-4); so benutzte er diese Befehlshaber im Krieg gegen Niger und ließ sie danach zusammen mit ihren Kindern hinrichten; dennoch wurde der General Aemilianus von Anfang an hingerichtet (SHA Severus VIII.16-17).

Während seines Aufenthaltes im Osten traf Kaiser Septimius Severus eine Reihe organisatorischer Maßnahmen. 1. Die Heere – die auf seiner Seite kämpften – wurden durch ehrenvolle stadtrömische Münzemissionen der Jahre

4. Die maurischen Bogenschützen, die auf Nigers Seite 20

Nicolae Gudea: Bezüglich der Europäischen Wanderschaft der legio III Gallica zu Beginn des 3. Jh.

Widerstand leisteten, wurden liquidiert (Hasebroek 1921: 64). Diese Einheiten hatten ein großes Blutbad in den Städten Laodikeia und Tyrus (Städte, die Severus Gefolgschaft leisteten: Herodian III. 3.3-6) veranstaltet.

Judaea) erlitt dasselbe Schicksal (SHA Severus IX.5-7). 13. Auch andere Städte Syriens (z. B. Berytus), Bithynien (Nicaea), Palästina mußten ihr treues Festhalten an der Sache Nigers mit dem Verlust ihrer Frieheiten und ihrer Stadtrechte büßen (SHA Severus XIV.6-7).

5. Es wurde eine Neuordnung der Provinz Syrien durchgeführt: die Provinz wurde in zwei neue Provinzen eingeteilt: Syria Coele und Syria Phoenice; die beiden neuen Provinzen wurden den neuen Legionslegaten unterstellt.

14. Die Familie Nigers (Frau und Kinder), die zu Beginn des Krieges noch beschützt wurden, wurden in der zweiten Phase (nach dem Widerstand des Albinus) hingerichtet und das Vermögen beschlagnahmt (SHA Severus VIII.1-2, 1011; IX.2: Niger VI.1-2).

6. P. Claudius Attalus Paterculianus, der Statthalter Thrakiens, wurde wegen seiner Beteiligung an den Taten Nigers in Thrakien, Griechenland und Makedonien (SHA Niger V.6-7) seines Amtes enthoben und aus dem Senat ausgeschlossen (Dio LXXIX.3.9).

15. Der Kaiser unternahm nur einen Strafzug gegen die barbarischen Verbündeten des Niger (Osrhoene, Adiabene, Hatra und Parthien: SHA Severus III.2-1; IX.9-11).

7. Von den Senatoren aus dem Orient, die Niger unterstützt hatten, wurden einige bestraft und viele hingerichtet (SHA Severus IX.8-9). Auch reiche Leute, die Niger geholfen hatten, erlitten dasselbe Schicksal (SHA Severus IX.7-8; Herodian III.4.7).

a. Das Königreich Osrhoene wurde zur römischen Provinz; der König Abgarus IX wurde unterworfen (SHA Severus XVIII.1-2); der Legat der neu eingerichteten Legion (I Parthica) erhielt die Prokuratur der Provinz. b. Es folgen die Unterwerfung des Gebiets bis zum Tigris, der Angriff auf Hatra, die Unterwerfung Adiabenes und wahrscheinlich die Einrichtung der Provinz Mesopotamien (hierbei sind sich die Historiker nicht einig).

8. Den Städten, die die von Niger auferlegten Kriegssteuern gezahlt hatten, erlegte Severus das Vierfache dieser Steuern auf (Hasebroek 1921: 64). 9. Die von Niger zerstörten Städte in Syrien (z. B. Laodikeia, Tyrus, Heliopolis; Herodian III.4-5) unterstützte der Kaiser mit Geldmitteln (Herodian III.6.910). Wie bekannt, ging Niger mit blutiger Strenge sowohl gegen die Städte, die er nicht halten konnte, als auch gegen die abtrünnigen Städte Syriens und Kleinasiens vor (Herodian III.4.1-3). Als Ausgleich für seine Opfer wurde Laodikeia zur colonia erhoben und erhielt, wie auch Heliopolis, das ius Italicum, die erste Stadt ob belli civilis merita, die zweite per bellis civili occasionem.

16. Um sich den Senat nicht zu entfremden, ließ Septimius Severus nur wenige Senatoren hinrichten. In dem meisten Fällen zog er ihr Vermögen ein und schickte sie in die Verbannung. Mit dem Geld konnte Severus die Forderungen der Soldaten erfüllen (Dio LXXIV.9). Erst 197 begann ein regelrechtes Schreckensregiment, dem zahlreiche Senatoren zum Opfer fielen (SHA Severus XIII.1-9). Noch 197, während des zweiten Partherkrieges, ergriff der Kaiser strenge Strafmaßnahmen gegen die Reste der Heere Nigers und gegen Nigers Freunde und Helfer (SHA Severus XV.4).

10. Die Stadt Byzantium wurde nach einer Belagerung von zweieinhalb Jahren systematisch zerstört (die Wehrmauern, Bäder, Tempel, Theater usw. wurden abgerissen); staatsrechtlich wurde sie zum vicus degradiert, verlor das ius civitatis und wurde der Verwaltung von Perinthus unterstellt (Dio LXXV.14.3). Die Belagerung und Zerstörung Byzantiums wurde von Legionsabteilungen aus Ober- und Untermoesien unter dem Kommando von L. Marius Maximus Perpetuus Aurelianus, dux exercitus moesiacus aput Byzantium et aput Lugdunum (CIL VI.1450), durchgeführt. Der Anfang der Belagerung ist Juli-August 193 anzusetzen; erst im Dezember 195 wird die Stadt infolge einer Hungersnot gefallen sein (Dio LXX.14.2).

All diese Taten des Severus – Belohnungen für Treue und Strafen für Untreue – lassen uns vermuten, daß möglicherweise eine solche strenge Strafe auch über die legio III Gallica verhängt wurde. Die Tatsache, daß die Legion in Porolissum ihren Namen auf Baumaterialien setzt, daß sie in Sirmium ihren Namen erwähnt, daß sie sich in Viminacium sogar einen kaiserlichen Beinamen zulegt, passen allerdings nicht gut zu einer Verurteilung zu damnatio memoriae (egal, zu. welchen Zeitpunkt dies stattgefunden haben mag). Sie könnte sogar eine Abweichung von einer Regel darstellen (falls es eine solche gab).

11. Die Stadt Antiochia – die ehemalige Haupstadt Syriens – wurde staatsrechtlich zur kome = vicus erniedrigt und Laodikeia untergeordnet (Herodian III.6.9; SHA Severus IX.4-5).

Wie diese Abteilungen der legio III Gallica nach Porolissum, nach Viminacium oder nach Sirmium gelangten, ist schwieriger festzustellen. Man kann annehmen, daß sie von den Abteilungen der dakischen und

12. Die Stadt Neapolis in Palästina (bzw. in der Provinz

21

Limes XVIII

moesischen Legionen eingegliedert wurden, die am Orientkrieg auf der Seite des Septimius Severus teilgenommen hatten, und daß sie dann mit diesen zu deren Standort oder in deren Aktionsgegend zurückkehrten (Saxer 1967: 46).

descoperiri arheologice pîna în anul 1977. (Zalau). Hasebroek J. 1921 Untersuchungen zur Geschichte des Kaisers Septimius Severus. (Heidelberg). Keppie L. 1984 The making of the Roman army from Republic to empire. (London). Le Bohec Y. 1989 L’Armée Romaine, sous le Haut-Empire. (Paris). Macrea M. 1969 Viata in Dacia romana. (Bucuresti). Mirković M. 1990 Sirmium et l’armée romaine. Arheološki Vestnik 411: 631-642. Parker H.M.D. 1958 The Roman legions. (2nd ed. with bibliography by G.R. Watson, Cambridge). Picard G. Ch. 1944 Castellum Dimmidi. (Algiers). Ritterling E. 1925 Legio (III) Gallica. In RE XII: Cols. 15171532, Nr. XXXIII. Tóth E. 1978 Porolissum. Das Castellum in Moigrad. Ausgrabungen von A. Radnóti 1943. Régészeti Füzetek ser. II, Nr. 19. (Budapest). Saxer R. 1967 Untersuchungen zu den Vexillationen des römischen Kaiserheeres von Augustus bis Diocletian. Epigraphische Studien 1. (Köln). Szilagyi J. 1946 A Dáciai erödrendszer helyörségei és a katona i téglabelyegek. (Budapest). Vulic N. 1905 Antike Denkmäler in Serbien. Jahreshefte des Österreichischen Archäologischen Instituts VIII. Bild 1-24. Watson G.R. 1969 The Roman soldier. (London). Wheeler E. L. 1996 The laxity of Syrian legions. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 229276.

Es ist bekannt, daß Ti. Claudius Claudianus ein Sonderkommando über Abteilungen der dakischen Armee (bzw. Abteilungen der Legionen V Macedonica und XIII Gemina) im Bürgerkrieg innehatte (CIL VIII.5349, 7178). 195 war er Legat der legio V Macedonica. Ob sein Kommando über die legio XIII Gemina in die vorangehende oder darauffolgende Zeit fällt, ist nicht zu entscheiden (Saxer 1967: 46). Es ist auch nicht eindeutig festzustellen, so R. Saxer, in welches Jahr sein Sonderkommando fällt. Es gibt zwei Möglichkeiten: entweder führte er die dakischen Vexillationen in den Krieg gegen Clodius Albinus (193), oder aber führte er das dakische Expeditionskorps in den Jahren 193-194 im Kampf gegen Pescennius Niger an. J. Fitz nimmt an. Claudianus habe das dakische Korps 193 in Italien übernommen und dann im Kampf gegen Niger bis 196 befehlit, wobei er zunächst Legat der legio XIII Gemina, dann der legio V Macedonica gewesen sei (Fitz 1961: 180). Auch falls die Fälle Viminacium oder Sirmium in Frage gestellt werden könnten, kann der Fall Porolissum nicht umgangen werden. Die Stempel der Einheit erschienen nicht nur im Kastell auf dem Gipfel Pomet. Sie traten auch in der Zivilsiedlung und im Amphitheater auf und lassen sich in ungefähr dieselbe Zeitspanne datieren; somit wurde die Einheit bei Bau-, Reparaturund Instandhaltungsarbeiten des Komplexes zu Beginn des 3. Jh. verwendet. Der Abzug der Abteilung von Porolissum kann zeitlich noch nicht festgelegt werden. Nur die Tatsache, daß die Stempel der Einheit in den späten Bauarbeiten (nach 240) nicht mehr erscheinen, könnte Hinweis darstellen. Bibliography Birley A. 1971 Septimius Severus, the African emperor. (London). Campbell B.D. 1994 The Roman army. 31 BC- AD 337. A source book. (London). Dabrowa E. 1996 The commanders of Syrian legions, 1st-3rd centuries AD. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 277-296. Daicoviciu C. 1951 Porolissum. RE XII. Sp. 265-270. Fitz J. 1961 Ti. Claudius Claudianus cursus honorumához. Archäeologia Értesitö 88: 180-183. French D. 1994 Legio III Gallica. In E. Dabrowa (ed.) The Roman and Byzantine army in the east. Proceedings of a colloquium held at the Jagellonian University, Krakow in September 1992: 29-46. Gudea N. 1978 Descoperiri arheologice si epigrafice mai vechi sau mai noi de la Porolissum. Acta Musei Porolissensis II: 65-72. Gudea N. 1989 Porolissum. Un complex daco-roman la marginea de nord a Imperiului Roman I. Cercetari si

22

Nicolae Gudea: Bezüglich der Europäischen Wanderschaft der legio III Gallica zu Beginn des 3. Jh.

Abb. 1. Die Karte des Römischen Reiches im 2.-3. Jhds. Auf der Karte markiert sind die Plätze wo die Inschriften der legio III Gallica gefunden wurden.

Abb. 2. Ziegelstempel der legio III Gallica von Porolissum und die zeitgenössischen Ziegelstempel der legio VII Gemina Felix und cohors III Dacorum (?).

23

Limes XVIII

Abb. 3. Die Inschrift von Viminacium (Zeichnung nach N. Vulić).

Abb. 4. Die Inschrift von Sirmium (Zeichnung nach N. Vulić).

24

Why did the Roman empire cease to expand? W.S. Hanson

Many ancient historians see the last years of Augustus’ reign as marking the end of any attempt to pursue the aim of continuing expansion. Even the conquest of Britain by Claudius and of Dacia by Trajan tend to be seen as aberrations. Yet expansion of the empire continued in an intermittent and piecemeal way until the early C3rd and the ideology of expansion was often apparent even when its physical expression was not achieved. For example, emperors not usually associated with the expansion of empire, such as Nero and Gaius, were responsible for pursuing not inconsiderable territorial gains, and the philosopher emperor, Marcus Aurelius, appears to have been contemplating the conquest and annexation of new territory in central Europe. Rather than seeing occasional territorial expansion as the aberration, it may be argued that the reverse is the case: that it was the reign of the emperor Hadrian, with its concentration on consolidation and frontier control, which represented the major break with the past. Roman emperors seem, wherever possible, to have pursued policies of continued, if limited, territorial expansion throughout the whole of the period from the later C1st BC to the early C3rd AD. Thus it cannot be argued that the empire ceased to expand because of any lack of desire to augment it or because it had reached some accepted optimum size. This paper will consider the underlying reasons why expansion came to a halt, placing particular weight on socio-economic factors entirely outside Roman control.

standard assumption, usually implicit rather than explicit, is that the empire had reached its natural limit, its optimum size, by the end of the reign of Augustus and, accordingly, expansion ended because there was no strategic need for it. Thus once the Rhine-Danube line had been established in the west, the desert fringes reached in the south and control extended to the Euphrates in the east, any further expansion was either an aberration on the part of a particular emperor, or a minor adjustment or rationalisation of the frontier.

Introduction By the second quarter of the C3rd both internal political instability and external pressure ensured that any thought of further imperial expansion was impossible. Indeed, we see major retrenchment by the third quarter of the century. In AD271 the emperor Aurelian was obliged to evacuate the province of Dacia in the face of external pressure, while a decade earlier under Gallienus control of the agri decumates was given up and the Roman army withdrew to the Rhine-Danube line. When Rome was constantly on the defensive in the east against the new Persian dynasty, and in the west was faced by the newly amalgamated tribes of north-western Europe, such as the Picts, the Franks and the Alamanni, or the new groups sweeping across central and eastern Europe, such as the Vandals and the Goths, it was all she could do to maintain control of her territory, particularly when she faced major problems on more than one front simultaneously. Though she did recover from the serious setbacks of the middle of the C3rd with the establishment of the tetrarchy, never again was Rome able to contemplate further territorial expansion.

However, the ideal limit of the empire clearly was not predetermined, understood and agreed by successive emperors. The establishment of the Rhine-Danube frontier alone should serve to demonstrate this fact, for it was chosen as a convenient demarcation line, intended originally to be no more than a temporary stopping-off point as the frontier advanced north and east towards the Elbe and beyond. Nor should it be assumed that the limits of empire were determined by a desire to establish ‘natural’ or ‘scientific’ frontiers, if, indeed, such things exist (vide Whittaker 1994: 60-3). To take a specific example, the decision to reconquer lowland Scotland by Antoninus Pius was a political one, an attempt to win military prestige and justify selection for imperial honours (Breeze 1975: 75-8; Hanson & Maxwell 1986: 59-69), not a strategic one taken simply in order to advance the frontier in northern Britain to a better line.

But for most commentators, the age of expansion was long over. Many ancient historians see the last years of Augustus’ reign as marking the end of any attempt to pursue the aim of continuing expansion (eg. Brunt 1990: 468). At best, reference may be made to the conquest of Britain by Claudius and of Dacia by Trajan, but these tend to be seen as aberrations. Though one may dispute the accuracy of this synthesis, and the current author certainly does so, nonetheless, the level of sustained expansion does fall away quite dramatically from the later C1st to the C3rd by contrast with the last years of the Republic and early years of the principate. This paper attempts some explanation for this phenomenon.

A lack of inclination It is inevitable that the concentration of political power in the hands of a single individual will result in fluctuations in policy. It is quite clear that some emperors, such as Hadrian, deliberately chose not to pursue any further expansion of empire, while others, such as Antoninus Pius, did so in only a very limited way. But most emperors seem to have harboured and pursued the desire to expand the empire, a process which continued in an intermittent and piecemeal way until the early C3rd. Indeed, the ideology of expansion was often apparent even when its physical expression was not, or could not be, achieved (Woolf 1993a: 181-5). The obvious examples of continuing expansion are the invasion of Britain by Claudius and of

The optimum empire The study of Roman frontiers during the principate is, all too often, predicated on the assumption of a largely static empire in which Rome’s primary motivation was defensive and any expansion was linked to a desire to improve security (eg. Luttwak 1976; Wheeler 1993a; 1993b). The 25

Limes XVIII

Dacia and Parthia by Trajan. But even emperors like Gaius and Nero, who are not normally associated with such attitudes, can be seen to fall into the same general mould and to have harboured expansionist tendencies.

is generally regarded as eschewing the physical expansion of the empire, yet the province of Syria was extended southwards to include Dura Europos on the Euphrates and garrisons were placed along the River Khabur in northern Mesopotamia (Birley 1987: 145; Kennedy 1987, 282-3), a situation which one commentator has described as occupation without annexation (Dillemann 1962: 197). As a result of the eventual success of his Marcomannic Wars, the Historia Augusta suggests that Marcus was contemplating the creation of two new provinces beyond the Danube, Marcomannia and Sarmatia (Marc. Aurel. XXIV.5-6; XXVII.10), and both he and Lucius Verus are referred to as propagatoribus imperii on an undated medallion probably from the mid-170s (Birley 1987: 2534).

Thus, Gaius annexed the client kingdom of Mauretania and seems to have undertaken some form of expedition beyond the Rhine which, it has plausibly been suggested, was intended to ensure the security of the German frontier before embarking on a expedition beyond the Ocean (Barrett 1989: 116-20, 129-32). Indeed, there can be little doubt that Gaius intended the invasion of Britain, having assembled large numbers of troops on the northern coast of Gaul (Tacitus Agric. 13), even if that expedition was subsequently aborted. Similarly, Nero continued the advance westwards in Britain, before it was halted for almost a decade by the Boudican revolt. He did expand directly controlled territory by annexing three small client kingdoms: that of Julius Cottius in the Alps, the territory of the Iceni in eastern Britain, which precipitated the Boudican revolt, and Pontus on the southern shore of the Black Sea. It seems that he also planned an expedition against the Albanians in AD68, having conscripted a new legion and assembled troops from Germany, Gaul and the Balkans, but was forestalled by the outbreak of civil war (Kolendo 1982; Isaac 1990: 42-3).

Septimius Severus was also responsible for the considerable expansion of Roman territory. In 197 he annexed most of Osrhoene and subsequently Mesopotamia as far south as the Jebel Sinjar mountain range. Subsequently we see Roman garrisons pushed down the Euphrates 80 miles or more beyond Dura Europos (Invernizzi 1986), and it may have been Severus’ intention to annex the whole of the land between Tigris and Euphrates, as Trajan had attempted. The considerable extension of the frontiers in North Africa also involved some territorial expansion (Daniels 1987: 250-4). Roman control was extended south of Mauretania Caesariensis and westwards from Numidia beyond the Ouled Naïl Mountains and out towards the Saharan Atlas along the southern edge of the High Plains. It may well have been the intention to re-occupy southern Scotland after a series of major campaigns there, but he died before his plans could be brought to fruition (Birley 1988: 170-87). Finally, Severus is styled propagator imperii on a number of inscriptions largely derived from the African provinces and dated throughout the middle and later years of his reign (Birley 1974: 23-4).

Thereafter, in the Flavian era we see a major commitment to continuing expansion in the province of Britain, with the complete conquest of the island all but achieved. Campaigns under successive governors in Wales, northern England and finally into Scotland, culminated in the defeat of the Caledonians and the circumnavigation of the island (Hanson 1991). In the same period we see substantial territorial gains in Germany, where direct Roman control was first extended into the re-entrant angle between Rhine and Danube under Vespasian, and the process was continued and consolidated with the construction of forts along the rivers Main and Neckar under Domitian (Schönberger 1969: 155-63). In the Taunus and Wetterau Domitian launched a successful campaign against the Chatti which resulted in the occupation of a large salient beyond the Rhine (Jones 1992: 128-131), though the importance of the campaigns and extent of the advance have tended to be downplayed because of the hostility of contemporary sources (eg. Tacitus Agric. 39). In Africa, too, we see a slow but steady advance. Under Vespasian the boundaries of both provinces of Mauretania were redefined and in Numidia military occupation was extended south into the northern slopes of the Aures Massif (Daniels 1987: 240-2). So it was too in the east. In AD72 Vespasian re-annexed the client kingdoms of Armenia Minor and Commagene, the latter involving the use of force. The presence of a legionary vexillation indicates that the Albanians, located on the south-western shores of the Caspian Sea, had also been brought within some form of Roman control (Jones 1992: 156-7; Bosworth 1976: 74-6).

It is difficult, therefore, to argue that there was any basic change in Roman ideology, which saw empire as a birthright, sanctioned by the gods, most famously expressed in the lines from Vergil’s Aeneid: I set no boundary in space or time. I have granted them dominion and it has no end. Yes, even the furious Juno, who now wearies sea, earth and heaven with the strain of fear, shall amend her plans, and she and I will foster the nation which wears the toga, the Roman nation, masters of the world (I: 278-9).

Nor do we see any sustained challenge to the fundamental importance within Roman society of military success and concomitant personal glory. Most emperors still considered it necessary to emphasise their military prowess and victories through the various media of propaganda available to them, including imperial salutations (Table 1), honorific titles derived from the names of conquered peoples, triumphs, celebratory coin issues and victory monuments. What could be more indicative of the imperial

Nor is the process restricted to the C1st. Marcus Aurelius 26

W.S. Hanson: Why did the Roman empire cease to expand?

ethos in this context than the actions of Claudius in relation to the conquest of Britain? After the initial landings and the defeat of the major forces of opposition, Aulus Plautius was ordered to await the arrival of the emperor. Claudius risked a lengthy and dangerous sea crossing to reach Britain as quickly as possible in order to march at the head of his victorious troops into Camulodunum, even bringing with him elephants to bolster the spectacle (Dio LX.21), and then departed after only 16 days. The Senate awarded the title Britannicus both to him and to his young son; gave permission for him to celebrate a triumph (held on his return to Rome in 44); approved an annual festival of celebration; and ordered the construction of triumphal arches in Rome and Gaul. Claudius subsequently issued a series of coins to commemorate his victory (eg. BMC 1.168) and boasted of his achievements in expanding the boundaries of empire beyond the ocean in a speech to the Senate (ILS 212). In practical terms Claudius’ personal involvement in the enterprise was minimal and entirely irrelevant to its success, but the propaganda associated with that involvement was immense (Levick 1990: 144). For example, a number of poems were written to commemorate the conquest and eulogise the emperor (Anthologia Latina I, 419; 423) and local communities set up their own monuments to celebrate the event (eg. ILS 217).

known world was larger than, for example, was understood at the time of production of the famous map of Agrippa early in the reign of Augustus was bound to impact on expectations of what was achievable. Indeed, it has been argued that it was the under-estimation of its overall size in Agrippa’s calculations which encouraged Augustus to believe that conquest of the world was feasible (Brunt 1963: 175-6; Wells 1972: 6-8). It must have been daunting, even to Roman confidence, to begin to realise the enormity of the distances involved and the reservoirs of manpower contained in, for example, central and northern Europe. Table 1 Imperial salutations Total number of salutations taken Augustus 21 Tiberius 8 Gaius 71 Claudius 27 Nero 122 Vespasian 20 Titus 17 Domitian 22 Nerva 2 Trajan 13 Hadrian 2 Antoninus Pius 2 Marcus Aurelius 10 Lucius Verus 5 Commodus 8 Pertinax 0 Didius Julianus 0 Septimius Severus 15 Clodius Albinus 1 Pescennius Niger 1 Caracalla 3 Geta 1 Emperor

There is some indication in the surviving contemporary writings of historians and orators of a change in attitude towards further expansion of empire from the mid-later C2nd AD. Thus, several Greek writers, such as Aelius Aristides, Appian and Pausanias, allude to the limits of empire, their view coinciding broadly with the period when the definition of Roman frontiers on the ground had reached its climax with the creation of linear barriers. For example, Appian states: Possessing the best part of the earth and sea the Romans have, on the whole, aimed to preserve their empire by the exercise of prudence rather than to extend their sway indefinitely over poverty-stricken and profitless tribes of barbarians. (Pr. 7).

1. According to Dio (LIX.22), though these are not featured on his coinage 2. At least 12

The restrictions of geography

But support for the principle of expansion continues to be maintained even into the C3rd in, for example, the writings of the Greek historian Herodian (eg. 1.2.5; 1.5.6). Thus, with reference to Commodus he writes:

It is also clear that as the empire increased in size, and particularly as it moved away from being entirely focused on the Mediterranean, the logistics of expansion and control became increasingly difficult. The physical constraints of distance had a two-fold impact. Basic communications were very slow, which had a detrimental impact on centralised decision-making. It might take as much as several months to receive a reply to a message from a frontier zone sent to the emperor in Rome, making it difficult to react swiftly to favourable or unfavourable circumstances. Arguably one direct result of this, which becomes apparent from the mid-C2nd AD, was the need for emperors to take direct command of major campaigns, both offensive and defensive (Millar 1982: 9-10, 21-2; Austin & Rankov 1995: 124-5, 205-213). The transport of resources and of personnel was even slower than the

“If you subdue them all and extend the northern empire to the ocean, it will be wonderful for you to return in triumph, leading home barbarian kings and governors as prisoners in chains. This was what made your Roman predecessors great and famous” (Herodian 1.6.6). There is no doubt that geographical knowledge increased as the empire expanded, as is demonstrated by a comparison between Strabo’s Geography and the world map of Ptolemy, another Greek geographer writing a century later. Increasing recognition that the size of the 27

Limes XVIII

relaying of messages. This made considerable long-term planning necessary in advance of any major campaign of conquest, since troops might need to be assembled from various parts of the empire. Thus, the invasion of Britain involved troops from the Rhine and from Pannonia (Frere 1987: 48); while in addition to the legions based in the eastern provinces, both Trajan and Lucius Verus employed troops from Pannonia, Germany and Moesia in their Parthian campaigns, Trajan using troops from Dacia also (Mitford 1979: 1196-9, 1203-4). Clearly the need for longterm preparations did not prevent such campaigns, but it was a constraint. Moreover, the movement of large bodies of troops from one frontier to another presupposes that the security of the province from which they were taken was sufficiently assured to allow a diminution of the forces in garrison. There are possibly two occasions when the occupation of parts of northern Britain was given up precisely because troops were required elsewhere. Thus, the transfer of a legion and associated auxiliaries to the Danube frontier in the late 80s was the determining factor in the abandonment of Agricola’s conquests in northern Scotland and a similar situation may have pertained in relation to Trajan’s campaigns of conquest in Dacia, when direct military control in Britain was withdrawn to the Tyne-Solway isthmus (Hanson 1991: 151-2, 166).

One such event is often regarded as a turning point in Roman history, when in AD9 Arminius led a major uprising and destroyed three legions in the Teutoberg Forest. As a result Augustus abandoned his plans to complete the subjugation of Germany beyond the Rhine. More usually, however, it was the unexpected demands for troops in one theatre of war which brought expansion to an end in another, as already alluded to above. Even the threat of warfare on another front could be sufficient to bring campaigning to a halt. Thus, Domitius Ahenobarbus penetrated deep into Germany crossing the Elbe between 6 and 1BC, but a Parthian war was thought to be pending so any further engagement with the Germans was broken off (Dio LV.10a.3). Similarly, in 175 Marcus Aurelius was obliged to come to terms with the Iazyges after the rebellion of Avidius Cassius threatened the stability of the eastern provinces (Dio LXXI.16-17). Social and economic development on the periphery Debate about the cessation of Roman imperial expansion tends to be couched primarily in political or military terms. But in recent years it has become increasingly clear that social and economic factors were of equal, if not greater, importance (eg. Groenman-van Waateringe 1980; Whittaker 1994), even if some doubt the validity of some of the arguments put forward (cf. Woolf 1993c).

The nature of the terrain beyond Rome’s frontiers was a further constraint to continuing expansion. The empire had reached the limits of the ocean in the west, but was bounded by deserts in the south and east, and dense forests in the north-west. Not only did these present constraints to movement, but they made the local population more difficult to conquer. The Romans were particularly susceptible to guerrilla tactics and often had difficulty dealing with mountainous and forested terrain in which enemy forces could disappear. According to Dio, Severus’ campaigns in north Britain suffered such a fate (LXXVI.13.1) and Quinctilius Varus lost three legions in AD9 in Germany when he was ambushed in the Teutoberg Forest. This is not to say that Rome was not capable of campaigning effectively in such terrain - the success of Trajan and Severus in Parthia and of Trajan also in Dacia makes clear that Rome could overcome such constraints if she wished - but rather that it was much more difficult and required greater sustained effort.

At the simplest level, peoples whose assets were not fixed, or at least not concentrated, are more difficult to conquer (Luttwak 1976: 45; see also below). The Roman army was supremely effective in high-intensity warfare, either in open battle against massed opposition or in major sieges of towns or cities. It was that much more difficult to achieve any single decisive victory faced with peoples who lacked central places, who did not depend on an urban-focused socio-economic structure, and whose resources were rural, scattered and diffuse. Conquest then involved a slow process of attrition, potentially even a war of extermination. These are precisely the conditions which Rome seems to have faced in much of north-western Europe and in the semi-desert of North Africa. Furthermore, after successful conquest an army of occupation needs to be fed. The literary and documentary evidence from the wider empire makes clear that the process of supplying the army was both complex and variable (Breeze 1984: 272-82). On the one hand, there is clear archaeological evidence of the provision of military supplies from long distances. On the other, it has long been recognised that in the ancient world, where overland transportation was slow and expensive, supplies would have been obtained locally whenever possible (Theodosian Code 7.4.15). Though Rome was capable of organising long-distance supply systems, and these did play an important part in her ability to extend her empire beyond the Mediterranean littoral (Fulford 1992), it was simply not practicable to rely on external supply on a permanent basis for all commodities on all the frontier zones. Rome was well aware of these logistical problems. Thus, for example,

Force of circumstances There is a tendency for historians analysing the earlier empire to assume, perhaps subconsciously, that the Romans were almost always in control of events, for from at least the mid-C3rd BC Rome had been the dominant power in the ancient world. This is partly a reflection of the nature of their literary source material which consistently reflects a confidence in Rome’s abilities, a belief that success was her manifest destiny, and gives an entirely one-sided picture. Rarely do we get the viewpoint of those who opposed Rome. But no power in the ancient or modern world can control events completely.

28

W.S. Hanson: Why did the Roman empire cease to expand?

it is precisely in the regions of high population and good food production along the Neckar, Main and Taunus that advance beyond the Rhine was accomplished (Whittaker 1994: 89). Similarly, the furthest limit of Roman occupation in northern Britain in the C1st closely followed the Highland boundary fault, the minimum area necessary to ensure the inclusion of all agricultural land in Scotland considered to be of first class quality and which had the ability, or at least the potential, to sustain Roman garrisons (Hanson 1997). But such potential had to be realised. Successful long-term occupation required that such regions were able to produce an agricultural surplus and had the infrastructure already in place for its collection and redistribution (Groenman-van Waateringe 1980; Whittaker 1994: 85-97). This was all the more important if the region was to progress from military control to civil administration with the establishment of urban centres with a non-food producing population.

particularly its ruling elite, to perceive the potential benefits of becoming Roman and seek to emulate Roman culture. This is not to say that the assimilation of less developed areas was impossible, but rather that the timescale involved was too great for Rome to maintain the impetus of continuing expansion. One archaeological manifestation of more advanced socioeconomic development in pre-Roman Iron age Europe is the use of coinage. Even if not generally accepted as indicating the establishment of a monetary system or early state formation, nonetheless a developed system of coinage with consistent weight standards reflects the existence of some centralised issuing authority and the development of a more sophisticated economic structure. In some cases coins even indicate the potential literacy of the issuer, or at least some awareness of the value of writing. Coin use in Europe ultimately derives from Greek originals, subsequently overlain in the C1st BC by silver and bronze coinage inspired by Roman traditions (Allen 1980: 22-3), so its adoption also indicates a willingness to respond to new ideas coming in from outside. The distribution of preRoman Iron age coinage is generally associated with tribes of Celtic origin, though distinguishing archaeologically tribal groups of Celtic or Germanic ethnicity can be difficult. By the C1st BC coin distribution extended from south-eastern Britain in the west across the whole of central Europe to the Black Sea. It stretched beyond the Rhine to the upper reaches of the Elbe and beyond the Danube to the Carpathians (Fig. 1). In the former respect in particular it extended well beyond the furthest established limit of the Roman empire. But absolute measurement of coin distribution does not take into account on-going south-westward tribal movements in the C1st BC/C1st AD. This brought a number of German and Sarmatian tribes closer to what became imperial territory. For example, the Suebi moved from the Elbe to the Rhine; the Marcomanni displaced the Boii eastwards from Bohemia to the middle Danube; and the Iazyges pushed into the Hungarian plain (Todd 1992: 23; Rieckhof 1995; Szabo 1992: 73). Thus German and Sarmatian expansion made significant inroads into what had been Celtic territory. Significantly, there is no indication that the major German tribes, such as the Chauci, Cherusci, Chatti, Hermunduri or Suebi, were coin using (Allen 1980: 68). Thus many of the major tribes occupying the periphery of the Roman empire in central Europe by the C1st AD were likely to have been less well developed in socio-economic terms than those further west, and concomitantly less amenable to the changes which would have been involved in their incorporation into the Roman empire.

However, the issue is not merely one of geographical determinism whereby the poorer areas incapable of producing a surplus were unable to sustain a Roman presence. More fundamentally, continued successful expansion of the empire into new territory was directly linked to the level of social development which had occurred in the region concerned. As the empire expanded, peoples who originally may have resisted Roman conquest gradually became assimilated and integrated therein – the process which is popularly known as romanization. Indeed, this process was essential to the successful expansion of that empire, for if it had been necessary to continue to station troops every few miles on a permanent basis in order to maintain control of the civilian population, the manpower requirements would have been such that there would soon have been no forces available to bring about and maintain further expansion (Hanson 1988; 1994: 156). Thus, ultimately, Rome depended on the acquiescence and co-operation of those peoples she conquered in order to sustain her empire and allow it to grow. As new peoples became fully integrated into the empire, so Roman troops became available for further conquest. Moreover, such newly assimilated areas, rather than being a drain on the resources of empire, served to enhance its capability to expand by augmenting that resource with further manpower. It was no accident that one of the immediate consequences of conquest was the recruitment of auxiliary troops to serve in the Roman army (eg. Willems 1984). Thus, successful romanization was essential to the continuing expansion of the empire. The exact nature of the mechanisms involved in that process of cultural assimilation remains much disputed (eg. Millett 1990; Woolf 1998 to quote the most extensive considerations of the issues), though it is clear that such change could not be forced upon unwilling recipients. Thus, the extent to which, and the speed with which, that process took place was a direct reflection of the pre-existing socio-economic development of the areas concerned. The more that any indigenous society had developed along the road to statehood, the more likely were its population, and

Finally, in order to administer occupied and pacified territories, Rome depended heavily upon the participation of the indigenous population, particularly the local elite. Roman local administration was based on self-governing urban centres or, more accurately, city-states with an urban focus. In the eastern empire this meant the pre-existing Hellenistic cities and in the western Mediterranean the 29

Limes XVIII

long-established Greek or Phoenician colonies. Beyond the coastal fringe of the Mediterranean in the west Rome was obliged to create urban centres. In the later Republic and early empire many colonies were founded as unofficial garrisons and exemplars of the Roman way of life, populated in the main by veterans of the Roman army. But there were insufficient time-expired soldiers or other candidates from within the imperial core to sustain such action as the empire expanded deep into Europe. In these areas successful integration into the Roman system was heavily dependent on the extent to which the local population could be encouraged to develop their own towns to act as administrative centres for self-governing communities, usually based on the pre-existing indigenous tribal structure.

for these criticisms, the sites still have value as predictive indicators of potential levels of socio-economic, or, to use Woolf’s phraseology (1993c: 17), techno-economic development. A brief case study may serve to illustrate the force of the argument concerning the impact of variation in levels of socio-economic development on the success of Roman occupation. In Britain there is a clear distinction between the civil part of the Roman province in the south-east, and the military zone in the north and west. After initial resistance the former was rapidly conquered by Rome and, judging by the paucity of Roman forts (Fig. 2), does not seem to have required much in the way of long-term military control. Thereafter the south-east shows all of the characteristics which typify a developed Roman province in the western empire, with administrative control vested in major urban centres, the so-called civitas capitals, an extensive road system linking a range of smaller urban settlements, a hierarchy of romanized rural sites, including villas and temples, and the ubiquitous presence, even on lower order rural settlements, of a distinctive romanized material culture. By contrast, the north and west was much slower to succumb to Roman arms and remained dominated by the military presence throughout the life of the province. Urbanisation was slow to develop and limited in nature and extent; rural settlements exhibit little sign of Roman influence with very few villas being established; and away from the military sites, the distribution of romanized material culture is relatively sparse. It is surely more than coincidence that the civil zone coincides almost exactly with the tribal areas in the pre-Roman Iron age which exhibit signs of a more developed socio-economic system, with the appearance of a complex political hierarchy, proto-urban centres (oppida), independent coin use, and both long-term diplomatic and trading contacts with the Roman empire. Moreover, many of the civitas capitals can be shown to have developed on or immediately adjacent to the sites of the pre-Roman Iron age oppida (Burnham 1986: 193-4).

Thus, it has reasonably been argued that successful permanent Roman occupation was most likely in those regions where a well-organised proto-urban structure was already in place as a base upon which Rome might readily project her own social and administrative system (Groenman-van Waateringe 1980). In this context, attention has been drawn to the broad correlation between the known distribution of the so-called oppida of the late pre-Roman Iron age and the developed extent of the Roman empire in the west (Fig. 1).1 Certainly, those areas of western Europe in which new Roman urban centres with a predominantly indigenous population were successfully established seem to have been already well advanced towards developed and centralised settlement forms of their own. As conventionally defined, these proto-urban sites tend to be very large, well-demarcated, and potentially well-defended, with evidence of economic sophistication, involvement in long-distance trade and signs of industrial specialisation and wealth distinction in their interiors (Collis 1984). However, the correlation of successful Roman urbanisation with Iron age oppida is not straightforward. Firstly, the definition of oppida is itself an issue so that plotting the distribution of proto-urban sites becomes problematic. Indeed, the very status of the sites as proto-urban and their recognition is a consistent phenomenon has been challenged (Woolf 1993b; 1993c). Secondly, like coinage, the correlation with the furthest extent of the Roman empire is only approximate, though the same arguments in relation to tribal movements apply as for the distribution of coinage. Thirdly, a number of the oppida east of the Rhine and north of the upper Danube seem to have been in decline shortly before the Roman conquest, though this too may have been the result of the movement of German tribes into what were previously Celtic areas (Rieckhof 1995) or, alternatively, of the disruption of their economic system linked to Caesar’s operations in Gaul (Wells 1999: 77-8). But even allowing

Conclusion Not surprisingly, a number of factors, geographical, political and socio-economic, combined to restrict the extent of Roman expansion. The size of the world, once it had been realised, and the nature of the terrain encountered were both major constraints, if not always determining factors. Historical circumstances had an important part to play and eventually brought about not merely the end of expansion, but the onset of retrenchment. But arguably the most important factor in the western empire was the preexisting level of economic development and social organisation with which Rome was confronted. Where this was not sufficiently advanced to provide the infrastructure of supply and administrative control, and to facilitate the cultural assimilation upon which Rome depended, then development beyond a purely military occupation was unlikely and further territorial advance impossible with the military resources available. It was primarily this latter

1

Since the primary concern here is with urban character, Fig. 1 has been extended to include a broader range of sites than Collis allows (1984: 68). It, therefore, includes larger enclosed sites in south-eastern England, which are generally regarded as proto-urban and often described as oppida (Cunliffe 1991, 366-70), and some smaller sites with proto-urban characteristics in eastern Europe (Petres 1976; Glodariu 1983).

30

W.S. Hanson: Why did the Roman empire cease to expand?

factor which slowed and eventually halted the previously rapid expansion in northern and western Europe even before the problems of the later C2nd and C3rd.

the Walls: essays on the prehistory and history of north Britain in honour of George Jobey. (Edinburgh): 264286. Brunt P.A. 1963 Review of H.D. Meyer Die Aussenpolitik des Augustus und die augusteische Dichtung. Journal of Roman Studies 53: 170-176. Brunt P.A. 1990 Roman Imperial themes. (Oxford). Burnham B.C. 1986 The origins of Romano-British small towns. Oxford Journal of Archaeology 5.2: 185-203. Collis J.R. 1984 Oppida: earliest towns north of the Alps. (Sheffield). Cunliffe B.W. 1988 Greeks, Romans and barbarians: spheres of interaction. (London). Cunliffe B.W. 1991 Iron age communities in Britain. (3rd edit., London). Daniels C.M. 1987 Africa. In J. Wacher (ed.) The Roman world. (London): 223-265. Dillemann L. 1962 Haute-Mésopotamie Orientale et pays adjacents. (Paris). Frere S.S.1987 Britannia: a history of Roman Britain. (3rd edit., London). Fulford M.G. 1992 Territorial expansion and the Roman empire. World Archaeology 23.3: 294-305. Glodariu I. 1983 Architectura Dacilor: civila si militara. (Cluj Napoca). Groenman-van Waateringe W. 1980 Urbanization and the north-west frontier of the Roman empire. In W.S. Hanson & L.J.F. Keppie (edd.) 1980 Roman Frontier Studies XII. Papers presented to the XIIth International Congress of Roman Frontier Studies. British Archaeological Reports Int. Ser. 71. (Oxford): 10371044. Hanson W.S. 1988 Administration, urbanisation and acculturation in the Roman West. In D. Braund (ed.) The administration of the Roman empire, 241 BC – AD193. (Exeter): 53-68. Hanson W.S. 1991 Agricola and the conquest of the north. (London). Hanson W.S. 1994 Dealing with barbarians: the romanization of Britain. In B. Vyner (ed.) Building on the past. (Royal Archaeological Institute, London): 149-163. Hanson W.S. 1997 The Roman presence: brief interludes. In K.J. Edwards & I.B.M. Ralston (edd.) Scotland: environment and archaeology, 8000BC - 1000AD. (London): 195-216. Hanson W.S. & Maxwell G.S. 1986 Rome’s north west frontier: The Antonine Wall. (Edinburgh). ILS Inscriptiones Latinae Selectae Dessau, H. (ed.), Berlin, 1892-1906. Invernizzi A. 1986 Kifrin and the Euphrates limes. In P.W.M. Freeman & D.L. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Report Int Ser.297. (Oxford): 357-381. Isaac B. 1990 The limits of empire: The Roman army in the east. (Oxford). Jones B.W. 1992 The emperor Domitian. (London). Jones G.D.B. & Mattingly D.J. 1990 Atlas of Roman Britain. (Oxford). Kennedy D.L. 1987 The East. In J. Wacher (ed.) The Roman world. (London): 266-308. Kolendo J. 1982 Le projet d’expédition de Néron dans le Caucase. In J.M. Croisille & P.-M. Fauchère (edd.) Neronia 1977. (Clermont Ferrand): 23-30. Levick B. 1990 Claudius. (London). Luttwak E.N. 1976 The grand strategy of the Roman empire from the 1st century AD to the 3rd. (London).

The interplay of these various factors does not mean that the limits of empire were in some way pre-ordained. Clearly they did fluctuate over time, for the situations which Rome faced on any of her frontiers were neither uniform nor static. Similarly, socio-economic and cultural distinctions between indigenous peoples are often blurred (eg. Cunliffe 1988: 116-17) and are further complicated by the considerable population movement attested in the late Iron age in central Europe (eg. Todd 1992: 22-3, 47-9). What these factors do mean, however, is that once Rome extended her interests beyond the Mediterranean littoral, successful long-term territorial expansion was more difficult to achieve, and involved greater determination and more consistent application. Any one of the specific factors outlined above might be sufficient to throw the process off course either temporarily or permanently. It has been suggested that the history of Rome is the story of a nation trying to keep pace with the situations produced by the incredible success of it army (Mann 1974: 509). Expressed differently, it could be argued that no state achieves the optimum balance between its range of conquests and its ability to rule them. Acknowledgements I am grateful for the support of the British Academy and the University of Glasgow in facilitating my attendance at the Congress in Amman. A more extensive study on this general theme is in progress. Bibliography Allen D.F. 1980 The coins of the ancient Celts. (ed. D. Nash, Edinburgh). Austin N.J.E. & Rankov N.B. 1995 Exploratio: military and political intelligence in the Roman world from the second Punic war to the Battle of Adrianople. (London). Barrett A.A. 1989 Caligula: the corruption of power. ( London). Birley A.R. 1974 Roman frontiers and Roman frontier policy: some reflections on Roman imperialism. Transactions of the Architectural & Archaeological Society Durham Northumberland 3: 13-25. Birley A.R. 1987 Marcus Aurelius: a biography. (2nd edit., London). Birley A.R. 1988 Septimius Severus: the African emperor. (2nd edit., London). BMC 1923-62 Coins of the Roman empire in the British Museum (ed. H. Mattingly, London). Bosworth A.B. 1976 Vespasian’s reorganisation of the north-east frontier. Antichthon 10: 63-78. Breeze D.J. 1975 The abandonment of the Antonine Wall: its date and implications. Scottish Archaeological Forum 7: 67-80. Breeze D.J. 1984 Demand and supply on the northern frontier. In R. Miket & C. Burgess (edd.) Between and beyond

31

Limes XVIII

Mann J.C. 1974 The frontiers of the Roman principate. In H. Temporini (ed) Aufstieg und Niedergang der Römischen Welt II.1 (Berlin): 508-533. Millar F.G.B. 1982 Emperors, frontiers and foreign relations, 31 BC to AD 378. Britannia XIII: 1-25. Millett M. 1990 The romanization of Britain. An essay in archaeological interpretation. (Cambridge). Mitford T.B. 1979 Cappadocia and Armenia Minor: historical setting of the limes. In H. Temporini (ed) Aufstieg und Niedergang der römischen Welt II.7.2 (Berlin): 11691228. Petres É. 1976 The late Iron age in Hungary with special reference to oppida. In B. Cunliffe & T. Rowley (edd.) Oppida: the beginnings of urbanisation in barbarian Europe, British Archaeological Reports Int. Ser. 11. (Oxford): 51-80. Rieckhof S. 1995 Süddeutschland im Spannungsfeld von Kelten, Germanen und Römern. (Trier). Schönberger H. 1969 The Roman frontier in Germany: an archaeological survey. Journal of Roman Studies 59: 144-197. Szabo M. 1992 Les Celtes de l’est: le second age du fer dans la cuvette des Karpates. (Paris). Todd M. 1992 The early Germans. (Oxford). Wacher J. (ed.) 1987 The Roman world. (London). Wells C.M. 1972 The German policy of Augustus. (Oxford). Wells P.S. 1999 The Barbarians speak: how the conquered peoples shaped Roman Europe. (Princeton). Wheeler E.L. 1993a Methodological limits and the mirage of Roman strategy: part I. Journal of Military History 57.1: 7-41. Wheeler, E.L 1993b Methodological limits and the mirage of Roman strategy: part II. Journal of Military History 57.2: 215-240. Whittaker C.R. 1994 Frontiers of the Roman empire: a social and economic study. (Baltimore). Willems W. 1984 Romans and Batavians: a regional study in the Dutch eastern river area. Berichten Rijksdienst Oudheidkundig Bodemonderzoek 32: 42-491. Woolf G. 1993a Roman peace. In J. Rich & G. Shipley (edd.) War and society in the Roman world. (London): 171194. Woolf G. 1993b Rethinking the oppida. Oxford Journal of Archaeology 12: 223-234. Woolf G. 1993c European social development and Roman imperialism. In P. Brun, S. van der Leeuw & C.R. Whittaker (eds.) Frontières d’Empire: nature et signification des frontières romaine. (Nemours): 13-20. Woolf G. 1998 Becoming Roman. The origins of provincial civilization in Gaul. (Cambridge).

32

Fig. 1. The distribution of pre-Roman Iron Age proto-urban settlements (after Collis 1984; Cunliffe 1991; Petres 1976 & Glodariu 1983 with amendments) and coin producing regions in Europe (after Allen 1980) plotted against the furthest extent of the Roman empire.

W.S. Hanson: Why did the Roman empire cease to expand?

33

Limes XVIII

Fig. 2. The distribution of Roman military sites in Britain (after Jones & Mattingly 1990 with amendments).

34

Mints and the Roman army from Augustus to Diocletian Johan van Heesch Two-thirds of Rome’s state expenditure, at least under the reign of Augustus, went to the army. It is therefore quite reasonable to suppose that to facilitate those payments and money transports, mints were established nearer to the military regions. In this paper a general survey will be given of the main mints of the principate and we will discuss the reasons why some were established outside Rome. It will be argued that the military explanation is only part of the story. After an initial phase under Augustus, where mints tend to be situated outside Rome for political and military reasons, state coinages are concentrated in Rome from the Flavians onwards. The creation of mints outside Rome in the C3rd is not only linked with the growth of military activities but also determined by the uncertain political situation of the emperors and continuous reminting of old coin used as raw material for the production of debased money. It can be presumed that the use of ‘old coins’ for state expenditure became gradually less important in the last century of the principate and that freshly minted coins became increasingly preponderant.

Two-third of Rome’s state expenditure, at least under the reign of Augustus (27 BC-AD14), went to the army (Wolters 1999a) whose legion’s were predominantly situated near the Roman frontiers on the Rhine, the Danube and in the east. Only a minority lay elsewhere in Spain, Dalmatia, Egypt and Africa and some garrisons where situated near Italian ports, in Gaul etc. (Le Bohec 1989: 173-189, 218-19). To pay all these soldiers, often stationed in the remotest corners of the empire, huge amounts of cash were needed and costly and time-consuming coin transports must have been a daily matter. It would then have been important for the imperial authorities to organise the paying of the stipendia, donativa and praemia in the most practical way.

money on the spot and were obliged to pay their troops with gold and silver coins struck at various mints moving with their masters. Well known are the legionary denarii of Mark Antony, struck at a travelling mint in 32-31BC. It does not need much to open a mint; all that is necessary is an oven, some workmen, metal and a few dies to strike the coins. If one knows that each obverse die could easily produce say 30,000 coins, it becomes understandable that even huge quantities could be produced by these moving mints. That it was easy to open such a workshop in any place of the empire is shown also by the coinage produced by usurpers like Clodius Macer in 68 or Carausius in 286 (or 287). They were able to produce high quality coins respectively in Africa and Britain, provinces that neither had a Roman nor a provincial mint at these times. Both struck gold or silver and Carausius even produced beautiful medallions of some artistic quality and originality.

What I will do in this paper is examine the location of the imperial mints from Augustus to Diocletian in order to see if their situation can be linked with an imperial policy to facilitate those payments. If this was not the case, we will try to answer the question what actually determined the choice of the place where these state factories were located.

When Octavian at the age of 15 took the toga virilis in 48BC, minting was not only confined to Rome; in full civil war, mints were operating at different locations. It should not surprise us then that Octavian, after peace was established in 31BC, did nothing to restore the Republican tradition of a central and unique mint in Rome. In the first part of his reign different mints were operating at several places (Fig. 1). Their identification is all but certain, but we might presume mints at Ephesus (29-24BC), Pergamum (28-23 and 19BC), Samos (22BC) and probably as many as five western mints: an unidentified Spanish mint, Emerita, Caesaraugusta, Colonia Patricia (but this mint might actually be situated in Nîmes (Volk 1997: 7778) and Lyons. None of these mints lay on the military frontiers and it is hard to explain their existence with certainty. Sutherland explained the minting of aurei and denarii at Pergamum c.19BC by the hypothetical passage of Roman legions (Sutherland 1976: 55) and military pay and bonuses could well explain the existence of the mint at Emerita, where P. Carisius, propraetor of Lusitania, struck aurei and denarii in 24-23BC. The latter coins, as well as the augustan bronzes with a small shield (caetra) on the reverse and certainly struck somewhere in the north-west of Spain, can be linked with the war of Augustus against the Cantabrians and the Asturians between 27 and 19BC. Augustus’ first massive issues however were those of Caesaraugusta and Colonia Patricia (or Nîmes). Dated

It should be remembered that different kinds of coinages were used throughout the empire. Besides the official state coinage, regional series consisting of local denominations were minted by some provinces in the east, and more than 300 cities, situated in the Greek speaking part of the empire, struck their own coins, mainly in bronze. These so called “provincial” mints ceased their activities in the second half of the C3rd, on the very moment that even Rome stopped minting bronze coins in favour of silver antoniniani issued at several mints in the different corners of the empire. The era of the denarius. From Augustus to Gordian III (27 BC - AD238) During the first centuries of the Republic, Rome, as all great city-states before her, barely needed travelling mints or decentralised minting. Coinage was usually struck in the political centre, as in all city-states of the ancient world. But when eg. after the murder of Julius Caesar (44BC) the Roman imperators struggled for supreme power, the Roman mint was not always at their disposal. They needed

35

Limes XVIII

between 19 and 16BC they were struck after these wars and represent 25% of Augustus’ gold and silver coinages (Volk 1997: 76). By that time (c.19BC), the senatorial mint in Rome, managed by the triumviri monetales, reopened in 19BC after decades of inactivity (on the rôle of the Senate, see Wolters 1999a). It is clear however that these issues of Rome were limited in volume and that the main minting activity lay outside Italy. The emperor as supreme military commander, probably tried to avoid senatorial influence in monetary matters and was eager to retain his imperatorial rights of coinage, just as the military leaders during the civil war did before Actium.

necessary to transfer the gold and silver ingots as well (as part of) the tax income to Rome to provide the emperor with freshly minted coins. In that case, these revenues would have fallen under the direct authority of the senatorial functionaries responsible for the aerarium, the main treasury and the mint. In placing the mint for gold and silver at Lyons, Augustus withdrew the main revenues and the paying of his armies in Germania from the direct control of the Senate. So Lyons stayed the only mint for gold and silver till Nero. Why it was transferred to Rome in 64, the year of an important monetary reform, is not very clear, but the expenditure linked with the reconstruction of Rome after the great fire of 64 and the situation of Lyons that far from the imperial residence in a period of growing resistance, might have been in itself sufficient reasons to explain the change of mint.

The most important step towards consolidation of his financial independence was taken in 15BC when the mint at Lyons opened. Although official dies, used to fabricate these coins, are found all over Gaul, and the existence of a travelling mint can not be excluded (Amandry 1992), we firmly believe that almost all official Roman silver and gold coins from 15BC (Rome stopped minting gold and silver in 12BC) till Nero in 64 were struck at the mint at Lyons. This central rôle of Lyons under the Julio-Claudian emperors is confirmed by Strabo (Geog.IV.3.2: Sutherland 1976: 46-48, Wolters 1999a) and also by inscriptions that mention a military cohort stationed near the mint. But why did they prefer Lyons ?

When Civil War broke out after Nero’s death in 68, the different pretenders struck coins in the several areas where the money, to pay their troops was necessary. Once Vespasian came to power, coinage was, after a while, centralised in Rome. This situation with a unique mint persisted till the reign of Gordian III in 238. Some exceptions however do occur, but it is not our intention to list them all. During Vespasians’ reign (c.69-74) denarii and aurei were also struck at Ephesus, at an unidentifiable Asian mint and in Antioch. The reason for these issues is not clear. Some series might be linked with military expenditure, others were perhaps simply struck by the local governors and intended to be nothing more than fractions of the local cistophori and tetradrachms or to compensate for the closure of the local mint at Caesarea in Cappadocia (Burnett, Amandry & Carradice 1999: 125). An eastern mint was also active under Hadrian (Strack 1933:192-199) and when Septimius Severus was at war with Pescennius Niger (193), he struck at different places, though his main eastern mints were at Laodicea-ad-Mare or at Antioch-adOrontem (Bickford-Smith 1994/1995; Mattingly 1975: cxiv-cxxv). These Severan mints stayed active for some years, probably till 202; and their activity can be linked with Severus’ Parthian wars (Christol 1997a: 23-24).

Rheinhard Wolters (Wolters 1999a; 1999b: 82-83) pointed out two main reasons. First of all Lyons, the capital of the three Gauls, lay not too far from the main Spanish gold and silver mines (Domergue 1990). But that on its own could never have been enough reason to transfer Spanish mints to Gaul as gold and silver revenues came not exclusively from Spain (Lehrberger 1995: 117; Edmonson 1989). Moulds for Roman gold ingots were found in Noricum (Piccotini 1994), gold might have come from Dalmatia and Africa also (Giovannini 2000; Duncan-Jones 1994: 103105) and although taxes in gold and silver were possibly exacted from all provinces, the contribution of the three Gauls, the richest part of the whole empire (Velleius II.39; Josephus Bell. Jud. II.371-373) must have been considerable. This is illustrated by the story of Licinus, procurator of Gaul, who presented to Augustus in 15BC the “many treasures of silver and gold” exacted from this province (Dio LIV.21). So we could say that Lyons lay almost in the centre of all these resources of precious metals.

The most important fact however is that no permanent state mint was established outside Rome from Nero until Gordian III in 238. The implication of this must be clear. Coins used for state expenditure, be it the pay of civilian functionaries or the armies, must frequently have been transported over long distances. Troops in Britain, Germany and along the eastern borders must have received regular shipments of fresh coin. That this was judged not to be an uneconomic or impractical solution is also shown by the fact that even some major local mints like Ephesus, Caesarea, Antioch, Alexandria, and mints on Cyprus and in Lycia (Burnett, Amandry & Carradice 1999: 11) had their coins, be it exceptionally, made in Rome.

Secondly, Augustus, staying in Gaul between 16 and 13 (Halfmann 1986: 159) might have judged it more favourable to transfer his main mint from senatorial territory (Nîmes in Gallia Narbonensis became a senatorial province in 22BC) to an imperial province which was under his sole rule. As the main mining areas and the revenues of these provinces were under his direct administration, it was easier for him to use these revenues directly for the payment of his troops. We should not forget that just in this last decade before Christ, important military actions were taking place in Germania. When the sole state mint had been in Rome, it would have been

Of course, soldiers were not only paid with coins and certainly not always with new ones. As Roman money was 36

Johan van Heesch: Mints and the Roman army from Augustus to Diocletian

very probably slightly overvalued, older and worn coins or those with a lesser silver content could circulate freely together and the Roman government would certainly have put these in circulation again after they had entered the states’ treasury. Together with Roman gold, silver and bronze coins, locally made bronze issues, as the imitative asses of the reign of Claudius or the civic issues of the eastern towns, could circulate freely even in Roman camps. Other silver issues, minted by local authorities in the East, could be used to make state payments also. The mints of Caesarea in Cappadocia and of Antioch not always coined in great quantities, but when they occasionally did so, these coins were probably used to cover state expenditure, especially as Roman gold and silver appears to be excessively rare in the orient during the C1st. The Syrian tetradrachms minted under Vespasian between 69 and 73 were struck in huge numbers and it was calculated that 6,500,000 tetradrachms were issued, enough to pay four Roman legions during five years, exactly the number of forces that were used by Titus when he was at war in the east (Burnett, Amandry & Carradice 1999: 274-275). The same situation occurred under Caracalla, whose Syrian tetradrachms are very numerous.

Cologne in Germania Inferior (opened in 257) and of Siscia in Pannonia Inferior (opened in 262) were on or near the German and Danubian limes where German tribes, Goths or even the own Roman generals formed a constant threat. Coin legends as Gallienus cum exer(citu) suo on antoniniani of Cologne mark clearly the military character of these new founded establishments. The second half of the C3rd also witnessed the emergence of the comitatensian mint, that travelled with the emperor and his train. All this might seem sufficient a reason to link the appearance of decentralised minting with the instable situation of the empire. But in my view this purely military explanation is only part of the story. As most of these new mints also struck coin in the absence of the emperor, this policy cannot have offered any more guarantees for safety than a central mint in Rome. It is true that mints nearer to the military regions are more practical for the paying out of the armies, but how can one explain that it almost took 250 years before this idea was being realised? The answer must be sought in the monetary policy of the C3rd. During the reign of Gordian III (238-244) the silver antoninianus replaced the old denarius. After almost 50 years of continuous weight and fineness reductions of the silver coins, Gordians’ antoninianus, worth two denarii, contained less than 2gms of pure silver, exactly the same amount as the denarius of Septimius Severus (Harl 1996: 127, 130). The antoninianus also deteriorated continuously and its silver content fell from about 48% under Gordian to 27% in the first part of Gallienus reign, to 3% (0.16gms of silver) at the end of his reign in the years 267-268. The pressure on the monetary system was enormous and probably for the first time in Roman monetary history, it became difficult to maintain the older and better coins for long in circulation. Taken into account the exhaustion and the loss of Spanish and other mines (Jones 1980; Domergue 1990; Andreau 1990), the dramatic increase in military expenditure (pay rises, more men under arms, subsidies to barbarian tribes, etc.) the government had to look for other resources. These were found in the continuous reminting of the older and better coins. With the continuous deterioration of the coins, old denarii and antoniniani often contained more silver than the newly minted ones. When in the first two centuries coins that entered the provincial treasuries could easily be spent again in the same region, this was disadvantageous in the third, especially the late C3rd. It was then in provinces where government expenditure was at its heighest, that newly created mints operated to withdraw older and better coins to remint these to the inflationary coinage that became so typical for the last decades of the principate.

Egypt is another case. This province lodged two legions during the first two centuries (Le Bohec 1989: 189) that could never have been paid with the ‘ordinary’ state coinages minted in Lyons or Rome, as these were not allowed to circulate there. Soldiers in this province must have been paid with the Greek inscribed Alexandrian coins made of bad silver or bullion. The era of the antoninianus. From Gordian III to Diocletian (238-284) All this changed drastically during the C3rd. All of Caracalla’s state coinage was struck at the mint of Rome only, but under Diocletian, at the very end of the same century, coin production was spread over 14 mints. This change occurred gradually: in 251 three mints were active, in 260 five, in 268 seven and in 274 Aurelian used 9 mints situated all over the empire, except Africa, Britain and Spain (for an overview see Christol 1977; Carson 1978). Most of these C3rd workshops lay fairly close to the limes or at strategic and administrative centres (Fig. 2). Linking this phenomenon to the increased military readiness and the threat of invasion seems in most cases the most obvious explanation. Faced with enormous dangers at the Danube and on the Persian frontier, Trebonianus Gallus (251-253) opened branch-mints at Viminacium in Moesia Superior and Antioch in Syria. The local Antiochene tetradrachms were gradually replaced by a massive output of Roman silver coins (antoniniani). Gallienus opened a mint in Milan in 258, the seat of his equites, whose commander can be considered as the most important man after the emperor (Alföldi 1967: 408). The closure of Milan under Aurelian in 274 and the opening of a new mint only 50kms south of this city was on the other hand probably due to the distrust of this emperor towards this same body of soldiers (Göbl 1993: 40). The mints of

In a first stage the Roman state tried to withdraw the denarii. Although they were valued at half the antoninianus, two denarii weighed 25% more than the actual antoninianus (under Gordian III two denarii weighed 6.03gms and an antoninianus only 4.50gms). Under Trajan Decius (249-251) and Trebonianus Gallus (251-253) old denarii were sometimes used as blanks and simply restruck to fabricate antoniniani (Mattingly 1939: 42-44; Callu 37

Limes XVIII

1969: 242). Later on, mints were opened in the provinces to remelt the old coins and mint new ones. The immediate impact of a newly created mint on the coin circulation of a region can be demonstrated with several examples based on coin hoards.

military activities or the uncertain political situation of the emperors. The main reason for their existence was the shortness of bullion and the continuing deterioration of the money which necessitated the reminting of old coins used as raw material for the production of more debased money. It can be presumed that the use of ‘old money’ for state expenditure became gradually less important in the last 50 years of the principate and that freshly minted coins became increasingly preponderant. This situation was maintained in the C4th, when decentralised minting became standard practice and recalls of the bullion coinage occurred continuously at shorter intervals than ever.

When Trebonianus Gallus opened his mint of Viminacium in Moesia Superior in 251 (Carson 1990: 90), denarii disappeared rapidly from circulation in Illyricum and only there (Christol 1977: 253; Callu 1969: 256; Bland 1996: 87). Old denarii were apparently recalled to fabricate antoniniani. When Gallienus opened a mint at Cologne in 257 (or 256 ? Christol 1997b), some months after his victory over the Germans, he started minting antoniniani that replaced the existing monetary stock in northern Gaul and the Germanies in a few years time (Christol 1977: 258; van Heesch 1998: 129. For other hoards in Gaul see Foucray 1998: 15). At that time, old denarii still circulated in fairly large numbers in northern Gaul. But as is shown in Fig. 3, these high value coins disappear suddenly from the hoards in 259-260 (van Heesch 1998: 128; Christol 1977: 260263). Their elimination seems to have taken place in such a short period, that it is difficult not to imagine government action to withdraw this denomination from circulation. Although written evidence for a compulsory withdrawal of old and better coins for newer but poorer ones hardly exist for Roman times, we think that this is exactly what happened here. Why should the Roman administration not be able to do what was possible in medieval Europe, where recalls of older coins were common practice in a society where the administration was certainly not more developed than that of the Roman empire (Spufford 1988)?

Bibliography Alföldi A. 1967 Studien zur Geschichte der Weltkrise des 3. Jahrhunderts nach Christus. (Darmstadt) Amandry M. 1992 Les coins monétaires et les monnaies. In Masques de fer. Un officier romain du temps de Caligula. (Saint-Germain-en-Laye): 88-99. Andreau J. 1990 Recherches récentes sur les mines à l’époque romaine, II, Nature de la main d’oeuvre; histoire des techniques et de la production. Revue numismatique 6 ser. 32: 85-108. Bickford-Smith R.A. 1994/1995 The imperial mints in the east for Septimius Severus. Rivista italiana di numismatica 96: 53-71. Bland R. 1996 The development of gold and silver coin denominations, AD 193-253. In C. King & D. Wigg (edd.) Coin finds and coin use in the Roman world. Studien zu Fundmünzen der Antike 10. (Berlin). Burnett A., Amandry M. & Carradice I. 1999 Roman provincial coinage, II. From Vespasian to Domitian. (London & Paris). Callu J.-P. 1969 La politique romaine des empereurs romain de 239 à 311. (Bibliothèque des écoles françaises d’Athènes et de Rome 214: Paris). Carson R.A.G. 1978 Mints in the mid-third century. In R.A.G. Carson & C.M. Kraay (edd.) Scripta nummaria Romana. Essays presented to Humphrey Sutherland. (London): 65-74. Carson R.A.G. 1990 Coins of the Roman empire. (London). Christol M. 1977 Efforts de guerre et ateliers monétaires de la périphérie au IIIe siècle apr. J.-C. L’atelier de Cologne sous Valérien et Gallien. In Armées et fiscalité dans le monde antique. (Colloques nationaux du CNRS 936: Paris). Christol M. 1997a L’empire romain du IIIe siècle. Histoire politique (de 192 à 325). (Paris). Christol M. 1997b Les déplacements du college imperial de 256 à 258: Cologne, capitale impériale. Cahiers du centre Gustave-Glotz 8: 243-253. Domergue Cl. 1990 Les mines de la péninsule ibérique dans l’antiquité romaine. (Collection de l’école française de Rome 127, Rome). Duncan-Jones R. 1994 Money and government in the Roman empire. (Cambridge). Edmonson J.C. 1989 Mining in the later Roman empire and beyond: continuity or disruption? Journal of Roman Studies 79: 84-102. Estiot S., Amandry M. & Bompaire M. 1993 Le trésor de Sainte-Pallaye (Yonne): 8864 antoniniens de Valérien à Carin. Trésors monétaires 14: 39-124. Foucray B. 1998 Le dépôt de deniers et d’antoniniens de

Our last example concerns the reign of Aurelian. With his reform of the coinage in AD274, his 8 or 9 mints became quasi-permanent institutions, preparing the way for Diocletian’s reforms at the end of the same century (Watson 1999: 132-136). Their main task must have been to execute the general recall of old coin mentioned by Zosimus (Zosimus I.61.3). Aurelian also closed the mints of the Gallic usurpers in Trier and Cologne and a new one was opened in Lyons once more. Internal trouble, successive Germanic invasions and the weakened position in the northern parts of the empire made it impossible for the new and distant mint of Lyons to do its job in renewing the old pre-reform coinage. In contrast with Italy and the Balkans, where mints as Rome, Ticinum, Siscia and Serdica replaced much of the pre-Aurelianic coins with new antoniniani within a decade, spurious and old money continued circulating in the Gauls and the Germanies till the first decade of the C4th (Fig. 4; Christol 1997a: 178; Estiot et al1993: 44-46; Callu 1969: 344-348). In conclusion: after an initial phase under Augustus, where mints tended to be situated outside Rome for political and military reasons, state coinages were concentrated in Rome from the Flavians onwards. The creation of mints outside Rome in the C3rd was not only linked with the growth of 38

Johan van Heesch: Mints and the Roman army from Augustus to Diocletian

Berméricourt (Marne). Cahiers numismatiques 136: 1129. Giovannini A. 2000 L’or africain. Un aspect méconnu de l’économie antique et de l’impérialisme romain. In La guerre dans les économies antiques. (Économie antique. Entretiens d’archéologie det d’histoire, SaintBertrand-de-Comminges): 253-276. Göbl R. 1993 Die Münzprägung des Kaisers Aurelianus (270-275). Moneta imperii Romam 47. (Vienna). Halfmann H. 1986 Itinera principum. Geschichte und Typologie der Kaiserreisen im Römischen reich. Heidelberger Althistorische Beiträge und Epigraphische Studien 2. (Stuttgart). Harl K.W. 1996 Coinage in the Roman economy, 300 B.C. to A.D. 700. (Baltimore). Jones G. 1980 The Roman mines at Riotinto. Journal of Roman Studies 70: 146-165. Le Bohec Y. 1989 L’armée romaine sous le Haut-Empire. (Paris). Lehrberger G. 1995 The gold deposits of Europe: An overview of the possible metal sources for prehistoric gold objects. In G. Morteani & J.P. Northover (edd.) Prehistoric gold in Europe. Mines, metallurgy and manufacture. (NATO series. Series E: applied sciences 280, Dordrecht). Mattingly H. 1939 The great Dorchester hoard of 1936. Numismatic Chronicle (Ser. 5). 19: 21-61. Mattingly H. 1975 Coins of the Roman empire in the British Museum, V, Pertinax to Elagabalus. (2nd edit., London). Piccotini G. 1994 Gold und Kristall am Magdalensberg. Germania 72: 467-477. Spufford P. 1988 Money and its use in medieval Europe. (Cambridge). Strack P. 1933 Untersuchungen zur Römischen Reichsprägung des zweiten Jahrhunderts, II, Die Reichsprägung zur Zeit des Hadrian. (Stuttgart). Sutherland C.H.V. 1976 The emperor and the coinage. Julio-Claudian studies. (London). van Heesch J. 1998 De muntcirculatie tijdens de Romeinse tijd in het noordwesten van Gallia Belgica (Monografie van nationale archeologie 11, Brussels). Volk T.R. 1997 Hispania and the gold and silver coinage of Augustus. In La moneda en temps d’August (Barcelona): 59-90. Watson A. 1999 Aurelian and the third century. (London & New York). Wolters R. 1999a Nummi Signati. Untersuchungen zur römischen Münzprägung und Geldwirtschaft (Vestigia 49, München). Wolters R. 1999b Die Organisation der Münzprägung in iulisch-claudischer Zeit. Numismatische Zeitschrift 106/107: 75-90.

39

Fig. 1. The location of the Augustan legions, mints and main mining areas. Mints: 1. Ephesus; 2. Pergamum; 3. Samos; 4. „Caetra“-mint; 5. Corduba; 6. Caesaraugusta; 7. Emerita; 8. Lugdunum; 9. Roma; 10. Nemausus

Limes XVIII

40

Fig. 2. Possible mints under Valerianus and Gallienus (AD254 - 260/268): 1. Roma; 2. Viminacium; 3. Colonia Agrippina; 4. Mediolanum; 5. Siscia; 6. Sirmium; 7. Antiochia; 8 Samosate; 9. Emesa; 10. Smyrna

Johan van Heesch: Mints and the Roman army from Augustus to Diocletian

41

Limes XVIII

Fig. 3: Percentage of denarii and antoniniani in coin hoards ending between 253 and 268 found in north-west Gaul (civitates Nerviorum et Menapiorum: data: van Heesch 1998).

Fig. 4 Aurelian’s recall of old coin (AD274). Hoards ending between 279 and 285 (data: Estiot 1993).

42

Problems in the study of Roman and native Fraser Hunter In comparing Roman imports found in different areas, the bias caused by differences in find circumstances has rarely been addressed. This paper looks at the effects of this taphonomic bias on our perceptions of contacts with Rome. In particular it compares areas with a rich burial tradition with those which lack one. It is argued that the rich burials of ‘Free Germany’ have had something of a blinkering effect on the picture, and that assessment of the fragmentary settlement evidence is needed to redress our views of Roman-native interaction.

neighbouring Wielbark culture (Czarnecka 1992: 41), while areas which favoured cremation can only be compared with caution to those where inhumation was predominant because of the differing survival of grave goods. Some of these problems have of course been realised (eg. Eggers 1951: 30-31), but the implications have not always been followed through. Most difficult of all are areas where there is little or no tradition of burial at all, as in Scotland and Ireland (Raftery 1981; 1994: 188199; Whimster 1981; Welfare 1983). An on-going reappraisal of Scottish Iron age burials by the writer has identified only some 100 burials for the whole period from 700BC to AD400 – far fewer than many individual gravefields on the Continent! – and barely a quarter of these have any grave goods. The few burials with Roman finds (Hunter 1997: fig 12.3) would hardly merit a second glance in Continental cemeteries. Yet this need not be evidence of limited access to rich Roman items; rather I would argue it is the lack of a significant tradition of accompanied burials which creates this perception.

Introduction The study of Roman and native in areas like Scotland can be a frustrating one. The material often seems poor and fragmentary in comparison to the Continent, a few scraps of worn samian or sherds of broken glass in contrast to graves full of bronze or hoards rich with weapons. A stroll through the galleries of the Nationalmuseet in Copenhagen can rapidly turn into a depressing comparative exercise. This is also the picture that comes through in most published commentaries – Roman material beyond the frontier in Scotland is generally seen as a poor relation to the ‘real’ contact with barbaricum in ‘Free Germany’ (Wheeler 1954: 16; Fulford 1985: 102-3). But is this picture valid? In my contribution to the Limes Congress (which is published elsewhere; Hunter 2001) I argued that there was a complexity to the Scottish situation which has not been appreciated because the material comes primarily from settlement contexts. Here I would like to develop this point, as I think it is an important methodological hurdle in the wider study of Roman and native.

Two examples will show why Scotland and Ireland need to be viewed as part of the wider European picture. The first is the occurrence of denarius hoards beyond the frontier, as studied for Continental Europe by Lind (1981, especially maps 1-3) and Berger (1996). It is clear that the Scottish and Irish hoards are part of the same pattern, with a dominance of late C2nd hoards (Bateson 1971; Robertson 1978). The same is true of Hacksilber hoards in the later C4th – early C5th; the Scottish hoard from Traprain Law provides the best example of a pattern which stretches from Ireland to Poland and beyond (Curle 1923; Grünhagen 1954; Birley 1955). In these instances Scotland and Ireland received the same material as the rest of the western barbarian world. Was this due solely to particular historical events? Or do they indicate a broader picture of high-quality material being more widely available than the grave finds indicate?

Both Scotland and Ireland lie on the fringes of most scholars’ mental maps of barbaricum. They barely feature in Wheeler’s classic synthesis, and most Continental works have focussed on ‘Free Germany’ without integrating the British and Irish evidence (eg. Eggers 1951; Hedeager 1978). Hedeager’s significant and highly influential study of patterns of Roman contact in ‘Free Germany’ developed the concept of the buffer zone near the frontier and the richer zone beyond. Subsequently this model has been applied more widely, with Scotland and Ireland both seen as ‘buffer zone’ areas which received a range of mundane, everyday Roman material (eg. Fulford 1985: 102-3). This is contrasted with the more prestigious material found in northern Germany, Poland and Scandinavia, where it marks the graves of the richest layers of society. A key difficulty with this model is that it conflates material from different sources, since it compares areas dominated by rich grave finds with those (such as Scotland) where settlement and stray finds predominate. This creates an intrinsic bias: grave finds inevitably appear wealthier than the more fragmentary and intractable material from settlements, but the tradition of accompanied burial is not a constant across Europe and there is considerable diversity in burial rite and grave goods (eg. Todd 1987: 39-76). For instance some areas favour weapons burial while others do not, as in the Polish Przeworsk culture compared to the

My intention here is not to rewrite Hedeager’s model, which retains much validity; if graves alone are compared, it is clear that many of the burials in northern Germany and Denmark are indeed richer than equivalent burials near the limes. However, I would like to explore the methodological point that some of the patterning comes from differences in the use and deposition of Roman material, not differences in access to it. This should affect our views of how different areas of Europe interacted with Rome.

43

Limes XVIII

Table 1. Occurrence of vessel types in different contexts in eastern Germany, Westphalia and Scotland. Sources: Laser & Voß 1994; Laser & Schultze 1995; Voß 1998; Eggers 1951; Robertson 1970; Hunter 2001.

Total sites

Sites with vessels

bronze Sites with vessels

glass Sites with samian vessels

Eastern Germany Grave Settlement Hoard/stray Total

362 118 23 503

199 27 17 243

78 10 1 89

4 14 1 19

Westphalia Grave Settlement Hoard/stray Total

22 17 5 44

21 1 0 22

3 1 0 4

15 10 1 26

Scotland Grave Settlement Hoard/stray Total

9 163 21 193

0 7 18 25

5 39 1 45

2 82 1 85

data, not just rich grave finds, makes the picture more balanced: for instance, comparison with Eggers (1951) shows a marked increase in the proportion of finds from settlement sites in the areas studied.1

Blinded by burials - the effects of taphonomy If we compare the find circumstances of Roman imports across Europe, it is clear that some areas (notably those furthest from the limes) are dominated by grave finds, while others (nearer the frontier) are predominantly settlement or stray finds (Fig. 1; cf. Eggers 1951: Karte 68). Clearly a number of variables are at work: for instance, in Ireland and Scotland there is a lack of a burial tradition, while in Holland burials are generally poorly furnished. Yet the lack of settlement finds in other areas cannot reflect past reality. Presumably the people in these rich burials were wealthy in life as well as death, but rich assemblages from farmsteads in Scandinavia are few and far between (see Lund Hansen 1987). There are a few ‘ports of trade’ emerging, such as Gudme/Lundeborg in Denmark and Uppåkra in Sweden, which are producing Roman finds (Kromann 1994; Petersen 1994; Thomsen 1994; Silvegren 1999: 95-99; Stjernquist 1999: 68-70; Hårdh 2000), but very little work has yet been done on the distribution of this material through the settlement system. In part this arises from a predominance of burial excavations, in part from a lack of published data on settlement finds (for instance, reports in Acta Archaeologica tend to focus on architecture, spatial arrangement and phasing rather than artefacts); there are also specific problems with settlement finds which will be discussed below. The work of the Corpus der römischen Funde im europäischen Barbaricum (henceforth CRFB; eg. Laser & Voß 1994; Laser & Schultze 1995; Voß 1998) has highlighted these defects. Its systematic study of all the

What are the effects of this on our knowledge? I would argue this burial/settlement divide distorts our perception of Roman imports. This can be seen by comparing the occurrence of bronze, glass and samian vessels in graves, settlements and hoards in three areas: eastern Germany, Westphalia and Scotland. These were chosen because they have a reasonable quantity of settlement data. In each case the data reflect the presence or absence of each type in each find group, not absolute numbers of finds (Fig. 2; Table 1). The different areas show distinctive regional patterns. For instance, bronze vessels predominate over samian in both graves and settlements in eastern Germany, in contrast to the situation in Westphalia; this is much as Hedeager’s model indicates. However if we look at the differences between different site-types, a number of biases are clear. As might be expected, high-quality material such as bronze and glass vessels is less common in settlements than in graves: valued objects were less likely to be thrown away as rubbish. Yet presumably all these Roman imports were 1

Brandenburg-Berlin, increased from 15% to 34%. Freistaat Sachsen, 11% to 26%. Mecklenburg-Vorpommern, 0% to 14%. In calculating these figures each discovery was counted as one regardless of the quantity of finds involved.

44

Fraser Hunter: Problems in the study of Roman and native

not just kept for funerals but did see some use in life – or are we to suggest they were imported specifically as grave goods? Samian, by contrast, is markedly more common in settlements than burials: even in Sweden, where it is very rare and the grave record predominates, two of the three finds are from settlements (Helander 1997). In sum, settlement finds present a markedly different picture from grave finds because of the preferential curation or selection of material in different contexts. The result is that settlements appear poorer than burials, and thus settlementdominated areas will automatically seem poorer than burial-dominated areas even if the same range of material was in circulation.

material was later introductions. However it requires better stratigraphic data than we currently have to resolve it. The results of recent Dutch terp excavations will be crucial in this respect; the interim results on the evidence from Wijnaldum, for instance, offer important insights into the various possible mechanisms (papers in Besteman et al 1999). One area where burial-poor areas compensate with quality material is in hoards. In Scotland and Ireland, the bronze vessels which elsewhere appear in burials are found in hoards or as single finds in rivers, lochs and other contexts (Fig. 2; Hunter 1997: 117-9, fig 12.5; Bateson 1973: 66). I have argued elsewhere that such stray finds of goodquality intact objects are unlikely to be casual losses and are best seen as part of a wider tradition of votive hoards (Hunter 1997). Similar finds occur in parts of the Continent, but are minor compared to the grave finds. This pattern of prestige objects being used in different contexts in different areas is familiar from other periods (eg. the Bronze age: Bradley 1990: 99-101). It emphasises the dangers of over-reliance on the burial record – different areas were dealing with the same range of prestige goods in different ways.

This is compounded by a number of specific problems with settlement evidence. Apart from this selection of what is discarded or deposited, there are recognition difficulties with fragmentary material. While the quality of objects in a burial is readily recognised, the scraps from a settlement are far harder to appreciate. In Scotland, only recent detailed work has revealed the quality of the finds. The best example is glass, where Dominic Ingemark’s study has identified a much wider range of imports than was previously realised, much of it good-quality items comparable with the Scandinavian grave assemblages (eg. Ingemark 1998 & pers. comm.). An example is the painted cups (Isings 1957: type 85b), some of the most striking of Roman imports: while they are best known from rich burials (eg. Jesendorf; Voß 1998: taf 56), fragments are known from four Scottish settlement sites (Fig. 3). Once we make this mental leap from fragments to once-intact vessels, the differences between areas become less apparent. We have been blinded by the burials into overlooking the fragments.

These examples show the care which must be taken in interpreting finds of Roman material in barbaricum. It is crucial that we assess the biasing effects of the archaeological record when we are comparing different areas, and as far as possible compare like with like. This will be illustrated by looking in more detail at settlement finds. Using settlement data

Evaluation of settlement finds is further complicated by the question of when the material arrived on site. Roman objects are often found in later levels, and some scholars have suggested that many are later introductions, imported as curios or relics in the early Medieval period (eg. Alcock 1963; Alcock & Alcock 1987: 131; 1990: 115-16; Warner 1976: 285-88). Alternatively this could be explained as residuality: as an occupation site is reused and rebuilt, this causes older material to be disturbed into younger layers. It is worth noting two Scottish examples. The Roman fragments from C7th-C8th AD sites at Dunadd in Argyll and Buiston in Ayrshire have been seen as such ‘reliquary’ finds (Alcock 1981: 167; Stevenson 1955: 283), but in both cases recent excavations have identified Roman Iron age occupation (Lane & Campbell 2000; Crone 2000). This evidence was elusive, and it is likely that antiquarian excavators or small-scale trenching would have missed it – a warning against too ready an acceptance of an ‘absence’ of Roman period occupation. There are of course shades of interpretation in between: Roman items may have survived in use for several hundred years (as Scandinavian grave finds suggest, eg. Helander 1997; Nielsen 1988: 148-155) or been found on local Roman sites and kept as curios. This remains an active topic for debate, although in my view it is unlikely that any significant quantity of the

I have considered elsewhere the finds of Roman objects from Iron age settlements in Scotland, a corpus of over 150 individual sites (Hunter 2001). In this study I developed techniques to circumvent some of the biases outlined earlier. The results suggested there was a strong selectivity in both the material coming into native hands and the sites which had access to it. The finds were not everyday Roman material culture, but a highly selective subset which was dominated by two categories: material for feasting and drinking; and jewellery. Both of these were areas of social display in the pre-Roman Iron age, and I argued that the Roman material being adopted was that which was useful within native society. The distribution of finds on different sites suggests that access to the material was controlled in a hierarchical system, with groups on certain key sites having access to much richer material and passing a selection to others, such as dependents, via the social ties which bound the communities together. It is also clear that the picture varies in different areas, with some showing a higher degree of centralised control than others. Overall the picture from these Scottish finds is one of a highly selective use of Roman material, which acted as a form of prestige good in the local societies. This is similar to the use of Roman objects in the ‘rich’ areas of ‘Free 45

Limes XVIII

Germany’, with a dominance of feasting equipment in graves. Again this casts doubt on the idea that the area immediately beyond the frontier was simply a partromanized buffer zone using everyday Roman objects. Furthermore, if we compare the Scottish picture to the situation within Britannia, in northern England and north Wales, we find very different site assemblages. These show much more everyday pottery such as coarse wares and mortaria, while the Scottish sites are dominated by samian (eg. Evans 1999). A study of Roman Iron age roundhouses also suggests that everyday Roman items such as nails are far commoner within than beyond the province (R. Pope pers. comm.; Hunter 1998: 366-7). In other words, when the distortions in settlement data are allowed for, a much more complex picture emerges – and a picture which has similarities with much of the rest of northern Europe in the use of Roman items as status markers.

Conclusions This article has been intended to expose some of the underdiscussed methodological complexities in comparative studies of Roman-native interaction. It has not been the aim to propose new overarching models – Hedeager’s work still has much to commend it. There are clear regional differences in access to and use of Roman material, and the Danish and north German material is notably rich when the burials alone are compared. All areas beyond the limes did not have equal access to Roman material; its value and exoticness would be different in different regions. The picture also varied through time, while Erdrich’s work has shown some of the political complexity in the way the Romans dealt with different native groups (Erdrich 2000). However, the settlement data should be considered more carefully. It is very easy to be blinded by the visibly rich finds, in the offerings and burials which were always intended to create a spectacle. Yet the more subtle traces from the settlements conceal equally interesting patterns and patterns which should lead us to a different and more complex picture of Roman - native interaction. The ‘poor relations’ of barbaricum deserve a more prominent place at the feast.

In other areas there are also suggestions that the buffer zone idea may have been misled by the dominance of settlement data. The Dutch province of Frisia lies within the ‘buffer zone’. Here too the record is dominated by settlement finds (from terps) and stray finds. A full catalogue of this material is in preparation for CFRB, but from the existing data it is clear there is some complexity in the use of Roman artefacts. Samian is the dominant find category, being far more prevalent than other forms of pottery (Eggers 1951: 122-5; Galestin 1997): this is at marked variance with most Roman sites within the frontier. There is also a concentration of Roman statuettes unparalleled beyond the frontier, although Denmark too has a significant number (Eggers 1951: Karte 63; ZadoksJosephus Jitta et al 1967; Galestin 1994). Rather than using everyday Roman objects, it seems better to interpret this as a specific local response to Rome, with selection of particular aspects of Roman material culture.

Acknowledgements I am grateful to Dominic Ingemark and Michael Erdrich for much useful discussion on this topic, and to them, David Clarke and Andrew Heald for commenting on earlier drafts of this paper. Fig. 3 is by Alan Braby. Bibliography Alcock L. 1963 Dinas Powys. (Cardiff). Alcock L. 1981 Early historic fortifications in Scotland. In G. Guilbert (ed.) Hill-Fort Studies. (Leicester): 150-180. Alcock L. & Alcock E.A. 1987 Reconnaissance excavations on early historic fortifications and other royal sites in Scotland, 1974-84: 2, Excavations at Dunollie Castle, Oban, Argyll, 1978. Proceedings of the Society of Antiquaries of Scotland 117: 119-147. Alcock L. & Alcock E.A. 1990 Reconnaissance excavations on early historic fortifications and other royal sites in Scotland, 1974-84: 4, Excavations at Alt Clut, Clyde Rock, Strathclyde, 1974-75. Proceedings of the Society of Antiquaries of Scotland 120: 95-149. Bateson J.D. 1971 The finding of Roman silver coins in the vicinity of the Giant’s Causeway. Ulster Journal of Archaeology 34: 50-57. Bateson J.D. 1973 Roman material from Ireland: a reconsideration. Proceedings of the Royal Irish Academy (Sect. C) 73: 21-97. Bateson J.D. 1976 Further finds of Roman material from Ireland. Proceedings of the Royal Irish Academy (Sect. C) 76: 171-180. Berger F. 1996 Roman coins beyond the northern frontiers: some recent considerations. In C.E. King & D.G. Wigg (edd.) Coin finds and coin use in the Roman world

Ireland too can be viewed in a similar way. The material is sparser than in Scotland, and much of it is stray finds – Irish archaeologists have a continuing difficulty in identifying settlements to populate their Iron age (cf. Raftery 1994: 112-3). The Roman finds have inspired radically different interpretations from those advanced for Scotland, focussing on the question of a Roman presence in Ireland or invoking Roman refugees, travellers or merchants (Warner 1976; 1995; Raftery 1994: 200-219). Yet the character of the material is essentially similar to the Scottish finds (Bateson 1973; 1976): a dominance of fine ware and brooches, some stray finds (probably votive) of bronze vessels, and a scattering of silver – two late C2nd denarius hoards and two late C4th Hacksilber hoards. This material could readily be interpreted much as the Scottish finds have been: exotic artefacts used as status goods in local society, with the selection of material which was locally useful. Our interpretations to date have rather separated the Scottish and Irish material – they may be closer in concept than we realise.

46

Fraser Hunter: Problems in the study of Roman and native

(SFMA 10): 55-61. Besteman J.C, Bos J.M., Gerrets D.A., Heidinga H.A. & De Koning J. 1999 The excavations at Wijnaldum vol 1. (Rotterdam). Birley E. 1955 Review of Grünhagen 1954. Antiquity 29: 246. Bradley R. 1990 The passage of arms. (Cambridge). Crone A. 2000 The history of a Scottish lowland crannog: excavations at Buiston, Ayrshire 1989-90. (Edinburgh). Curle A.O. 1915 Account of excavations on Traprain Law in the parish of Prestonkirk, county of Haddington, in 1914. Proceedings of the Society of Antiquaries of Scotland 49: 139-202. Curle A.O. 1923 The Treasure of Traprain. (Glasgow). Czarnecka K. 1992 Two newly-found Roman swords from the Przeworsk culture cemetery in Oblin, Siedlce district, Poland. Journal of Roman Military Equipment Studies 3: 41-55. Eggers H.J. 1951 Der Römische Import im Freien Germanien. (Hamburg). Erdrich M. 2000 Rom und die Barbaren: Das Verhältnis zwischen dem Imperium Romanum und den Germanischen Stämmen vor seiner Nordwestgrenze von der späten römischen Republik bis zum Galischen Sonderreich. Römisch-Germanische Forschungen 58. (Mainz). Evans J. 1999 The Roman pottery. In D. Longley, N. Johnstone & J. Evans Excavations on two farms of the Romano-British period at Bryn Eryr and Bush Farm, Gwynedd. Britannia XXIX: 206-226 (185-246). Fulford M.G. 1985 Roman material in barbarian society c.200 B.C.– c.A.D. 400. In T.C. Champion & J.V.S. Megaw (edd.) Settlement and society: aspects of west European prehistory in the first millennium B. C. (Leicester): 91108. Galestin M.C. 1994 A new Mars from the Netherlands. In J. Ronke (ed.) Akten der 10. internationalen Tagung über antike Bronzen. (Stuttgart): 155-159. Galestin M.C. 1997 Romans and Frisians: analysis of the strategy of the Roman army in its connections across the frontier. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (ed.) Roman Frontier Studies. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 347-353. Grünhagen W. 1954 Der Schatzfund von Gross Bodungen. Römisch-Germanische Forschungen 21. (Mainz). Hamilton J.R.C. 1968 Excavations at Clickhimin, Shetland. (Edinburgh). Hårdh B. 2000 Uppåkra – a centre in south Sweden in the 1st millennium AD. Antiquity 74: 640-48. Hedeager L. 1978 A quantitative analysis of Roman imports in Europe north of the Limes (0-400 A.D.) and the question of Roman-Germanic exchange. In K. Kristiansen & C. Paludan-Müller (edd.) New directions in Scandinavian archaeology. (Copenhagen): 191-216. Helander A. 1997 Ett Terra Sigillata-kärl i Linköping – det hittills nordligaste fyndet. Fornvännen 92: 49-56. Hunter F. 1997 Iron age hoarding in Scotland and northern England. In A. Gwilt & C. Haselgrove (edd.) Reconstructing Iron age societies (Oxford): 108-133. Hunter F. 1998 Iron. In L. Main Excavation of a timber roundhouse and broch at the Fairy Knowe, Buchlyvie, Stirlingshire, 1975-8. Proceedings of the Society of Antiquaries of Scotland 128: 293-417 (356-367). Hunter F. 2001 Roman and native in Scotland: new approaches. Journal of Roman Archaeology 14. 289-309.

Ingemark D. 1998 Öl eller vin? Kulturkontakter och maktspel under romersk järnålder i Skottland, norra Northumberland och Skandinavien. Vetenskapssocietetens i Lund Årsbok 1997-1998: 1630. Isings C. 1957 Roman glass from dated finds. (Groningen). Kromann A. 1994 Gudme and Lundeborg – the coins. In P.O. Nielsen, K. Randsborg & H. Thrane (edd.) The archaeology of Gudme and Lundeborg. (Copenhagen): 64-67. Lane A. & Campbell E. 2000 Dunadd: an early Dalriadic capital. (Oxford). Laser R. & Schultze E. 1995 Corpus der römischen Funde im europäischen Barbaricum. Deutschland Band 2: Freistaat Sachsen. (Bonn). Laser R. & Voß H.-U. 1994 Corpus der römischen Funde im europäischen Barbaricum. Deutschland Band 1: Bundesländer Brandenburg und Berlin. (Bonn). Lind L. 1981 Roman denarii found in Sweden 2. Catalogue text. Stockholm Studies in Classical Archaeology 11:2. (Stockholm). Lund Hansen U. 1987 Römischen Import im Norden. (Copenhagen). Nielsen S. 1988 Roman denarii in Denmark – an archaeological approach. Nordisk Numismatisk Årsskrift 1987-1988: 147-169. Petersen P.V. 1994 Excavations at sites of treasure trove finds at Gudme. In P.O. Nielsen, K. Randsborg & H. Thrane (edd.) The archaeology of Gudme and Lundeborg. (Copenhagen): 30-40. Raftery B. 1981 Iron Age burials in Ireland. In D. Ó Corráin (ed.) Irish Antiquity. (Cork): 173-204. Raftery B. 1994 Pagan Celtic Ireland. (London). Robertson A. 1970 Roman finds from non-Roman sites in Scotland. Britannia 1: 198-226. Robertson A.S. 1978 The circulation of Roman coins in north Britain: the evidence of hoards and site-finds from Scotland. In R.A.G. Carson & C.M. Kraay (edd.) Scripta nummaria romana: essays presented to Humphrey Sutherland. (London): 186-216. Silvegren U.W. 1999 Mynten från Uppåkra. In B. Hårdh (ed.) Fynden i centrum: keramik, glas och metall från Uppåkra. Act Arch Lundensia Ser 8o 30. (Lund): 95112. Stevenson R.B.K. 1955 Pins and the chronology of brochs. Proceedings of the Prehistoric Society 21: 282-294. Stjernquist B. 1999 Glass from Uppåkra: a preliminary study of finds and problems. In B. Hårdh (ed.) Fynden i centrum: keramik, glas och metall från Uppåkra. Act Arch Lundensia Ser 8o 30. (Lund): 67-94. Thomsen P.O. 1994 Lundeborg – an early port of trade in south-east Funen. In P.O. Nielsen, K. Randsborg & H. Thrane (edd.) The archaeology of Gudme and Lundeborg. (Copenhagen): 23-29. Todd M. 1987 The northern barbarians 100 BC – AD 300. (Oxford). Voß H.-U. 1998 Corpus der römischen Funde im europäischen Barbaricum. Deutschland Band 3:Bundesland Mecklenburg-Vorpommern. (Bonn). Warner R.B. 1976 Some observations on the context and importation of exotic material in Ireland, from the first century BC to the second century AD. Proceedings of the Irish Academy (Sect. C) 76: 267-292. Warner R.B. 1995 Tuathal Techtmar: a myth or ancient literary evidence for a Roman invasion? Emania 13: 23-32. Welfare H. 1983 Review of Whimster 1981. Scottish

47

Limes XVIII

Archaeological Review 2/1: 75-79. Wheeler M. 1954 Rome beyond the imperial frontiers. (London). Whimster R. 1981 Burial practices in Iron age Britain. British Archaeological Reports Brit. Ser. 90. (Oxford). Zadoks-Josephus Jitta A.N., Peters W.J.T. & Es van W.A. 1967 Roman Bronze statuettes from the Netherlands I: Statuettes found north of the limes. (Groningen).

48

Fraser Hunter: Problems in the study of Roman and native

Fig. 1. Proportion of finds from graves and settlements across barbarian Europe, ranked by burial quantity. Each site is counted as one regardless of the quantity of finds; coins are excluded. Data from Eggers (1951), Robertson (1970), Hunter (2001) and Bateson (1973; 1976). While recent studies have updated parts of Eggers’ work, the general picture remains valid. Some areas with similar proportions have been combined to give at least 20 findspots in each area.

Fig. 2. Relative proportions of vessels of bronze, glass and Samian in different contexts. Calculated on a presence/absence for each find, i.e. the quantity of objects in a find is not considered. The vessel hoards category includes all multiple deposits, water or moor finds, and single stray finds of intact vessels. There is a single stray find of a complete Samian vessel from Westphalia, omitted from the plot because the sample is so small. For data see Table 1.

49

Limes XVIII

Fig. 3. Intact painted cup from a burial at Jesendorf (1; from Voß 1998: taf 37 & 56) and Scottish fragments from settlements at Traprain Law, East Lothian (2; Curle 1915: 108; NMS GV 41); Dunollie, Argyll (3; from Alcock & Alcock 1987: ill 9); and Clickhimin, Shetland (4; from Hamilton 1968: fig 62). A fragment from Scatness, Shetland, is not yet published. Drawn by Alan Braby.

50

Die Legionen des Augustus. Probleme der römischen Heeresgeschichte nach dem Ende des Bürgerkrieges: Die Truppengeschichte Galatiens und Moesiens bis in Tiberische Zeit und das Problem der Legiones Quintae Karl Strobel sich ohne Zweifel um eine außerordentliche Mission. Aus seinem prätorischen Rang ist deshalb weder darauf zu schließen, das er keine Legionen unter seinem Kommando gehabt habe noch daß er maximal eine Legion befehligt haben könnte. Zudem ist zu beachten, daß wir uns noch in der unmittelbaren Folgephase der Bürgerkriege und erst am Anfang der Entwicklung regelhafter Schemata befinden. Lollius war ein Mann des besonderen Vertrauens des Herrschers. Die Annexion des galatisch-lykaonischpisidischen Reiches des toten Amyntas, zu dem auch Pamphylien und Kilikia Tracheia gehörten, erfolgte rasch und ohne Widerstand. Die neue Ära der Provinz beginnt mit dem Jahr 25/4 (Herbst 25 - Herbst 24) v. Chr. Dabei hatte das Amyntas-Reich durchaus über beachtliche Streitkräfte und ein Netz von Festungen und festen Städten verfügt. Hinzu kam von Anfang an die Aufgabe, den offenen Widerstand in den südlichen Gebirgszonen des Reiches niederzuwerfen. Wir können mit gutem Grund davon ausgehen, daß für die Aktion eine erhebliche Truppenmacht aufgeboten wurde, wobei man wahrscheinlich auch auf Einheiten in Makedonien bzw. im Balkanraum zurückgreifen konnte, wo L. Licinius Crassus, der im Jahre 30 gemeinsam mit Octavian oder besser Caesar dem Sohn den Consulat bekleidet hat3, in den Jahren 29 und 28 entscheidende Erfolge gegen Bastarner, Thraker, Geten und Daker errungen und dafür am 4.7.27 v. Chr. seinen Triumph ex Thracia et Geteis gefeiert hatte. Damit war die römische Herrschaft bis zur unteren Donau vorgeschoben. Die nach römischem Vorbild formierten Kerntruppen des Amyntas wurden, wie mit gutem Grund anzunehmen ist, gemäß einer Absprache in das römische Heer übernommen. Ein Teil wurde als Legio XXII 24 v. Chr. nach Ägypten zur Verstärkung der dortigen Garnison geführt.4 Der Beiname Deiotariana ist für diese Legion, die in tiberischer Zeit mit dem Epitheton Cyrenaica erscheint5, erst seit flavischer Zeit belegt, dürfte ihr aber inoffiziell von an Anfang an zu eigen gewesen sein, da die Einheit auf die nach dem Vorbild der Legion formierte

Traditionell wurde die nach dem Tode des Amyntas, des letzten Tetrarchen aller Galater und Königs von Pisidien, 25/4 v. Chr. und der Annexion seines Reiches eingerichtete Großprovinz Galatia, welche alle Länder des Reiches des Amyntas umfaßte, als eine provincia inermis angesehen. Dagegen hat S. Mitchell 1976 die Auffassung vertreten, daß die Legio VII, die spätere Legio VII Claudia, die Legionsgarnison der Provinz Galatia gewesen sei.1 Wir können jedoch zu Recht davon ausgehen, daß diese Großprovinz, deren Annexion nicht ohne politischmilitärische Risiken war und in deren Süden bis 6 n. Chr. schwere Kriegshandlungen stattfanden, bis zur Errichtung der Provinz Cappadocia 17/18 n. Chr. eine größere Garnison beherbergt hat, zu der Legionsverbände in wechselnder Stärke gehörten (Strobel 2000: 516ff). Dies bleibt nicht ohne Auswirkungen auf die Heeresgeschichte der augusteischen Zeit nicht nur im Osten des Reiches. Zudem haben wir insbesondere für die Jahre 25 v. Chr. (Annexion des Amyntas-Reiches, Ende des Kantabrerkrieges), 20 (Armenien- und Parthermission des Tiberius) und 1 v.- 4 n. Chr. (Mission des C. Caesar) mit erheblichen Umgruppierungen, mit teilweise weitgespannten Truppenverlegungen und zeitweiligen Heereskonzentrationen zu rechnen. Im Jahre 25 war der mächtigste Vasallenkönig in Kleinasien, Amyntas, trotz erheblicher Anfangserfolge auf seinem Feldzug gegen die Homonadeis, der kriegerischen Gebirgsbevölkerung zwischen Ostpisidien und dem westlichen Isaurien, ums Leben gekommen. Anstelle einer dynastischen Nachfolge ordnete der damals noch in Spanien weilende Augustus die Annexion des Reiches und seine Umwandlung in eine römische Provinz an. Es war dies offensichtlich eine grundsätzliche politische Entscheidung, der das Konzept zugrunde lag, das zentrale Kleinasien bei den sich rechtlich dazu anbietenden Anlässen unter direkte römische Herrschaft zu stellen, eine Politik, die Tiberius in den ersten Jahren seiner Herrschaft mit der Annexion von Kappadokien (17 n. Chr.) und Kommagene (18 n. Chr.)2 zu einem Abschluß führte, der die römische Grenze zum oberen Euphrat vorgeschoben hatte und eine unmittelbare römische Aufmarschzone gegen die Parther im Norden Syriens schuf. Mit der Annexion des Amyntas-Reiches wurde M. Lollius, ein enger Vertrauter und Parteigänger des Augustus, als Legat beauftragt. Bei diesem Kommando des damals prätorischen Rang bekleidenden M. Lollius handelte es

3

Vgl. zu ihm PIR² L 186; zu seinen Donaukriegen zuletzt Ivanov 1997: 473-475. Vgl. zu der Legion auch Keppie 2000: 225-232. Die aus Kleinasien stammende bronzene Votivtafel des Markos Antonis Silbanos, Centurio der Legio XXII (Speidel 1992: 304f.) bezeugt einen von Antonius in eine seiner Legionen rekrutierten und dabei mit dem Bürgerrecht versehenen Kleinasiaten, der, wie mit guten Gründen anzunehmen ist, bei der Umformung der Kerntruppe des Amyntas zur Legio XXII in diese abkommandiert worden ist. Die Annahme von Schmitthenner 1958: 149f. die Legion könnte 31 v. Chr. zum Heer des Pinarius Scarpus gehört haben, ist unzutreffend. 5 ILS 2690. In der wahrscheinlich aus claudischer Zeit stammenden Inschrift AE 1976: 200 eines Tribunen in den Legionen III und XXII wird nur erstere mit dem Epitheton Cyrenaica offensichtlich zur Unterscheidung von der III Gallica benannt; letztere, bis zur Gründung der XXII Primigenia unter Caligula zudem ohne Doublette, .bleibt hier ohne Beinamen. 4

1

Mitchell 1976; gefolgt von Keppie 1984: 207f., allerdings irrtümlich mit dem Ansatz „30-20?”; 2000: 91f., 183. 2 Tacitus Anns. II.42 (Tod des angeklagten Königs Archelaos Herbst 17 in Rom und Einzug seines Reiches; Tod des kommagenischen Königs und folgende Unruhen 17 n. Chr.).56. Beide Maßnahmen wurden von Germanicus im Jahre 18 organisatorisch umgesetzt.

51

Limes XVIII

Kerntruppe des Galatertetrarchen und Königs Deiotaros zurückgeführt werden kann. Sehr auffallend ähnelt der Zusammensetzung der Legio XXII in mittelaugusteischer Zeit jene der Legio III Cyrenaica; auch diese Legion hat in jenen Jahren offensichtlich Kerntruppen des Königs Amyntas, aber auch kleinasiatische und galatische Rekruten peregriner Herkunft in großem Umfange erhalten.

an den erfolgreichen Operationen in Africa teilgenommen, für die der Proconsul L. Autronius Paetus am 16.8..28 einen Triumph feierte.11 Die Legio VII kann mit einigem Recht als eine weitere Garnisonstruppe der Provinz Galatia bereits unter Lollius angesehen werden, der sich bei der Annexion des Amyntasreiches sehr wahrscheinlich auf die beiden Legionen V Gallica und VII als Kern seines Heeresaufgebotes stützen konnte. Die Legion bildete nach den vorliegenden Zeugnissen bis zu ihrem endgültigen Abzug aus Galatien wohl um 14 v. Chr. die Garnisonslegion Galatiens12; ihre Verlegung war vermutlich durch die in diesem Jahr begonnene, beim Tod des Augustus abgebrochene Mission des Tiberius im mittleren Donauraum und ein entsprechendes Truppenrevierement veranlaßt gewesen. Ein weiterer möglicher Anlaß für den Transfer der Legion, der dann angesichts des wiederaufgenommenen Germanienkrieges perpetuiert wurde, könnte die Militärrevolte der pannonisch-illyrischen Legionen nach dem Tod des Augustus sein. Nach dem Jahre 14 war die Legion in Dalmatien stationiert, wo sie seit 16/17 n. Chr. sicher belegt ist. Ob eine Legion unter dem praetorischen Legaten Sex. Sotidius Strabo Libuscidianus, der in Galatien c.1315/16 n. Chr. amtierte13, als Ersatz für die abgezogene Legio VII in der Provinz stationiert wurde, ist unbekannt. Im Jahre 17, als Tiberius den kappadokischen König Archelaos nach Rom lockte und dort unter Anklage stellte, wobei die Absicht, sein Reich einzuziehen, eindeutig in Hintergrund stand, ist jedoch mit Sicherheit von einem größeren Truppenaufgebot in der Provinz Galatia auszugehen, welches die erfolgreiche Annexion Kappadokiens nach dem Tode des alten Königs im Herbst des Jahres 17 gewährleisten sollte. Zugleich stellten diese Truppen in Galatien zweifellos einen Teil der Heereskonzentrationen für die Orientmission des Germanicus im Jahre 18 dar. Da sich Cn. Piso im Jahre 18 weigerte, syrische Legionen nach Armenien zu bringen (Tacitus Anns. II.57.1), muß sich Germanicus bei seinem Zug nach Armenien auf die in Galatien wie Kappadokien bereitstehenden Truppen gestützt haben. Für seine Orientmission ist in den Jahren 17-19 n. Chr. insgesamt mit umfangreichen Truppenbewegungen im Osten des Reiches zu rechnen, denen dann 20 eine Neuverteilung der entsprechenden Verbände folgte.

Die Statthalterschaft des M. Lollius in Galatien ist 25/4-22 v. Chr. anzusetzen. Aus dem pisidischen Antiocheia, das nach der Errichtung der römischen Provinz Galatia als Colonia Caesarea Antiochia, neu gegründet wurde, kennen wir heute die Grabdenkmäler von 9 Veteranen der Legio V Gallica.6 Die Deduktion der Veteranenkolonie, ist mit einiger Sicherheit von M. Lollius nach der Einrichtung der Provinz 25/4 und vor dem Ende seiner Statthalterschaft, jedenfalls noch vor dem Abschluß der Binnenorganisation der Provinz 21/0 v. Chr. durchgeführt worden. Während L. Keppie noch zuletzt die Stationierung der V Gallica, die er mit der späteren V Macedonica gleichsetzen möchte, in der Provinz Galatia nicht als gesichert sehen und an der Legio VII als der eigentlichen galatischen Legionsgarnison festhalten möchte7, kann die Legio V Gallica als Teil der Garnison der Provinz Galatia 25 bis vermutlich 20 v. Chr. (s.u.) und wahrscheinliche Kerntruppe der Annexionsstreitmacht des M. Lollius als gesichert gelten. Die durch die Münzprägung bezeugte Deduzierung von Veteranen der Legio VII, der späteren VII Claudia p. f., ist offenkundig erst später, vermutlich unter Agrippa oder anläßlich der endgültigen Verlegung der Legion aus Galatien an die Donau 13/14 n. Chr. (s.u.) erfolgt. Die Legio VII, die seit 42 n. Chr. den Ehrentitel Claudia pia fidelis trug8, gehörte seit 58 v. Chr. zu den Legionen Caesars. 44 wurde die Legion von Octavian aus ihren in Kampanien angesiedelten Veteranen neu aufgestellt; nach der Heeresreorganisation nach dem Sieg bei Philippi, aus dem ihr nur selten geführter Beiname Macedonica resultierte9, gehörte sie zu den drei Stammlegionen Caesar des Sohnes des Göttlichen. Ihre Teilnahme an der Schlacht bei Actium dürfte wahrscheinlich sein. 30-25 v. Chr. ist der Aufenthaltsort der Legion im Westen des Reiches, wo ihre Veteranen zwischen 27 und 25 in 3 besonders privilegierten nordafrikanischen Kolonien deduziert wurden.10 Die Legion hat, wie anzunehmen ist, 30-29/28

Doch bereits zuvor müssen in der Provinz Galatia größere Truppenverbände in Einsatz gewesen sein, so in der Okkupationsphase neben der Legio V Gallica sehr wahrscheinlich auch die Legio VII und vermutlich noch Detachements des syrischen Heeres sowie bundesgenössische Aufgebote zumindest des Archelaos I.

6

Vgl. Strobel 2000: 520-522; Christol & Drew-Bear 1998; AE 1998: 1386-1389. 7 Keppie 2000: 91f., allerdings unter Betonung, daß die Anwesenheit der Legio V Gallica in augusteischer Zeit in irgendeinem Teil Kleinasiens nicht mehr in Frage zu stellen ist. 8 Vgl. zu ihr Strobel 2000: 526ff.; auch Keppie 1984: 207f.; 2000: 75, 87. 9 Bauinschrift der 10. Kohorte auf der Thrakischen Chersonnes unter Calpurnius Piso 13-11 v. Chr. (CIL III 7386); in Italien CIL X 1711; 4723; 8241; AE 1938: 141; die These, diesen Beinamen mit einem entsprechenden markanten Aufenthalt in Makedonien zu verbinden (angedeutet noch bei Keppie 2000: 87, wo dieser Legionsbeiname aber zugleich in seiner weiten Verbreitung (Anlässe: Pharsalos, Philippi, Triumphe ex Macedonia im Jahre 39 v. Chr.) aufgezeigt wird. 10 Vgl. Laporte 2000, dessen Ausführungen ebd. 556ff. zur Geschichte der Legion allerdings sehr problematisch sind. Die Vorbehalte bei Keppie 2000: 90 gegen die Identifizierung der Veteranendeduktion mit dieser

Legio VII erscheinen unbegründet. Veteranen der antonischen Legio VII wurden offensichtlich in der Colonia Iulia Gemellum Parium in Mysien angesiedelt (vgl. Ritterling 1924-25: 1615; Keppie 2000: 89). 11 Inscr.It. 13, 1, 87; PIR² A 1680. 12 Vgl. im einzelnen Strobel 2000: 518, 520ff., 526ff. 13 AE 1976: 653; 1978: 789; CIL VI 1237, 31544a; Vgl. Eck 1978: 569f.; Rémy 1989: 137f.

52

Karl Strobel: Die Legionen des Augustus von Kappadokien. Von den Nachfolgern des Lollius (Strobel 2000: 518ff ) kennen wir L. Calpurnius Piso, Consul 15 v. Chr., für die Jahre 14-13 v. Chr. Im Jahre 13 wurde er aus Pamphylien, das zu seinem Amtsbereich gehörte, auf den Balkan beordert, wo er als Legat des Augustus einen dreijährigen Krieg gegen die Rebellion in Thrakien führte, die zur Flucht des Rhoimetalkes, zur Verwüstung der thrakischen Chersones und zu Einfällen in das römische Gebiet geführt hatte. Für seinen Erfolg, der im Jahre 11 gefeiert wurde, erhielt Calpurnius Piso die Ornamenta Triumphalia. Bei Ausbruch der Krise im thrakischen Vasallenreich waren offensichtlich keine ausreichenden Reserven verfügbar, da 14/13 durch den Skordiskereinfall der große illyrische Krieg ausgelöst worden war, dessen Oberbefehl Tiberius 12 v. Chr. übernahm, und andererseits die Kräfte im Westen durch den beginnenden großen Germanenkrieg gebunden waren. Das Heer des Calpurnius Piso, der offenkundig seine Kräfte für Operationen im Taurus-Raum und gegen die noch nicht unterworfenen Homonadeis konzentriert hatte, was auf dem Seewege am raschesten zur Stabilisierung der Lage herbeizuführen. Die galatischen Kontingente konnten wohl erst nach dem siegreichen Ende des Illyrienkrieges des Tiberius im Jahre 10 nach Kleinasien zurückkehren. Neben der Legio VII, deren Detachement damals auf der verwüsteten thrakischen Chersones Baumaßnahmen durchführte (CIL III 7386), stand damals mit Sicherheit noch mindestens eine weitere, uns unbekannte Legion unter dem konsularen Kommando des Calpurnius Piso. Für Cornutus Aquila, der im Jahre 6 als Statthalter in Galatien belegt ist und die logistische Vorbereitung für eine großangelegte militärische Operation im Süden der Provinz abschloß, kennen wir den Zeitpunkt seines Konsulates und somit seinen exakten Rang nicht. Sein Nachfolger P. Sulpicius Quirinus, Consul 12 v. Chr., ist in den Jahren 5 bis c.3 v. Chr. anzusetzen; er führte den Krieg gegen die Homonadeis zu einem siegreichen Ende und erhielt die Ornamenta Triumphalia. Auch er hatte eine Streitmacht von sicher weit mehr als einer Legion unter seinem Kommando. Nach dieser Unterwerfung der kriegerischen Bergregionen wurde die Legionsgarnison Galatiens wohl für die Orientmission des C. Caesar auf ihre Stammlegion, die Legio VII, verringert. M. Servilius Nonianus führte die Provinz vor der Jahre 3 n. Chr. entsprechend als Prätorier. Dagegen ist mit M. Plautius Silvanus, Consul 2 v. Chr. zusammen mit Augustus selbst, nochmals ein Statthalter konsularischen Ranges in der Zeit c.5-7 n. Chr. belegt. Die Provinz hatte offensichtlich bei der Auflösung der Truppenkonzentration für C. Caesar 4/5 n. Chr. zusätzliche Legionstruppen erhalten, da sich der Ausbruch von Unruhen im isaurischen Raum wahrscheinlich schon abzeichnete. Mit dem verstärkten Provinzheer führte Plautius Silvanus 6 n. Chr. einen siegreichen Feldzug gegen die Erhebung der Isaurier durch. Im folgenden Jahr führte er dann, wie Calpurnius Piso 13 v. Chr., starke Truppenverbände ex transmarinis provinciis in die sich krisenhaft zuspitzenden Kämpfe gegen die pannonischen Aufständischen (Velleius II.112.46), darunter die gesamte, offenkundig als verfügbare Reserve angesehene Legionsstreitmacht seiner Provinz.

Das vereinigte Heer des A. Caecina Severus und des Plautius Silvanus umfaßte 5 Legionen, von denen 3 aus dem Osten herangeführt wurden, davon zumindest 2 (Legiones VII und ?), wenn nicht gar alle 3 damals aus dem Süden der Großprovinz Galatien. Die heute mögliche Rekonstruktion der Truppengeschichte der Provinz Galatia veranlaßt uns, für die Zeit bis zur Stabilisierung nach der Mission des Germanicus um 20 n. Chr. mit einer hohen Beweglichkeit der Legionen im Osten zu rechnen, während sie andererseits die Geschichte der Legionen VII (Claudia) und V Gallica in dieser Zeit zu klären vermag. Andererseits ist es kaum zu bezweifeln, daß die Veteranen der Legionen V und VIII, die Agrippa 15/14 v. Chr. in Berytos deduzierte, die Anwesenheit der Legionen V Macedonica und VIII Augusta ca. 20 bis 14 v. Chr. im Osten, genauer im syrischen Raum belegen, wo Veteranen dieser Legionen auch in Heliopolis (Baalbek) angesiedelt wurden.14 Die Legio V Macedonica dürfte im Zusammenhang der Reise des Augustus in den Osten 22 v. Chr. bzw. der Orientmission des Tiberius in den Osten gekommen sein. Tiberius führte 20 v. Chr. ein Heer durch Makedonien und Thrakien in den Osten (Suetonius Tib. 14.3; Horace Epist. I.3.3), für das mit größter Wahrscheinlichkeit die Legio V Macedonica aus ihrem bisherigen Wirkungsbereich auf dem Balkan herangezogen wurde. Ein Abzug der Legionen V Macedonica und VIII Augusta aus dem Osten ist, wie wir mit gutem Grund annehmen können, anläßlich des großen illyrischen Krieges 13 v. Chr. erfolgt. Die Geschichte der Legiones Quintae nach Actium Die Notwendigkeit, für die Provinz Galatia über längere Zeit eine Garnison unter Einschluß von Legionsverbänden annehmen zu müssen, und die Frage der Identifizierung der Legio V Gallica bringen erhebliche Konsequenzen für die Rekonstruktion der Truppengeschichte der augusteischen Zeit mit sich. Diese Konsequenzen beziehen sich jedoch in gleicher Weise auf die Rekonstruktion der Maßnahmen Octavians oder besser Caesar des Sohnes des Göttlichen nach dem Sieg bei Actium und auf die Rekonstruktion der Legionsliste in den Jahren 30-27 v. Chr. Ja die Konsequenzen reichen bis in die Truppengeschichte der caesarischen Zeit zurück. In der frühen Principatszeit kennen wir drei Legionen mit der Ordnungszahl V, die mit den folgenden Beinamen verbunden sind: V Alaudae, V Gallica und V Macedonica (Strobel 2000: 522ff). Für die mehrfach erwogene Gleichsetzung der V Gallica mit der V Macedonica15 können keine stichhaltigen Argumente vorgebracht werden. Die andere vermutete Identifizierung mit der 14 Vgl. CIL III 14165,6 (bei Beirut) noch mit Nennung des auf die caesarische Tradition weisenden Epithetons VIII Gallica; Strabo XVI.16.2, 19; Josephus Bell. Jud. II.5.1; RPC I, 650f.; Ritterling 1924-25: 1573, 1643f. (allerdings zu früher Ansatz für Berytos); Keppie 2000: 91, 183, 234. 15 Etwa Ritterling 1924-25: 1572; Schmitthenner 1958: 134; vgl. dazu Strobel 2000: 522 Anm. 50.51

53

Limes XVIII Legio V Alaudae16 ist mit Sicherheit abzulehnen. Diese Legion des Antonius war nach der Entlassung ihrer Veteranen und der Eingliederung von Soldaten anderer, aufgelöster Legionen 31 v. Chr. (s.u.) wohl 30 v. Chr. nach Spanien verlegt worden, ebenso wie die antonische Legio X Equestris, die 31 v. Chr. unter Zusammenlegung der im Dienst verbleibenden Mannschaften zweier Legionen als Legio X Gemina neu formiert worden war.17 Die Annahme, es habe in frühaugusteischer Zeit nur 2 Legionen mit dieser Ordnungsziffer gegeben, ist in keiner Weise begründet. Die Legio V Gallica (Strobel 2000: 522f. ) wurde 25/24 v. Chr. wahrscheinlich aus dem Balkanraum nach Kleinasien beordert, wo sie den Kern des Okkupationsheeres des M. Lollius für das Reich des Amyntas bildete, zu der mit großer Wahrscheinlichkeit, wie oben ausgeführt, auch die Legio VII gehörte. Die Annahme erscheint nicht unbegründet, daß die V Gallica 20 v. Chr. für die Operationen M. Lollius als Proconsul von Macedonia 20/19-18 v. Chr., der Rhoimetalkes zu unterstützen und die Besser niederzuwerfen hatte (Dio LIV.20.3), wiederum auf den Balkan zurückkehrte. Vermutlich als eine von diesem engen Vertrauten des Augustus hoch geschätzte Truppe wurde sie dann nach Gallien verlegt, wo Lollius seit der 2. Hälfte des Jahres 17 als Legat der Gallia Comata amtierte. Im Frühsommer 16 v. Chr. erlitt er eine schwere Niederlage gegen Sugambrer, Usipeten und Tenkterer, die als Clades Lolliana in die Überlieferung einging und Augustus veranlaßte, selbst nach Gallien zu eilen (Dio LIV.20.4-6; Velleius II.97.1). Bei dieser Niederlage ging der Adler einer Legio V verloren. Diese Legion können wir zu Recht mit der V Gallica identifizieren, die offenkundig nach den schweren erlittenen Verlusten nicht mehr ergänzt, sondern aufgelöst wurde, was durchaus nicht als eine Tilgung cum infamia zu verstehen sein dürfte. Zudem wurde 15/14 eine groß angelegte Veteranenentlassung durchgeführt. Durch die Eingliederung der Legio XXII Deiotariana war die Zahl der Legionen auf 29 angestiegen und wurde nun bei diesem Anlaß augenscheinlich wieder auf 28 zurückgenommen. Zu dieser Zeit stand die V Macedonica im Osten des Reiches, die V Alaudae, wie im Grunde immer zu vermuten war, noch in Spanien.

Punkten präzisiert worden sind.19 Gleiches gilt für die Maßnahmen zur Reorganisation des Heeres, die Augustus nach dem Sieg bei Actium am 2.9.31 und nach dem Fall von Alexandria am 1.8.30 v. Chr. durchgeführt hat. Die von Th. Mommsen in seinem Kommentar zu den Res Gestae Divi Augusti20 vorgelegte, in den Quellen nirgends belegte These, Augustus habe die Zahl seiner Legionen nach Actium auf 18 verringert, dann seien 6 n. Chr. die Legionen XIII-XX und nach dem Verlust der Legionen XVII-XIX 9 n. Chr. die Legionen XXI-XXII aufgestellt worden, wurde bereits von Ritterling21 mit vollem Recht verworfen, auch wenn diese von Y. Le Bohec unverständlicherweise wieder aufgegriffen worden ist.22 Ritterling nimmt nach der augusteischen Reform eine Zahl von 27-28 oder sogar mehr Legionen an, Schmitthenner rechnet zwischen 30 und 25 v. Chr. mit 27-28 Legionen, Keppie mit 28. Die Untersuchungen Ritterlings in seinem monumentalen RE-Artikel bilden weiterhin den Ausgangspunkt für jede Untersuchung der Geschichte der römischen Legionen nach 31/30 v. Chr. Dagegen sind die Beiträge in Le Bohec 2000 teilweise von sehr unterschiedlicher Qualität und bisweilen sogar ein Rückschritt hinter Schmitthenner und Ritterling. Die Legio V Gallica ist, wie bereits ausgeführt, seit 25/4 v. Chr. in der neu errichteten Provinz Galatia stationiert gewesen, wo ihre Veteranen ca. 22 v. Chr. in der Kolonie Antiochia in Pisidien deduziert wurden. Da die Veteranen sämtlich italischer Herkunft sind, stand die Legion nach Philippi zweifellos im Westen des Reiches und gehörte zu den Legionen Caesar des Sohnes des Göttlichen. Die Legion nahm offensichtlich an den Schlachten von Philippi und Actium nicht teil. Nach Philippi wurden die im Dienst gehaltenen Mannschaften der dort von Antonius und Caesar dem Sohn des Göttlichen eingesetzten Legionen nach der Entlassung der Veteranen in 11 Legionen neu organisiert, von denen Antonius 8 in den Osten mitnahm, während Caesar der Sohn mit drei, den verbliebenen Stammlegionen des Jahres 44, IIII Macedonica, VII (Macedonica, später Claudia, s.o.) und VIII Gallica (später Augusta)23, nach Italien zurückkehrte (Appian Bell. Civ. V.3.). Zu den Legionen des Antonius gehörten die alten caesarischen, nun neu formierten Legionen III Gallica, V Alaudae, VI Ferrata24, X Equestris und XII Antiqua25,

Entgegen meiner früheren Skepsis (Strobel 2000: 522, 523) sind für die Frühgeschichte der Legionen V Gallica und V (Macedonica) aufgrund der vorliegenden Informationen neue Rekonstruktionen möglich. Hierzu ist es notwendig, die Grundzüge der Heeresentwicklung der Triumviralzeit zu berücksichtigen, wie sie insbesondere von W. Schmitthenner in seiner leider ungedruckt und oft unberücksichtigt gebliebenen Oxforder Dissertation herausgearbeitet18 und zuletzt von L. Keppie in einigen

19 Vgl. Schmitthenner 1958; Keppie 2000 (bes. Mark Antony’s legions 75ff., 233ff., 249ff.); 1983; 1984: 114ff. zu 44-30 v. Chr.; 132ff. Heeresorganisation nach Philippi und Actium; 199ff. Truppenverzeichnisse; ferner Brunt 1971: 498ff, 502ff.; auch Hahn 1969. 20 Mommsen 1883: 68-76; die Berechnungen von Hardy 1920 und Cavaillac 1952 führten nicht weiter. 21 Ritterling 1924-1925: 1213-1238, bes. 1216f., 1225f.; auch Kubitschek 1924: 1209f.; vgl. Keppie 1984: 136; 2000, 161ff. 22 Le Bohec 1989a: 19, 25, 34, 178, 189 (widersprüchlich zu S. 34), der zudem annimmt, die Zahl sei 31 v. Chr. auf 15 verringert worden, dann hätten 30 v.- 6 n. Chr. 18 Legionen bestanden (Neuaufstellung der II, III, VIII Augusta). 23 Die Legio Martia war 42 auf See verloren gegangen. 24 Nach der Teilnahme an der Schlacht von Philippi ist auch der Beiname Macedonica durch die Inschrift ihres Tribunen in Ephesos belegt (ILS 8862).

16

Noch von Keppie 1984: 203 betont. Vgl. bereits Ritterling 1924-25: 1566, 1678f.; zuletzt Keppie 2000: 78, 84ff., 90, 251f.; dens. 1984: 206, 209. 18 Schmitthenner 1958: 8ff. Truppen 44 v. Chr.; 57ff. Reorganisation nach Philippi; 61f. Armeen im Westen in Jahre 42; Armeen vor Actium 107ff., 119ff.,126ff.; nach Actium 139ff: bisweilen zu korrigieren die Legionsliste 151f. 17

54

Karl Strobel: Die Legionen des Augustus ferner die für ihn 43 neu aufgestellte Legio IIII (später Scythica), das bewußte Gegenstück zu der zu Octavian abgefallenen Legio IIII Macedonica. Obwohl nicht genauer zu fassen, ist zu diesen Legionen sicherlich auch die Legio II zu zählen, die Caesar im Jahre 48 aufgestellt hatte und die 44 von Antonius aus Makedonien herbefohlen in der Folge treu zu ihm gestanden war. Unbekannt bleibt die letzte dieser 8 Legionen; die 44-43 auf Antonius’ Seite stehende Legio XXXV blieb aufgrund ihrer hohen Ziffer nach Philippi sicher nicht bestehen. In der kurz vor Actium in großem Umfange ausgemünzten Prägeserie seiner Legionsdenare (Sear 1998, bes. 229ff.) zählte Antonius dann in durchgehender Reihe die Legionsziffern I-XXIII, dürfte aber insgesamt mindestens 27 Legionen (23 davon bei Actium) unter Waffen gehabt haben (Vgl. jetzt Keppie 2000: 76), so daß mit mehreren Doppelbesetzungen zu rechnen sein dürfte. Caesar der Sohn hatte etwa 24 Legionen für den Kampf nach Griechenland übergesetzt.26

die Frage der octavianischen Legio XII Victrix, die sich in Philippi ausgezeichnet hatte und am Perusinischen Krieg teilnahm und der offensichtlich auf Caesar zurückgehenden Legio XII Antiqua des Antonius, da wir später nur eine Legio XII mit dem Beinamen Fulminata kennen. Während sich die Veteranen der octavianischen Legio XII in Italien, so in Ateste (Keppie 1983: 212f. Nr. 7.20), ansiedelten, wurden jene der antonischen Einheit in Patrai, der nun insbesondere mit Veteranen der X Equestris neu gegründeten Colonia Augusta Achaia Patrae31, deduziert, wo auf den Veteranengrabsteinen auch der Legionsname Fulminata erstmals bezeugt ist.32 Als Lösung erscheint naheliegend, daß die Legio XII Victrix nicht nach Actium gebracht worden war, sondern als Reserve im Westen verblieb und zwischen 30 und 27 v. Chr. nach einer großen Veteranenentlassung aufgelöst wurde. Somit ist die aus Caesars Aufstellung im Jahre 58 zurückgehende Legio XII Fulminata zu Recht zu den 31 nicht aufgelösten antonischen Legionen zu zählen33; die in der Inschrift eines Tribunen genannte Legio XII in Ac(h)aia34 weist auf einen zeitweiligen Aufenthalt dieser Legion nach Actium in der Provinz Achaia hin, der sicher zeitgleich mit der Amtszeit des L. Licinius Crassus 29-28/7 als Proconsul von Macedonia und Achaia ist. Durch das Weiterbestehen dieser Legionen waren die entsprechenden Legionsziffern nun doppelt besetzt.

Nach dem Sieg von Actium kapitulierten die 19 Legionen des Landheeres des Antonius nach entsprechenden Verhandlungen und Zusagen.27 Die altgedienten Soldaten der octavianischen wie der besiegten antonischen Legionen wurden entlassen und dann nach Italien transferiert; die Ansiedlung der ersteren erfolgte in Italien, letztere wurden in den Provinzen deduziert.28 Die im Dienst verbleibenden Mannschaften wurden auf die nicht aufgelösten Legionen verteilt, zu denen entgegen Cassius Dio (LI.3.1) auch die traditionsreichen caesarischen Legionen auf der Seite des Antonius gehörten, nämlich die Legionen III Gallica, V Alaudae und VI Ferrata. Die Legio X Equestris wurde unter Zusammenlegung mit einer weiteren Legion als X Gemina neu formiert.29 Weitere Neuformationen durch die Zusammenlegung der Mannschaften zweier Legionen waren die XIII Gemina und XIV Gemina. Caesar der Sohn hat damit wahrscheinlich drei seiner eigenen Legionen durch Zusammenlegung aufgelöst; die Mannschaften der aufgelösten antonischen Einheiten wurden dagegen auf die fortbestehenden octavianischen verteilt.30 Ebenso müssen Mannschaften octavianischer Legionen auf die weitergeführten antonischen, mit dem großen Prestige der caesarischen und philippischen Tradition ausgestatteten Legionen verteilt worden sein, nicht zuletzt um deren Loyalität zu sichern. Die unkompensierte Auflösung eigener, siegreicher Legionen unmittelbar nach Actium ist dagegen, wie wir mit guten Gründen annehmen können, nicht erfolgt. Problematisch ist in diesem Zusammenhang

Im Winter 30 mußte Caesar der Sohn wegen der Unzufriedenheit der nach Italien gebrachten Veteranen nach Brundisium eilen, wo er die Unruhe durch Geldgeschenke und die Anordnungen für die Ansiedlung der Veteranen beseitigte35 und zugleich die Verteilung der nunmehr unter seinen Fahnen stehenden Legionen ordnete (Orosius VI.19.14). Nach einem nur 30 tägigen Aufenthalt in Italien brach er nach Syrien auf. Für den Angriff auf Ägypten wurden nur eigene zuverlässige Legionen aufgeboten, während die verbliebenen antonischen im Jahre 30 auf die Provinzen verteilt wurden (Spanien: V, X; Syrien: III, VI; Achaia/Makedonien: IIII, XII). Wichtige Zusicherungen müssen auch dem Heer des Pinarius Scarpus, der mit 4 Legionen zur Deckung Ägyptens in der Kyrenaika stand und sich 31 von Antonius lossagte, gemacht worden sein; Cornelius Gallus konnte die Legionen im folgenden Jahr problemlos übernehmen und gegen Alexandria führen.36 Es ist mit gutem Grund 31

Vgl. Strabo VIII.7.5; Rizakis 1997, 1998; Keppie 2000: 83ff. 4 Veteranen; Rizakis 1998: 200ff. Nr. 151-154. Aus dem Gebiet von Naupaktos kommen 2 Grabsteine ihrer Veteranen (ebd. 302f. Nr. 368f.); dieses Gebiet wurde wahrscheinlich in augusteischen Zeit, wahrscheinlich 14 v. Chr. (RGDA XVI.1-2) zum Territorium der Kolonie geschlagen. Es ist m. E. durchaus möglich, daß im Rahmen der zweiten Deduzierungswelle 15/14 v. Chr. hier erneut Veteranen der XII Fulminata angesiedelt wurden. Eine frühe Deduzierung wird auch durch die Datierung der Grabsteine etwa ab mittelaugusteischer Zeit nahegelegt. 33 Vgl. Auch Keppie 1984: 209; 2000: 85f.; Schmitthenner 1958: 69 plädierte dagegen für die XII Victrix als Ursprung. Ritterling 1924-25: 1705, 1710 sieht alle Benennungen nur als unterschiedliche Bezeichnungen einer einzigen Legion. 34 CIL III 6097 = Rizakis 1998: 304 Nr. 370 (wahrscheinlich Patras). 35 Suetonius Aug. 17, 3; Dio LI.4.2 – 5.1; RGDA XVI.1f. 36 Dio LI.5.6; 9.1; Plutarch Ant. 69.1; 74.1; Orosius VI.19.15. Zum Problem der Identifizierung der dortigen Legionen vgl. Schmitthenner 32

25

Die Legion trug offensichtlich 43/42 auch den Namen XII Paterna; ihr Veteran wurde sehr wahrscheinlich nach Philippi in Parma deduziert (CIL XI 1058). 26 Plutarch Ant. 61.2 mit Orosius VI.19.8. Den Zahlen liegt offenkundig der schematische Ansatz einer Legion zu 5000 Mann zugrunde. 27 Plutarch Ant. 68.2; Dio LI.1.4; vgl. zum weiteren bes. Keppie 2000: 75ff., 81ff.; auch Schmitthenner 1958: 139ff. 28 Dio LI.3.1-4; Hygin. limit. const. 177, 8-13 L. betont die Gleichbehandlung der Veteranen des Lepidus und des Antonius mit den eigenen (milites colonos fecit, alios in Italia, alios in provinciis); auch Appian Bell. Civ. V.128. 29 Epitheta Gemina Equestris auf der Weihung ihrer Centurionen im Mars Ultor-Tempel (AE 1934: 152). 30 Entsprechend Dio LI.3.1. Daß Caesar der Sohn auch die eigenen im Dienst verbliebenen Mannschaften neu gruppierte, belegt Dio LI.3.2.

55

Limes XVIII

anzunehmen, daß nach der Entlassung ihrer Veteranen 30 v. Chr. zu gleichen Bedingungen wie jenen nach Actium die verbleibenden Soldaten in 2 Legionen zusammengefaßt wurden, also in den Legionen III (Cyrenaica) und IIII, wobei erstere sogar in Ägypten verblieb, während letztere mit größter Wahrscheinlichkeit schon in diesem Jahre nach Makedonien verlegt wurde (s.u.).37 Dort war eine militärische Initiative zur Wiederherstellung des Ansehens Roms und zur Durchsetzung der römischen Kontrolle bis zur Donau geplant, mit der Caesar der Sohn des Göttlichen seinen Mitconsul des Jahres 30, L. Licinius Crassus, beauftragte. Erst nach dem Fall von Alexandria und der Rückkehr Caesar des Sohnes 29 nach Rom sind, wofür alle Argumente sprechen, auch die 31 zur Sicherung und als Reserven im Westen verbliebenen Legionen in die Versorgung der Veteranen und die Neuordnung des Heeres einbezogen worden. Dabei können die Kämpfe in Gallien 30-28 v. Chr. durchaus als retardierendes Element gewirkt haben. In Spanien wurde C. Calvisius Sabinus für seine siegreichen Operationen zum Imperator ausgerufen und feierte am 26.5.28 seinen Triumph (PIR² C 352); hinzu kamen die Operationen in Africa, die der Proconsul Autronius Paetus (29/28 v. Chr.) erfolgreich abschloß (s.o.). Auf allen Kriegsschauplätzen waren Truppen gebunden. Wahrscheinlich wurde im Jahre 28 v. Chr. die Reorganisation des Legionsheeres abgeschlossen, so daß von einem fixierten Truppenstand und dessen fester Verteilung auszugehen ist, als Caesar der Sohn des Göttlichen am 13.1.27 v. Chr. im Senat erklärte, seine außerordentliche Gewalt niederzulegen und alle Provinzen mit ihren Heeren in die Hände von Senat und Volk zurückgeben zu wollen, worauf ihm das Imperium Proconsulare für 10 Jahre und die Provinzen Galliens, Syrien und Spaniens und natürlich Ägypten übertragen wurden (RGDA XXXIV; Dio LIII.9.6; 12.1-7; 13.1-2). Durch die Neuordnung wurde eine durchgehende Serie von Legionen mit den Ziffern I-XXI geschaffen, die in dieser Weise auch 25 v. Chr. bestand, als sie durch die Errichtung der Legio XXII (Deiotariana) erweitert worden ist. Zu einer Neuaufstellung von Legionen ist es nach dem Zeitraum 40-32 v. Chr. unter Augustus mit Ausnahme dieser sicher politisch bedingten Maßnahme entgegen zahlreichen Hypothesen nicht mehr gekommen, vielmehr wurde die Zahl der Legionen in mehreren Schritten, deren umfangreichster unmittelbar nach Actium durchgeführt wurde, bis zum Jahre 28 v. Chr. auf 28 Legionen gesenkt. 24-16 v. Chr. standen dann 29 Legionen unter Waffen, nach der Auflösung der V Gallica 16 v.-9 n. Chr. wieder 28. Nach dem Verlust von 3 Legionen im Jahre 9 blieb die Zahl bei 25 Legionen.

Zu der durchgehenden Reihe der Ordnungsziffern der eigenen Legionen (I-XXI) kamen Doppelt- (IIII, VI, X) und Dreifachbesetzungen (III, V), die auf die historischpolitischen Rahmenbedingungen der Neuordnung Caesar des Sohnes des Göttlichen zurückzuführen sind. Zum einen wurde die Reihe um die antonischen Legionen mit caesarischer Tradition ergänzt, deren Fortbestand nach Actium zugesichert worden war: III Gallica, V Alaudae, VI Ferrata, V Equestris bzw. Gemina. Die XII Fulminata trat wohl erst nachträglich allien an die Stelle der Doublette mit der octavianischen Legio XII. Die Legionen III Cyrenaica und IIII Scythica waren aus dem Heere des Pinarius Scarpus zu übernehmen. Zu klären bleiben die beiden offenkundig octavianischen Legiones V, die neben der V Alaudae weiterbestanden. Im Jahre 43 wurde eine der neu aufgestellten konsularischen Legionen beim Abmarsch des Pansa in Rom zur Deckung der Stadt zurückgelassen, wo sie auf Octavians Seite übertrat, als dieser am 19.8.43 Rom besetzte.38 Diese Legion können wir zu Recht mit der Legio V Urbana identifizieren, deren Veteranen nach Actium in Ateste deduziert wurden.39 Die Legion kann sehr wohl an der Schlacht bei Actium und dem anschließenden Zug gegen Ägypten teilgenommen haben. Jedenfalls ist diese Legio V wie die Legio VIII Augusta 16/14 v. Chr. im Osten stationiert gewesen, wo ihre Veteranen von Agrippa in Berytos und Heliopolis angesiedelt wurden (s.o.). Der durch den Skordiskereinfall von 14/13 ausgelöste große illyrisch-pannonische Krieg war, wie mit gutem Grund zu schließen ist, der Anlaß, beide Legionen in den Donauraum zu führen. Es ist zu erwägen, ob eine der beiden..Einheiten zu dem 13 v. Chr. gegen die Thraker geführten Heer des L. Calpurnius Piso gehörte, mit dem dieser 14/13 eigentlich den Wiederstand der südgalatischen Bergvölker brechen sollte, was dann aber aufgrund der Konzentration auf die Kriegsschauplätze an Rhein und Donau bis 7 v. Chr. (Triumph des Tiberius) zurückgestellt wurde. Die Legio V Macedonia gehörte seit der Errichtung des eigenständigen moesischen Militärbezirkes im Rahmen der Provinz Macedonia spätestens 10/9 v. Chr. zu dessen Garnison (Strobel 2000: 523ff.). Der zuerst nur selten belegte Beiname Macedonica wird auf diesen ständigen Aufenthalt zurückzuführen sein, kann aber auch bereits auf eine Teilnahme an den Donaukriegen des L. Licinius Crassus zurückgehen. Die Veteranen der Legio V Alaudae waren von Caesar offensichtlich 45 v. Chr. relativ geschlossen in Italien angesiedelt worden, wo sie nach der Ermordung Caesars in enge Beziehung zu Antonius traten, der ihre Centurionen, aber auch Mannschaftsgrade neben den Centurionen anderer angesiedelter Veteranenlegionen für die

1958: 128ff.; verfehlt die Vermutung bei Keppie 2000: 80 Anm. 54, auch die später als Deiotariana bezeichnete Legio XXII könnte zu seinem Heer gehört haben. Als weitere Legionen unter Pinarius Scarpus sind die antonische Legio VIII (Legionsdenare des Scarpus; Sear 1998: 238f.) und vermutlich die Legio XVIII Lybica, die in der Serie der Legionsdenare mit Beinamen erscheint (vgl. Schmitthenner 1958: 149).. 37 Vgl. zur Legion jetzt Speidel 1998; dens, 2000: 327ff . Auf die Rolle der Legion bein Übergang zu Caesar dem Sohn dürfte die ehrende Verleihung des Capricon-Emblems (dazu Speidel 1998: 221 mit Abb. 11, 1) zurückgehen.

38 Appian Bell.Civ. III.91-92. Nicht überzeugend zur Legion Ritterling 1924-25: 1587. 39 Vgl. Crawford 1989; Buchi 1993: 65ff.; zu den einzelnen Inschriften Bassignano 1997; Keppie 1983; 2000: 249f. Nr. 1-4, 258; für die Identifizierung mit der V Macedonica auch Keppie 1984: 203, 207. Eine Aufstellung erst 41-40 ist wenig wahrscheinlich. Ein Centurio der V Urbana ist in Oriculum bezeugt (AE 1996: 591).

56

Karl Strobel: Die Legionen des Augustus Wiederaufstellung der dritten Richtercenturie heranziehen wollte, wogegen Cicero am 2.9.44 polemisierte und dabei sicherlich auf die gallische Herkunft der Truppe zielte.40 Ende 44 hatte Antonius die Legion aus den Veteranen neu aufgestellt; sie kam in Mutinensischen Krieg zum Einsatz.41

44/43 von Lepidus aus einem Kern neu Veteranen aufgestellt worden sein, der die Zahl seiner Legionen von 4 Einheiten Ende 44, wie Appian Bell. Civ. III.46 belegt, bis Mai 43 sein Heer auf 7 Legionen erhöht hatte.44 Die Legion ist aller Wahrscheinlichkeit 42 mit Lepidus im Westen verblieben und erst bei der Ausschaltung des Lepidus im Jahre 36 in Sizilien zu Caesar den Sohn des Göttlichen übergetreten45, wobei wohl ihre Fortexistenz garantiert wurde. Die Truppe blieb im Westen, vermutlich in Gallien stationiert und war somit bereits vor Actium eine Zifferndoublette im Heere Caesar des Sohnes des Göttlichen.

Offen ist nun die Frage der Legio V Gallica. Caesar hatte im Jahre 58 die Legionen VII, VIII, IX und X zur Hand, noch im gleichen Jahr stellte er die XI und XII auf; 57 kam die XIII, 53 die XIV und XV hinzu, wobei letztere 50 an Pompeius über geben wurde.52 kam die Legio VI neu zum Einsatz und aus peregrinen gallischen Aufgeboten dieses Jahres wurde die Legio Alaudae formiert (Suetonius Caes. 24; Pliny NH IX.121), die ihre Ordnungsziffer V sicher frühestens mit der Bürgerrechtsverleihung an ihre Angehörigen und der damit gegebenen Erhebung zu einer vollwertigen Legio während des Bürgerkrieges erhalten hat (cf. Ritterling 1924-1925: 1564). Es wurde nun zuletzt von L. Keppie zu Recht betont, daß Caesar die Legionsziffern des traditionellen consularen Aufgebotes (I-IV) erst im Jahre 48 in seinem Heer aufgefüllt hat. Es stellt sich aber die Frage, warum Caesar 53/52 neben der neu erscheinenden Legio nicht auch eine Legio V formiert haben sollte, um an die traditionellen consularen Legionszahlen anzuschließen. Am 19.12.50 verfügte er jedenfalls nach der Abgabe von zwei Legionen an Pompeius über 11 Legionen, wie Cicero belegt.42 Die Legionen VI-XIV und die Alaudae ergeben jedoch nur 10 Einheiten. Es ist naheliegend, daß Caesar damals über eine Legio V verfügte. Es stellt sich somit die Frage, welche Legion unter der veterana legio quinta zu verstehen ist, die 46 mit Bravour in Africa zum Einsatz kam und bisher immer mit der V Alaudae identifiziert wurde.43 Die sehr wahrscheinlich auf das Geschichtswerk des Asinius Pollio zurückzuführende Nachricht Appian Bell. Civ. II.96 (Dazu Bell. Afr. 84) über die 5. Legion, die bei Thapsus sich freiwillig gegen die Kriegselefanten gestellt und dafür das von ihr „bis heute” getragene Emblem des Elefanten erhalten hatte, kann sich sehr wohl auf diese, nach 44 erneut aufgestellte und unter der Bezeichnung V Gallica bekannte Legion beziehen. Diese bestand zum Zeitpunkt der Abfassung des Geschichtswerkes noch (Siehe Horace Carm. II.1). Die Legion kann nach dem Sieg von Thapsus wie andere Legionen aufgelöst und ihre Veteranen deduziert worden sein. Die Legionsziffer V kann damit 46 auf die Legio Alaudae übertragen worden sein und dabei mit der Legio VI an Caesars consulare Serie bis zur Veteranendeduktion nach Munda angeschlossen haben.

Die Truppenkörper des frühaugusteischen Heeres (2725 v. Chr.) Die Legionen lassen sich somit nach dem heutigen Kenntnisstand wie folgt zusammenstellen, wobei die Details hier nicht alle diskutiert werden können: Legio I Macedonica46 (später Germanica): in der konsularen Serie des Jahres 43 aufgestellt, von Octavian nach Mutina übernommen, unter Licinius Crassus an der Donaufront, dann Spanien; Legio II Gallica47, dann Augusta: ebenfalls in der konsularen Serie des Jahres 43 als Legio II Sabina aufgestellt48, 40 offensichtlich im Heere des Fufius Calenus49 in Gallien50, nach Actium in Spanien; 44 Lepidus stellte die Legionen VI und X wieder auf, wie Appian Bell. Civ. III. 83; Cicero ad. Fam. X.18 (11), 2, wo Munatius Plancus an Cicero erwähnt, daß die decima legio veterana „mit den übrigen” (cum reliquis) wieder aufgestellt worden sei. Letzteres muß sich auf mehr als eine Legion beziehen. Vgl. auch Schmitthenner 1958: 35f. 45 Appian Bell. Civ. V.123-126. Lepidus hatte nach dem Anschluß der 8 Legionen des Sex. Pompeius nun 22 Legionen unter seinen Fahnen, hatte also 14 Legionen von Africa nach Sizilien übergesetzt. 46 Der Beiname ist nun durch den Grabstein des L. Licinius L. f. Aem(ilia tribu) aus Dyrrhachium belegt (AE 1994: 1563). Die Legion kämpfte 36 im Krieg gegen Sex. Pompeius (Appian Bell. Civ. V.112). Der Veteran aus Dyrrhachium wurde offenkundig als Peregriner von L. Licinius Crassus während seiner Donaukriege 29/27 rekrutiert, in denen die Legion auch das ehrende Epitheton erhalten haben dürfte. Aus einer antonischen Legion muß der ebenfalls als Peregriner von Ventidius Bassus wohl anläßlich des Partherkrieges rekrutierte Veteran P. Ventidius P. f. (AE 1994: 1562, Dyrrhachium) in die Legion gelangt sein. Seine Entlassung und Ansiedlung gehört wahrscheinlich zu der Deduktionswelle von 16/14 v. Chr. (anders Keppie 2000: 86f.). In Dyrrhachium hatte Augustus Bürger der 30 aus Italien vertriebene Gemeinden, die für Antonius Partei ergriffen hatten, angesiedelt (Dio LI.4.6), darunter sicher auch nach Philippi angesiedelte Veteranen des Antonius; die Ansiedlung der beiden Veteranen ist jedoch entgegen Keppie 2000: 87 kaum mit den Maßnahmen nach Actium zu verbinden. Unrichtig zur Legion Ritterling 1376. 47 AE 1952: 44; Arausio, Gründung der Colonia 35 v. Chr. 48 ILS 2227, Venafrum; C. Aclutius Gallus war zuerst Militärtribun der Legio Prima, dann der Legio Secunda Sabina, beide Posten sind sicher den Legionen Pansas zuzuweisen. Vgl. auch Keppie 2000: 123ff., bes. 125, 253. 49 Parteigänger des Antonius, 41-40 erst Statthalter der Gallia Narbonensis, dann auch der Gallia Comata mit einem Heer von 11 Legionen, das nach seinem Tode vor dem Vertrag von Brundisium Caesar dem Sohn des Göttlichen ausgehändigt wurde (Appian Bell. Civ. XXIV.51). 50 Die Legion kann sicher nicht auf die 48 v. Caesar aufgestellte Legio II zurückgehen, die seit ihrer Verlegung aus Makedonien nach Italien 44 v. Chr. treu zu Antonius gestanden war; diese Legion hat Antonius aller Wahrscheinlichkeit nach in den Osten mitgenommen, wie bereits oben ausgeführt. Mit den Veteranen seiner Legio II, die damals bereits den auf

Die Legio V Gallica wird mit einiger Wahrscheinlichkeit 40 Cicero Phil. I.20; Schmitthenner 1958: 9; Keppie 1984: 115, 132 schließen zu Unrecht, daß die Legion 44 noch im Dienst gewesen sei. Für die Neuformierung einer Rittercenturie kamen nur im näheren italischem Umland lebende entpflichtete Soldaten in Frage. 41 Cicero ad Att. XVI.9 (8), 2; Appian Bell. Civ. III.46. 2; Cicero ad Fam. X.31 (33), 4; 34.1. 42 Cicero ad Att. VII.7.6; dieses Zeugnis wird von Keppie 1984 nicht diskutiert. Vgl. ansonsten zur Entwicklung der caesarischen Legionen ebd. 80ff., bes. 97, 100, 102 (ohne Berücksichtigung einer 11. Legion). 43 So Ritterling 1924-1925: 1565; Franke 2000: 39.

57

Limes XVIII

Legio III Augusta: ebenfalls in der konsularen Serie des Jahres 43 aufgestellt; 42 im Westen verblieben, vielleicht seit 36, wohl sicher aber seit 32 in Africa stationiert51; Legio III Cyrenaica: von Antonius zwischen 41 und 32, wahrscheinlich 41/40 aufgestellt, 31/30 im Heer des Pinarius Scarpus in der Kyrenaika, seit 30 in Ägypten; Legio III Gallica: 48 von Caesar aufgestellt, sehr wahrscheinlich 48-43 Teil der Heere in Gallien, 45 in der Schlacht von Munda eingesetzt, 42 Teilnahme an der Schlacht von Philippi; seit Philippi mit Antonius im Osten, wo sich die Einheit im Partherkrieg auszeichnet (Tacitus Hist.III.24.2; Plutarch Ant.42.4); wohl seit 30 in Syrien stationiert (Dabrowa 2000: 309ff.).; Legio IIII Macedonica: Aufstellung durch Caesar 48 v. Chr., 47-44 in Makedonien, seit 44 auf Seiten Octavians; nach Actium in Spanien (Gómez-Pantoja 2000: 105ff.; Cerdán 2000: 609ff.); Legio IIII (Scythica): 43 von P. Bagiennus für Antonius in Ligurien bzw. Nordwestitalien formiert52; 31/30 sehr wahrscheinlich im Heer des Pinarius Scarpus in der Kyrenaika, 30 auf den Balkan verlegt; Legio V Alaudae, von Caesar 52 aus peregrinen Galliern formiert; Wiederaufstellung 44 durch Antonius, zu dessen Stammlegionen sie gehört, seit 30 in Spanien; Legio V Gallica: wahrscheinlich 53/52 von Caesar aufgestellt, von Lepidus 43 wieder errichtet, 36 zu Caesar dem Sohn des Göttlichen übergetreten, nach 30 vermutlich im Balkanraum, 25/4-20 in Galatien, dann wahrscheinlich Makedonien und Thrakien, 18/17 nach Gallien, schwere Verluste in der Clades Lolliana und dann 16 v. Chr. 16 aufgelöst; Legio V (zuerst Urbana, dann Macedonica): Teil der Aushebungen Pansas, 43 zu Octavian übergetreten, vermutlich 30 nach Makedonien, 22/20-15/14 im Osten, seit 14/3 an der unteren Donau stehend; Legio VI Ferrata: von Caesar 53/2 aufgestellt, 44/3 von Lepidus wieder aufgestellt, seit 43 im Heere des Antonius, vermutlich seit 30 in Syrien; Legio VI (Hispaniensis, später Victrix): von Caesar dem Sohn des Göttlichen 41 aufgestellt; seit 30 in Spanien; Legio VII (Macedonica, dann Claudia): Caesarische Stammlegion, Oktober 44 von Octavian aus ihren Veteranen wieder aufgestellt, nach 30 in Afrika eingesetzt, 25/4 v.-13/14 n. Chr. in Galatien; Legio VIII Augusta: Caesarische Stammlegion, Oktober 44 von Octavian aus ihren Veteranen wieder aufgestellt, nach 30 vielleicht kurzzeitig in Africa53,

c.22/20-15/4 im Osten belegt, seit 14/13 auf dem Balkan; Legio VIIII Hispaniensis (später Hispana): 43 von P. Ventidius Bassus im Picenum aufgestellt und nach Mutina mit Antonius vereinigt54; die Legion gehörte wohl zu den beiden, 42 von Antonius zur Deckung Italiens zurückgelassenen und nach Philippi Caesar dem Sohn des Göttlichen überlassenen Legionen55; nach Actium Epitheton Actiaca 30-14/13 in Spanien; Legio X Fretensis: von Caesar dem Sohn wohl 41 aufgestellt56, Epitheton für die Teilnahme am Krieg gegen Sex. Pompeius (CIL X 3890; Ritterling 1924-1925: 1671), 15/14 unter L. Tarius Rufus dem Proconsul von Makedonien und propaetorischen Legaten des Augustus in Makedonien belegt57, dann in Syrien stationiert; Legio X Gemina: nach Actium durch Zusammenlegung mit einer octavianischen Legion aus der caesarischen X Equestris formiert, seit 30 in Spanien (Gómez-Pantoja 2000: 169ff.): Legio XI (Claudia): Aufstellung durch Caesar den Sohn des Göttlichen vor dem Perusinischen Krieg, an dem sie im Heere des Salvidienus teilnahm; für die Beteiligung an der Schlacht von Actium Epitheton Actiaca getragen, dann Stationierung in Illyricum; Legio XII Fulminata: von Caesar 58 aufgestellt, Wiedererrichtung 44/43, seit 41 mit Antonius im Osten, nach Actium zuerst kurzzeitig in Achaia belegt, schließlich wohl seit der Mission des C. Caesar in Syrien stehend; Legio XIII Gemina: nach Actium durch die Zusammenlegung zweier octavianischer Legionen aus dessen Legio XIII gebildet, Stationierung in Illyricum; Legio XIIII Gemina: nach Actium durch die Zusammenlegung zweier octavianischer Legionen aus dessen Legio XIII (während des Perusinischen Krieges aufgestellt)58 gebildet, Stationierung in Illyricum; Legio XV Apollinaris: Aufstellung durch Octavian wahrscheinlich 40 während des Perusinischen Krieges59, nach Actium in Gallien, Einsatz im Kantabrerkrieg (AE 54 Cicero ad Fam. X.30 (31), 4. Ventidius Bassus hatte zuerst in den Veteranenkolonien Caesars 2 Legionen aufgestellt, dann in seiner Heimat Picenum eine weitere Legion (Appian Bell. Civ. III.66); bei der Wahl der Legionsziffern VII, VIII, die wir sicher mit den beiden ersten Legionen verbinden können, hat Ventidius Bassus offenkundige Gegenstücke zu den von Octavian im Oktober 43 aus caesarischen Veteranen in Kampanien wiederaufgestellten Legionen VII und VIII schaffen wollen; zudem hat er einen Teil der Veteranen der legionen VII und VIII für sich gewinnen können (Cicero ad Fam. XI.11 (10), 3; Phil. X.22; XI.37). Die Ziffer VIIII war im Frühjahr 43 nicht besetzt. Die Annahme Schmitthenners 1958, 41, er hätte auch im Picenum Veteranen aufgerufen, also der caesarischen Legio IX, bleibt unbegründet. Anfang März stand Bassus mit seinen 3 Legionen nahe Ancona (Cicero Phil. XII. 23) und stieß dann zur ligurischen Küste durch. 55 Appian Bell. Civ. IV.3. Die Möglichkeit, daß die Legion auf die caesarische, 46/5 aufgelösten Legio IX zurückzuführen ist, kann entgegen Keppie 2000: 201, 253 ausgeschlossen werden. 56 Er verfügte bei Ausbruch des Perusinischen Krieges über 12 Legionen (Appian Bell. Civ. V.24; bei ihm selbst 4, bei Salvidienus 6, je eine in Sardinien und Süditalien anzunehmen; vgl. Schmitthenner 1958: 41), die wohl kontinuierlich gezählt wurden (I - XII Victrix). 57 AE 1936: 18; Dio LIV.20.3 mit Groag 1932: 2321f. 58 Vgl. zur Aufrüstung während der Auseinandersetzung mit L. Caesar: Appian Bell. Civ. V.33, 35; Schmitthenner 1958: 67. Die Legio XIII kämpfte gegen Sex. Pompeius (Appian Bell. Civ. V.87). 59 Der Grabstein eines Veteranen aus Cremona belegt entgegen Keppie 2000: 66 keinen früheren Ursprung.

ihre Stationierung seit 43/2 in Gallien und wohl auf dortige militärische Bewährung hinweisenden Beinamen Gallica trug, gründete Caesar der Sohn des Göttlichen 35 v. Chr. nach dem Sieg über Sex. Pompeius und der Ausschaltung des Lepidus die Colonia Firma Iulia Secundarum Arausio (Pliny NH III.36; CIL XII 3203). 51 Le Bohec 1989b: 335ff.; 2000: 373ff. nimmt ohne Grund an, daß die Legion zur Armee des Lepidus gehört habe. 52 Ein Veteran aus dieser Rekrutierung ist offensichtlich in der Grabinschrift aus Piozzo in Ligurien (Colonia Augusta Bagiennorum) belegt (AE 1996: 679; Menella 1996; Keppie 2000: 257). 53 ILAfr 414, 471,472; vgl. Ritterling 1924-25: 1646f.; Reddé 2000: 119ff. (nicht immer überzeugend zur Frühgeschichte der Legion).

58

Karl Strobel: Die Legionen des Augustus 1954: 252; vgl. Wheeler 2000: 306ff.), dann in Illyricum stationiert; Legio XVI Gallica: Aufstellung wahrscheinlich 43, als Garnison in Gallien verblieben; Legiones XVII-XIX: Aufstellung wahrscheinlich 40 durch Caesar den Sohn (Keppie 2000: 161ff.), nach 30 vermutlich in Gallien (Wiegels in Le Bohec 2000: 74ff ); Legio XX (Valeria Victrix): Aufstellung wahrscheinlich 40 durch Caesar den Sohn60, nach Actium zuerst vielleicht in Makedonien, 15 in Gallien oder im Alpenraum61, dann in Illyricum62 und Moesien63; ab 9 n. Chr. am Rhein stationiert, wohl kurzzeitig auch in Spanien im Einsatz64; Legio XXI Rapax: Aufstellung wahrscheinlich 40 durch Caesar den Sohn, nach 30 in Gallien. Die frühe Donaulinie

Truppengeschichte

der

thrakisch-getischen Raumes möglich gemacht. Die versuchte Einmischung dakischer Fürsten in das Geschehen der Bürgerkriege auf der Seite des Antonius sowie die bereits von Caesar gesehene Bedrohung67 der römischen Position auf dem Balkan ließen Caesar den Sohn nach Actium den Plan fassen68, eine gesicherte römische Kontrolle im Vorfeld der Provinz Macedonia bis zur unteren Donau herzustellen. Die maiestas des römischen Volkes hatte gegenüber den dortigen Völkern durch die Bürgerkriege 44-31 v. Chr. stark gelitten; demonstrative Aktionen mußten hier wie in anderen Regionen des Reiches als notwendig erscheinen. Auch der thrakische Raum war durch mangelnde Stabilität gekennzeichnet. Die Statthalter des Antonius in Makedonien hatten zweimal für ihre militärischen Erfolge Triumphe erlangt.69 Mit der Aufgabe, die Situation im Vorfeld Macedoniens nachhaltig im Sinne der Errichtung einer effektiven römischen Kontrolle zu bereinigen70, entsandte Caesar der Sohn des Göttlichen, seinen Mitconsul des Jahres 30, L. Licinius Crassus71, der seinen Konsulat zum 30.6.30 v. Chr. niederlegte, als Proconsul von Macedonia und Achaia und damit als Kommandeur des in Makedonien und Achaia stehenden Heeres (Legiones IIII Scythica, V Gallica, V Urbana, XII Fulminata, X Fretensis?).

moesischen

Bald nach der Ermordung Caesars 44 v. Chr war auch der Dakerkönig Burebista, der eine kurzlebige Herrschaft über die getischen und dakischen Stämme aufgebaut und eine Machtexpansion im Karpatenbecken wie an der unteren Donau betrieben hatte, durch eine innere Revolte beseitigt worden; der von dakischen und getischen Völkerschaften bewohnte Raum zwischen Theiß und Tyras (Dnjestr) zerfiel nun in vier politisch-herrschaftliche Einheiten, deren Zahl sich in mittelaugusteischer Zeit auf fünf erhöht hatte, wobei das dakische Gebiet in Südwestsiebenbürgen un Sarmizegetusa nur einen dieser Komplexe bildete.65 Damit war auch die Stabilität der Region verloren gegangen66 und auch das Eingreifen bastarnischer Heeresund Wanderzüge in die inneren Auseinandersetzungen des

Wohl 31, nicht erst 30 v. Chr., wie meist angenommen, hatte ein bastarnischer Heeres- und Wanderzug die Donau überschritten, die im mittleren Moesien lebenden Stämme sowie die nördlichen Teile der Dardaner besiegt, und war dann aber 30/29 über das Haemus-Gebirge bis in das Gebiet der Dentheleten, die an die römische Provinz grenzten, vorgedrungen. Crassus hatte seine Provinz sicher noch im Spätsommer 30 übernommen, eine Eröffnung militärischer Operationen noch in diesem Jahre war damit kaum mehr nicht zu erwarten. Der Vorstoß der Bastarner und der Hilferuf des Königs der Dentheleten boten dann in der 1. Hälfte des Jahres 29 die Notwendigkeit und den Anlaß zum sofortigen Eingreifen an.72 Crassus hatte mindestens 4 Legionen unter seinem Kommando. 29 schlug er die Bastarner, die völlig aufgerieben wurden, und die mit ihnen verbündete Daker vernichtend. Dann begann er mit der Unterwerfung der moesischen Stämme zwischen

60 Das Epitheton Sicil(iaca), das nunmehr als gesichert gelten darf (AE 1988: 396; auch CIL IX 1625; 1608) verweist auf ihre Teilnahme am Krieg gegen Sex. Pompeius; damit ist die Legionsziffer XX für Caesar den Sohn belegt (anders noch Keppie 1984: 202); Veteranen der Legion wurden nach dem Sieg über Sex. Pompeius oder nach Actium in Benevent angesiedelt; vgl. jetzt Keppie 2000: 250f., der früher (1984: 143) für die Legionen XX und XXI eine Aufstellung erst vor 27 erwogen hat. 61 Nach dem Alpenfeldzug wurden Angehörige des Stammes der Trumpilini in die Legion rekrutiert (CIL III 7452; V 4923) rekrutiert; vgl. Ritterling 1924-25: 1770f.; Polaschek, 1939: 707-710. 62 Vgl. Brusin 1992: Nr. 2800-2801. 6 n. Chr. unter Tiberius im dalmatisch-pannonischen Aufstand, ab 9 n. Chr. in Germanien kämpfend (Velleius II.112; Tacitus Anns. I.42). 63 Die Grabinschrift eines Angehörigen der Legion, der ca. 15 v. Chr. aus dem damals zum Territorium von Brixia geschlagenen Stammesgebiet rekrutiert worden sein dürfte und mit 17 Dienstjahren verstarb (CIL III 7452 = ILS 2270 = ILBulg 179), ist etwa 3/4 n. Chr. zu datieren. Ihr Fundort ist das untere Oescus-Tal. 64 Der Spanienaufenthalt ist umstritten, erscheint aber doch belegt; vgl. Ritterling 1924-25: 1769f.; Le Roux 1982: 59f.; Perea Yébenes 2000, 581ff.; Keppie 2000b: 26; Cerdan & Marcos 2000: 589ff. 65 Strabo VII.3.5; 11.13; vgl. Strobel 1987; 1998; Lica 2000: 62ff., 93ff., gegen dessen teilweise problematische Thesen die Besprechungen durch den Verf. in Klio und SCIVA. Licas Rekonstruktion sowohl zur Geschichte des Burebistas wie zur Zeit nach dessen Ende sind in mehreren grundlegenden Punkten zu korrigieren. 66 Dio LI.22.8; Die Thesen rumänischer Forscher über das Weiterbestehen einer dakischen Zentralstaatlichkeit nach dem Ende des Burebista o. ä. sind zurückzuweisen. Das Burebista-Reich kann zudem nicht mit den Maßstäben eines im Inneren ausgebauten hellenistischen Staates betrachtet werden.

67

Vgl. Vergil Georg. II.497; Horace Sat. II.6.53; Carm. III.6.13-16. Nach Appian Illyr. XXII hatte Caesar der Sohn den Krieg gegen Daker und Bastarner schon während seiner dalmaisch-pannonischen Feldzüge 35-33 v. Chr. beschlossen. 69 L. Marcius Censorinus (42-40) und Asinius Pollio (40/39), beide im Jahre 39. 70 Bleicken 2000: 311ff. geht von einer weitgehenden Eigeninitiative des Crassus aus, räumt jedoch die Möglichkeit ein, daß Caesar der Sohn des Göttlichen seine Operationen angeregt haben könne; daß Crassus weitgehend selbständig gehandelt hat, soll nicht in Abrede gestellt werden. Innenpolitisch war der herausragende militärische Erfolg des Crassus sicher zu einem Problem geworden. 71 Horace Carm. III.8.18; Dio LI.23.2-27, 3; Livy Per. 134,135; Florus II.26; vgl. ILS 8810 (Huldigung der Athener an den Proconsul nach seiner vom Senat nicht anerkannten imperatorischen Akklamation durch das Heer nach dem Sieg über die Bastarner); Groag 1926: 270- 285; Stein 1940: 10ff.; PIR² L 186; Patsch 1932 (mehrfach überholt); Mócsy 1966; Ivanov 1997: 473-475; Bleicken 2000: 310ff., 723f. 72 Die Feldzüge waren nur durch ein Winterlager unterbrochen, fallen also ausschließlich in die Jahre 29 und 28 v. Chr. 68

59

Limes XVIII

Haemus und Donau. Schon im Jahre 29 wurden gefangene Daker in Rom bei den Gladiatorenspielen anläßlich der Einweihung des Tempels des Divus Iulius (18.8.29) vorgeführt. Die Vernichtung des Heeres des dakischen Teilkönigs Cotiso, die Horace (Carm. III.8.18) anspricht, kann dabei nicht sicher dem Jahre 29 oder 28 zugewiesen werden. Im Frühling 28 zerschlug Crassus einen erneuten Vorstoß von Dakern und bisher noch nicht südlich der Donau auftetretenen bastarnischen Gruppen über die vereiste Donau, die wiederum bis in das obere StrymonGebiet vorgedrungen waren.73 Bastarner und Daker wurden zum Frieden und zur Annahme der von Crassus diktierten Bedingungen gezwungen. Anschließend wurden in Zentralthrakien Serden, Meden und Bessen sowie die übrigen Gebiete mit Ausnahme der verbündeten Odrysen unterworfen. Eine nochmalige Erhebung der Moesier wurde niedergeschlagen. Das Eingreifen in eine innergetische Auseinandersetzung in der Dobrudscha und in Nordostbulgarien brachte Crassus weitere Erfolge und die Unterwerfung des westpontischen Städte unter die Oberherrschaft Roms, die sich nun über das gesamte Gebiet zwischen Margus (Morava), Donau und Schwarzmeerküste erstreckte, ohne daß man aber eine direkte römische Beherrschung dieser Gebiete eingerichtet hat. Nach seiner Abberufung feierte Crassus am 4.7.27 einen Triumph ex Thraecia et Geteis. Die Niederlegung der Spolia Opima im Tempel des Iupiter Feretrius hat Augustus aber verhindert. Der Friede mit den Bastarnern wurde durch eine Gesandtschaft zu Augustus bestätigt und diese in ein Vertragsverhältnis zu Rom gebracht.74

notwendige Administration und Kontrolle des jeweiligen Gebietes ausgestattet gewesen sein mußte, wurde entsprechend seit 27 von einem Legaten des Augustus geführt.76 Eine Vereinigung der Funktionen des Proconsuls und des Legatus Augusti in einer Person war nach dem Jahre 23, als Augustus das Imperium Proconsulare Maius erhalten hatte, schließlich überhaupt kein Problem mehr. Doch war die Konstituierung eines formal eigenständigen Militärbezirkes mit administrativer Zuständigkeit für das zugehörige Territorium außerhalb der Provinz Macedonia vorgezeichnet und nach einer Vorverlegung der Legionen über die Provinzgrenze hinaus auch notwendig. Dieser Distrikt erhielt den Namen Moesia. Die Maßnahme ist mit großer Wahrscheinlichkeit am Ende des des großen illyrischen Krieges 10/9 v. Chr. erfolgt (s. auch u.), als die Besetzung der Donaulinie vom südlichen Pannonien bis ins westliche Untermoesien organisiert werden mußte. Im Jahre 12 war jedenfalls C. Poppaeus Sabinus Statthalter der Provinz Moesia geworden, eine Funktion, in der ihn Tiberius bis zum Jahre 35 beließ, der ihm im Jahre 15 noch die Provinzen Macedonia und Achaia übertrug, die nun vom moesischen Legaten mitverwaltet wurden (Tacitus Anns. I.80.1; VI.39.3). Die Einrichtung der Provinz Moesia ist, wie wir mit guten Gründen annehmen können, parallel zu der Teilung und Neuorganisation des Illyricum nach dem Ende des großen dalmatisch-pannonischen Aufstand 9 n. Chr. angegangen worden. Eben Poppaeus Sabinus dürfte mit ihrer Organisation beauftragt gewesen sein.77 Thrakien blieb einschließlich der Gebiete des östlichen Untermoesien unter der Kontrolle von thrakischen Vasallenkönigen. Die Zuständigkeit für die westpontischen Küstenstädte lag beim moesischen Legaten.

Bei der Aufteilung der Provinzen zwischen der Zuständigkeit des Senates und der auf 10 Jahre festgelegten Verfügungsgewalt des Augustus (imperium proconsulare) wurden die Provinzen Macedonia und Achaia mit Epirus (Vgl. auch Errington 1999: 732-739 mit Karte) dem Senat übergeben. Damit stellte sich aber für Augustus das Problem, daß damit auch das Kommando über die starken, hier stationierten Truppenteile an den Senat gefallen wäre.75 Dies war jedoch gerade nicht die Absicht der Neuordnung. Wir können deshalb zu Recht annehmen, daß bereits im Jahre 27 für die in Makedonien und vielleicht auch in der bundesgenössischen Randzone der Provinz stehenden Legionen und Auxilien ein eigenes militärisches Kommando für den exercitius in Macedonia außerhalb der Ebene der regulären Provinzadministration geschaffen wurde, ein System, das wir ähnlich in Numidien und bei den germanischen Heeresbezirken sowie zu Beginn auch für Illyricum kennen. Dieses militärische Oberkommando, das mit einer Zuständigkeit für die

Etwa 15/14 kämpfte L. Tarius Rufus, Consul suffectus 16 v. Chr. siegreich gegen die Sarmaten (Dio LIV.20.3; dazu Groag 1932: 2321f.). Das Geschehen ist sicherlich in der Dobrudscha zu lokalisieren. Die damals unter seinem Kommando stehende Legio X Fretensis baute eine Brücke bei Amphipolis (AE 1936: 18). In dieser Bauinschrift erscheint Tarius Rufus als Legatus pro praetore des Augustus. Wir können seine Stellung als die des Legaten des „exercitus in Macedonia” erkennen. Die Bautätigkeit der Legion im Namen des Augustus und in seinem Namen bei Amphipolis widerspricht dem nicht. Der Skordiskereinfall von 14/13 v. Chr., an dem sich aus die Dentheleten beteiligten, löste den großen illyrischpannonische Krieg aus78, dessen Leitung nach dem ersten 76 Zur kontroversen Debatte um die Anfänge des römischen Moesien vgl. Ivanov 1997: 475ff.; Strobel 2000: 523ff.; Filow 1906: 1ff.; Fluss 1932: 2370ff., bes. 2372; Stein 1940: 9ff.; Syme 1971: 40-70, bes. 67ff.; Mócsy 1974: 32ff.; Kos 1977; Wachtel 1977; Papazoglou 1979; Burian 2000: 328-332 mit Karte. 77 Dies erklärt die Aussage Appian Illyr. XXX, die Tiberius die Errichtung einer Provinzherrschaft in Moesien zuschreibt. Bereits der Vorgänger des Poppaeus Sabinus, Aelius Catus (s.u.), verwaltete 7-11/12 neben dem Militärdistrikt auch Makedonien. 78 Dio LIV.20.3; RGDA XXX. Vgl. Zu den augusteischen Feldzügen nach 15 v. Chr. Suetonius Aug. 21, 1; Stein 1940: 12ff.; Mócsy 1974: 34ff.; Kienast 1982: 301f.; Gruen 1996: 171ff., bes. 175f.; Strobel 1998: 61ff., bes. 76ff.; auch Syme 1995: 242ff.

73 Dio LI.25.2-3; Florus II.28. Die Bastarner waren nach ihren schweren Verlusten im Vorjahr entgegen der Darstellung bei Dio mit Sicherheit nicht die einzigen, wahrscheinlich nicht einmal die hauptsächlichen Gegner des Crassus. Die Siege gegen die Daker wurden in der Überlieferung offenkundig unterdrückt, die eine Tendenz zur Schmälerung der Erfolge des Crassus zeigt. 74 RGDA XXXI Gesandte der Bastarner, der Skythen und der Könige der Sarmaten die dies- und jenseits des Tanais wohnen. 75 Vgl. auch Kienast 1982: 73f.; Bleicken 2000: 326f., der allerdings das Problem des Heeres in Makedonien nicht diskutiert.

60

Karl Strobel: Die Legionen des Augustus Gegenschlag, der von M. Vinicius in Illyricum/Pannonien geführt wurde, zuerst Agrippa übernehmen sollte, führte dann Tiberius 12-10/9 siegreich, wofür er wohl am 16.1.9 v. Chr. die Ovatio ex Pannonia feierte. Der römische Machtbereich bzw. der auch durch diplomatische Maßnahmen geschaffene, von Rom beanspruchte Raum war nun auch in Illyricum bis zur Donau ausgedehnt. Die Operationen der ersten Phase hatten das Skordiskergebiet und das Flußsystem der Morava fest in römische Hand gebracht. Bereits im Jahre 12 v. Chr. leisteten die Skordisker Tiberius loyal die bundesgenössische Waffenhilfe, zu der sie verpflichtet worden waren. Bereits 13/10 v. Chr. ist mit der Errichtung einer römischen Militärbasis in dem strategisch wichtigen Raum von Singidunum zu rechnen; die archäologischen Kenntnisse zur Frühgeschichte des römischen Belgrad sind allerdings bis heute völlig unzureichend geblieben. Nachdem gegen Ende des Jahres 10 bereits die dritte Schließung des IanusTempels beschlossen worden war, mußte diese dann aber unterbleiben, da die Daker im Winter 10 v. Chr. die Donau auf dem Eis überschritten und im südostpannonischen wie wohl auch im obermoesischen Raum Plünderungszüge unternommen hatten.79 Als Antwort darauf sind die Operationen des Cn. Cornelius Lentulus (Augur)80 zu sehen, der c.10/9-6 v. Chr. als Legat des moesischen Heeres seine Feldzüge unternahm und dabei über die Donau vorstieß; für seine Erfolge erhielt er die Ornamenta Triumphalia. Cornelius Lentulus sicherte im Osten die Donaulinie gegen die Sarmaten, deren Einfallswege über Moldavien und Ostmuntenien verliefen, zum anderen Führte er einen Feldzug nach Dakien hinein durch, dessen Operationen das Banat, Südwest- und Westsiebenbürgen betroffen haben. Damals enden, wie eine kritische Bestandsaufnahme des in der rumänischen Forschung bisher meist einseitig auf die Dakerkriege Traians hin interpretierten archäologischen Materials ergibt, wichtige befestigte Zentren im Banat und am mittleren Mureş-Lauf wie auch die imposante Festung von Piatra Craivii nahe dem späteren Apulum. Der südwestdakische Raum mit seinen politischen Formationen, darunter auch das Reich von Sarmizegethusa, das vielleicht gerade in der Folge dieser Ereignisse im 1. Jh. n. Chr. einen Aufschwung nehmen konnte, wurde damals durch Vertragsverhältnisse in das von Rom indirekt kontrollierte Vorfeld der Donaulinie eingebunden. Die zentrale Basis für die Operationen des Cornelius Lentulus in Dakien müssen wir im Bereich der Morava-Mündung sehen, über deren Talsystem die rückwärtigen Verbindungen nach Makedonien liefen. Zugleich war das Gebiet der MoravaMündung das geographisch vorgegebene Einfallstor nach Obermoesien und eben in dieses Flußsystem. Dabei war der traditionelle Übergang etwas östlich der MoravaMündung gegeben, da sich westlich davon ausgedehnte versumpfte Flußauen bis in das Vorfeld von Singidunum erstreckten. Dieser traditionelle Übergang ist ohne Zweifel

bei Viminacium zu suchen, wo wir bereits seit 10/9 eine römische Sperrstellung und zugleich einer Ausgangsbasis für römische Vorstöße über die Donau voraussetzen müssen, von der aus die traditionellen Kommunikationslinien durch das Banat verliefen. Die zweite stategisch und naturräumlich vorgegebene Basisstellung ist in Oescus gegeben gewesen. Auch ihre Besetzung bereits unter Cornelius Lentulus wird im Grunde zwingend nahegelegt. Problematisch ist die zeitliche Einordnung der weit ausgreifenden Operationen des M. Vinicius, der aber offenkundig von Pannonien aus im nördlicheren Raum des Karpatenbeckens gegen die Kotiner und Anarten, keltische, in der materiellen Kultur dakisch beeinflußte Völkerschaften in Nordungarn, der Südostslowakei und im oberen Theißgebiet operierte.81 Ferner kämpfte er gegen die Daker, deren Siedlungsgebiete im Nordwesten, wahrscheinlich um Somesch und Krisch lagen und die von Verbündeten jenseits des Karpatenkammes, die in den Quellen als „Bastarner” erscheinen, unterstützt wurden. Der Bastarner-Name dürfte damalas eine Sammelbezeichnung für die östlich respektive nordöstlich des Karpatenkammes anschliesßenden Völkerschaften gewesen sein, die auf die schematisierenden Vorstellung dr antiken Geographie und Enthnodrapgie zurückging, wobei man sich räumlich grob auf den Herkunftsraum jenes Ethnikons bezog, das zu dieser Zeit durch die Konfrontation mit der griechischen und römischen Seite zuerst bzw. vorrangig bekannt war. Die Operationen des M. Vinicius sind, wie man mit guten Gründen annehmen kann, parallel und wohl auch konzeptionell abgestimmt zu denen des Cornelius Lentulus ebenfalls ca. 9-6 v. Chr. durchgeführt worden.82 Jede dakische Bedrohung suchte Rom bereits im Vorfeld der Donaulinie, die nun in Moesien durch Flotteneinheiten kontrolliert wurde, was zur Einrichtung der moesischen Flotte auf der Donau führte, zu bannen, wie auch Florus (II.28) hervorhebt. Dies dokumentieren in besonderer Weise die Operationen des Sex. Aelius Catus, ordentlicher Consul im Jahre 4 n. Chr., der 50,000 Menschen aus der südlichen Zone von Kleiner und Großer Walachei nach Moesien umsiedelte83 und damit ein geräumtes Glacis für die römische Donaulinie schuf. Die Stellung des Aelius Catus vereinigte, wie die inschriftlichen Zeugnisse nahelegen84, die Funktionen des Legatus Augusti des moesischen Heeres und des Proconsuls von Macedonia. Dies ist wohl aus der besonderen Situation seiner Amtszeit heraus zu erklären. Ebenfalls in die Amtszeit des Aelius Catus erfolgte der von Ovid als getisch bezeichnete, aber 81

Vgl. Auch Ptol. III.8.5; Mocsy 1974: 18f., 35 ILS 8965; Florus II.24; Syme 1971: 26ff. (Zu spät 6 v.- 4 n. Chr. datiert); Hanslik 1961: 112-116, der Symes zwingenden Ansatz erst nach dem Ende der Kämpfe 12-10/9 zu Unrecht ablehnt. 83 Strabo VII.3.10; die Datierung der Maßnahmen hat in der Forschung ganz unterschiedliche Ansätze erfahren (vor respektive nach seinem Konsulat). 84 IGR 1, 654; AE 1949: 10; 1960: 378; AE 1966 = SEG 22: 158; vgl. PIR A 157; Mrozewicz 1999 (mit umfassender Bibliographie), der Aelius Catus eher als prätorischen Proconsul in Macedonia interpretiert. 82

79

Dio LIV.36.3; RGDA XXX.2; vgl. Strabo VII.3.11; Florus II.28.29. Tacitus Anns. IV.44, 1; vgl. PIR² C 1379; Syme 1986: 290ff.; 1989: 435ff.; Eck 1997: 194 Nr. II 25. Die vorliegenden Quellen enthalten keine genauere Definition seiner Stellung. 80

61

Limes XVIII

sicher von transdanubischen Dakern getragene Angriff auf die Festung Aegyssos, die dem odrysischen Klientelherrscher verloren ging. Sie wurde von ihm mit der Unterstützung römischer Truppen zurückgewonnen.85 Letztere hat der offensichtlich für den Schutz der östlichen untermoesischen Donaulinie zuständige Legionslegat P. Vitellius auf Schiffen herangeführt. Als Ausgangsplatz kann ihm nur Oescus gedient haben, wo wir zu Recht von der Existenz eines großen augusteischen Uferkastells mit Landeplatz ausgehen können, auch wenn archäologische Befunde noch fehlen.

durchzuführen, wobei zugleich die Einfallswege nach Süden durch die Flußsysteme von Isker, Utus (Vit) und Asamus (Osam) zu decken waren. Für das Vorgehen im Bereich des unteren Jiu war eine Basis an der Mündung der Ogosta erforderlich, die zugleich die von hier über Ogosta und Skat nach Süden führenden Wege gegen Einfälle aus dem Norden sperrte. Für beide zentrale untermoesische Stellungen ist eine Besetzung bereits seit der Zeit des Cornelius Lentulus anzunehmen. Während das östliche Untermoesien bis zum Ende des thrakischen Vasallenreiches zu diesem gehörte, ist das Gebiet wahrscheinlich bis zur unteren Vit oder eher Osam zum Militärbezirk respektive zur frühen Provinz Moesia zu rechnen. Die fünf in mittelaugusteischer Zeit bestehenden dakischen Königreiche waren durch die Feldzüge des Cornelius Lentulus und des Aelius Catus in ein Vertragsverhältnis zu Rom übergeführt worden, das die Anerkennung der römischen Hoheit enthielt und sie in das klientele Vorfeld der Reichsgrenze an der Donau eingliederte (Strabo VII.3.12.13). Diese Politik führte aber zu keiner dauerhaften Stabilisierung der Verhältnisse, was zu dem römischen Urteil führte: Dacorum gens numquam fida (Tacitus Hist. III.46.2).

Dakische Gruppen hatten zusammen mit den Sarmaten die Gelegenheit zu einem Einbruch in Moesien genutzt, als A. Caecina Severus (cos. suff. 1 v. Chr.), der Legat des moesischen Heeres (Dio LV.29.3.) mit seinem Legionen 6 n. Chr. zum Entsatz der belagerten römischen Positionen im südlichen Pannonien abmarschiert war. Caecina mußte deshalb umkehren und schlug die Eindringlinge in der zweiten Jahreshälfte zurück, um danach auf den pannonischen Kriegsschauplatz zu eilen (Dio LV.30.4). Er hat damals mit Sicherheit einen Teil seines Heeres zur Sicherung gegen erneute Einfälle in seinem Rücken zurücklassen müssen. Wir können zu Recht davon ausgehen, daß im Jahre 7 nur zwei Legionen aus Moesien in das vereinigte Fünflegionenheer des Caecina und des Plautius Silvanus (s.o.) inkorporiert wurden (Anders z. B. Syme 1995: 247ff.). Daß das moesische Heer 6 n. Chr. 3 Legionen umfaßt hat, können wir aber daraus nicht mit Sicherheit folgern, da auch Legionsdetachements zurückgeblieben sein könnten. Die Legio XX, die um 3 n. Chr. in Moesien belegt ist, war zweifellos im Jahre 6 für den Markomannenfeldzug des Tiberius herangezogen und in Pannonien stationiert. Sie könnte als Ersatz für die Verlegung einer moesischen Truppe, vermutlich die Legio V Macedonica, anläßlich der Truppenkonzentration für die Orientmission des C. Caesar nach Moesien gekommen sein. Es bleibt aber naheliegend, daß des pannonischdalmatischen Aufstandes 6-9 n. Chr. eine Legion in dem damals strategisch wichtigen Raum im Rücken des Kriegsschauplatzes belassen wurde. Als Nachfolger des Caecina in der Führung des moesischen Heeres können wir im Jahre 7 n. Chr. zu Recht, obwohl dies mehrfach verneint worden ist, Sex. Aelius Catus sehen, dessen Amtszeit, wie mit gutem Grund zu vermuten ist, bis 11/12 n. Chr. dauerte, als Moesien dem Poppaeus Sabinus übergeben wurde und nun zu einer regulären Provinz umgewandelt werden sollte. Die großangelegte Umsiedlungsaktion, die eine massierte Truppenpräsenz und ein entsprechendes militärisches Vorgehen nördlich der Donau enthalten haben muß, kann erst nach dem Ende des dalamatisch-pannonischen Aufstandes etwa 9-11 n. Chr. angesetzt werden. Damals endeten alle dakischen Zentren im unmittelbaren Vorfeld der walachischen Donaulinie. Als zentrale Basis muß Aelius Catus ein römischer Militärkomplex an der Mündung des Oescus (Isker) gedient haben. Von hier aus waren die Operationen im Bereich des unteren Alutus (Olt) und westlich davon 85

Für die augusteischen Stellungen an den Flüssen Rhein und Donau, die den Römern als zentrale Kommunikationslinien dienten, zeichnen sich gemeinsame strategische Prinzipien ab. Sie sind als Basis und zugleich Sperrpositionen gegenüber den zentralen Verkehrswegen des Vorlandes angelegt worden. Sie blockierten also die möglichen Einfallswege aus dem „Barbaricum”, waren aber primär als Basen für eine offensive Militär- und Sicherungspolitik gedacht. Ebenso waren sie der Endpunkt wesentlicher rückwärtiger Kommunikationslinien. Als Beispiele können hier etwa Nimwegen (Rhein-MaasDelta), Vetera (Lippe-Linie), Novaesium (Ruhr), Bonn (Sieg), Mainz (Mainlinie), Vindonissa (Schweizer Mittelland, Aare; Bodenseegebiet) und Carnuntum („Bernsteinstraße”, Mähren) genannt werden. Für die augusteischen Lager im moesischen Raum, wo wir noch weitgehende Forschungslücken haben, kann von denselben strategischen Prinzipien ausgegangen werden: große Militärbasen am Ende von Flußsystemen, die das Hinterland mit entsprechenden Kommunikationslinien erschließen, Positionen, die zugleich Einfallstore an traditionellen, von Norden kommenden Verkehrswegen verschließen, um die sich zudem die Besiedlung des Vorfeldes gruppierte, und die eben über diese Verkehrswege das Vordringen der römischen Truppen in das transdanubische Vorland ermöglichten. Im frühen römischen Moesien treffen diese Kriterien primär auf Viminacium (Morava und auch Mlava), Augusta (Ogosta) und Oescus (Isker) zu, ferner auf die Mündung des Timacus (Timok). Ratiaria weist dagegen nur eine sekundäre strategische Bedeutung auf, zumal das von Wasserläufen und -flächen durchzogene Vorland nördlich des Donau Kommunikationslinien unmöglich machte. Wesentlich größere Bedeutung ist für die Kontrolle des westlichen Oltenien kam der Donaustrecke zwischen Drobeta und Vidin und hier insbesondere der Timok-

Ovid Pont. I.8.11-24; 4, 7. Vgl. auch Schuster 1961: 385f.; Syme 1978.

62

Karl Strobel: Die Legionen des Augustus Mündung zu. Die traditionelle Annahme ein frühes moesisches Legionslager in Ratiaria zu suchen (Ivanov 1997: 540ff.), hat nur wenig Wahrscheinlichkeit für sich. Ratiarias Bedeutung lag wahrscheinlich von Anfang an in seiner Rolle als Flottenbasis. Daß bereits unter Cornelius Lentulus römische Militärpositionen an der moesischen Donaulinie eingerichtet worden waren, steht außer Zweifel (cf. Florus II.28).

Raumes von Morava-Mündung und Viminacium ist bisher völlig unzureichend. Die neueren archäologischen Forschungen richten sich primär auf einzelne Nekropolen. Auch können Ortsverlagerungen in der Entwicklung dieses militärischen Komplexes nicht ausgeschlossen werden, wie wir dies von der Rheinlinie, aber auch von Carnuntum kennen. Im Oescus ist die Anwesenheit der Legio V Macedonica durch die Bauinschrift des Jahres 44 n. Chr. Gesichert.90 Durch die Ausgrabungen der letzen Jahre in Oescus konnten hiererstmals ältere Befunde aufgedeckt werden, doch blieb die Interpretation lange Zeit eher hypothetisch und zuerst durch die flächenmäßig sehr geringen Sondagen auch wenig fundiert.91 Erst in den Kampagnen 1999-2000 konnte ein gesicherter Befund durch die Aufdeckung einer als Schalenmauer mit Quaderfront und Erdschüttung aufgeführten Befestigung, offensichtlich ein Teil der Südmauer, aufgedeckt werden. Diese Ummauerung könnte allerdings erst zu der Anlage der traianischen Colonia Ulpia nach der Auflassung des Legionslagers gehören, doch sprechen die begleitenden Funde für eine flavische Datierung (vgl. auch Goguey & Reddé 1995: 33ff.). Es zeigt sich heute, daß weder das ursprüngliche Legionslager noch die im 2. Jh. ausgebaute Colonia mit dem pentagonalen Mauerzug in Zusammenhang steht, sondern dieser erst einer späterkaiserzeitlichen Neubefestigung zuzuweisen ist. Archäologisch ungeklärt bleibt bisher aber die Frage des frühkaiserzeitlichen Legionslagers, auch wenn aus dem Areal der späteren Colonia frühe italische bzw. norditalische und südgallische Importkeramik bekannt ist. Denn diese können ebenso wie die sich in zwei respektive wohl drei Hauptphasen gliedernden Gebäudereste auch mit der als sicher vorauszusetzenden Zivilsiedlung bei dem frühen Legionslager in Verbindung gebracht werden. Nach der Aufdeckung einer in Stein errichteten Lagerbefestigung, die wir mit gutem Grund als domitianischen Neubau nach dem Dakereinfall von 85 n. Chr. (Strobel 1989) interpretieren können, dürften diese frühen Schichten mit ihrem Fundbestand an Militaria jedoch zu einem vorflavischen respektive vordomitianischen Lager gehören. Allerdings bleibt die Möglichkeit offen, die auch durch die Topographie des Ortes nahegelegt wird, in Donaunähe ursprünglich nur ein großes Uferkastell zu suchen und das Lager auf der strategisch dominierenden Höhe im Rücken der antiken Stadt zu vermuten. Vergleiche mit den Anlagen anderer augusteischer Militärstützpunkte sprechen für diese Vermutung, der bisher von der bulgarischen Forschung nicht nachgegangen worden ist.

Als Stammlegionen des moesischen Militärbezirkes können wir mit guten Gründen seit 10/9 v. Chr. die Legionen IIII Scythica und V Macedonica annehmen. Im Jahre 23 bestand die Provinzgarnison aus 2 Legionen, wie Tacitus (Anns. IV.5.3) ausführt. 33/4 n. Chr. bauten Baukolonnen beider Legionen den westlichen, tiberischen Abschnitt der Kunststraße in der Donauenge des Djerdap.86 Die Legio XX ist offensichtlich um 3 n. Chr. für eine gewisse Zeit im Raum von Oescus gelegen. Als Legionsbasen an der Donau sind ebenfalls seit 10/9 der Raum von Viminacius und Oescus, erstere für die IIII Scythica, letztere für die V Macedonica, zu erschließen. Hier lagen die Schlüsselstellungen zur Kontrolle des dakischen Raumes; die Daker galten den Römern von Anfang an als bedeutende Gegner.87 Eine dritte Legionsbasis, wenn vorhanden, sollte eher an der rückwärtigen Morava-Linie gesucht werden; hier gabelten sich die von Makedonien und der Adria kommenden Verbindungslinien nach Viminacium und zur untermoesischen Donau. Der Raum von Naissus muß weiterhin als naheliegend für eine große rückwärtige Militärbasis in augusteischer Zeit gesehen werden. Die frühen römischen Lager an der unteren Donau Unser archäologisches Wissen über die frühkaiserzeitlichen Legionslager an der unteren Donau war bisher im wesentlichen auf das Lager von Novae beschränkt, das in claudischer Zeit nach der Annexion Thrakiens errichtet worden ist. Doch auch hier sind die konkreten archäologischen Befunde für die Frühphase eher gering.88 Der Versuch von M. Mirković, die Errichtung des Legionslagers von Viminacium erst in domitianischer Zeit muß zurückgewiesen werden. anzunehmen89, Selbstverständlich ist auch in Singidunum von einer frühen römischen Präsenz seit 13/12 v. Chr. auszugehen. Allerdings kann dieser Raum in früher römischer Zeit wohl eher zur römischen Save-Linie und damit zu Illyricum gerechnet werden. Singidunum war strategisch auf das Savetal, Slavonien und die südliche Donau-Theiß-Ebene ausgerichtet. Der Anschluß an Moesien kann sehr wohl erst bei der Neugliederung Illyricums nach dem Ende des dalmatisch-pannonischen Aufstandes 9 n. Chr. erfolgt sein. Das archäologische Wissen über die Frühgeschichte des

In Augustae, dem namensgebenden Kastell an der Osgosta-Mündung, sind die beiden Steinbefestigungen 90 ILBulg 1. Vgl. auch Gerov 1967. Die historischen und truppengeschichtlichen Ausführungen bei Ivanov 1997: 503ff. sind in mehreren Punkten nicht überzeugend. 91 Kabakchieva 1994-2000. Für weitere Informationen danke ich bulgarischen Kollegen, insbesondere Herrn R. Ivanov. Die historischen und truppengeschichtlichen Ausführungen von Frau Kabakchieva sind z. T. problematisch oder zu korrigieren.

86

ILIug 57.60. Zur IIII Scythica in Moesien vgl. auch Speidel 2000: 328ff. 87 Vgl. Horace Sat. II.6.53; Carm. I.35.9; II.20.18; III.6.14; III.8.18. 88 Vgl. zusammenfassend Ivanov 1996; 1997. 89 Mirković 1996. Vgl. ferner Zotović & Jordović 1990.

63

Limes XVIII

(Augustae I und II) erfaßt (Mašov 1991; 1994; Ivanov 1997: 543ff.). Unklar ist jedoch die Datierung der Spuren einer Holz-Erde-Phase an der Westseite von Augustae I, ebenso jene des Verteidigungsgrabens. Jüngste Untersuchungen des Fundmaterials legen jedoch einen Beginn der Belegung des Platzes des späteren Auxiliarkastells in spätaugusteischer Zeit nahe. Ausdehnung und Lage der augusteischen Anlagen, die nicht auf die Fläche des späteren Kastells beschränkt gewesen sein dürften, bleibt derzeit offen.

Gómez-Pantoja J. 2000 Legio IIII Macedonica. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 105-117. Gómez-Pantoja J. 2000 Legio X Gemina. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 169190. Groag E. 1926 Licinius Crassus. Paulys Real-Encyclopädie der classichen Alterwissenschaft 13, 1. (Stuttgart): Cols. 270- 285. Groag E. 1932 Taurius. Paulys Real-Encyclopädie der classichen Alterwissenschaft 4 A. (Stuttgart): Cols. 2321-2323. Gruen E.S.1996 The expansion of the empire under Augustus. Cambridge Ancient History² X. (Cambridge): 147-197. Hahn I. 1969 Die Legionsorganisation des Zweiten Triumvirats. Acta Antiqua Academiae Scientiarum Hungaricae 17: 199-222. Hanslik R. 1961 M. Vinicius. Paulys Real-Encyclopädie der classichen Alterwissenschaft 9 A 1. (Stuttgart): Cols. 112-116. Hardy E.G. 1920 Augustus and his legionaries. Classical Quarterly 14: 187-194. Ivanov R. 1996 Der Unterdonaulimes von Dorticum bis Durostorum (1.-6. Jh.). In P. Petrović (ed.) Roman limes on the middle and lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 161-171. Ivanov R. 1997 Das römische Verteidigungssystem an der unteren Donau zwischen Dorticum und Durostorum (Bulgarien) von Augustus bis Maurikios. Berichte der Römisch-Germanischen Kommission 78: 467-590. Kabakchieva G. 1994 Archäologische Angaben über die Frühgeschichte von Ulpia Oescus. In D. Draganov (ed.) Settlement life in ancient Thrace (Cabyle III). (Jambol): 148-170. Kabakchieva G. 1996 Frührömische Militärlager in Oescus (Nordbulgarien). Germania 74: 95-117. Kabakchieva G. 1997 Die Gründung des Militärlagers bei der Mündung des Flusses Oescus und die Entstehung der Provinz Moesien. In W. Groenman-van Waateringe, B. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVth International Congress of Roman Frontier Studies. (Oxford): 387-392. Kabakchieva G. 1999 Neue Angaben zum frührömischen Legionslager am Oescus-Fluß (Nordbulgarien). In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalǎu): 487-494. Kabakchieva G. 2000 Oescus I. Castra Oescensia. Das frührömische Militärlager bei der Mündung des Flusses Iskar. (Sofia). Kabakchieva G. 2001 Die Frührömische Militärgeschichte der Provinz Moesia an der unteren Donau. Jahrbericht der Gesellschaft Pro Vindonissa: 3-8. Keppie L. 1983 Colonisation and veteran settlement in Italy, 47-14 B.C. (London). Keppie L. 1984 The making of the Roman army. (London). Keppie L. 1995 Soldiers and veterans at the colony of Forum Julii (Fréjus). In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 367-372. (=Legions and veterans. Roman Army Papers 1971-2000. (Stuttgart): 263-300). Keppie L. 2000 Legions and veterans. Roman Army Papers

Literatur Bassignano M. S. 1997 Ateste. Supplementa Italica 15: 11-376. Bleicken J. 2000 Augustus. (Berlin). Brunt P.A. 1971 Italian manpower, 225 B.C.-A.D. 14. (Oxford). Brusin G.B. 1992 Inscriptiones Aquileiae. Pars altera. (Udine). Buchi E. 1993 Venetorum Angulus: Este dà communità poleoveneta a colonia romana. (Verona). Burian J. 2000 Moesi, Moesia. Die Neue Pauly 8. (Stuttgart): Cols. 328-332. Cavaillac E. 1952 RGDA et l’armée. Revue Etudes Latines 30: 285-296. Cerdán A.M. 2000 La Legio IIII Macedonica en la Peninsula Ibérica: el campamento de Herrera de Pisuerga (Palencia). In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 609-624. Cerdan A.M. & Marcos V.G. 2000 Nuevos testimonios acerca de las legiones VI Victrix y X Gemina en la región septentriol de la Península Ibérica. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 589607. Christol M. & Drew-Bear T. 1998 Vétérans et soldats légionnaires à Antioche en Pisidie. In G. Paci (Hrsg.) Epigraphia romana in area adriatica. (Macerata): 303332. Crawford M.H. 1989 Ateste and Rome. Numimaicae antichità classiche (Qaderni Ticinesi) 18: 191-200. Dabrowa E. 2000 Legio III Gallica. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 309315. Eck W. 1978 Subatianus. Paulys Real-Encyclopädie der classichen Alterwissenschaft Suppl. 15. (München): Cols. 569-570. Eck W. 1997 Cn. Cornelius Lentulus. Die Neue Pauly 3. (Stuttgart): Col. 194. Errington M. 1999 Makedonia/Makedonos. Die Neue Pauly 7. (Stuttgart): Cols. 726-739. Filow B. 1906 Die Legionen der Provinz Moesia von Augustus bis auf Diokletian. (Leipzig). Fluss M. 1932 Moesia. Paulys Real-Encyclopädie der classichen Alterwissenschaft 15.2. (Stuttgart): Cols. 2350-2412. Franke T. 2000 Legio V Alaudae. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 39-48. Gerov B. 1967 Epigraphische Beiträge zur Geschichte des moesischen Limes in vorclaudischer Zeit. Acta Antiqua Academiae Scientiarum Hungaricae 15: 85-105. Goguey R. & Reddé M. 1995 Le camp légionnaire de Mirebeau. (Mainz).

64

Karl Strobel: Die Legionen des Augustus 1971-2000. (Stuttgart). Keppie L. 2000b Legiones II Augusta, VI Victrix, IX Hispana, XX Valeria Victrix. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 25-37. Kienast D. 1982 Augustus. (Darmstadt². 1992). Kienast D. 1996 Römische Kaisertabelle. (Darmstadt²). Kos M. 1977 The military role of Macedonia from the Civil Wars to the establishment of the Moesien Limes. In J. Fitz (ed.) Akten des 11. Internationalen Limeskongresses. (Budapest): 286-292. Kubitschek W. 1924 Legio. Paulys Real-Encycyclopädie der classichen Alterwissenschaft 12, 1: Cols. (Stuttgart): 1186-1210. Laporte J-P. 2000 La Legio VIIa et la déduction des trois colonies augustéennes de Césarienne. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 555579. Le Bohec Y. 1989a L’armée romaine. (Paris). Le Bohec Y. 1989b La troisième légion Auguste. (Paris). Le Bohec Y. 2000 Legio III Augusta. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 373381. Le Roux P. 1982 L’armée romaine et l’organisation des provinces ibériques. (Paris). Lica V. 2000 The coming of Rome in the Dacian world. (Konstanz). Mašov S. 1991 Das spätantike Kastell und die frühbyzantinische Stadt Augustae beim Dorf Harletz, Nordwestbulgarien (bulg.). Severozapadna Balgarija 16: 21-43. Mašov S. 1994 Das spätantike Kastell und die frühbyzantinische Stadt Augustae beim Dorf Harletz, Nordwestbulgarien. In G. Susini (Hrsg.) Limes (Studia di Storia 5) (Bologna): 21-29. Menella G. 1996 Legionari bagienni in età triumvirali. In Studi in onore A. Garzetti. (Brescia): 257-269. Mirković M. 1996 The Iron Gates (Derdap) and the Roman policy on the Moesian limes AD 33-117. In P. Petrović (ed.) Roman Limes on the middle and lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 27-40. Mitchell S. 1976 Legio VII and the garrison of Augustan Galatia. Classical Quarterly 70: 298-308. Mitchell S. 1993 Anatolia. Land, men, and gods in Asia Minor. (Oxford). Mócsy A. 1966 Der vertuschte Dakerkrieg des M. Licinius Crassus. Historia 15: 511-515. Mócsy A. 1974 Pannonia and Upper Moesia. (London). Mommsen T. 1883 Res Gestae Divi Augustae ex Monumentis Ancyrano et Apolloniensi. (Berlin). Mrozewicz L. 1999 Prosopographia Moesiaca II: Sex. Aelius Catus. Eos 86 103-105. Papazoglou F. 1979 Quelques aspects de l’histoire de la province de Macedoine. In H. Temporini (ed.) Aufstieg und Niedergang der römischen Welt II.7.1. (Berlin): 302369. Patsch C. 1932 Beiträge zur Völkerkunde von Südosteuropa V. Aus 500 Jahren vorrömischer und römischer Geschichte 1. Bis zur Festsetzung der Römer in Transdanuvien. (Wien). Perea Yébenes S. 2000 Hispania y la legio XX. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 581-587.

Petrović P. 1996 Roman limes on the middle and lower Danube. (Beograd). Polaschek E. 1939 Trumpilini. Paulys Real-Encyclopädie der classichen Alterwissenschaft 7 A 1. (Stuttgart): Cols. 707-710. Reddé M. 2000 Legio VIII Augusta. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 119126. Rémy B. 1989 Les carrières senatoriales dans les provinces romaines d’Anatolie au Haut-Empire (31 av. J.-C. 284 ap. J.-C.). (Istanbul). Ritterling E. 1924-25 Legio. Paulys Real-Encyclopädie der classichen Alterwissenschaft 12, 1-2. (Stuttgart): Cols. 1211-1829. Rizakis A.D. 1997 Roman colonies in the province of Achaia: territories, land and population. In S.E. Alcock (ed.) The early Roman empire in the east. (Oxford): 15-36. Rizakis A.D. 1998 Achaïe II: la cité de Patras, épigraphie et histoire. (Athens). Sanders H.A. 1941 The origins of the Third Cyrenaic legion. American Journal of Philology 62: 84-86. Schmitthenner W. 1958 The armies of the Triumviral period. A study of the origins of the Roman imperial legions. (Unpublished D.Phil. thesis, University of Oxford). Schuster M. 1961 P. Vitellius. Paulys Real-Encyclopädie der classichen Alterwissenschaft 9 A 1. (Stuttgart): Cols. 385-391. Sear D.R. 1998 The history and the coinage of the Roman Imperatores, 49-42 B.C. (London). Sherk R. 1980 Roman Galatia: the governors from 25 B.C. to A.D. 114. In H. Temporini (ed.) Aufstieg und Niedergang der römischen Welt II.7.2. (Berlin): 9541052. Speidel M.A. 1998 Legio IIII Scythica, its movements and men. In D. Kennedy (ed.) The twin towns of Zeugma on the Euphrates: Rescue work and historical studies. Journal of Roman Archaeology Supplementary Series 27: 163204. Speidel M.A. 2000 Legio IV Scythica. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 327337. Speidel M.P. 1992 Roman Army Studies II. (Stuttgart). Stein A. 1940 Die Legaten von Moesien. (Budapest). Strobel K. 1987 Die politische und militärische Führung Dakiens während der Kämpfe mit Rom. Balkan-Archiv 12: 149-166. Strobel K. 1988 Die legio V Alaudae in Moesien. Eine Phantomtruppe der römischen Militärgeschichte. Historia 37: 504-508. Strobel K.1989 Die Donaukriege Domitians. (Bonn). Strobel K 1998 Die Daker. Zur Komplexität ethnischer, politischer und kultureller Größen der Geschichte des unteren Donauraumes I-II (rum. mit dt. Res.). In Studia si Cercetari de Istorie Veche (si Arheologie) Buceresti. 49: 61-95, 207-227. Strobel K. 2000 Zur Geschichte der Legiones V (Macedonica) und VII (Claudia pia fidelis) in der frühen Kaiserzeit und zur Stellung der Provinz Galatia in der augusteischen Zeit. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 515-528. Syme R. 1971 Danubian Papers. (Bukarest). Syme R. 1978 History in Ovid. (Oxford). Syme R. 1986 The Augustan aristocracy. (Oxford). Syme R. 1989 Roman Papers VI. (Oxford).

65

Limes XVIII

Syme R. 1995 Anatolian studies in Strabo. (Oxford). Wachtel K. 1977 Zum Militärkommando an der Unteren Donau in Augusteischer Zeit. In J. Fitz (ed.) Akten des 11. Internationalen Limeskongresses. (Budapest): 377383. Wheeler E.L. 2000 Legio XV Apollinaris. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les légions de Rome sous le Haut-Empire. (Lyon): 259-308. Zotović L. & Jordović C. 1990 Viminacium. Nekropole „Vise Grobalja”. (Beograd).

66

The northern, southern and eastern frontiers and the climate c.AD200 Wolfgang Vetters & Heinrich Zabehlicky The eruption of the Taupo volcano on North Island, New Zealand can be dated to the period AD175-180. Compared to other eruptions a deterioration of the climate for two to four decades can be identified. Unexplained changes at some military sites might have been reactions to the resultant periods of bad weather. As one of a number of reasons this could explain changes in frontiers and frontier policy at just this time.

The topic of archaeological evidence which could be explained by a deterioration of climate caused by a volcanic eruption has been discussed by the authors several times (Vetters 1994; Zabehlicky 1994; Kandler et al 1995; Vetters & Zabehlicky 2001). There is some archaeological evidence which seems to be best explained by increasing humidity, bad weather and rising ground water levels all which can be dated to the second half of the C2nd.

The Taupo volcano is situated at the centre of North Island, New Zealand and is now marked by a lake 624sq kms in size, the result of a violent explosive eruption. Thick pumice deposits in an area of more than 1000sq kms bear witness to that explosion. The event is attributable by various methods of age determination to about AD185. The effect of the eruption on the (global) climate can be compared with other eruptions with associated documentation of their climatic impact.

Geological and climatological evidence A geological answer to the suspicion of a deterioration of climate was hinted at by the eruption of the volcano Taupo on New Zealand. The results are summarized in the following paragraphs.

Based on the petrography and the chemical composition of the pyroclastics the Taupo eruption can be equated very closely with that of Tambora in 1815. This eruption on Sumbabwe Island in the Little Sunda Islands (Indonesia) had similar dimensions with respect to volume whilst the chemical composition was comparable. The eruption of 1815 was documented and interpreted in numerous climatic observations in America and Europe. Climatic charts of the following years give a characteristic picture of weather deterioration. In 1816 a remarkable drop in the average summer temperatures and in the average annual value occurred.

In general the climate is determined by solar irradiation whose long-term shifts are caused by astronomical factors. Short-term shifts are often caused by geogenic factors in the atmosphere and the stratosphere. Such atmospheric factors are: 1. the percentage of aerosols (eg. sulphur) 2. dust or ash particles 3. water vapour.

The regular summer average was not regained before 1840. The fine ash component of the volcanic pyroclastic material as well as the relatively high sulphur component within the gases was held responsible. The finest ashparticles served as nuclei for condensation for the development of clouds and for the higher rate of precipitation. Ash and dust reduced solar irradiation and the sulphur component was important for splitting the sunlight to single spectra. The result was a reduction of the infra-red thermal radiation, which was absorbed in the atmosphere by the sulphur component.

Violent and explosive eruptions of volcanoes would be sources of dust and gas in the atmosphere, while the anthropogenic factor - discussed so much nowadays might be discounted for the less populated and not so industrialized world in this period. The retention period of volcanic products in the stratosphere and atmosphere is crucial for the shifts of irradiation and therefore for the climate. The local climate, the height of the eruption column and the chemical as well as the petrographic composition of the eruption cloud are also important factors. Year 1400 BC AD 185 AD 1815 AD 1883 AD 1982

Volcano Santorin Taupo Tambora Krakatoa El Chichon

SO2, H2S ? 1.4-1.8% 0.3% 0.4% 2-2.3%

Volume (km³) c.10 80-100 60-150 16 0.1

Table from Vetters 1994

67

Ash/Pumice 40%/40% 40%/20% 15%ca.40% 45%/35% 25%/45%

Climate ? ? c. –1.5°C -0.5-0.8°C -0.3-0.5°C

Limes XVIII

The 25 years from 1815 to 1840, when the temperature had ‘recovered’, fell in a period of generally increasing temperatures. In contrast the C2nd was the beginning of a period of declining warmth, which had a maximum at c.AD100 but which became cooler subsequently until medieval times. So the climatic effect of the Taupo eruption could have lasted longer and higher precipitation and lower average temperatures could be estimated for a period of 30 to 50 years.

sleeves was replaced by the tunica manicata (Ubl 1969: 510), the paenula by the longer and wider sagum (Ubl 1969: 565) and the breeches by longer trousers (Ubl 1969: 597). From this starting point the frontier-policy and military history should be checked for the period from Commodus to Caracalla. The literary sources for this time (Historia Augusta, Herodian and Cassius Dio) enjoy little scholarly respect and have to be treated with reserve.

Though we have to accept that the reasons for these changes would have been unclear to the Romans and that they had no weather-statistics as far as we know, some reaction to this period of some decades should be recognizable in the archaeological evidence.

Commodus Commodus is reported in the SHA (XVI.7) being responsible for the organization of a new African grain fleet following the model of the Egyptian one, in case the Egyptian fleet should fail. Perhaps this was the result of a famine-revolt in AD189, which was mostly attributed to grain speculation. Though speculation should not be doubted, and although the Historia stresses the ridiculous fact that the fleet was called classis Africana Commodiana Herculia, Commodus’ actions seem to be wise and prudent. The negative aspect could easily be assigned to the tendency of the Historia.

Archaeological evidence from military contexts We have dealt elsewhere with the evidence from one military context, the case of the garrison of legio II Italica at the confluence of the Enns with the Danube. Here, at Albing, the large and regular fortification of a legionary fortress was never completed and used, but a new fortress was built only 6kms distant on a place 5m higher, on a dryer location, protected from inundation (Kandler et al 1995: 562; Zabehlicky 1994).

In general the policy of Commodus was surely not offensive. His terminating of the Marcomannic wars without reaching an expansive success was heavily criticized. However it is hard to say whether this criticism was due to the same tendency of the senatorial historians.

The fortress of Eining-Unterfeld on the Danube may also be relevant (Schönberger 1970; Fischer 1983). An area of 328m x 320m was surrounded on three sides by ditches outside a stone wall, whereas the riverside-front is lacking. It is unclear whether this wall ever existed or whether it was destroyed by the river. Following the coin evidence, the fortress was short-lived after AD161/169 (Schönberger 1970: 78) and the tile-stamps are the earliest of the legio III Italica. Internal buildings are known only from aerial photographs (Fischer 1983: 104). The interpretation of its function is not certain, varying between a fortified supplybase (Schönberger 1970: 78) or a short-lived half-legionary fortress (Fischer 1983: 104). The location at the only place to access the river from a slightly higher position along the otherwise flat riverside would fit both explanations. The plan suggests at any rate that the riverside front was destroyed by the Danube and the coins suggest that this happened not too long after AD169.

Septimius Severus Policy changed very much with the reign of Septimius Severus, who came to power with the help of the army on the Danube and who had to fight for his reign during several campaigns. Maybe he was more a politician than an army-leader, but he cared for the soldiers. For his reign we shall follow mostly Birley 1988. In the spring of 195 Septimius mounted an invasion of Mesopotamia, probably not only from a ‘desire for glory’ as Dio says, but following a design to extend the eastern frontier (Birley 1988: 115f.). In AD197 he started a new campaign. A reorganization of the eastern frontiers took place in 198, establishing Mesopotamia as a new province with its capital at Nisibis. Osrhoene was also organized as a province and the frontiers in Syria/Phoenike and Arabia were reorganized (Birley 1988: 129ff.).

Another case with a military aspect is the ditch of a temporary camp for the Marcomannic wars of the 160s or170s at Bernhardsthal, which was filled in after its abandonment. At the end of the C2nd the rebuilt farm was surrounded by a wall without a ditch, which for this reason can be called a flood-protection dam (Adler 1986).

This activity continued and in the north-east of what is now Jordan the early decades of the C3rd “…are marked by significant work” including the building of a new road from Syria to the Azraq Oasis (Kennedy 2000: 47). This is attested by building inscriptions from Qasr el-Uweinid (Kennedy 2000: 59) and Qasr el-Hallabat (Kennedy 2000: 93) and by milestone fragments (Kennedy 2000: 55 & 65), altogether dating between 200 and 214.

A hint of cooler weather could be also expected in more or less warm clothing. While civil costumes are determined by regional or fashion-dependent differences, the military uniform could be checked as a more or less global standard. In the time of the emperor Caracalla the uniform of the Roman army was reformed. Generally these changes tended towards warmer clothing: The tunica without

We are indebted to A. Schmidt-Colinet for the suggestion 68

Wolfgang Vetters and Heinrich Zabehlicky:The northern, southern and eastern frontiers and the climate c.AD200

that political and military activities in the Severan period extended as far to the east as Palmyra. A cohors I Flavia Chalcidenorum equitata sagittariorum was transferred there between 162 and 212 (Herzig & Schmidt-Colinet 1991: 66f.) and dedicated one of the rare Latin inscriptions of Palmyra to Severus and his sons. From the same place the building of a covered stoa with 8 columns along the main road is reported by a building inscription of 218 (Will 1983: 71 ff.). The building of a stoa is strongly reminiscent of the case of the Damianos stoa at Ephesos, which was built in the period after 185 explicitly to stop the worshippers at the Artemision staying away, “if it is raining” (Vetters & Zabehlicky 2001: 10.; Knibbe forthc.).

At Eburacum/York Septimius’ life ended on 4 February 211, with his well-known last words to his sons: “Do not disagree between yourselves, give money to the soldiers and despise everyone else”. Quoting Kornemann (1963: 309): “More than elsewhere in the north, the empire was here, on the furthermost point of its frontier, in a retrograde movement.” Caracalla As is well known, Septimius Severus’ sons (Caracalla and Geta) did not follow the first part of his bequest as arcana imperii, but the survivor of the struggle, Antoninus III/Caracalla, kept to the rest of it. Both the military uniform (mentioned above) and the pay for the legionary soldiers were improved. Again in 213 he had to campaign on the Upper Danube, on the Raetian limes. Here the ‘Limestor of Dalkingen’ - a singular structure as a gate in the limes, facing towards the empire in a representative manner - could mark his triumph over the Alemannic tribes (Planck 1980). I am indebted to Ernst Künzl for his suggestion that the replacement of the wooden palisade by a masonry wall on the Upper German and Raetian limes might also be attributed to rising ground-water level. This would have damaged the wooden piles and stimulated its replacement by a masonry construction. In terms of the military and frontier aspects of Caracalla’s reign I see one more expedition on the lower Danube against the Carpi 214 and a war against the Parthian empire 216/217 (Kornemann 1963: 313).

In Africa too the military frontier was extended and newly fortified from Dimmidi to Cydamus and Gholaia (Birley 1988: 147). In the wide-ranging ‘Libyan Valleys Survey’ a change of climatic conditions in the Roman period was sometimes considered in order to explain the surprising ‘olive-boom’ (Mattingly 1988: 23). But climatic changes in Roman times were always rejected because of the results of geological research within the scope of the same project (Barker et al 1983: 80). It would be interesting to know from the researchers concerned whether a deterioration of climate in terms of bad weather for a period no longer than 25, or at most 50, years might be expected to be detectable by the methods applied. As part of the Libyan survey, a lot of effort was made to expand into a pre-desert region, which at first sight offered no attractions or at any rate less than a trans-Danubian region, which should have been quite well-known to the former governor of Pannonia and where probably the installation of two new provinces Marcomannia and Sarmatia had been thought of some 20 years previously (Dobias 1966: 121ff.).

Summary In assessing the factors which occasioned the advance of the imperial frontiers, first one should look at the attitude of the emperors in question. Perhaps the Historia Augusta exaggerates Commodus’ preference for luxury, but this emperor was not one of the ‘notorious’ expanders of the empire. This is different to Septimius Severus and Caracalla. Septimius literally was accused of being ‘desirous of glory’. Even if this is a simplification of psychology and fact, he surely looked back to Trajan, the last emperor to labour to enlarge the precincts of the empire. Caracalla in this aspect of his reign surely followed the heritage and the example of his father.

In contrast to the east and African frontiers, Severus’ attitude to frontier policy and military action was different on the north. In Britain, the pressure from the barbarian tribes increased, the supply for the troops had to be improved and secured by rebuilding the granaries at Birdoswald (Wilmott 1989: 35) and by building new ones at Arbeia/South Shields (Hodgson 1999: 75f.). The Antonine Wall was abandoned within 20 years of its construction. Even the contemporary authors, Dio and Herodian, do not agree about the real frontier line, Hadrianic or Antonine Wall, but they do concur on the swampy and muddy character of the country and the tattooed barbarians. The importance of bridges and pontoons is reported by Herodian and Dio and reiterated by Birley (1988: 179). Finally, the campaign of 209 resulted in Roman arms reaching the northernmost point of the island. Septimius Severus’ intention to occupy all the north of Britain is probable (Birley 1988: 182) but after a second campaign in 210/213, conducted mainly by Caracalla, the ultimate effect was only to stabilize the line of Hadrian’s Wall.

A second factor that requires analysis is the nature of the different regions of the imperial frontiers. In Britain strong efforts were made, but ultimately they succeeded only in keeping a line already reached some decades earlier. In Upper Germany and Raetia the situation was very similar. Incursions by the barbarians were punished, but no new achievements were made. The plan to expand beyond the middle Danube, marked by the projected provinces of Marcomannia and Sarmatia, was never realized. Quite different is the situation in the east and along the southern imperial frontiers. The province of Mesopotamia was re-established. This included the stabilization of the ‘hinterland’ including the region of Osrhoene and the city 69

Limes XVIII

of Palmyra. The African region, used for agriculture, was extended substantially and protected by a strengthened military frontier.

Planck D. 1980 Das Limestor von Dalkingen. In W. Beck & D. Planck Der Limes in Südwestdeutschland. (Stuttgart): 132-138. Schönberger H. 1970 Das Römerlager im Unterfeld bei Eining. Bericht über die Grabung 1968. Germania 48: 66–84. Ubl H. 1969 Waffen und Uniform des römischen Heeres der Principatsepoche nach den Grabreliefs Noricums und Pannoniens. (Ungedr. Diss. Wien). Vetters W. 1994 Der Taupo und das Klima um 200 A.D. in Europa. In H. Friesinger, J. Tejral & A. Stuppner (edd.) Markomannenkriege - Ursachen und Wirkungen. VI. Internationales Symposium ‚Grundprobleme der frühgeschichtlichen Entwicklung im nördlichen Mitteldonaugebiet’, Wien 23-26. November 1993. (Spisy Archeologického Ústavu AV ČR Brno 1): 457461. Vetters W. & Zabehlicky H. 2001 Eine Klimakatastrophe um 200 n. Chr. und ihre historisch-archäologische Nachweisbarkeit. In M. Frey & N. Hanel (edd.) Archäologie Naturwissenschaften Umwelt. Beitrage der Arbeitsgemeinschaft ‚Römische Archäologie’ auf dem 3. Deutschen Archäologenkongreß in Heidelberg 1999. British Archaeological Reports Int. Ser. 929. (Oxford): 9-12. Will E. 1983 Le développement urbain de Palmyre: Témoignages épigraphiques anciens et nouveaux. Syria 60: 69–81. Wilmott T. 1989 Birdoswald. In C. Daniels (ed.) The eleventh Pilgrimage of Hadrian’s Wall 1989. (Newcastle): 35f. Zabehlicky H. 1994 Kriegs- oder Klimafolgen in archäologischen Befunden? In H. Friesinger, J. Tejral & A. Stuppner (edd.) Markomannenkriege - Ursachen und Wirkungen. VI. Internationales Symposium ‚Grundprobleme der frühgeschichtlichen Entwicklung im nördlichen Mitteldonaugebiet’, Wien 23-26. November 1993. (Spisy Archeologického Ústavu AV ČR Brno 1): 463-469.

At the actual state of knowledge the frontier policies of Commodus, Septimius Severus and Caracalla are in any case inconsistent. Accepting a weather factor, of declining temperatures and increasing humidity, the less effective attempts to protect or enlarge imperial territory at the northern and central European frontiers might be explained. There was no good reason to conquer these regions, whose tribal economies would have suffered heavily by poor harvests (which might also explain the frequent incursions into the better-fed Roman empire). Equally, and even perhaps more apposite, the expansion into pre-desert or desert regions in the east and the south might have been justified by the climatic changes here discussed. Bibliography Adler H. 1986 Bernhardsthal. In M. Kandler & H. Vetters (edd.) Der römische Limes in Österreich. Ein Führer. (Wien): 244-247. Barker G.W.W., Gilbertson D.D., Griffin C.M., Hayes P.P. & Jones D.A. 1983 The UNESCO Libyan Valleys Survey V: Sedimentological properties of Holocene wadi floor and plateau deposits in Tripolotania, north-west Libya. Libyan Studies 14: 69– 85. Birley A.R. 1988 The African emperor: Septimius Severus. (London). Dobias J. 1966 Rom und die Völker jenseits der mittleren Donau. In Corolla Memoriae Erich Swoboda dedicata. Römische Forschungen in Niederösterreich, Band V. (Graz): 113-125. Fischer T.H. 1983 Kastelle und vicus bei Eining-Abusina, Landkreis Kehlheim. In Führer zu römischen Militäranlagen in Süddeutschland. Herausgegeben aus Anlaß des 13. Internationalen Limeskongresses in Aalen (Stuttgart): 101–107. Herzig H.E. & Schmidt-Colinet A. 1991 Two recently dicovered Latin inscriptions from Palmyra. Damaszener Mitteilungen 5: 65–69. Hodgson N. 1999 South Shields-Arbeia. In P. Bidwell (ed.) Hadrian’s Wall 1989–1999. (Carlisle): 73-82. Kandler M., Vetters W. & Zabehlicky H. 1995 Fragile towns in the north of the Roman empire. A geoecological impact for the last quarter of the 2nd century and earthquakes. In La Città Fragile in Italia. Geologia applicata e idrogeologia 30. Atti Primo convegno del Gruppo Nazionale di geologia applicata con la partecipazione dell’International Asssociation Engineering Geology (I.A.E.G.), Sezione Italiana. Giardini Naxos (ME 11-15 Giugno). Preprint published by Geologia Applicata e Idrogeologia: 561-568. Kennedy D. 2000 The Roman army in Jordan. (London). Knibbe D. forthc. Via sacra III. Kornemann E. 1963 Römische Geschichte. Zweiter Band. Die Kaiserzeit. 5. Auflage, bearbeitet von Hermann Bengtson. (Stuttgart). Mattingly D.J. 1988 The olive boom. Oil surpluses, wealth and power in Roman Tripolitania. Libyan Studies 19: 2141.

70

Similarities and differences in the late Roman defense system on the European and eastern frontiers Zs. Visy The Roman empire incorporated the entire region around the Mediterranean Sea and the most significant territories of Europe and Asia. A relief map demonstrates quite clearly the rôle the major mountains and rivers played in the development of its extension. By the end of the era of the Republic, Rome had conquered all its larger neighbours, leaving no significant enemy, except the Parthian kingdom. By this time, at the height of its power, the Roman empire slipped into conflict with itself, and from the crisis of the civil war only the principate could lead it out. The then established Augustan empire set the basis of the state of the imperial period and it was Augustus who defined the optimum extent of the orbis Romanus.

the oldest and most important issues in historical research. The question is – taking into account its huge territory – what was the reason for the fact that the extent of the empire did not change significantly? One can also ask what was the reason for an offensive policy, which was performed by a combination of diplomacy and a strong army with offensive characteristics. How can this contradiction be re-solved? The explanation is to be found in the imperialistic structure of the state, in the strongest and only professional army of its age, in economic interests and in infrastructural elements, and of course in the personal interests and the emperor’s enthusiasm. Isaac (1990) has demonstrated that the main factor was the last one; emperor’s striving for victories, instead of supposing a rational preparation for war (Campbell 1984: 133). This fact was the engine for new and more offensives, their aspiration and their inner compulsion for proving their aptitudes to be an emperor. There is almost no one in the long line of Roman emperors who did not play a decisive role as an initiator of war or offensives to acquire new territories and provinces. So did Claudius, and above all Traianus who mobilised once again the Roman eagles.

This Mediterranean empire penetrated deeper into continental territories in order to occupy them. The Atlantic Ocean and the Black Sea and major rivers such as the Rhine, the Danube or the Euphrates made it possible. Maritime and riverine traffic had the greatest significance for troops and goods in most places, but especially in regions lacking formal roads. According to Tacitus, Augustus admonished his successors not to enlarge the already huge territorial extent of the empire, but only to keep it safe. Some scholars have interpreted the statement addideratque consilium coercendi inter terminos imperii (Anns. I.11), to mean that Augustus favoured a kind of peace policy, but this was not the case. He achieved a relatively high level of social peace and therewith created peaceful conditions for most inhabitants of the empire but he ran an offensive policy for almost all his life. He put the central European region into Roman hands. He wanted to realize his greatest dream, to occupy Germania up to the River Elbe. When he had not succeeded in this when he was an old man he turned to the idea that it was better solely trying to keep and maintain what was already held.

Through the occupation of new territories, huge masses of people came under Roman rule and became inhabitants of an empire. They received some rights and opportunities within the provinces along with the prospect of becoming a Roman citizen. Isaac (1990: 395) underlines the fact that the ancient sources spoke first of all about the subjugation of peoples and not about the occupation of territories. From this fact, the main question for the Romans was the mode and level of integration of the subdued peoples and the occupied territories. A. Mócsy (1959; 1970), G. Alföldy (1964; 1986; 1988), H. Wolff (1997) and other scholars (eg. Balsdon 1979) have examined the most important components of the integration and romanisation from both the Roman and the provincial perspectives. One of the most important questions is what induced the newly subdued peoples to take on Roman customs?; what led them to identify themselves with the political doctrines formulated by a crude Roman imperialism? What was the magic power that made, for example, the Illyrian and Celtic population of Pannonia ready for it? And from what inner compulsion did they, within one or two generations, identify themselves with the same important political principles and aims? The simple, but of course simplified answer is that Rome allowed its inhabitants to follow their way of life in their communities according to their own rules and with a degree of self-government. Rome only siezed control of the most important principle elements but still she demanded strict obedience without delay and condition. Those salient principles were political subjugation, loyalty and taxation for the state. In these activities their partners were not the mass of the subugated peoples in general but rather their aristocracy.

Despite Augustus’ advice, almost every one of his successors tried to enlarge the imperium romanum. It is worth emphasizing, however, that the extent of the Roman empire did not change significantly over the subsequent centuries. This could not have happened by chance. In comparison to its huge territory, the addition of Britannia by Claudius, Dacia by Traianus and the provinces created from the allied kingdoms, can not be regarded as significant enlargements. This point is reinforced by the example of the oriental front where, despite many wars and more territorial changes, the frontiers of the Roman empire do not seem to have been pushed forward during the four centuries following the death of the creator of the principate. The expansion and retreat of the Roman empire is one of 71

Limes XVIII

Thus the limits of the empire depended on numerous factors on both sides. It can be stated that the Roman factors were about largely consistent in the different regions, as portraying the unity of the Roman empire but the factors of the other side were quite different, dependant on the nature of language, culture, economy and their variations. To be effective, Roman policy had to adapt its system to these different structures.

according to their nature as sea-frontiers (ora), riverfrontiers (ripa) or artificial ones (limites). After a change in meaning, the word limes in the imperial age came to signify a fortified road along the frontiers. Isaac (1988; 1990: 408) has shown that limes, in the changed sense did not automatically mean military forces and a military system based on this but that it could mean a relatively free boundary as well, originally in the desert regions of the Near East, where the partly settled, partly nomadic people and the Romans had the same interest in preserving peace1.

The frontiers of the empire demonstrate these differences. The frontier system, its character and role seem not to be identical empire-wide. The similarity of the situation is only superficial on the borders of Europe, Africa and in the orient. Though the peoples living in Europe, outside of the empire were poorly organized, they were numerous and they had been slowly but continuously advancing for centuries, from the north of Europe to the south and from the steppe region to the west. On the European frontiers, Rome had to face an ‘enemy’ that consisted of lesser organized groups who did not form states or coherent political units. The level of their civilization was less developed than that of the Mediterranean world but their number increased quickly. Nor did their movement towards the Mediterranean diminish. They pressed the Roman frontiers more and more and from time to time, when they could unite their power, they overran and occupied Roman provinces.

However, by the C4th, the word limes meant a strip of land on the boundary which was under strong military control. This zone could be so extensive as to cover enormous territories as well. Since E. Kornemann, it bears the name ‘the invisible frontier of the empire’. Rome refused to recognize ideological boundaries to their imperialism or practical limits to their control beyond the administrative borders of the provinces. The political network consisting of different treaties with the neighboring peoples was used by Rome with extraordinary skill to save peace and to control such peoples as dependents. Thus the actual boundary area altogether could make up a wide strip of land, the new provinces and the territories in advance of them; a potentially huge and wide region. The existence of such a region to be controlled within Roman territory is proved by the fact that the Roman army was often stationed deep inside it. It had an interior political reason and it was suspended in most provinces in the Flavian era, as the romanisation and the inner peace of these provinces achieved a higher security. The change to a linear frontier structure ordered the troops next to the boundary and kept the federal system alive. Thus in some cases, only a very narrow strip could exist along the boundary. In the late Roman times the expeditionary army stood in the rear of the limitanei.

In Africa, Rome held a favorable situation. Neither Egypt nor the frontiers of Africa Proconsularis, Numidia or Mauretania were crisis areas which could endanger the security of empire as a whole. Although a large population lived in the Nile valley and western Egypt, they could derive support from anywhere else. For the region’s security was insured by the immense Sahara desert. Therefore invasions of peoples from Africa was never a real danger for Rome.

In the east, a different situation developed. There was no built limes as in Europe and Africa. In Britain, the walls of Hadrian and Antoninus were built across the narrowest sectors of the island in order to exclude barbarians. In Germania, a similar system was built to connect the Rhine and the Danube. The remains of a man-made limes are to be found also in some parts of Dacia and there was a wall system on the Hungarian Plain too. The limes Sarmaticus, as it was named by S. Soproni, was established in the C4th and formed a kind of demarcation line between allied peoples and barbarians under Roman control. Then there was a 250kms long built frontier in Africa: the fossatum Africae, built on the most threatened sector of the province. Other boundaries were river lines as the Rhine, the Danube and the Euphrates. They served first of all as roads connecting for the army and to facilitate trade but later when the linear defense system was developed, they served as true boundaries consisting of a formal road and a chain

As much as the European frontier, the oriental one was of primary importance to Rome. Though its characteristics are apparently similar to those of the European ones, there is in fact a great degree of difference between them. The enemy here was a rival empire, the strong and wellorganized Parthian, later Persian, state whose inhabitants did not want to go to and live in Roman territory, as the Germans and others in Europe did. It was a state with occasional hostility towards Rome but mostly in a static position between east and west. A long series of treaties between the empires paved the way to creating partly peaceful conditions that could be preserved for long periods. The key to this political balance was Armenia whose king was inaugurated by the power which was the more influential at the time. The kind and the degree of the protectorate was always the exact monitor of the real power of both empires. Just as the condition of the frontiers in the orient can not be compared to the European ones on a political level, there was also a significant difference in the structure of the frontiers. The frontiers (termini, fines) could differ

1

Parker 1991 emphasises the military character of the border line. Graf 1997 doubts the existence of a via militaris in the central sector of the limes Arabicus.

72

Zs. Visy: Similarities and differences in the late Roman defense system on the European and eastern frontiers of fortifications as well.2 There was a similarity between ripa and limes, with a difference in the nature of the demarcation line (river or wall). But the aim of both systems was to exclude barbarians from Roman provinces in rather densely populated areas where the establishment of a system of defense would not have been the best option because territory was needed for civil purposes.

again the economy of Roman action and the ability to adapt the system to the political, social, economic and geographical circumstances of the region. This system can only be analyzed by reference to the issues of romanisation and economic and social development, considering the special role of regions and provinces and the link between central government and regions. Through the occupation of new territories Rome governed newer and newer peoples. Isaac is right in stating that Rome did not want to occupy territories but rather to control peoples. The shape and exact extension of the new gained territories played a secondary role. The most important thing was the integration of the conquered peoples, and this was accomplished through a variety of different methods, depending on the mode of the occupation and the strength of the resistance, itself a product of the local social system, to the structure of the Roman settlements and their culture. Last but not least, the federal system established a huge circle of allied nations around the empire where the condition and quality of the federal contacts defined the necessary level of the defense.

Certain long boundaries of the empire were not, or else were only feebly, fortified. At such places, even the exact line of the boundary can not be located. There was only a road running along the border of the Roman territory with some military sites situated in these regions. The system has been comprehensively described by Isaac, who analyzed the desert boundaries in the Near East (Isaac 1993, following Mann 1974: 520). His main points are that there was no corridor-like boundary and the not very populous nomadic tribes lived in close co-existence with provincials. He could not draw a definite line for the frontier but it does not mean that there was no territorial changes through the centuries or that Rome did not take care of its own political interests. His conclusion, however, is valid for the desert regions of the other sectors of the empire and above all for Africa.

All this is relevant for a static system defending an empire. The type and level of it can not be connected to strategic questions of offensives, which concerned primarily the political decision makers: first of all the emperor, but one can not exclude the rational planning as well, a central direction in Roman policy (Potter 1990). “Rome never stopped fighting aggressive wars, even if expansion did slow down” (Woolf 1993: 14), but from the openness of some sectors one may not conclude that the reason for it was a permanent plan of offensives.

One can conclude that the Roman defensive system was adapted to local circumstances in every case and in every sector. One can not speak of a unified system of wellfortified lines but only of a looser or tighter line of military sites along the boundary line. The system was organized not only on the basis of political or military principles but on taking into consideration every other condition. The strength of the military forces depended on the level of danger caused both by provincials not yet pacified and by the barbarians. The proximity of the army to the boundaries of the empire depended partly on the aforementioned reason and partly on the geographical situation as well as on the nature of the contacts to the barbarians. The boundary line was demarcated through a waterway or a built road or even by both. It is the only common feature of the different kinds of Roman frontiers. The seashores and rivers served the traffic first of all but at the same time, they composed a demarcation line as well and in such cases they were fortified and thus became a real barrier. The built limites had the same function in the sectors where one had to reckon with greater barbarian invasions or raiding. These lines had the task of keeping undesired groups out of the empire. Finally, there were sectors where the circumstances did not demand a strong and chain-like military presence and the creation of a demarcation line. For these areas, mostly desert-like, a relatively weak military presence and patrols were sufficient along the roads running on the border of the province.

The examination of strategic issues is mainly due to E. Luttwak’s famous monograph (Luttwak 1979). His approach and analysis of the border protection of the Roman empire was done purely from the military aspect. He advanced the thesis that the Romans envisaged border protection in a grand strategy that evolved in time. At the beginning, there was a forward-looking strategy which incorporated in-depth considerations and the federal system of the C1st. The second phase was a corridor-like defence that corresponded to a linear border protection which took into consideration the federal system as its forebear had done. The third phase was defence-in-depth, created from the C3rd. And yet, defence-in-depth did not use the territory of the allies as an advance guard but the real border territories of Rome were transformed again into buffer zones. It is a characteristic feature of this age that the army was divided, with one part of it stationed in the inner territories where chains of fortified towns were set up. Many researchers have not accepted this conscientious reconstruction of (Roman) border protection, one that assumes the analysis of strategic considerations and actual political situations was paramount (Millar 1982; Isaac 1990). Related to this, it is an important question of to what extent the Romans strived to conquer and protect. The defensive strategy means that an attempt was made to

The physical system of the Roman frontiers demonstrates 2

Against Isaac 1990: 410 rivers did not help only connections but served as real boundaries too: Mócsy 1974: 35 emphasises both characteristics; cf. Dåbrowa 1997.

73

Limes XVIII

preserve the existing territories. The offensive strategy aimed at conquering new territories. It can be shown that besides the occasional smaller conquests, Rome was only very rarely offensive in the age of the ‘high’ empire. The relatively persistent conquests are primarily a characteristic of the early imperial period, while the Roman empire became more and more defensive from the middle imperial period.

obviously aimed at the protection of the enormous area already gained although different emperors and the decisions of particular periods provide the information about offensive attitudes. Thus, there was a contradiction between the concrete decisions as intentions and the reality of the actual changes in the imperial territories. The strength of the Roman empire and of the opposing significant forces, that of the Sarmatians and Germans in Europe and the Parthian and Sassanid empires in the east, gradually came closer to each other in the C2nd and C3rd. After the crisis of the C3rd, a temporarily successful conquest took place only in the reign of Diocletian. The balanced situation, maintained until the third-quarter of the C4th, was broken by the appearance of the Huns and by the Germans escaping from them into the empire (Burns 1994; Bóna 1993). The lessening power of the western part of the empire was definitely liquidated by the Huns moving into the Carpathian basin.

The follow-up of this strategy in archaeological facts can be examined in epigraphic materials, through different military establishments and findings. By knowing thoroughly the eastern front, Isaac draws the conclusion that the other border lines were quite similar (Isaac 1990: 410-415). Nevertheless this is not exactly true, because the indigenous population were little tribes and alliances of tribes in the western part of the empire (Woolf 1990). Isaac basically relies on written historic sources and epigraphic materials whereas the archaeological finds have got a secondary rôle in his train of thoughts. He presumes, for instance, that the selection of the place for military establishments is accidental, often changing, from which no strategic issues can be concluded. However we know that site selection was made very consciously which can be proved by the fact that for centuries most military establishments are to be found at the same locations. Another point is the openness of the eastern border from which he justly draws the not entirely correct conclusion of the offensive character: as there is no border line fixed, the Roman forces could start from here any time as from a stepping stone. In the east, where they faced a static domination which is valid not only for the Persian but also for the Sassanid empire, sometimes leading offensives against the Romans, there was no actually marked border line, but there were Roman offensives also in the C3rd and C4th. It means that the empire was not in defence. They did not do it there to a lesser extent than at other places but the situation is different as in other places. In the area of the Danube and the Rhine Rome was forced for real defence by the C3rd. The barbarians struck these territories and after a while, these areas were given up.

Bibliography Alföldy G. 1964 Bevölkerung und Gesellschaft der römischen Provinz Dalmatien, mit einem Beitrag von A. Mócsy. (Budapest). Alföldy G. 1986 Die römische Gesellschaft. Ausgewählte Beiträge. (Stuttgart). Alföldy G. 1988 Die Romanisierung in den Donauprovinzen Roms. In P. Kneissl & V. Losemann (Hrsg.) Alte Geschichte und Wissenschaftsgeschichte. Festschrift für Karl Christ zum 65. Geburtstag. (Darmstadt): 1-21. Balsdon J.P.V.D. 1979 Romans and aliens. (London). Bóna I. 1993 Die Hunnenreiche. (Budapest). Burns T.S. 1994 Barbarians within the gates of Rome. A study of Roman military policy and the barbarians, ca. 375425 A.D. (Bloomington). Campbell J.B. 1984 The emperor and the Roman army 31 BC AD235. (Oxford). Dåbrowa E. 1997 The rivers in the defensive system of Roman Syria (from Augustus to Septimius Severus). In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 109-114. Dobson B. 1986 The Roman army. Wartime or peacetime army ? In W. Eck & H. Wolff (Hrsg.) Heer und Integrationspolitik, Die römischen Militärdiplome als historische Quelle. Passauer Historische Forschungen 2: 10-25. Graf D.F. 1997 The via militaris and the limes Arabicus. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 123-133. Isaac B. 1988 The meaning of the terms limes and limitanei in ancient sources. Journal of Roman Studies 78: 125-147. Isaac B. 1990 The limits of empire. The Roman army in the east. (Oxford) Isaac B. 1993 An open frontier. In P.S. Brun, S. van der Leeuw & C.R. Whittaker (edd.) Frontières d’empire. Nature et signification des frontieres romaines. Actes de la Table Ronde Internationale de Nemours 1992. Mémoires du Musée de Préhistoire d’Ile-de-France 5: 105-114. Luttwak E. 1979 The grand strategy of the Roman empire.

The issue of military strength is also related to the question what was the rôle of the economically operated Roman mercenary army in high politics. The subsistence of this army cost significant amounts of money but the balanced situation was maintained very nearly until the very end. In case of either an offensive somewhere or an invasion to be rejected, as there was no provision available, supplies had to be taken from other provinces. In these cases an attack could happen again at the temporarily weakened border sections. According to Dobson (1986), the army itself was not a war time institution but a peace time one. The task of the army was to keep the existing territories and not to conquer new ones. It is a contradictory phenomenon that the empire as a whole - according to the available source documents 74

Zs. Visy: Similarities and differences in the late Roman defense system on the European and eastern frontiers

From the first century A.D. to the third. (Baltimore). Mann J.C. 1974 The frontiers of the principate. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II 1. (Berlin): 508-533. Millar F. 1982 Emperors, frontiers and foreign relations, 31 BC to AD 378. Britannia XIII: 1-23. Mócsy A. 1959 Die Bevölkerung von Pannonien bis zu den Markomannenkriegen. (Budapest) Mócsy A. 1970 Gesellschaft und Romanisation in der römischen Provinz Moseia Superior. (Budapest). Mócsy: A. 1974 Pannonia and Upper Moesia. (London). Parker S.T. 1991 The nature of Rome’s Arabian frontier. In V. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 498504. Potter D.S. 1990 Review of B. Isaac 1990. Bryn Mawr Classical Review 1: http://ccat.sas.upenn.edu/bmcr/l/l/12 Whittaker C.R. 1993 What happens when frontiers come to an end? In P.S. Brun, S. van der Leeuw & C.R. Whittaker (edd.) Frontières d’empire. Nature et signification des frontieres romaines. Actes de la Table Ronde Internationale de Nemours 1992. Mémoires du Musée de Préhistoire d’Ile-de-France 5: 133-142. Woolf G. 1990 World-system analysis and the Roman empire. Journal of Roman Archaeology 3: 44-58. Woolf G. 1993 European social development and Roman imperialism. In P.S. Brun, S. van der Leeuw & C.R. Whittaker (edd.) Frontières d’empire. Nature et signification des frontieres romaines. Actes de la Table Ronde Internationale de Nemours 1992. Mémoires du Musée de Préhistoire d’Ile-de-France 5: 13-20. Wolff H. 1997 Civitas Romana. Die römische Bürgerrechtspolitik vom Bundesgenossenkrieg bis zur Constitutio Antoniniana. (Köln).

75

Limes XVIII

76

The Roman frontier in Jordan: An overview S. Thomas Parker pastoral nomads. Hunting is frequently mentioned in the texts (Winnett & Harding 1978: 23). Others refer to launching raids and prayers for booty (MacDonald 1992: 421). Some of the drawings that often accompany these graffiti seem to illustrate scenes of camel raiding (MacDonald 1990). Others depict warriors fighting on foot with bow and arrow, spears, or swords and shields (Winnett & Harding 1978: 7-28; MacDonald 1993: 328).

Introduction Since the Roman frontier in Jordan is largely terra incognita to many colleagues at this congress, my purpose in this paper is to provide a brief summary of the history of the frontier. In particular, I wish to highlight areas of current scholarly debate. After a brief review of the sources, we will proceed in chronological fashion.

With whom were these tribes in conflict? There was surely some inter-tribal warfare. A few texts refer to individuals escaping from the Romans (MacDonald 1992: 331) but there are few texts that could refer specifically to warfare with the Romans. Some do mention conflict with the Nabataeans. The references referring to migration, hunting, raiding, and warfare generally corroborate many details in classical sources.

The sources Until the mid-1970s, our knowledge of the Arabian frontier in Jordan was largely limited to some stray remarks in literary sources, a handful of inscriptions, and some usually cursory site descriptions by western travellers. Not a single Roman military site had ever been excavated in Jordan. The last 25 years has witnessed an explosion of new evidence from archaeological surveys and excavations, aerial photography, as well as new inscriptions and papyri (Fig. 1). Although in quantitative terms the amount of new evidence is impressive, various caveats must be raised. The bulk of the new evidence has been obtained from surveys. Some sectors of the frontier have now been intensively surveyed, but other sectors have received only superficial coverage at best. The number of excavated military sites is still relatively small, the number of extensively excavated sites is even fewer, and those adequately published smaller still. The collected evidence is still dwarfed by comparison to most European frontiers. Nevertheless, the amount of new evidence collected in the last 25 years is significant and has provided a basic outline of the frontier’s history. Needless to say, it has also raised many new questions.

In addition to the epigraphic evidence, recent surveys have also located many sites interpreted as nomadic campsites. Contrary to the notion that nomads are largely invisible archaeologically, literally hundreds of such sites have been identified in both Jordan and the Negev, although few as yet have been excavated (eg. Clark 1987). Some of these sites, offering water and grazing on seasonal basis, seem to have experienced intermittent reuse over long periods. Some are located in proximity to Roman military structures. The historical implications of this proximity may naturally be debated. Finally, recent ethnographic research on nomads generally offers more varied and sophisticated models that may prove useful in understanding the Arabian frontier (Khazanov 1984). These models range from environments where the nomads completely dominate the local sedentary population to situations of complete nomadic subservience to the sedentary folk. For example, it seems clear that most nomads are dependent upon the sedentary population for certain crucial items and thus there is nearly always some economic exchange. But the nature of this exchange depends heavily upon their relative power. The relationship is usually dynamic, often fluid, and may change rapidly due to factors beyond the control of local populations, such as severe drought. A study of this process in the modern Levant exploits extensive documentary sources for this region that simply do not exist for the Roman period (Lewis 1987). These sources reveal that at the beginning of the C19th the Arab nomads were in complete control of the frontier of Syria and Jordan. They grazed their herds throughout areas otherwise suited for dry farming, waged frequent inter-tribal wars over grazing lands and water sources, and raided the small and scattered sedentary population with impunity. But the introduction of Ottoman military forces in the late C19th slowly reversed the situation. The nomads were defeated, forced to abandon lands suitable for cultivation, and accept

The nomads No issue among scholars of the Arabian frontier has proven more contentious than the nature of the relations between the nomadic and sedentary populations. There is much new evidence on this issue. Most important are the thousands of graffiti, often accompanied by drawings, produced by the nomads themselves. These inscriptions, dubbed Safaitic and Thamudic by modern scholars, are typically laconic and notoriously difficult to date. Most specialists date the texts rather broadly, from the C1st BC to the C4th AD (MacDonald 1992: 421). A recent synthesis of the “Safaitic” inscriptions notes that the vast majority are concentrated in the basalt desert of southern Syria and northeastern Jordan, with a few found much further afield (MacDonald 1993: 304). So-called Thamudic texts are concentrated in the south, mostly in the deserts of southern Jordan, Wadi Sirhan, the Hejaz, and the Negev. The Safaitic texts mention seasonal migrations, grazing herds of camels, sheep, and goats, watering places, and encampments. In short, this evidence portrays the tribes as

77

Limes XVIII

some measure of Ottoman authority. Sedentary settlers, including some former nomads, quickly expanded right up to the desert fringe. Are there useful parallels here for the Roman period?

Arabian army was based in the cities of the province, presumably to maintain control over the provincial population and to facilitate supply. Yet actual evidence for such urban bases is slim. Most of the major urban centres of Roman Jordan have been extensively excavated, including the Decapolis cities of Philadelphia, Gerasa, Capitolias, Abila, Gadara, and Pella, as well as other cities such as Esbus, Madaba, and Petra. Yet the evidence for any military presence in all these cities is confined to a few inscriptions usually mentioning only individual soldiers (Kennedy 2000: 104). Research has thus far failed to identify a military quarter, such as at Dura Europus, in any of these cities. And it may also be notable that none of these same cities is listed as garrisoned in the later Notitia Dignitatum (Or. XXXVII).

The Nabataean kingdom Limited space requires perfunctory treatment of the Nabataean kingdom. Recent research in various parts of Nabataea reinforces the notion of a densely populated region, largely integrated into the wider Roman world. The Nabataeans maintained substantial military forces that were posted throughout the kingdom (Graf 1994). They were periodically used against rival client kings and as allies on Roman campaigns. But one wonders about relations between the nomadic tribes and the Nabataeans on their desert frontiers. The lucrative caravan traffic in luxury goods that passed through the kingdom surely required protection from nomadic raids. Recent surveys suggest that some forts along major routes may originate with the Nabataeans (Parker 1986: 115-120). And a few of the so-called “Safaitic” inscriptions may refer to occasional conflicts along the frontier, but the evidence is sketchy at best. The Nabataeans and the nomadic tribes may well have often enjoyed generally peaceful relations in this period. We simply are largely in the dark about this issue.

In contrast, the limited evidence for Roman military deployments in the C2nd attests deployments along the via nova Traiana and towards the desert. There were garrisons at Bostra (Bosra), the provincial capital and legionary base (Sartre 1985), Areopolis (Rabba), where a praefectus equitum is attested in 127 in the Babatha papyri (Lewis in Yadin 1989: no. 16.11), and at Avara (Humeima), where a fort sufficient for a full-strength auxiliary unit or legionary vexillation has been dated by excavation to the early C2nd (Oleson et al 1999). All these settlements lie on the Trajanic road. The location of Avara is particularly noteworthy, since it lies in the Hisma desert and its excavator has estimated the total civilian population of the site at less than 1000. Inscriptions also attest soldiers from legio III Cyrenaica and two alae in the northern Hejaz, even deeper within the Arabian Desert (Speidel 1977: 70306; Sartre 1982: 30-34). The most famous evidence for Roman concern with the nomads in the C2nd also lies in the Hejaz at Ruwwafa (Sartre 1982: 27-29; Bowersock 1975; 1983: 96-97). Here a bilingual inscription from the reign of Marcus Aurelius mentions a temple built by a confederation of the Thamudeni, encouraged by the Roman governor, Antistius Adventus, who “made peace among them.”

The 2nd century This century, so well known on other frontiers, paradoxically remains one of the darkest periods in the history of the Arabian frontier. The motivation behind the Trajanic annexation of Nabataea in 106 remains controversial. Various economic, geo-political, and military explanations have been offered but there is as yet no consensus (Bowersock 1983: 78-82; Parker 1986: 12324; Fiema 1987; Freeman 1996). The possibility of Nabataean resistance to the Roman annexation, long thought to have been minimal, has again been raised by new evidence, such as that obtained by recent excavations at Petra (Schmid 1997). But the use of terminology such as Arabia Adquista (not Capta) on Trajanic coins and the absence of Arabicus from Trajan’s official titles still suggest that resistance was fairly limited.

In short, we are largely ignorant of the security situation in C2nd provincial Arabia. There is no evidence of any internal resistance to Roman rule, in stark contrast to neighboring Judaea. There is also no explicit evidence for any nomadic attacks on the frontier in this period. Yet one may wonder why Arabia remained so strongly garrisoned if there was in fact no perceived security threat. Perhaps the threat of force by Roman units based on the edge of the desert and even in the desert itself, combined with skillful diplomacy such as that implied in the Ruwwafa inscriptions, was sufficient to insure generally peaceful conditions.

The unarguable facts for the C2nd are few. Much of the Nabataean army was incorporated into the Roman auxilia and transferred out of the province. The via nova Traiana was constructed between 111 and 114 between southern Syria and the Red Sea, terminating at Aila (modern Aqaba). The new province of Arabia included most or all of the former Nabataean kingdom. Whether the northern Hejaz (now in north-west Saudi Arabia) was part of the province remains in dispute (Bowersock 1983: 97). The Roman provincial garrison was a middle-sized force among the provincial armies, consisting of one legion and probably 10-12 auxiliary units, perhaps 10,000-12,000 men in all (Speidel 1977).

The Severan period This period witnessed Roman military activity around the key oasis of Azraq, near the north-western outlet of Wadi Sirhan, a major nomadic migration route between southern Syria and the interior of the Arabian peninsula (Kennedy

It is commonplace to suggest that the bulk of the provincial 78

S. Thomas Parker: The Roman frontier in Jordan: An overview

1982). Inscriptions refer to the construction of several forts and extensive road construction as well as the presence of detachments of legio III Cyrenaica, several auxiliary cohorts, and Gothic gentiles in the region (Parker 1986: 129-31). This activity has reasonably been compared to Severan policy in North Africa, which also witnessed an advance deep into the desert. It must be admitted that there is no explicit literary evidence for warfare against the nomads on the Arabian frontier in the Severan period. But such military construction is otherwise difficult to explain. And, as Kennedy reminds us in the handbook for this congress, Wadi Sirhan was known in early modern times as the “Way of Raids” (Kennedy 2000: 52).

tetrarchic inscription from Azraq that attests elements of six Roman legions constructing a military road from Bostra along the entire length of Wadi Sirhan, a distance of over 250 Roman miles, deep into the Arabian peninsula (Speidel 1987; Zuckerman 1994; Bauzou 1996). A number of forts, including the legionary base at el-Lejjun, were constructed east of the Dead Sea. In the south, an apparent legionary base (for legio VI Ferrata?) was erected at Udruh, just east of Petra, and legio X Fretensis was transferred from Jerusalem to Aila to anchor the southern end of the Trajanic road (Parker 2000b: 128-34). Many milestone inscriptions date to this period and several building inscriptions attest the construction of forts. Other forts have been dated to this period by excavation (Parker 2000a: 372-74). The forts fall into three broad categories: legionary bases, auxiliary forts about 1ha in area, and small forts c.30 x 30 – 60 x 60m. Several watchtowers also date to this period.

The 3rd Century The late C3rd was clearly a turning point in the history of the Arabian frontier as elsewhere along the eastern frontier. The collapse of Parthia, the rise of Sassanian Persia, and the meteoric rise and fall of Palmyra brought fundamental change to the region. There is evidence of major nomadic tribal migrations in this period. Along the Arabian frontier it is now the Saracens, described by Ammianus Marcellinus (XXIII.6.13) as “tent-dwelling Arabs”, who reach centre stage. The Saracens remain a shadowy folk. The descriptions in Ammianus, while admittedly biased, were based upon direct observation as both allies and enemies (XIV.4.1-7). He reports that they were completely nomadic with an economy based on pastoralism, hunting and gathering, and brigandage. Mounted on horses and camels, they were more suited for lightening raids rather than set-piece battles (XXXI.16.5). It seems possible that they now combined into larger federations, albeit temporary, that increased their military capabilities. In the C4th we begin to hear of “kings” and “queens” of the Saracens, perhaps reflecting this development. There is also the possibility that adoption of the North Arabian camel saddle enhanced their military prowess (Bulliet 1975). Another scholar has suggested that the use of camels as transport combined with the Arab cavalry horse was a decisive enhancement of Saracen military capabilities (Kuhnen 1991).

Some kind of military build-up under the tetrarchy along the Arabian frontier can no longer be seriously doubted. But scholars continue to debate the causes of this development (Banning 1986; 1987; Parker 1987b; Mayerson 1986; 1989). Several have downplayed the seriousness of any Saracen threat and instead argued for internal security problems, such as a restive sedentary population or brigandage (Graf 1989; Isaac 1992). Although I cannot rehearse this debate here, I do wish to stress one fundamental fact that strongly suggests that the perceived threat was external, not internal. The most compelling evidence for the nature of the principal security threat in the C4th is the Notitia Dignitatum. The chapters covering the eastern frontier are widely agreed to reflect the deployments of Diocletian and the C4th. If the principal security threat was internal, one would expect to find Roman garrisons deployed throughout Palestine and Arabia in major population centres, in the agricultural heartland, and at key points along major interior roads to police the provincial population and guard against brigandage. Further, basing military units in the agricultural heartland would also greatly facilitate their supply. But analysis of their deployments in Arabia and Palestine reveals otherwise. Although the deployment of some units remains unknown, most are located either with certainty or with some degree of confidence either to a specific site or at least to a limited area. The map (Fig. 2) illustrates the deployment of these units, some of which are conjectural.

Diocletian and the 4th and 5th centuries Diocletian, in c.295, partitioned the province of Arabia and assigned the southern portion to Palaestina. The northern area remained as a truncated province of Arabia. Henceforth a dux commanded the military forces in each province. How serious was the Saracen threat in this period? Only a brief summary of the evidence is possible here. Above all is the campaign waged by Diocletian himself against the Saracens in Syria in 290 (Latin Panegyrics 11.5.4, 7.1). This was followed by construction of the strata Diocletiana, a fortified road through the desert between Damascus and the Euphrates via Palmyra (Poidebard 1934: 27-94; van Berchem 1952: 10-17; Konrad 2001). Farther south is evidence of extensive military construction from the Azraq oasis to Sinai (Parker 1986: 135-43). Of particular interest is a fragmentary

In Palestine a group of elite cavalry units (equites Illyriciani) is deployed on an east-west line from south of Gaza to the Dead Sea, on the northern edge of the Negev desert. It is notable that Palestine north of this line, including nearly all the major cities, is practically denuded of garrisons. A similar picture emerges east of the Jordan Rift Valley, where the populous and urbanized areas of Jordan also lack garrisons. Instead, the vast majority of the military units 79

Limes XVIII

are based either in the desert or close to the desert fringe either on or just east of the Trajanic road. This includes all three legions, most élite cavalry vexillations, and many alae and cohortes. In short, the disposition of both Roman military forces and fortifications in the C4th clearly suggests that the perceived threat was external, from the desert, and not from within the agricultural zone.

Thus it is striking to note that the numismatic corpus of over 700 coins from el-Lejjun ends with issues of Justinian, supporting this assertion (Betlyon 1987; forthc.). Other archaeological evidence from the Arabian frontier also supports the literary sources. The latest military building inscription from Arabia dates to 529, or a full century before the end of imperial rule (Parker 1986: 32). More significant is evidence for the widespread abandonment of forts along the frontier, including all five forts excavated east of the Dead Sea (Parker 1986: 149-55; 1991: 134-42). Of special importance in this region is the fact that none of the many watchtowers in this sector, the "eyes and ears" of the frontier system, yielded any evidence of C6th or early C7th occupation (Parker 1987a: 819-23). Some military sites with evidence of continued occupation in the C6th may have been converted to monastic or other civilian use, such as the so-called “Barracks” at Umm el-Jimal (Parker 1998a). The defenses of the Syrian frontier south of the Euphrates also appear badly weakened in this period (Liebeschuetz 1977).

From the nomadic perspective the Roman military buildup along the frontier was probably seen simply as naked aggression into their traditional grazing lands and water sources. The recent regional surveys agree that the C4th and C5th were a high water mark of sedentary settlement along the frontier. One scholar has perceptively suggested that the extension of agriculture to the desert fringe might have resulted in the exclusion of the nomads, rather than in symbiosis with the sedentary population (Villeneuve 1989). This expansion to the desert fringe may itself have exacerbated conflict. The archaeological evidence of intensified settlement that extended right up to the edge of the desert strongly suggests that the security system developed by Diocletian functioned successfully through the C4th and C5th. This is not to deny periodic conflicts with the Saracens, as attested by literary sources in this period. Several inscriptions attest continued military construction in the late C4th and early C5th (Parker 1986: 143-46). The construction of the city wall at Aila (Aqaba) around the turn of the C5th, for example, may reflect concern about Saracen raids (Parker 1998b: 383-85; 2000c: 386-90). Throughout this period the Romans continued to rely not only upon regular military forces but also upon the time-tested methods of diplomacy and financial subsidies. Some Saracens were incorporated as distinct units into the regular Roman army, as seen in the Notitia and other sources. Other Saracen leaders were recognized as phylarchs who provided foederati under treaty with Rome (Shahîd 1984a; 1984b; 1989; 1995). Each phylarch was paired with a dux in each province. Together they were supposed to insure local security against hostile Saracens and other threats.

This is not to suggest that there were no longer any regular Roman military forces on the frontier in the C6th. The newly discovered Petra papyri of this period mention several soldiers and some places undoubtedly remained garrisoned (Gagos & Frösén 1998). The C6th papyri from the military unit at Nessana in the Negev suggest that these soldiers actually lived outside their fort in the neighboring village and engaged in various civilian occupations in the C6th (Kraemer 1958: 19-24). A few new fortifications may even have been constructed. Magness, for example, now redates the construction of two fortlets in southern Palestine to the C6th (Magness 1999). But both these fortlets are too small (c.20 x 20m) to have contained any significant number of soldiers and both lie north of the main military frontier as defined by the line of posts manned by equites Illyriciani. The evidence suggests that only minimal Roman forces remained along the frontier, especially after the Persian invasion and occupation of the region between 613 and 628. There was simply too little time to rebuild any effective frontier defense before the Muslim conquest a few years later.

Justinian and the end of the Roman frontier Conclusion A decisive moment in imperial policy on the Arabian frontier came in the early C6th under Justinian. Procopius claims that Justinian, prompted by the devastating Lakhmid raids on the eastern frontier, created a Ghassanid monarchy over most of the Roman Saracens to counter the Lakhmids, who led Saracens allied to the Persians. Apparently the divided Roman military command was no match for these Saracens united under the Lakhmid king (Procopius Bellum Persicum 1.17.45-48). Then, after concluding the so-called Eternal Peace’ with Persia in 532, Justinian demobilized many eastern limitanei (Procopius Anecdota 24.12-14). Admittedly, this last assertion is contained in the Secret History, so strongly biased against Justinian that any such allegation must be evaluated independently.

When assessing the last quarter century of research on the Arabian frontier, one is struck by how much we have learned, particularly compared to the prior state of knowledge. Large portions of the frontier have now been intensively surveyed, a number of military and civilian sites have been excavated, and many new documentary sources (especially inscriptions) have come to light. We now have access to relatively complete aerial photographic coverage of the region. All the new evidence has engendered several noteworthy syntheses and vigorous scholarly debate. But a closer look reveals how far we have to go, particularly when compared to other imperial frontiers. The totality of the evidence is still relatively small. Large portions of the frontier remain only superficially surveyed at best. Only limited soundings have 80

S. Thomas Parker: The Roman frontier in Jordan: An overview

been conducted at most excavated sites. Much of the recent fieldwork remains unpublished. Nevertheless, in terms of modern scholarship the Arabian frontier, with all its enormous potential, has begun to enter the light of day.

V. (Mainz). Kraemer C. J. (ed.) 1958 Excavations at Nessana. v. 3: Nonliterary papyri. (Princeton). Kuhnen H.-P. 1991 Der Sarazenensattel: Zu den voraussetzungen der Sarazeneinfalle am Limes Arabiae. In V.A. Maxfield & M.J. Dobson (edd.) Roman frontier studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 326-334. Lewis N.H. 1987 Nomads and settlers in Syria and Jordan, 1800-1980. (Cambridge). Liebeschuetz W. 1977 The defences of Syria in the sixth century. In Studien zu den Militärgrenzen Roms II. (Cologne): 487-499. MacDonald M.C.A. 1990 Camel hunting or camel raiding? Arabian Archaeology & Epigraphy 1.1: 24-28. MacDonald M.C.A. 1992 Inscriptions, Safaitic. The Anchor Bible Dictionary 3. (New York): 418-423. MacDonald M.C.A. 1993 Nomads and the Hawran in the late Hellenistic and Roman periods: reassessment of the epigraphic evidence. Syria 70: 303-413. Magness J. 1999 Redating the forts at Ein Boqeq, Upper Zohar, and other sites in SE Judaea, and the implications for the nature of the Limes Palaestinae. In J.H. Humphrey (ed.) The Roman and Byzantine Near East. Vol. 2. Journal of Roman Archaeology Supplementary Series 31: 189-206. Mayerson P. 1986 The Saracens and the Limes. Bulletin of the American Schools of Oriental Research 262: 35-47. Mayerson P. 1989 Romans and Saracens: micro-macro relationships. Bulletin of the American Schools of Oriental Research 274: 71-79. Oleson J.P., `Amr K., Foote R., Logan J., Reeves M.B. & Schick R. 1999 Preliminary report on the alHumayma Excavation Project, 1995, 1996, 1998. Annual of the Dept. of Antiquities of Jordan 43: 411450. Parker S.T. 1986 Romans and Saracens: a history of the Arabian frontier. (Winona Lake, Ind.). Parker S.T. 1987a History of the Roman frontier east of the Dead Sea. In S.T. Parker (ed.) The Roman frontier in central Jordan. Interim report on the Limes Arabicus Project,1980-1985. British Archaeological Reports Int. Ser. 340 (2 vols). (Oxford): 793-823. Parker S.T. 1987b Peasants, pastoralists, and Pax Romana: A different view. Bulletin of the American Schools of Oriental Research 265: 35-51. Parker S.T. 1991 Preliminary report on the 1989 season of the Limes Arabicus Project. Bulletin of the American Schools of Oriental Research Supplement 27: 117-154. Parker S.T. 1998a The later castellum (‘Barracks’). In B. de Vries (ed.) Umm el-Jimal: a frontier town and its landscape in northern Jordan. Vol. I: Field work, 19721981. Journal of Roman Archaeology Supplementary Series 26: 131-142. Parker S.T. 1998b The Roman Aqaba Project: The 1996 campaign. Annual of the Dept. of Antiquities of Jordan 42: 375-394. Parker S.T. 2000a The defense of Palestine and Transjordan from Diocletian to Heraclius. In L.E. Stager, J.A. Greene & M.D. Coogan (edd.) The archaeology of Jordan and beyond: essays in honor of James A. Sauer. Studies in the archaeology and history of the Levant 1. (Winona Lake, Ind.): 367-388. Parker S.T. 2000b Roman legionary fortresses in the east. In R.J. Brewer (ed). Roman fortresses and their legions: Papers in honour of George C. Boon. Occasional

Bibliography Banning

E. B.

1986

Peasants, pastoralists and Pax

Romana: Mutualism in the southern highlands of Jordan. Bulletin of the American Schools of Oriental Research 261: 25-50. Banning E. B. 1987 De Bello Paceque: A reply to Parker. Bulletin of the American Schools of Oriental Research 265: 52-54. Bauzou T. 1996 La praetensio de Bostra à Dumata (el-Jowf). Syria 73: 23-35. van Berchem D. 1952 L’armée de Dioclétien et la réforme constantinienne. (Paris). Betlyon J.W. 1987 Coins, commerce, and politics: Coins from the Limes Arabicus Project, 1976-1985. In S.T. Parker (ed.) The Roman frontier in central Jordan: Interim report of the Limes Arabicus Project, 1980-1985. (2 vols). British Archaeological Reports Int. Ser. 340 (ii). (Oxford): 655-689. Betylon J.W. forthc. The Coins. The Roman frontier in central Jordan: Final report on the Limes Arabicus Project 1980-1989. (2 vols.). (Washington). Bowersock G.W. 1975 The Greek-Nabataean bilingual inscription at Ruwwafa, Saudi Arabia. Le monde grec: Hommages à Claire Préaux. (Brussels): 513-522. Bowersock G.W. 1983 Roman Arabia. (Cambridge, Mass.). Bulliet R.W. 1975 The camel and the wheel. (Cambridge). Clark V.A. 1987 The Desert survey. In S.T. Parker (ed.) The Roman frontier in central Jordan: Interim report of the Limes Arabicus Project, 1980-1985. British Archaeological Reports Int. Ser. 340 (i). (Oxford): 107163. Fiema Z. 1987 The Roman annexation of Arabia: a general perspective. The Ancient World 15: 25-35. Freeman P.W. 1996 The annexation of Arabia and imperial grand strategy. In D. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 91-118. Gagos T. & Frösén J. 1998 Petra Papyri. Annual of the Dept. of Antiquities of Jordan 42: 473-82. Graf D.F. 1989 Rome and the Saracens: reassessing the nomadic menace. In T. Fahd (ed.) L’Arabie préislamique et son environment historique et culturel: Actes du Colloque de Strasbourg 24-27 Juin 1987. (Leiden): 341-400. Graf D.F. 1994 The Nabataean army and the Cohortes Ulpiae Petraeorum.. In E. Dabrowa (ed.) The Roman and Byzantine army in the east.. (Kraków): 265-311. Isaac B. 1992 The limits of empire: The Roman army in the east. (Oxford). Kennedy D.L. 1982 Archaeological explorations on the Roman frontier in north-east Jordan. British Archaeological Reports Int. Ser. 134. (Oxford). Kennedy D. 2000 The Roman army in Jordan. (London). Khazanov A. M. 1984 Nomads and the outside world. (Cambridge). Konrad M. 2001 Der Spätrömische Limes in Syrien. Archäologische Untersuchungen an den Grenzkastellen von Sura, Tetrapyrgium, Cholle, und in Resafa. Resafa

81

Limes XVIII

Papers of the Society of Antiquaries of London 20: 121-138. Parker S.T. 2000c The Roman Aqaba Project: The 1997 and 1998 Campaigns. Annual of the Dept. of Antiquities of Jordan 44: 373-394. Poidebard A. 1934 La trace de Rome dans le désert de Syrie. 2 vols. (Paris). Sartre M. 1982 Trois études sur l’Arabie romaine et byzantine. Collection Latomus 178. (Brussells). Sartre M. 1985 Bostra, Des origins à l’Islam. (Paris). Schmid S. G. 1997 Nabataean fine ware pottery and the destructions of Petra in the late first and early second century AD. In G. Bisheh, M. Zaghoul & I. Kehrberg (edd.) Studies in the History and Archaeology of Jordan VI. (Amman): 413-420. Shahîd I. 1984a Rome and the Arabs. A prolegomenon to the study of Byzantium and the Arabs. (Washington DC). Shahîd I. 1984b Byzantium and the Arabs in the fourth century. (Washington DC). Shahîd I. 1989 Byzantium and the Arabs in the fifth century. (Washington DC). Shahîd I. 1995 Byzantium and the Arabs in the sixth century. (Washington DC). Speidel M.P. 1977 The Roman army in Arabia. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II.8. (Berlin): 687-730. Speidel M.P. 1987 The Roman road to Dumata (Jawf in Saudi Arabia) and the frontier strategy of Praetensione Colligare. Historia 36: 213-221. Villeneuve F. 1989 Citadins, villageois, nomades. Le cas de la Provincia Arabia. Dialogues d’histoire ancienne. 15.1: 119-140. Winnett F. V. & Harding G. L. 1978 Inscriptions from fifty Safaitic cairns. Near and Middle East Series 9. (Toronto). Yadin Y. 1989 The documents from the Bar-Kokhba period in the Cave of Letters: Greek Papyri. Ed. N. Lewis. (Jerusalem). Zuckerman C. 1994 Aur. Valerianus (293-305) et Fl. Severinus (333), commandants en Arabie, et la forteresse d'Azraq. Antiquité Tardive 2: 83-88.

82

S. Thomas Parker: The Roman frontier in Jordan: An overview

Fig. 1. Map of the Arabian frontier.

83

Limes XVIII

Fig. 2. Map of the Roman frontier in the provinces of Arabia and Palestine in the C4th. The location of some units is conjectural.

84

Galerius and the eastern frontier1 Bill Leadbetter ports of Berenike and Myos Hormos, the former in particular. The Coptus-Berenike road passed porphyry quarries, emerald mines and gold mines. An inscription discovered at Coptus lists the costs of travel permits to the Red Sea ports (Petrie 1896: 27f.). Desert wayfarers included sailors, artisans, wagons with various loads including masts, yardarms, corpses and women. Prostitutes were charged a considerable sum, indicating the profitability of a trade plied in remote communities.7 A series of ostraka lists some of the articles flowing eastwards through Coptus as a consequence of the trade: wheat, barley, oil, hemp, skins, bullion and coins.8 The natural consequence of this had been that Coptus had held a place as an important centre of Roman trade. Its wealth was reflected in great buildings and embellishments.9 It was also the home to a garrison. Troops from the III Cyrenaica were permanently stationed there until they took their expertise in desert patrols to Arabia Petraea (CIL III Supp 2. 13580). They were replaced by a unit of Palmyrene horse archers.10 Other detachments came through from time to time to build or repair the fortifications along the desert road (CIL III 6627; Bernard 1984: 178 no. 56).

Convention has placed Galerius on the Danube between 293 and his Persian campaign, whence he returned to his European duties.2 Based upon the political description of the tetrarchy provided by Aurelius Victor, there had been little reason to doubt the accuracy of this proposition until, in 1976, T.D. Barnes demonstrated that Galerius had spent the years 293 - 299 in the eastern provinces. Barnes’ arguments were based upon the victory titulature of the tetrarchs (Barnes 1976: revised 1982: 27; Arnaldi 1972), but were confirmed in some part by the publication of a papyrus which attested the presence of Alogius, an adiutor memoriae in Galerius’ comitatus in Egypt in late 293.3 At first glance, Galerius’ foray into Egypt seem puzzling, a local matter at most. Two Egyptian towns defied the empire. Jerome’s Chronicle, reports tersely: Busiris et Coptos contra Romanos rebellantes ad solum usque subversae sunt4 Busiris and Coptos, rebelling against the Romans, were utterly destroyed Yet we have some evidence of significant military investment. Papyri attest equites promoti of the emperors in Egypt on 23 December 293.5 Three legionary vexillationes and an imperial protector are attested at Oxyrhynchus in January 295. These were drawn from the IV Flavia, the VII Claudia and the XI Claudia.6 Another papyrus document records the presence of a detachment of the V Macedonica (P. Oxy. 2953). An emperor with a significant number of legionary vexillationes seems excessive in the light of the meagre literary testimionia. Of the two towns attested as in revolt, both were in the Thebaid. Coptus is well known; Busiris eluded identification until Bowman re-identified it with ‘Boresis’, a town of the Thebaid known from the Cornelius Gallus inscription (P. Erlangen 52).

But the late C3rd brings evidence of desuetude. Piracy on the Red Sea had already been a problem in the C2nd. The eastern trade itself seems to have dried up by the C3rd (Sidebotham 1989: 385-513). The latest evidence of imperial euergetism in Coptus dates from the time of Caracalla.11 Shabby and down-at-heels after decades of recession, it still remained of great strategic and economic significance. The southern sea-route through Berenike or Myos Hormos by-passed the Persian empire entirely, avoiding both a number of tariff barriers and the closure of borders in time of war (Warmington 1974: Ch. I, and the map in the endpapers). Its strategic location was one which astute traders had spotted and exploited. In particular, the influence of Palmyrene merchants had been considerable in Coptus where their interests were guarded by a troop of Palmyrene cavalry. Inscriptions attest their presence at both ends of the Red Sea road: both at Coptus and at Berenike. Moreover, more recent excavation has disclosed an entire temple to the Palmyrene gods at Berenike.12 Just

It is fair to say that Coptus had been a place of considerable value. It stood at the head of the Wadi Hamammat and commanded secure routes to the Red Sea 1

This version of the paper is substantially as delivered, although with minor revisions. I am grateful to a number of congress participants, in particular Walter Cockle, David Graf, David Kennedy, Anne Haeckl, Ariel Lewin, Tom Parker and Steve Sidebotham for their comments and advice. In addition, I am grateful for the advice and assistance of Catherine Arends. Such imperfections and infelicities as remain are my doing, not theirs. 2 For example: Mócsy 1974: 268f.; Seston 1946: Ch. VI; Mattingly 1939: 334f.; Ensslin 1930: 2518-2521; Altenorf 1950: 786-7; Jones 1964: 39. 3 SB 18.13851; see Rea, Salomon & Worp 1985. 4 Jerome, Chron. p. 226 (Helm), s.a. 293; see too Zonaras, 12.31; Cedrenus, p. 226D (SHB); Theophanes, Chron. s.a. 282. The curtness of this reference has led some scholars to identify the earlier revolt in Upper Egypt with the later one based in Alexandria: Seston 1946: 141-2; also Chastagnol 1982: 102. 5 P. Grenfell II, 110. The document is dated VII Ides of January 293. 6 P. Oxy. 43. recto; see Bowman 1978: 27; van Berchem 1952: 105f.

7

Bernand 1984: 199f, no. 67; see too Johnson 1936: 593f.; the cost for wives was twice that of their menfolk; the cost for prostitutes was 12 times the average cost for a man. 8 Bernand 1984: 391f., no. 230. For the instability of bullion prices in Coptus, P. Geiss. 47. 9 See Weill Koptos 1911: 110f.; Petrie 1896: Ch. IV. The latest piece of sculptural embellishment is a head of Caracalla. 10 See Petrie 1896: 33; also Reinach 1911: 46-50 lists a number of Palmyrene grave stelae found at Coptus. 11 Sidebotham 1986: 166. 12 On the inscriptions from Coptus, see Bernand 1984: 85, 103; from Berenike, see Verhoogt 1998: 193–198; Wendrich 1998: 243-252, 248. I am indebted to Professor A. Haeckl of Kalamazoo College, who presented a report of the excavation of the Berenike temple to the Congress, for both fruitful discussions and an advance copy of her paper.

85

Limes XVIII

as Palmyrene patrols had ensured the safe passage of commerce through the deserts of Arabia, so too they ensured the orderly conduct of traffic along the CoptusBerenike road (Raschke 1978: 643-5; Matthews 1984: 166-169; Isaac 1992: 144-6).

(Zosimus 1.70; P. Oxy 43. recto.). This would seem to indicate a more general remit in Egypt: to find a workable solution to the problem of regional disloyalty and commence the renovation of infrastructure which had fallen into decay.

Aurelian’s destruction of Palmyra must have had its effect. Palmyrenes had been active in Egypt. Once the protective penumbra of Palmyra was withdrawn, traders had to shift for themselves or find alternative routes. The consequence in Egypt was that towns took matters into their own hands. After the death of Aurelian, Coptus’ neighbour, Ptolemais, allied itself with the Blemmyes and attacked Coptus. The town was only rescued by imperial intervention from Probus (Zosimus 1.71.1).

Some structural changes were made. Coptus was, at least briefly, removed from the equation. Jerome attests its complete destruction. This is something of an overstatement, however, since it is referred to in the correspondence of Paniskos.14 Perhaps what is called here Coptus is the fort attested elsewhere as “Potecoptos”15 (ie: “once-Coptus” or “formerly Coptus”), where there was certainly a garrison of mounted archers at this fort under the command of a praepositus (Bowman 1978: 27; P.Oxy. 2673.9).

If Egyptian commerce could not be safeguarded after the fall of Palmyra, then the situation in Arabia was even more acute. By the end of the 280s the nomadic tribes of the Arabian desert, known collectively as Saraceni, posed a particular threat to ordered commerce and on at least one occasion, Diocletian himself was obliged to mount an expedition against them (Pan. Lat 3 (11) 4.4; Parker 1986: 136; for the date, see Barnes 1982: 51). It may be in this context that the legio X Fretensis was moved from Jerusalem to Aila (modern Eilat-Aqaba) at the head of the Red Sea. In noting its presence there, Eusebius stated of Aila:

While the fortress provided a base for desert patrols, a new economic centre was needed. The new town of Maximianopolis served this purpose, its location probably ancient Caenopolis, modern Qena (Meredith 1953: 132). Nomenclature renders the nature of its renascence clear; “Maximian” is Galerius’ regnal name. His newly named town still existed in the C6th.16 Ancient Kainopolis had already been employed as a base for the soldiers and masons of Mons Claudianus and Mons Porphyrites. Prized and decorative grey granodiorite and ruby porphyry had been embarked from its quays for the embellishment of buildings all over the empire (Cuvigny 1998). The road to the quarries did not terminate there, but went on to reach the coast at a fortified port facility on the Red Sea now called ‘Abu Sha’ar (Sidebotham, Zitterkopf & Riley 1991). This fortress was long thought to be of Ptolemaic date, but it has been recently shown that it was constructed in the last decade of the C3rd (Sidebotham, Riley Hany, Hamroush & Barakat 1989: 133-146). Numismatic evidence from the site provides Alexandrian coinage of Galerius struck before the monetary reform of 296.17 The desert road is fortified on the model of a desert limes, and carried traffic between the Red Sea and the Nile and also connected the Nile with the porphyry quarries (Sidebotham 1991: 494-498; 1994: 158).

“Ailam at the far bounds of Palestine, by the southern desert and the Red Sea, which is sailed by those coming from Egypt and India.”13 Eusebius’ observation illustrates the economic significance of the southern route and the need to protect it. It is in this context that the expedition to Coptus must be perceived. Moreover, Coptus’ economic significance went far beyond its historic place in the eastern trade. The eastern desert of Egypt was a major quarry for the empire. Monumental buildings were embellished with the grey granodiorite from the Mons Claudianus; imperial statues were carved from the rich red porphyry of the Mons Porphyrites; luxuriant emeralds and other precious stones were hewn from the Mons Smaragdus (Meredith 1952; Peacock 1992; Shaw, Bunbury & Jameson 1999). It was not simply an operation designed to quell the disloyalty of distant peoples. It was an assertion of control over key resources and trade routes.

Galerius’ expedition to Upper Egypt cannot be seen as an isolated incident. While there are particularities unique to the situation, when placed in a wider context it encompasses far more than a merely punitive expedition 14 P. Mich. 214ff. It is worth noting that the letter is from Paniskos, at that time a rebel in the army of the corrector Achilleus. As such, the use of the name “Coptus” may be a deliberate archaism. Athanasius lists the city as a bishopric in the middle of the C4th (cf. Lallemand 1964: 104f.) and Phoibammon, the Bishop of Coptus city was a signatory to the acta of the Council of Ephesus in 431 (Amelineau 1893: 214). 15 P. Beatty Pan. II, 162; Skeat 1964: 145 also notes the existence of a kastra Koptou (Stud. Pal. xx. 84 verso 1.5). Skeat’s conjecture as to the origin of the name does not take into account the adjacent and newlyfounded Maximianopolis. 16 Two cavalry units are recorded as stationed there in the Notitia Dignitatum (Or. 31.29, 48), see Kees RE XIV: 2484f. 17 The authors note (p.144) that, so far, this coin is unique, being the only evidence that the Alexandrian mint struck for Galerius.

Just as Saraceni needed to be controlled in the Syrian desert, so too did the desert nomads of Upper Egypt. The Blemmyes had already assisted in rebellion at Ptolemais, and Galerius’ expedition did not concentrate upon the Thebaid alone but seems also to have visited Elephantine 13 Eusebius Onomastikon (E. Klostermann, ed.), p. 6; on the date of the Onomastikon, see Barnes 1982: 106-111, a dating most recently accepted by Millar 1993: 175f. Roll 1989: 239-299, prefers a Constantinian date for the transfer of the legion based on the absence on inscriptional evidence from the tetrarchic period. He does not, however, challenge Barnes’ earlier dating for the Onomastikon, now well accepted.

86

Bill Leadbetter: Galerius and the eastern frontier

against a distant and rebellious region. It was, instead, a part of a process which was already under way to renovate and revitalize the mercantile economy and its infrastructure throughout the Roman east. This was a process commenced by Diocletian with his campaign against the Saracens, in which context, perhaps, the X Fretensis was redeployed to Aila.18 It was continued by Galerius in Egypt and, in the following 18 months or so, in Roman Palestine and Syria.

frontiers. The eastern provinces were of critical concern to Diocletian. He had become emperor in the east in the aftermath of an unfinished war; once free to, he had spent considerable time in Palestine and Syria. In 286, he was in Palestine.19 In 287, he was back in the east, possibly residing at Antioch.20 In May 290, he was in Palestine again, overseeing his campaign against the Saracens (Barnes 1982; Pan. Lat. 1.9.1). After this, he appears to have remained on the Danube detained by projects there.21 A series of inscriptions from Transmarisca, Durostorum and Kladovo attest a renovation of frontier fortifications at about this time. Diocletian certainly visited these places in the course of 294.22 To this journey ought to belong the renovation of Danubian fortresses. These rebuilding activities were completed at varying times, but the first seems to have been either Durostorum or Kladovo, which were completed before the Persian victory, followed by Transmarisca soon after it.23 The victory title Carpicus Maximus does not appear on any dedicatory inscriptions, which might indicate an uneasy peace and a diplomatic settlement in the area.24 Along with these specifically attested frontier sites, there is evidence of similar work at 8 other sites along the Danube at this time (Gudea 1974: 173-181).

A heavily scrutinised inscription from Qasr al-Azraq, in north-east Jordan, attests detachments from the IV Flavia, the VII Claudia and the XI Claudia engaged in roadworks (Kennedy & MacAdam 1985: 100-104; Speidel 1987; MacAdam 1989; Lewin 1990: 152f.). Vexillations from the same legions are attested at Oxyrhynchus in January 295 (P. Oxy 43. recto; see Ensslin 1952; Bowman 1978: 27f; Rea et al 1985: 108). The appearance of such vexillations in two adjacent locations within the same timeframe cannot be simple coincidence. The likelihood is that these were the same units, and that they belong to Galerius’ mobile field army. The inscription from Qasr al-Azraq attests works on a network of roads between centres on the edge of Roman Palestine. From there, roads ran north and west into Roman territory. From there also ran a long and important route through the Wadi Sirhan to Dumata (modern Jawf). From there, this road continued on to the Persian Gulf, and was used in antiquity as both a military road and a trade route (Speidel 1987: 213). In many ways, the situation in this region was as vulnerable and important as that on the Red Sea coast. Both regions carried considerable long-range caravan traffic; both regions had lacked a strong military presence to ensure the safe conduct of trade. The soldiers stationed here were not guarding against Sassanid invasion, but serving as patrol officers against predation.

Galerius’ task then, was to continue work in the eastern provinces already commenced by Diocletian. Little had been done for years and there was much to accomplish.25 While milestones have been recorded which identify roadworks in the region dating to before 293, the bulk of

19

See Barnes 1982: 50f., n.25. Barnes 1982: 51; Malalas records that Diocletian built a palace at Antioch (Chron. 305). 21 Barnes 1976: 186-7 (1982: 63) suggests on the evidence of the victory titulature that in 295, Galerius was already campaigning in Persia. The Persicus title in question, however, was assumed by Diocletian after his diplomatic victory of 287. The title appears in 290 (ILS 618), thus predating (and obviating) an extra and otherwise unattested campaign in 295. 22 Barnes 1982: 63. Diocletian was at Transmarisca on 18 October, and at Durostorum on the 21st and 22nd. Diocletian would have been in the vicinity of Kladovo between the 5th and 8th, perhaps resting there on the evening of the 6th. 23 The Durostorum and Transmarisca inscriptions are so similar as to be virtually identical, the only difference in their wording being that the Durostorum inscription has Gothici Max where the Tramsmarisca inscription records Persici Max II (on the Durostorum inscription, see Brennan 1984: 144). Both of these inscriptions record Sarmatici Maximi IIII and Germanici Maximi V, which dates them to the end of 299 or beginning of 300, after Galerius’ campaign against the Sarmatians of 299, and before his campaign against the Marcomanni later in that year (see below). The Kladovo inscription cannot be securely dated, because it records only victory titles taken against the Germans and Sarmatians, and those without numeration. 24 Brennan 1984 suggests that the apparent abandonment of Gothicus Maximus as a victory title was the consequence of a diplomatic agreement with the Goths (loc. cit.). The same policy may have applied here in Carpic territory. 25 The previous great fortification of this frontier was under Septimius Severus and although attempts were made in the intervening years to strengthen the frontier, it was left to Diocletian to actually achieve this (see Bowersock 1983: Ch. VIII, 131ff.). 20

Further evidence of Galerius’ involvement in the renovation of the eastern frontier comes from nomenclature. Less than 100kms north of Qasr al-Azraq lies ancient Saccaeum. During the tetrarchic period it was elevated to the status of a city and renamed Maximianopolis (SEG VII (1927) 1055; Millar 1993: 184, 543f.). As in Egypt with Kainopolis, Galerius took an opportunity to stamp his own name upon the geography of the empire, thereby creating a memorial of sorts of his own contribution to Diocletian’s greater work. Indeed, a third place also bears Galerius’ name: Capharcotna in the Jezreel Valley, the settlement which had accumulated around the sometime camp of the VI Ferrata, also took the name of Maximianopolis (It. Burd. 586.3; Avi-Yonah 1976: 170; Isaac 1992: 432f.). These activities of Galerius which can be inferred from this slight evidence were not isolated, but part of a grander plan which Diocletian was implementing for the eastern 18 Isaac 1998: 71-74 has recently suggested that a newly discovered road through the Wadi Aravah was constructed in connexion with this transfer.

87

Limes XVIII epigraphic material is tetrarchic.26 Dedicatory inscriptions from completed fortifications belong to the first tetrarchy, except that at Deir el-Khaf which belongs to the second.27 A new legion, the IV Martia, was raised and stationed at the site now called Lejjun.28 Tetrarchic milestones are to be found on the roads running between Phaeno and Damascus (CIL III 197); Philippopolis and Petra (CIL III 14149, 34, 36, 54b (?)); Damascus and Galilean Maximianopolis (AE 7.145; 33.144, 145; 36. 145); Palmyra and Bostra (AE 31. 101–110); and between Emesa and Damascus (AE 34.262). In addition, the fortress at Palmyra was probably completed during this period (CIL III. 6661). Diplomatically, links with the Lakhmid Arabs and their king ‘Amr ibn ‘Ali were perhaps strengthened.29 Fortifications were also built and refurbished on the road from Damascus to Palmyra and, at some stage, on the roads from Palmyra and Sura. From the evidence of the milestones and the Palmyrene inscription of Hierocles, it most likely occurred at this time.30

The likely context for such a visit was a consultation between Augustus and Caesar. The two had not met since Galerius left for Egypt in the latter part of 293. It is reasonable to suppose that Galerius would be asked to report in person to his Augustus. Such consultations featured in Diocletian’s relationship with Maximian during the period of the dyarchy. The panegyrics attest at least two such ‘summit conferences.’33 Galerius had much to report to Diocletian. Not only had he quelled provincial dissent, but he had taken action to renovate military and economic infrastructure. He had established a series of guardposts and way stations to reignite Rome’s trade with the east and reasserted Rome’s local authority. His activities in Egypt and in Palestine should not be seen as two different projects: the first, the suppression of a revolt; the second, the renovation of infrastructure. They are in fact both part of the same project: Diocletian’s grand design to reinvigorate the imperial economy and reassert the ideological eternity of Rome. Diocletian was engaged in the same business in his tour of the Danube, and in the east, Galerius was continuing a work already well begun there. The revolts of Coptus and Boresis did not, then, prompt Galerius’ visit to Egypt; they merely punctuated it.

The result of Diocletian’s work was a web of roads and fortresses from the Red Sea to the Euphrates. Damascus was the key. It linked the limes of the south, which protected traders and farmers from predatory nomads of the Arabian desert, and those which ran to the Euphrates which apprehended the grander threat from Rome’s imperial rival. According to Malalas, Diocletian established an arms factory there (Malalas Chron. 306). On 1 May 295 Diocletian issued an edict at Damascus on marriage31 which fixes his presence there at that time.32

Bibliography Altenorf H.D. 1950 Reallexicon für Antike und Christentum 8. (Stuttgart): 786-787. Amelineau E. 1893 Géographie de l’Egypte à l’Epoque Copte. (Paris). Arnaldi A. 1972 La successione dei cognomina gentium e le loro iterazione nella titolatura dei primi tetrarchi. Reconditi dell’Istituto Lombardo, Classe di Lettre, Scienze e Storiche. 106: 28-50. Avi-Yonah M. 1976 Gazeteer of Roman Palestine. (Jerusalem). Barnes T.D. 1976 Imperial campaigns A.D. 285–311. Phoenix 30: 174-193. Barnes T.D. 1982 The new empire of Diocletian and Constantine. (Cambridge, Mass.). Bernand A. 1984 Les Portes du Désert: receuil des inscriptions grecques d’Antinooupolis, Koptos, Tentyris, Apollinopolis Parva et Apollinopolis Magna. (Paris). Bowersock G.W. 1983 Roman Arabia. (Cambridge Mass.). Bowman A.K. 1978 The military occupation of Upper Egypt in the reign of Diocletian. Bulletin of the American Society of Papyrologists: 27: 25-38. Brennan P. 1984 Diocletian and the Goths. Phoenix 37: 142146. Chastagnol A. 1982 L’évolution politique, sociale et économique du monde romain 284-36. (Paris). Corcoran S. 1996 The empire of the tetrarchs: imperial pronouncements and government AD 284–324. (Oxford). Cuvigny H. 1998 Kainè, ville nouvelle: Une expérience de regroupement familial au IIe s.è. chr. In O. Kaper (ed.) Life on the fringe: living in the southern Egyptian deserts during the Roman and early-Byzantine periods.

26 CIL III. 6267 (between Palmyra and Edessa); 14152. 48a (between Amman and Petra); 14382 (near Gerasa); AE 34.262 (between Palmyra and Emesa); 77.833 (near Bostra). For other milestones of the dyarchy in the region of Bostra, see Littmann, Magie & Stuart 1904-5; 1909; 1921: xx-xii, xxiv, xxvi-vii. 27 For example: CIL III, 6661; AE 30.105; 31.86; 57.272. 28 See Speidel 1977: 699. Hoffmann’s suggestion (idem. n.37) that the legion’s name reflects some relationship between Galerius and Mars and thus belongs to this period is enticing, but must be rejected. The question of Galerius’ divine comes is a difficult one, complicated, rather than clarified by discussions of the question (see, for example, Nicholson 1984. On Lejjun, see Parker 1986: 136f. 29 See Bowersock 1983: Ch. X. It is possible that at this time ‘Amr died. He was apparently long-lived and appears on the Paikuli inscription of Narseh in 293. He must have died not long after and was succeeded by Imru’l-qais. 30 This is an inference from the fact that in 303, Hierocles was ex vicario and praeses of Bithynia (Lact. de mort. pers. 16.4). The dedication in all probability predates 300. Further, it is reasonable to assume that work ceased on the limes during the Persian War when troops were more urgently required to meet the threat of Persia, and later, the revolt in Egypt. Therefore, the dedication at the fortress of Palmyra should be dated to the period 293-6. 31 Mos. et Rom. legum collatio. 64; CJ. 5.4.17. 32 Barnes’ 1982: 62n. 76 argument that the edict was issued by Galerius on the instructions of Diocletian, simply because a visit by the senior emperor to Damascus seems difficult to fit in at this time is not compelling. Diocletian was in Nicomedia on 18 March of that year. Damascus is less than six weeks from Nicomedia by road, and Diocletian’s whereabouts in the intervening period are unknown. It is therefore quite possible for Diocletian to have been in Damascus by early May. For further discussion which does not rule out Barnes’ argument, see Corcoran 1996: 270. It is worth noting that, if the edict were issued by Galerius, it is the only case of a Caesar issuing an edict on his own, Barnes’ response that he was doing so at Diocletian’s direction is simply unconvincing.

33 Pan. Lat. 2: 9.1 speaks of a conference with Maximian at an unknown location, pehaps in Rhaetia. Pan. Lat 3: 8.1 speaks in grander terms of the better known conference between the two in Mediolanum.

88

Bill Leadbetter: Galerius and the eastern frontier

(Leiden): 87-94. Ensslin W. 1930 Maximianus. Paulys Real-Encyclopädie der classischen Alterwißenschaft 14.2 (Stuttgart). Cols. 2518-2521. Ensslin W. 1952 Zu Pap. Oxyrhynchus I 43, Recto. Aegyptus 32: 163-178. Gudea N. 1974 Befestigungen am Banater Donau-Limes aus der Zeit der Tetrarchie, Actes de IXe Congres International d’Études sur les Frontières Romaines. (Bucharest): 173-181. Isaac B. 1992 The limits of Empire: the Roman army in the east. (Rev. ed., Oxford). Isaac B. 1998 The Near East under Roman rule: Selected papers. (New York). Johnson A.C. 1936 Roman Egypt to the reign of Diocletian. In T. Frank (ed.) An economic survey of ancient Rome II. (Baltimore). Jones A.H.M. 1964 The later Roman empire. (Oxford). Kees H. 1930 Maximianopolis Paulys Real-Encyclopädie der classischen Alterwißenschaft 14.2. (Stuttgart). Cols. 2484-2485. Kennedy D. & MacAdam H. 1985 Latin inscriptions from the Azraq Oasis, Jordan. Zeitschrift für Papyrologie und Epigraphik 60: 97-107. Lallemand J. 1964 L’administration civile de l’Egypte 284-382. (Brussels). Lewin A. 1990 Dall’Eufrate al Mare Rosso: Diocleziano l’esercito e I confini tardoantichi. Athenaeum n.s. 68: 141-165. Littmann E., Magie D. & Stuart D.R. 1904-1921 Syria III Greek and Latin Inscriptions, (Publications of the Princeton University Archaeological Expedition to Syria). MacAdam H.I. 1989 Epigraphy and the Notitia Dignitatum (Oriens 37). In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire: proceedings of a colloquium held at Ankara in September 1988. British Archaeological Reports Int. Ser. 553 (Oxford): 295-309. Mattingly H. 1939 The imperial recovery. In H. Last, F.E. Adcock, M.P. Charlesworth & N.H. Baynes (edd.) Cambridge Ancient History XII. (Cambridge): 297-351. Matthews J.F. 1984 Tax law of Palmyra: evidence for economic history in a city of the Roman east Journal of Roman Studies 74: 157-180. Meredith D. 1952 The Roman remains in the eastern desert of Egypt. Journal of Egyptian Archaeology 38: 94-111. Meredith D. 1953 Eastern Desert of Egypt: notes on inscriptions. Chroniques d’Egypte 28: 126–141. Millar F. 1993 The Roman Near East 31 B.C. - A.D. 337. (Cambridge, Mass.). Mócsy A. 1974 Pannonia and Upper Moesia. (tranl. S.S. Frere, ondon). Nicholson O.P 1984 The wild man of the tetrarchy: a divine companion for the Emperor Galerius. Byzantion 54: 253-275. Parker S.T. 1986 Romans and Saracens: a history of the Arabian frontier. (Winona Lake, Ind.). Peacock D.P.S 1992 Rome in the desert: a symbol of power. (Southampton). Petrie Flinders W.M. 1896 Coptos. (London). Raschke M.G. 1978 New studies in Roman commerce with the east. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II 9.2. (Berlin): 6041361. Rea J., Salomon R.P. & Worp K.A. 1985 A ration warrant for

an adiutor memoriae. Yale Classical Studies 28: 101115. Reinach A. 1911 Catalogue des Antiquités Egyptiennes de Koptos. (Paris). Roll I. 1989 A Latin inscription from Yotvata. Israel Exploration Journal 39: 239-299. Seston W. 1946 Dioclétien et la Tétrarchie. (Paris). Shaw I., Bunbury J. & Jameson R. 1999 Emerald mining in Roman and Byzantine Egypt. Journal of Roman Archaeology 12: 203-215. Skeat T.C. 1964 Papyri from Panopolis in the Chester Beatty Library, Dublin. (Dublin). Sidebotham S.E. 1986 Roman economic policy in the Erythean Thalassa 30 BC-AD 217. (Leiden). Sidebotham S.E. 1989 Ports of the Red Sea and the Arabia-India trade. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman Empire: proceedings of a colloquium held at Ankara in September 1988. British Archaeological Reports Int. Ser. 553. (Oxford): 385-513. Sidebotham S.E. 1991 A limes in the eastern desert of Egypt: myth or reality? In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989: Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 494-498. Sidebotham S.E. 1994 Preliminary report of the 1990-1991 season of fieldwork at ‘Abu Sha’ar (Red Sea coast). Journal of the American Research Center in Egypt 31: 133-168. Sidebotham S.E., Riley Hany J.A., Hamroush A. & Barakat H. 1989 Fieldwork on the Red Sea coast: the 1987 season. Journal of the American Research Center in Egypt 26: 127-166. Sidebotham S.E., Zitterkopf R.E. & Riley J.A. 1991 Survey of the ‘Abu Sha’ar-Nile road. American Journal of Archaeology 95: 571-622. Speidel M. 1977 The Roman army in Arabia. In H. Temporini & W. Haase (edd.) Aufsteig und Niedergang der Römischen Welt II.8. (Berlin): 687-730. Speidel M. 1987 The Roman road to Dumata (Jawf in Saudi Arabia) and the Roman frontier strategy of Praetensione Colligare. Historia 36: 213-221. van Berchem D. 1952 L’armée de Dioclétien et la reforme constantinienne. (Paris). Verhoogt A.M.F.W. 1998 Greek and Latin texts. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996, report of the excavations at Berenike (Egyptian Red Sea coast and the survey of the eastern desert). (Leiden): 193–198. Warmington E.H. 1974 The commerce between the Roman empire and India. (2nd ed., London). Weill 1911 Koptos. Annales du Service XI: 97-141. Wendrich W.Z. 1998 Fringes are anchored in warp and weft: the relations between Berenike, Shenshef and the Nile Valley. In O. Kaper (ed.) Life on the fringe: living in the southern Egyptian deserts during the Roman and early-Byzantine periods. (Leiden): 243-252.

89

Limes XVIII

90

Diocletian: Politics and limites in the Near East Ariel Lewin The aim of this paper is to discuss how and when Diocletian reorganized the Near East. In this context it appears profitable to examine again literary, epigraphical and archaeological evidence. The results of this research will be evaluated in order to provide an assessment of the various interpretations given by modern scholars about the function of the Roman army in the area from the Euphrates to the Red Sea. Discussion of other literary sources pertaining to the post-Diocletianic period will also help to clarify some aspects of the question.

the desert were withdrawn, if this had not already happened in the C3rd.

The aim of this paper is to discuss to what extent Diocletian reorganised the army in the Near East. In this context it is profitable to re-examine the literary, epigraphic and archaeological evidence. It will be of great interest to focus our attention on different sectors of this area in which, during the tetrarchic period, many military settlements and roads were renovated or built. It is difficult to give up the idea that Diocletian meant to occupy the most marginal areas on the fringe of the desert.

In an influential paper Ritterling advanced the idea that it was the emperor Aurelian who reorganised militarily the Near East. He pointed out that the Notitia Dignitatum records the presence of cavalry units of Mauri, Dalmatae scutarii and promoti, called Illyriciani in all the provinces of the Near East. He recalled that Zosimus, describing Aurelian’s decisive battle against the Palmyrenians, confirmed that the Roman emperor employed the Dalmatian cavalry, the Moesii, the Pannonias, the troops from Noricum and Raetia, the praetorians and the Moorish cavalry. The correspondence between the two lists implies that it was Aurelian himself who deployed the élite units in the Near Eastern provinces. Ritterling also suggested that legio IV Martia might have been created by Aurelian, unless it had been previously created by Alexander Severus or by some other emperor in the C3rd.

The idea that Diocletian’s reign represented a fundamental milestone in the history of the frontier of the Near East has been argued by many scholars. It has been asserted that Diocletian renewed the military settlements and the routes in the Near East after the decadence of the C3rd.1 Although there is no solid evidence of any particular problem the Saracens caused to the empire during the C3rd crisis, we might easily suppose that the weakness of the imperial authority caused conspicuous and frequent episodes of banditry and plundering. Furthermore the Tanukh, an important Arab tribe from the Arabian peninsula, came into contact with the Romans. We know that Vaballathus bore the title Arabicus: this is to be related to the conflict between the Palmyrenians and the Tanukh, led by king Jadhima, and mentioned in Arabic sources.2

Ritterling stressed that though legio I Illyriciana had established itself in Palmyra only in the tetrarchic period, Aurelian had already placed a garrison in the city. Therefore Ritterling was inclined to believe that the general reorganisation of the Roman armies in the Near East was made by Aurelian. Consequently Diocletian found the Illyrian cavalry unit on the spot as well as most of the legionary set-up.3

The most dramatic aspect of the conflicts was the usurpation committed by the Palmyrenians that caused losses in the Roman army and destruction in the cities. The governor of Arabia was killed by the Palmyrenians while opposing the rebels with his troops, and a famous inscription recalls that the Temple of Zeus at Bostra, dear to the legio III Cyrenaica, was destroyed by the Palmyrenians (Malalas 249; IGLS XIII 1907; cf. Graf 1989).

The issue of who was responsible for the reoganization has recently been re-examined by scholars, some who have asserted more cautiously that Aurelian gave great impulse to the restoration of the armies. M.P. Speidel merely took into consideration the Illyrian troops and their possible settlement in the eastern provinces: “The case for Diocletian rests largely on less reliable general arguments and leaves one wondering why at that time a province like Arabia should have received so many elite units; but the question cannot be answered decisively before more documentary evidence will be found” (Speidel 1977; cf. Graf 1978).

We can assume that this war weakened the Roman army. The effectives had been decimated, and it cannot have been easy to replace them nor to restore the situation. A possible result of such a chaotic situation might have been that the soldiers quartered in the fortlets and towers near

A reconsideration of the documentary evidence will allow us to establish to what extent Aurelian or Diocletian contributed to the restoration of the military set-up in the Near East. In order to evaluate the impact of tetrarchic renovation, this paper will take into account the documentation from other sectors of the Near East.

1

van Berchem 1952: 3-30; Parker 1986: 133-143; Roll 1989; Lewin 1990; Millar 1993: 174-193; Isaac 1992:161-171 is slightly more cautious. I am grateful to Prof. David Kennedy for his comments on earlier drafts of this paper. 2 For the documentation pertaining to the titlature of Vaballathus see Bauzou 1998: 247-250, who put it in relation with the war against the Tanukh. For the title of Arabicus bore by Aurelian, see Kettenhofen 1986: 138-139. The appearance of the Tanukh is examined by Bowersock 1983: 132-137. There is a scanty and difficult to assess documentation about the Saracen threat in the C3rd. See Sartre 1982: 122-136.

3

Ritterling 1903; for the legions I Illyricorum and IV Martia see also Ritterling 1925 col. 1506; 1556.

91

Limes XVIII

fighting the Goths, and literary sources confirm that detachments of the legions from this province were able to reach the east only for Galerius’ second Persian campaign in 297, after drawing up a peace treaty with the Goths (Speidel 1987: 220-221; Brennan 1980).

North Jordan I will start with the north Jordan sector and a famous text: the so-called praetensio inscription from Azraq, the object of several important studies (Kennedy 1982: 179-193; Kennedy & MacAdam 1985: 100-104; Speidel 1987; MacAdam 1989; 1990; Zuckerman 1994b; Bauzou 1996). The inscription presents a number of problems: the possible existence of a road leading to Dumata dotted with military units, the identification of Basienis and Amatha, the meaning of the term praetensio, and the object of the same inscription. According to one scholar the inscription could recall the building of the castra at Azraq (Zuckerman 1994b). Another scholar insisted that the inscription was meant to affirm that the castra of Azraq was part of the wider context of the organisation of a military road (praetensio) leading from Bostra to al-Jawf (Bauzou 1996).

There is a chance that the aforementioned legions participated in the campaign of Diocletian against the Saracens in 290 (mentioned in Pan. Lat. XI. 4;7), and that the works mentioned in our text refer to that occasion. On the other hand, a tetrarchic milestone found near Umm elQuttein in the direction of Bostra is proof that this route was the object of imperial attention sometime between 293-305. Umm el-Quttein was a nodal point on the BostraAzraq axis, and some of the roads radiating from it were taken under imperial control in the tetrarchic period, even if there were pre-existing routes or tracks.6 The presence at Azraq of the so-called ‘altar of the tetrarchs’ confirms the idea of new, important imperial activity in the area between 293 and 305 (Kennedy 1982: 91-92). In conclusion, if the Bostra-Azraq route was reorganised militarily in the tetrarchic period, the praetensio inscription cannot be related to the campaign of 290, but to a slightly later time. As already stated, Galerius was in Damascus in May 295, where he issued a law: the activity at Azraq of Galerius’ Egyptian expeditionary legionary detachments could be consistent with his presence in the area at that time.7

The inscription records that the organisation was accomplished by five legions: XI Claudia, VII Claudia, IV Flavia, I Italica and I Illyricorum. Pap. Oxy. 43 records that the first three legions were part of Galerius’ mobile army in his Egyptian campaign, that went on at least from the end of 293 to January 295.4 The legio I Italica too was supposed to have participated in the same campaign, although there is no mention of it in the papyrus. In fact the mention of a praepositus of XI Claudia in Pap. Oxy. 43 was thought to be a proof of the participation of the I Italica in the campaign. Normally a praepositus was in charge of two units, and it is very likely that the two legions of Moesia Inferior were under his command (Jones 1964: 54-55).

An important issue of the inscription must be taken into consideration. The praetensio inscription mentions four legions that went to Egypt with Galerius, but the presence of the I Illyricorum causes problems. In a footnote in his seminal paper, Speidel tried to overcome them and advanced some suggestions: “The detachment of the I Illyricorum from Palmyra may have been present either because of the legion’s desert experience, or because it retained something of its Illyrian elite quality” (Speidel 1987: 220 n. 33).

This suggestion is reinforced by a new interpretation of the inscription of Aurelius Gaius. The inscription recalls the service of a soldier in the tetrarchic army, who happened to operate in Egypt. A scholarly study demonstrated that Aurelius Gaius was in Egypt with Galerius in 293-5 with I Italica.5 We can now state that the four legions from the two Moesiae were in Egypt with Galerius.

We could give credit to this idea assuming that it was this same legion that built the strata Diocletiana and the military installations at Palmyra. But we know from an inscription that the camp of the I Illyricorum at Palmyra was founded in the tetrarchic period under Sossianus Hierocles (CIL III 133 = 6661). All the milestones placed

Galerius is still attested in Egypt in January 295, but on 1 May he was in Damascus (Barnes 1982: 62). Spring 295 is clearly a terminus post quem for dating the Azraq inscription. On the other hand, at the beginning of 297 Galerius’ army was already fighting the Persians in Mesopotamia. It appears logical to infer that the activities mentioned in our text date back to the previous period. This point is strengthened by one more observation: it was acutely pointed out that the four legions of the two Moesiae were mentioned in the inscription, unlike the two legions of the Dacia Ripensis located between the two Moesiae. At that time the army of the Dacia Ripensis was

6

Full documentation in Kennedy 1996: 257-258; 1997: 77-80. Bauzou 1997:28, relying also on AE 1948: 136, identifies Umm el-Quttein with Basienis, mentioned in the praetensio inscription. For a different view, see Zuckerman 1994a. Isaac 1996: 159-161 states that the placing of milestones along a road was connected with its officialisation by the imperial authorities. It is important to stress that already Septimius Severus had official routes and military structures built in this area. See Kennedy 1980; 1982; 2000a: 55-59, 64-67. It is logical to think that the tetrarchic renewal occurred after a period in which they were neglected. 7 Eadie 1996:78 rightly doubts any connection of the praetensio inscription with the campaign undertaken by Diocletian against the Saracens. It is interesting to observe that in his titulature Diocletian is never called Arabicus. See Ensslin 1942: 19 for an interesting suggestion about it. But see also Graf 1999: 672; “...inscriptions do not necessarily denote road construction or repairs, as milestones were convenient kiosks for purely imperial propaganda.”

4

The connection of the Azraq text with the campaign of Galerius was noted by Kennedy 1982: 182-183; Speidel 1987: 220-221. For the movements of Galerius, see Barnes 1982: 62. 5 Zuckerman 1994a. The inscription of Aurelius Gaius was published with a rich commentary by Drew-Bear 1981.

92

Ariel Lewin: Diocletian: Politics and limites in the Near East

along the strata Diocletiana in the Palmyrenian sector show that it was built in the tetrarchic period.8 Following Speidel we would have to squeeze the events to be able to infer that in less than two years the I Illyricorum received its base at Palmyra, acquired experience in road building and fortlets in the desert, and eventually arrived in Azraq in the spring of 295.

the city or around it, is redundant. The III Diocletiana, a creation of Diocletian that participated to the imperial campaigns in 301, was settled definitively at Luxor, in the area of the old temples (Reddé 1986). Possibly I Illyricorum might have come to its base from the expeditionary army of the tetrarchs. The camp at Palmyra was surely built between 293 and 302/3. Sossianus Hierocles after being praeses of Phoenicia, the province to which Palmyra belonged, is attested in 303 as vicar of an unnamed province, probably Pontica, and inciting the emperor to persecute the Christians (Barnes 1996: 550). However there are good reasons to shorten the time lapse for the construction of the camp for I Illyricorum at Palmyra. Epigraphic texts assert that detachments of the two Phoenician legions, III Gallica and I Illyricorum, were in Upper Egypt in 316 and in 323. But they were probably there in 300 if they are to be identified with the miliarienses mentioned in another inscription from Elephantine, and with the vexillations of oriental legions attested in the second papyrus from Panopolis (Brennan 1989: 200-201).

A different scenario might be imagined: the legio I Illyricorum too was part of the expeditionary force in Egypt with Galerius. In 295 this legion did not join the four Moesian legions coming from the east, but was already with them as part of the expeditionary force. In Pap. Oxy. 43, besides the two praepositi in charge of the vexillations of the four Moesian legions, at least 7 more praepositi are mentioned. Unfortunately the names of the units are not preserved, but we could posit that the I Illyricorum was one of them. But one important question remains unsolved; where was the I Illyricorum based before Diocletian ordered it to build the new camp at Palmyra? There are two possibilities: (i) under Aurelian the legion was quartered somewhere in the Near East, but not at Palmyra. A detachment joined Galerius in Egypt and then returned east. In the end it moved to Palmyra with the rest of the legion; or (ii) under Diocletian the legion was transferred from the west to Egypt to operate under Galerius in the Egyptian campaign; later it was settled in Palmyra. On the other hand we must take into account what M. Christol and M. Lenoir affirm in a forthcoming article: according to them the praetensio inscription dates back to the time of Aurelian. It does not imply the existence of any military work on forts or roads, but it proclaims the renovated control of the area by the army of Aurelian.

Scholars have credited Diocletian with building a military route, the strata Diocletiana, connecting the Euphrates with Arabia, probably with Azraq (Van Berchem 1952: 15). But a recent study argues that a route called the strata Diocletiana is attested only in the milestones between Arak and Khan el-Qattar. A different name, the strata Diocletiani et Maximiani (or even of strata of the four tetrarchs) is attested until Khan el-Trab (Valle Diocletiana). Besides, it was pointed out that the strata Diocletiana was not merely a linear route bordering the desert: in Phoenicia some lateral segments had the same name. In Arabia a lateral road leading to Nemara, in the most marginal part of the settled land near the steppe, received formal organisation and was called strata Diocletiana.11

From the Euphrates to Damascus In an inscription found in the principia of the camp at Palmyra, the praeses Sossianus Hierocles declared he had built the castra in the time of Diocletian, Maximian and the two Caesars. The military area was included in the western sector of the city, and the Temple of Allat, rebuilt after being partially destroyed in the seizure of the city by Aurelian, became part of it. The construction of the castra and the related spatial reorganisation must be connected with the settlement of legio I Illyricorum in the city.9 Some structures, one of which lies quite far from the city, are attributed to Aurelian’s soldiers10, but there is no evidence that at that time legio I Illyricorum was already at Palmyra, or around it. Diocletian surely had psychological and propagandistic aims when he included the Temple of Allat in the camp of I Illyricorum in western Palmyra; but the idea to build a new camp for a unit that already had one in

A text from Bkhara in Phoenicia is of particular interest. It is a dedication to the tetrarchs made by a praepositus of the equites promoti indigenae and of another unit whose name is not entirely preserved in the text (equites sagittarii or equites indigenae). Bkhara must be identified with the Auatha mentioned on some milestones. This matches with the distances marked on the milestones and with recalling Auatha in the Notitia Dignitatum (Or. XXXII 22) as the base of the equites promoti indigenae (Bauzou 1993: 4648). Probably the dedication was written to celebrate the construction of the camp of the promoti. It was suggested that the equites promoti indigenae were a late development of legionary cavalry; with time they were detached from their mother unit and settled elsewhere (Jones 1964: 53).

8

However now it appears that it was not so. There were

See Bauzou 1993 for a new discussion of the characteristics of this route. 9 CIL III 133 = 6661; Gawlikowski 1984: 62-63; Barański 1994; Fellmann 1995: 240 casts doubts whether the camp of Palmyra was built for the legio I Illyricorum. There is no ground for such doubts. 10 Troops left before the revolt of 272: Zosimus I.60 (generic); Vita Aurel. 31 (archers): military buildings Barański 1994: 9.

11 See Bauzou 1993 who argues that the milestone found near Sa’neh in Arabia is not an evidence that the strata Diocletiana led to Arabia. According to him, the milestone does not belong to the main route touching Sa’neh, but to a lateral one leading to Nemara.

93

Limes XVIII

indeed equites promoti in the legions, but as far as the equites promoti indigenae are concerned, it must be stated that they were locally recruited cavalry. It is interesting to note that in central Jordan we can posit the existence of a similar case: a unit of equites promoti indigenae settled in a fort on the fringe of the desert approximately in the same years (in 306/307).12 The creation of these units in the first tetrarchy, or in the following years, is strong evidence of the effort Diocletian and his colleagues made to recruit new soldiers to strengthen the Roman presence in the most marginal areas of the empire.

the site identified as the base of the IV Martia, put the date of the building of the fort into the late tetrarchy. A coin of 304/5 found in the foundation of the primary legionary barracks provides an important terminus post quem.16 Consequently, we can take it for sure that Aurelian had nothing to do with the transfer of this legion. As well as this, we can presume that the minor military settlements around el-Lejjun have a meaning only if they are considered as part of a global reorganisation of the sector (Parker 1986: 37-86). An epigraphic text reveals that Qasr Bshir was built in the tetrarchic period. It has been suggested that Bshir was not a fort but a praetorium used as a fortified road-station for passing troops (CIL III 14149; Isaac 1992: 172-174). Nonetheless its construction must be seen as part of the building-up of routes and military settlements in this part of the pre-desert. It is important to stress that among all the military settlements in this area, only Khirbet el-Fityan shows signs of previous occupation, in the Nabataean period.

Under Diocletian other units must have been settled in their forts. The Notitia Dignitatum places at Veriaraca an ala Nova Diocletiana (Or. XXXII 34) and at Valle Diocletiana a cohors secunda Aegyptiorum (Or. XXXII 43). At Valle Alba, according to the Notitia, was quartered a cohors prima iulia lectorum (Or. XXXII 42). Between Palmyra and Sura only one milestone on the strata Diocletiana was found at Arak. The impact of tetrarchic activity in this area cannot be evaluated with certainty, but it could be argued that the greatest part of military settlements between Sura and Oriza were built/rebuilt approximately at that time.13 If we rely on the hagiographic text of the life of S. Sergius the forts mentioned as the stages of his martyrdom - Barbalissus, Sura, Tetrapyrgium, Resafa - already existed in 312.14

El-Lejjun and other minor sites were built in the last years of the first tetrarchy or a few years later (Parker 1986: 3786). The peculiarity of the gigantic towers at el-Lejjun has been stressed: such a size is not attested in the east before 310 or so. The fan-shaped corner towers are a distinctive feature of the time of the tetrarchs.17 The IV Martia must have been settled at el-Lejjun in the last period of the first tetrarchy, or at the beginning of the second tetrarchy. As for the fort at Udruh, we are still awaiting an excavation clarifying more precisely the date of its construction. The campaigns led by Killick have not been followed-up with definitive publication (Killick 1983; 1986). Nonetheless, some observations based on the typology of the site point to a date in the tetrarchy. Its fan-shaped corner towers and the round interval towers are similar to those of el-Lejjun. It was noted that “…there is also an interesting correlation of dimensions between the plans of Lejjun, Udruh, Diocletian’s camp at Palmyra, Luxor and Split, which might imply a standardisation of plan for the ceremonial areas of legionary bases at that period” (Gregory 1995: 195). It has also been stressed that the towers might be a secondary addition to a pre-existing structure (Parker 1986: 94-98). In general it has been noticed that it is difficult to detect a Diocletianic build-up in southern Jordan (Fiema 1995).

According to a recent excavation, the fortlet of Tetrapyrgium (Qusair as-Sayla), mentioned in Sergius’ life was not a creation of Diocletian. Coins and pottery are indicative of a building erected around 320 (Konrad 1999: 401-404). If we rely on this interpretation of the data, it would be confirmed that the age of Constantine was not a break in the military occupation of the borders of the empire. Tetrapyrgium could be a later addition (Konrad 1999: 402); otherwise we should look at the date of 320 with a bit more flexibility.15 Central and southern Jordan As we have seen, Ritterling took into serious consideration the possibility that the legion IV Martia was a creation of Aurelian or of an earlier emperor of the C3rd. On the other hand some scholars stress that the name Martia is connected with the god Mars, frequently associated with Galerius (Hoffmann 1969: 173-174; Parker & Lander 1982). The excavations conducted by Parker at el-Lejjun,

Aravah An inscription attests that the praeses Priscus settled an ala in a fortlet at Yotvatah in the time of the tetrarchs. Moreover, on two sides of the inscription in the tabula ansata, a celebration of the already held vicennalia of the tetrarchs is carved together with vows for 40 years of

12

See Lewin forthc. for another probable example of a detachment of equites promotii from their legionary base. 13 The milestone from Arak: CIL III 6719 = Thomsen 1917: 28 n. 52; see Bauzou 1993: 28-29. A study of some of the military installations is now available: see Konrad 2001. 14 For the year 312 as the most probable for the martyrdom of Sergius, and for the reliability of the text, see Key Fowden 1999: 8-25. Traces of the camp of the equites promoti indigenae at Resafa where Sergius was tried: Konrad 1992. 15 For this last suggestion see Key Fowden 1999: 1, who means to overcome the difficulty of the presence of Tetrapyrgium in his life of Sergius.

16 Betlyon 1987: 657 coin n. 8. I am grateful to S.T. Parker for explaining to me the importance of that find. The main sequence of the numismatic finds (700 pieces) at el-Lejjun starts from Diocletian, but two unstratified coins of Probus and one Nabataean coin were found too. 17 For a discussion on the diffusion of the fan-shaped corner towers in the C4th, see Lander 1984: 246-252; Gregory 1996: 195.

94

Ariel Lewin: Diocletian: Politics and limites in the Near East

jubilee. Thus, this fort should have been erected in 303/4, but the publisher of the inscription, I. Roll, has pointed out that the proclamation of the anniversary and the main text were carved by two different hands. Consequently the texts in the ansatae were a later addition. The name of the province governed by Priscus is erased: this means that some time after carving the main text this region had become part of another provincial entity.18 Roll argued that the name of the province originally carved was Syria Palaestina. Two inscriptions found at Caesarea mention the vir perfectissimus Aufidius Priscus as praeses of Palestine, partly confirming his interpretation.19

In north Jordan the military organisation initiated by Diocletian followed the policy of Septimius Severus, who had occupied the sites in the pre-desert reaching Azraq. The sector east of the via nova Traiana in northern Jordan was renewed through the reconstruction of military settlements, and through the official adoption of a road network. Great care was given to the reconstruction both of the military settlements at Umm el-Quttein and Azraq, and of the road from Bostra. A fort was built at Deir el-Kahf, along the secondary route that crossed the basalt desert from Azraq to Imtan (Motha; cf. Kennedy 1980; 1982). The Aravah road was surely recognized by the emperors long before Diocletian’s reign. Its tetrarchic milestones are of secondary use, and the documentation found at the military site of Yotvatah attests an occupation during the first half of the C3rd. The fort at Hatseva was erected in the first decades of the same century (Avner & Roll 1997; Meshel 1989; Kindler 1989; Cohen & Israel 1996). A fundamental difference may be detected between the SuraPalmyra road and the Palmyra-Damascus road: at the military sites along the Sura-Palmyra road were found traces of an occupation that could date back to the Flavian period (Konrad 1992; 1996). Thus, the presence at Arak of the famous milestone of Traianus’ father (AE 1993: 205) would make sense: at that time an official route and military structures were built on the border with the steppe.

We must accept also Roll’s deduction that the vicennalia text inscribed in the tabula ansata was an addition to the original inscription in which the foundation of the fortlet was proclaimed. In fact Eusebius mentions that in 303/4 it was Flavianus and not Priscus, as the governor of Palaestina. The construction of the fort at Yotvatah must date back to a previous year in the tetrarchic period.20 In more recent years three groups of milestones have been found along the Aravah, approximately 20kms north of Yotvatah. Some date back to the first tetrarchy, some to the following years up to 324 (Avner & Roll 1997: 135; Wolff 1996). We can thus argue that the renewal of a route in the Aravah and the construction of new military settlements were undertaken under the tetrarchs. This matches with the assertion of the excavators of Hatseva: the fort was rebuilt in the time of Diocletian, after a hiatus in the occupation of the C3rd.21

As far as the Palmyra-Damascus axis is concerned, there is nothing that gives evidence of a route with military settlements south of the Jebel Rawaq in the preDiocletianic period. Van Berchem (1952: 12-13) recalled an inscription found at Basiri dating back to the first centuries of the empire that mentions the cohors VI Hispanorum. Furthermore he supposed that the fort at Manqoura too related to the same period. He pointed out that both forts are located on a strategic spot at the outlet of two main hills of the Jebel Rawaq, but denied the existence of a coherent system of a road organised militarily. It is significant that the Tabula Peutingeriana, apparently the expression of a period between 120 and 160 in the Near East, does not depict the existence of a road south of the Jebel Rawaq. It attests nevertheless the stations between Palmyra and Sura.22

Discussion Important building activity was undertaken in various areas of the Near East in the time of the tetrarchs. Some routes bordering the empire received formal recognition and along them military settlements were built. One question is still unsolved: was the construction of the routes and the military settlements introduced for the first time during the tetrarchic period, or was it simply a renewal after the decadence of the C3rd? At present we are not able to answer fully this question, but must take into consideration each area separately.

Quite probably not even at the time of the Severi was an official road beyond the Jebel Rawaq organised. Along this route no milestones of previous emperors have been found, nor the tetrarchic ones seem to be secondary. We can appreciate their difference from the tetrarchic milestones along the Aravah and the via nova Traiana, which are secondary and show the activity of the emperors of the C2nd and C3rd.23

18

Roll 1989. For the archaeological and numismatic documentation concerning Yotvatah, see Meshel 1989; Kindler 1989. For relevant observations about the inscription Eck 1992. 19 Burrell 1993: 290-294 points out that in the two inscriptions from Caesarea, in the Onomastikon of Eusebius and in the list of Verona, the province is called Palaestina and not Syria Palaestina. 20 Contra Barnes 1996: 548; 551 who argued that in the inscription from Yotvatah Priscus’ province was Arabia or Arabia Petraea. According to Eusebius Mart. Pal. L, 1.1, 3.1, Flavianus was governor of Palaestina in 303/304; consequently Barnes argued that Priscus could not have been praeses Palaestinae. But Barnes seems unaware of the inscriptions published by Burrell and of the fact that the vicennalia text is a probable second-hand addition to the main text. 21 Cohen & Israel 1996: 110-116 argue that the towers were added to the original building at the end of the C3rd.

22 Bowersock 1983: 177-179 for dating the archetype of the Tabula to the C2nd; Bauzou 1989: 208 stressed the years 120-160, but with a preference for the beginning of Marcus Aurelius’ reign. See also Millar 1993: 135-136. 23 Avner & Roll 1997; Bauzou 1989: 211-212; Graf 1995. But at Arak, where milestones were put in Flavian times, a new tetrarchic milestone was erected. See Bauzou 1993: 28-29.

95

Limes XVIII

The documentary evidence I have taken into account shows that it was only under the tetrarchs that the most marginal and external areas again became the object of imperial interest after the hard times of the C3rd. It is worth pointing out that the existence of a similar pattern can be found in the Egyptian Eastern desert, where the military settlements along the main roads show a gap in the occupation during the C3rd.24 Diocletian restored the imperial presence moving his troops into the pre-desert. He mostly occupied areas already reached by the Severans, but for what it concerns the Palmyra-Euphrates axis, it is likely that at the time of the Flavians there was already a route organised militarily. On the other hand, the sectors of central Jordan around el-Lejjun were occupied for the first time by the tetrarchs. It is quite probable that the route beyond the Jebel Rawaq was given a coherent system of forts connected to each other by an official route only in that same period. Such observations are of great relevance: they could dismantle the idea that in late antiquity the whole of the empire was in serious difficulty, withdrawing from previously occupied lands. A more nuanced view should take into consideration the great effort undertaken by the tetrarchs: they moved to the most marginal areas and occupied them.

of Neocaesarea with the ancient Athis and with the modern Dibsi Faraj seems well-founded (Harper 1977: 457-460). These two towns were on the bank of the Euphrates, upstream of Sura. Occariba is ‘Agerbat on the Apamea-Palmyra road (Mouterde & Poidebard 1945: 49). Seriane was identified with Isriye, to the rear of the axis Sura-Palmyra. It lies on an important junction of routes: one leading to Androna and Chalcis, the other one to Salamias and Emesa. Seriane was connected with Resafa too. The site has a temple of the C2nd-C3rd, and a fort that was probably the one where the equites scutarii were stationed (Dussaud 1927: 273; Mouterde & Poidebard 1945: 89, tav. 58; 59.2; Gogräphe 1993: 45-61). The Illyrian units listed in the Notitia Dignitatum in Phoenicia are: the equites Mauri Illyriciani at Otthara (Or. XXXII 18); the equites scutarii Illyriciani at Euhari (XXXII 19), the equites Dalmatae Illyriciani at Lataui (XXXII 21). This last toponym is still unidentified, but the identification of Euaria-Aueria with Hawarin is certain (IGLS V 2696; Dussaud 1927: 280). The Notitia mentions the equites promoti indigenae at Saltatha (Or. XXXII 20); they are listed after the equites scutarii illyriciani but before the equites Dalmatae Illyriciani. In his edition of the Notitia Seeck noted that it was not a unit of indigenae, but an Illyrian one. Usually in the Notitia the indigenae are not listed before the Illyriciani. Saltatha is still unidentified. Dussaud argued that Otthara/Otthora was Ghunthur (Dussaud 1927: 268), a nodal place at the intersection of routes leading to Emesa, Palmyra, Qaryatein and Hauwarin. The two sites Euaria and Otthara are on the route connecting Palmyra with Damascus, north of the Jebel Rawaq. This route crossed an area located in an easier environment than the southern one where the strata Diocletiana was built.

The first question we face is the detection of the eventual impact of the activity of Aurelian. The assumption that Aurelian or Probus settled military units or gave new recognition to the most marginal areas of the settled land is not supported by the sources. In particular, the documentation reveals that at least two legions, the I Illyricorum and the IV Martia, were based there only in the tetrarchic age. Milestones of Aurelian were found only along the via nova Traiana or in the rear of Phoenicia.25 Then what remains of Ritterling’s theory and, in a wider sense, what can be attributed to Aurelian?

At first sight the pattern in Arabia seems quite different. Two sites in the southern Hawran housed élite units of Illyriciani: the equites scutarii Illyriciani were based at Motha (Or. XXXVII 14), the equites promoti Illyriciani at Tricomia (Or.XXXVII 15). The identification of Motha with Imtan and Tricomia with Salkhad seems sure (Dunand 1934: 108-109). Motha itself appears to lie on the extreme edge of the agricultural zone. Both sites are beyond the main axis of the via nova Traiana.

The Notitia Dignitatum, whose pars orientalis dates back to 400 (Zuckerman 1998), is likely to reflect to a good degree the reality of the sector between the Euphrates and the Red Sea during the tetrarchic age (Jones 1964: 56-57). In this list a fundamental feature of the deployment of the Illyrian troops in Syria and in Phoenicia is discernible: their known bases were in the rear of the provinces. In the Notitia Dignitatum we find the equites scutarii illyriciani at Seriane (Or. XXXIII 16), and the equites promoti Illyriciani at Occariba (XXXIII 17) under the disposition of the dux Syriae; whereas in Augusta Euphratensis the equites Dalmatae Illyriciani are quartered at Barbalissus (XXXIII 25) and the equites Mauri illyriciani at Neocaesarea (XXXIII 26). The identification

The Notitia Dignitatum lists two more Illyrian units at the disposition of the dux Arabiae: the equites Dalmatae Illyriciani at Ziza (XXXVII 16), and the equites Mauri Illyriciani at Areopolis (XXXVII 17). Areopolis lies on the via nova Traiana, but Ziza is beyond it. It should be noted that in Arabia the deployment of Illyrian units is different from Syria and Phoenicia, where they were deployed on the rear of the Sura-Palmyra axis and of the Jebel Rawaq. In Arabia the sites of Motha, Tricomia and Ziza are east of the main route, the via nova Traiana. The reason is that in north Jordan the via nova Traiana is separated from the desert by an agricultural zone 15-40kms wide. Remarkably enough, an extensive ancient settlement survived at Ziza at the beginning of the C20th (Brünnow & v. Domaszewski

24 S. Sidebotham pers. comm., who nonetheless remarked on the difficulty of establishing the length of such a hiatus. T. Erickson-Gini is studying the gap in the occupation of the military settlements in the Negev and along the Aravah. 25 Thomsen 1917: 125 b2 = CIL III 14149 48b (Diban); Thomsen 1917: 151 = CIL III 14149 48 (Dat Ras); AE 1936: 139 (near Tyrus in direction of Caesarea Paneas).

96

Ariel Lewin: Diocletian: Politics and limites in the Near East

1904-1909, vol. II: 91-103). We may once more infer that the Illyrian units were not settled on the extreme edge of the empire.26

Literary sources have stressed Diocletian’s serious concern with strengthening the army and building military structures along the frontiers. Zosimus praised his deployment of the army in cities, forts and towers along the frontiers, to make the empire a safe place. Zosimus might have exaggerated in praising Diocletian; his aim was to compare the ‘good’ pagan emperor with the ‘bad’ Constantine, destroyer of the army, the frontier and the cities. According to Zosimus (II.34), Constantine’s decision to withdraw troops from the frontiers in the end undermined the empire’s safety.

Motha was a village that had a chance to develop thanks to hydraulic devices. It is interesting to point out that, if compared with Deir el-Kahf where a fort was rebuilt in 306 (Littmann, Magie & Stuart 1910: 228), Motha appears to have a slightly more favourable environment. Deir elKahf seems really an isolated military settlement in the basalt desert. However a modern visitor to the sites needs to be cautious in his judgement. Changes in the climate and in the systems of human irrigation can affect the appearance of the two sites even in a short period of time (M. Sartre pers. comm.). At the beginning of the C20th remains of agricultural settlements were noted around the fort of Deir el-Kahf (Butler 1909: 145). More recently a couple of tombs were found recalling the smaller ones at Petra and at other Nabataean sites (Kennedy & Freeman 1996). Nonetheless the impression that Deir el-Kahf as a more marginal location cannot be avoided (cf. Kennedy 2000a: 67).

Lactantius (de mort. pers. 7) criticised Diocletian for the impact the growth of the army had on the empire, probably referring to the creation of four different armies, one for each tetrarch. Although his aim was merely polemical, we should take into account that Lactantius could really have witnessed the effort made by Diocletian and his colleagues to strengthen the army with massive recruitments. John Malalas is an embarrassing source in another sense. His report of the tetrarchic period has many wrong dates and pieces of information. Nevertheless, when he describes Diocletian’s policy in the east, he is worth attention. According to Malalas, Diocletian built forts in all frontier provinces, from Egypt as far as the borders of Persia. He also settled limitanei under the command of the duces in each province and put stelae to the Augustus and the Caesar on the frontier of Syria. This last assertion must refer to the milestones of the strata Diocletiana. Even though Malalas seems not to be completely correct about the creation of the limitanei and about a generalised appointment of duces, nonetheless he relates the important events that occurred in the east which, as we saw, are confirmed by the inscriptions. Malalas also recalls that Diocletian built ammunition factories in Antioch and Edessa for supplying the armies fighting against the Persians, and in Damascus for the army facing the Saracen threat along the frontier from the Euphrates to the Red Sea.29 These factories are mentioned in the Notitia Dignitatum, and scholars do attribute them to Diocletian’s initiative (James 1988).

The attestation of Illyrian units at Salkhad and Imtan is remarkable. If we assume they were deployed by Aurelian, we cannot help remarking on the difference between the policies of Aurelian and Diocletian. Aurelian settled troops in agricultural lands, but it was only in the tetrarchic age that the army moved to sites bordering the desert. This leads us to a comparison between Diocletian’s and Septimius Severus’ policies. This emperor had settled military units in most marginal sites of north Jordan, reaching Azraq and Deir el-Kahf; whereas Salkhad and Imtan had already been connected with Bostra by an official road at least since 162 (Kennedy 1980; 1982; 2000a: 65). Still no document can tell us who was the emperor behind the diffusion of the illyriciani in the eastern provinces. Aurelian or Diocletian are the most likely.27 We cannot help observing that if it was Aurelian who settled these troops in the east, he did not deploy them in the most marginal areas of the empire. If it was Diocletian, he settled élite troops to the rear as a tactical support to the forces along the borders – as already argued by some scholars.28

According to Ammianus Marcellinus Diocletian, when “…cum in ipsis barbarorum confiniis interiores limites ordinaret”, strengthened the fortress of Circessium with the soldiers deployed on the borders. For an exact translation of this passage see Isaac 1988: 141. 29 Malalas 307; 308; see also ibid. 313. Isaac 1988: 141 points out that at the time of the first tetrarchy the duces were not instituted in all the provinces yet. See now Zuckerman 1994b: 83-84 who dates back to the first tetrarchy an inscription from Bostra (IGLS XIII 9062) mentioning Aurelius Valerianus v(ir) p(erfectissimus) dux. Nonetheless Zuckerman prefers to see this man as an expeditionary dux rather as a military provincial commander. The importance of Malalas for our understanding of the problems in the area between the Euphrates and the Rea Sea has been pointed out by Millar 1993: 180, 184. On the problems the Saracens caused on the frontiers in late antiquity see also the keen observations by Whittaker 1984: 137-138. Kennedy & Riley 1990: 238 state that “under the tetrarchy… for the first time there are indications of a fully planned system in a frontier area, in this case the desert south-west of the Euphrates.”

26

See Parker’s 1989 remarkable observations. 27 Jones 1964:55 did not choose. 28 In favour of a Diocletianic deployment of the Illyriciani in the Near East: Van Berchem 1952: 15-17 (Phoenicia); at p. 26 he argues the existence of a tactical deployment of the military units in Palaestina, where the Illyrician cavalry “en arrière de la frontière, sur les grandes voies de pénétration, y remplit la tâche d’un réserve mobile”; Hoffmann 1969:256; Gichon 1971; 1997 developing the idea that Diocletian displayed a tactical and strategic conception in the Negev; Luttwak 1976: 176-178 stressing the concept of a “Diocletian’s defence-in-depthstrategy”; Williams 1985; Parker 1989; Nicasie 1998: 38. An important point of Van Berchem’s (1952: 17-19) reconstruction of the military organisation under Diocletian must be corrected: according to his misleading translation of Malalas. 308, he argues that the duces of Diocletian commanded the élite troops and other units on the rear, but not

97

Limes XVIII

towers and walls to prevent raids by the Persians (XXIII.5.1-2). It is important to underline that our source seems to support our view on a feature of Diocletian’s policy: the military occupation of the most marginal areas.

construction of these structures and of similar ones beyond the Danube, ought not to be seen as extraordinary events, as they were built at a short distance from the river (cf. Brennan 1980). But they were the expression of a renewed strength that was leading the Roman empire to move forward into barbarian territory.

Finally, it is important to mention a source scholars generally disregard, but which shows clearly how the renewal of the armies and of the military structures had a tremendous impact on the contemporaries. It is a pagan source that cannot be suspected of exaggerating the greatness of the military apparatus under Diocletian for panegyric purposes. Libanius hated Diocletian because he had killed his grandfather, a leading councillor of Antioch, and had taken possession of his property. Libanius mentions Diocletian calling him “that emperor that had erected a wall of men in order to defend the Roman empire” (Or. XX.17; see also Or. XIX. 13).

Further evidence of the extraordinary military efforts of the tetrarchs is given by the inscription of Aurelius Gaius (Drew-Bear 1981). In his career Gaius served in most of the provinces of the empire, as well as having fought against the barbarians in their own land. In the inscription he recalled having been in Carpia, Sarmatia (four times), Gothia (twice) and Germania. As the editor of the inscription has pointed out, the word Germania refers to the land of the ‘Free Germans’, and not to the Roman province (Drew-Bear 1981: 120). One more consideration is worth stressing: the wars fought in Gothia imply a Roman presence in ancient Dacia, therefore the statement “Dacia restituta” pronounced in a panegyric in 297 – or more probably in 298 (Pan. Lat. VIII.3.3) – acquires greater relevance.

In addition to what we have said, it is important to note that Diocletian and his colleagues did not confine themselves to a defence line along the borders of the empire. Tetrarchic propaganda laid stress on the renewed offensive activity of the Roman army. This policy was put into practice on many frontiers of the empire. Maximianus’ panegyrist stressed that before the tetrarchs, the Gauls had the habit of thanking the gods for protecting them through the Rhine. The orator himself emphasised the fact that the degree of safety offered by the Romans was strictly linked to the level of the river: if it lowered the provincials were exposed to an enemy attack, whereas if it swelled, it secured them from it. The crossing of the Rhine into Germany by Maximianus was hailed as a extraordinary deed ensuring a safety the Romans were proud of: “Exinde igitur soluto animo ac libero sumus. Licet Rhenus arescat tenuique lapsu vix leves calculos perspicuo vado pellat, nullus inde metus est; quidquid ultra Rhenum prospicio, Romanum est “ (Pan. Lat. X.7).

Diocletian clearly meant to restore Roman authority in these areas, and the treaty drawn up with the Goths after his glorious campaigns affirmed Rome’s predominant position on the barbaric world.30 The reason why the two legions of Dacia Ripensis did not come to the aid of Galerius, who was fighting against the Persians at the beginning of 297 (Speidel 1987: 220-221; Brennan 1980), could be thus explained: they were fighting in Dacia to help Rome restore its hegemony in this former province. The achievement of hegemony, and the drawing-up of treaties with the barbarians from a position of supremacy, must be considered among the most representative features of tetrarchic policy. In the Near East control of Wadi Sirhan, the proliferation of military structures around elLejjun, and probably around Udruh too, the strata Diocletiana, and the renewed control of areas on the border with the desert, are evidence of this relentless will.31

To Diocletian are also ascribed the revival of an imperialistic attitude and the enlargement of the boundaries in the sector of the Raetia: “Ingressus est nuper illam quae Raetiae est obiecta Germaniam similique virtute Romanum limitem victoria protulit” (Pan. Lat. X.9.1); “ transeo limitem Raetiae repentina hostium clade promoto” (Pan. Lat. XI.5.4). Besides, victories were obtained in the heart of the territory of Germany and Sarmatia was devastated (Pan. Lat. XI.5.3; 4).

Bibliography Avner U. & Roll I. 1997 Southern ‘Arava, Roman milestones. Excavations and surveys in Israel 16: 135. Barański M. 1994 The Roman army at Palmyra: a case of adaptation of a pre-existing city. In E. Dabrowa (ed.) The Roman and Byzantine army in the east. (Krakow): 9-17. Barnes T.D. 1975 The composition of Eusebius’ Onomastikon. Journal of Theological Studies 26: 412-415.

At first sight these assertions might seem unconvincing, the expression of propaganda rather than the evidence of real, significant facts. However there are other sources that point not to neglect of the relevant military effort undertaken by the tetrarchs.

30

For a different interpretation of this sentence see Demougeot 1983: 103. According to her the panegyrist was praising the tetrarchs because they had united the two provinces Dacia Mediterranea and Ripensis into a single province called Dacia. In my opinion this view is misleading. Demougeot suggested that the Roman hegemony in ancient Dacia was due to the military activity of Constantine. 31 On the reverberations in the Hejaz of the imperial power, see Lewin 1990: 149-152.

It is known that under Diocletian military bases were built beyond the Danube, but some examples are worth being recalled: in 294 “castra facta in Sarmatia contra Acinco et Bononia” (Chron. min. I, p, 230; cf. Brennan 1980; Bertók 1997). In the Notitia Dignitatum this last fort is mentioned again: “contra Bononiam in barbarico.” The 98

Ariel Lewin: Diocletian: Politics and limites in the Near East

Barnes T.D. 1981 Constantine and Eusebius. (Cambridge, Mass.). Barnes T.D. 1982 The New Empire of Diocletian and Constantine. (Cambridge, Mass.). Barnes T.D. 1996 Emperors, panegyrics, prefects, provinces and palaces (284-317). Journal of Roman Archaeology 9: 532-552. Bauzou T. 1985 Les voies de communication dans le Hauran à l’époque romaine. In J.M. Dentzer (ed.) Hauran I. Recherches archéologiques sur la Syrie du sud a l’époque hellénistique et romaine. (Paris) 137-165. Bauzou T. 1989 Les routes romaines de Syrie. In. J.M. Dentzer & W. Orthmann (edd.) Archeologie et histoire de la Syrie. (Saarbrücken): 205-221. Bauzou T. 1993 Epigraphie et toponymie: le cas de la Palmyrène du sud-ovest. Syria 69: 27-50. Bauzou T. 1996 La praetensio de Bostra à Dumata ( el-Jowf). Syria 73: 23-35. Bauzou T. 1998 La via nova en Arabie. Le secteur nord, de Bostra à Philadelphie. In J.B. Humbert & A. Desremaux A. (edd.) Khirbet es-Samra 1. (Turnhout): 101-255. Bertók G. 1997 Ripa Sarmatica: late Roman counterfortifications on the left bank of the Danube. In W. Groenmann-van Waateringe, B.L. van Beek, W.J.H. Willems & S. L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 165-172. Betlyon J.W. 1987 Coins, commerce, and politics: coins from the Limes Arabicus project, 1975-1986. In S.T. Parker (ed.) The Roman frontier in Central Jordan: interim report on the Limes Arabicus Project 1980-1985. British Archaeological Reports Int. Ser. 340. (Oxford): 655-689. Bowersock G. 1983 Roman Arabia. (Cambridge, Mass.). Brennan P. 1980 Combined legionary detachments as artillery units in late-Roman Danubian bridgehead dispositions. Chiron 10: 553-567. Brennan P. 1984 Diocletian and the Goths. Phoenix 38: 142146. Brennan P. 1989 Diocletian and Elephantine: a closer look at Pococke’s puzzle (IGRR 1.1291 = SB 5.8393). Zeitschrift für Papyrologie und Epigraphik 76: 193205. Brünnow R. & von Domaszewski A. 1904-1909 Die Provincia Arabia. 3 vols. (Strassburg). Burrell B. 1993 Two inscribed columns from Caesarea Maritima. Zeitschrift für Papyrologie und Epigraphik 99: 287-295. Butler H.C. 1909 Ancient architecture in Syria. The Southern Hauran. (Leiden). Cohen R. & Israel Y. 1996 En Hazeva 1990-1994. Excavations and Surveys in Israel 15: 110-116. Demougeot E. 1983 Constantin et la Dacie. In E. Frézouls (ed.) Crise et redressement dans les provinces européennes de l’empire. (Strasbourg): 91-112. Drew-Bear T. 1981 Les voyages d’Aurélius Gaius, soldat de Dioclétien. In La géographie administrative et politique d’Alexandre à Mahomet (Strasbourg): 93-141. Dunand M. 1934 Le museé de Soueïda. (Paris). Dussaud R. 1927 Topographie de la Syrie antique et mediévale. (Paris). Eadie J. 1996 The transformation of the eastern frontier, 260-305. In R.P. Mathisen & H.S. Sivan (edd.) Shifting frontiers in late antiquity. (Aldershot): 72-82. Eck W. 1992 Alam Costia constituerunt: zum Verständnis

einer Militärinschrift aus dem südlichen Negev. Klio 74: 395-400. Ensslin W. 1942 Zur Ostpolitik des Kaisers Diokletian. (München). Ensslin W. 1952 Zu Pap. Oxyrinchus 43 recto. Aegyptus 32: 163-177. Fellmann R. 1995 L’inscription d’un “optio princeps” au temple de Ba’alshamin à Palmyre. In Y. Le Bohec (ed.) La hieérarchie (Rangordnung) de l’armée romaine. (Paris): 239-240. Fiema Z. 1995 Military architecture and the ‘defense’ system of Roman-Byzantine southern Jordan - A critical appraisal of current interpretations. In K. ‘Amr, F. Zayadine & M. Zaghloul (edd.) Studies in the History & Archaeology of Jordan 5: 261-269. French D. & Lightfoot C.S. (edd.) 1989 The eastern frontier of the Roman empire. British Archaeological Reports Int. Ser. 553. (Oxford). Gawlikowski M. 1984 Palmyre VIII. Les principia de Dioclétien. “Temple des Enseignes”. (Warszawa). Gichon M. 1971 The military significance of the Limes Palaestinae. In S. Applebaum (ed.) Roman Frontier Studies 1967. (Tel Aviv): 191-200 Gichon M. 1997 The strategic conception and the tactical functioning of the Limes Palaestinae. In W. Groenmann-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 25-31. Gogräphe R. 1993 Die Datierung des Tempels von Isziye. Damaszener Mitteilungen 7: 45-61. Graf D.F. 1978 The Saracens and the defense of the Arabian frontier. Bulletin of the American Schools of Oriental Research 229: 1-26 (= Graf 1997a). Graf D.F. 1989 Zenobia and the Arabs. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. British Archaeological Report Int. Ser. 553. (Oxford): 143-167. Graf D.F. 1995 The via nova Traiana in Arabia Petraea. In J. Humphrey (ed.) The Roman and Byzantine Near East. Some recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 241267. Graf D.F. 1997 Rome and the Arabian frontier: from the Nabataens to the Saracens. (Aldershot). Graf D.F. 1997 The via militaris in Arabia. Dumbarton Oaks Papers 51: 271-281. Graf D.F. 1999 Roads. In G.W. Bowersock, P. Brown & O. Grabar (edd.) Late antiquity. A guide to the postclassical world. (Cambridge, Mass.): 671-673. Gregory S. 1995-1997 Roman military architecture on the eastern frontier. (Amsterdam). Gregory S. 1986 Was there an eastern origin for the design of late Roman fortifications? Some problems for research on forts of Rome’s eastern frontier. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 169-209. Hoffmann D. 1969 Das spätrömische Bewegungsheer und die Notitia Dignitatum. (Düsseldorf ). Isaac B. 1988 The meaning of the terms limes and limitanei. Journal of Roman Studies 78: 125-147 (reprinted with additions in Isaac 1998a: 345-387). Isaac B. 1992 The limits of empire. The Roman army in the east. (rev. edit. Oxford). Isaac B. 1996 Eusebius and the geography of Roman provinces. In D.L. Kennedy (edd.) The Roman army in

99

Limes XVIII

the east. Journal of Roman Archaeology Supplementary Series 18: 53-167 (reprinted with additions in Isaac 1998a: 284-309). Isaac B. 1998a The Near East under Roman rule. (Leiden). Isaac B. 1998b The eastern frontier. In A. Cameron & P. Garnsey (edd.) The Cambridge Ancient History Vol. XIII. (Cambridge): 437-460. James S. 1988 The fabricae: state arms factories of the later Roman empire. In J.N. Coulston (ed.) Military equipment and the identity of Roman soldiers. Proceedings of the fourth Roman military equipment conference. British Archaeological Reports Int. Ser. 394. (Oxford): 257-331. Jones A.H.M. 1964 The later Roman empire. (Oxford). Kennedy D.L 1980 The frontier policy of Septimius Severus: new evidence from Arabia. In W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies XII. British Archaeological Reports Int. Ser. 71. (Oxford): 879-888. Kennedy D.L. 1982 Archaeological explorations on the Roman frontier in north-east Jordan. British Archaeological Reports Int. Ser. 132. (Oxford). Kennedy D.L. (ed.) 1996 The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18. Kennedy D.L. 1997 Roman roads and routes in north-east Jordan. Levant 29: 71-93. Kennedy D.L. 2000a The Roman army in Jordan. (London). Kennedy D.L. 2000b The frontier of settlement in Roman Arabia: Gerasa to Umm el-Jimal...and beyond. Meditezraneo antico 3: 397-453. Kennedy D. & al-Husan A.G. 1996 New milestones from northern Jordan 1992-1995. Zeitschrift für Papyrologie und Epigraphik 113: 257-262. Kennedy D.L. & Freeman P. 1996 Southern Hauran survey 1995. In P. Bikai & V. Egan (edd.) Fieldwork in Jordan. American Journal of Archaeology 100: 512513. Kennedy D.L. & MacAdam H.I. 1985 Latin inscriptions from the Azraq Oasis. Zeitschrift für Papyrologie und Epigraphik 60: 97-107. Kennedy D. & Riley D. 1990 Rome’s desert frontier from the air. (London). Kettenhofen E. 1986 Zur Siegestitulatur Kaiser Aurelians Tyche. 1: 138-146. Key Fowden E. 1999 The barbarian plain. Saint Sergius between Rome and Iran. (Berkeley). Killick A.1983 Udruh. The frontier of an empire. 1980 and 1981 seasons, a preliminary report. Levant 15: 110-131. Killick A. 1986 Udruh and the southern frontier. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297. (Oxford): 44-57. Kindler A. 1989 The numismatic finds from the Roman fort at Yotvata. Israel Exploration Journal 39: 261-266. Konrad M. 1992 Flavische und spätantike Bebauung unter der Basilika B von Resafa. Damaszener Mitteilungen 6: 313-402. Konrad M. 1996 Frühkaiserzeitliche Befestigungen an der Strata Diocletiana ? Neue Kleinfunde des 1. Jahrhunderts n. Chr. aus Nordsyrien. Damaszener Mitteilungen 9: 163-180. Konrad M. 1999 Research on the Roman and early Byzantine frontier in north Syria. Journal of Roman Archaeology 12: 392-410. Konrad M. 2001 Der spätrömische Limes in Syrien. Resafa V. (Mainz).

Lander J. 1984 Roman stone fortifications. Variation and change from the first century A.D. to the fourth. British Archaeological Reports Int. Ser. 206. (Oxford). Lewin A. 1990 Dall’Eufrate al Mar Rosso: Diocleziano, l’esercito e i confini tardo-antichi. Athenaeum 78: 141165. Lewin A. forthc. Kastron Mefaa, the equites promoti indigenae and the creation af a late Roman frontier. In M. Piccirillo (ed.) Umm er-Resas II. Littmann E., Magie D. & Stuart D.R. 1910 Greek and Latin Inscriptions. The Southern Hauran. (Leiden). Luttwak E.N. 1976 The grand strategy of the Roman empire. From the first century A.D. to the third. (Baltimore). MacAdam H.I. 1989 Epigraphy and the Notitia Dignitatum (Oriens 37). In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. British Archaeological Reports Int. Ser. 553. (Oxford): 295309. MacAdam H.I. 1990 Ptolemy’s geography and the Wadi Sirhan. In P.L. Gatier, B. Helly & J.P. Rey-Coquais (edd.) Géographie historique au Proche-Orient. (Paris): 5575. Meshel Z. 1989 A fort at Yotvatah from the time of Diocletian. Israel Exploration Journal 39: 229-238. Millar F. 1993 The Roman Near East 31 B.C. - A.D. 337. (Cambridge, Mass.). Mouterde P. & Poidebard A. 1945 Le limes de Chalcis. (Paris). Nicasie M. 1998 Twilight of empire. The Roman army from the reign of Diocletian until the Battle of Constantinople. (Amsterdam). Parker S.T. 1986 Romans and Saracens. A history of the Arabian frontier. (Winona Lake, Ind.). Parker S.T. (ed.) 1987 The Roman frontier in Central Jordan. Interim report on the Central Limes Arabicus Project 1980-1985. British Archaeological Reports Int. Ser. 340. (Oxford). Parker S.T. 1989 The fourth century garrison of Arabia: strategic implications for the south-eastern frontier. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. Proceedings of a colloquium held at Ankara in September 1988. British Archaeological Reports Int. Ser. 553. (Oxford): 355372. Parker S.T. & Lander J. 1982 Legio IV Martia and the legionary camp at el-Lejjun. Byzantinische Forschungen 8: 185-210. Reddé M. 1986 L’installation du camp et la réorganisation du système militaire en haute-Egypte sous la tétrarchie. In M. El-Sanghir, J.C. Golvin, E. El Sayed & G. Wagner (edd.) Le camp romain de Louqsor. (Le Caire): 20-31. Ritterling E. 1903 Zum römischen Heerwesen des ausgehenden dritten Jahrhunderts. In Festschrift zu Otto Hirschfeld sechzigstens Geburtstage. (Berlin): 345-349. Ritterling E. 1925 Legio. RE 23-24: cols. 1211-1837. Roll I. 1989 A Latin imperial inscription from the time of Diocletian found at Yotvatah. Israel Exploration Journal 39: 239-260. Sartre M. 1982 Trois études sur l’Arabie romaine et byzantine. (Bruxelles). Southern P. & Dixon K.R. 1996 The late Roman army. (London). Speidel M.P. 1975 The rise of the ethic units in the Roman imperial army. In H. Temporini & W. Haase (ed.) Aufstieg und Niedergang der Römischen Welt II.3. (Berlin): 202-231. Speidel M.P. 1977 The Roman army in Arabia. In H. Temporini

100

Ariel Lewin: Diocletian: Politics and limites in the Near East

& W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II.8. (Berlin): 687-730. Speidel M.P. 1987 The Roman road to Dumata (Jawf in Saudi Arabia) and the frontier strategy of praetensione colligare. Historia 36: 213-221. Thomsen P. 1917 Die römischen Meilensteine der Provinzen Syria. Zeitschrift des Deutschen-Pälastina Vereins 40: 1-103. Van Berchem D. 1952 L’armée de Dioclétien et la réforme constantinienne. (Paris). Watson A. 2000 Aurelian and the third century. (London). Whittaker C.R. 1984 Frontiers of the Roman empire. A social and economic study. (Baltimore). Williams S. 1985 Diocletian and the Roman recovery. (London). Wolff S.R. 1996 Archaeology in Israel. American Journal of Archaeology 100: 725-768. Zuckerman C. 1994a Les campagnes des tétrarques, 296-298. Notes de chronologie. Antiquité Tardive 2: 65-70. Zuckerman C. 1994b Aur. Valerianus (293/305) et Fl. Severinus (333), commandants en Arabie, et la fortresse d’Azraq. Antiquité Tardive 2: 83-88. Zuckerman C. 1998 Comtes et ducs en Egypte autour de l’anne 400 et la date de la Notitia Dignitatum Orientis. Antiquité Tardive 6: 137-147.

101

Limes XVIII

102

The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau Chaim Ben-David From the northern bank of the Wadi Hasa, west of Khanzireh, and the slope down to Zoar-Safi, one can see the impressive remains of a well-built Roman road which leads up from Zoar to the area of Khanzireh, covering a difference in elevation of c.1300m over a distance of 8kms as the bird flies. The road, 3-6m wide with well-built curbs, passes along a limestone outcrop of the lower Jebel Khanzireh across to a sandstone cliff on the slope of Wadi Sarmuj. Crossing over to the sandstone cliff meant the road had to ascend a steep cliff c.300m high. At the foot of the cliff the road must traverse a most extraordinary route - the cliffs of the Wadi Sarmuj to the north and the steep ravine of Wadi Hasa to the south leaves between them a narrow outcrop c.50m wide over which passes the road. At the western end of this outcrop there is a square fort measuring c.20m x 20m. The remains of the road are clearly discernible on the sandstone slopes, sometimes hewn, sometimes constructed. They can be easily followed until a distance of 1.5kms from Zoar-Safi. Apparently the lower section of the road and the junction with Safi has been washed away in the many floods which sweep through this area. After going along this road with slides and maps, we will discuss its importance in re-evaluating the location and role of Zoar as a major junction in the network of Roman roads between Iudea Palestina and Arabia on the way east to the via nova Traiana and the eastern frontier of Arabia.

steps built across it. These steps are found in the limestone zone (Fig. 2) as well as in the sandstone zone. This road thus joins the other stepped roads known from provincia Iudaea-Palaestina (Kloner 1996). The margins, retaining walls, as well as the steps were constructed using the local rock in each zone. In the sandstone zone, some sections of the road were cut in the rock. The course of the descending road can be divided into three main sections, according to the geological formations it crosses.

In all the studies to date concerning the Roman road system east of the Dead Sea (Mittmann 1982; Roll 1994, 1999; Dearman 1997), only one ancient road was marked ascending from the south-eastern end of the Dead Sea to the Moabite plateau. Following an excursion1 east of the Dead Sea, impressive remains of a hitherto unknown, well built Roman road were discovered on the northern bank of the Wadi Hasa, west of Khanzireh and down to Safi-Zoar, ascending from Zoar to the Moabite plateau. The discovery of this road has important implications for the understanding of the road system in the area and has an interesting contribution for the understanding of the Madaba Map.

In the first section, the road descends moderately over the limestone slopes, and exploits fully the soil cover of the limestone bluffs. This trend of ‘hugging’ the limestone slopes as much as possible is very characteristic of the routes or naqbs going down from the mountains of southern Jordan, as the limestone slopes are more convenient for building roads than the sandstone formations. This can be seen in Naqb Dahal north of Feinan, Naqb Mas‘uda south of Petra and especially along Naqb Rubai on the southern slopes of Jabal Harun near Petra where a 4m wide built ancient road descends over a limestone slope to the Wadi Khusheiba and the ‘Arava Valley. This phenomenon can be observed also in the ancient roads descending from the Moabite plateau to the Dead Sea, eg. at the lower end of the Roman road from Esbus to Livias (Piccirillo 1998), on the roads descending from Kerak (Worschech & Knauf 1985) and along the road descending from Kathrabba to the Dead Sea (Mittmann 1982). In the limestone zone, two parallel sections were discovered. The remains at the junction between the two sections suggest that the western one is the earlier. Unmistakable traces of steps and bends in the road can be discerned from a distance (Fig. 3). The passage from the limestone range to the sandstone formations requires the road to descend a nearly 300m high cliff (Fig. 4). Since a modern road was made on and parallel to the ancient one, only few traces of retaining walls and steps were discovered in this section. At the foot of the cliff, the road passes through an unusual topographic formation - a 50m wide ridge between the cliffs of the Wadi Sarmuj and a tributary of the Wadi Hasa. At the western end of this ridge, there is a 20m square fort (Fig. 5). Near the fort,

The general course of the road The road connecting the Safi-Zoar area to the Moabite plateau (Fig. 1) has to overcome a vertical elevation difference of some 1500m across a linear distance of less than 10kms. On its way, the road crosses the geological formations characteristic of southern Jordan: igneous rock at the bottom, sandstone cliffs in the middle and steep limestone slopes and plateau at the top. The road has been identified to date along some 12kms and is usually 4–5m wide. It is bounded on both sides by a 0.5– 0.6m wide built margin. Retaining walls, up to 2m high at places, were constructed as necessitated by the topography. One of the remarkable features of the road is scores of built 1

The road was first seen incidentally during a tour to the Kerak plateau. We went to a beautiful observation point, called Medinet er-Ras on the south-western end of the plateau and apart from the beautiful view of the Dead Sea area we saw a dirt road going down the steep cliffs towards Safi. The road, not marked on the maps we held, went steeply down 800m through spectacular scenery but did not reach the Dead Sea. Climbing back was not easy but with the afternoon sun in our backs we saw the remains of a well-built ancient road along sections of the modern dirt road. We informed the Dept. of Antiquities of Jordan and were more than happy to be invited by the former Director general Dr. Ghazi Bisheh, and together with Dr. Fawzi Zayadine we returned to the area and showed them remains of the road. They agreed that this Roman road is indeed a new discovery. The participants in the excursions along the ancient road were mainly G. Pel’i, A. Kloner, Z. Meshel and Y. Fixler.

103

Limes XVIII

pottery was observed on the ground, covering the early Roman through the Byzantine periods, including painted Nabatean ware and late Roman red ware.

Dead Sea connects to the via nova Traiana between Philadelphia and Thornia, near Rababatora, which, according to the number of miles from Philadelphia (62) should be south of the River ’Arnon .

The traces of the road are clearly discerned on the sandstone slopes - sometimes cut in the rock and sometimes built (Fig. 6). On the final, steep descent into the Wadi Sarmuj very impressive remains of the road are preserved (Fig. 7). The road can be easily followed up to the alluvial terrace on the south bank of Wadi Sarmuj some 1.5kms from Safi-Zoar. The remainder of the road and its connection to Safi was carried away by the many floods that hit the area.

What then, is the route of the road between Thamaro and Rababatora? Aharoni (1963: 34-35) and Mittmann (1982: 179) suggested that the connection passed through Zoar and continued north along the eastern shore of the Dead Sea, and from there ascended through Kathrabba to Kerak and Rabba-Rababatora. Rothenberg (1967: 164), who accepted that Thamaro was at ‘Ain el ‘Arus, suggested that the connection marked on the tabula ascends from Qasr EtTelah (Toloha in the Byzantine sources), through the Wadi Hisha and Naqb Dahal to Edom. Finkelstein (1979: 31-32) believes that the author of the tabula intended to mark the main road from Jerusalem to ’Aila, which connected to the via nova Traiana of Transjordan at its southern end. Thus, claims Finkelstein, it is not plausible that whoever wishes to go from Thamaro to ’Aila through the mountains of Edom should have to ascend to Rabba, which is situated far to the north-east. Following Rothenberg, he too supposes that the road from Thamaro continued east to Qasr Et-Telah and from there through Naqb Dahal to the environs of Buseira in Edom, where it joined the via nova Traiana. This suggestion certainly forces him to claim that the connection marked on the tabula is mistaken, and it is south of Thornia station and not north of it.

The road and its contribution to the understanding of the road system between Arabia and Iudaea-Palaestina On the Tabula Peutingeriana (Weber 1999) (Fig. 8) a connection is marked between the road system of Arabia and that of Iudaea-Palaestina south of the Dead Sea. On the road that starts at Elusa a place called Thamaro is marked, whence the road arrives at Rababatora, which is marked on the via nova Traiana between Philadelphia and Thornia. Of these five sites, which relate to the identification of the road marked south of the Dead Sea, three are identified beyond any reasonable doubt: Elusa (Tsafrir et al. 1994: 191) in the northern Negev, Philadelphia in modern ‘Amman (Schmitt 1995: 279), and Thornia in Tuwaneh (Fiema 1993, 1997), in northern Edom, east of Tafileh. For the identification of Thamaro, which is mentioned among other sources by Eusebius (Onomasticon 8,8), as “…a village at one day's journey from Mampsis, on the way from Hebron to ’Aila, which is today a fort of the soldiers”, and is marked on the Madaba map as well, two main sites were proposed. Alt (1935: 34) suggested identifying it near the southern end of the Dead Sea, at Qasr el Juheiniye and ‘Ain el ‘Arus in the ‘Arava, a fort and a village respectively. Aharoni (1963) suggested a more southerly site, at ‘Ein el Husb in the ‘Arava. Alt’s suggestion was accepted by Rothenberg (1967: 163), Donner (1992: 69), Gichon (1993) and Schmitt (1995: 324) and Aharoni’s by Avi-Yonah (1976: 99), Finkelstein (1979: 31) and Roll (1989: 260).2 As for the site of Rababatora3, which is mentioned nowhere else, Domaszewski (1898: 69) suggested that it may be a combination of two names: Rabba (Schmitt 1995: 287) in Moab, which lies north of Kerak and Betthoro (ibid: 110) which is commonly identified with the Roman camp at Lejjun, some 13kms south-east of Rabba. His suggestion was followed by others (Alt 1935: 33; Aharoni 1963: 34; Bowersock 1983: 175; Schmitt 1995: 286; Graf 1999: 232). In any case, based on the points shown on the tabula, it is clear that the road marked as passing south of the

The discovery of the Zoar ascent presents an opportunity to suggest that it is in fact the connection in the Tabula Peutingeriana, because it is the shortest built Roman road that connects the roads of Iudaea-Palaestina to the via nova Traiana in Arabia. In any event, the discovery places Zoar as a major junction in the connection between the road systems of the provinces of Arabia and Iudaea-Palaestina. The road going south from Jerusalem and that coming from Gaza via Mampsis, descended from the Negev to the northern ‘Arava through Naqb Juheiniye and Naqb Mezell (on today’s maps Ma‘ale Tamar and Ma‘ale Peres). On this road, there is a large Roman fort (38m square), Qasr Juheiniye (Gichon 1993), which, as pointed out, has been identified by some scholars as Thamaro. Along this route, remains of a built road, small forts and watchtowers were observed4 (Frank 1934: 257-259, 279-280; Marcus 1973), and it is marked as a Roman road on Roll’s map (1994). From this road, there is a direct connection to Zoar and from there to Moab. Thus, a quick connection was created between provincia Iudaea-Palaestina to the via nova Traiana and the military centre at Lejjun.

2

4

Tsafrir et al 1994: 247 presented the two views without deciding in favour of either, and so did Alliata 1999: 84. 3 For a detailed linguistic analysis of the name Rababatora and its connection to names appearing in ancient Egyptian sources, as well as in Bronze and Iron age place names in Moab, see Weippert 1995.

Lately, a new survey was conducted along this road section, including accurate measurements of the remains and their comparison to the Zoar ascent. This survey will be published under the studies conducted by Yoel Fixler: ‘Mapping and analysis of Roman roads and paths in Eretz Israel based on aerial photographs’, Ph.D. Diss., Bar Ilan University (in prep.).

104

Chaim Ben-David: The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau

Tharais with the site whose remains stand at the top of the Roman road ascending from Zoar, presented in this paper. This site, not mentioned in previous surveys, comprises of a number of buildings, some constructed of large, dressed stones, alongside a large water cistern.6 These remains are situated at the foot of the highest peak in the area, called Medinet er-Ras, located above the village of Khanzireh. On its summit is a fortified site (Miller 1991: 149) containing pottery from the Bronze, Iron and Nabatean periods. Glueck (1939: 95) viewed this place as the key to the defense of the south-western corner of the Moabite plateau in the Iron age, and the scholar Van Zyl (1960: 6465), who studied biblical Moab, suggested identifying the place with Luhith.

The road and its contribution to the understanding of the Madaba Map On the edge of the surviving section of the Madaba Map, east of the Dead Sea, is marked the city of Charachmoba, today Kerak. Between a mountain ridge to the east and another one forming a cliff above the Dead Sea, two villages are indicated, ‘Aia and Tharais (Fig. 9). The two sites are located on the map between the two largest rivers east of the Dead Sea, the ’Arnon River to the north and the Wadi Hasa, identified by some scholars with biblical Nahal Zered5, to the south. The artist who created the map located the settlements on the topographical terrace between the steep cliff to the east of the Dead Sea and, the mountain ridge above, portrayed with a moderate gradient.

Is there a meaning to the name Medinet er-Ras? In Arabic ‘medina’ means town and ‘er-Ras’ means the head (in all its senses). Is the meaning, therefore, ‘the town at the top of the mountain’, or ‘the main city’? There is apparently no such instance in the Arabic onomastica.

The accepted identification of ‘Aia is at Khirbet ‘Ay (Donner 1982: 184), today the village ‘Ay, located 6kms south-west of Kerak, at the northern end of the intermediate terrace, rich in springs, below the upper ridge of the Kerak plateau. At Khirbet ‘Ay remains from the Roman and Byzantine periods were found (Miller 1991: 109), and the identification of the place, based on the preservation of the name and the archaeological finds, is quite convincing. The identification of Tharais, however, is more complicated. Donner (1982: 184-188) summarized all the earlier suggestions, rejected them and suggested instead identifying Tharais at Khirbet Tarin, today within the village of Irak located c. 5kms south of ‘Ay on the same intermediate terrace.

We suggest that the name Medinet er-Ras represents the preservation of the name Tharais: Medinet er-Ras = the town (Tha) rais. The city itself can possibly be identified with the large settlements at the foot of the hill of Medinet er-Ras: Khanzerih-Tayyibeh, located 1km east of the mountain, or at Khirbet Dubab (Miller 1991: 148-149), 3.5kms to the south-east. At both sites Nabatean, Roman, and Byzantine pottery was found.7 In any case, Tharais must be identified, in our opinion, as the city that stood at the top of the paved ascent leading up from Zoar eastward toward the King’s Highway, and this is why it was marked on the Madaba Map.

Donner states that he found there sherds from the Roman and Byzantine periods, a fact borne out by the survey of Miller (1991: 117). In any case, the question arises - why do these settlements appear on the map in the first place? They are not known as biblical or Christian sites, and they are not mentioned in any other historical sources.

Bibliography Aharoni Y. 1963 Tamar and the roads to Elath. Israel Exploration Journal 13:30–42. Alliata E. 1999 The legends of the Madaba Map. In M. Piccirillo & E. Alliata (edd.) The Madaba Map Centenary. (Jerusalem): 47-101. Alt A. 1935 Aus der Araba II. Zeitschrift des Deutschen Palästina-Vereins 58: 1-59. Avi-Yonah M. 1954 The Madaba Mosaic Map. (Jerusalem). Avi-Yonah M. 1976 Gazetteer of Roman Palestine. (Jerusalem). Ben-David C. 1999 Identifying Aia and Tharais to the east of the

In the opinion of the scholars who have studied the Madaba Map (Avi-Yonah 1954: 16; Donner 1982: 25), the artist had before him a map of the roads of his period. The marking of milestone sites on the way to Jerusalem, the marking of sites on the Scythopolis-Jerusalem road that have no biblical significance, and the marking of sites along the coastal road from Gaza to Egypt comprise the evidence for this view. We suggest that ‘Aia and Tharais were marked on the Madaba Map because of their location at the top of the Roman and Byzantine ascents which led up from the Dead Sea eastwards (Fig. 10).

6

While showing the remains of the Roman road to Dr. Bisheh, then Director-General of the Dept. of Antiquities of Jordan and Dr. Zayadine in November 1998 we visited this site. Pottery from the early Roman period and painted Nabataean fineware found at the site are important for dating the first phase of the ancient road. 7 Khirbet Dubab is located at an important junction from which one can descend the slopes of the Wadi Hasa and cross over it southwards. It is possible that a branch of the biblical King's Way passed this way, as would appear from the recently published map of late Bronze and Iron age sites in the area of the Wadi Hasa by Bienkowski 1995: 30, Fig. 1. It appears that this secondary Roman road also passed through Khirbet Dubab which crossed the Wadi Hasa to the west of the via nova Traiana, and even to the west of the modern Kerak-Tafila road. For recent excavations in Kh. Dubab, see Bienkowski et al. 1997; 1999.

‘Aia-‘Ay stood at the top of the ascent of the paved Roman road, which is called by scholars to this day the Ascent of Kathrabba or the Ascent of Luhith, and this explains its appearance on the Madaba Map. It is suggested to identify 5

For an overview on the identification of Nahal Zered in the Wadi Hasa based on the Madaba Map and a well-grounded argument for the rejection of this identification, see Elitzur 2000.

105

Limes XVIII

Dead Sea. Cathedra 93: 57-66 - in Hebrew. Bienkowski P. 1995 Observations on late Bronze-Iron age sites in the Wadi Hasa, Jordan. Levant 27: 29-37. Bienkowski P., Adams R., Philpott R. A. & Sedman L. 1997 Soundings at Ash-Shorbat and Khirbat Dubab in the Wadi Hasa, Jordan: the stratigraphy. Levant 29: 41-70. Bienkowski P. & Adams R. 1999 Soundings at Ash-Shorbat and Khirbat Dubab in the Wadi Hasa, Jordan: the pottery. Levant 31: 149-172. Bowersock G.W. 1983 Roman Arabia. (Cambridge). Dearman A.J. 1997 Roads and settlement in Moab. Biblical Archaeologist 60: 205-213. Domaszewski A.M. von 1898 Die Namen Römischer Kastelle am Limes Arabicus. In Beitrage zur Alten Geschichte und Geographie: Festchrift für H. Kiepert. (Berlin). Donner H. 1982 Mitteilungen zur topographie des Ostjordanlandes anhand der mosaikkarte von Madaba. Zeitschrift des Deutschen Palästina-Vereins 98: 174190. Donner H. 1992 The mosaic map of Madaba. (Kampen). Elitzur Y. 2000 Zared or Αρεα ? – One significant detail in the Madaba Map. Scripta Classica Israelica XIX: 155– 162. Fiema Z. T. 1993 Tuwaneh and the via nova Traiana in southern Jordan. Annual of the Dept. of Antiquities of Jordan 37: 549–550 Fiema Z. T. 1997 At-Tuwana – The development and decline of a classical town in southern Jordan. In G. Bisheh, M. Zaghloul & I. Kehrberg (edd.) Studies in the History and Archaeology of Jordan VI: 317-322. Finkelstein I. 1979 The Holy Land in the Tabula Peutingeriana – a historical-geographical approach Palestine Exploration Quarterly 111: 27-34. Frank F. 1934 Aus der 'Araba I. Zeitscrift des Deutschen Palästina-Verein 57: 191-280. Gichon M. 1993 Mezad Tamar. The New Encyclopedia of Archaeological Excavations in the Holy Land (Jerusalem): 1437-1440. Glueck N. 1939 Explorations in Eastern Palestine III. Annual of the American Schools of Oriental Research 18-19. Graf D.F. 1999 Roman roads east of the Jordan. In M. Piccirillo & E. Alliata (edd.) The Madaba Map Centenary (Jerusalem): 230-234. Kloner A. 1996 Stepped roads in Roman Palestine. ARAM 8: 111–137. Marcus M. 1973 Ma’ale Tamar-Peres: an ancient road on the Hatzera Ridge. Hazeva field school. Israel Nature Protection Society - in Hebrew. Miller J. M. (ed.) 1991 Archaeological survey of the Kerak Plateau. (Atlanta). Mittmann S. 1982 The ascent of Luhith. In A. Hadidi (ed.) Studies in the History and Archaeology of Jordan 1: 175-180. Musil A. 1907 Arabia Patraea I: Moab. (Wien). Piccirillo M. 1998 The Roman Esbus-Livias road. In M. Piccirillo & E. Alliata (edd.) Mount Nebo – New Archaeological Excavations 1967–1997. (Jerusalem): 133–149. Roll I. 1989 A Latin imperial inscription from the time of Diocletian found at Yotvata. Israel Exploration Journal 39: 239-260. Roll I. 1994 Roman roads. In Y. Tsafrir et al. (edd.) Tabula Imperii Romani – Iudaea Palaestina, Maps and Gazetteer (Jerusalem): 21-22 and maps. Roll I. 1999 The roads in Roman-Byzantine Palestina and Arabia. In M. Piccirillo & E. Alliata (edd.) The Madaba Map

Centenary (Jerusalem): 108-113. Rothenberg B. 1967 Negev (Ramat Gan) – in Hebrew. Schmitt G. 1995 Siedlungen Palästinas in griechischrömischer Zeit. (Wiesbaden). Tsafrir Y., Di Segni L. & Green J. 1994 Tabula Imperii Romani – Iudaea Palaestina, Maps and Gazetteer. (Jerusalem). Weber E. 1999 The Tabula Peutingeriana and the Madaba Map. In M. Piccirillo & E. Alliata (edd.) The Madaba Map Centenary. (Jerusalem): 41–46. Weippert M. 1995 Rababatora. In M. Weippert & S. Timm (edd.) Meilenstein – Festgäbe für Herbert Donner. (Wiesbaden): 333–338. Worschech U. & Knauf E. A. 1985 Antike Strassen in der nordwestlichen Ard-el-Kerak. Zeitschrift des Deutschen Palästina-Vereins 101: 128-133. Van Zyl A. H. 1960 The Moabites. (Leiden).

106

Chaim Ben-David: The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau

Fig. 1. The ascent of Zohar – view from the west of the Dead Sea.

Fig. 2. Built steps across the road in the upper, limestone zone.

107

Limes XVIII

Fig. 3. The built bends of the road at the lower end of the limestone ridge.

Fig. 4. The new dirt road cut on the course of the ancient road that descends through the 300m high sandstone cliff.

108

Chaim Ben-David: The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau

Fig. 5. View from the top of the sandstone cliff towards the fort that controls the passage at the foot of the cliff.

Fig. 6. A section of the road cut in the sandstone at the descent into Wadi Sarmuj.

109

Limes XVIII

Fig 7. Remains of the road and the impressive bends in the lower section of the road.

Fig. 8. The tabula Peutingeriana - Iudaea-Palaestina and Arabia.

110

Chaim Ben-David: The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau

Fig. 9. ‘Aia and Tharais in the Madaba Map.

111

Limes XVIII

Fig. 10. ‘Aia and Tharais and the Roman roads in the vicinity of the Dead Sea.

Fig. 11. Sites east of the Dead Sea mentioned in the text.

112

Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations Tali Erickson-Gini Past limited archaeological excavations at the Nabataean/Roman site of Avdat (Oboda) in the central Negev have produced inconclusive results for dating the construction and occupation of a military camp. The difficulty in obtaining satisfactory means for dating this camp appears to be the result of the secondary deposition of pottery sherds and coins, dated to the Hellenistic and mainly to the early C1st AD (derived from the soil in nearby middens and used in the construction of the camp) and minimal finds dating to the actual occupation of the camp in the late Roman period. Recent excavations have produced evidence that the camp was constructed in the late C3rd or early C4th over pockets of ash deposits and debris dated to the Hellenistic and early C1st AD and over remains of a Nabataean structure dated to the second half of the C1st AD. The camp appears to have been occupied briefly in the early C4th AD and contains no evidence of reoccupation.

myrrh through the Negev from Petra to the port of Gaza.3 Avdat appears to have been abandoned for several decades, possibly dating from the time of the conquest of Gaza by the Hasmonean king, Alexander Jannaeus at the beginning of the C1st BCE. It was reoccupied in the later part of the C1st BCE, the Nabataeans established a fortified route through the Ramon Crater in order to connect Petra with Gaza by way of Avdat.4 Avdat was continously occupied from this time and into the late Byzantine period. This town appears in the Tabula Peutingeriana (Fig. 1.2). The site was totally destroyed by a local earthquake in the C7th CE (Fabian 1996: 25).

Introduction This paper reconsiders the date of the camp at Avdat (Oboda) in light of recent archaeological fieldwork at this site. Avdat is situated on a high flat plateau, 600m ASL, in the heart of the Negev highlands (Fig. 1.1).1 At this site there are archaeological remains of Hellenistic period camp grounds, a Nabataean settlement, over 200 man made caves and tombs, a military camp and a late Roman and Byzantine town. This settlement may have been named Avdat in honour of Avdat (Obodas) I, the Nabataean king who defeated Alexander Janneus in a battle in the Golan Heights (Josephus Ant.Jud. XXIII.375) or in honor of Avdat (Obodas) II (Negev 1997: 3).

History of the research of the camp Remains of a large, square structure, approximately 100 x 100m in size and located 300m north-east of the acropolis was noted in early surveys. This structure was identified as a ‘Roman’ army camp and was drawn and published by Jauseen, Savignac and Vincent in 1904 (Jaussen et al 1905: 414) and again by A. Musil in 1908 following his visit to the site in 1902 (Musil 1908: 122-124, Figs. 65, 88); (Fig. 2.1). According to the French team, the camp was stripped of building stones in order to construct the fort located on the acropolis. Woolley missed the camp during his survey of Avdat in 1914 but the omission was noted in The Wilderness of Zin (Woolley & Lawrence 1914-1915: 95).5 Based on the layout of the architecture he believed that it was constructed in the “late imperial” period and questioned the French team’s suggestion (later repeated by Negev) that the Byzantine fort was constructed

The site was originally chosen by the Nabataeans in the second half of the first millennium BC as an easily defended camping ground and strategic watchpost overlooking major transportation and trade routes that ran through the central Negev.2 At Avdat, Hellenistic period remains were found all over the site but especially in camping grounds on the northern side of the plateau near the military camp and in the area of the acropolis (Negev 1996: 67; 1997: 1). These camp grounds were probably occupied during seasonal excursions by the Nabateans who transported expensive goods, such as frankincense and

1

I would like to thank Prof. J. Magness and B. Dolinka for their suggestions in preparing the text and also for their support and encouragement in presenting this paper. I would also like to thank Dr. U.Dahari and Dr. G. Avni of the Israel Antiquities Authority as well as Dr. Z. Feima and Dr. B. Saidel for their support and encouragement in addressing the subject of this paper. 2 These same routes are lined with a series of Iron age forts, and one of these forts is located directly opposite Avdat (the Har Arkov fort, Israel Map Ref.1258/0217; Cohen 1986: 139-141). At strategic locations along the major trade routes running through the central Negev there are forts and fortresses from the Iron age, Persian, Hellenistic (early Nabataean) and Roman periods. The proximity of Persian period and Hellenistic Nabataean sites to Iron age forts is found elsewhere in the Negev and Hellenistic Nabataean reoccupation of one these forts has been found at ‘Qasr Ruheibeh on the road between Halutza (Elusa) and Rehovot-in-theNegev. C1st AD Nabataean reoccupation was found in the Iron age fortress at Ein Hazeva in the Central Arava and possibly in a small, unexcavated fort on the ancient road between Rehovot-in-the-Negev and Nessana (Israel Map Ref. 103030/040995; UTM 644152/425918). Nabataean reoccupation of Iron age installations has also been noted in Jordan (Parker 1987: 64).

3

Diodorus (XIX.94.5) quotes Hieronymous of Kardia (C3rd BCE) on Nabataean trade of myrrh and other perfumes from Arabia to the Mediterranean. 4 This has been confirmed by excavations at the Hellenistic-Nabataean fort of Ein Rahel in the Arava. The early route by-passed the Ramon Crater and went through the Darb es-Sultan to the Avdat area and onwards towards Gaza and north Sinai through Nessana (Erickson-Gini, Israel & Nachlieli forthc.). In the 1914 survey, Woolley noted the importance of this route: “The best Petra road runs direct from Khalasa to Nagb el Gharib and W. Merzaba (being here called the Darb Sultaneh, the King’s Highway)…” Woolley & Lawrence 1914-1915: 99. A section of the new route through the Ramon Crater between Avdat and Sha’arRamon was surveyed in detail by Z. Meshel and Y. Tzafrir (1974; 1975). 5 Woolley visited the site between the 9 and 15 February 1914, a few days after he parted ways with Lawrence at Ain el-Guderat in northern Sinai (Chapman & Gibson 1996: 97,100).

113

Limes XVIII

with stones stripped from the army camp, pointing to the fine stone dressing of the fort as opposed to the rough building material used in the army camp.6 Joint excavations were carried out in the camp in 1975 and 1976 under the direction of A. Negev on behalf of the Hebrew University, Jerusalem and R. Cohen on behalf of the Israel Department of Antiquities. Although the excavations have been referred to by both of the excavators in several publications, to date no final report of the excavations has been published. The excavators indicated in their preliminary publications that they did not agree on the date of the army camp. In the year after the excavations Negev wrote:

before AD106 by the Nabataeans. He mentions that the ceramic finds from the excavations were “…minimal and insufficient to determine whether the camp was occupied only during the C1st AD or after the annexation of the Nabataean realm” (Cohen 1982c: 245). Elsewhere Cohen reports that not a single complete vessel was found and claimed that the pottery could be dated to the C1st-C2nd AD. He also pointed out that the external plan of the camp appeared to be reminiscent of Roman fortresses and suggested that it may have been constructed in the early C2nd AD by Roman forces, presumably after the Roman annexation of Nabataea in 106 (Cohen 1980: 44). In response, Negev stood by his assertion that the pottery from the camp dated to the first half of the C1st AD and pointed to the similarity between the pottery found in camp and the pottery he excavated and published from the nearby “potter’s workshop”. Negev also stood by his theory that the camp was abandoned in the middle of the C1st AD, possibly after the transfer of troops and their deployment against invading Arab tribes (Negev 1996: 7 1-72). Significantly, Negev and Cohen also excavated a caravanserai dated to the C2nd and C3rd AD, located next to the potter’s workshop in the vicinity of the army camp.7

“During our work at Oboda in 1959, we established that it was a Nabatean rather than a Roman military installation. We renewed our investigations in 1975 and 1976 with the aim of gaining more information. The camp is 100 m. square, with two gates, the main gate facing east, opening on the Petra-Gaza road, the other on the south, facing the Nabatean town. At the four corners of the camp there were large rectangular towers (and not round ones, as drawn by the earlier scholars). In the interior there are two broad streets, running east-west, north-south. In each of the quarters thus formed, there are two barracks, ten rooms to each, arranged in two rows of five. In the interior of the south gate, there are two large guardrooms. The pottery, lamps and coins found in the barrackrooms were all not later than the middle of the 1st century C.E. Save for the fragmentary pottery and the few coins, nothing else was found in the excavated loci, and it seems that the camp had been abandoned in an orderly manner by the middle of the 1st century. At the southern gate fragments of charred wood, probably of the wooden gates, were found covering the threshold.

The 1999 excavation of the army camp Between March and December 1999, the army camp was excavated extensively by the present author and my codirector, P.Fabian, on behalf of the Israel Antiquities Authority (Fig. 2.2).8 Nearly 50% of the camp was excavated and work was concentrated on the southern half of the area. This area included the previously unexcavated principia located along the back of the western wall of the camp, as well as four barrack blocks, a series of casement rooms and towers along the southern and eastern walls of the camp (Pl. 1.1). In addition, we excavated the main gate of the camp on the eastern side of the structure, located directly opposite the principia (Pl. 1.2) and the streets surrounding the barrack blocks were excavated. Two casemate rooms located along the eastern wall as well as rooms and what may be a wicket located between the principia and the southern wall were not excavated.

The walls and towers of the camp were built of large blocks of hammer-dressed stones. Many of these were uprooted at a later date, at the times of Constantine the Great, as witnessed by coins of that emperor found in the debris left by stone-robbers. These stones were re-used in the construction of the much smaller, late-Roman, citadel erected on the acropolis of Oboda. This may be taken as positive evidence that this camp, abandoned by the middle of the 1st century C.E., could never have formed part of a Roman limes system..” (Negev 1977:622-624).

Elsewhere, Negev reports that the barrack rooms in the camp were completely clean, indicating that the troops were removed from the camp, possibly in order to defend the Nabataean border in northern Arabia sometime in the second quarter of the C1st AD (Negev 1988: 42).

Compared to Negev’s description of the camp, our initial research showed that Musil’s survey of the camp was the more accurate rendering, even prior to excavation (see Fig. 2.1). True to that plan, the camp contained a principia area with rooms along the western side of the enclosure wall. The additional space used in the principia required a

Cohen, however, rejected the idea that the camp was built

7

Cohen (1982c: 245-246) reports that this structure contained a “…rich ceramic assemblage characteristic of the 2nd and 3rd centuries CE” as well as coins of Antonius Pius (138-161), Philip the Arab (244-249) and Claudius Gothicus (268-270). He also reports that nearly 80 coins dating to the second half of the C4th AD were found “overlying” the rubble of the structure. 8 P.Fabian is preparing the final report which will contain in depth analysis of the pottery and other finds. Other members of the team included our two field supervisors, G. Sarai and N.S. Paran. Photographs of the camp were made by Z. Segiv and C. Amit. Section drawings and top plans were drawn by V. Essman, S. Persky and T. Cornfeld. Computer graphics were produced by Y. Sherenko.

6

Woolley & Lawrence 1914-1915: 99. In Woolley’s words, “ I have not mentioned here the Roman camp lying to the north of the town. Judging by its plan, it is of quite late imperial date. Though I did not see it, I venture to doubt the theory put forward by its discoverers that it was dismantled and that the fortress and wall-tower were built out of its material. The stone dressing of these buildings is precisely in keeping with their period, identical with that of Esbeita and El Auja; with all due caution I suggest that the camp was a rather rough and temporary affair not much earlier in date than the fort itself, possibly even used while the fort was building.”

114

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

reduction in rooms of the two barrack blocks located in the south-west quadrant of the camp and each of these blocks contained only 6 rooms, compared to the regular blocks of 10 rooms. In addition, Musil’s description of the main gate and adjacent towers on the east side of the camp, facing the principia, was also correct. However, as Negev noted, the interval and corner towers of the camp are rectangular and not semi-circular (Negev 1977: 624). It should be noted that the towers of the camp bond with the exterior walls.

Room 66 and in the next room, Room 65 (W175), (Fig. 2.4). The upper part of a Gaza wine jar, dated between 300 and 450 (Majcherek 1995: Form 2; 166-168; Pl.5:1-5), was found embedded into the floor of Room 66. A long baulk section that extended from the western enclosure wall of the camp to the street outside of the principia was excavated directly opposite the main gate of the camp (Fig. 4.3). This section clearly revealed a thick plaster floor in Room 80 and in the adjoining room Room 74 (Pl. 1.3). It should be pointed out that the plaster floors in these rooms were the only of their kind found anywhere in the camp. The construction of this type of floor in these rooms indicate the importance of their functions and Fabian has suggested that one of these may have served as the aedes or shrine of the camp.10 These floors were covered with a heavy layer of collapsed stones (probable earthquake collapse), some of which penetrated and punctured the surface (Pl. 1.4). The pottery found in this layer and under the debris of the punctured floor in Room 80 included a restorable late Roman cooking pot, fragments of a Beit Natif style lamp and rims belonging to another Gaza wine jar (Majcherek 1995: Form 2; op.cit.). Two tabuns or clay lined cooking pits, were sunk into the floor of Room 80 next to where the late Roman cooking pot was recovered.

a. The middens Our excavation confirmed that this camp contained only one occupational phase. Some of the streets and rooms appeared to have been built over quarried soft limestone (Fig. 4.1). The camp appears to be constructed so that the centre lies at lower point than the outer walls, some of which were constructed on “nari” or hard limestone outcrops. Pockets in the limestone outcrops were filled with heavy ash deposits and pottery (Fig. 2.3). Based upon our present knowledge there are two middens underneath the southern portion of the camp located in the south-east area of the camp and partially under the south-west barrack block and under one of the streets. The first deposit, Midden 1, contained Hellenistic pottery and several Hellenistic Nabataean coins.9 This midden was clearly disturbed by the construction of the late Roman structure, and foundation trenches belonging to several walls (eg. Walls 88, 101 and 103) located around the entrance room (Room 77) and the street leading to the south-east corner tower (Room 54) cut through the midden. In this area the street was constructed from a thick layer of crushed limestone that was designed to cover the ash and debris (Fig. 4.2). Midden 1 appears to have extended as far as the north-east corner room of the south-east barrack block (Room 16). Here a Hellenistic lamp was recovered in the ash layer directly under what was probably the dirt floor of the late Roman structure. Midden 2 was located near the centre of the camp under the eastern side of the south-west barrack blocks and part of the street (see Fig. 2.3). The pottery found in this midden, unearthed in our excavations, included partially intact Nabataean painted ware bowls CE corresponding to Schmid’s Phase 2b-c (Schmid 1996: 645) and early Roman lamps dated to the first half of the C1st (Fig. 3.1). Barrack walls 10, 11 and 12 were constructed in a foundation trench that cut into this midden.

The east and west walls of Room 80 (W146 and W148), oriented north to south, were constructed, albeit offset, on the foundations of two earlier walls (W201 and W200 respectively) dating to the second half of the C1st AD (see Fig. 4.3). The foundations of yet a third wall (W155), belonging to the early structure, was revealed joining W200 (under W148) below the street level between the principia and a room in the barracks block (Room 20), (Pl. 1.5). Under Room 80, the baulk section revealed a layer of fill (ash and soil with pottery sherds and coins) overlying the early Nabataean floor. The Nabataean floor was constructed over a layer of small stones overlying smooth quarried limestone bedrock. The construction trench of the late Roman wall (W146) can be clearly traced cutting through the fill as far as the top of the Nabataean foundation wall (W201) in Room 80. The same stratigraphy was found on the opposite side of the balk section in Room 74 located between W145 and W147. A coin dated to Year 2 of the Jewish Revolt (AD68) and Nabataean painted pottery sherds corresponding to Schmid’s Phase 3a (Stucky et al 1994: 283; fig.11; Schmid 1996: fig. 11) were excavated by the author in the Nabataean layer above the bedrock in Room 74.11 A second coin dated to the 2nd year of the Jewish Revolt was found nearby under the late Roman strata in Room 80. The fill of ash and soil separating the earlier Nabataean strata from that of the late Roman floor contained large amounts of pottery that appear to have originated in dumps near the

b. Nabataean structures under the camp, late Roman occupation in the camp Evidence for earlier structures was found mainly under the central area of the principia located along the western enclosure wall, and wall foundations (W161;W162) revealed at floor level in the south-west casemate room, 9

The numismatic evidence reported here is based on a preliminary reading of the coins by H. Sokolov of the Israel Antiquities Authority. A final report of the coins will published by Ms.Sokolov at a future date. 207 coins were found in the excavation of the camp and it was possible to identify 143 of these after cleaning in the preliminary reading.

10

P. Fabian pers. comm. Oct. 1999. Nabataean painted fine ware of this type, fabric and decoration has also been found in the main siege camp, Camp F, at Masada in recent excavations (J. Magness pers. comm.). 11

115

Limes XVIII

camp. This layer also contained Hellenistic pottery and coins and a mixture of later Nabataean pottery, dating mainly from the first half of the C1st AD.

Preliminary analysis of the artifacts indicates that the numismatic evidence corresponds in date with that of the ceramics (Fig. 3.2). The majority of coins date to the reign of Aretas IV or are generally described as ‘Nabataean’, ie. pre-106 (Fig. 2.5).13 Only one coin could be assigned to the reign of the last Nabataean king, Rabbel II and only one Roman coin dated to the early C2nd AD was recovered. Two or three other coins appear to date to the late C2nd and early C3rd AD. The remainder of the coins (16% of the assemblage) found in the camp dates to the late C3rd and first half of the C4th AD. These coins were found at floor level throughout the camp and in a few cases they were found below the living surface in the principia (Room 61) and in the entrance to an interval tower located along the southern enclosure wall (Room 52; cf. Fig. 2.6). Contrary to Negev’s assertion that late Roman coins were left in the debris by stone-robbers in the C4th, these coins were found in several locations at floor level or street level throughout the camp and below layers of collapse.14 In addition, there was no evidence to show that any secondary use of the camp or any of its rooms took place in a period later than the original construction and occupation. Thus no suitable explanation can be provided for the provenance of these coins other than that they were deposited during the occupation of the camp in the early C4th AD.

After the abandonment of the camp most of the walls in the centre of the principia appear to have been robbed down to their foundations and even to the bedrock. These walls were evidently constructed from finely dressed stones in secondary use obtained from earlier Nabataean structures in the immediate area. Only one or two of these stones were left in the principia, but they testify to the high quality of building stone used there as opposed to the roughly cut stones used throughout the rest of the camp. In addition, part of a Nabataean inscription was found in the collapse layer in the principia which also appears to be a building stone in secondary use in the camp. The area of the main gate was similarly robbed out, possibly for the same reason (Pl. 1.2).12 Thus, it appears that fine building stones from earlier Nabataean buildings on the plateau were robbed twice. First, these finely dressed stones were reused in the military camp for specific architectural features. Second, following the abandonment of the camp these stones were removed from portions of the camp and reused in the construction of the Byzantine town. c. Primary or secondary depositions?

Conclusions A mixture of Hellenistic and early C1st pottery and coins was found throughout the camp in all the rooms and the streets, from the surface and to what may be assumed to be the living surface of the camp. Coins of Aretas IV (9BCAD40) were found throughout the camp at every level, as well as C1st AD “Herodion” style lamp fragments (Fig. 2.5). Based on this evidence Negev proposed that the camp was built and occupied by the Nabataeans in the first half of the C1st AD. The pottery found in the recent excavations was initially sorted and recorded by the author and awaits final publication by Fabian. However, along with the Hellenistic and C1st AD, a very small amount of late Nabataean painted sherds, described as Nabataean II “debased” painted ware (Gunneweg et al 1988: 333-336; fig.7.2, 4-8) was observed. This type of pottery was found in fieldwork conducted by the present author in more recent excavations of late C2nd and early C3rd structures elsewhere at Avdat (Erickson-Gini forthc.) and also at Mampsis (Erickson-Gini 1999: 22-24, 84-85, fig. 1.8.112). It was also found in the late Roman caravanserai excavated by Negev and Cohen (Cohen 1982c: 245-246). In addition, a large amount of Nabataean course wares was also recovered which have been dated generally to th C1st and C2nd AD. The pottery found throughout the camp closely parallels the pottery found in the surrounding dumps and the “potter’s workshop” published by Negev (Negev 1974; 1986).

1: Archaeological data The materials used in the construction of this camp have complicated the dating and the understanding of the occupational history of this fort. The source of Nabataean pottery and coins uncovered throughout the camp is to be found in the same middens and debris located in the immediate area of the camp and even in pockets below the camp itself. These artifacts were probably introduced into the camp when the builders utilized deposits of ash and soil as construction material, particularly in the interior of the walls. Compared with the small quantity of late Roman ceramic material found in the camp, the large quantity of Nabataean sherds appears to be overwhelming.15 Upon closer inspection, it becomes evident that this material is unrelated to the actual occupational phase of the camp. Nabataean pottery and coins were found in middens underneath the camp, in fills under floors and throughout all the layers, including collapse layers in the rooms and streets of the camp. Building stones for the camp were easily obtained from the bedrock outcrops on this plateau. However, the construction of the camp demanded a large 13 16% of the coins found in the camp dated to the Hellenistic period. Notably, Hellenistic coins were found either in the area of Midden 1 or in layers of fill under the principia (Fig. 2.7). 14 Due to the nature of the dirt floors found throughout the camp, it is probable that even more late Roman coins were actually located below what were assumed to be the floor surfaces. The exact level of the floors was difficult to determine. 15 The paucity of occupation finds has been observed at other Roman military sites such as Mezad Tamar, Arad and Tel Beer Sheva (Gichon 1976: 86).

12

The large stones used to construct the two guard towers flanking the east gate were also robbed out. This included the wall foundation in the northern guard tower. In the south guard tower it appears that the work was in progress and left unfinished. A group of stones stripped from the tower was found 1.5m east of this tower.

116

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

amount of soil for various purposes. For example soil was used for levelling floor surfaces, for mud plaster on interior walls, and on the light roofing over the rooms. 16 Soil was also utilized for bonding the interior of the walls, which were constructed with two facings and filled with rubble and earth.17 Due to the small amount of soil deposits located on the plateau, most of the soil located in the immediate vicinity of the camp had to be obtained in middens. These middens contained artifacts that had been in use hundreds of years prior to the construction of the camp. These deposits contained a large amount of Nabataean sherds, worn Hellenistic and eastern sigillata sherds, stamped Rhodian amphora handles and coins. This is mixture is identical to that found by Negev in his excavations of the surrounding area in the dumps and ‘potter’s workshop’. Cohen’s observation that their excavations of the camp did not produce any restorable pottery forms is justified: the pottery and coins from earlier periods were mixed and weathered, and together with soil and ash they were redeposited in the camp as construction material.

of some of the rooms. This earthquake is probably the same that partially destroyed the town in the late C4th or early C5th.19 Thus it would appear that the camp was abandoned sometime before this event. Negev’s contention that the camp was built and manned by the Nabataean army in the first half of the C1st AD and his proposal that it was abandoned in the mid-C1st, or as late as AD70 has been proven incorrect by the recent excavations. These excavations clearly revealed that the camp’s principia was constructed over a Nabataean building dated to the second half of the C1st AD. This structure was well dated with coins from the second year of the First Jewish Revolt as well as Nabataean painted ware pottery dated to the third-quarter of the C1st AD. Although Negev believed that the camp was Nabataean, this theory has been met with scepticism by some researchers (Kennedy & Riley 1990: 170-172). It has nonetheless influenced others, for example Gichon, who changed his original dating for the establishment of the fort at Mezad Tamar from late Roman to Nabataean, mainly on the basis of the similarity of the masonry and wall construction with that of the camp at Avdat (Gichon 1976: 91). It appears that his excavation of Mezad Tamar was plagued with the same lack of material clarity as that of the camp at Avdat, including a well cleaned late Roman fort (Gichon 1976: 86) and Nabataean middens from earlier periods under and next to the fort (Gichon 1976: 91). These factors may also be responsible for the uncertainty in dating military installations elsewhere and particularly in Jordan.20

In contrast to the standard construction of stone slab ceilings found elsewhere at Avdat and other parts of the Negev, no stone ceiling slabs were found anywhere in the camp and the only evidence of arches was found in the southern and eastern gates. This suggests that the ceilings of all the occupied rooms in the camp were constructed with wooden beams. In addition it appears that the rampart walkway running over the casemate rooms was constructed from wood and was held up by wooden pillars standing on low stone pilasters found in these rooms.18 Considering that wood is an exceptionally valuable commodity in this desert area, it may be assumed that all the wooden superstructures in the camp were dismantled close to the time of the camp’s abandonment, either by the army itself or by the inhabitants of the town. This action would necessarily bring about the immediate collapse of the roofing material over the floors of the rooms, sealing the living surface over which the late Roman coins were found.

Cohen’s suggestion that the camp was constructed by the Romans in the early C2nd AD, presumably after the Roman annexation of Nabataea in 106 is also unsupported. It appears that Cohen’s suggestion was based on circumstantial considerations that grew out of his research on Nabataean and Roman installations along the PetraGaza road (between Moa and Avdat through the Ramon Crater). At those sites Cohen found hard evidence for continued use of the road, originally created by the Nabataeans, into the Roman period as late as the C3rd and C4th (Cohen 1982c: 246). However, this same evidence is wholly lacking in the army camp at Avdat. Compared to those sites, only one Roman coin (of Hadrian) from the

Yet another factor is the evidence for earthquake damage, mainly found in the principia. Walls in this area were thrown down and some sank into the plaster and dirt floors 16 A brittle hydraulic plaster covered the spaces between building stones on the exterior walls of structures inside the camp. No evidence of lime plaster was found in any of the barrack rooms although some traces were found in one of the casemate rooms along the eastern side of the camp. It is probable that the interiors of rooms in the camp, and particularly the interior of the barracks, were covered with a mud plaster. The excavator of the fort at Humayma has suggested that mud plaster was used on the exterior walls of the fort there (Oleson et al 1995: 327-328). 17 Similar to the fort at Humayma, no mortar was observed in the core of the walls (Oleson et al 1995: 325). 18 The function of the casemate rooms remains unclear. The interior walls of these rooms are all inclined downwards from the outer wall and this may indicate that these rooms were not roofed over. They may have been used for housing mounts, either camels and/or horses, or for storing equipment and supplies, but no hard evidence was found to indicate their function. Kennedy & Riley (1990: 171) have suggested that the casemate rooms served as stables.

19 The original town at Avdat was partially destroyed by an earthquake, probably at the end of the C4th or early C5th on the basis of recent excavations in a late Roman quarter carried out by the present author in early 2000 (Erickson-Gini forthc.). 20 The fort at Humayma appears to share many of the features that have created difficulties in dating the camp at Avdat. For example, this fort is built in a pre-existing Nabataean/Roman settlement and is constructed in a similar manner as the Avdat camp (Oleson et al 1995: 321-330; Kennedy 2000: 183). Although the fort at Humayma is larger, they share similar features such as a recessed courtyard in the principia, square corner and interval towers and evidence of a rampart walkway inside the outer wall. We should consider the possibility that the Hadrianic coins, which have been used to date the fort to the early C2nd AD, may be part of a secondary deposition of material from outside the fort. The same criticisms and problems may also be applicable to the fortress at Udruh (see n. 26).

117

Limes XVIII

early C2nd AD was found anywhere in the camp and that was found in a fill containing pottery and coins from mixed contexts. I suggest that the nearby caravanserai (dated by Cohen to the C2nd and C3rd) may have filled the same function in that period at Avdat as the caravanserai at Sha’ar Ramon in the Ramon Crater.21 This caravanserai, excavated by Cohen, was a principal stop on the Petra-Gaza road in the C2nd and C3rd (Cohen 1982c: 244). An inscription found at Sha’ar Ramon indicates that it was used by a unit of the cohors VI Hispanorum (Figueras 1992: 178).22

To these three we may add a fourth, that of the old PetraGaza route leading through the Ramon Crater. However, to date little evidence has been found to suggest that this road was used to any great extent after the mid-C3rd AD.24 This type of distribution of military installations has also been noted in southern Jordan by Fiema, who observed that the Roman forts there are usually located along main roads, at crossroads, high ground passes or in the vicinity of major settlements of areas of economic importance (Fiema 1995: 265). Avdat is suited for all of these conditions, both in the pre-Roman era and in the Roman and Byzantine periods, and it would have been a particularly suitable site for a military installation in the late Roman period.

The evidence provided by the recent excavations point to the camp’s construction and shortlived use in the late Roman period, ie. the very late C3rd and early C4th AD.

The establishment of the camp at Avdat in the Tetrarchic period fits the general scheme of the construction of such installations in the region along major routes. The similarity of the plan and size of the camp in Avdat to known late Roman camps in Jordan such as Da’ajaniya, Kh. Khaw and Umm el-Jimal has already been noted by Kennedy in a recent publication (Kennedy 2000: 212). However, how are we to explain the early abandonment of the camp?

2. Historical considerations During the Tetrarchy a resurgence of Roman military activity and construction is evident in the Negev and Arava and in Jordan (Kennedy 2000: 38). The Roman emperor, Diocletian, implemented far reaching reforms that had a direct effect on this area. These reforms included the administrative transfer of the Negev, Sinai and southern Transjordan from the Provincia Arabia to Provincia Palaestina as well as the transfer of the legio X Fretensis from Jerusalem to the Red Sea port of Aila (Aqaba). This redeployment has been linked with the administrative changes undertaken by Diocletian (Tsafrir 1986: 82-83). The transfer of an entire legion to the Red Sea coast called for the establishment of secure roads through this desert region. The presence of late Roman military installations here reveals that they are usually connected with main roads leading from Beersheva, which probably also had a large camp, (Fabian 1995) to Transjordan and to Aila. Three main routes are known to us: the first connected Beersheva with Zoar by way of Mampsis and Mezad Tamar and a line of small forts located along to the modern Dimona – Sdom highway. The second branched off from Mampsis to Mezad Hazeva in the central Arava by way of the Scorpion’s Pass and a line of small forts. The third, which is recorded on the Tabula Peutingeriana, leads through the central Negev highlands from Beersheva to the fort of Yotvata (Ad Dianam)23 by way of Avdat (Oboda).

The fact that the camp was clean at the time of its abandonment was already observed by Negev (Negev 1977: 624). One surprising feature of our excavation was the lack of hard evidence for almost any daily activity in the rooms and even in the streets. The primary exceptions were the tabuns found in Room 80 of the principia, some evidence of metal work (lead fragments) in one of the barrack rooms and large quantities of ash in another barrack room (Room 6). True to Negev’s observation, the picture is that of an orderly abandonment of the camp but it also appears to have been in very short-lived use. Compared to late Roman military installations in other parts of the Negev, which were apparently occupied into the C5th, such as Mezad Hazeva25 and Mezad Tamar (Gichon 1976: 84, 90), the large legionary camp at Lejjun in Jordan (Parker 1987: 819), and the smaller camp at Da’ajaniya (Kennedy 2000: 163), it appears that the camp at Avdat was abandoned after a brief occupation sometime

identified with Lysa on the Peutinger Table (Meshel 1981: 366). 24 Although Cohen claims in one publication that the road was in use in the C4th (Cohen 1986: 246), an examination of his preliminary reports of sites on that route show that the latest coins found there date to the C3rd. Sites on the Petra – Gaza road west of the Arava include those located between Moa (Moyat Awad) and Avdat: Har Masa, Qazra, Mezad Nekarot, Sha’ar Ramon, Mezad Mahmal, and Mezad Grafon. With the exception of Har Masa (Cohen 1983b: 70) these sites were active in the C2nd and early C3rd AD. Coins from the reigns of Trajan, Commodus and Caracalla were found at Moa (Cohen 1981: 37). Coins from the reign of Caracalla (212-217) were found at Qazra (Cohen 1981: 36). Coins from the reign of Elagabalus (218-222) were found at Mezad Nekarot (Cohen 1982b: 54). Coins from the reigns of Antoninus Pius, Commodus and Caracalla were found at Sha’ar Ramon (Cohen 1982a: 52). Coins found at Mezad Ma’ale Mahmal date from the mid-C3rd to the reign of Gallienus (253-268; Cohen 1983a: 65). 25 The current processing of the Roman period finds from Mezad Hazeva by the present author suggests a continued presence into the C5th. One of the excavators believes that the fort may have been abandoned prior to the 363 earthquake (Y. Israel pers. comm.).

21

The station at Sha’ar Ramon (Qasr el Mahle or Qasr el ‘Ein) has been identified as Maliatha on the Peutinger Map (Meshel 1973: 206). This caravanserai is located one days journey south-east of Avdat (Meshel & Tsafrir 1974: 118). The route between the two sites is lined with a series of milestones (see, Meshel & Tsafrir 1974; 1975). 22 Evidence for the presence of Roman army units was also found in C2nd AD contexts in a military cemetery at nearby Mampsis. Here a centurion from the legio III Cyrenaica and a knight of the cohors I Augusta are buried (Negev 1969: 9). Evidence of the cohors VI Hispanorum was found in C1st AD contexts in Syria at el-Asiri and at Qasr el-Hallabat on an inscription from 212 (Kennedy 2000: 45). 23 Recent surveys of tetrarchic milestones in the southern Arava have raised the possibility that the fort at Yotvata may have actually belonged to a site with a different name (Cosia, Osia or Bosia) and that Ad Dianam of the Peutinger Table is to be located somewhere else on the road to Aila (Kennedy 2000: 193-194, 209). In regard to this road, Meshel has published the plan of a C4th structure with a central courtyard on a bank of Wadi Lussan (Israel Map Ref. 0891/9801) which may possibly be

118

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations in the C4th.26

55 - in Hebrew. Cohen R. 1982c New light on the date of the Petra-Gaza Road. Biblical Archaeologist 45: 240-247. Cohen R. 1983a Mezad Ma’ale Mahmal. Hadashot Arkheologiyot: 65 - in Hebrew. Cohen R. 1983b Har Masa. Hadashot Arkheologiyot: 70 - in Hebrew. Cohen R. 1986 The settlement of the Central Negev in light of archaeology and literary sources during the 4th – 1st millennia BCE. Vol. 1. Unpublished Ph.D. dissertation, The Hebrew University of Jerusalem – in Hebrew. Erickson-Gini T. 1999 Mampsis: A Nabataean Roman settlement in the Central Negev highlands in light of the ceramic and architectural evidence found in archaeological excavations during 1993 – 1994. (Unpublished M.A. dissertation, Tel Aviv University). Fabian P. 1995 The late Roman military camp at Beer Sheba: a new discovery. In J. Humphrey (ed.) The Roman and Byzantine Near East: some recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 235-240. Fabian P. 1996 Evidence of earthquake destruction in the archaeological record – the case of ancient Avdat In Big Cities World Conference on Natural Disaster Mitigation in Conjunction with the Tenth International Seminar on Earthquake Prognostics, Abstracts, Jan. 510, 1996. (Cairo): 25 Fiema Z.T. 1995 Military architecture and the defense “system” of Roman-Byzantine southern Jordan - a critical appraisal of current interpretations. In K. ‘Amr, F. Zayadine & M. Zaghloul (edd.) Studies in the History and Archaeology of Jordan V. (London): 261269. Figueras P. 1992 The Roman worship of Athena-Allat in the Decapolis and the Negev. ARAM 4:1/2 173-183. Gichon M. 1976 Excavations at Mezad Tamar-“Tamara” 1973-1974, preliminary report. Sonderdruck aus dem Saalburg-Jahrbuch XXXIII. Gunneweg J., Perlman I. & Asaro F. 1988 The origin, classification and chronology of Nabatean painted fine ware. Jahrbuch des Romisch-Germanischen Zentralmuseums 35: 314-345. Jaussen A., Savignac R. & Vincent H. 1905 Abdeh (4-9 fevrier 1904). Revue Biblique 14: 78-89, 235-244. Kennedy D. 2000 The Roman army in Jordan. (London). Kennedy D. & Riley D. 1990 Rome’s desert frontier from the air. (London). Majcherek G. 1995 Gazan amphorae: typology reconsidered. In H. Meyza & J. Mlynarezyk (edd.) Hellenistic and Roman pottery in the eastern Mediterranean - advances in scientific studies. Acts of the II Nieborow pottery workshop. (Warsaw): 163-178. Meshel Z. 1973 The roads of the Negev according to the Geography of Ptolemy and the Tabula Peutingeriana. In Y. Aharoni (ed.) Excavations and Studies: Essays in Honor of Professor Shemuel Yeivin. (Tel Aviv): 205210 - in Hebrew. Meshel Z. 1981 The history of ‘Darb el-Ghaza’ – the ancient road to Eilat and southern Sinai Eretz Israel 15: 358371 - in Hebrew. Meshel Z. & Tsafrir Y. 1974 The Nabatean road from Avdat to Sha’ar Ramon. Palestine Exploration Journal 106: 103-118. Meshel Z. & Tsafrir Y. 1975 The Nabatean road from Avdat to Sha’ar Ramon. Palestine Exploration Journal 107: 321.

The camp may have been established for troops temporarily engaged in reinforcing the town’s defences. Two towers were built in the town at the end of the early C3rd and the first contained an inscription stating that the builder was from Petra and dates to 293 (Negev 1977: 660). The Byzantine town was also fortified with a wall, however, this wall was probably built in the late C4th or early C5th.27 In addition, a large rectangular fortress was built on the acropolis adjoining the temenos area. Negev dated the construction of this fortress to the early C4th (Negev 1988: 50). However it was probably constructed after the C4th or early C5th earthquake wrecked the town.28 In addition, the early abandonment of the camp may have been due to the difficulty in maintaining a large military force of several hundred men (a cohort?) for an extended period of time in such an isolated town the size of Avdat or it may have simply been due to a policy of military cutbacks.29 Indeed, the abandonment of the camp at Avdat seems to lend credit to statements made by Zosimus who praised Diocletian for the military build-up and construction of installations in frontier areas and condemned Constantine for withdrawing troops from there in order to strengthen the comitatenses (Zosimus II: 4).30 Bibliography Chapman III R.L. & Gibson S. 1996 A note on T.E. Lawrence as photographer in the Wilderness of Zin. Palestine Exploration Quarterly 128: 94-102. Cohen R. 1980 The excavations at Avdat in 1977. Qadmoniot XII: 49-50 - in Hebrew. Cohen R. 1981 Moa. Hadashot Arkheologiyot: 36-38 - in Hebrew. Cohen R. 1982a Mezad Sha’ar Ramon. Hadashot Arkheologiyot: 52-53 - in Hebrew. Cohen R. 1982b Mezad Nekarot. Hadashot Arkheologiyot: 5426 One important Roman military site (as yet unpublished), appears to share with Avdat camp many of the problematic features in determining the date of its establishment and abandonment: the fortress at Udruh (Adrou/Adroa) in southern Jordan. This fortress may have been built as a legionary fortress paralleling that at Lejjun and, similar to the camp at Avdat, this fortress was also abandoned in the early C4th (Kennedy 2000: 169-170). Other Roman military sites in Jordan that appear to have been abandoned in the early C4th include Qasr el-Uweinid, Qasr ez-Za’faran, Qasr Saliya, Khirbet Shudayyid and Qeseir al-Medeifi (Parker 1986: 145). 27 See n.17 28 The sloping revetments walls found along the southern wall of the temenos are not present anywhere around the rectangular (see Negev 1997: Figs. 16-17; Pls 23, 33). This would suggest that the fortress was erceted after the earthquake of the late C4th or the early C5th BCE (see n.19) 29 In light of recent excavations at Avdat by the present author (see n.8) it appears that Avdat experienced an economic resurgence in the C4th after the collapse of international trade through the town on the Petra-Gaza road. The C4th economy was based on agricultural production and interregional trade. However, it remains unclear how well established this economy was in the town at the beginning of the C4th. 30 Parker 1987: 816 has suggested that this may account for a reduction in the number of troops at Lejjun in the C4th.

119

Limes XVIII

Musil A. 1908 Arabia Petraea 2 (Edom). (Vienna): 25-28. Negev A. 1969 The chronology of the middle Nabatean period. Palestine Exploration Quarterly 101: 5-14. Negev A. 1974 The Nabatean potter’s workshop at Oboda. (Bonn). Negev A. 1977 The Nabataeans and the provincia Arabia. In G. Haase & H. Temporini (edd.) Aufstieg und Niedergang der Romischen Welt II.8. (Berlin): 520686. Negev A. 1986 The late Hellenistic and early Roman pottery of Nabatean Oboda. Qedem 22. Negev A. 1988 Nabataean cities in the Negev. (Jerusalem) in Hebrew. Negev A. 1996 Oboda: a major Nabatean caravan halt. ARAM 8, 1/2: 67-87. Negev A. 1997 The architecture of Oboda. Qedem 36. Oleson J.P., ‘Amr K., Foote R.M. & Schick R. 1995 Preliminary report of the Humayma excavation project, 1993. Annual of the Dept. of Antiquities of Jordan XXXIX: 317-354. Parker S.T. 1986 Romans and Saracens. A history of the Arabian frontier. (Winona Lake). Parker S.T. (ed.) 1987 The Roman frontier in Central Jordan: Interim report on the Limes Arabicus project, 19801985, vol.2. British Archaeological Reports Int. Ser. 340. (Oxford). Schmid S.G. 1996 Nabataean fine ware from Petra. In K.’Amr, F. Zayadine & M. Zaghloul (edd.) Studies in the History and Archaeology of Jordan V. (London): 637647. Stucky R.A., Gerber Y., Kolb B. & Schmid S. 1994 The SwissLiechtenstein excavations at Ez-Zantur in Petra 1993: The fifth season. Annual of the Dept. of Antiquities of Jordan XXXVIII: 271-292. Tazfrir Y. 1986 The transfer of the Negev, Sinai and southern Transjordan from Arabia to Palestine. Israel Exploration Journal 36: 77-86. Woolley L. & Lawrence T.E. 1914-1915 The Wilderness of Zin. Palestine Exploration Fund Annual 1914-1915 (London).

120

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

Fig. 1.1. General map.

121

Limes XVIII

Fig. 1.2. Avdat (Obada) on the Tabula Peutingeriana.

Fig. 2.1. Musil’s plan of the camp.

122

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

Fig. 2.2. The 1999 excavations of the camp

Fig. 2.3. Middens 1 and 2

123

Limes XVIII

Fig. 2.4. Early walls under the camp

Fig. 2.5. The provenance of the coins of Aretas IV

124

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

Fig. 2.6. The provenance of late Roman coins

Fig. 2.7. The provenance of Hellenistic coins

125

Fig. 3.1. Chronology of Nabataean Painted Fine Ware (Stucky et al 1994: fig. 11).

Limes XVIII

126

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

Fig. 3.2 Preliminary analysis of the coins found in the 1999 excavations of the camp.

Fig. 4.1. Section 1-1: Barrack Block B (facing east).

Fig. 4.2. Section 2-2: Street Section V – III (facing south).

127

Limes XVIII

Fig. 4.3. Section 3-3: Principia Section Rooms 80 & 74 (facing south).

Pl. 1.1: 1999 Camp excavations (facing south-east).

128

Tali Erickson-Gini: Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

Pl. 1.2. Main Gate (facing east)

Pl. 1.3. Late Roman plaster in Room 80 (facing north-west).

129

Limes XVIII

Pl. 1.4. Wall collapse from W146 puncturing the floor in Room 80 (facing north).

Pl. 1.5. Nabataean walls (W155 and W200) under the street in front of Room 80 (facing south).

130

The military presence in the countryside of Petra in the C6th1 Zbigniew T. Fiema It has become a common wisdom to assume that in the C6th the Byzantine military presence in southern Jordan was minimal, and that the political/military control over that area was largely left in care of Arab phylarchs, the leaders of local foederati. That was seemingly confirmed by the evidence of drastic troop reduction and widespread abandonment of military sites during the reign of Justinian. Furthermore, that supposed weakening of the frontier defences in the C6th was often understood as responsible for the subsequent military defeats suffered by the Byzantines in their early C7th wars with the Persians and the Muslims. This paper reviews new historical and archaeological evidence which strongly indicates that the presence of the regular army units, in addition to the Arab foederati, is well attested up to the end of the C6th, if not later. The military dispositions in the Petra area are specifically highlighted and updated, as derived from the recently discovered Petra Papyri. Further, the paper argues that the military organization and defensive measures in the C6th reflected the stability and security of the area during that century but became inadequate when facing large scale invasion by the Persians in the early C7th. The impact of the Persian occupation of the Middle East appears crucial in examining the causes of the Byzantine military collapse when facing the conquering Muslim forces in the 630s.

Kithara, 'Ain Gharandal, Rujm el-Sadaqa, and Rekhemtein exhibit C4th to C5th pottery, but always in addition to Nabataean and/or Roman sherds.

The C4th dispositions At the end of the C4th, the period of extensive but poorly understood administrative changes (Brünnow & Domaszewski 1909: 273-80; Mayerson 1984: 1987; Tsafrir 1986; Fiema 1991: 128-133), the original Trajanic Arabia received its final territorial form. Its northern part, with Bostra as a major city, retained the name of Arabia, while the entire area south of the Wadi al-Hasa, emerged as a new province of Palaestina Salutaris/Tertia, with Petra as its main administrative centre. This province included southern Transjordan, southern Palestine and the Sinai. With its location in a remote corner of the empire, vulnerable to internal disorder as much as to a potential external threat, the security of the Petra area would have been a primary concern of the authorities in order to assure the proper maintenance of local administration and economy. This paper will review the evidence of the military dispositions in the Petra area during the Byzantine period (C4th-early C7th), and will discuss new evidence which calls for the reconsideration of the previously held views.

As for the Petra area, the available information is, unfortunately, incomplete. According to the Notitia Dignitatum (Or. XXXIV; 1876: 72-74), major garrisons were stationed in Admatha (modern al-Hammam) to the east of Petra, in Arieldela ('Ain Gharandal) to the southwest, in Robatha (Khirbat Ruwath?) to the north, and in Zodocatha/Kastron Zadacathon (Sadaqa) and Hauana = Hauarra (Humeima), both to the south of Petra. Additionally, if the identification of ancient toponyms is correct, military units were stationed in Bi'r Madhkur, Wadi Sabra, and in the environs of Udhruh, if not in Udhruh itself.2 Other military installations - fortlets or watchtowers - around Petra, Udhruh and Humeima, which cannot be otherwise identified with ancient names, include Qasr Umm al-Rattam, Qasr Wadi at-Tayyiba, al-Mutrab, Ail, and Fardhakh.3 On the basis of the survey material, it appears that all these sites were occupied, and presumably garrisoned, during the C4th. That means that the countryside around Petra and the associated communication network were under a firm military control during that century and most probably in the later period too.

During the tetrarchy, the military presence in the east was reorganized and strengthened. A massive military build-up in some frontier areas, is well-attested in both archaeological and historical sources (eg. John Malalas XII.40). Following the separation of civil from military organization, the army of Palestine (before and after the creation of Palaestina Salutaris) was commanded by a dux. Under his disposition were 4 units of equites Illyriciani and 1 of equites Thamudeni Illyriciani, 2 units of equites promoti indigenae, 5 of equites sagittarii indigenae, and six alae, all of them cavalry. Palestine was also garrisoned by 11 cohorts of infantry and one, possibly two, legions. The army of Palestine totalled between 13,500 and 19,500 (Parker 1986: 136). However, the military build-up in southern Transjordan appears less impressive. The only firmly-dated Diocletianic foundation is the fort at Yotvata (Meshel 1989: 238). Larger castella at Khalde and Quweira may have been C4th reconstructions, since in both places, Nabataean pottery has been found (Parker 1986: 142). Other military sites along and in the vicinity of the via nova, the major Roman highway in the area, including Ail, Khirbet Qirana, Khirbet es-Semeira,

A review of the concepts and a reassessment of the reasons for this military presence should include an analysis of the troops' distribution and the nature and direction of the potential threat. Obviously, military arrangements in Palaestina Salutaris were not of an offensive character. No further expansion to the east or south-east was envisaged, nor is one evidenced. Hence, if these arrangements were indeed defensive, the locations of troops and fortifications should reflect two primary considerations: the nature and direction of the threat, and the location of the 1

I am thankful to David F. Graf, Philip Freeman, Jodi Magness and Ariel Lewin for their comments and suggestions concerning the subject of this paper. I am equally grateful to Jaakko Frösén, Ludwig Koenen and Marjo Lehtinen for their assistance in interpreting the military terms as derived from the Petra Papyri. All errors of omission and interpretation are mine. 2 For site identifications, dating of occupation and further information, see Fiema 1991: Appd. 3; 1995: 263-266, Tables 1 and 2. 3 Graf 1992; 1995; 1997a; 1997b for a comprehensive review of these and other sites.

131

Limes XVIII

economically/politically sensitive areas to be defended. The defences in Palaestina Salutaris do not exhibit concern for a large-scale, well-organized invasion. Such a threat did not in fact materialize in the area until the early C7th Persian invasions, and even then it is uncertain whether or not the defensive system in southern Jordan was actively ‘tested’ by the invaders.

Hammam and el-Mutrab protected agricultural settlements of the Udhruh-Ma'an region. The fortifications extending north-east from Udhruh toward the Qala'at al-Hasa region (and parallel to the via nova) marked the greatest extent of agricultural and settlement expansion in southern Jordan during the Byzantine period. It is evident that forts and troops were located in areas with large population clusters, in areas of economic importance, and where main communication lines were situated. It is also doubtful if these military measures were simultaneously undertaken in the manner of a planned build-up. Forts were constructed, rebuilt, regarrisoned and abandoned throughout the C4th and later, presumably with regard to a constantly changing political and economic environment, and neither a fully linear, nor a defense-in-depth system, were ever established.

Following his designation of the Roman defence in Transjordan as the so-called Limes Arabicus, Parker conceived it to be a broad, fortified zone in the northern, central, and (to a degree) southern sectors, with a single fortified line along the via nova Traiana in the Hisma. During the C4th, that portion of the Limes Arabicus in southern Transjordan was a main line of defence, while élite troops stationed along the fortified line of southern Judaea or the inner zone of defence served as a strategic reserve (Parker 1986: 142-3; 1987: 41). The function of that system was to hold off Arab nomads, but since smallscale raids and periodic nomadic movements would infiltrate the provincial countryside anyway, Parker interpreted the Roman frontier in Palestine and Arabia as a broad zone monitoring nomadic movements (Parker 1987: 48). These interpretations have been challenged by some scholars, who concluded that the nature, distribution and localization of troops and fortifications in Palestine and Arabia were far more compatible with an effort to maintain internal security, rather than to regulate external pastoralist movements, thus the direction of an external threat and the organization of defenses had little in common (eg. Isaac 1984; 1992; Graf 1989). Forts and towers served to police the countryside, to guard the main roads, and to control populations in the main settlements. It was the endemic banditry, nomadic raids and social unrest in the countryside, as attested in historical records describing the conditions of Palestine, the Negev and Sinai, which were the principal threat, and all of these would have originated to the west of the so-called Limes Arabicus.

The C5th-C6th changes During the later Byzantine period, the military alliances arranged between the Byzantine state and the local Arab tribes increased in importance. The foedus with the Tanukhids in the later C4th, the Salihids in the C5th and the Ghassanids in the C6th included the defence of eastern provinces against non-aligned nomads, the preservation of peace and order in the countryside, the containment of the Lakhmid clients of Persia, and the participation of foederati as auxiliaries in Byzantine campaigns. After the elevation of Harith b. Jabala, the Ghassanid to the status of the patrician and the king of all Arab foederati in the Near East, his brother Abu Karib (Abochorabos) declared his vassalage to Justinian in 529, and was granted the phylarchate of Palestine, including southern Transjordan and Petra. However, the phylarchates were neither created, nor invested with the power to replace fully the existing administrative and military systems in the frontier provinces but rather to implement them. The Novellae 102 and 103, issued by Justinian in 536 and referring to the provinces of Arabia and Palaestina Tertia, specify that all military forces in Palestine, including the comitatenses, limitanei and Arab foederati (in this order) were to remain under the command of the Byzantine dux.

Indeed, the defensive arrangements in southern Transjordan differ from those in Byzantine Arabia. The idea of a line or lines of progressive defenses (as in a defense-in-depth) in this area to protect the whole territory from outward incursions is difficult to accept. Also, the distribution of the military infrastructure was conspicuously uneven and regionalized. In southern Transjordan, the via nova Traiana was still a major communication line connecting the major settlements of the countryside, such as Aila, Humayma, Sadaqa, Petra and Tuwaneh. Therefore, the distribution of forts along the road guaranteed security of communication. The overall strategy concentrated on the protection of major clusters of settlements located in different regions. A good example of this is Udhruh, the region located farther east from the via nova, which experienced an expansion of agricultural land and settlements during the Byzantine period. Not surprisingly, that region possessed its own micro-defensein-depth in the form of several parallel lines of towers running north-south (Killick 1986: 444). South of Udhruh, in the environs of Ma'an, two early Byzantine forts at el-

The results of recent surveys claim the evidence of a widespread abandonment of fortifications in the east during the later Byzantine period. South of the Wadi elHasa, the major military sites, such as Jurf ed-Darawish, Da'janiya, el-Mutrab, el-Hammam, Qirana and Quweira, had been abandoned in the late C5th or early C6th. Occupation probably continued throughout the C6th only at the forts of Khalde, Kithara, and the garrisoned sites of Udhruh and Humayma. Inevitably, scholars have regarded this evidence as a sign of military decline and a neglect of security by the central government.4 All this was further reinforced by the malicious Procopius’ remark (Anecdota 24.12-14) on the neglect of limitanei under Justinian. Concurrently, the military and administrative control over 4

See Parker 1986: 149, 152-3, 158, for the survey evidence; 1986: 152, for his historical interpretation of this evidence.

132

Zbigniew T. Fiema: The military presence in the countryside of Petra in the C6th

southern Jordan was thought to have been largely relinquished to the Ghassanid phylarchs.5 But even this arrangement appears to have been ended by the Byzantines who dismantled the Ghassanid phylarchy in 581.

southern Transjordan, such as the 'Amila, Judham and Bali. These tribes remained actively on the Byzantine side even after the interlude of the Persian invasions. New evidence and interpretations

At first, it is debatable, whether the criticism of Justinian's policy is indeed warranted. Inevitably, the emperor, who faced enormous financial strains due to his western campaigns and the Persian wars, tended to economize spending. This policy may indeed be perceived as a weakening of frontier defences and a general decline of military arrangements, but only when numerically compared against the arrangements of the previous centuries. In the light of extant political and economic circumstances, such comparisons seem inappropriate. The political and economic situation in the provinces had greatly changed since the reigns of Septimius Severus or Diocletian, and hence military arrangements had to be modified as well. The state could no longer afford to defend all of its provinces by a protective cordon of troops and fortifications along the frontiers. Instead, the government in Constantinople encouraged local defensive measures, and extended its direct military protection only to those areas which were politically and economically sensitive, and where a direct threat was expected. The guidelines of former frontier arrangements were basically preserved, although, considering the political circumstances of the Justinianic period, the new arrangements were made on a more modest scale, and with some modifications. Procopius recorded extensive constructions and repairs of fortifications in Mesopotamia, northern Syria and Armenia under Justinian (De. Aed. 2.3). The military arrangements in southern Syria and Transjordan were of a different sort. No large-scale invasion was envisaged, and small raids could be easily fended off by the loyal Ghassanids. The smaller communities had increasingly to depend upon themselves, either through private arrangements with local phylarchic groups or friendly nomads, or by constructing their own defences.

Furthermore, recent research and investigations indicate that southern Jordan and the Petra area were not as completely deprived of military protection as previously suggested. For example, of importance is the re-evaluation of the Beersheva Edict which contains a portion of an imperial edict and a schedule of annual payments in cash, assessed community by community. The edict was initially interpreted as concerning three taxes; the old annona militaris, a supplementary tax for civil servants and a special tax for the vicarius.6 Although the payments are generally modest and not excessive, three communities were clearly differentiated from the rest. Adhruh (Udhruh) was assessed 65 solidi, Auara (Humayma) 43, and the military site of Robatha, 42 solidi, all in the Petra area. Notably, Udhruh does not appear to be listed in the Notitia Dignitatum at all. The unknown date of this fragmentary imperial edict is speculated to range from the late C4th/early C5th to the early C6th, and even more specifically around 536.7 But if the Edict is dated to the early C5th, and if the payments represent return of the part of the annona by the military personnel back to the dux Palaestinae, as recently suggested (Isaac 1995: 138-9, 141 - on Udhruh), then the high tax assessed on Udhruh means that the legionary base in there had been organized in the early C5th, after the redaction of the Notitia Dignitatum. In such a case, there is no reason to suspect its abandonment until the C6th. The absence of Petra would indicate that the city was not garrisoned at that time, although an inscription dated to 446 mentions a numerus, stationed there either permanently or temporarily.8 Secondly, while some military sites indeed feature archaeologically detected discontinuity in the C6th, no wholesale abandonment occurred in southern Transjordan. To the abandoned sites probably belong Ail, Khirbet alQirana, B'ir Madhkur, Qasr Umm al-Rattam, Qasr Wadi at-Tayyiba, and 'Ain Gharandal. Major sites such as Udhruh and Humeima were occupied, and quite probably with a military component. Mutrab and al-Hammam were claimed to have been abandoned, yet the name of Admatha occurs in the Petra Papyri (infra). Zodocatha was certainly occupied. The recent re-evaluation of extant historical sources related to sites in the neighboring Negev also demonstrates that a substantial number of military sites

Likewise, the end of the Ghassanid phylarchy seems, on first sight, to be a major blunder of Byzantine policymakers, yet the effect of that has been greatly exaggerated by scholars. The policy of the Byzantine emperors shows clear signs of an adjustment to the political conditions of the late C6th. The 50 year peace treaty of 562 seemingly improved Byzantine-Persian relations. Hence, the existence of a large tribal confederacy under Byzantine control which was obviously directed against Persia and its allies became politically embarrassing and burdensome. Actually, the Byzantine phylarchical system was transformed rather than eliminated, returning to the earlier form of independent alliances with individual local groups; eg. the foedus with the Salih was revived (Shahîd 1989: 303, 476). Similar independent arrangements probably prevailed with the tribes of the northern Hejaz and

6

Alt 1921: 7. See Mayerson 1986: 146-8, for a different interpretation. Alt 1921: 1-13 for the edition of the Beersheva texts. Dating: the C4th/early C5th, specifically before AD 443 (van Berchem 1952: 33-36; Isaac 1995: 138-39), the early C6th (Abel 1967 II: 177-8) and around AD 536 (Mayerson 1986: 148). Di Segni 1993: 499-500 doubts both early and late date for the edict. 8 The inscription is related to the conversion of the Urn Tomb into a church in 446, (Brünnow & v. Domaszewski 1909: 345). The unit is reconstructed as numerus Tertiodalmatarum (Sartre 1993: 82). If Isaac’s hypothesis is correct (1995: 138-139), this unit would have come to Petra only after 443. 7

5 For phylarchoi and the meaning of the title, see Grouchevoy 1995; Mayerson 1991.

133

Limes XVIII

(including garrisons in civilian settlements) were still occupied in the C6th (Isaac 1995: 137-145). Not only the evidence of continuity of military occupation is mounting but also new military sites were apparently constructed in the area during the Justinianic period.9 Although these were of modest size and apparently of local function, their construction and occupation well into the C7th attest to the concern of the local, if not imperial authorities over the safety of the countryside. This evidence confirms the policing the countryside as a primary rôle of the army, in a way to maintain the proper functioning of the political and financial apparatus in the provinces.10

καθωcιωµένος (devotus, devotissimus), often of military connotation, also occurs in the Petra texts.19 It is not without significance that Abu Karib ibn Jabala, the Ghassanid phylarch of Palaestina Tertia appears in the Petra Papyri as a negotiator in a private dispute (Bikai 1996: 534; Kaimio & Koenen 1997: 462; inv. 83). Furthermore, of potential military interest is the appearance in the Petra texts of one Fl. Dusarios, who, in one of the documents, is styled as α[π]Ä πραιφñκτων Καστρου Αµµαθων (Kastrou Ammayvn) - ex-praefect of Kastron Ammatha (Gagos & Frösén 1998: 475). Generally, the α[π]Ä -επαρχων is a designation of a former prefect,20 as well as an honorific title specifically denoting the social class.21 The title of Fl. Dusarios indicates his current social rank or status in relation to his previous appointment, and he does not find it necessary to mention his actual current appointment; the status of an ex-official is sufficient. It may also be symptomatic that Dusarios formally specifies his origins (from Petra) directly after his status. The use of πραιφέκτων rather than επαρχων was perhaps meant to emphasize the former military rather than civilian appointment. By the C6th, the original military prefect of the early empire continued, although on limited scale, to denote the commander of an auxiliary alae or cohortes as well as the numeri, legiones and vexillationes legionis,22 all being primarily the limitanei forces. As for Kastron Ammatha, it specifies the location of former post of Dusarios, where he could have been the commander of a military unit.23

Finally, the Petra Papyri, discovered in 1993,11 provide an important contribution to the clarification of the military presence in the Petra area during the C6th. Apparently, the local tax-system mentioned in the texts also included extra payments in kind for example οινόκρεον (meat marinated in wine)12 which is known from Egypt to form a part of annona militaris payments.13 Although annona was also disbursed to the Arab allies, at least until 581/2,14 the presence of regular army personnel is attested in Kastron Zadacathon, even toward the end of the C6th.15 These were apparently limitanei rather than the foederati. judging from the military ranks such as prior, and ordinarios (πρίωρ, Ôρδινάριος), both being grades of non-commissioned officers,16 as well as cτρατιώτης.17 Even if equites promoti indigenae were no longer in Zodocatha, the unit stationed there in the C6th would be not unlike a numerus known from Nessana18 - the locally recruited soldiers who had served in the garrison of their hometown where they could own houses and landed property. An honorific title

Conclusions Although the pattern of the military presence in the Petra

9

See Magness 1999: 193-4, 199, 205 for redating the forts at Ein Boqeq and Upper Zohar to the mid-C6th. See also Isaac 1992: 210-12 10 Isaac 1992: 420. For the rôle of forts as producing and protecting food sources (also in imperial estates), see Graf 1997a: 280-81. 11 To date (March 2000), the more extensive published reports on the Petra papyri include Koenen 1996a; 1996b; Bikai 1994; 1996; 1997; Daniel 1998; Fiema, Koenen & Zayadine 1997; Kaimio & Koenen 1997; Gagos & Frösén 1998. 12 P. Petra inv. 69.3,2 and 4. I am grateful to L. Koenen for this information and clarification of the term. 13 See P. Berol. 21891, Aphrodito, C6th, where οίνόκρεον clearly occurs in the context of annona payments (Syrcou 1996: 86-91). See Bagnall 1990: 90-92 for further clarification of the term. Notably annona militaris appears to have still survived in Byzantine Syria, Mesopotamia and Egypt at the time of the Muslim conquest, and was still a form of government disbursement rather than the produce of the soldier-farmers left for their upkeep (Kaegi 1985: 593-595). 14 Kaegi 1985: 592. Some units of comitatenses appear in Palestine even after the Persian wars (Kaegi 1975: 65-66). 15 P. Petra inv. 33a, AD593. I am grateful to J. Frösén & M. Lehtinen for this information. 16 Grosse 1918: 126; Jones 1964: 662, 674; but ordinarii could also be civil officials. Priores from Kastron Nessana are attested in Pap. Ness. 26 and 35. See the list of military titles from Nessana in Rubin 1997: 6667. 17 Also in Pap. Ness. 15, 16, 18, etc. For the use of this term, as well as that of katalogos, which also occurs in the Petra text, see Müller 1912: 101-107. 18 Isaac 1995: 145-47, and Rubin 1997: 64-66, argue against the identification of the Nessana unit as Theodosiani. The unclear reference to that unit in Pap. Ness. 15, and the available information on the Nessana unit (not unlike an unit of equites sagittarii indigenae) favor Isaac’s suggestions.

19

Also applied to denote soldiers of numerus in Nessana. It occurs most often in relation to elite imperial officer corps and palace guards (scholae, protectores, domestici, silentiarii), the militarized secret police (agentes in rebus) but also for various officials of the imperial chancellery - Koch 1903: 78-81. 20 eg. high rank officials like απὸ επαρχων πολεος. Also granted to consistoriani upon their retirement (Delmaire 1995: 14-15). 21 See Guilland 1982: 30-31. In fact, these honorifics were also applied to toll collectors, notaries, supervisors of state workshops, or chartularioi, who all belonged to the class of low-level provincial administration (Kazhdan 1991: 133-34). The difference between the high and low-level officials holding this title could be further discerned by a specific rank added, eg. the inscription no. 91 from Aphrodisias - apo eparchon (person being Ñνδοζότατος; late C6th-early C7th?). Other examples from Aphrodisias: no. 10 - apo primipilarion, and no. 152- apo protektoron all examples in Roueché 1989. See the ex-domesticus (απο δοµ[εστικον]) in Pap. Ness. 26. The use of απὸ, which began in the C3rd, indicated that the ex-holder of a particular office was automatically entitled to a particular rank or status with all its privileges and exemptions formally attached to that. For discussion, see Millar 1983: 90, 94, n. 78; Roueché 1989: 23-24. 22 Southern & Dixon 1996: 60-61; Jones 1964: 640; Haldon 1995: 6. Early Roman praefectus castrorum is not meant here, being not in use by the C6th, still less praefectus praetorio vacans –‘the prefect of the army’an extraordinary appointment discussed by Kaegi (1982: 103-13) and Scharf (1991). Praefectus is often used in the context of a commander of a military unit, who could be also titled tribunus or praepositus (Grosse 1918: 152). 23 See the example of Fl. Abinnaeus, praefectus alae at Dionysias, who is often addressed as praepositus, or preafectus castrorum Dionisiados (from the Abinnaeus Archive, mentioned by Southern & Dixon 1996: 61)

134

Zbigniew T. Fiema: The military presence in the countryside of Petra in the C6th

area in the C6th is more ambiguous than in the C4th, yet it indicates its continuity there. The selective abandonment or withdrawal of troops was probably related to a changed emphasis as to the political and economic vitality of the area for the empire. However, the major towns in the area were still accorded the protection by the military, which apparently included the regular army units. Furthermore, local security concerns could also have been shared with the local foederati.

Ishaq in Ibn Hisham 1955: 525-26; al-Baladhuri 1966: 9295). The prior subjugation by the Muslims of Tabuk, Tayma and other commercial settlements in the northern Hejaz meant a Muslim control over the inter-regional trade in which settlements like Aila and Udhruh participated. Thus, the measures taken by these church and civil officials were realistic and beneficial for the economic life of their towns. The conquest of southern Transjordan passed without any specific references in either Byzantine or Arab sources, except that in no place between the Hejaz and the Wadi 'Araba (ie. roughly corresponding to the area of southern Transjordan) did the Muslim forces pass without firmly establishing their authority (Al-Baladhuri 1966: 168).

That such defensive disposition was not in existence in the early C7th, is hardly surprising. One cannot consider the emperors of the C6th to have been responsible for the vicissitudes of the C7th. Neither Justinian nor his successors could envisage the scale of warfare which would be carried out by the Persians in the early C7th, as previous conflicts had been restricted to the regions of Armenia or northern Syria. The Persians of Heraclius' time were no longer interested in conducting war simply as a means to obtain the most profitable peacetime conditions. Instead, they planned and pursued the actual occupation of Byzantine territories through conquest. The shock of the military defeat and subsequent occupation must have seriously affected provinces like Palestine, which had not experienced any large-scale, non-Roman, invasion from the east since the Palmyrene episode in the later C3rd. It is not surprising then that Byzantine civil administration and the military presence simply vanished during the Persian occupation.

Bibliography Abel, P. F. 1967 Geographie de la Palestine I. Geographie Physique et Historique. II. Geographie Politique. Les Villes. (Paris). Alt A. 1921 Die Griechischen Inschriften der Palaestina Tertia westlich der Araba. In T. Wiegand (ed.) Wissenschaftliche Veröffentlichungen des DeutschTürkischen Denkmalschutz-Kommandos. (Berlin). al-Baladhuri Abu-l’Abbas Ahmad ibn-Jabir 1966 The origins of the Islamic state. (Transl. of Kitab futuh al-Buldan by P. Hitti: Beirut). Bagnall R.S. 1990 Five problematic fourth-century pieces. Bulletin of the American Society of Papyrologists 17: 79-93. Berchem D. van 1952 L’armée de Dioclétien et la réforme constantinienne. (Paris). Bikai P.M. 1994 The Petra Papyri. Annual of the Dept. of Antiquities of Jordan 38: 509-11. Bikai P.M. 1996 Petra Church Project, Petra Papyri. Annual of the Dept. of Antiquities of Jordan 40: 487-489. Bikai P.M. 1997 The Petra Papyri. In G. Bisheh, M. Zaghloul, & I. Kehrberg (edd.) Studies in the History and Archaeology of Jordan 6: 343-4. Brünnow R & von Domaszewski A. 1909 Die Provincia Arabia III. (Strassburg). Daniel R. 1998 Toponomastic Mal in P. Nessana 22 and P. Petra Inv. 10 (Papyrus Petra Khaled & Suha Shoman). Zeitschrift für Papyrologie und Epigraphik 122: 195-6. Delmaire R. 1995 Les institutions du Bas-Empire romain, de Constantin à Justinien. I. Les institutions civiles palatines. (Paris). Di Segni L. 1993 Review of M. Sartre Inscriptions de la Jordanie, Tome IV. Pétra et la Nabatène méridionale du Wadi al-Hasa au golfe de 'Aqaba. Liber Annuus XLIII: 496-515. Donner F. M. 1981 The early Islamic conquests. (Princeton). Fiema Z. T. 1991 Economics, administration and demography of late Roman and Byzantine southern Transjordan. (Ph.D. Dissertation, University of Utah. Salt Lake City). Fiema Z.T. 1995 Military architecture and the defence “system” of Roman-Byzantine southern Jordan - a critical appraisal of current interpretations. In K. ‘Amr, F. Zayadine & M. Zaghloul (edd.) Studies in the History and Archaeology of Jordan 5: 261-69. Fiema Z.T., Koenen L. & Zayadine F. 1997 Petra Romana, Byzantina et Islamica. In N.T. Weber & R. Wenning (edd.) Petra. Antike Felsstadt Zwischen Arabischer

The interlude of the Persian invasion and occupation (61429) effectively disrupted and eliminated any military arrangements which generally existed in Palestine in the C6th. A brief respite (629-633/4), following the Byzantine victory and the Persian withdrawal apparently did not suffice to reconstruct fully the military strength and regarrison outlying parts of the empire, such as the Petra area. Notably, the elite equites Illyriciani appear to have been stationed in Caesarea on the eve of the Islamic invasion (Kaegi 1975: 65-67). The soldiers of a cohort, who were eventually executed in Gaza by the Muslims, c.640, were probably stationed somewhere in Palestine before the conquest. Likewise, some sort of a regional defense, most probably of local foederati, was apparently reconstituted in southern Moab just north of the Wadi elHasa, since the first Muslim raid in 629 resulted in a sound Byzantine victory near Mu'ta. Arab sources imply that tribal segments of the Bali, Judham, Lakhm, Ghassan and al-Qayn fought at Mu'ta on the Byzantine side (Donner 1981: 103-107, 115-116). But during the major invasion of 634, southern Palestine was deprived of effective defence. When Gaza was directly threatened, the nearest forces, sent under the patrician Sergius, came from Caesarea. Similarly, the regular army presence in southern Jordan and the Petra area on the eve of the Muslim conquest was minimal or non-existent. Even before the main invasion, three larger towns in the area - Aila, Udhruh and Jarba peacefully capitulated to the Muslim forces in 630 (Ibn

135

Limes XVIII

Tradition und Griechischer (Mainz): 145-160. Gagos T. & J. Frösén J. 1998 Petra Papyri. Annual of the Dept. of Antiquities of Jordan 42: 473-81. Graf D. F. 1989 Rome and the Saracens: Reassessing the nomadic menace. In Géographie Historique au ProcheOrient. Notes et Monographs Techniques No 23. (Paris): 171-211. Graf D.F. 1992 Nabataean settlements and Roman occupation in Arabia Petraea. In M. Zaghloul, K. ‛Amr, F. Zayadine, R. Nabeel & N. Tafaq (edd.) Studies in the History and Archaeology of Jordan 4: 253-260. Graf D.F. 1995 The via nova Traiana in Arabia Petraea. In J. Humphrey (ed.) The Roman and Byzantine Near East: Some recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 241265. Graf D.F 1997a The via militaris in Arabia. Dumbarton Oaks Papers 51: 271-282. Graf D.F. 1997b The via militaris and the Limes Arabicus. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd). Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 123-133. Grosse R. 1918 Die Rangordnung der römischen Armee des 4.-6. Jahrhunderts. Klio 15: 122-161. Grouchevoy A.G. 1995 Trois “niveaux” de phylarques. Étude terminologique sur les relations de Rome et de Byzance aves les Arabs avant l’Islam. Syria LXXII, fasc. 1-2: 105-131. Guilland R. 1982 Etudes sur l’histoire administrative de L’Empire Byzantin. III L’apoeparque – απο επαρχων. Byzantinoslavica XLIII: 30-44. Haldon J. F. 1995 Administrative continuities and structural transformations in east Roman military organisation c.580-640. In J. Haldon State, army and society in Byzantium. (London): 1-20 [reprint from F. Vallett & M. Kazanski (edd.) L’armee romaine et les barbares du 4e au 7e siècle. (Paris 1993]. Ibn Hisham 1955 Sirat Rasul Allah. Mustafa al-saqa et al, eds. (Cairo). Isaac B. 1984 Bandits in Judaea and Arabia. Harvard Studies in Classical Philology 88: 171-203. Isaac B. 1992 The limits of empire. The Roman army in the east. (Oxford). Isaac B. 1995 The army in the late Roman east: the Persian Wars and the defence of the Byzantine provinces. In A. Cameron (ed.) The Roman and early Islamic Near East III. (Princeton): 125-155. Jones A.H.M. 1964 The later Roman Empire, 284-602. A social, economic and administrative survey. (Norman). Kaegi W. E. 1975 Notes on hagiographic sources for some institutional changes and continuities in the early seventh century. Byzantina 7: 61-70. Kaegi W.E. 1982 Two studies in the continuity of late Roman and Byzantine military institutions. Byzantinische Forschungen 8: 87-113. Kaegi W.E. 1985 The Annona Militaris in the early seventh century. Byzantina 13/1: 591-596. Kaimio M. & Koenen L. 1997 Reports on decipherment of Petra Papyri (1996/97). Annual of the Dept. of Antiquities of Jordan 41: 459-462. Kazhdan A. 1991 Apo Eparchon. In Oxford Dictionary of Byzantium (Oxford): 133-134. Killick A. C. 1986 Udruh and the southern frontier. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological

Reports Int. Ser. 297. (Oxford): 431-446. Koch P. 1903 Die Byzantinischen Beamtentitel von 400 bis 700. (Jena). Koenen L. 1996a The carbonized archive from Petra. Journal of Roman Archaeology 9: 177-188. Koenen L. 1996b The phoenix from the ashes: The burnt archive from Petra. Michigan Quarterly Review 35: 513-531. Koenen L. 2000 The decipherment and edition of the Petra Papyri: Preliminary observations. Paper presented at the Edelman Conference “A climate of creativity: Semitic papyrology in context” in honor of Baruch A. Levine March 5-7, 2000. (New York). Lehtinen M. 1998 News on the prosopography of the Petra archive. Paper presented at the XXII Congresso Internazionale di Papirologia, Firenze, Italy. Magness J. 1999 Redating the forts at Ein Boqeq, Upper Zohar, and other sites in south-east Judaea, and the implications for the nature of Limes Palaestinae. In J. Humphrey (ed.) The Roman and Byzantine Near East II. Journal of Roman Archaeology Supplementary Series 31: 198-206. Mayerson P. 1984 “Palaestina” vs. “Arabia” in the Byzantine sources. Zeitschrift für Papyrologie und Epigraphik 56: 223-230. Mayerson P. 1986 The Beersheba Edict. Zeitschrift für Papyrologie und Epigraphik 64: 141-48. Mayerson P. 1987 Libanius and the administration of Palestine. Zeitschrift für Papyrologie und Epigraphik 69: 251260. Mayerson P. 1991 The use of the term phylarchos in the RomanByzantine east. Zeitschrift für Papyrologie und Epigraphik 88: 291-295. Meshel Z. 1989 A fort at Yotvata from the time of Diocletian. Israel Exploration Journal 39 (3-4): 228-238. Millar F. 1983 Empire and city, Augustus to Julian: Obligations, excuses and status. Journal of Roman Studies 73: 76-96. Müller A. 1912 Das Heer Justinians (nach Procop und Agathias). Philologus 71: 101-138. Notitia Dignitatum 1876 Notitia Dignitatum. (O. Seeck, ed. Berlin). Parker S.T. 1986 Romans and Saracens: A history of the Arabian frontier. (Winona Lake). Parker S.T. 1987 Peasants, pastoralists and pax Romana: a different view. Bulletin of the American Schools of Oriental Research 256: 35-51. Roueché C. 1989 Aphrodisias in late antiquity. (London). Rubin R. 1997 Priests, soldiers and administrators: society and institutions in the Byzantine Negev. Mediterranean Historical Review 12: 56-74. Sartre M. 1993 Pétra et la Nabatène méridionale du Wadi alHasa au golfe de 'Aqaba. Inscriptions de la Jordanie. Tome IV. Inscriptions Grecques et Latines de la Syrie. Tome XXI. (Paris). Scharf R. 1991 Praefecti Praetorio Vacantes Generalquartiermeister des Spätrömischen Heeres. Byzantinische Forschung 17: 223-233. Shahîd I. 1989 Byzantium and the Arabs in the fifth century. (Washington DC). Southern P. & Dixon K. 1996 The late Roman army. (London). Syrcou A. 1996 Six Byzantine documents. AFP 42: 84-99. Tsafrir Y. 1986 The transfer of the Negev, Sinai and southern Transjordan from Arabia to Palaestina. Israel Exploration Journal 36: 77-86.

136

Limes Arabicus, via militaris and resource control in southern Jordan George Findlater Recent scholarship on the classical period in Jordan has moved away from the interpretation of Roman military sites as parts of a defensive static system to one that emphasises these monuments as part of a wider socio-economic system. Also more rigorous archaeological definitions of Roman military sites have resulted in a revision of what constituted the military impact on the landscape. Within this new interpretative framework, the Dana Archaeological Survey carried out a three-year project to investigate these monuments as material correlates of resource control from the Iron age to Byzantine period. An area stretching from the Wadi Arabah to the desert fringe in the east, centred on the areas of Dana and Wadi Faynan, was chosen. Encompassing several different terrains, the area was already known to have many Roman military sites. The present survey purposively concentrated on the communication routes that connected these sites. By viewing these sites in the wider landscape, new discoveries have been made and alternative interpretations of these monuments can be advanced.

resource areas which states exploited, maintained and taxed. Due to the presence of major wadis that deeply dissect the steep scarp on the western edge of the plateau, travel between the Wadi Arabah and the uplands is fairly easy.

Introduction This paper sets out to show how evidence, accumulated from almost 400 surveyed sites in southern Jordan, by the Dana Archaeological Survey (hereafter DAS), may impact on Roman frontier studies in the region. DAS was designed as a three year project to understand the processes of resource control and state organisation in the Faynan, Dana and Shaubak areas of Jordan within the historically known periods of the Iron age to Byzantine. The Council for British Research in the Levant, the Seven Pillars of Wisdom Trust, and the Palestine Exploration Fund support the project. The aim of the project is to produce a record of the relationship between state/military sites and settlements within a landscape setting of resource areas. In doing so it seeks to understand imperial, colonial and national processes in the development of cultural systems during the research period. As textual evidence of such relationships varies considerably within this time-scale, the project focused on the archaeological correlates of state activities to allow a more balanced comparison of state control across a wider time-span. It is the intention of this paper to present new evidence, which alters the existing picture of Roman military sites on the frontier, and places such sites within a wider landscape setting of socioeconomic exploitation.

Previous research in project area Until recently, little systematic archaeological survey had taken place in this area of southern Jordan. Musil (1907) and Brünnow & v. Domaszewski (1904-1905) documented highly visible remains along main routes or presumed Roman roads. Glueck’s surveys (1934; 1935; 1939), although much more detailed, were highly purposive and biased towards his view of the Edomite kingdom. Bennett carried out excavations at the site of Buseirah (Bennett 1966). Killick’s (1983a; 1983b; 1986; 1987) survey and selected excavations of a large area between Shaubak and Ma’an have not been published except for a few sites in the immediate vicinity of Udhruh (1983a: 127-131). Hart’s survey of an area to the west of Killick’s project area focussed on the Iron age and was highly purposive (Hart 1986a; 1986b; 1987; Hart & Falkner 1985). Parker surveyed the major Roman/Nabataean sites to the east of the via nova Traiana (Parker 1986). Fiema surveyed a section of the via nova Traiana to the north of the DAS area (Fiema 1993), while Graf surveyed the sections to the south of Shaubak in his extensive study of the via nova Traiana in southern Jordan (Graf 1995). In 1997 Walmsley began a project investigating the site and environs of Gharandal (ancient Arindela; Walmsley 1998: Walmsley et al 1999). Similarly, in 1999 MacDonald, began a survey project extending his Hasa Survey to the south, around Buseirah to the Desert Highway area (MacDonald 1999). In contrast, the Faynan and Fidan areas have been the subjects of intensive work (see Levant 1998). In the wider area, King and MacDonald have carried out purposive surveys in the area (MacDonald 1992; King 1985; King et al 1987; 1989), following Glueck (see above) and Frank (1934).

Research area The survey area was a large section between the Wadi Arabah in the west and the Desert Highway on the plateau to the east (Fig. 1). In the north it was bounded by the Wadi Dahal and the Gharandal to Jurf Ed-Darwish road. The southern boundary was delimited by the Wadi Arja on the Shera’a plateau. This area not only encompasses Jordan’s two major trade and communication routes but also contains a variety of natural environments from Mediterranean to Semi-Arid ranging in elevation from 100m to 1500m ASL. Such variation means that diverse settlement sites and resources are contained within a relatively finite area. Furthermore, the research area was known historically to contain massive natural resources such as timber in the upland plateau areas (Shaubak etc) and copper in the Faynan areas in the Wadi Arabah. These routes and these natural resources form the basis of the

Project background The present survey grew out of the original Dana Archaeological Survey, undertaken by the Centre for Field

137

Limes XVIII

Archaeology, University of Edinburgh, as part of the Royal Society for the Conservation of Nature’s (RSCN) GEF project. This was a baseline archaeological survey of the RSCN Dana Nature Reserve (Finlayson & Baird 1995). During the course of the survey it became apparent that previous archaeological surveys had not adequately described or realised the significance of the monuments in the area. Previous research in the Dana/Faynan area had revealed the massive exploitation of copper resources from the Iron age to Roman period (Hauptman & Weisgerber 1992). However, the control and management of such a valuable resource in this period, and its relationship to two of the major trade and communication routes in the area (namely the Wadi Arabah and Kings’ Highway) had not been studied in any depth. This is in direct contrast to the extensive work carried out on the nature of copper exploitation and resource control during the early Bronze age in this area (Adams 1998; Wright et al 1998).

However, Banning (1986) attempted to review certain aspects of the nomadic-Roman interaction using a mainly archaeological data set on the basis of anthropological models. As with the results obtained by Rosen (1987), both studies suggested that settled communities and pastoralists acted in complement with each other. Although criticised by Parker (1987b) as ignoring substantial historical evidence, Mayerson (1989) pursued a middle course by denying the overt military functions of controlling nomads while questioning the apparent ‘mutualism’ between settled and pastoralist populations. Other scholars have since begun to provide alternative theories on the tactical and strategic rôle of the military in the Roman period in southern Jordan (Graf 1997a; 1997b; Isaac 1992; 1998; Fiema 1995). They contend that historical sources do not reflect such a threat and that many of the historically attested military sites (known primarily from the Notita Dignitatum) are situated well behind the supposed frontier system proposed by Parker. Furthermore, they see the location and variability of many military sites as being determined by economic conditions. Fiema (1991; 1995) attempted to integrate such sites into the overall settlement pattern, linking military location to economic policy or trends, where fluctuations over time in military systems are seen to be indicative of larger socio-economic processes.

As this paper deals predominately with the imperial Roman and Byzantine landscape, the state has been viewed as an essentially exploitative agency, maintaining and building a material infrastructure to tax, exploit and control its chosen area (Sinopoli 1994). The orthodox view of Roman/Byzantine state activities in this frontier province, certainly in the southern sector, has emphasises the overt military nature of one aspect of its material correlate; namely the forts of the so-called limes Arabicus. This system defended the province against external attacks and also policed the internal area. This view has been heavily questioned. The nature of the external threat - nomads from the eastern steppes - is unclear and the pattern of sites does not fully represent a system for the defence of a geographically circumscribed area. The data that is emerging now strains the old view. New theories emphasising the socio-economic framework of these forts seem to fit the evidence better. It is within this more recent framework that DAS carried out the present work.

The tension between both schools of thought is, in part, due to a lack of models to integrate and explain the data sets. Too often the research strategies of classical archaeologists have relied on the use of historical data to direct research. Aside from Fiema and Banning, the debates have rested predominately on mono-causal explanations of system origin and change (ie. in response to a perceived threat or perceived policy) which are based mainly on historical interpretations. If one utilises an annales perspective (see Knapp 1992) to view the data, the varying explanations can be seen as part of a larger cultural system. The nomadic raiding or changes in Roman policy can be viewed as the ‘events’ that intersected with the medium-term cycles of economic behaviour or settlement patterns, conditioned by the long-term effects of climate and geography in this marginal environment. Within this model the present project’s focus on resource control was to explore the imperialism of the Roman impact. The aim was to identify correctly the evidence of socio-economic exploitation and the domination of the colonised population. Recent studies have begun to move away from such a view and have started to explore the colonial process from the colonised perspective where the power relations are seen as unstable (Hall 1993). Although, undoubtedly this would provide a more balanced view of the imperial process in southern Jordan, it is argued here that the Roman military, as the prime agency of imperialism, has not been properly integrated into such a model. Overtly military debates on the nature of the legions have obscured other facets of the Roman military imperial experience. Following Fiema (1991), this project seeks to demonstrate other aspects of Roman socioeconomic exploitation and place them within a more dynamic cultural system.

Traditionally, scholars have concentrated on the larger military sites situated on the supposed frontiers of empire and integrated this with historical sources following the modern interpretation of the term limes (Isaac 1998: 345346). Such research produced grand strategic explanations concentrating almost totally on the defensive systems and the perceived external threat that these systems countered (eg. Luttwak 1979). Such an approach, although perhaps valid in certain key areas such as Syria where the various Persian states posed a clear strategic threat to Rome, does not fully explain the organisation of military sites in southern Jordan. Although Parker’s massive Limes Arabicus Survey has done much to document such sites, the underlying trend of his work has been to reinforce the concept of the frontier zone as a military one (Parker 1986; 1987a). This concept rests on the identification and location of such sites (and the road network) and the direct threat which supposedly inspired the system. In southern Jordan it is presumed to be from nomadic tribes (Parker 1986: 136).

138

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

The ability of states to protect, manage and exploit resources is fundamental to the sustainability and stability of that type of organisation. In the context of this study, the term resource refers not only to natural resources for economic exploitation but also control of trade and communication routes. Resource control in the marginal area under study is critical to the success of any state/entity. In fact, the control and management of a resource like copper (which exists in great quantity in the study area) may have led to state formation in the Iron age (Knauf & Lenzen 1987: 86) or perhaps the raison d’être for the presence of military sites in the Wadi Arabah (Rothenberg 1972). The control of a resource may be managed at a variety of levels and the hierarchy of sites associated with such processes must be explicitly demonstrated. In archaeological terms this may be achieved by observing patterns in the spatial proximity of key archaeological sites (here military sites) to resources and settlement types (Harfield 1988). In this regard, archaeological evidence may show signs of clustering around resource areas, demonstrating a high degree of control. Alternatively, control of resources by states may have been parasitic where only access to the area was regulated. Such a pattern may distribute along routes or distinct landscape features. This approach, however, demands the careful identification of state correlates in the archaeological record, which have not been fully elucidated in previous studies of the area.

only (Gregory 1996: 189-190). Using this kind of methodology, similar sites, such as Al-Hammam (Parker 1986: 101, fig. 45) and Al-Mutrab (Parker 1986: 103, fig. 46), have been included in most lists of military sites (eg. Kennedy 2000). Similarly, within work on the Iron age, many scholars have tended to assign a military function on the basis of ‘square-ness’ and location (eg. Hart & Falkner 1985) backed up by biblical evidence (Glueck 1934; 1935; 1939). The lack of an explicit methodology in this regard hampers such studies. Earlier surveyors such as Glueck (see above), who carried out such massive surveys of Jordan in the 1930s, would, for example, record a square structure on a hill, designate it a fort, link it with others, build a military system and thus construct states/kingdoms etc. This, however, is not to criticise the overall approach. The identification of patterning in the archaeological record is fundamental to the analysis of sites in the landscape. However, the methodology in the initial interpretation of a site’s function has to be more rigorous and logical. Furthermore, adequate settlement hierarchies need to be established if the site type is to be seen in the correct landscape perspective. DAS has shown the critical need to view these sites in the actual landscape, not as dots on a map. Fundamental to this was the need to view military systems within a wider settlement system. This approach has yielded considerable results and, in the few examples that follow, will demonstrate the project’s success at re-interpreting some features of the frontier/limes system.

Material correlates of the state The major focus of the present project began with the most obvious archaeological correlate of ancient imperial state systems - military sites. It is argued here that the historical data for state control of resources in southern Jordan is not sufficient to map changes across the Roman/Byzantine period (Parker 1986: 4-5). However, as the project primarily focussed on aspects of power-relations in a classic ‘colonialist perspective’ (Lightfoot & Martinez 1995: 473), the emphasis was on power-relations. Such power-relations are expressed primarily through domination of economic and military fields. These areas can be mapped through the archaeological correlates of these behaviours, which are military sites in this case (Bartel 1980/81: 14-15). However, the identification of such sites in the archaeological record has always been a matter of debate. Isaac (1992) has pointed out the fallacy of assuming that every major building in a frontier zone is a military one. Gregory’s (1995-7) mammoth study of Roman military sites on the eastern frontier has led to a rigorous re-evaluation of the data set, which provides a basis for more precise interpretations of military systems. Although Parker (1995), and Kennedy and Riley (1990) have established a broad typology of Roman military sites, Gregory has pointed out that many sites may have had different functions belying their military appearance, eg. Qasr Tilah (Gregory 1996: 190). Furthermore, she points out that a site like Aseikhin - a substantial building of rectangular plan around a courtyard - is assumed to have a military character due solely to its geographical position

Ceramic chronology and historical frameworks Underlying all reconstruction of settlement histories is the development of ceramic typologies to date the material remains. As frontier studies debates in southern Jordan follow within tightly defined historical periods, it is useful to explain the basis of the dating mechanisms for settlement reconstruction in this area. Although the pottery of most periods in northern Jordan is well known, nearly all that is published from southern Jordan is from surveys or preliminary reports (eg. Parker 1987a; Brown 1991). Moreover, these publications base their periodisation on stratified sites in northern Jordan, the most notable being Hesban (Sauer 1973). There is, however, considerable regional variation in the pottery of most periods in Jordan, with the published assemblages from northern Jordan not being entirely applicable in the south (Bienkowski 1998; Walmsley 1998). The typology of the published site assemblages from the Limes Arabicus Project (Parker 1987a), which provide one of the few sources of comparative excavated material for Nabataean through Byzantine pottery in the DAS project area, is based on the Hesban system. This and other work (eg MacDonald 1988) presents pottery ranges within tightly defined historical correlates. Within the classical period this follows a tradition of providing tightly defined periods (eg. Hayes 1972; 1997) for fine wares. However, recent work at Petra is providing closer checks on the ceramic chronology of the area (Bignasca et al 1996; Stucky et al 1994). 139

Limes XVIII

Thus, most reconstructions of settlement histories are usually based on ceramic evidence, and which produce a wave effect where settlement histories are perceived to rise and fall. Unfortunately, coarse pottery types (such as cooking wares) rarely change over time and are therefore harder to date. Thus, they are usually neglected in the reconstruction process. This may have the effect of emphasising shorter settlement durations (based on fine wares) whereas, in fact, the coarser wares may point to longer settlement use. This has many implications for the reconstruction of settlement histories during the classical periods. At worst, change may be interpreted when there is none. Schick (1994: 135-137) has elaborated on the limited use of ceramic dating on sites in southern Jordan but views this within a historical perspective. These ceramic difficulties are usually presented as failings of the archaeological data. However, recent data from the site of Gharandal, near Buseirah, is beginning to show the longevity of classical wares into the Islamic period (Walmsley & Grey forthc.). The present paper does not attempt to solve these issues but rather emphasises the internal consistencies of the archaeological data. Using the model discussed above, the erratic snapshots of historical data, and the broad patterns of archaeological process, are only successfully integrated if one views them as different rhythms of a cultural system.

only association with a military feature is its proximity to a ‘military road’ that supposedly led to the fort of Dajaniya and then to the fortress/town of Udhruh. Certainly, Thomsen’s plan of milestones in the area, mainly based on Brünnow and v. Domaskewski’s data, would point towards such a route (Thomsen 1917: Tafel 1). Graf however, (1997a: 128) suggests that the direction of the milestones does not indicate a route leading to the fort of Dajaniya (Parker 1986: 93-94; Freeman 1990). More importantly the lack of a formal route associated with the milestones, created clear problems. Although one could not expect a paved route, a cleared desert track would be usual and is paralleled elsewhere. Additionally, if the track did not go to Dajaniya as the topography indicates, the lack of military/state structures directly to the south (such as towers etc.) clearly does not demonstrate a major route heading south – military or not. However, DAS located two sites to the south: DAS 211, a small tower was located 5kms south of Jurf Ed Darwish and one large site, DAS 217, further to the south. Both contained classical pottery. More significantly though, located immediately to the west of these sites was a stretch of cleared track that was clearly not made by vehicles or graders (Fig. 1). Eight metres wide, lined with stones, situated to the east of the Hejaz railway, it was first thought to be a service track linked with the railway. However in close proximity three milestones were located (Thomsen 1917: 58, nos. 182, 183, 184). Importantly, the milestone just south of Jurf Ed Darwish had two milestones. To the north of Jurf Ed Darwish en route to Wadi Hasa, similar fragments of track were located which were not associated with the railway track. Along this track were located all of Thomsen’s 5 milestone groups 177-181. DAS located one further set of milestones. It is worth emphasising that groups (up to five) of milestones were located, suggesting repeated use of the route.

Via militaris The road system has been rightly been emphasised by Isaac as underpinning the system of forts which linked the frontier area. To determine a system of forts without demonstrating the road system, is, as Isaac called it, an “unstructured procedure” (Isaac 1992: 128). He postulates that, in the area around the Wadi Hasa, there should be evidence for a route to the east of the via nova Traiana. A line of milestones had been found between Wadi Hasa and the fort of Dajaniya (Thomsen 1917: 57-8, nos. 177-184) as well as some near the great fortress site of Udhruh to the south (Thomsen 1917: 56, nos. 172-173). This led scholars to suggest that this was an outer branch road of the limes (Parker 1986: 91). The logic, somewhat crudely summarised, is: line of milestones plus proximity of forts equals road system therefore military system.

It would be reasonable to assume that this classical route followed the same route as used today and is heading towards Ma’an. Unfortunately, to the south of Wadi Hasa/Jurf Ed Darwish area, the modern Hejaz railway has obliterated any evidence for the presence of this route. Moreover, it is difficult to prove this when there are no reported classical sites along this track. Earlier travellers (Brünnow & v. Domaszewski 1905: 7-8 quoting Doughty) reported that this stage could be achieved in half a day from Ma’an to our area, so it may be that sites were not required. However, this does not explain the sites to the north and the length of track with associated milestones. The only site, south of DAS 211 and 217, known along this route was Qal’at Unaiza.

While acknowledging the existence of the milestones, David Graf (1997a; 1997b) has questioned the presence of a road system on the edge of the desert areas linking military sites: a so-called via militaris. In reviewing the evidence, Graf suggested that the line of milestones leading to and from the site of Jurf Ed Darwish (cf. Parker 1986: 91; Kennedy 2000: 159), a small castellum, really only marked the approaches to this fort. However, it must be pointed out that there are no obvious military features apparent in its layout. The thickness of walls noted by Brünnow and v. Domaszewski (1905: 14) at 1.7m which may suggest a military feature, were considered by DAS to be a confusion with internal features. Thus, the ‘fort’ at Jurf Ed-Darwish may correspond to a courtyard layout as the function can be interpreted in many ways. In fact, the

Qal’at Unaiza (DAS 389 fig. 1) was built, according to historical sources, in 1576 as part of the Hajj system to protect pilgrims on the way south to Mecca (Peterson 1986; 1989). On a black basalt hill (Jebel Unaiza) to the west and overlooking the site, DAS located several structures disturbed by modern military use containing classical pottery. This led to a re-examination of the 140

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

Ottoman fort. Several features were immediately apparent to have not been noted in earlier surveys (Peterson 1986: 78-85). The asymmetric entrance of the courtyard structure is unusual, as entrances in this type of plan were usually placed halfway along walls (Fig. 2). At the corner southwest curtain wall an obvious re-build phase was apparent. This meant that the southern wall would have continued west as part of a larger structure not noted in earlier surveys. From the re-surveyed plan it is obvious that the 1576 construction is actually a drastic smaller rebuild of an earlier structure. Further examination noted the differences in room size between the rooms on the east and south faces, and those on the north and east. Additionally, it was also noted that different cement was used in the construction of both types of room size. The re-surveyed plan now shows an earlier larger structure where the entrance fits more symmetrically into the plan. The plan suggests a large courtyard outline (perhaps a caravanserai layout). DAS carried out a careful sherding of the area and found ceramics from the classical, early Islamic and C19th periods.

102-103), of which the latter has long been identified as the site of Admatha (Parker 1986: 101 n.69; Fiema 1991: 295, 301) where a camel unit was stationed. Further surveys of the area by Stein (Gregory & Kennedy 1985: 295-301) have shown that the sites were associated with water management/agricultural techniques (Fig. 3). Parker’s survey demonstrated that they were in use from the later Byzantine into the Islamic period (Parker 1986: 175, 179). Served by a massive reservoir the water was thought to be used for Al-Hammam and then transported through an aqueduct to the site of Al-Mutrab and then on for some distance (Brünnow & v. Domaszewski 1905: 3-4). Kennedy’s recent aerial description of the sites still assigns a military function although notes Gregory’s scepticism (Kennedy 2000: 175). Parker comments on the lack of towers and thin walls compared to other military sites (Parker 1986: 102). Several other features were noted by DAS. Within the debris of Al-Hamman, DAS located numerous red and white polished marble fragments. Some of these were clearly pillar fragments. In addition, a massive grinder (prob. 1.20m diameter) and large limestone weight were found in the debris of Al-Mutrab. This grinder may have been the circular hearth feature noted by Brünnow and v. Domaszewski and is located in the south-west corner (Brünnow & v. Domaszewski 1905: 5, fig. 554). Such features are not usually found in military sites in the area although the lack of similar data from excavated military sites makes comparison difficult.

It is now highly probable that Qal’at Unaiza was another large classical site on an ancient road that leads to the Ma’an oasis. The evidence DAS has accumulated clearly shows a definite cleared track associated by proximity with 9 milestones and over 6 military/state sites or at least regular square structures of varying size. However, one should note that this of course represents a palimpsest of activity where none of the sites should be viewed as contemporary. The main point to draw is the spatial association of sites (which all contain Roman/Byzantine pottery) that proves the existence of a formal Roman route to the east of the via nova. Furthermore, this route can now be seen as the precursor to the Hajj route (Peterson 1989). This route is clearly not associated with the sites of Dajaniya or Udhruh (Parker 1986: 94-98). One cannot now argue for a via militaris since the two major military sites in the area have no physical connection with it. However, this is not to deny a connection between these sites and this route. That the personnel of these obvious military sites patrolled or monitored this route is without question, but they were not part of a fortified military road in the way the Strata Diocletiana was constituted (Whittaker 1994: 136-137). It is more probable that these sites fulfilled a variety of functions that will be discussed below. However, it must be noted that the position of larger military sites like Dajaniya and Udhruh may have been sited with greater consideration of the need to harvest large quantities of water than of specific tactical requirements.

One kilometre to the north of Al-Mutrab, however, DAS located a new site, Khirbat Samra (DAS 332), that has almost exactly the same layout as Al-Mutrab and AlHammam (Fig. 4 and plan Kennedy 2000: 176, fig. 18.3 Site D [Site C, not D, should be Al-Mutrab on Kennedy’s plan]). The pottery was of the same date as the other two sites, ranging from later Byzantine to early Islamic. It was complete in plan and, unlike Al-Mutrab and Al-Hammam, not damaged by bulldozer activity. Measuring 50m x 50m, the walls were of a similar thickness to the other sites. There was a discernible lack of rubble around the site and the even height of the walls suggests that the superstructure was made of mudbrick. Using the methodology followed by earlier projects, the plan shows another courtyard structure, which linked to the identification with Admatha, suggests a military function. However, as noted above the identification of Admatha with Al-Hammam is not secure. Moreover, three (apparently contemporary) forts in a 5kms x 3kms area plainly make no sense. The plans of these buildings have no obvious diagnostic military features. Although Kennedy (2000: 174-176, figs. 18.2 & 18.3), following Gregory, noted the association with hydraulic features, the association of these features was not fully understood. The ground survey by DAS, to the east of the main reservoir and the site of Al-Hammam, located a channel (Channel III) that heads east for 5kms (Fig. 4). This was not the aqueduct thought to be between Al-Hammam and Al-

Forts or farms? That a formal route is directed south to Ma’an should come as no surprise as it was and still is the natural stopping place for travellers heading into the Arabian interior. However, the only tangible classical remains are the sites of Al-Hammam (DAS 391; Brünnow & v. Domaszewski 1905: 3-4; Parker 1986: 101-102) and Al-Mutrab (DAS 331; Brünnow & v. Domaszewski 1905: 4-6; Parker 1986: 141

Limes XVIII

Mutrab (Parker 1986: 102). The aqueduct commented on by earlier surveyors was, in fact, a rubble wall that enclosed a massive 5kms x 3kms area that was irrigated by this central channel. Along the channel were located several sluices from which ran perpendicular channels to the interior of the area. Situated by each of these sluices were small, usually singular, structures (DAS 333, 336, 334). Near the end of the enclosed area a large reservoir was found (DAS 337). Several phases were noticed in the construction of the central channel, suggesting a long period of use. Moreover, the channel carried on past the eastern enclosure wall and thus it seems that an even larger area was to be irrigated.

empire may be invalid given the specific social and environmental factors at play here. However, within the DAS area there is some evidence for a boundary system that, if correctly identified, may prove a useful contrast to the Roman system. The Khatt Shebib was a 20/30kms stretch of walling noted by Kirkbride (1947-48) during the 1930s (Fig. 5). Running from the edge of the Ras En Naqb to just east of Ma’an, the date and function was unknown. Parker and Killick both suggest it could be similar to walls seen in the North African provinces (Parker 1986: 86; Killick 1986: 432436). On the ground, the Khatt Shebib manifests itself as a low rubble wall (or in places orthostatic stones) about 0.60m – 0.90m wide. Harding (1967: 154) postulated that it was a defence against cavalry but this was dismissed by DAS as in many places the wall was clearly not high enough or situated at the bottom of slopes. DAS tracked this line from Kirbride’s area to the Wadi Hasa where it stopped (see also MacDonald 1999). This means that the wall is over 90-100kms long (Fig. 1). It is located along the eastern upland edge of the Jibal and Shera’a mountains. Furthermore, it is associated by proximity with a line of towers that occur along this upland edge (Fig. 1: DAS 231, 230, 238, 226, 202, 205, 203). While classical pottery is found on some sites, Iron II pottery was found on all sites. Although the wall is not physically associated with the line of towers, it would be reasonable to assume that it dates to the period when all towers were in use. The main point here is to observe that the placement of the wall along the edge of the mountain range and the siting of the towers, all in line of site and with very little hidden ground, strongly suggests the maintenance of a boundary area. Kirkbride also noted that it runs along the 100m isohyte line for dry farming (Kirkbride 1947-48: 266-267).

The scale of this irrigation is immense and represents a huge investment. All sites not only contained later Byzantine pottery but early Islamic as well. The enclosure of the walls and the location of the sites; Al-Hammam (containing high status stonework) and the reservoir at the western end of the site - Al-Mutrab and Samra on either side of the central channel III; suggest a centrally planned massive irrigation network. Clearly it was begun in the later Byzantine period and extended. Such a pattern can be discerned elsewhere (Kennedy 1992). Although many sites of this nature have been assigned to the Islamic period, the clear presence of Byzantine pottery on all three sites suggests a classical foundation for this agricultural complex. They should not, however, be termed military sites nor included in any theory regarding the military location of Byzantine forces. If any military site was located in this area during the classical period, it is more likely to be in the environs of Ma’an. Landscapes of defence? The identification of the Ma’an sites as an agricultural estate clearly demonstrates the need to assess archaeological evidence on its own terms before integrating it with historical data. Similarly, the location of the classical route described above, which follows today’s Desert Highway, must be demonstrated on the ground, without reference to postulated connections between known military sites. In this context, much of the problem of the archaeological identification of a frontier zone in southern Jordan is that it is very hard to operationalise in material terms. Namely, what set of physical remains constitutes a frontier zone? As more recent studies have attempted to view frontier areas as zones of crosscutting social networks (Lightfoot & Martinez 1995: 474) it is hard to see many frontiers in a purely boundary maintenance framework. The evidence of nomadic raiding and the location of military sites on the steppe fringe, however, has led to the strong impression that this type of frontier existed during the classical period (Parker 1997). Nomadic threats are seen as ‘external’ although the evidence for tribal dynamics in this area is very unclear (Millar 1993: 428-436). Similarly, it assumed that the Roman military would seek to protect the settled (internal) Mediterranean areas on the plateau edge. The evidence can be used both ways and comparisons with other areas in the

Thus, possibly during the Iron age, the Jibal and Shera’a area, with its Mediterranean environment, good springs and agricultural land, was delimited and monitored. In contrast, the Romans did not re-occupy all the sites along this line. In fact, they only re-used DAS 226, a tower Jebel Dajaniya (Tawil Ifjeij), which acted as look-out for the fort at Dajaniya. As Fiema (1995: 266-267) has noted, the Romans positioned most military sites on nodal communication points. The only major forts that appear in the survey area that do not lie on a major route are Dajaniya and Khirbat Qannas. Dajaniya, as noted above, can no longer be seen as part of a via militaris. However, its relationship with Khirbat Qannas, and their place in a wider landscape setting of settlements and land use may suggest other functions for Roman military sites. The line of forts during the later Roman/Byzantine period along the desert fringe and linked by a road system is not a frontier in administrative or military terms. As Whittaker (1994: 93-94) points out it is an environmentally determined road system. Khirbat Qannas is the only major Roman fortification that appears on the line of the Khatt Shebib (Fig. 1, DAS 193; also Fig 16.9 Kennedy 2000: 163). Discovered by Dr 142

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

Fawzi Zayadine and David Graf (Graf 1997a), Khirbat Qannas has a pottery range from the Iron age to the early Islamic period. It is not connected to the wall since the wall stops a kilometre from the site, which suggests it was robbed for construction of the fort. Kennedy notes this feature in classical sites to the south (Kennedy pending). DAS surveyed the site and associated features, revealing a fort layout with external towers to which courtyard building was later added (Fig. 6). In addition, a small cross-wadi field system was situated immediately to the south of the structure. To the north and west of the site over 40 large bell-shaped cisterns were noted as well as a small cemetery to the east. It is highly probable that the design of the fort dates to the Roman period. The connection of a clear fort site with a courtyard site does suggest different uses for each. This may call into question similar courtyard buildings termed forts.

but believed to be north of Petra. The modern village of Megdes in Shaubak was suggested as the possible site but DAS did not find any features suggesting a large agricultural entity (Graf 1997a: 131 quoting Honigmann 1939: 43-44) Within the DAS area, however, a set of sites and features may be interpreted as belonging to a large diverse managed landscape that could reflect an imperial estate. To the east of Dana, and straddling the via nova Traiana, a series of roads were located which form fields and connect small farms and single structures. Several tracks may have been roughly paved. Kennedy (1998) has noted two elements of this landscape. Firstly, a large circular structure located to the east of the via nova Traiana lies within this field system (Kennedy 1998: 579, 581). Secondly, the fort noted by Kennedy (1998: 579-582) is really only a massive field clearance wall, but it does serve to demonstrate the substantial and widespread clearance and thus investment that occurred in this area. Most of the structures have a pottery range from Nabataean C1stBC/1stAD to the Byzantine period with occasional Iron age and Islamic sherds. The majority of structures occur on nodal route points. They range from single simple rectangular structures and farmsteads to small settlements. It is a massive and intensive use of the landscape. Furthermore it was clearly linked with the via nova, although sections of field tracks probably predate it.

Khirbat Qannas is only 10kms west of the fort Dajaniya that dates from the C2nd to early C6th AD (Parker 1986: 94). The connection between the two and their development is unclear. Both sites lay on the banks of the Wadi Dajaniya, an effective route from the Desert Highway, across the Ifjeij plain on which both sites lie, to the King’s Highway. Thus, one could argue that it fulfils the function of protecting a route. In fact, Khirbat Qannas, with its additional courtyard building, may have been a caravanserai. However, neither is located on nodal communication points (Fig. 1), although they are probably on a route from Petra to the desert areas (Zayadine 1992: 229). Both sites, situated as they are on a wide plain offering easy access to the agricultural areas, would fit the purely military model proposed by Parker. They would have contained relatively substantial garrisons. Moreover, the fort at Dajaniya was large enough to contain a large cavalry force that could have easily dominated the plain (Parker 1986: 94). It can be argued that although this may be seen as a response to incursions, the size of both garrisons (based on the size of site only) are larger than most military sites in this area. There may be other factors to account for the presence of such large military sites, which will be discussed below.

To the west of these fields, situated on the plateau edge, is a massive single structure, Khirbat El-Bir (DAS 1, Fig. 7), roughly the same size as Dajaniya, which may have been associated with this road system. Unfortunately, the growth of the modern village of Al-Qadisiyeh has obliterated the area around this site. The ceramics range from the Nabataean to Byzantine period. However, a Ptolemy II tetradrachma (Finlayson & Baird 1995: 31 Fig. 12) was also retrieved. The site is a massive construction and has been altered through several phases of use. The size and plan of the structure is unusual within the project area. Located to the west of Khirbat El-Bir, on a wooded plateau area overlooking the Wadi Dana, is a probable temple site DAS 85 (Fig. 8) dating to the later Nabataean and Roman periods. A relatively small site (25 x 15m) it was built of large limestone blocks and 4 fragments of pillars were found within the rubble debris. A couple of smaller structures abutted the north and south walls. A 5m wide podium may be interpreted at the western end of the structure. A small tomb (DAS 86) is located just to the south carved into the limestone rock. Other sites have also been noted that would fit in this ritual landscape. Below the modern village of Al-Qadsiyeh, about 5kms south-west of Dana lies the site of Sheikh er-Rish which was examined by Glueck (1939: 38-42). He noted the existence of burial chambers which he termed a “sacred burial place” (1939: 41).

Imperial estates and the military Graf has stressed the connection between imperial estates and the Roman military (Graf 1997a: 129-131). Until more excavations of military sites and investigation of the adjacent agricultural hinterlands are carried out, it is difficult to ascertain the exact relationship of rural practices and forts. However, data from the excavations at Lejjun contributes greatly to this area (Parker 1995: 118). Although there are historical references to estates in southern Jordan the archaeological correlates are not easily discernible at a landscape level (Graf 1997a: 131). A royal estate Salton Hieratikon was mentioned by the Byzantine geographer, George of Cyprus, as being in the metropolis of Petra. Possibly comprising a sacred area, it was associated with the Nabataean royal family. This of course was confiscated upon annexation. The location is unknown

Although it is clear that, without direct physical relationships between the sites, the interpretation of this landscape is difficult, it is possible to infer that this set of 143

Limes XVIII

features is unusual enough to suggest a different pattern of landscape use. The inter-relationship of sites and tracks with fields through which the via nova runs is a clear sign of a well-managed landscape, which does not occur elsewhere in the area. The connection of this landscape with the large site of Khirbat El-Bir is unsure but the uniformity of ceramic evidence from the sites suggests a broad contemporary pattern. The relationship with DAS 85 is, of course, impossible to ascertain although it too existed during the Nabataean to Byzantine periods. However, it is argued here that the association of these sites and features, which suggests a large, managed area of inter-linked features, coupled with the information from the text, is highly persuasive. All these features suggest that we may have the setting of the imperial estate, referred to in the sources. The proximity of these sites to the metal resources in the Faynan area should also be noted.

agricultural/irrigation projects were carried out, supposedly in a climate of shrinking imperial control. Thus, economic variability cannot be linked directly to the presence of imperial units. It is clear that the growing body of data suggests dynamic and diverse cultural systems within the Roman/Byzantine period. Older models of colonial control, stressing dominant coloniser power relations, must be modified to explore the varying scales of Roman control and exploitation. Conclusion This project has clearly demonstrated that the archaeological data set cannot continue to provide evidence for the overt military nature of the Roman legions in southern Jordan. More than that, it shows that the use of more dynamic interpretative frameworks allows a broader view of the legions and their monuments in the landscape. The move away from the simple interpretation of the frontier system as a reflection of purely military needs would engender more fruitful research avenues in the study of the Roman military in this province. Such research, reflecting the socio-economic, technological and ideological dimension of military presence, could not only document their impact but also their contribution to the province of Arabia and the development of material culture in Jordan.

Indeed, within this context, one may better explain the setting of sites like Dajaniya and Qannas (excluded as they are now from their position on major routes). If the military policed or controlled the desert system road, one would expect a site like Dajaniya to be placed on a nodal communication point. Its situation, just off the two main Roman roads, however, does not suggest this. Alternatively, one could say that Dajaniya played a more military rôle, defending or controlling an area. Defending what? It is situated on a plain and clearly commands the area, but for what reason? The Ifjeij plain is a clear open approach into the richer resource areas of the plateau fringe. If the Khirbat el Bir sites are seen as part of an Imperial estate, the location of several forts (such as Khirbat Qannas or Dajaniya), contemporary or not, within the area tends to suggest a link with defence and estates as Graf (1997a: 131) has suggested for other areas in Jordan and Syria. Although this paper has been critical of the defensive military system as envisaged in a limes system, it does not deny the placing of sites like Dajaniya as reflecting defensive needs. However, it is argued that their location has more to do with the position of resources rather than purely military needs. The threat may be similar but the purpose of defence quite different. Fiema (1991; 1995) is right to argue for the connection of military sites and economic resources, but while he stresses that military viability fluctuated with the economics of long distance trade, it is argued here that it could reflect direct imperial economic exploitation of the area. The later Byzantine decline in military sites in the research area (Parker 1986: 152) could be as plausibly attributed to the decline of imperial interests in say, the copper extraction in the Faynan area than the fluctuations of long distance trade. The excavations carried out by Walmsley in Gharandal (Walmsley & Grey forthc.; Walmsley, Karsgaard & Grey 1999) to the north of the research area shows clear rural continuity and settled occupation through to the early Islamic period. Clearly, rural settlement and its stability were not linked to military security. Furthermore, the later date of the Ma’an sites discussed above and others such as Jebel Tahuna (Kennedy 2000: 173, fig. 17.8; Killick 1986: 438-440), demonstrate that massive

Acknowledgements The author is grateful to the University of Edinburgh’s Abercromby Fund for financial support towards travel to the 2000 Limes Congress. Bibliography Adams R. 1998 The development of copper metallurgy during the early Bronze age of the southern Levant: Evidence from the Feinan region, southern Jordan. Unpublished PhD thesis, Sheffield University in CBRL Amman library. Banning E.B. 1986 Peasants, pastoralists and pax Romana: Mutualism in the southern highlands of Jordan. Bulletin of the American Schools of Oriental Research 261: 2550. Bartel B. 1980/81 Colonialism and cultural responses: problems related to Roman provincial analysis. World Archaeology 12: 11-26. Bennett C-M. 1966 Fouilles d’Umm el-Biyara: Rapport Preliminaire. Revue Biblique 73: 372-403. Bienkowski P. & Adams R. 1998 Soundings at AshShorabat and Khirbat Dubab in the Wadi Hasa, Jordan: the pottery. Levant XXXI: 149-172. Bignasca A. et al. 1996 Petra. Ez-Zantur 1. Ergebnisse der Schweizerish-Liechtensteinischen Ausgrabungen 19881992. (Mainz). Brown R. 1991 Ceramics from the Kerak Plateau. In J.M. Miller (ed.) Archaeological survey of the Kerak Plateau. (Atlanta): 169-280. Brünnow R. & v. Domaszewski A. 1904-1905 Die Provincia Arabia (two vols). (Strassburg). Fiema Z.T. 1991 Economics, administration and demography

144

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

of late Roman and Byzantine southern TransJordan. Unpublished PhD, University of Utah in ACOR library. Fiema Z.T. 1993 Tuwaneh and the via nova Traiana in southern Jordan: A short note on the 1992 season. Annual of the Dept. of Antiquities of Jordan 37: 549550. Fiema Z.T. 1995 Military architecture and the defense ‘system’ of Roman-Byzantine Jordan – a critical appraisal of current interpretations. In K. Amr, F. Zayadine & M. Zaghoul (edd.) Studies in the History and Archaeology of Jordan V. (Amman): 261-269. Finlayson W. & Baird D. 1995 Dana Archaeological Survey, Phase 1 Report. Report No. 204. Centre for Field Archaeology, University of Edinburgh unpublished report on file at the Royal Society for the Conservation of Nature Amman. Frank F. 1934 Aus der ‘Arabah I: Reiseberichte. Zeitschrift der Deutschen-Palästina Vereins 57: 191-280. Freeman P.W. 1990 Recent work on a Roman fort in south Jordan. In H. Vetters & M. Kandler (eds.) Akten des 14. Internationalen Limeskongresses 1986 in Carnuntum. (Vienna): 179-191. Glueck N. 1934 Explorations in Eastern Palestine, I. Annual of the American Schools of Oriental Research 14. Glueck N. 1935 Explorations in Eastern Palestine, II. Annual of the American Schools of Oriental Research 15. Glueck N. 1939 Explorations in Eastern Palestine, III. Annual of the American Schools of Oriental Research 18-19. Graf D. 1978 The Saracens and the defense of the Arabian frontier. Bulletin of the American Schools of Oriental Research 229: 1-26. Graf D. 1995 The via nova Traiana in Arabia Petraea. In J. H. Humphrey (ed.) The Roman and Byzantine Near East: some recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 141167. Graf D. 1997a The via militaris and the Limes Arabicus. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 123-133. Graf D. 1997b The via militaris in Arabia. Dumbarton Oaks Papers 51: 271-281. Gregory S. 1995-1997 Roman military architecture on the eastern frontier. (Amsterdam, 3 vols). Gregory S. 1996 Was there an eastern origin for the design of late Roman fortifications? Some problems for research on forts of Rome’s eastern frontier. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 169-210. Gregory S. & Kennedy D. 1985 Sir Aurel Stein’s Limes Report. The full text of M. A. Stein’s unpublished Limes Report (his aerial and ground reconnaissances in Iraq and Transjordan in 1938-39. British Archaeological Reports Int. Series 272. (Oxford). Hall M. 1993 The archaeology of colonial settlement in southern Africa. Annual Review of Anthropology 22: 177-200. Harding G.L. 1967 The antiquities of Jordan. (London). Harfield C.G. 1988 Control of resources in the medieval period. In J.Gledhill, B. Bender & M.T. Larsen (edd.) State and society: The emergence and development of social hierarchy and political centralization. One World Archaeology 4: 137-148.

Hart S. 1986a Nabataeans and Romans in southern Jordan. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297. (Oxford): 337-342 Hart S. 1986b Some preliminary thoughts on settlement in southern Edom. Levant XVIII: 51-58. Hart S. 1987 Five soundings in southern Jordan. Levant XIX: 33-47. Hart S. & Falkner K. 1985 Preliminary report on a survey in Edom, 1984. Annual of the Dept. of Antiquities of Jordan 29: 255-273 Hauptman A. & Weisgerber G. 1992 Periods of ore exploitation and metal production in the area of Feinan, Wadi ‘Arabah, Jordan. In S. Tell (ed.) Studies in the History and Archaeology of Jordan IV. (Amman): 6166. Hayes J.W. 1972 Late Roman pottery. (London). Hayes J.W. 1997 Handbook of Mediterranean Roman pottery. (London). Honigmann E. 1939 Les Synekdemos d’Hierokles et Georges de Chypre. Corpus Bruxellense Historiae Byzantinae, Forma Imperii Byzantini I. (Bruxelles). Isaac B. 1992 The limits of empire: The Roman army in the east. (rev. ed. Oxford). Isaac B. 1998 The Near East under Roman rule. Myemosyne Supplementum 117 (Leiden). Kennedy D. 1998 Gharandal survey 1997: Air photo interpretation and ground verification. Annual of the Dept. of Antiquities of Jordan 42: 573-585. Kennedy D. 2000 The Roman army in Jordan. (London). Kennedy D. pending Nabataean archaeology from the air. In H. Qatami (ed.) Nabatean Research and Studies. (Amman). Kennedy D. & Riley D. 1990 Rome’s Desert frontier from the air. (London). Kennedy H. 1992 The impact of Muslim rule on the pattern of rural settlement in Syria. In P. Canivet & J.-P. ReyCoquais (edd.) La Syrie de Byzance a L’Islam VIIeVIIIe siecles. Actes du Colloque International, Maison de l’Orient Mediterranean, Paris, Institute du Monde Arabe, 11-15 September1990. (Damascus): 291-297. Killick A. 1983a Udhruh – The frontier of an empire: 1980 and 1981 seasons, a preliminary report. Levant XV: 110131. Killick A. 1983b Udruh – 1980, 1981, 1982 seasons, a preliminary report. Annual of the Dept. of Antiquities of Jordan 27: 231-243. Killick A. 1986 Udhruh and the southern frontier. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297. (Oxford): 431-446. Killick A. 1987 Udhruh and the trade route through southern Jordan. In A. Hadidi (ed.) Studies in the History and Archaeology of Jordan III. (Amman): 173-179. King G. 1985 A survey of the southern Ghawr, the Wadi ‘Araba and western Trans-Jordan, 1981-2. Proceedings of the Society for Arabian Studies 15: 41-47. King G., Lenzen C.J., Newhall A., King J.L. & Deemer J.D. 1987 Survey of Byzantine and Islamic sites in Jordan: Third season preliminary report (1982) The southern Ghor. Annual of the Dept. of Antiquities of Jordan 31: 439-459. King G., Lenzen C.J., Newhall A., King J.L., Deemer J.D. & Rollefson G.O. 1989 Survey of Byzantine and Islamic sites in Jordan: Third preliminary report (1982)

145

Limes XVIII

The Wadi ‘Arabah (Part 2). Annual of the Dept. of Antiquities of Jordan 33: 199 –215. Kirkbride A. 1947-8 Shebib’s wall in Transjordan. Antiquity 21-22: 151-154. Knapp B. 1992 Archaeology, annales and ethnohistory. (Cambridge). Knauf E.A. & Lenzen C.J. 1987 Edomite copper industry. In A. Hadidi (ed.) Studies in the History and Archaeology of Jordan III. (Amman): 83-88. Lightfoot K.G. & Martinez A. 1995 Frontiers and boundaries in archaeological perspective. Annual Review of Anthropology 24: 471-492. Luttwak E. N. 1979 The grand strategy of the Roman empire: From the first century AD to the third. (Baltimore). MacDonald B. 1988 The Wadi el-Hasa archaeological survey 1979-1983, west-central Jordan. (Waterloo, Ont.). MacDonald B. 1992 The southern Ghors and northeast ‘Arabah archaeological survey. (Sheffield). MacDonald, B. (1999) Tafila-Busayra Archaeological Survey (TBAS), Jordan – Home Page. viewed http://stfx.ca/people/bmacdona/TBASWEB/, 25/01/01. Mayerson P. 1989 Saracens and Romans: micro-macro relationships. Bulletin of American Schools of Oriental Research 274: 71-79. Millar F. 1993 The Roman Near East 31B.C. – A.D. 337. (Cambridge, Mass.). Musil A. 1907 Arabia Petraea II. (Vienna). Parker S.T. 1986 Romans and Saracens: a history of the Arabian frontier. (Winona Lake). Parker S.T. 1987a The Roman frontier in central Jordan: Interim report on the Limes Arabicus Project, 19801985. British Archaeological Reports Int. Ser. 340: (Oxford). Parker S.T. 1987b Peasants, pastoralists and the pax Romana. Bulletin of the American Schools of Oriental Research 265: 35-51. Parker S.T. 1995 The typology of Roman and Byzantine forts and fortresses in Jordan. In K. Amr, F. Zayadine & M. Zaghoul (edd.) Studies in the History and Archaeology of Jordan V. (Amman): 251-260. Parker S.T. 1997 Geography and strategy on the south-eastern frontier in the late Roman period. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 115-112 . Parker S.T. 1998 The Roman Aqaba project: the 1998 campaign. Annual of the Dept. of Antiquities of Jordan 44: 375-394. Peterson A.D. 1986 Early Ottoman forts on the Hajj route in Jordan. Unpublished M.Phil. Oxford, held in CBRL Amman library. Peterson A.D. 1989 Early Ottoman forts on the Darb al-Hajj. Levant XXI: 97-117. Rosen S.A. 1987 Byzantine nomadism in the Negev: results from the emergency survey. Journal of Field Archaeology 14: 28-41. Rothenburg B. 1972 Timna: valley of the Biblical copper mines. (London). Sauer J.A. 1973 Heshbon pottery 1971. (Berrien Springs: MI). Schick R. 1994 The settlement pattern of southern Jordan: the nature of the evidence. In G.R.D. King and A. Cameron (edd.) The Byzantine and early Islamic Near East II. Land use and settlement patterns. (Princeton): 133-154.

Sinopoli C.M. 1994 The archaeology of empires. Annual Review of Anthropology 23: 159-180. Stucky R.A. et al 1994 Swiss-Liechtenstein excavations at Es-Zantur in Petra 1993. The fifth campaign. Annual of the Dept. of Antiquities of Jordan 38: 271-292. Thomsen P. 1917 Die römischen Meilensteine der Provinzen Syria, Arabia und Palastina. Zeitschrift der DeutschenPalästina Vereins 40: 1-120. Walmsley A. 1998 Gharandal in Jibal: first season report. Annual of the Dept. of Antiquities of Jordan 42: 433441. Walmsley A., Karsgaard P. & Grey T. 1999 Town and village: site transformation in south Jordan. Annual of the Dept. of Antiquities of Jordan 43: 459-478. Walmsley A. & Grey T. forthc. An interim report on the pottery from Gharandal (Arindela), Jordan. Levant. Whittaker C.R. 1994 Frontiers of the Roman empire: a social and economic empire. (Baltimore). Wright K., Najjar M., Last J., Moloney N., Flender M., Gower J., Jackson N., Kennedy A. & Shafiq R. 1998 The Wadi Faynan fourth and third millennia project, 1997: report on the first season of test excavations at Wadi Faynan 100. Levant XXX: 33-60. Zayadine F. 1992 L’espace urbain du grand Pétra, les routes et les stationes caravaniès. Annual of the Dept. of Antiquities of Jordan 36: 217-239.

146

Fig. 1. Map of DAS project area.

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

147

Limes XVIII

Fig. 2. Plan of Qal’at Unaiza.

Fig. 3. Stein’s map of Ma’an and environs (From Gregory & Kennedy 1985, 297 fig. 28).

148

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

Fig. 4. Plan of agricultural system in Ma’an.

Fig. 5. Kirkbride’s map of Khatt Shebib (From Kirkbride 1947/48, 152).

149

Limes XVIII

Fig. 6. Plan of Khirbat Qannas.

150

George Findlater: Limes Arabicus, via militaris and resource control in southern Jordan

Fig. 7. Plan of Khirbat El-Bir.

151

Limes XVIII

Fig. 8. Plan of temple DAS 85 & 86 (From Finlayson & Baird 1995, 14 fig. 4).

152

Nomads and the Arabian frontier: the epigraphic perspective David F. Graf Discussion of Rome's Arabian frontier has emphasized that the nomads were exterior to the limes or defensive system. The primary evidence of the presence of the nomads has been the pre-Islamic North Arabian inscriptions called 'Safaitic' and 'Thamudic' which have been found mainly in the desert regions of the harra in north-east Jordan and the Hisma of southern Jordan respectively. However recent finds have been accumulating in the settled area of Transjordan near such urban centres as Jerash, Amman, Madaba, Kerak and Petra. In fact, the longest and most revealing of these texts have been found in the Madaba region. These texts are written in a language that closely approaches classical Arabic. In sum, the assumption that these texts are the product of only nomads is now to be rejected, as the texts clearly emanate from both the sedentary and pastoralist population of Arabia. From the content of these graffiti, they also appear to date mainly from the C1st and C2nd AD and show intimate acquaintance with the Roman cultural sphere of that period.

Nomadic incursions frequently are assumed to be the primary exterior threat to the defensive system or so-called limes of Rome’s Arabian frontier. As such, the nomads have been characterized as a constant low-intensity threat to the Roman frontier (Luttwak 1976: 160, cf.112; Parker 1986: 8; Millar 1993: 435; Whittaker 1994: 135-139; Ball 2000: 241). The basic evidence often cited for the presence of these nomads is the pre-Islamic North Arabian inscriptions called ‘Safaitic’ and ‘Thamudic’. The Safaitic texts are found mainly in the desert regions of the |arra or ‘black basalt desert’ in north-east Jordan and southern Syria (Macdonald 1994). The Thamudic texts are primarily found in the Ḥismā, the desert region of southern Jordan and north-west Saudi Arabia (Macdonald & King 1999). Both names are misnomers. None of the texts appear in the øafa region proper that is south-east of Damascus and few, if any, of the texts can be definitely associated with the legendary tribe of Thamud (Macdonald 2000: 33-35). Nevertheless, both conventional labels are still generally employed for the sake of convenience, in full awareness that further refinement is needed. The Safaitic texts number now more than 20,000 (Macdonald 1994: 761). About two-thirds of these are from the black basalt area of north-eastern Jordan, with the remaining third in the adjacent regions of southern Syria, Iraq, and north-west Saudi Arabia, with a few stray finds outside of these areas. Many more are yet to be recorded. The Thamudic texts are more complex, and have been arranged into various categories or types: Taymanite (‘A’), Najdi (‘B’), Hijazi (‘C’ and ‘D’), and Tabuki (‘E’). The latter is the most relevant for the Roman period, and the Arabian frontier, as they are primarily found within Jordan. About 8,000 ‘Thamudic E’ texts have been recorded. Both the ‘Thamudic E’ texts and Safaitic texts are generally regarded as dating between the C1st BC and C4th AD (Macdonald 1994: 761; Macdonald & King 1999).

source material. Its potential for illuminating the situation on the Roman frontier from the perspective of the indigenous population has virtually been ignored (but cf. Bowersock 1983; Sartre 2001). Parker is quite right to insist that a monograph is still needed that “synthesizes the huge corpus of Safaitic and Thamudic texts” (1987: 152). If such a repository existed for the British tribes or German tribes on the northern frontier, one only can imagine the interest it would create. But the reason why these preIslamic north Arabic texts have not been addressed adequately is quite simple. These texts are written in a language that has been designated as “Ancient North Arabian” in what is considered a version of the South Semitic script that consists of 28 letters that are comparable to those of the Arabic alphabet, in contrast to the 22 letters of such north-west Semitic scripts as Aramaic or Hebrew (Müller 1982: 17-25; Macdonald 2000: 29-37; cf. Beeston 1981: 181-184). This is a complex and challenging corpus for most Semitic scholars, yet filled with abundant references to the peoples and places of the Roman Near East. Nevertheless, few Roman historians are Semiticists, and few Semiticists have the necessary grasp of the Roman world to illuminate and elucidate the appropriate meaning of the texts. In this context, I merely wish to draw attention to this important archive of material. In addition, I hope to demonstrate that this corpus is just as important for the Roman provincial world of the Near East, as the ‘external’ world of the nomads. This becomes apparent by just the distribution of the texts. Let us deal with each separately. Safaitic The Safaitic inscriptions have often been viewed as the product of ‘external nomads’, insulated and isolated from the Roman world in the Syro-Arabian desert (Macdonald 1992: 303). Enno Littmann’s dictum that “the fewer the traces of real civilization are, the more numerous are Safaitic inscriptions” (1904: 104) has cast a determining shadow over the interpretation of the corpus. Recent treatment of this subject represents a simple rehearsal and reassertment of Littmann’s old maxim (Macdonald 1993: 311-322). But it should be emphasized, when Littmann made his comment, only several thousand of the Safaitic texts were known. Today, there are approximately 20,000 that have been published. It would be strikingly ironic that an observation made when only 10% of the evidence was

As a result, the Safaitic and ‘Thamudic E’ texts presumably are contemporaneous with the early Roman imperial era. Yet, as important as this material may seem for Roman frontier studies, it generally has been neglected by most Roman historians of the east. Tom Parker’s Romans and Saracens (1986), David Kennedy and Derrick Riley’s Rome’s desert frontier from the air (1990), Ben Isaac’s The limits of empire: The Roman army in the east (1992), Fergus Millar’s The Roman east (1993), and Warwick Ball’s Rome in the east (2000), only allude to this 153

Limes XVIII

available should be axiomatic for the corpus when it had expanded ten-fold. In reality, the size of the corpus has not only increased, but the geographic range of the texts has also expanded considerably, including finds in the contexts of major settlements, villages and districts within the sedentary zone. In order to minimize this development, Macdonald has asserted that the “Safaitic inscriptions are found almost entirely in the desert and only rarely in or around settled areas”, and then relegates to a footnote the ‘few’ texts he knows from settled areas (1993: 311 n.50). In reality, the list is larger than he suggests and the depiction of a barrier between the sedentary and nomadic spheres not as distinct as he assumes.

Roman fort at Azraq itself (CIS V 5200-5208). Further south, Safaitic texts have been found at the village of Sahab just east of Amman (Harding 1951: no.1), the Roman-Byzantine fort at Khan el-Zebib (1951: no.2) and just east of Qatraneh (1951: no.8). The date of the texts has always been troublesome, due to the lack of precise chronological clues, but they have been accumulating recently. First, a number of the recently published texts from the Basalt Desert Rescue Survey (King 1990b) have dating formula involving the Herodian dynasts that clearly must be placed in the late C1st BC or C1st AD (cf. al-Khraysheh 1997). One is dated to the “year Herod was engaged in war” (Macdonald 1995: 286, snt qttl hrd¡), referring perhaps to the war between Herod the Great and the Trachonitis rebels during the reign of Augustus (Josephus Ant. Jud. 16.4.6 [130]; 9.1 [273-276]) or that between Herod Antipas and Aretas IV in AD36 (Ant. Jud. 18.5.1 [109-115]), who is also referred to as just ‘Herod’ in the New Testament (Matt. 14.3-12; Mark 6.1729; Luke 13.31-33). Macdonald is correct to take the word qttl as the VIII Form of qatala, but his preference for the passive meaning (“to die mad,” Lisān XI 550a; cf. Hava 587), and translation of the phrase as “the year Herod died mad” is not compelling. This seems like an attempt to bring the phrase in conformity with Whiston’s less than accurate translation of Ant. Jud. 17.6.6 [180] that Herod’s illness “brought him to do all things like a madman”. But the primary and natural sense of the VIII Form is “to fight with one another, combat each other” (cf. Lane 2984c; Kazimirski II 674; cf. Wehr 743), and this seems more likely than the passive meanings of “to die from love, madness, or an evil spirit”. The conflict alluded to is probably connected with the Herodian intrusions into the ›aurān under Herod or his successors. Another text is dated to “the year the people of the ›aurān complained to Caesar about Philippus” (KRS 2618 = Macdonald 1993: 341; 1995: 288, snt qbl ’[l] Hrn q¡ral flf [¡ ]), perhaps some unknown event associated with the rule of Philip the Tetrarch, or the offenses of Philippus, son of Iachimus, son of Zamaris, the general of Agrippa II (Josephus Vita 74 [407-410]). Finally, the author of another text indicates “he rebelled against the king Agrippa” (KR 3, wmrd al hmlk grf¡ = King 1990b: 62), which must be either Agrippa I or Agrippa II, between AD37 and 100 (Kokkinos 1998: 396399). These references to Herod the Great and/or the later Herodian dynasts must be firmly dated to the end of the C1st BC and no later than the end of the C1st AD.

Finds of Safaitic texts now range from the Phoenician coast between Tyre and Sidon (Ryckmans 1947-8: 20) to Dura-Europos on the Euphrates (CIS V 5175-5180), and at primary settlements and places in between. For example, Safaitic texts have been found in the Biqāa Valley just 35kms north-east of Baalbek (Ghadban 1971; Harding 1971b; 1975), at the excavations at the Temple of Allāt at Palmyra (Drijvers 1976: 34, but still not published), and on the famous Triad relief now in the Louvre, where a Palmyrene Aramaic-Safaitic text is engraved, suggesting the presence at the oasis of a population that utilized the script (J. Starcky in Seyrig 1949: 35-41). This is confirmed by other Safaitic-Palmyrene Aramaic texts left by Zubaida son of H.̣awmal in the Wadi Hauran, 65kms west of Hit, near the Euphrates in Iraq, that are dated to AD98 (Safar 1964). This individual is probably to be identified with the Zebeidos Aumolos mentioned in Greek texts at Palmyra, in dedications by his grandsons who were serving in the Ala Thracum Herculana in Syria (Inv X, 1-2 and 117, as discussed by Milik 1972: 257-261). This unit appears in Syria during Vespasian’s reign, and is first attested at Palmyra in AD157, and thereafter until it was transferred to Egypt in AD183 (Spaul 1994: 132-143). Ten other Safaitic texts were found in the villages north-west of Palmyra by Schlumberger, who emphasizes they were found in a ‘sedentary’ context (1951: 133-133). Other Safaitic texts have been found in Syria in the Lejja (CIS V 1 = R 1198), at Bostra, the capital of the province of Arabia (Littmann 1904: 104; Sartre 1985: 148), and nearby Suweida in the northern ›aurān (Littmann 1904: 104, which Macdonald, 1993: 311 n. 50, incorrectly assigns to Sīa), and the villages of Rushayda (Sartre 1992: 41), Diyatheh (CIS V, 5157-58), and Mithayet (CIS V 51595160) on the eastern edge of Jebel Druze. In north Jordan, Safaitic texts have been found at the Roman-Byzantine settlements in the southern ›aurān just west of the via nova Traiana at Raoÿat al-Roeay (Attalah 1995: 398), Khan (Attalah 1997: 11) and Zamat al-Amir Gazi nearby (Atallah forthc.), and at ›ayyan al-Ġarbī, just a few kilometres north of Khirbet al-Samra (Knauf 1998: 535536). Just east of the Trajanic road, they appear at the Roman forts at Umm al-Jimal (LP 1269-1279 = CIS V 5162-5174), and Deir al-Kahf (CSP 1; Macdonald 1982: 172). Others have been found further south between Deir al-Kahf and Azraq (CIS V 5161), and in the environs of the

Secondly, a group of Safaitic texts were recently discovered scratched on the northern wall of the corridor behind the small theatre at Pompeii in Italy (Calzini Gysens 1990). They appear on the same plastered wall that is covered with Latin graffiti cut into the red painted sections. These texts represent the first discovery of Safaitic texts outside of the Levant, and their context indicates a date sometime before the city was destroyed in AD79. These recent discoveries suggest that the texts flourished in the C1st AD. Finally, historical allusions in 154

David F. Graf: Nomads and the Arabian frontier: the epigraphic perspective

the texts suggest a similar date in the early Roman imperial era, or possibly just after the annexation of Arabia in 106. For example, one Safaitic text possibly mentions “the year the troops of Germanicus were at Nq t” (LP 653 as reread by Macdonald 1993: 333, snt lgy[n] grmnq¡ bnq t). Another is dated to “the year Caesar’s son died; and he heard that Philippus had been killed,” but the historical context is enigmatic (Macdonald 1995: 286-287). Another text speaks of “the year the Nabataeans revolted against the people of Rome” (WH 2815, snt mrdt nb¥ l l rm). Additional wars conducted by the Nabataeans are mentioned against the Jews (Abbadi 1996), Palmyrenes (al-Khraysheh forthc.), and the Safaitic tribe of the Hawilat (Graf & Roussan forthc.). Other conflicts in North Arabia include a war of the “the tribe of Jašm against the people of the Thamud” (WH 3792a and 3792c), conflicts with a Li|yan, the kingdom centered at al- Ulā in the ›ijaz (KR 2), including an attack by Li|yan against a group in Syria (Macdonald, Al-Mu azzin & Nehmé 1996: 458-463), and the struggles of a number of Safaitic tribes against the people of ¦ayyi (CIS V 2795, 5089, jaS 10; CSNS 1004, 1011, 1046), perhaps the famous Arab tribe that played such a pivotal role later for the Lakhmids in Iraq, as the Persian clients that waged war against the Byzantine client kingdom of the Ghassanids. Such texts reveal the authors were not in some secluded desert zone that isolated them from the larger cultural sphere. The desert obviously penetrated by people and parties from the sedentary zone.

security” (l s bn m bn s d l bšt w mrd l l rm fh b lsmn lh s slm, CNNS 424). The tribe of Ubaishat is known from Nabataean and Greek texts at Sī to have been instrumental in erecting the sanctuary (Graf 1989: 360361), demonstrating that the author is inherently a member of the same cultural tradition. There are other texts that refer to various well-known populations in the region, including the ‘Jews’ (CIS V 1270 bis, 2732, SIJ 688 with Macdonald 1995: 285-286), the ‘Ituraeans’ (Y©r, with Graf 1989: 373-374), and the Romans (Macdonald 1993: 328333). In essence many of the important places and peoples of the Near East who surrounded the |arra to the west are mentioned in Safaitic texts. If these texts are the product of mere ‘external nomads’, they can hardly be considered as isolated and insulated from the major players and centers of the Near East. Thamudic The ‘Thamudic E’ texts present problems of similar complexity. The earliest finds of these texts were concentrated west of Tabuk, with scattered finds extending from Māda in Ṣāli| to Wadi Ramm. As a result, they were renamed ‘Tabuki’ (Winnett & Reed 1970: 70), although a sizeable corpus of such texts had been published earlier from Wadi Ramm (TIJ). In subsequent decades, thousands more of the ‘Thamudic E’ texts were discovered in the Jordanian part of the ›ismā by W. Jobling, Geraldine King, Saba Fares-Drappeau and Fawzi Zayadine and myself, greatly expanding the number of previous finds now to almost comparable numbers with the Safaitic texts. As a result, the Ḥismā texts now greatly outnumber all the other ‘Thamudic E’ texts published from other regions. This fact has led to their being designated ‘Hismaic’ (Macdonald & King 1999: 437; Macdonald 2000: 44-45). Another proposal attempts to assign the ‘Thamudic E’ texts to a sub-group of Safaitic and therefore classify them as ‘South Safaitic’ (Knauf 1983). But whatever similarities they may enjoy with Safaitic, it is now clear that the ‘Thamudic E’ texts have distinctive paleographical and orthographical features which distinguish them from Safaitic (Winnett 1985; King 1990a; Macdonald 2000: 44). Such geographical labels distort the extent of their distribution which is far beyond the regional categories proposed. The few stray finds of ‘Thamudic E’ texts in northern Transjordan that originally prompted their being labelled ‘South Safaitic’ have now vastly increased. Substantial finds of ‘Thamudic E’ texts have emerged on the Transjordanian plateau north of the Ḥismā, with discoveries extending from the region of Petra in the south (Nehmé 1997: 126) to as far north as Mafraq near the Syrian border (Clark 1976-77: 39-40, no. 2). These extensive Transjordanian finds call for the rejection of regional geographical designations like the former label of ‘Tabuki’ and the more recent ones of ‘South Safaitic’ and ‘Hismaic’. ‘Thamudic E’ remains the best neutral category for designating the script.

A number of texts also show familiarity with the sedentary zone, even if their provenance is in the ḥarra desert to the east. The city of Edessa ( Urhay) in North Syria (rendered as rhy in CIS V 742, 2670, WH 3605, 3736a), and the oasis town of Palmyra in the Syrian desert (tmr, CIS V 663, 1649, 1664-65, LP 717, WH 2833a, KR 1, and perhaps CIS V 2473, as interpreted by Macdonald 1993: 333) appear in texts. At least 10 texts mention the ›aurān (see Macdonald 1993: 339-342, but with caution on the readings), as well as villages in the region. Several texts mention the village of Dhakir on the eastern edge of the Lejjā (CIS V 2233, 2549), Qanawat (CIS V, 3095, as re-read by Macdonald 1993: 333 n. 195), and perhaps S.̣alk²ad (King 1990b: 67 n. 18) in the ›aurān. Several other texts specifically mention Bostra, the capital of Arabia. One is dated to “the year the Persians (Mdy) came to Bo¡rā” (SIJ 78), and another to the “the year the Persians (Mdy) fought the people of Rome at Bo¡rā” (CIS V 4448). Other texts which refer to “the city” (hmdnt) must also refer to Bosrā. One is dated to “the year that the Roman army delivered the city” (snt njy q¡r hmdnt, WH 1698, 1725b), and another to “the year that Caesar sent aid to the city and placed the village in good order” (snt md qyṣr hmdnt w swy hmdnt, Macdonald, AlMu azzin & Nehmé 1996: 453-458). Both may allude to the same event. Perhaps the most dramatic reference in the texts to the ›aurān is that found south of H4, that mentions the sanctuary of Baalshamïn at Sī , near Qanawat by a member of the tribe BŠT: “and he rebelled against the people of Rome, and so, O B LSMN, god of Sī [grant]

In fact, ‘Thamudic E’ seems to dominate the total

155

Limes XVIII

an even lengthier and perhaps even more eloquent style. The elegant language, effusive, rhetorically versatile, and sometimes redundant, is also filled with lexical and grammatical features best known in classical Arabic (Graf & Zwettler forthc.). These texts written in a language conspicuously more elegant than other known Thamudic text suggest a more sophisticated culture utilized the script than just nomads. Since these discoveries, more than a dozen more ‘Thamudic E’ have been discovered in this same region of northern Moab by the Wadi ath-Thamad Project, including the lengthiest text now known, a monumental text pounded into bedrock near Khirbat alMudayna (Daviau, Mulder-Hymans & Foley 2000: 279280). Clearly, the ‘Thamudic’ texts are more deeply seated in this region than formerly anticipated. Such finds from within the settled region at or in proximity to villages and settlements in central Transjordan demonstrate the currency of the script exceeded far beyond the parameters of the H.ismā desert and that some hesitancy should exist before ascribing the Thamudic script to just “external nomads.” Both the quantity and quality of the new finds suggest they are the product of well-established traditions in the sedentary realm.

landscape of Transjordan, not just the ›ismā desert in the south. They appear well within the sedentary zone in the environs of such urban centres as Madaba (Clark 1980: no. 2-3, which he designates as ‘Safaitic-Thamudic’ but is best considered ‘Thamudic E’), Dibon (Clark 1980: no.1, which again is best viewed as ‘Thamudic E’), Ader (TIJ 494) and Smākiyyah near Kerak (Milik 1958: 342-358; and Graf in Miller 1991: 46, site 44, 6kms north-east of Smākiyyah). In addition, they appear at such fort sites as Khirbet alSamra on the via nova Traiana (Knauf 1998), Ziza (Röllig 1987: 43-45, and pl. LIII), Khan Zebib (TIJ 522), Umm er-Rasas (TIJ 495-498, and Clark 1976-77: 40, no.3), and Qasr Bshir (Clark 1987: 752, but illegible). In fact, during the excavations at the legionary camp at Lejjun, a survey discovered them within 10-15kms radius of the site, to the north, east, south and west of the fort (Clark 1987: sites 62b, 62d, 117, 140, 151, 159). As V. Clark observed, these texts came from “within the territory of the Nabataean kingdom” or “provincia Arabia” (1987: 724). On the basis of such discoveries in the sedentary zone, Dr. Fawwaz alKhraysheh, George Mendenhall, and myself began surveying in 1996-1998 the region between Amman to Qatrana to the south, and Amman to Azraq to the east. This survey yielded several hundred of the ‘Thamudic E’ texts, with only a few stray finds of Safaitic intermingled among them. It should be noted that most of these texts were found near Kharaneh, Qasr Amra, and Qasr Uwaynid, between Muwaqqar and Azraq, just 25kms east of Amman. Such finds still might be interpreted as the penetration of the sedentary zone by migrating nomads, but there is now every reason to believe that the population that produced these texts was largely resident within the Roman provincial sphere.

Conclusion The corpus of Safaitic and Thamudic inscriptions is then an important archive for both the sedentary and pastoralist populations that inhabited provincia Arabia, not just ‘external nomads’. Admittedly, the interpretation of the material is delicate and precarious, and my own ventures into this arena have drawn criticism (Macdonald 1993: 334-346). Partially, this has been the result of misunderstanding. I have tried both earlier (1989) and now only to emphasize those texts which seem to defy characterization as the product of external nomads. Even earlier, I indicated that “I am aware that at most they [my interpretations] apply to only a substantial part of the tribal world represented by the texts, not all of it” [emphasis mine] (1989: 367). Such an extensive corpus of north Arabian texts could hardly be the product of a ‘homogeneous community’ or a single ‘cultural entity’. The substantial overlap in provenance and content of some texts with the sedentary world now clearly demonstrates the authors were an inherent part of the provincial sphere, not divorced from it.

At the same time these discoveries were being made, even more exciting discoveries were being made further to the west, between Amman and Madaba, well within the sedentary zone. At the town of Madaba, just 25kms southwest of Amman, during excavations conducted in 1996 by Pierre Bikai, a stone engraved with 11 lines of ‘Thamudic E’ script was discovered near the opening of an ancient water cistern. But what was distinctively different about this text was not just its provenance and length; both its content and language were unique. It was written in an Arabic dialect that exhibits a number of striking features that have their best parallels in the language of Old or even Classical Arabic (al-Kraysheh 2000: 60). In addition, rather just a string of names with a brief petition to a deity, this text contains a lengthy supplication to the god øa b, known from Aramaic texts from Palmyra, Petra, and Madā in Ṣāliḥ to be the “Fortune deity of the Nabateans” (Healey 2001: 153-154; cf. Milik 1972: 211-212). In the same year, during the Wadi ath-Thamad Project directed by Michèle Daviau, another even lengthier ‘Thamudic E’ was discovered at West Uraynibah, just 6kms south-west of Jiza. It was originally discovered by E. Knauf in 1984, and ‘re-discovered’ in 1986 by G. Bisheh. The text is remarkably similar to that of the Madaba text, containing an eloquent invocation to the same Nabataean deity, but in

Other disagreements with my interpretations have been due to a lack of understanding of the chronology and historical context of the texts. As we have seen, most of this material emanates from the first several centuries of the Roman imperial era. Few allusions in the texts can be assigned to a period after the C2nd. This has been ignored by some critics. For example, Whittaker notes that “it is hard to talk away all the references to attacks from beyond the frontiers” (1994: 213), and then cites material from the fourth century and later to prove his point. But in my discussions (1978; 1989), I have made it clear that the period of time that I am focusing on is the pre-Diocletianic period. Parker even admits that “there is no evidence of 156

David F. Graf: Nomads and the Arabian frontier: the epigraphic perspective

serious nomadic threats” in the second century, in spite of arguing that nomads were a constant threat from preclassical times into the Islamic period (1986: 635). It is not until the dramatic Diocletianic military buildup in the late C3rd that one can postulate a major Arab threat. At this time, Parker is aware that the “the [Safaitic] texts cease during this time of turmoil in the North Arabian desert in the late third century” (2000: 371). In fact, they seem to have flourished centuries earlier. As a result, the massive Roman buildup takes place when this so called evidence for nomads disappears, and the texts flourish during a time before there was a Roman defensive system.

al-Khraysheh F. 1997 Safaïtische Inschriften mit Jahresangaben. Sūdarabien. In R. Stiegner (ed.) Memoriam Maria Höfner. (Graz): 69-79. al-Khraysheh F. 2000 An Arabic inscription written in Thamudic scrip[t] from Jordan. Adumatu 2: 59-70 - in Arabic with English summary. Attalah N. 1995 Inscriptions inédites du ›awran (Raod.at alRoey). Syria 72: 387-399. Attalah N. 1997 Trois épitaphes du nord-est de la Jordanie trouvées à al-Khan. Abhath Yarmouk, Humanities and Social Sciences Series 13/1A: 9-13. Attalah N. forthc. Des inscriptions du nord de la Jordanie. Epigraphia Anatolica. Ball W. 2000 Rome in the east: The transformation of an empire. (London). Beaulieu A. & Mouterde R. 1947-48 La grotte d’Astardé à Was¥a. Mélanges de Université Saint Joseph 27: 120. Beeston A.F.L. 1981 Languages of pre-Islamic Arabia. Arabica 28: 178-186. Bowersock G.W. 1983 Roman Arabia. (Cambridge, Mass.). Calzini Gysens J. 1990 Safaitic graffiti from Pompeii. Proceedings of the Seminar for Arabian Studies 20: 1-7. Clark V.A. 1976-77 Some new pre-Islamic Arabian inscriptions from northern Arabia. Abr-Nahrain 17: 35-42. Clark V.A. 1980 Three Safaitic stones from Jordan. Annual of the Dept. of Antiquities of Jordan 24: 125-129. Clark V.A. 1987 The Semitic inscriptions. In S.T. Parker (ed.) The Roman frontier in central Jordan: Interim report on the Limes Arabicus Project, 1980-1985. British Archaeological Reports Int. Ser. 340; ii. (Oxford): 723-755. Daviau P.-M.M., Mulder-Hymans N. & Foley L. 2000 Preliminary report of excavations at Khirbat alMudayna on the Wadi ath-Thamad (1996-1999): The Nabataean buildings, with a contribution by C. J. Simpson. Annual of the Dept. of Antiquities of Jordan 44: 271-282. Drijvers H J.W. 1976 Das Heiligtum der arabischen Göttin Allât im westlichen Stadteil von Palmyra. Antike Welt 7/3: 28-38. Ghadban C. 1971 Un Site Safaïtique Dans l’Antiliban. Annual of the Dept. of Antiquities in Jordan 16: 77-82. Graf D.F. 1978 The Saracens and the defense of the Arabian frontier. Bulletin of the American Schools of Oriental Research 229: 1-26. Graf D.F. 1989 Rome and the Saracens: Reassessing the nomadic menace. In T. Fahd (ed.) L'Arabie préislamique et son environnement historique et culturel. Travaux du Centre de Recherche sur le Proche-Orient et la Gréce Antiques, vol. 10. (Leiden): 341-400. Graf D.F. & Roussan M. forthc. A new Safaitic inscription from Wadi Salma. Graf D.F. & Zwettler M. forthc. The North Arabian ‘Thamudic E’ inscription from Uraynibah West. Harding G.L. 1951 New Safaitic texts. Annual of the Dept. of Antiquities of Jordan 1: 25-29. Harding G.L. 1952 Some Thamudic inscriptions from the Hashemite kingdom of Jordan. (Leiden). Harding G.L. 1971a An index and concordance of pre-Islamic Arabian names and inscriptions. (Toronto). Harding G.L 1971b Safaitic inscriptions from Lebanon. Annual of the Dept. of Antiquities in Jordan 16: 83-85.

Moreover, if I am right in what I have argued above, it must also be emphasized that this material is not just the product of ‘outsiders’. Far too often the study of a frontier has just been the description of the forts that comprised what is considered the ‘frontier defensive system’, ignoring the people who resided both within and without the provincial borders. Partially, this is the result of lack of evidence, at least elsewhere in the Roman world, but this is clearly not the case for this region of the east. The north Arabian texts represent an essentially untapped archive of the native indigenous people. For the region of provincia Arabia, the impression has been given is that Arab nomads were the outsiders, and Aramaic and Greek speakers the insiders of the provincial realm. It should now be clear that from these recent finds of North Arabian texts that Arabs were not just outsiders, but insiders. The thousands of Safaitic and Thamudic texts must now be considered as an archive produced by sedentaries as well as nomads. Abbreviations The abbreviations for north Arabian texts cited in this discussion are listed in G.L. Harding (1971a: ix-xxxiii) and Macdonald (1993: 389-391) unless otherwise noted. CIS

=

Hava

=

Inv.

=

Kazimirski

=

Lane

=

Lisan

=

Corpus Inscriptionum Semiticarum J.G. Hava Al-Farā id ArabicEnglish Dictionary. (5th edit. Beirut, 1982. Inventaire des inscriptions de Palmyre A. de Biberstein Kazimirski Dictionnaire arabe-français. (2 vols. Beyrouth: Librairie du Liban, 1860). E.W. Lane An Arabic-English Lexicon. (Beirut, 1980). Ibn Man©ūr Lisān al- arab. (Beirut).

Bibliography Abbadi S. 1996 Mention of the Nabataean and Jewish War in the Safaitic inscriptions. Mu’tah Journal for Research and Studies, Series A, Volume 11/2: 239-253 - in Arabic with English summary.

157

Limes XVIII

Harding G.L. 1975 Further Safaitic texts from Lebanon. Annual of the Dept. of Antiquities in Jordan 20: 99-102. Healey J.F. 2001 The religion of the Nabataeans: A conspectus. (Leiden). Isaac B. 1992 The limits of empire: The Roman army in the east. (Rev. edit. Oxford). Kennedy D.L. & Riley D.N. 1990 Rome’s desert frontier from the air. (Austin). King G.M.H. 1990a Early North Arabian Thamudic E. Ph.D. dissertation, University of London, School of Oriental & African Studies. King G.M.H. 1990b The Basalt Desert Rescue Survey and some preliminary remarks on the Safaitic inscriptions and rock drawings. Proceedings of the Seminar for Arabian Studies 20: 55-78. Knauf E.A. 1983 Südsafaitisch. Annual of the Dept. of Antiquities of Jordan 27: 587-596. Knauf E.A. 1998 Die Altnordarabischen Inscriften. In J.-B. Humbert & A. Desreumaux (edd.) Khirbet esSamra en Jordanie I. (Turnhout): 523-544. Kokkinos N. 1998 The Herodian dynasty: Origins, role in society and eclipse. Journal for the Study of the Pseudepigrapha, Supplement Series 30. Littmann E. 1904 Semitic Inscriptions. Part 4 of the Publications of the American Archaeological Expeditions to Syria in 1899-1900. (New York). Studien zur Littmann E. 1940 Thamūd und øafā. altnordarabischen Inschriftenkunde. Abhandlungen für die Kinde des Morgenlandes 25/i. Leipzig: Deutsche Morgenländische Gesellschaft. Luttwak E.N. 1976 The grand strategy of the Roman empire. From the first century AD to the third. (Baltimore). Macdonald M.C.A. 1982 The inscriptions and rock drawings of the Jawa area. Annual of the Dept. of Antiquities of Jordan 26: 159-172. Macdonald M.C.A. 1992 The distribution of Safaitic inscriptions in northern Jordan. In M. Zaghloul, K. ‘Amr, F. Zayadine & R. Nabeel (edd.) Studies in the History and Archaeology of Jordan IV. (Amman/Lyon): 303-307. Macdonald M.C.A. 1993 Nomads and the ›awrān in the late Hellenistic and Roman periods: A reassessment of the epigraphic evidence. Syria 70: 303-413. Macdonald M.C.A. 1994 Safaitic. Encyclopaedia of Islam.2 (Leiden): VIII: 760-762. Macdonald M.C.A. 1995 Herodian echoes in the Syrian Desert. In S. Bourke & J.-P. Descoeudres (edd.) Trade, contact, and the movement of peoples in the eastern Mediterranean: Studies in honour of J. Basil Hennessy. Mediterranean Archaeology Supplement 3. (Sydney): 285-290. Macdonald M.C.A. 2000 Reflections on the linguistic map of pre-Islamic Arabia. Arabian Archaeology and Epigraphy 11: 28-79. Macdonald M.C.A. & King G.M.H. 1999 Thamudic. Encyclopaedia of Islam.2 (Leiden): X: 436-428. Macdonald M.C.A., Al-Mu azzin M. & Nehmé L. 1996 Les inscriptions safaïtiques de Syrie, cent quarante ans après leur découverte. Comptes Rendus Académie des Inscriptions & Belles-Lettres: 435-494. Milik J.T. 1958 Nouvelles inscriptions sémitiques et grecques du pays de Moab. Studii Biblici Franciscani Liber Annus 9: 330-358. Milik J.T. 1972 Dédicaces faites par des dieux. (Paris). (257261).

Millar

1993 The Roman Near East 31 BC-AD 337. (Cambridge, Mass.). Miller J.M. (ed.) 1991 Archaeological survey of the Kerak Plateau. (Atlanta). Müller W.W. 1982 Das Frühnordarabische. In W. Fischer (ed.) Grundriß der arabischen Philologie. I. Sprachwissenschaft. (Wiesbaden): 17-29. Néhme L. 1997 La géographie des inscriptions de Pétra (Jordanie). In Antiquitiés Sémitiques II: Des Sumériens aux Romains d’Orient. La perception géographique du monde. (Paris): 125-143. Parker S.T. 1986 Romans and Saracens: A history of the Arabian frontier. (Winona Lake). Parker S.T. 1987 The Roman limes in Jordan. In A. Hadidi (ed.) Studies in the History and Archaeology of Jordan III. (London): 151-164. Parker S.T. 2000 The defense of Palestine and Transjordan from Diocletian to Heraclius. In The archaeology of Jordan and beyond: Essays in honor of James A. Sauer. (Winona Lake): 367-388. Röllig W. 1987 Appendix A: The Thamudic inscriptions. In M. Ibrahim & R. Gordon (edd.) A cemetery at Queen Alia International Airport. (Wiesbaden): 43-45. Ryckmans G. 1947-48 Appendice: graffito safaitique. P. 20. Mélanges de Université Saint Joseph 27: 3-20. Safar F. 1964 Inscriptions from Wadi Hauran. Sumer 20: 9-27. Sartre M. 1985 Bostra: Des origines à l’Islam. (Paris). Sartre M. 1992 Transhumance, economie et societe de montagne en Syrie du Sud. In La Montagne dans l’Antiquité. (Université de Pau): 39-54. Sartre M. 2001 D’Alexandre à Zénobie: Histoire du Levant antique, IVe siècle av. J.-C.-IIIe siècle ap. J.-C. (Poitiers). Schlumberger D. 1951 La Palmyrène du Nord-Ouest. (Paris). Seyrig H. 1949 Antiquities Syriennes 41: Nouveaux monuments palmyréniens de Baalshamin. Syria 26: 29-41. Spaul J. 1994 Ala2: The auxiliary cavalry units of the preDiocletianic imperial Roman army. (Oxford). Whittaker C.R. 1994 Frontiers of the Roman empire: A social and economic study. (Baltimore). Winnett F.V. 1937 A study of the Lihyanite and Thamudic inscriptions. University of Toronto Studies, Oriental Series 3. Winnett F.V. 1985 Studies in Thamudic. Journal of the College of Arts, King Saud University 12: 1-52. Winnett F.V. & Reed W.L. 1970 Ancient records from North Arabia. Near & Middle East Series 6. (Toronto).

158

F.

David F. Graf: Nomads and the Arabian frontier: the epigraphic perspective

Fig. 1. The Basalt Desert in northern Jordan (after King 1990b: 56).

159

Limes XVIII

Fig. 2. The location of Safaitic and Thamudic texts in Jordan.

160

A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus in the second century AD Theodor Kissel After the annexation of the Nabataean kingdom in AD106, the Romans began to cover the new province with a close-meshed network of roads and stationed several military units in forts and fortlets on strategic points. The development of the provincial infrastructure was set up not only to faciliate communication between military sites but also to protect the Trans-Arabian caravan routes which passed the mountainous area of the Hauran. A recently discovered milestone from the southern edge of the Gebel el-Arab (formerly Gebel Druz) sheds new light upon the region’s road network. At first sight, its Antonine origin seems surprising since the earliest evidence for road building north of Azraq, as Kennedy has shown, dates from Severan times. But if we put the stone in a broader context, ie. comparing it with other road building activities in the early 160s in other parts of the Near East, the Hauran milliarium fits very well in a massive road building programme set about at the beginning of Marcus Aurelius and Lucius Verus‘ co-regency.

system (Kissel & Stoll 2000: 109-125).

After a lengthy hiatus, Roman roads in the so-called ‘lava lands‘ of the Hauran in southern Syria have again become the focus of archaeological research. This renewed interest in a formerly neglected region was stimulated by the ground-breaking work of Glen Bowersock whose seminal article in the Journal of Roman Studies (1971) and the subsequent monograph Roman Arabia (1983) redirected the researchers‘ attention to the eastern provinces and gave the impetus to further archaeological activity and historical analysis. The Roman road system, too, profited from that renaissance. Substantial contributions to the topic were provided by the commendable research of Thomas Bauzou (1985; 1988; 1989; 1998), David Graf (1995b) and David Kennedy (1982; 1997b; 1998; 2000c; Kennedy & Riley 1990) which tied into the important ground survey of the 1920s (Butler et al 1911-1943; Dussaud & Macler 1901; Brünnow & v. Domaszewski 1904-1909) and the pioneering aerial surveys of the enterprising French Jesuit Priest Père Antoine Poidebard and the Hungarian-British orientalist Sir Aurel Stein in the 1930s (Poidebard 1934; Gregory & Kennedy 1985; Kennedy 2000b: 28-36). Their extensive fieldwork illuminated the subject enormously and substantially contributes to our knowledge about the formerly only rudimentary understanding of the Roman road system1 in that area. Together with numerous regional field surveys2 in recent years these remarkable examinations have shown that the road system of provincia Arabia was much more complex than the Tabula Peutingeriana, with its focus on the via nova Traiana as the only highway of the province, makes us think. But despite the euphoria about recent activities, not all routes have been reconstructed. Some areas, for instance the black basalt desert of the Harra east of the Gebel el-Arab or the adjoining vulcanic landscape of the Safa in the north, are still a terra incognita while, on the other hand, remains of surviving roads are falling victim to human and natural influences.3 To make matters worse, many questions in several areas of the Near East remain unsolved, such as internal communications in smaller areas and their connection with a broader trans-regional network

During recent fieldwork in the mountainous region of the southern Hauran, the modern Gebel el-Arab (Fig. 1), hitherto unpublished evidence of Roman road building activity has been discovered in the village of Imtan (ancient Mothana), which – after meticulous observation – could be identified as a Roman milestone.4 At first, the site of discovery of the milliarium – 32kms south-east of the Roman provincial capital of Bostra – caused some amazement since hitherto there has been found only one tetrarchic milestone on the eastern fringe of the Gebel elArab, 35kms further north near the village of as-Sa’ane.5 Even more surprisingly, its time of erection could be dated exactly to AD162 and the tribunicial year of the two Augusti recorded in the inscription. The milestone from Imtan is therefore the earliest evidence of Roman road building east of the Gebel el-Arab. It also demonstrates that the ‘lava lands‘ of the Hauran mountains were much more densely covered with a close-meshed road network than modern findings seem to have shown. Until now, the earliest milestones from the desert fringe of the northern part of the limes Arabicus came from the Azraq oasis in the south and dated to the Severan period (Kennedy & MacAdam 1985; Kennedy 1997a; 2000a: 64-66). That north-south artery - recently called the via Severiana by Kennedy – was probably a former road of the strata Diocletiana which was built during the Diocletianic reorganisation of the eastern frontier, from Azraq in the south along the eastern escarpment of the Gebel el-Arab to Palmyra and from there up to the Euphrates. The tetrarchic milestone from as-Sa’ane with its formula STRATA DIOCLETIANI (sic!) and the nearby castellum seem to corroborate that assumption (Kissel & Stoll 2000: 114). Description and chronology The initial autopsy of the milestone during the initial visit in 1999 concentrated on the archaeological context of the object. The milestone is a fragmentary piece of a 4

The fieldwork was carried out within the framework of the Dept. of Special Projects in the Humanities (SFB 295): Cultural and Linguistic contacts: Modes of Change in Historical Areas of Tension in Northeast Africa and Western Asia, financed by the National Endowment of Research (DFG). For the topics of that research project, see Kissel & Beyer 1999: 68-71. 5 Still invaluable is the milestone corpus of Thomsen 1917:30 no. 61.

1

On the importance of roads for a society, see Kissel 1998: 149-51. 2 Graf 1997a: 123-133; 1997b: 271-281; Rasson-Seigne & Seigne 1995: 193-210; Kennedy & al-Husan 1996: 257-262; Kennedy 1997a: 71-93; Mittmann 1999: 24-44; Kissel & Stoll 2000: 109-125. 3 On the problem of the rapidly advancing decay of the ancient remains, see Kennedy 1995: 221-226.

161

Limes XVIII Arabian governorship can be dated to 162.6 Unfortunately the lettering on the lower left third is badly damaged7, which particularly affects line 10 which originally contained information about the construction of the road. At the beginning of the line vestiges of slightly incised letters are still visible in the basalt which can be read as [..]RAVER[...]. The perfect form of the verb sternere – here in the third-person plural (straverunt) usually used for newly-built roads, might be taken into account as a possible addition (cf. the formula aperuit et stravit at milestones of the via nova Traiana [CIL III 14.14921 = ILS 5834]). The bilingual distance m.p. XXII = KB at the lower end of the inscription clearly stands out from the remaining not very harmonic script, both in the size of the letters and in the care of their design. It indicates a distance of 22 Roman miles (milia passuum) between the place of erection and the road’s point of departure (caput viae). Since the distance of m.p. XXII = 32.56kms corresponds exactly to the one between Bostra and Imtan, it can be assumed that in all probability, a road was built from the legionary camp at Bostra towards the eastern edge of the Gebel el-Arab in 162, leading (via Salkhad) to Imtan (Bauzou 1985: 143).

cylindrical basalt column, 1.7m high, which had been reused as a doorpost in the masonry of a modern house (Fig. 2). Due to the severe deformation of the rough and pockmarked surface because of progressive weathering, it is very difficult to decipher the irregularly sized letters. To make matters worse, only two-thirds of the text are readable since the view of the rest of the document on the right side is obstructed by the adjoining masonry. In spite of these unfavourable circumstances it was possible to obtain a more detailed reading of the inscription by making a copy of liquid latex during a return visit in summer 2000 (Fig. 3): IMP(erator) CAE[S(ar)] M(arcus) AURELIUS ANTO[NINUS] AUG(ustus) PONTIF(ex) MAX(imus) TRI[B(unicia) POT(estate)] XVI COS III [ET] IMP(erator) CAES(ar) L(ucius) AURELI[USVERUS]5 AUG(ustus) TRIB(unicia) POT(estate) II CO(n)S(ul) [II DIVI] ANTONINI FIL(ii) DIVI [HADRIANI)] NEPOT(es) DIVI TRAIANI [PARTH(ici) PRO] NEPOT(es) [DIV]I NERV[AE ABNEPOT(es)] [ST]RAVER[UNT PER P(ublium) IUL(ium) GEMINIUM]10 MARC[(IANUM LEGATU]M PRO [PRAETORE)] XXII KB

Being the only testimony of Roman road building activity east of Bostra in the C2nd, placing the milestone in the broader context of the Arabian road system raises problems. In order to remove this otherwise single monolith from its supposed isolation, the following discussion will concentrate on the road system of the provincia Arabia at the time of its annexation in AD106.

The emperor Caesar Marcus Aurelius Antoninus Augustus, Chief Priest, holding tribunician power for the 16th time, thrice consul and the emperor Caesar Lucius Aurelius Verus Augustus, holding tribunician power for the 2nd time, twice consul, sons of the deified Antoninus (Pius), grandsons of the deified Hadrian, great-grandsons of the deified Trajan, victor over the Parthians, great-greatgrandsons of the deified Nerva, have built (this road), through the agency of Publius Iulius Geminius Marcianus, imperial governor (of the province of Arabia), 22 miles.

The second century AD – a poorly documented period in provincia Arabia The relevant publications on the ancient road present the Roman road system as a close-meshed and nearly completely developed network. However, this network was the result of a lengthy process of more than 100 years and can not be reconstructed in all of its periods (Kissel 1999: 71-73, 86-90). Comparatively well documented are the periods of consolidation of the eastern frontier under the Severans (Kennedy 1979) as well as the period of the reorganisation of the frontier’s defense system under Diocletian during whose reign the strata Diocletiana was built.8 Much more problematic on the other hand is the situation in the C2nd AD where the evidence is unequal to that of most the other Roman provinces.9 Although having an epigraphically and archaeologically well documented and in the meantime sufficiently researched highway for the early period of occupation in the form of the via nova Traiana – the great trunk highway of the provincia Arabia

The milestone, consisting of 13 engraved lines, can be dated by several different factors. The first date is given by the dies imperii of the two adoptive emperors (7 March 161) and the death of Lucius Verus in January 169 (Kienast 1990: 137; 143f). The missing otherwise obligatory epitheta of victory and imperial acclamations of the principes (ARMENIACUS 163/64, PARTHICUS MAXIMUS 165 and MEDICUS 166) (Kneissl 1969: 97-99) further limit the time range. The decisive criterion for dating the pillar is the tribunician power of the two emperors which places the inscription in the early reign of their fraternal co-regency. The senior emperor held the 16th tribunician power between 10 December and 9 December 162 while his junior partner, Lucius Verus, held the 2nd tribunician power, exactly as mentioned in the inscription. Finally, circumstantial evidence is given by the name of the imperial governor during whose governorship the milestone was erected. Based on the available evidence, the initial letters of his hardly visible name MARC can be reconstructed as Publius Geminius Marcianus as other milestones show, eg. from the via nova Traiana. His

6

PIR2 I 340; RE X 610, no. 259; Thomasson 1984: 328-29; Sartre 1982: 83-84. 7 This damage could also have been caused by wearing away the surface when the milestone was taken away from the ground. 8 Dunand 1931: 227-48, 416-34; 1931: 579-84; Mouterde 1931: 220-33; Bauzou 1993: 27-50. For the sector between Palmyra and Sura, see Konrad 1996: 163-80; Konrad 1999: 392-410; Bauzou 2000: 79-91. 9 Parker 1986b: 635; 2000: 369f., esp. for the C2nd AD.

162

Theodor Kissel: A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus

constructed under the emperor Trajan between 111-114 and running the length of the province from the capital at Bostra to Aela on the Gulf of Aqaba – the rest of the infrastructural building activities during the C2nd within the province is still beyond our knowledge. With the exception of milestones, there is for this time an unfortunate lack of epigraphic evidence in the form of military (building) inscriptions or military diplomas. Although having some roads and road installations like bridges, castella, halting posts, watchtowers, cisterns etc. from the meanwhile extensive fund of archaeological testimonies, these artifacts of annexation (Eadie 1985: 414: for lack of explicit chronological setting) are causing considerable chronological problems.10 Recent discoveries at the Roman fort at Humayma in the Hisma, just north of the Hedjaz, by Oleson has now slightly lifted the curtain of mystery (Oleson et al 1994; 1997). On the basis of the numismatic and pottery evidence found there he has dated the site to the first quarter of the C2nd AD. With the newly discovered milestone from Imtan in the northern part of the province we have more evidence of Roman policy in the C2nd AD, which puts new light on the ancient traffic network in that part of the limes zone.

archaeological remains dating them still causes considerable difficulties. So, unfortunately, we have hardly any knowledge of the early period of Roman annexation in the provincia Arabia. After a longer period of relative stagnation under Hadrian12 and Antoninus Pius there was an extensive construction of roads and a general strengthening of the defensive system in the Syro-Arabian border region in the latter years of the Antonines as reflected, for instance, by some east-west arteries which, according to the epigraphic evidence, seem to date to the reign of Marcus Aurelius and Lucius Verus. As a result of the focus on the via nova Traiana, these routes have been given little consideration in the last decades. The obvious concentration of horizontal roads in the early 160s may on one hand be explained by the strategic and infrastructural needs of a frontier zone which had been threatened by the persistent pressure and perennial menace of incursions by the north Arabian tribes, and on the other hand by the surprisingly massive Parthian incursions in to Cappadocia and northern Syria in 161. As an immediate reaction to those problems, for logistical and strategic reasons Rome increased her efforts in building up the horizontal communication lines in the hinterland of the Near Eastern provinces.13 These infrastructural efforts are well documented by numerous milestones from Judaea from 161/62 which have been pointed out by Benjamin Isaac. Many milestones from the adjacent provinces of Syria and Arabia, set up in the same year, confirm the assumption that the milestone from Imtan was apparently part of a comprehensive and wide-ranging road building programme which was implemented in the course of the reorganisation of the eastern frontier immediately after the devastating Parthian incursions in to Syria and Cappadocia. Additional threats by marauding nomads14 finally induced the Roman authorities to reinforce and enlarge the former fortification system further to the east by a ring of military installations (eg. Nemara15) which had been stretched out into the Harra for monitoring the desert tribes. Part of the vast and elaborate fortification system were the numerous military installations (at-Diyateh on the Wadi Gharaz or As-Sa’ane

As the construction of the via nova Traiana shows, almost immediately after the annexation of the Nabataean kingdom in 106, the Romans must have started improving the pre-existing communication lines by setting up a huge network of fortified roads linking the new province with adjacent provinces. Most of the initial road building activities were concentrated on the north-south arteries (the via nova Traiana and the Ledja road) which is also indicated by the bridges north of Bostra, at Jemarrin and Kharaba.11 However, in spite of the multitude of these 10 We have the same problem with determining the identity of the occupying forces of the provincia Arabia. Since there is not even a fragment of a military diploma amongst the more than 400 examples from the empire as a whole, the exercitus Arabicus of the C 2nd can only be determined insufficiently. For a better understanding of the C4th, see Parker 1989: 355-72. 11 MacAdam’s 1986: 26-27 assumption that both bridges were built at the same time as the bridge of at-Taybeh (164) can not be proved. So reservations must be expressed about his view since the bridges of Kharaba and Jemarrin are not dated by inscriptions. Therefore I would suggest the following hypothesis. It is based on the fact that the bridges can be dated to different periods by their different alignments (Kharaba north-west and Jemarrin north-east of Bostra). According to this hypothesis, there was only one bridge north of Bostra during the early period of annexation which was possibly built under Trajan in connection with the building of a road leading to Damascus conceived as a northern extension of the via nova Traiana. The location of the Kharaba bridge some 5kms north-west of the provincial capital may point to a road running in that direction crossing the fertile plain of the Nuqra, bypassing the Ledja to its western edge. This road possibly met the Damascus-Nawa-Gadara road of the Itinerarium Antonini (196-197) in the area of as-Sanamein as already supposed by Butler 1919: 440 – „(The road) passes around the southwest angle of the Ledja and then turned northward toward Damascus“; against MacAdam (1986: 21) who denies the existence of a paved road through the western Hauran in the C2nd – he only speaks of a „dirty track“. Actually, however, some traces of an ancient paved road are still visible on the ground between Bostra and the Kharaba bridge. Only some decades later, during the reign of Marcus Aurelius, a more direct connection between Damascus and Bostra came into being by cutting a road through the Ledja was there need for a second crossing over the Wadi Zedi near Jemarrin.

12 Hadrian’s visit to Arabia in 129/30 seems to have resulted in connecting the via nova Traiana with the Palestinian road system linking Scythopolis with Philadelphia, see Avi-Yonah 1950/51: 56. For the Gerasa-Adraa road which is dated by milestones to the year 120, see Mittmann 1964: 113-136; Isaac & Roll 1979: 61f. 13 In this context one should also mention the milestone from Akinyol near the banks of the Beuyuk Qara-tschai river on the Seleucia Pieria – Antiochia road (162/63), see Mécerian 1962: 301 and the bridge over the Beuyuk Qara-tschai and Kutchup Qara-tschai, the latter, however, epigraphically attested not until the C6th (IGLS III 2, 1142 – έγένοντο αί γέφυραι τ´ν δύο Μελάυτωυ. Against the background of the enormous logistical requirements, the link roads between the Mediterranean coast and the regions further inland were of utmost importance, see Adams 1976: 37-38; Kissel 1995: 30-31, 56-59; Roth 1999: 169 „supply lines“. 14 This was inspite of friendly relations between Rome and a few nomad tribes, as seen in the Rawwafa inscriptions of the Hedjaz in the Arabian peninsula, see Milik 1971; Bowersock 1975; 1983: 96-98; Graf & O’Connor 1977; Graf 1978: 8-12; Beaucamp 1979; Sartre 1982: 27-29. 15 Waddington 2271 = IGR III 1257 (Mesamarus eques); Waddington 2267 = IGR III 1259 (Γαδδος δροµεδάρις), see also the auxiliary unit of the ala dromedariorum from Meda’in Salih in the Hedjaz. For these indigenous elements in the Roman frontier defense, see Dabrowa 1991: 365 for the area of responsibility of these units. Graf 1995a: 296ff.

163

Limes XVIII

on the Wadi as-Sham) which clustered around the edges of the wadi banks that penetrate the desert frontier (Bowersock 1983: 94-109; Freeman 1996: 108-109). These intervisible mountain-top forts also served as signalling posts, where beacons could be lit to carry urgent, although necessarily very restricted information, from the frontier zone to the legionary headquarters at Bostra.16 The effectiveness of this militarized zone was to a high degree dependent on rapid communication and co-operation between the frontier forces which could only be guaranteed by fully developed roads (Fig. 4). However, the state of the eastern frontier defensive system already marked the end of a process which for the first time is partially provable in Severan times and was finally well developed in the tetrarchic period. The milestone from Imtan may possibly indicate the beginning of that enlarged frontier policy whose origin therefore can be traced back to 162. Therefore Septimius Severus and his successors would have reaped what Marcus Aurelius has sown.

interpreted the building activity in a broader sense, as routine road building carried out for completing the existing road network in Judaea. Whatever the reasons for the enormous efforts, they did not remain an individual case. In the adjacent provinces of Syria and Arabia19, there is quite a lot of evidence for pushing up with the infrastructure of the Roman territory which was essentially directed at maintaining efficient communication lines. The simultaneous road building in all three provinces elucidates its special importance, and, from my point of view, cannot be explained as regular repair work but rather as a reaction to an exceptional situation. What event would provide a better solution than the Arsacid attack on Syria and Cappadocia in 161/62 which literally caught the two newly proclaimed emperors on the wrong foot? Building activities on the Antiochia-Ptolemais road The earliest proven evidence of road building under Marcus Aurelius and Lucius Verus comes from the SyroPalaestinian coastal road. This longitudinal road, commonly known as the via Maris, running the length of the provincia Syria for about 463kms was the first road in the east attested by milestones. Its construction goes back to Nero under whose Syrian governor C. Ummidius Quadratus the road had been built „ab Antiochea [ad n]ovam colon[ia]m [Ptolemai]da“ in AD56.20 On some stretches of that main artery some road repair can be proven immediately after Marcus Aurelius and Lucius Verus came to power. There is the milestone from Maameltein from 16121 which was found about 5kms north of the mouth of the Nahr el-Kelb at milestation 215 on the Byblos-Berytus section. In the same year – most probably parallel to the activities in the north – further road repair took place on the same road further south on the Caesarea-

In spite of extensive field work in the last decades, our knowledge of Roman frontier policy in northern Arabia is incomplete. This peculiar lack stands in striking contrast to the remainder of Arabia. In the following, the milestone of Imtan will be used as an opportunity to subject Roman frontier policy in the second half of the C2nd to a more detailed analysis. To appreciate the milestone’s importance for the road system in Arabia, it is first advisable to consult the other evidence of the same period in that region. Evidence of Roman road building under Marcus Aurelius and Lucius Verus The milestones of the provincia Judaea At the beginning of our stock-taking we should look at the milestones from Judaea which date to 162. In his instructive article on the imperial road system in Judaea, Isaac drew attention to the unusually high number of these markers.17 With 31 specimens, its corpus by far represents the most extensive milestone series in Judaea (Graf, Isaac & Roll 1992: 786). The remarkably high number of stone pillars finally led Isaac to believe that the reason for the road building in 162 might be found in the outbreak of the Parthian War in 161.18 Henry MacAdam on the other hand

forces in the area or that it played any role in Near Eastern military affairs at all“. May it be as it be, but, in my opinion MacAdam underestimates the importance of infrastructure and logistical preparations for troop movements during military campaigns. A close connection between road building and military campaigns was convincingly demonstrated by Instinsky 1938: 33-50 with the example of Raetian milestones. According to this study the improvement of the Raetian road network was carried out as a run-up to the preparations of Septimius Severus against his British antagonist Clodius Albinus. As a result, Severan troops were able to reach the Gallic theatre much quicker, see Kissel 1995: 279. 19 via nova Traiana, see Bauzou 1998: 160-63 [no. 014-021]). For milestones on the Pella-Gerasa road, see CIL III 141769; Thomsen 1917: no. 226; Mittmann 1970: 153-59; Moors 1992: 98-101. For the GerasaPhiladelphia road, see CIL III 13.613; Moors 1992: 108. 20 Thomsen 1917: no. 9a2; Goodchild 1949: 91-127. The site named after Ptolemaios II Philadelphos (266 BC) was elevated to the rank of a colony (colonia Claudia Stabilis Germanica Felix Ptolemais) by Claudius. Veterans of the four Syrian legions were settled there as the founder’s coins with vexilla show, see Kadman 1961: no. 92ff. For the foundation of the colony, see Pliny NH V.17.75 – „Colonia Claudi Caesaris Ptolemais, quae quondam Acce [...]“; Millar 1990: 23-26; Applebaum 1989: 70; Isaac 1980/81: 37f.; 1990: 322f. 21 CIL III 208; Thomsen 1917: no. 4 - „Imp(erator) Caes(ar) / M(arcus) Aurel(ius) Antoninus / Aug(ustus) pont(ifex) max(imus) trib(uniciae) / pot(estatis) [---] / co(n)s(ul) II[I] divi Antonini / fil(ii) divi Hadriani / nep(os) divi Traiani pron(epos) / divi Nerv(ae) abnep(os) / m(ilia) [p(assuum)] CCXV / m(ilia) p(assuum) XI“ (the distances are referring to Antiochia and Byblos); Goodchild 1949: 107; 118.

16 For the optical signalling system of the Roman army, see Rebuffat 1978: 829. An early warning system is also attested on the Hadrian’s Wall: Woolliscroft 1991 and on the German frontier cf. Woolliscroft 1997. 17 The roads are listed by Isaac 1978: 49-50. Especially remarkable is the improvement to the transversal routes in particular: CAESAREA – LEGIO – SCYTHOPOLIS – PELLA – GERASA - PHILADELPHIA and DIOSPOLIS – JERUSALEM – JERICHO – LIVIAS - ESBUS which functioned as connecting arteries between the major longitudinal routes running on both sides of the deeply entrenched depression of the Jordan Rift Valley – the via nova Traiana and the via Maris. 18 Isaac 1978: 50 – „it was not the result of routine route repair“. In his opinion the foster father of Lucius Verus and former praefectus vehiculorum, Nicomedes, can be regarded as co-ordinator of the road building, see also Isaac 1990: 111, contrary to MacAdam 1986: 20 who categorically refuses such a causal connection – „There is as yet no evidence that the new road construction resulted in a buildup of military

164

Theodor Kissel: A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus

Ptolemais section as indicated by several milestones found north of Caesarea Maritima.22

The Damascus-Caesarea Panias road The building activities on the coastal road can also be seen further inland, east of the Jordan Rift Valley. During a field survey in northern Golan in 1973, two milestones came to light near the modern village of Taranĝe, one of which was inscribed.26 Originally set up on the DamascusBanias road which passed the nearby village of Taranĝe, the milestone dates to 162. An observation point on a nearby commanding peak found 2kms south-east of the village near the ancient site of Khirbet es-Sarbûkh apparently was used for the protection and surveillance of that important Trans-Gaulanitan link with its western terminus at Tyros.27 An Aurelian milestone found 5kms north-east of Tyros marks the western starting point of this link road which met the Antiochia-Ptolemais road at mile station 281.28 This inland road running south of the River Litani was still in use in the C5th as mentioned by a letter of Bishop Eucherius of Lyon addressed to a presbyterian called Faustus (?Faustus of Reji).29 A few years ago, MacAdam drew attention to the nearly identical wording of the milestone inscription from Taranĝe and the building inscription from the Roman bridge of at-Taybeh which led the author to assume a close connection between the Bostra-Der’a and Damascus-Banias road (MacAdam 1986: 20). It seems as if parallel to the extensive road work in Palestine the improvement of the road network further north in the Syro-Arabian frontier area was also forced.

The Bostra-Adraa road Because of its importance to the entire communication system in the Near East, this horizontal inland road was recorded in the Itinerarium pictum of the Tabula Peutingeriana. Although having only scanty visible remains on the ground23, the ancient Roman road can be definitely identified by a building inscription from atTaybeh which was built into the Roman bridge over the Wadi Zedi in 164.24 This strategic artery connecting northern Palestine via the Decapolis with northern Arabia is one of the most important east-west routes in the Near East that intersected the various north-south routes (eg. the via Maris and the via nova Traiana). The details of the itinerary and stations the Tabula Peutingeriana provides are as follows: after Tiberias, a station is listed at Gadara XVI m.p. further east; at XVI m.p. further it lists the station at Capitolias, and at XVI m.p. the station of Adraha, followed by Bostra at XXIII, the terminus of the route.25 For a long time the Roman bridge at at-Taybeh was regarded as the only evidence for this Transjordan highway. During recent fieldwork in north-western Jordan, new evidence has been revealed which permits a more detailed reconstruction of that road. Ground survey in the so-called plateau of the Ard el-Ala west of Gadara has recently brought to light four milestones of the early C3rd AD which indicate the initial course of the Gadara-Tiberias route (Riedl 1999: 45-48). On the other hand, east of Gadara an uninscribed milestone found on the eastern edge of the Wadi es-Sellale in the vicinity of the modern village of Hirbet el-Mugaiyir es-Serqi can be regarded as possible evidence for the reconstruction of the Capitolias-Adraa road (Mittmann 1999: 31-34, plate 6C). After leaving the Hauran plain directly west of Adraa, the Roman road after a few kilometres entered the north-western foothills of the ‘Ağlun mountains (Müller 1996: 93-99) where Siegfried Mittmann only a short time ago succeeded in finding more evidence of Roman road building (with the bridge over the Wadi aš- Šallale; Mittmann 1999: 24-44).

The Ledja road Extensive road-building activity in southern Syria during the reign of Marcus Aurelius and his junior emperor Lucius Verus is also well documented by a wealth of epigraphical and archaeological evidence (Isaac 1990: 138). The building inscription from at-Taybeh indicating road work on the Der’a-Bostra route in 164 has already been mentioned, and, in addition, some kilometres further north there is simultaneously evidence of road repair in the Barada valley on the Damascus-Baalbek road: viam fluminis vi abruptam interciso monte restituerunt, as in an inscription chiselled into the steep rock face above the road shows (CIL III 199). This conspicious concentration of infrastructural efforts in southern Syria in the early 160s led some scholars to the assume that the Ledja road could have been built at that time as well.30 This longitudinal

22 Roll 1996: 552-55; for Marcus Aurelius, see AE 1971: 470: „Imp(eratori) / Caesari M(arco) Aur(elio) / Antonino Pio / fel(ici) Aug(usto) co(n)s(uli) II[I] / filio divi Ti(ti) An / tonini Aug(usti)“; Isaac (1978: 50). The same section had already been renovated by the Syrian governor L. Caesennius Paetus in the early Flavian period (AD72), Thomsen 1917: no. 9a1. 23 During fieldwork in 1999 the author was able to find some isolated traces of ancient paving beside the modern railway near Ghasm. Here, the ancient road was apparently used as a foundation for the modern railway line from Bostra to Adraa. 24 PAES III A5, no. 622; Dussaud & Macler 1903: 290, no. 154; GermerDurand 1904: 32-33, no. 51; Brünnow & v. Domaszewski II 1905: 338; Thomsen 1917: 32, no. 66; Bauzou 1985: 143. 25 Tabula Peutingeriana Segmentum IX; Miller 1964: 830, fig. 260 [route 120]; Bauzou 1989: 207-09. For the further course of the road between Adraha and Gadara, see Mittmann 1970: 133-37; 1999: 24-44; Kettenhofen 1991: 87-89.

26

„[Imp(erator) Caes(ar) M(arcus) Aurelius Anto / ninus Aug(ustus) pont(ifex) max(imus) trib(uniciae) / pot(estatis) XVI co(n)s(ul) III et Imp(erator) Caes(ar) L(ucius)] / Aure[li]us [V]eru[s Aug(ustus) tri] / bunicia(e) pot(estatis) II [co(n)s(ul) II] / divi Ant[o]nini f[il(ii) divi] H / adriani nepot[es divi T] / raiani Parthici [prone] / potes d[ivi Nervae / a]bnep[otes]“; Urman 1985: 109-10; 133 n. 54; Moors 1992: 147. 27 Tabula Peutingeriana Segmentum IX: Tyro XXXII Caesarea Paneas XXVIII ad ammontem XXVIII Damaspo. 28 Cagnat (1936: 99-100, l. 10-13) – „A Tyro / metropoli / Panea M / E“; Goodchild 1949: 99. 29 Eucherii de situ Hierusolimae epistula ad Faustum Presbyterum XIV (= ed. Geyer 1965: 239) – „in quarto a Paneade miliario Tyrum pergentibus“. 30 Dunand 1933: 552-53; Poidebard 1934: 32-33, agreed with by MacAdam 1986: 23; Bauzou 1985: 151; Millar 1996: 108-109.

165

Limes XVIII the caravan trade through the Ledja.37 All in all, the obvious accumulation of military activities in southern Syria during the reign of Marcus Aurelius finally seems to corroborate the assumption that the Ledja road must have been built under that emperor.

artery constructed against all natural adversities traverses the inhospitable plate of broken volcanic rock known as Trachonitis and can still be seen from the air and on the ground in many parts with its almost straight alignment.31 The road is lined with watchtowers32 and milestones, one of which records that in 185/86 the road was restored ‘from Phaena to Aerita’, that is, from the northern edge of the rock plate at Mismiyeh almost to the southern edge at Ariqah. The term refecerunt indicates an earlier date of construction. A Greek inscription from the Nabataean sanctuary at Menara Henu on the Ledja road recently commented on by M.A. Speidel mentions a centurio of the legio IV Scythica who supposedly had been engaged with the security of the roads in the Trachonitis33, an area notorious for robbers. This dedication can be dated to the period between the 7 March 161 – the dies imperii of the principes - and 162, the end of the Syrian governorship of L. Attidius Cornelianus.34 If, as some scholars assume (Moors 1992: 76; Bauzou 1988: 294), there is a direct connection between the dedication at the sanctuary and the building of the Ledja road, there is no convincing reason not to regard Marcus Aurelius and Lucius Verus as the initiators of that road, too. The presence of the centurio mentioned above may therefore be explained by military road work to which his comrades of the legio IV Scythica had been appointed.

Mothana - an important military site of the limes Arabicus Imtan, the ancient Mothana of the Notitia Dignitatum (Or. XXXVII.3), is situated on the southern foothills of the Gebel el-Arab. This area in the southern Hauran, thickly strewn with basalt boulders, is the final area of fertile soils (after clearance) and relatively good rainfall before it gives way to the drier and inhospitable Black Desert of eastern Syria and Jordan. Due to these conditions the area was densely settled in antiquity. Confronting the desert and its nomads it is no surprise that this marginal land had to be policed and protected by the Roman army. In the course of further improvement of the military infrastructure in Arabia following the annexation of the province, the development of the mountaineous area of the Gebel Hauran was carried out in the early 160s. In the course of this the important transversal inland route connecting northern Palestine and north Arabia by way of Tiberias and Bostra was extended to its eastern terminus at Imtan as the newly discovered milestone now elucidates. Remains of that road are still rudimentarily visible on the ground below the Tell al-Khidr. Here, north-west of Imtan, the milestone must have been once set up to mark the last Roman mile on the Bostra-Mothana road before being removed and ending up in the wall of the entrance gate of the al-Atrash family’s house.38 There is weighty reason to believe that either or both former Syrian auxiliary units epigraphically attested at Imtan in the C2nd AD, the mil(ites) et equ[i]tes coh(ortis) I [Fl(aviae)] Can[at]hen(orum) [CIL III 14.379] and coh(ors) II Aug(usta) Thr(acum) eq(uitata) [CIL III 109], moved into their new fort immediately after completing the new road (Speidel 1977: 716). On the other hand the Gothic gentiles (PES no. 223), attested in the Imtan area in the first decade of the C3rd AD, may have been transfered to the limes Arabicus as part of the Severan oriental policy. The construction of another military road which runs from the Azraq oasis to the southern foothills of the Gebel el-Arab appearently belongs to the Severan era. Accordingly, one has to assume that this road, called the via Severiana by Kennedy, met the horizontal Bostra-Imtan road at Mothana where the latter bi-furcates (Fig. 5). Imtan therefore played a major part in the Roman defense system on the northern sector of the limes Arabicus. As an important road station

Further epigraphic evidence confirms a Roman military presence in the Ledja which unquestionably can be explained by security purposes.35 Numerous soldiers from different Syrian military units36 are attested in Phaena during the reign of Marcus Aurelius principally to protect

31 Dunand 1933: 521-56; Poidebard 1934: 29-34; Miller 1984: 196; Bauzou 1985: 139; 1989: 220-34. 32 The rôle of that roadside watchtowers was to facilitate the monitoring and protection of travellers, officials, and soldiers, progressing along routes. In such cases the objects of surveillance, the travellers, can be kept under continuous observation. The watchtowers therefore were one component of official attempts to suppress, or at least contain, petty brigandage in rugged, mountainous regions like the Trachonitis where the terrain faciliated the work of the bandits. 33 Dunand 1933: 539, no. 4; Speidel 1998: 170, 186 for the police duties in the Ledja; Bauzou 1989: 149 thinks there is a connection between the dedication and the road-building. 34 For the governorship of Cornelianus, see PIR2 A 1341; Thomasson 1984: 312, no. 57; Dabrowa 1998: 107f. who supposes that Cornelianus remained at his post up to the arrival of Lucius Verus at the end of 162. 35 For the problem of robbery in ancient Trachonitis, see Strabo Geog. XVI.2.20 (756); Josephus Bell. Jud. I.20.4 (398-400); Ant. Jud.XV.10.1 (343-348); Isaac 1984: 171-203; 1992: 62-63; Millar 1996: 36-37. 36 Two centurions of the legio III Gallica from Raphanea, Waddington 2525 = IGR III 1113 (AD166-169); Waddington 2528 = IGR III 1114 (168/69?). Another centurio of the same legion is mentioned in an inscription from Aerita/Ariqah where he supervised the building of a city gate dedicated by the veteran T. Claudius Magnus to his hometown, Waddington 2438 (168/69). Two centurions of the legio XVI Flavia Firma from Samosata, Waddington 2531-2 = IGR III 1121-2 (both undated, but the same officer, Petusius Eudemus, is mentioned in an inscription dated 177-79 from Philippopolis/Shahba [Waddington 2071 = IGR III 1195], and on another with the same date from Capra/al-Kafr in Hauran [IGR III 1290]. A centurio who cannot be placed in any particular legion [Waddington 2533 = IGR III 1132: Severus Askaion, χειλίαρχος] may be dated somewhat later).

37

According to Strabo Geog. XVI.2.20 (756) merchants from Arabia Felix were often victims of raids in the Trachonitis. The Roman authorities responded to this by policing mountainous zones as mentioned in an inscription from Canatha/Qanawat. Here a centurio of the legio IV Scythica was honored by the members of a collegium mercatorum from nearby Si‘/Seia due to his merits in safeguarding the local traffic, IGR III 1230; Speidel 1998: 186f. 38 Unfortunately the present house-owner could not give details about the origin of the milestone.

166

Theodor Kissel: A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus

conveniently situated on the desert fringe, Mothana was one of a string of fortifications which had been outlined east of the long-distance trade routes traversing the Hauran39 as an outer limes. The garrison at Mothana therefore was responsible for monitoring and surveillance of the border area and keeping up the communication lines to the rear, especially to the provincial capital of Bostra – headquarters of the legio III Cyrenaica (Lenoir 1998: 523528; Gatier 2000: 341-349) which served as a major nexus for a number of other routes.

de Recherche Archéologique 20 directed by Jean-Marie Dentzer of the CNRS in Paris this formerly neglected border area has finally been woken from its slumber. However, east of Bostra there is still very little known beyond a handful of undated archaeological sites and a few scattered inscriptions. The milestone from Imtan now opens up the possibility of obtaining more precise insights into Roman frontier policy on the northern limes Arabicus in the second half of the C2nd AD. On the basis of the current evidence there are strong signs that even at the beginning of the co-regency of Marcus Aurelius and Lucius Verus measures must have been taken east of Bostra with the final aim to integrate the Gebel el-Arab into the limes Arabicus. This assumption can be supported by a rapid increase of epigraphic evidence from the Hauranitian ‘world of the villages‘ where several military officers are documented at Nela/Mushennef42, 43 44 Philippopolis/Shahba , Capra/al-Kafr , Bosana/Busan45, and Canatha/Qanawat.46 Some of these inscriptions reveal close ties between the Roman army and the village communities in the Hauran mountains (Sartre 1987; 1993: 133-135; Sartre & Fauriat 1998; Kennedy 1999: 101). They testify to the general flourishing of these for the most part self-governing communities, a development which 70 years after the annexation of the province is reflected in advancing urbanization fostered by the Roman administration. Such a rapid development in civilization which resulted in an astronomical increase in monumental buildings (Ball 2000: 241-43) during the C3rd AD can be observed in the case of Shahba, ancient Philippopolis, on the north-eastern edge of the Gebel el-Arab which was given a magnificent architectural design by Philippus Arab, its native son who was declared emperor in the middle of the C3rd AD (Freyberger 1999: 263-269).

To what extent the north-south artery had been pushed forward further north beyond Imtan under the Severans as documented for the early C4th AD remains difficult to assess since the well preserved paved stretch of a Roman road near Khirbet Safiyeh (4kms north of Imtan) and the Roman bridge as well cannot be dated (Fig. 6). In any case, Mothana did not lose its military importance in late antiquity as shown by the C4th military handbook of the Notitia Dignitatum where the equites scutarii Illyriciani were recorded at Motha(na) (Or. XXXVII 14; cf. Sartre 2001: 629). Placing the milestone in the historical context Throughout more than 300 years of occupation in the Near East, the Romans outlined a massive defense system consisting of a deeply echeloned zone of military installations set in the steppe zone between the SyroArabian desert and the eastern edge of the Fertile Cresent at a length of nearly 750kms. This so-called limes40 - by no means a fully fortified obstacle consisting of artificial and natural barriers connected by military roads41 stretched from Sura on the Euphrates via Palmyra to the Hedjaz in the north Arabian peninsula. One of those natural obstacles was the Gebel el-Arab east of Bostra whose towering mountains provide a physical barrier to any invasion from the east. It also acted as a cultural buffer between the nomads living in the barren desert steppe of eastern Syria and the sedentary population settled on the fertile soils of the Hauran (Sartre nd: 39-54). This desolate volcanic area is interspersed with abundantly rich agricultural land and settlements. Its wealth and stability is reflected in the number and opulence of the public buildings at village level. One is therefore not amazed that this important albeit a long time neglected area has become the focus of archaeological investigation in the last two decades. Due to the stimulating and significant contributions of the Unité

Conclusion The newly discovered milestone from Imtan has shed new light on the regional traffic network of the northern sector of the limes Arabicus. It helps elucidate the scanty literary material and provides additional evidence for the complex traffic lattice that existed in Arabia as part of the Roman imperial network. This evidence permits a better understanding of military deployment and thus has implications for Roman objectives and strategy in this period. In a broader context the milestone from Imtan can be interpreted as follows: after reviewing the road building in adjacent provinces its putative isolation turned out to be an integral part of a comprehensive road building project on the south-eastern Roman frontier zone which after the Parthian threat in 161/62 had become a pressing necessity. In this context, however, the road building activities on the horizontal arteries as the Tyros-Paneas-Damascus road, the Tiberias-Bostra-Mothana road, the Baalbek-Damascus

39 Apart from monitoring of seasonal movements of transhumants in and out of the sedentary areas (Parker 1986a: 8; 129; Banning 1986: 25-50), the other major task which confronted the Roman army on the desert fringe was the protection of the great caravan routes coming in to Arabia and Syria from the Red Sea and from the Arabian peninsula (through the Wadi Sirhan) which also skirted the borderland between desert and sown. 40 This conventionally used term has recently been viewed as a modern misconception by Isaac 1990: 408-09. 41 Isaac 1990: 119-23, 171, 199-206; contra Luttwak 1976 in his analytical synthesis of the eastern frontier zone has argued with particular reference to roads such as the via nova Traiana and the strata Diocletiana that garrisoned roads in the frontier zone were utilised as communication and supply lines, not ‘frontier lines‘ or preclusive military barriers.

42 Waddington 2221 = IGR III 1261 (AD170/1); 2213 = IGR III 1262; Waddington 2214 (AD166-171). 43 Waddington 2071 = IGR III 1195 (AD177-79). 44 IGR III 1290 (AD177-79). 45 Waddington 2237 (AD170/1). 46 Waddington 2331 (AD170).

167

Limes XVIII

road, and the many roads in Judaea leading from the Mediterranean Sea towards the deeply entrenched depression of the Jordan Valley are most evident. The Bostra-Mothana connecting artery therefore should be seen as part of an extensive road building programme throughout the whole Near East which was directed at integrating the road networks of the three provinces into one regional traffic system.

J.-F. Salles (edd.) Aux origines de l’Archéologie aérienne. A Poidebard (1878-1955). (Beirut): 79-91. Beaucamp J. 1979 Rawwafa et les Thamoudéens. Súpplement aú dictionaire de la Bible 9: 1467-75. Bowersock G.W. 1971 A report on Arabia provincia. Journal of Roman Studies 61: 219-42. Bowersock G. 1975 The Greek-Nabataean bilingual inscription at Ruwwafa, Saudi Arabia. In J. Bingen, G. Cambier & G. Nachtergael (edd.) Le monde grec: Hommages à Claire Préaux. (Brüssel): 513-522. Bowersock G. 1983 Roman Arabia. (Cambridge, Mass.). Brünnow R.E. & v. Domaszewski A. 1904-09 Die Provincia Arabia, 3 vols. (Strassburg). Butler H.C. et al 1907-1949 Publications of the Princeton University Archaeological Expeditions to Syria in 1904-05 and 1909, 4 vols. In many parts. (Leyden). Cagnat R. 1936 Un nouveau milliaire de Syrie. Syria 17: 99100. Dabrowa E. 1991 Dromedarii in the Roman army: A Note. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 364366. Dabrowa E. 1998 Governors of Roman Syria from Augustus to Septimius Severus. (Bonn). Dentzer J.-M. (ed.) 1985 Hauran I. Recherches archéologiques sur la Syrie du Sud à l’époque hellenistique et romaine. (Paris). Dunand M. 1931 La strata Diocletiana. Revue Biblique 40: 227-248; 416-434. Dunand M. 1931 A propos de la strata Diocletiana. Revue Biblique 40: 579-584. Dunand M. 1933 La voie romaine du Ledjâ. Mémoires présentés par divers savants à l’Academie des Inscriptions et Belles-Lettres 13.2: 521-557. Dussaud R. & Macler F. 1901 Voyage archéologique au Safa et dans le Djebel ed-Druz. (Paris). Dunand R. & Macler F. 1903 Mission scientifique dans les régions désertiques de la Syrie moyenne. (Paris). Eadie J.W. 1985 Artifacts of annexation: Trajan’s grand strategy of Arabia. In J.W. Eadie & J. Ober (edd.) The craft of the ancient historian. Essays in honor of C.G. Starr. (Washington DC): 407-423. Freeman P. 1996 The annexation of Arabia and imperial grand strategy. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplmentary Series 18: 91-118. Freyberger K.S. 1999 Die Bauten von Philippolis: Zeugnisse imperialer Selbstdarstellung östlicher Prägung. In E.-L. Schwander & K. Rheidt (edd.) Stadt und Umland. Neue Ergebnisse der archäologischen Bauund Siedlungsforschung. (München): 263-269. Gatier P.-L. 2000 La legio III Cyrenaica et l’Arabie. In Y. Le Bohec (ed.) Les Légions de Rome sous le Haut-Empire. (Lyon): 341-349. Germer-Durand J. 1904 Rapport sur l’exploration archéologique en 1903 de la voie romaine entre Amman et Bostra (Arabie). Bulletin du Comité des Travaux historiques et scientifiques. (Paris). Geyer P. 1965 Itineraria et alia Geographica, vol. 1. (Turnhout) (= Corpus Christanorum Series Latine 175). Goodchild R.G. 1949 The coast road of Phoenicia and its Roman milestones. Berytus 9: 91-127. Graf D.F. 1978 The Saracens and the defense of the Arabian frontier Bulletin of the American Schools of Oriental Research 229: 1-26.

We can only hope that further evidence of comparable provenance will come to light in future which will set free the still imaginary shape of the northern frontier zone of the limes Arabicus in the late Antonine period. Acknowledgements I would like to express my appreciation to Prof. Dr. L. Schumacher (University of Mainz) for valuable comments on the reading of the milestone and to my Syrian driver Ghassan al-Shamat who was an always helpful companion and friend in the almost impassable area of southern Syria, Moreover I owe the Syrian Dept. of Antiquities, particularly Prof. Dr. M. Maqdissi, a special debt of gratitude for making it possible for me to carry out my research in Syria. References Adams J.P. 1976 Logistics of the Roman imperial army: Major campaigns on the eastern front in the first three centuries A.D. (Phil. Diss.: Ann Arbor). Applebaum S. 1989 The Roman colony of Ptolemais – Ake and its territory. In S. Applebaum Judaea in Hellenistic and Roman times. (Tel Aviv): 70-96 (= Studies in Judaism in late antiquity 40). Avi-Yonah M. 1950-51 The development of the Roman road system in Palestine. Israel Exploration Journal 1: 54-60. Ball W. 2000 Rome in the east. The transformation of an empire. (London). Banning E.B. 1986 Peasants, pastoralists and Pax Romana: Mutualism in the southern highlands of Jordan. Bulletin of the American School of Oriental Research 261: 2560. Bauzou T. 1985 Les voies de communication dans le Hauran à l’époque romaine. In Dentzer (1985): 137-165. Bauzou T. 1988 Les voies romaines entre Damas et Amman. In T. Fahd (ed.) Géographie historique au ProcheOrient (Syrie, Phénicie, Arabie greques, romaines, byzantines). (Paris): 293-300. Bauzou T. 1989 A finibus Syriae. Recherches sur les routes des frontières orientales de l’Empire Romain, Paris. (unpubl. masch. Diss.). Bauzou T. 1993 Épigraphie et Toponymie: le cas de la Palmyrène du sud-ouest. Syria 70: 27-50. Bauzou T. 1998 Le secteur nord de la via nova en Arabie de Bostra à Philadelphia. In J.-B. Humbert & A. Desreumaux (edd.) Khirbet es-Samra 1. La voie romaine, le cimetière, les documents épigraphiques. (Tournhout): 101-255. Bauzou T. 2000 La ‘strata Diocletiana’. In L. Nordiguian &

168

Theodor Kissel: A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus

Graf D.F. 1995a The Nabataean army and the Cohortes Ulpiae Petraeorum. In E. Dabrowa (ed.) The Roman and Byzantine army in the east. Proceedings of a colloquium held at the Jagiellonian University (Krakòw in September 1992). (Kraków): 265-311. Graf D.F. 1995b The via nova Traiana in Arabia Petraea. In J. Humphrey (ed.) The Roman and Byzantine Near East: Recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 141-167. Graf D.F. 1997a The via militaris and the Limes Arabicus. In W. Groeman-van Waateringe, B. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 123-133. Graf D.F. 1997b The via militaris and Arabia. Dumbarton Oak Papers 51: 271-281. Graf D.F. & O’Connor M. 1977 The origin of the term ‘Saracen’ and the Rawwafa Inscriptions. Byzantine Studies 4: 52-66. Graf D.F., Isaac B. & Roll I. 1992 Roman roads. In D.N. Freedman et al. (edd.) The Anchor Bible Dictionary, vol. 5. (New York): 782-787. Gregory S. & Kennedy D.L. 1985 Sir Aurel Stein’s limes report. British Archaeological Reports Int. Ser. 272. (Oxford). Instinsky H.U. 1938 Septimius Severus und der Ausbau des rätischen Straßennetzes. Klio 31: 33-50. Isaac B. 1978 Milestones in Judaea from Vespasian to Constantine. Palestine Exploration Quarterly 110: 4760. Isaac B. 1980/81 Roman colonies in Judaea: The foundation of Aelia Capitolina. Talanta 12/13: 31-53. Isaac B. 1984 Bandits in Judaea and Arabia. Harvard Studies in Classical Philology 88: 171-203. Isaac B. 1990 The limits of empire. The Roman army in the east. (Oxford). Isaac B. & Roll I. 1979 Judaea in the early years of Hadrian’s reign. Latomus 38: 54-66. Kadman L. 1961 The coins of Akko-Ptolemais. (Jerusalem). Kennedy D.L. 1979 The frontier policy of Septimius Severus: new evidence from Arabia. In W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies XII. British Archaeological Reports Int. Ser. 71. (Oxford): 879-88. Kennedy D.L. 1982 Archaeological explorations on the Roman frontier in north-east Jordan. British Archaeological Reports Int. Ser. 132. (Oxford). Kennedy D.L. 1995 The via nova Traiana in northern Jordan: A cultural resource under threat. Annual of the Dept. of Antiquities of Jordan 39: 221-226. Kennedy D.L. 1997a Roman roads and routes in north-east Jordan. Levant 29: 71-93. Kennedy D.L. 1997b Aerial archaeology in Jordan: air photography and the Jordanian southern Hawran. In G. Bisheh, M. Zaghloul & I. Kehrberg (edd.) Studies in the History & Archaeology of Jordan 6: 77-86. Kennedy D.L. 1998 Aerial archaeology in Jordan. Levant 30: 9196. Kennedy D.L. 1999 Greek, Roman and native cultures in the Roman Near East. In J.H. Humphrey (ed.) The Roman and Byzantine Near East, vol. 2. Some recent archaeological research Journal of Roman Archaeology Supplementary Series 31: 76-106. Kennedy D.L. 2000a The Roman army in Jordan. (London). Kennedy D.L. 2000b Relocating the past: Missing inscriptions from Qasr el-Hallabat and the air photographs of Sir Aurel Stein for Transjordan. Palestine Exploration

Quarterly 132: 28-36. D.L. 2000c Ancient Jordan from the air. Aramco World Magazine May-June: 36-46. Kennedy D.L. & MacAdam H. 1985 Some Latin inscriptions from the Azraq Oasis. Zeitschrift für Papyrologie und Epigraphik 65: 97-107. Kennedy D.L. & Riley D. 1990 Rome’s desert frontier from the air. (London). Kennedy D.L. & al-Husan A.G. 1996 New milestones from northern Jordan: 1992-1995. Zeitschrift für Papyrologie und Epigraphik 113: 257-262. Kettenhofen E. 1991 Zur Geschichte der Stadt Der’a in Syrien. Zeitschift des Palästina-Deutschen Vereins 107: 77-91. Kienast D. 1990 Römische Kaisertabelle. Grundzüge einer römischen Kaiserchronologie. (Darmstadt). Kissel T. 1995 Untersuchungen zur Logistik des römischen Heeres in den Provinzen des griechischen Ostens (27 v. Chr. – 235 n. Chr.). (St. Katharinen). (= Pharos. Studien zur griechisch-römischen Antike, vol. VI). Kissel T. 1998 Konstanten der Infrastruktur. Historische Wegekontinuität im nordsyrisch-obermesopotamischen Kulturraum am Beispiel der Flußübergänge am Mittleren Euphrat. In L. Schumacher (ed.) Religion – Wirtschaft – Technik. Althistorische Beiträge zur Entstehung neuer kultureller Strukturmuster im historischen Raum Nordafrika/Kleinasien/Syrien. (St. Katharinen): 147-78 (= Mainzer Althistorische Studien, vol. I). Kissel T. 1999 Ugarit – ‘Internationale’ Handelsmetropole im Schnittpunkt des vorderasiatisch-ostmediterranen Verkehrsnetzes. In M. Kropp & A. Wagner (edd.)‚ Schnittpunkt‘ Ugarit. (Frankfurt/M): 69-96 (= Nordostafrikanisch/Westasiatische Studien, vol. 2). Kissel T. 2001 Der Meilenstein von Imtan. Ein bislang unpubliziertes Zeugnis der militärischen Infrastruktur am nördlichen Abschnitt des limes Arabicus – ein Vorbericht (in press for Antike Welt 32.5). Kissel T. & Beyer J.M. 1999 Contact and change in the Near East: Studies in ancient economy and infrastructure at the Dept. of Special Projects in the Humanities (University of Mainz, Germany). Occident & Orient 4. 1&2: 68-71. Kissel T. & Stoll O. 2000 Die Brücke bei Nimreh. Ein Zeugnis römischer Verkehrspolitik im Hauran, Syrien. Antike Welt 31.2: 109-125. Kneissl P. 1969 Die Siegestitulatur der römischen Kaiser. (Göttingen). Konrad M. 1996 Frühkaiserzeitliche Befestigungen an der strata Diocletiana? Neue Kleinfunde des 1. Jahrhunderts n. Chr. aus Syrien. Damasszener Mitteilungen 9: 163-180. Konrad M. 1999 Research on the Roman and early Byzantine frontier in North Syria. Journal of Roman Archaeology 12: 392-410. Lenoir M. 1998 Bosra (Syrie): le camp de la légion IIIe Cyrénaique. Mélanges d’Archéologie et d’Histoire de l’École Française de Rome 110: 523-528. Luttwak E.N. 1976 The grand strategy of the Roman empire. From the first century AD to the third. (Baltimore). MacAdam H.I. 1986 Studies in the history of the Roman province of Arabia. The northern sector. British Archaeological Reports Int. Ser. 295. (Oxford). Mécerian J. 1962 Les inscriptions du Mont Admirable. Mélanges de l’Université Saint Joseph 38: 295-330. Millar F. 1990 The Roman coloniae of the Near East: A study of cultural relations. In H. Solin & F.M. Kajava (edd.) Kennedy

169

Limes XVIII

Roman policy in the east and other studies in Roman history. (Helsinki): 7-58. Millar F. 1996 The Roman Near East 31 BC – AD 337. (Cambridge, Mass.). Miller D.S. 1984 The Lava lands of Syria: Regional urbanism in the Roman empire. (phil. diss. New York). Miller K. 1964 Itineraria Romana. Römische Reisewege an der Hand der Tabula Peutingeriana. (Nachdruck Rom). Milik J.T. 1971 Inscriptions greques et nabatéennes de Rawwafah. Bulletin of the Institute of Archaeology 10: 54-58. Mittmann S. 1964 Die römische Straße von Gerasa nach Adraa. Zeitschrift des Deutschen-Palästina Vereins 80: 113136. Mittmann S. 1970 Beiträge zur Siedlungs- und Territorialgeschichte des nördlichen Ostjordanlandes. (Wiesbaden). Mittmann S. 1999 Römerstraßen in Nordwestjordanien und ihr Nachleben in der Kreuzfahrerzeit. Zeitschrift des Deutschen-Palästina Vereins 115: 24-44. Moors S. 1992 De Decapolis. Steden en Dorpen in de Romeinse Provincies Syria en Arabia. (s-Gravenhage). Mouterde P.R. 1931 La strata Diocletiana et ses bornes milliaires. Mélanges de l’Université Saint Joseph XV 6: 220-33. Müller D. 1996 Die topographische Vermessung einer antiken Straßentrasse bei El-Mugaiyir/Jordanien. Karlsruher Geowissenschaftliche Schriften, Reihe B, Band 3: 93-99. Oleson J. 1997 Landscape and cityscape in the Hisma: The resources of ancient al-Humayma. In G. Bisheh, M. Zaghloul & I. Kehrberg (edd.) Studies in the History & Archaeology of Jordan 6: 175-188. Oleson J. et al 1994 Excavations at Humeima, 1993. Echos du Monde Classique/Classical Views 38 (N.S 13): 141179. Parker S.T. 1986a Romans and Saracens: A history of the Arabian frontier. (Winona Lake). Parker S.T. 1986b Retrospective on the Arabian frontier after a decade of research. In P. Freeman & D.L Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297 ii. (Oxford): 633-660. Parker S.T. 1989 The fourth century garrison of Arabia: Strategic implications for the south-eastern frontier. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. Proceedings of a colloquium held at Ankara in September 1988. British Archaeological Reports Int. Ser. 553 ii. (Oxford): 355372. Parker S.T. 2000 The defense of Palestine and Transjordan from Diocletian to Heraclius. In L. L. Stager J.A. Greene & M.D. Coogan (edd.) The archaeology of Jordan and beyond: Essays in honor of J.A. Sauer. Studies in the Archaeology and History of the Levant 1. (Winona Lake): 367-388. Poidebard A. 1934 La trace de Rome dans le désert de Syrie. Le limes de Trajan à la conquête arabe. Recherches aeriennes (1925-1932). (Paris). Rasson-Seigne A.M. & Seigne J. 1995 Note préliminaires à l’étude de la voie romaine Gerasa/Philadelphia. Annual of the Dept. of Antiquities of Jordan 39: 193-210. Rebuffat R. 1978 Végèce et le Télégraphe chappe. Mélanges d’Archéologie et d’Histoire de l’École Française de Rome 90: 829-849. Riedl N. 1999 Eine neu entdeckte Meilensteingruppe in

Nordwestjordanien. Zeitschrift des PalästinaDeutschen Vereins 115: 45-48. Roll I. 1996 Roman roads to Caesarea Maritima. In A. Raban & K.G. Holum (edd.) Caesarea Maritima. A retrospective after two millennia. (Leiden): 549-558. Roth J.P. 1999 The logistics of the Roman army at war (264 BC – AD 235). (Leiden). Sartre M. 1982 Trois Etudes sur l’Arabie Romaine et Byzantine. Collection Latomus 178. (Brüssel). Sartre M. 1987 Villes et villages du Hauran (Syrie) de Ier au Vie siècle. In E. Frézouls (ed.) Sociétés urbaines, sociétés rurales dans l’Asie Mineure et la Syrie hellénistiques et romaines. (Strasbourg): 239-257. Sartre M. 1993 Communautés villageoises et structures sociales d’après l’épigraphie de la Syrie du Sud. In A. Donati (ed.) L’Epigrafia del villagio. (Rome): 133-135. Sartre M. 2001 D’Alexandre à Zenobie. Histoire du Levant antiquue IVe siècle avant J.-C. IIIe siècle après J.-C. (Fayard). Sartre M. n.d. Transhumance, economie et société de montagne en Syrie du Sud. In G. Fabre (ed.) La Montagne dans l’antiquite. Actes du colloque de la Sophau: 39-54. Sartre-Fauriat A. 1998 Culture et société dans le Hauran (Syrie du Sud) d’après les épigrammes funéraires (IIIe-Ve siècles ap. J.-C. Syria 75: 213-224. Speidel M.A. 1998 Legio IIII Scythica, its movements and men. In D.L. Kennedy (ed.) The twin towns of Zeugma on the Euphrates. Rescue work and historical studies. Journal of Roman Archaeology Supplementary Series 27: 163232. Speidel M.P. 1977 The Roman army in Arabia. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II.8 (Berlin): 687-730. Thomasson B.E. 1984 Laterculi Praesidium I. (Göteborg). Thomsen P. 1917 Die römischen Meilensteine der Provinzen Syria, Arabia und Palaestina. Zeitschrift des DeutschenPalästina Vereins 40: 1-103. Urman D. 1985 The Golan. A profile of a region during the Roman and Byzantine periods. British Archaeological Reports Int. Ser. 269. (Oxford). Woolliscroft D.J. 1991 Das Signalsystem an der Hadriansmauer und seine Auswirkungen auf dessen Aufbau. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 148-152. Woolliscroft D.J. 1997 Signalling and the design of the German limes. In W. Groeman van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 595602.

170

Theodor Kissel: A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus

Fig. 1. Map of the Hauran showing the location of Imtan.

171

Limes XVIII

Fig. 2. A milestone reused in a masonry of a modern entrance gate.

172

Theodor Kissel: A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus

Fig. 3. Drawing of the milestone’s inscription made in liquid latex

Fig. 4. The recently discovered Roman road leading from Rusheideh to the outer fortlet of Nemara.

173

Limes XVIII

Fig. 5. Map of Roman roads converging at Imtan

Fig. 6. The Roman bridge at Khirbet Safieh (north of Imtan)

174

Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes. Maurice Lenoir La présence sous le Haut-empire de la légion IIIa Cyrenaica à Bostra, capitale de la province d’Arabie, est une donnée stratégique qui ne peut être mise en doute, quelle que soit la date retenue pour l’installation définitive de la légion dans la province créée par Trajan1; la localisation de son camp au nord-est de la ville n’a cependant été établie que dans les années 1970 grâce aux travaux de S. Moughdad, publiés de façon succincte en 1976 dans le cadre d’une étude sur l’urbanisme de Bostra à l’époque romaine (Moughdad 1976). L’existence même du camp est donc encore considérée comme douteuse par Speidel (1977), dans son article sur l’armée romaine en Arabie. Quelques années plus tard, Fr. E. Peters reprend la description de Moughdad, mais considère que l’ ‘enclos’ au nord de la ville ne peut être identifié à un camp romain, mais plutôt à une zone extra-urbaine destinée au traitement et au stockage des céréales et autres produits agricoles destinés au marché damascène et syrien, zone qu’il caractérise comme un « commercial suburb » ou «suburban quarter [with] storage and marketing facilities ». Cet enclos aurait été ultérieurement inclus dans une enceinte romaine enveloppant Bostra, que Fr. E. Peters date hypothétiquement du IIIe siècle.2

1. Forme et superficie du camp Sur les photographies aériennes (Fig. 1) sont bien lisibles les murs modernes mémorisant, de façon irrégulière, le mur nord, une partie (les deux tiers?) du mur est et une partie (la moitié?) du mur ouest. Au sud, la voie rectiligne qui semble reproduire la limite du camp est récente, mais elle régularise une limite plus sinueuse figurant sur des clichés plus anciens, qui, elle, pourrait signaler à quelques (dizaines de?) mètres près le mur sud du camp. Toujours sur le côté sud, la large bande qui, sur la photographie ici publiée, apparaît occupée uniquement par des jardins est aujourd’hui densément construite. L’aménagement moderne de la source pérenne au sud-ouest et les constructions diverses sur tout le côté sud occultent les vestiges éventuels de la courtine; seules deux parcelles au sud-ouest sont susceptibles d’investigation (voir cidessous). L’estimation des dimensions du camp est rendue difficile en raison des habitudes locales d’aménagement des champs cultivés. Les limites de parcelle, dont la plupart sont déjà visibles sur les clichés aériens anciens, sont constituées de murs au tracé incertain, en blocs de basalte, hauts parfois de 2m à 3m, dont la base a été chemisée par des pierriers aménagés petit à petit lors de l’épierrement des parcelles et larges de 3m à 5m. Même si certains murs modernes semblent reproduire assez fidèlement le tracé des murs antiques, ceux-ci sont toujours masqués par l’accumulation des pierres.

Les photographies aériennes ne permettent cependant pas le doute et la zone au nord de la ville doit bien être identifiée au camp légionnaire de la IIIa Cyrenaica. L’une d’entre elles, datant des années 1970, est publiée ici (Fig. 1); une autre, provenant d’une couverture photographique de la Royal Air Force vers 1930, a déjà été publiée par Kennedy et Riley (1990: 125). La récolte de timbres légionnaires sur tuile dans la même zone, effectuée et publiée par Brulet en 1984, avait d’ailleurs confirmé, si besoin en était, l’existence du camp (Brulet 1984).

C’est en particulier le cas de l’enceinte. Les dimensions données par les différents auteurs varient beaucoup: la longueur est estimée de 400m à 470m et la largeur de 300m à 370m. Les conditions de terrain évoquées plus haut font que les dimensions très précises, au mètre près, données par certains auteurs ne peuvent être adoptées.4 Dans tous les cas et même si l’on adopte les mesures les plus fortes, la superficie du camp serait alors à peine supérieure à 17ha. Freeman (1996: 101), qui admet les mesures proposées par Kennedy et Riley et pour qui la superficie serait d’environ 16.50ha, note justement qu’elle est assez faible pour un camp légionnaire, dont la superficie moyenne est de 20ha environ.5

Dans le cadre d’une coopération institutionnelle établie en 1996 entre l’École française de Rome et l’Institut français d’archéologie du Proche-Orient (IFAPO), plusieurs campagnes de recherche ont eu lieu depuis mai 1997, dont j’exposerai ici brièvement les principaux résultats.3

1

On trouvera l’état de la question, avec toute la bibliographie antérieure, dans l’étude de Gatier (2000). 2 Fr. E. Peters 1983: 275-276 et fig. 5. Le plan de Bosra publié par Peters est beaucoup moins précis que celui publié par Moughdad (1976), en particulier pour la zone du camp. 3 Les principaux collaborateurs ont été MM. J. Leblanc (CNRS, Paris) pour les divers travaux de recherche sur le terrain et A. Kermorvant (Laboratoire d’archéométrie de l’Université de Tours) qui a réalisé la prospection géophysique en 1999. Mme J. Dentzer-Feydy (CNRS, Paris) étudie les éléments d’architecture décorés; M. P.-M. Blanc (IFAPO et CNRS, Paris) a expertisé la céramique découverte lors des prospections et des sondages; Th. Fournet (architecte à l’IFAPO) a réalisé les relevés topographiques d’ensemble. Des chroniques annuelles de ces recherches ont été publiées dans les MEFRA 110. 1 (1998): 523-528; 111. 1 (1999): 527-529; 112. 1 (2000): 496-502; la chronique des fouilles de l’année 2000 est à paraître dans les MEFRA 113. 1 (2001).

L’examen des photographies aériennes d’une part, la constatation d’une importante occupation médiévale d’autre part nous ont amenés, dès le début des recherches, 4

Kennedy et Riley (1990) fixent les dimensions à 463m x 363m, chiffres repris par Kennedy 2000: 206. Kh. Mukdad, dans sa thèse de doctorat de troisième cycle sur l’urbanisme de Bosra à l’époque romaine (thèse dactylographiée (Paris 1984) donne pour les petits côtés 300m à l’est et 360m à l’ouest, pour les grands côtés 422.50m au nord et 426m au sud. 5 Lambèse, Caerleon, Lauriacum ont une superficie d’environ 21ha; certains camps atteignent 25 à 27ha, comme Novaesium, Deva, Bonn. Voir les plans publiés par v. Petrikovits 1975.

175

Limes XVIII

à formuler l’hypothèse d’un camp primitif plus étendu vers l’est, dont la limite orientale coïnciderait approximativement avec la route moderne visible sur les photographies aériennes (B sur la Fig. 1) et dont la superficie, de 21ha environ, serait plus appropriée à celle d’un camp légionnaire que celle du camp « réduit ». Une campagne de détection géophysique a donc été mise en place en 1999; elle a porté essentiellement sur deux zones: au nord-est du camp sur trois secteurs (A, B, C sur la Fig. 2) pour explorer une éventuelle extension primitive à l’est ; au sud-ouest et à l’intérieur du “camp” (secteurs D et E sur la Fig. 2) pour tenter de cerner les limites ouest et sud et de mettre en évidence, au moins partiellement, son organisation interne.

perpendiculaires et parallèles à cet axe nord-sud, ont été mis en évidence dans ce secteur. La densité des vestiges repérés est notablement différente selon les divers secteurs de la zone explorée (secteur D sur la Fig. 2 ; voir également MEFRA 112. 1 (2000): 499; fig. 50). Un peu plus au nord, à l’occasion du défoncement du chemin longeant le “camp” à l’ouest par des bulldozers, la possible fondation de ce mur occidental a été découverte, sur une seule assise et une dizaine de mètres de long, dans le prolongement exact du mur repéré par la détection géophysique (voir MEFRA 111. 1 (1999): 528; fig. 44). Si le mur ouest du camp peut donc être situé avec assez d’exactitude, les murs nord et est ne le sont qu’avec une certaine marge d’incertitude. La localisation du mur sud, elle, nous échappe. On donnera donc au camp de Bosra les dimensions approximatives de 440m d’est en ouest et de 350m du nord au sud. Sa superficie serait donc d’environ 15.40ha, ce qui en ferait un des camps légionnaires les plus petits, mais proche de Carnuntum (c.14.50ha) et Noviomagus (c.15ha).

Au nord-est, les images magnétiques obtenues, très dynamiques, ont mis en évidence des contrastes significatifs, révélant un sous-sol hétérogène En particulier, dans la moitié ouest du secteur exploré, une zone de forts contrastes organisés est-ouest pouvait être interprétée comme la trace du prolongement vers l’est du mur nord du camp et de l’angle arrondi nord-est de celui-ci (secteur A sur la Fig. 2). Le retour oriental du rempart n’apparaissait cependant pas dans les explorations complémentaires (secteurs B et C sur la Fig. 2). Un sondage implanté en 2000 de façon à recouper perpendiculairement cette anomalie magnétique n’a cependant pas confirmé la nature anthropique de cette anomalie, qui est causée par une brusque élévation du niveau du basalte naturel (+ 0.8m sur 1m de longueur). L’hypothèse d’un camp primitif plus étendu vers l’est doit donc être abandonnée.

Fr. E. Peters suggère l’existence aux angles nord-est et nord-ouest de ce qu’il considère comme le “pseudo-camp” de deux tours carrées saillantes. Ni l’examen des photographies aériennes ni l’observation de surface ne confirment cette hypothèse. Il semble que le camp ait conservé les angles arrondis que l’on s’attend à trouver dans un camp légionnaire du Haut-empire. 2. La porte nord Depuis les fouilles de S. Moughdad, les murs alors dégagés avaient été en grande partie occultés par un pierrier qui s’était constitué sur et autour de l’espace fouillé. Dans un premier temps, en 1997, les vestiges anciennement découverts ont été nettoyés.

Au sud-ouest, un second sondage a été implanté à une vingtaine de mètres au nord de la rue moderne qui limite la zone du ‘camp’, de façon à recouper perpendiculairement une anomalie magnétique qui pouvait représenter la limite sud du camp (secteur E sur la Fig. 2). Sous une couche d’humus quasiment stérile, épaisse d’1m en moyenne, et un remblai épais de 0.3m, qui recouvrait au nord un éboulis peu compact de pierres non taillées, on a effectivement dégagé un mur large de 1m. Il est composé de blocs réguliers; ses joints sont largement beurrés d’un enduit de chaux qui inclut de gros tessons (Fig. 3). Le matériel recueilli dans la fouille de l’éboulis est varié; il comprend cependant une notable proportion de tessons peints attribuables à l’époque omeyyade. La fondation du mur n’a pu être atteinte pour des raisons de sécurité, mais aussi bien le matériel de l’éboulis que la technique de l’enduit ne permettent pas de le considérer comme le mur d’enceinte du camp romain. La limite sud de ce dernier n’est donc pas archéologiquement attestée : elle doit probablement être recherchée entre le mur dégagé cette année et une rue antique qui avait été vue sous la rue moderne en 1995 lors de travaux de voirie.

La porte elle-même est large de 2.70m; elle est munie de deux piédroits en appareil de pierre de taille soigné mais comportant des remplois. Le passage est encadré par deux tours rectangulaires, saillantes de 5m par rapport aux piédroits (le départ de l’enceinte n’a pu être repéré et on ignore son rapport exact avec ces deux tours), qui délimitent un passage antérieur large de 5m. Les murs est et nord de la tour ouest ont été dégagés sur toute leur largeur; ils sont composés de deux parements de blocs taillés, dont la face externe est assez bien dressée et d’un bourrage interne de blocs de diverses tailles (dont un fragment d’angle de corniche moulurée) mêlés à de petites pierres, cailloux, tesselles de mosaïque. Le passage antérieur a ensuite été bouché par un mur de facture semblable quoique moins soignée, large de 1.50m, délimitant une cour devant la porte et qui représente l’ultime état de l’occupation (Fig. 4).

À l’ouest, la représentation cartographique de l’ensemble des données de la détection géophysique suggère le tracé du mur ouest du camp sur une longueur de 100m. De nombreux indices de construction orthogonaux,

Ultérieurement, lors de la campagne de l’année 2000, un sondage a été implanté dans l’avant-cour de la porte, dans sa partie ouest, sur toute sa longueur nord-sud, pour tenter 176

Maurice Lenoir: Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes.

de préciser sa chronologie.

grand pierrier (75m sur 60m) a été constitué sur les thermes du camp, seul bâtiment identifiable et dont l’élévation est probablement presque intacte. Les dalles de basalte formant l’extrados des voûtes sont encore visibles au sommet de ce pierrier.

La zone a été très perturbée par les fouilles antérieures (on a retrouvé un fragment de sac plastique à 0.15m environ de la surface actuelle) et par la végétation: de grosses racines, certaines s’enfonçant jusque dans les anfractuosités de la roche mère, voire dans la fondation du seuil de la porte, ont été arrachées sur toute la superficie du sondage.

La zone du ‘camp’ se caractérise par l’abondance des blocs architecturaux décorés découverts dans les pierriers. Certains blocs ont été certainement déplacés.8 Cependant, dans un grand pierrier situé à l’emplacement probable de la porte orientale, la découverte d’un fragment de chancel et d’une pierre d’imposte décorée d’une croix pattée invite à formuler, avec beaucoup de prudence, l’hypothèse d’une église byzantine dans ce secteur, donc vraisemblablement après l’occupation militaire. Un aigle sculpté a été découvert dans un pierrier au sud de la porte nord ; sa présence n’étonne pas dans un camp légionnaire. Non loin de là a été découvert un fragment d’inscription impériale latine.9

La fouille a très rapidement rencontré la roche mère; celleci est très irrégulière et présente de nombreuses anfractuosités. Les murs du piédroit ouest de la porte et de la tour ouest ont été implantés directement dans celle-ci; on a pu retrouver la tranchée de fondation de ces murs, large de 0.07 à 0.12m, creusée irrégulièrement devant le seuil et le piédroit ouest de la porte et le long de la fondation du mur ouest de la tour; elle a été comblée par un calage d’éclats de basalte. La roche mère était recouverte d’une terre brune, présentant diverses nuances, à l’analyse non significatives, creusée au centre du sondage contre la fondation de la tour ouest d’une fosse destinée à accueillir trois grosses pierres plates, qui sont le seul vestige de la fondation d’un arc, accolé au mur de la tour après la construction de celui-ci (Fig. 5). Suite à l’identification de cette structure tardive, un sondage restreint a été implanté symétriquement contre le mur ouest de la tour orientale: à une faible profondeur, un ensemble identique y a été découvert, ce qui confirme l’hypothèse de l’adjonction d’un arc appuyé contre les murs des deux tours.6 On notera que cette structure a été découverte au-dessous du niveau du talon de la fondation de la tour. On peut donc envisager que les couches antérieures ont été détruites systématiquement lors de la construction de cet arc.

Les blocs décorés sont inégalement répartis sur l’ensemble de la superficie du ‘camp’; trois zones principales de concentration de ces éléments peuvent être distinguées: à la porte nord, dans le pierrier des thermes et les parcelles situées immédiatement à l’est, dans deux parcelles contiguës situées approximativement au centre du ‘camp’ et au sud du chemin moderne de direction est - ouest signalé plus haut (Fig. 6). À la porte nord, ces éléments d’architecture sont très variés, chapiteau, blocs de corniche corinthienne, tambours de colonne, dalle de plafond. Il s’agit évidemment de blocs réemployés dans la construction omeyyade qui constitue le dernier état d’occupation de la porte. En revanche, dans la zone des thermes et dans la zone centrale, les blocs décorés sont presque exclusivement des chapiteaux et des bases de colonne. Parmi les premiers, moins nombreux (respectivement 3 et 6 éléments dans chacune des deux zones), seuls trois éléments peuvent être datés approximativement et semblent relativement tardifs (IIIe/Ve siècles). En revanche, sur les dix-neuf bases découvertes (respectivement 9 et 10 éléments), treize (13) peuvent être datées de façon assurée de l’époque d’Hadrien (Fig. 7) et les six (6) autres appartiennent à un type dont la production commence en Syrie du sud également à la même époque, même si elle semble s’être prolongée dans le IIIe siècle.10 La présence d’éléments architecturaux comme ces bases n’est pas a priori étonnante dans des thermes monumentaux ; dans la seconde zone, leur

L’examen des rares tessons découverts a en effet montré qu’aucune trace d’occupation antérieure au Ve siècle n’est sûrement attestée, malgré la présence de tessons attribuables aux IIe et IIIe siècle et, dans les couches de remblai supérieures, de trois monnaies du milieu du IIe siècle à la fin du IVe siècle.7 La réoccupation très probablement civile de l’espace formé par l’avant-cour semble pouvoir être attribuée à l’époque omeyyade. 3. L’intérieur du “camp” À l’intérieur de la zone explorée, à l’exception des thermes situés dans le quart sud-ouest, aucun bâtiment n’est directement repérable. Des alignements de murs semblent marquer alternativement les deux rives d’une rue reliant les portes nord et sud ; cette dernière, souvent mentionnée, n’a pas été retrouvée. Un chemin moderne de direction générale est-ouest, malgré ses sinuosités, signale certainement sur les trois quarts de la longueur du camp une voie longitudinale. Dans le quart sud-ouest, à côté d’une salle chaude dont la voûte est encore conservée, un 6 7

8

On ainsi redécouvert une inscription funéraire grecque, vue par l’expédition de Princeton et considérée depuis comme perdue (IGLS 9405). 9 L’inscription était gravée sur plusieurs blocs successifs, de petite taille; un seul a été retrouvé. Trois lignes sont conservées: Antonini / Aug Pii / Fel Aug. La présence du titre Aug(ustus) aux deux L. 2 et 3 montre qu’il s’agit d’une inscription impériale avec indication d’au moins une filiation. La mention à la L. 1 d’un empereur Antonini [f(ilius)] permet d’exclure Antonin le Pieux, mais ne permet pas de décider entre la plupart des emperereurs depuis Marc Aurèle jusqu’à Sévère Alexandre, soit entre 161 et 235. 10 L’identification et la datation des éléments architecturaux décorés sont dues à J. Dentzer-Feydy, que je remercie.

Pour laquelle je remercie vivement P. -M. Blanc. Monnaies identifiées par Chr. Augé (CNRS, Paris), que je remercie.

177

Limes XVIII

accumulation incite à penser qu’elles marquent au moins approximativement l’emplacement des principia du camp légionnaire. Leur datation à l’époque d’Hadrien, soit peu de temps après l’arrivée de la légion à Bostra, voire même contemporainement à celle-ci11, indique qu’elles proviennent de bâtiments qui font partie de la première époque de la monumentalisation du camp. Les chapiteaux plus tardifs découverts dans les thermes pourraient alors indiquer une phase de réfection.

centaine de fragments de tuiles (tegula ou imbrex) marqués d’un timbre de la légion IIIe Cyrénaïque ont été ramassés.12 Seuls quinze (15) timbres ont été découverts dans le camp même, en particulier à l’occasion des sondages effectués dans la partie sud-ouest (voir supra). Soixante-quatorze (74) timbres ont été découverts dans la zone des entrepôts et neuf (9) exemplaires seulement dans la zone des fours. Le formulaire de ces timbres est d’une désespérante banalité ; il ne comporte que le nom de la légion; les seules variantes résident dans l’abréviation du surnom Cyrenaica, le plus généralement abrégé en Cyr(enaica) avec une liaison Y+R (voir MEFRA 111. 1 (1999): 529; fig. 45). Une typologie basée sur la forme du timbre et celle des lettres a été établie, comportant cinq groupes principaux. À titre de curiosité, on notera la relative abondance des timbres rétrogrades (15% des timbres classables).

4. Prospections autour du camp La prospection de surface a été étendue à la campagne environnante du camp. À l’est, des murs-pierriers ayant sensiblement les mêmes orientations que celui-ci peuvent être la mémoire d’un parcellaire rural. L’essentiel de la prospection a eu lieu au nord du camp, dans une zone constituant une légère proéminence, orientée est-ouest et s’abaissant progressivement selon cette direction, limitée au nord et au sud par deux légers thalwegs (voir MEFRA 112. 1 (2000): 498; fig. 49).

Au terme de nos recherches, la localisation du camp de la IIIa Cyrenaica, parfois mise en doute, nous paraît désormais amplement confirmée; la structure du camp primitif était apparemment très classique, avec un bâtiment assimilable à des principia au centre du camp, longé par une probable voie nord-sud, sans doute la via principalis. Sa forme générale se rapproche du carré13 et sa superficie relativement réduite en fait un des plus petits camps légionnaires connus, mais ces deux caractères ne sont pas contradictoires avec l’identification proposée. Les quelques sondages pratiqués montrent, en revanche, que le camp a connu une très longue occupation et qu’il faut tenir compte de réoccupations, accompagnées probablement de rénovations et/ou modifications des structures, aux époques byzantine et omeyyade, voire ayyoubide. Ces modifications, vraisemblables, nous échappent dans le détail; elles ont également occulté le camp du Haut-empire que nous ne faisons donc qu’entrevoir.

Dans une bande d’une trentaine de mètres de large sur presque toute la longueur du camp, sur le versant sud de cette proéminence, les pierriers sont caractérisés par l’abondance de fragments de tuiles, en particulier au nordouest, autour d’un gros pierrier construit sur la limite de diverses parcelles; ce pierrier et ses abords se sont révélés particulièrement riches en tuiles marquées (50 exemplaires, de tous les types; voir infra). Aucun échantillon surcuit n’a été découvert. En 1999, une brève opération de détection magnétique visant à mettre en évidence la présence de fours à tuiles a été effectuée sur cette zone dans le cadre du programme de détection géophysique (voir supra). Aucune anomalie magnétique révélatrice n’a été repérée. Cette zone, où aucun bâtiment n’est repérable par la prospection au sol, correspond plutôt à une zone d’entrepôts.

Bibliographie Un peu plus au nord ont été récoltés de nombreux fragments de tuiles surcuits ou vitrifiés, certains marqués. Ils ont été retrouvés sur une zone assez large, concentrée autour d’un gros pierrier; dans la même zone a été découvert un grand réservoir à ciel ouvert (birkê), approximativement circulaire, creusé dans le rocher et présentant deux niveaux d’utilisation ou de construction marqués par des terrasses. Il mesure une vingtaine de mètres de diamètre pour six à sept mètres de profondeur. Dans les parcelles adjacentes, plusieurs blocs de basalte allongés, à section en U et appartenant donc à un système de canalisation, ont été découverts dans plusieurs pierriers de séparation des champs. Ces vestiges pour l’heure dispersés témoignent de l’existence probable d’un système de stockage et de distribution de l’eau qui conviendrait bien avec celle d’un ensemble de tuileries, grosses consommatrices d’eau.

Brulet R. 1984 Estampilles de la IIIe légion Cyrénaïque à Bosra. Berytus 32: 175-179. Freeman P. 1996 The annexation of Arabia and imperial grand strategy. Dans D.L. Kennedy (ed.) The Roman Army in the East. Journal of Roman Archaeology Supplementary Series 18: 91-118. Gatier P.-L. 2000 La legio III Cyrenaica et l’Arabie. Dans Y. Le Bohec (ed.) Les Légions de Rome sous le HautEmpire. (Lyon): 341-349. Kennedy D. 2000 The Roman Army in Jordan. A handbook prepared on the occasion of the XVIIIth International Congress of Roman Frontier Studies, Amman, Jordan, 2-11 September 2000. (London). Kennedy D. et D. Riley D. 1990 Rome’s desert frontier from the air. (London). Moughdad S. 1976 Bosra. Aperçu sur l’urbanisation de la ville à l’époque romaine. Felix Ravenna 111-112: 65-81. Peters E. 1983 City planning in Greco-Roman Syria: Some new

Les marques sur tuile. Au cours des prospections, une

12

Brulet (1984) n’en connaissait que 19. La proportion de la longueur estimée à la largeur estimée est de 1,26, soit environ 2,5/2 alors que le pseudo-Hygin recommande, pour les camps de marche il est vrai, 3/2 (Mun. cast. 21). 13

11

Sur la question de la date d’arrivée de la légion à Bostra, voir ci-dessus n. 1.

178

Maurice Lenoir: Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes.

considerations. Damaszener Mitteilungen 1: 269-277. v. Petrikovits H. 1975 Die Innenbauten römischer Legionslager während der Prinzipatszeit. (Opladen). Speidel M.P. 1977 The Roman army in Arabia. Dans H. Temporini & W. Haase (ed.) Aufstieg und Niedergang der Römischen Welt II. 8. (Berlin): 687-730.

179

Limes XVIII

Fig. 1. Le camp de la légion IIIa Cyrenaica au nord de la ville moderne de Bosra. Photographie aérienne (1970 environ).

180

Fig. 2. La zone actuelle du camp, avec report des résultats de la détection géophysique. Fond de plan Th. Fournet (IFAPO) ; détection géophysique A. Kermorvant (Univ. de Tours).

Maurice Lenoir: Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes.

181

Limes XVIII

Fig. 3. Le mur omeyyade et son retour découverts dans le sondage 2 (secteur E sur la fig. 2). Cliché ML.

Fig. 4. Porte nord du camp après nettoyage, vue du sud-est. Au premier et au dernier plans, le pierrier établi sur les ruines de la porte ; au centre, à gauche, la porte (sont visibles une partie du seuil et le piédroit ouest) ; à droite, le bouchage ultérieur appuyé sur la tour ouest en saillie. Cliché ML.

Fig. 5. Porte nord, sondage dans la moitié ouest. Au premier plan, la roche-mère ; contre la fondation du mur de la tour ouest, trois dalles (une est cassée en deux) constituant la base de la fondation d’un arc (aménagement omeyyade). Cliché ML.

182

Fig. 6. Répartition des blocs d’architecture décorés dans la zone du camp. La taille des pastilles est approximativement proportionnelle à la densité des blocs découverts dans le secteur. Fond de plan Th. Fournet (IFAPO).

Maurice Lenoir: Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes.

183

Limes XVIII

Fig. 7. Bases de l’époque d’Hadrien dans un pierrier immédiatement au nord-est des thermes. Cliché ML.

184

45 years of research on the limes Palaestinae – the findings and their assessment in the light of the criticisms raised (C1st – C4th) Mordechai Gichon The author of this paper has been conducting intermittent research into the limes Palaestinae since the second half of the C20th. The following conclusions will be discussed in the light of the criticism raised against them. • the existence of a fortified border zone in Judaea-Palaestina to guard and police its desert boundary. • the Romans stepping into the belts existing, probably in Flavian times. • the fact that more often than not, the same sites were occupied, and that at times the same fortifications were utilisied. • the additional depth which was provided by the fortified hinterland of southern Judaea and the fortified line on the eastern edge of the Hebron mountains. • the courtyard type castellum having been applied since the Roman take-over and inspired by local precedents (going back to the C10th BCE and more). • the dictates of the geography influencing the strategy and tactics of the defence: the desert fringes, the nomads, the landbridges between the Red Sea and the Mediterranean, and between Arabia and the Mediterranean. • the common problems of policing and guarding against east and south: while the Nahal Beer Sheba line was given up after the Bar Kokhba revolt, the more southern portions of the limes were maintained by provincia Arabia. • Diocletian's reorganisation of the limes Palaestinae reflected the need to unite into one all of the limes, including the former Judaean line and its hinterland. • the consolidation necessitated the creation of Palaestina III to sustain the limes. • the existence of Illyrian fortresses prior to Diocletian, others having possibly been of later C4th date. • the exact layout of the border defence under Justinian and after is not yet clear. En Boqeq and Tamara were being held.

Former research and problems

its military aspect.2

In the generation after the Roman conquest (c.80-120CE), Rabban Gamaliel, head of the Jewish community of Palestine, mentioned Ḥeger as a region of reduced security, where orderly life could be jeopardized and a witness was sworn to a specific formula in certain legal matters. The late Benjamin Mazar proposed in an insufficiently recognized pioneering paper (Mazar 1949) to translate Ḥeger with limes, ie. the limes Iudaeae, and subsequently, in 1957, persuaded the present writer to make this limes the theme of his Ph.D. thesis (Gichon 1967a).1

It must be emphasized that until the mid-1950s the arguments of both sides were based mainly on narrow historical considerations, without reference to the small finds (pottery, coins, etc.) from the sites, nor to typological considerations of the architectural remains.3 To remedy the latter fallacies as much as possible, I started a survey of large tracts of the ancient limes in the Negev and on its confines. The outcome is the main base for my thesis, submitted in 1967, in which Alt’s and Mazar’s views were, to the best of my understanding, definitely sustained. Yet, only now, many doubters are being gradually convinced by

When commencing my research, I was able to base myself on the pioneering fieldwork of Alois Musil (Musil 1907), Fritz Frank (Frank 1934) and Nelson Glueck (Glueck 1953–1960). Frank’s survey served Albrecht Alt in the last part of his basic and seminal researches on the limes Palaestinae (Alt 1930; 1931; 1935. cf. 1955). His conclusions, concerning the Flavian date of the limes Palaestinae, were at first accepted by Avi Yonah (1949), but later, (idem 1958), rejected, and substituted by comprehensive arguments for a Diocletian foundation. With the noteworthy exception of Shimon Applebaum, who held to the “early date” (Applebaum 1961/62), many Israeli scholars tended to follow Avi Yonah. Extreme opinions even denied the existence of a limes, respectively

2 Shatzman 1983 denies the existence of the “early limes” and allows for some kind of military presence then and mainly in later periods. Tsafrir (in Baras et al 1982) admits the existence of earlier defences, but seems reluctant to name them ‘limes’ before the Diocletian reforms. Bowersock, in his magnum opus Roman Arabia (1983), chose to ignore the problem altogether by abstaining to mention the military sites and their discussion. In a former research (Bowersock 1976: 27) he speaks of a fortified road through the northern Negev. The reluctance to term the pre-Diocletian border defences in the Negev and southern Iudaea limes makes no sense once the borders are recognized as having been fortified and thus were under military control, whatever their exact functioning (see the following). Agreeing to the Roman presence from the C1st onward (Safrai 2001: 115-118) but negating its being called ‘limes’ prior to Diocletian, Safrai follows Parker 1986 (vide infra). He does not give any reasons, beyond quoting Shatzman’s doubting the dating of my survey sites, though in all cases, where they could be verified by excavation, the dates proved subsequently correct. He may have taken umbrage at calling the defences limes Palaestinae from the beginning. Up to 135, the designation should of course be limes Iudaea and was usually thus called by me (eg. Gichon 1967b: 40) when not (uncautiously) generalizing. Magness (1999) is the most recent gainsayer of the existence of a limes in Israel as well as of its military aspects. Unfortunately she has not discussed any of the basic problems raised by me beyond pottery discussions. Neither does Mayerson (1986), negating the Roman army’s task “to monitor and control the movement of Arab tribes”. 3 Frank has not published his pottery and Glueck, at best, very insufficiently.

1

Mishna Gittin 1.1: “If a man brought a bill of divorce from beyond the sea, he must say it was written in my presence and it was signed in my presence (by the husband staying there). Rabban Gamaliel says: “or even if he brought it from Rekem or Hagar”. Danby trad. p.307; Mazar 1949: 317 proves the term Rekem applied to regions/sites in the 1st Temple Negev and was used in the Aramaic Targum Unqelos to denote the regions of Shur and Qerar (Gen. 20.1); cf. Safrai 1994: 31-33; 2001: 115118.

185

Limes XVIII

evidence from sundry excavations, that the existence of a system of pre-Hadrianic border defences on the Judaean marches cannot be denied.

sounds to the ‘progressive scholar’, should not be overemphasized. Nathan Schur has collected the available written evidence and drawn a very bleak picture of Bedouin aggression throughout the C17th-C19th in the Golan, which has much in common with our parts (Schur 1993). Lithgew, Seezen, Burckart, Thomson and Schumacher are among those that underline this feature of life on the desert fringes.

Realities of border life and defence on border fringes Trying to define the problems and my arguments, I cannot but cover the same ground as I did in a number of earlier papers. For this, as well as for the frequent mention of these publications, the reader’s patience is requested (cf. Gichon 1967a; 1986; 1991; 1999a).

Yet even if border control was attempted, drought, cattle disease or pressure of other tribes from behind often formed a vital compulsion to transgress from desert and semi-desert into the sown. This implies the need for a permanent guard.

As a basis of my present deliberations, I wish to emphasize my belief, that, because of its location on the desert fringes, at no time, no government, sufficiently strong and sufficiently caring, wishing to safeguard orderly life, could neglect border defence and control on the western side of the Jordan.

Let it be said already now that the severer the environmental pressure - the severer the in-roads. Marcellus, comes excubutorum, has preserved the case of the invasion in the year 536 of 15000 (sic) Saracen tribesmen into Euphratensis, driven by extremely severe drought (nimiam sicitatem) after they had been denied pasture in Persian territory (pastura in Persica denignata) by the Lakhmid Mundir (Mommsen Chron. Min. II, p.105). Time does not permit to repeat my citations of the plethora of pre-Roman and post-Byzantine instances of the Cain versus Abel incidents recorded in Cisjordan. Experience in the C20th as to the attitude of the desert fringe tribes is sufficiently available. Let me quote two statements:

Let me repeat here two historical testimonies from my very first (English) paper dealing with the limes Iudaeae in its entity (Gichon 1967b): “...and so it was when the children of Israel had sown, that the Midianites…the Amalekites and the children of the East…came up…and encamped against them, and destroyed the increase of the earth, till they came into Gaza and left no sustenance for Israel, neither sheep nor ox nor ass. For they came with their cattle and tents, and they came as grasshoppers for multitude, both they and their camels were without number, and they entered the land to destroy it” (Judges VI.3-5).

(a) Aref el Aref, of proud Bedouin origin, British governor of Beer Sheba, told Haefeli in 1935: “The ghazu (marauding expedition) is a Bedouin habit, customary until recently. Many are the reasons: hunger, avarice, thirst, wish for revenge…beyond this, - as a pastime!” (Haefeli 1938: 153).

Nearly 3000 years later, the American consul in Jerusalem laconically reported, that, in his days, under the constant pressure of the Bedouins, the sedentary population had deserted its footholds in the plains and had been forced back into the natural bastion of the Judaean hills (De Haas 1934: 431).

(b) Touvia Ashkenazi, former officer in the Transjordan Frontier Force, who spent many years among the, much admired by-him, Bedouins, concluded in 1938 that as ever before, “Palestine was the destination of the thirsting and hungering nomadic tribes”. One of the many examples cited from his experience was “the migration of tribes from the Hedjaz and southern Jordan into the Beer Sheba valley in 1922, owing to the drought and the pressure of stronger tribes… ” (Ashkenazi 1938: 6-7 - author’s translation).

The explanation for the constant danger of nomad in-roads (Gichon 1991: 320-21 and bibliography there) is of course the fact that Iudaea has always been a climatic border. The 100mm isoyet passes through the Beer Sheba valley and any fluctuation in the yearly rainfall will endanger the very existence of the tribes on the desert fringes (ibid.). This, of course, also holds true for Transjordan. It does not exclude the possibility, of which, concerning the Negev, we have no proof, that from time to time, there developed a peaceful and mutual beneficial co-existence between the sons of Cain - the nomad husbandmen - and the sons of Abel - the sedentary tillers of the soil. However, only effective control, able to prevent nomadic transgression, was the conditio sine qua non for preserving this equilibrium.4 This peaceful coexistence, as appealing as it

The climatic conditions in ancient Palestine, coming within the wider scope of possible global changes, have been hotly argued. Baly (1957: 41ff.) has provided an apt survey of the problems involved. Archaeological finds tend to show that the climate has not changed sufficiently to cause a change in the flora (eg. Gichon 1992: 395-396). This, of course, has nothing to do with the fluctuations in precipitation, as implied by Palestine being a climatic border both in the east and in the south – the region of our

4

This, by me frequently reiterated necessity has been emphasized by other scholars too: (eg. Parker 1986: 8; 1991). Other desert borders had similar problems: Sidebotham 1991: 495 and sources there; Janon 1991, who argued the administrative and peacekeeping rôle of the army in Numidia, still leaves us with the need of the military presence to oversee and control. The same holds true for Wells 1991 and the French

colleagues cited by him. The fact that pastoral farming commodities begin to be in demand by sedentary communities does not negate their attempts at trespassing and raiding.

186

Mordechai Gichon: 45 years of research on the limes Palaestinae

Date 45 BCE +/- 68 CE 80-115 80-115 100-130 110-135 115-140 +/- 150 +/- 150 +/- 150 End of C2nd End of C2nd –C4th End of C2nd – C4th Ante 220 220-240 +/- 260 +/- 300 311/12 320-350

Event

Source

General drought and famine. Queen Helene of Adiabene came to the rescue Springs around Jerusalem dried up Rabbi Samuel the Little ordained two nation-wide days on two occasions because of the cessation of rainfall Ditto, Rabbi Eliezer ben Hyrkanos – 13 days of fasting During the drought years, Rabbi Tarphon celebrated marriage with several women, so as to feed them honourably (and not by alms) Eleasar be Perata said: “from the day the Temple was destroyed, the rains have become irregular” Rabbi Eliezer and Rabbi Aqiba ordained fasts and the rains came Rabbi Isaak recalls years of dried-up fields Rabbi José forbids passage on footpaths through private fields because of possible damage to poor crops in those days, as against former permission to do so, until the second (“late”) rains Rabbi Simeon ben Gamaliel: “…the Jews have been aversely affected” The Patriarch ordained fasting and publicly prayed for rain, but in vain. “From the late 2nd century onwards, throughout the whole of the 3rd and 4th century, we find frequent references of drought” Rabbi Levi…prayed for rains, but no rains came. Said he before him: “…Thou showest no consideration for Thy children”. Rains came, but Rabbi Levi became lame. The Patriarch Rabbi Jehuda opened his storehouse for free food distribution in the years of drought Severe droughts in north and south Rabbi Johanan blames the recurrent droughts on the rich, who did not fulfill their pledges for charities and alms Rabbi Levi recalled the farmer who converted half of his field into a water reservoir. During drought, the situation deteriorated so much that he sold 1 seah of wheat for 1 sela and 1 seah of water for 3 selas Drought – many villages wiped out Jewish fasts for drought to cease. cf. Rabbi Immi: “Rain is being withheld…

Josephus Ant. Jud. XX. 49.51 Josephus Bell. Jud. V. 40 B.Taanit 26.b Taanit ibid. Tos. Ketuvot 5.1 B.Taanit 19b Jer. Taanit 19.b Ber. Rabba 3.13 Tos. Sheviit 7.18 Tos. Sota 15.3 Taanit 24.a Sperber 1978: 72 B. Taanit 25.a B.B.B. 8.a B. Taanit 26 Ibid. 8.b Tanhuma Deut. Eus. HE. I B. Taanit 14.a-b B. Taanit 7.b

Table 1: Selected examples of drought in Palaestina during the C1st to C4th CE, based on the list compiled by Cippora Klein 1986

concern (cf. Baly 1957).

Table demonstrates.5

Table 1 enumerates 40 instances of severe drought that occurred in Palestine during the 700 years from about Herod’s reign to the advent of Islam. The true frequency of drought is of course considerably larger, since many of the incidents recorded spread over more than one year. Moreover, many local droughts, that severely affected the desert fringes, did have little effect on Iudaea-Palaestina as a whole, and did not find their way into the written sources.

Between 1863 and 1932, the Negebite Bedouins counted about 30 years of drought, which made Aref el Aref comment that the mainspring of the Bedouin raids was utter financial calamity and the need to rob for survival. These were coupled with greed and quest for prestige (Haefeli 1938: 152-153). The vital influence of the climatic changes on the politico-strategic developments in the late Byzantine east has been recorded by Butzer 1957 (cf. Gichon 1991: 320-322).

We note that the nature of a possible rhythm of dry and wet cycles in our region has not yet been established, and even less so, one of minor cycles and single season variations within the large ones. The existence of these variations is however a fact, and the Biblical story of the seven ‘fat’ and the seven ‘lean’ years (Genesis 41) is certainly set in a correct climatic environment, as our

5

C19th to C20th fluctuations of precipitation in Jerusalem have been examined by Striem 1985. Low rainfall averages (Streim 1985: fig.10) indicate a completely dry year south of the 100mm isoyet, passing through Beer Sheba, and the run-off waters, that were a major item of water replenishment for the desert fringe water catchments (Baly 1957; Evenari et al 1971). Striem’s table shows the occurrence of rainless, respectively low rainfall years within “cycles” that may be termed “rainy”.

187

Limes XVIII

Besides droughts and the other incentives for clashes between sedentary farmer and semi-nomad and nomad in the Beer Sheba valley and all other arable tracts of the Negev, there existed a further serious irritant that has not received attention by the students of desert-fringe symbiosis. The more these areas were reclaimed for sedentary agriculture and permanent settlements sprang up, the more the tribes were deprived of pastures and water. Even if they managed to survive, their sustenance depended on the even scarcer availability of these vital commodities and the most minor drought might compel them to invade the more northerly parts. Nabataean, late Roman and even more so, Byzantine times witnessed the great ages of agricultural expansion into the Negev and thus posed an additional security burden (cf. Villeneuve 1989).

largely fed by their economic rivalry, Iudaea by its very existence endangering the free outlet of the flow of the Arabian trade to the Mediterranean shores. Uneasy peace was very much an outcome of Roman pressure. The massive, enthusiastic participation of Nabataea in the final conquest of Iudaea in 67-73 and the equation of Idumaea as arch-enemy of Iudaea in Mishnaic and Talmudic thinking illustrate this attitude (Gichon 1996a: 25-26 contra Shatzman 1991). To the C1st CE fort and barracks (?), excavated at Sharuhen (Petrie 1930; Gichon 1991: 322; Yisraeli 1993) and bearing coins from 42 to 59CE, we can now add the excavated forts at Beer Sheba, ‛Ira, Arad, ‘Aroer and ‘Uza. All of them have 1st Jewish War destruction levels (Fritz 1973; Beit Arieh 1999; M. Aharoni 1993; Biran 1993; Beit Arieh 1986; 1993a). To these we need add the fortified hinterland, with strongholds as far west as Tel Shera (Oren 1993), as far north as the region of Hebron (Barouch 1995; 1997) and as far east as the edge of the Judaean mountain plateau (Gichon infra). A network of fortifications, largely distinct from those in the desert edge zone covered the whole of the southern Judaean hill country (cf. Gichon 1999b and bibliography there).

Nabataeans and Herodians (Fig. 1) Anyway, during the C1st CE, the Nabataeans felt the consistent threat from Saffaite and/or Thamudian script writing tribes inside and outside their realm. Suffices to read the citations by Littmann, Winnett and others in Graf’s deep-surfing 1989 study Rome and the Saracens, to be convinced that the tribesmen were frequently in hostile confrontation with the Nabataean government. Among the clashes mentioned on the graffiti were numerous raids for booty.

The garrisons for the rear fortifications of the Herodian limes were drawn, to my thinking, primarily from the local Idumean population. According to Josephus, they were organized as a well functioning militia-type force, in permanent, fully armed brigades, each under a hegemon. I have traced elsewhere the development of the Hasmonean frontier region of Idumaea. The performance of the Idumeans was sufficiently satisfactory to have Herod transfer 3000 of them to Batanaia and settle them there for similar purposes (Josephus Ant. Jud. XVIII. 9.2; Gichon 1967c; 1968: 328-333).

The ever-increasing number of Nabataean strongholds discovered by all surveys east of the Jordan and the Arava is sufficient proof (Graf 1989: 368ff.; Gichon 1986: 585586). We cannot think of any other enemy on the eastern marches of the Nabataean realm than these and other Arab tribes. As Graf pointed out, for the necessity of maintaining a constant border guard it made no difference whether some of these tribes had their grazing grounds partly or wholly within territory claimed and even administered by the Nabataeans. The deeper these tribes infiltrated into the country, the more complex and necessary the need for their policing became.

Still unravelled are the finds from the great civil settlement at the feet of the fortress of Malatha, but the rich pottery finds there testify to a strong occupation on the late Hellenistic, Hasmonean and early Roman (Herodian and later ?) periods. So far, the excavations on the tel have not brought evidence for all the post-Israelite phases. The excavators dated the small section of the Roman fort excavated there from the mid-C3rd to the late C4th. On the surface of the tel I did indeed pick up some earlier Roman sherds, but the relation between civil settlement and fortress is not yet well established (Gichon 1980: 847; Beit Arieh 1998).6

Though the laconic Saffaic inscriptions from the Negev do not speak of particular happenings, they must be attributed to the hundreds of proto-Bedouin encampments discovered by now in the Negev that are dated by their pottery to the C1st and C2nd CE (Avni 1982; Kaminer 1983; Gichon 1967b: 191). Nabataean fortified road stations, towers and castella, such as those along the axes Moa-Mampsis or Ovdat-Elusa, are proof that the situation west of the Arava was similar to the one in the east, ie. that the tribes that occupied these encampments and those behind had to be kept under control.

6

So far, financial strain has hindered the publication of our excavation in the vicus. The excavators of the tel have excavated only two small rooms of the fortress on the top of the mound (cf. Beit Arieh 1998; Kochavi 1993) and have identified the scarce pottery in two of the rooms cleared, as early Islamic. Clearly, this was a secondary occupation. Since the fortress was the only one overlaying its C6th BCE Judaean predecessor, it must have been identified with the base of the cohors I Flavia mentioned there by the Notitia Dignitatum (Or. XXXIV.45. Its relation to the Herodian pyrgos mentioned by Josephus (Int. XVIII.147) is not yet clear: does the fort overlie or incorporate the pyrgos or is it to be located elsewhere on the tel? No fitting structure from Hellenistic times has so far been discovered by the excavations outside the castellum, in the interior

This desert fringe watch formed also a serious consideration for the raison-d’être of the Herodian limes, the other reason being the guard against the Nabataeans themselves. The hostility between the two kingdoms was 188

Mordechai Gichon: 45 years of research on the limes Palaestinae

preparedness of one kind or another, when war, the ultima ratio regum was forced or decided upon. As a rule, the greater the threat of border transgressions, the more and better the border districts were being prepared by appropriate physical means to furnish the logistic and administrative requirements, necessitated at all times by the special status of these districts.

Flavian to Severan defences (Fig. 2) Critics of mine have contained that, granting the existence of Herodian border defence, it did not continue into the Roman pre-Diocletian Negev. Excavations at Beer Sheba, Ira, ‘Uza, ‛Arad (see Fritz 1973; Beit Arieh 1999; 1986; M. Aharoni 1993. For Beer Sheba, cf. Gichon 1980: 845-856; 1999a: 243) and En Tamar, on the Herodian borders and in the rear, in forts such as Tsalit and Rasm Jatir, have proved them wrong (Alon 1986; 1987). Coins at ‘Uza commence with Vespasian, at Beer Sheba with not later than Domitian (Beit Arieh 1986; 1993a; Aharoni 1973: 86-87), at Tsalit they continue with Vespasian until Hadrian, equally so at Jatir, and so on (Gichon 1999b, cit. there Alon]. They add to undoubted ceramic evidence (Fig. 2).

No wonder that we commence border installations and their unified command with the Pharaohs of the 3rd dynasty, who erected and maintained the “Wall of the Rulers” (Middle Kingdom appellation) to control the eastern approaches to the Delta (Helk-Otto 1956: 133). The C18th to C19th Austro-Hungarian Militärgrenze is only a more recent example (Allg. Militär Encyclopädie 1870: 6: 93-94) of the chain of defended border districts, into which the limites imperii fit as one further, though prominent, example.

The excavations in all the above sites have proved the correctness of my pottery analysis from my surveys (Gichon 1967a etc.). It is therefore with much certainty that we may assign castella, such as Sharuhen, H Kasif, Malatha and its vicus to this belt, as well as a number of turres. According to excavations, Hadrian’s reign, or more precisely the Bar Kokhba War, was the date of drastic reduction in these border defences. So much is sure! What we do not know as yet is whether, throughout all these years, part of these strong points, and mainly those in the hinterland, remained in Jewish hands with Roman approval. Had they been reoccupied by the Jews or had they become Roman and the destruction levels indicate their capture by the insurgents? One way or another, Hadrian’s reorganization after 135 will have included movement of forces inland into Iudaea to oversee its ‘pacification’ and deeper into the parts of the Negev pertaining to the provincia Arabia, to police this great expanse along its roads.

It seems to me that Isaac’s argument for the basically civilian nature of the limes (Isaac 1988) and the support he has received from various scholars, including Mayerson, stem from the multifaceted qualities of the limes, inherent in its being a ‘Militärgrenze’. Mayerson himself has underlined these qualities. It might be argued that even Isaac’s definition of the limes from the C3rd onward as a “frontier district administered by a military commander” (ibid.: 378) leads to my interpretation. The detailed discussion of the term limes and its interpretations are beyond the scope of this paper. A separate study is in preparation. However, I must add, respectively reiterate, the following: (a) that the simple facts of the existence of duces limitum and their administrative powers and the military frontier zone militias stamp the limites with a special military, administrative and strategic character, and (b) that various mentions of the term cannot be reconciled with a civilian interpretation. Take for instance the famous passage from the Vita Hadriani XII.6: …in pluribus locis, in quibus non fluminibus sed limitibus dividuntur stibibus magis in modum muralis … barbaros separavit. How can it otherwise be interpreted logically than that the limites are barriers and demarcation lines, respectively zones, (artificial and of human making) that continue the natural ones, on which the empire strives to lean its borders. Or, take the statement of Ammianus XXIII.5.2: In ipsis barbarum confinis interiores limites ordinaret…ne vagarentur per Syriam Persae. Without application of a tour de force, the simple answer is that the limes served the military aim to deny the Persians access into Syria.

Yet even after the existence of C1st-C2nd Roman border defences, that took over from the Herodian border guard, could not anymore be denied, the argumentation against their being a limes still continued. But what is a limes other than a centrally policed, defended and administered border, organized so as to regulate its passage, and, where deemed necessary, deny this transit by force? The physical nature of the limes may change according to time and border. To the best of my understanding, all imperial Roman mentions of this term fit this definition, as well or better than the proposed alternatives. It is my contention that the above definition of the limes expresses one of the major tasks incumbent on any sovereign government to ensure life in sedentary commonwealths, not threatened by unwanted and unregulated interference from abroad. Usually these tasks included defensive and at times, offensive military

Mayerson (1986; 1989a; 1989b) correctly underlined the limes’ composite nature, as I have already mentioned above. This however does not make an argument against the distinct and vital military connotation, uppermost in the conception of the limites imperii, whether ‘military’ indicates border control or backbone of defence against major aggression – or both. The comparison with the Carolingian Marken (Mayerson 1989b: 288) is apt. It

of which Hellenistic, Herodian, Roman, pre-Byzantine pottery was collected by me. The Byzantine period re-occupation cannot yet be ascertained (Gichon 1967a: II:1: 23-27b; Elder & Baumgarten 1993).

189

Limes XVIII

means that to maintain its military functions, the heads of the border sectors had to combine them with functions of civil administration. In Rome, from Diocletian times (?) onward, when much of border defence revolved on the, semi-civilian in character, limitanei, the civilian bureaucratic facet of the military administration of the duces grew in measure with the advancing volume of the state administration in general.

Iudaea to see to its ‘pacification’ after the war of 66-73CE and the battle against brigandism have been deemed a sufficient explanation for the disposition of the Roman forces on the Judaean borders, then, and in later periods. Accordingly, there was no need to explain the Order of Battle there as garrisons for frontier duties indicating the existence of a limes (Isaac 1984; 1988). Indeed, both tasks had to be undertaken, yet for both the displacement of troops on the borders was largely useless (cf. Parker 1991: 302). By the way, Iudaea then was not a specially brigandstricken province, if we remember “ληστής” frequently to mean, like ‘latro’, not simply robber and highwayman, but the irregular fighter against Rome.8 Therefore, every mention of the term must be checked against its true meaning. Yet, so or so, existing robbers and guerillas had to be combated mainly inland. The deployment of troops in border-region castella would have been of no help.

Bowersock (1983: 104-105), followed by Mayerson (1989b: 291, etc) emphasizes the depth as a necessary hallmark of the limes, of which more in the following. This links up with the above: had the limes been a line, (denied by me from 1967a onwards), its wider administrative rôles would not apply. Mayerson (1989b: 291; 1986: 39) aptly raised the problem of terminology: ‘Boundary’ is the actual demarcation line, while ‘frontier’ is the border region. This definition agrees with the one of leading geographers: (such as Prescott’s authoritative 1965: 30) ‘Boundary’ refers to a line, ‘frontier’ to a zone’, cf. Moodie 1947: Ch.V).

In the case of Nabataea, where we are speaking of desert regions, the Roman government must have been conscious of the fact that the Nabataeans, even if they were ready to do so, could not effectively prevent the tribal in-roads. The Nabataeans had their hands full with controlling the tribes and even uprisings east of the Jordan and the Arava, not to mention revolts such as the one of Damasi. In their endeavours to safeguard the trans-Negebite roads and their agricultural enterprises there, one may even ask whether tribal raids on Iudaea were not a welcome diversion from targets in the Negev proper. Therefore, it is wrong to assume that the fact of Nabataean presence in the Negev made the defence of the Judaean south unnecessary, as assumed by Bowersock (1976: 282), followed by Parker (1986: 119).9

The problem really arises with the word ‘border’, which actually is short for both ‘border district’, ‘border region’ and the like, or for ‘border line’, which lays it open to confusion. I have consistently used it in its former sense, but shall, in the future, make use of its more equivocal equivalent. Some scholars would use ‘border’ when emphasizing the non-physical aspects of the frontier region and give it its special qualities. These include the impact of cultures, languages, religion, social and political organization, economy and security arrangements from both sides of the boundary. These then assuredly form a vital component for any frontier studies.

Colleagues have drawn attention to the absence of literary sources for border incidents and unrest in the early postRoman conquest period. This absence of written evidence is contrasted with the mention of raiding from the C4th onwards and cited as proof for the non-existence of an ‘early limes’, even when fortifications did exist. My colleagues did however not pay attention to the fact that the writers recording this evidence were devoted Christians - more often than not clerics - who recorded the attacks on hermits, monks and priests in the Holy Land. Logically however, razzias and the like could not have been

For our special field of interest, it means that we have to consider the continuous influence of the desert fringes as a constant factor that cannot be dismissed on the ground that “quod non est in acta non est in vita”.7 A word on the use of the term limes in the pre-Diocletian period, in its military sense. The debate, when this definition of the term was introduced, has not yet been decided. Schleiermacher (1959: 14) and Forni (1967: 9) are to my mind still valid! I have however reiterated my need for the use of this term, even if it is chronologically premature, so as to convey concisely, in a single term, the ‘Roman borders’ as zones of physical, ethnic and cultural contact with the outside world, containing systems of control and bulwarks of defence. This multi-faceted nature of the term makes it the best available semantic vehicle to convey its composite nature whenever mentioning it.

8 The Roman equivalent of ληστής ‘latro’ is aptly defined in one of its meanings by Festus hostes hi sunt qui nobis, aut quibus nos publicum bellum decrevimus, ceteri latrones aut praedones sunt (Paulus Diaconus ex Festo, p.118 Müller). For sundry nuances of ληστεία, ληστικός as used by Josephus, that underline the non-robber definition of this term, see Gichon 1996b: 79-83. Other scholars have commented on this phenomenon in other provinces, (eg. Alföldi 1941: 43ff.). The wrong deployment, if the main task was internal security: see also Parker 1997: 116-117. See also p.14 supra. 9 Avi Yonah 1958: 135-137 antedates Bowersock in marshalling the argument against the existence of a Flavian and later pre-Diocletian limes in Iudaea/Palaestina. Much of Gichon 1967b is concentrated to refuting these notions. Nabataean in-roads into Roman territory, when, in the Augustan period it suited them, are reported by Strabo (Geog. XVI.IV.21). These might conceivably have been made by tribal elements and under their own initiative.

The need for a strong military presence in the interior of 7

Galestin (1997) is a good random example for the environmental factors influencing the military situation on a border quite different from ours. See the summing up on p.353: “…the Roman relation with the Frisians was not an accidental affair, but…a well planned frontier policy, in which the Frisian area was kept under military and economic control…”

190

Mordechai Gichon: 45 years of research on the limes Palaestinae

restricted to the clergy alone, but their occurrence, per se, was of no major interest to them. Moreover, Christian historiography and other writings proliferated only from the Constantinian age onward. Before that ‘pinpricks’, as far as imperial security was concerned, would not make the ‘headlines’, being, so to say, void of general interest.

with pre-70CE, and the Jerusalem Talmud, codified about 300CE, also partly based on early sources, written as well as oral, both have preserved precious authentic material relevant to our research. Referring you to the comprehensive article to appear hopefully soon, I must content myself here with some examples, partly quoted in former articles of mine, but hardly as yet discussed by my critics.

But what happened in the centuries before? Even if conditions in and beyond the tribal grazing grounds were less turbulent, border incidents must have occurred. Suffice the severe droughts mentioned in the rabbinical sources recurrently during the period after the Roman conquest and all through the C2nd (see above). To explain the silence of the official sources and the historians, be it permitted to illustrate the lack of interest in minor border incidents, though potentially dangerous, from personal experience. When serving as I.O., Southern Command, of the IDF in the mid-1950s, I compiled monthly summaries of the daily border transgressions, pilfering and theft. None of these were mentioned in the GHQ reports and appreciations, nor did they make the non-local press. While today there exists a special corps, the ‘Border Guard’ (Mishmar Hagvul), under the command of the police, in my time, it was, as in Roman days, the responsibility of the army to deal also with these matters, though unmentioned beyond local level.

Life on the C2nd desert borders in Iudaea was so insecure that the dwellers of adjacent settlements were dispensed from the customary ceremonial of declaring their innocence whenever an unidentified corpse was found in their fields. Dogs, that, in the settlements, had to be kept on leases at all time, were permitted to be set free at night time in ‘towns’ and villages on the border zone (Tos. Baba Kama 9:18, etc.; Mishna, Sota 9.2). In the mid-C2nd, these villages and towns were urged to construct a town wall, contrary to those inland, that were deemed sufficiently secure and did not have to take upon themselves the financial burden involved (Bavli B.B. 7,b). A town wall has been discovered in Sussia (Negev 1993d), the only southern Judaean town systematically excavated so far. A propos, not later than C3rd walls and towers were made out at Ovdat (Negev 1993c) and Mampsis (Negev 1967) and probably also at Elusa (Negev 1993a). Against whom? The answer is provided by rabbinical sources.

One wonders how much of this ‘small scale’ activity has been recorded for other provinces in written, respectively epigraphic sources. Syria, for which as kingpin of the Roman presence in the east one would expect a rather meticulous recording and clear archaeological evidence, Michaela Konrad had to marshal indirect evidence and logical deductions to prove the frühkaiserzeitliche presence on the north Syrian desert fringes (Konrad 1999). Her deliberations and findings are an important complement to my researches just because of Syria’s immediate frontline position vis-a-vis Parthia.

The Jewish settlements were apt to be surrounded by “Ismaelite” raiding parties (Gajis – ‫)גייס‬. The inhabitants were suspended from various observances to be able to repel them. Special “raider-watchers”, (Shomre Gajassot – ‫)שואדי גייסות‬, as distinct from ordinary town and field watchmen, were positioned, usually in towers, and sounded the alert by shofar (ram’s horn - ‫)שופר‬. The Tosephta Eruvin 4:5, ie. ante 300, has one of the most astonishing reference to border life: “if a razzu is announced, the townsmen are commanded to draw forth and apprehend it”. This, even on Sabbat, when leaving one’s living place is strictly forbidden and even if known that the raiders are heading to pilfer only “straw”. As a matter of fact, Jewish armies did not march on Sabbat without special dispensation, needed likewise (cf. Josephus Ant. Jud. XIX.63) for defensive actions in besieged places, nor was the carrying of weapons permitted. Obviously, the promulgators of this ruling were conscious of the fact that once assured that the Sabbat law hamstrings Jewish resistance, the tribal raiders would choose that day for constant razzias into the sown and undermine orderly life.

On the Arabian border, the positioning of the Trajanic to late C3rd castra at Humayma in the Hisma must be explained by conditions similar to those in Iudaea. Moreover, Humayma’s siting is a clinching argument against notions of inherent peacefulness of the desertfringe tribes, respectively of those in their rear, or the reasoning that troops posted there could have primarily had missions of internal security (Oleson 1999: 414ff.). Yet, in spite of all the above, there was one literary source that could not avoid mentioning the situation on the Judaean border. This was the rabbinical literature and mainly those legal writings that, in the limes hinterland, upto and including Hebron, regulated everyday life of the observant Jews, under all conditions. We recall that the Jewish villages formed, if not the majority, then a major part of settlement in southern Iudaea (Baras et al 1982: I: 192-198).

The Negebite borders between Iudaea/Palaestina and Nabataea/Arabia For a better understanding of borderlife in C2nd-C4th Iudaea/Palaestina, it behoves to define the provincial borders. The trace of the borderline between the Herodian realm, later the provincia Iudaea and Nabataea, hinges on

Therefore, the Mishna and the Tosephta, codified about 200CE, based on precedents and deliberations beginning 191

Limes XVIII

The Notitia mentions Carmel as the base of the equites scutari Illyriciani, who clearly were the immediate rear reserve of the main limes region forces as well as the main guard of the rear proper (Or. XXXIV.20; cf. Gichon 1971: 199, phase IV). Mention of the Notitia brings us to the situation in later periods. The reigns of Diocletian and Justinian are the termini ab quo and ad quem. Unfortunately, Magness (1999) has omitted to examine all issues of the problems concerned and has raised to my mind uncalled-for question marks. We refer for these two also to Gichon 1999b.

three strongholds, proven by excavation to have belonged to the former (see Fig. 1): Sharuhen, ‛Aroer and ‛Uza (Petrie 1930; Biran 1993; Beit Arieh 1986; 1993a). This means that it ran along the Naḥal Besor (Wadi Gaza), then on the hills bordering the Beer Sheba valley in the south as far as ‛Uza, and thence probably, including the Sohar ascent, down to the southern tip of the Dead Sea. Tamara, with its abundance of egg shell and painted Nabataean wares is thereby proven to be already outside the provincia and stamps its ascent as the most northerly pass into the Nabataean Negev (Gichon 1976). It seems that Trajan did not take to the idea of leaving the new provincia Arabia mistress over both the highway named after him and the other axis, Zoara-Tamara-Mampsis-Gaza, also of great international commercial importance. Consequently, he initiated the border arrangements between Iudaea and Arabia that are documented, according to my best understanding, in the list of his contemporary geographer, Marinus of Tyre, copied by Ptolemaios in his Geographia Hyphegesis (Der Kleine Pauly III: 1027-1028, s.v. Marinos von Tyros; IV: 1232, s.v. Claudius Ptolemaios). There, Elusa, Mampsis and Tamara are listed as belonging to Iudaea (V.15: 991-993] (See Fig. 3). No other explanation for this otherwise unrecorded border line comes to my mind. Following the Bar Kokhba War, Hadrian re-adjusted the borders with Elusa, Mampsis and probably the ruined Tamara within Arabia. The Geographia did not take these changes into account, but recorded the arrangements as described by Marinus. Only Malatha (if correctly identified with Μαλιαθα) was correctly assigned to Arabia, thus making the allocation of above three fortresses to Iudaea/Palaestina of Ptolomaios’ days anachronistic.

From Hadrian to the Illyrian emperors in the Negev and the Arava (Figs. 4 & 5) In the Negev, south of Beer Sheba and in the Arava, 14 excavated military sites, dated to the C2nd and C3rd, are situated south of the largely abandoned former main belt. The numismatic evidence, though not large in quantity, produced a sufficiently clear picture, summarized in the following table. To these findings we must add those from the two towns Ovdat (Oboda) and Kurnub (Mampsis) that cover all of our period. While the epigraphic evidence seems to prove the military presence at the latter town (p.22 infra), the situation at Ovdat after the abandonment of the camp and before the building of the fortress in Diocletian times, is not sufficiently clear in this respect. However, defensive measures are apparent (Negev 1993c: 1161ff). Locating the above named castella on the Figs. 4-6, these sites line the two main trans-Negebite roads and their approaches:

From Biblical times onwards, a major component of the Judaean border defences has been its hinterland – the Hebron hills. Most of what has been quoted from rabbinic literature pertains to this region. The present author has neglected sufficient personal survey of this region. This was remedied when, Hirschfeld (1979), followed by Barouch (1995; 1997), and others, began surveying and excavating there.

(a) the road Ḥazeva (Eisiba ?)–Rogem Zafir–Mezad Zafir (Cohen 1993h) - Ḥorvat Zafir and possibly Mezad Yeraq ‛am, to Mampsis, which had a garrison as late as in Eusebius’ time, ie. the early C4th (cf. p.22 infra; Eusebius Chromasticon 8.8, Klosterman). (b) Moa–Qazra-Mezad Neqarot–Sha‛ar Ramon–Ma‛aleh Mahmal to Oboda. This fact has been used by some scholars, (who could not deny any more the existence of the pre-Diocletian defences), to designate them as road defences. This, of course, is correct, but not so the faulty deduction that this hindered their being at the same time the main strongholds of the border defence system.10

Based on their findings, I was able to reconstruct the general picture of the hinterland including discussion of the peculiar ‘tower cum courtyard’ fortifications that complement the typical courtyard castella. The former like the latter may go back to 1st Temple period prototypes (Gichon 1967a; 1967b: 189-191; 1986; 1989; see also Guvrin 1996, who stresses the close inter-relationship between the Jewish farming communities and the Roman garrisons). Space necessitates my referring the reader to my 1999b article and the maps there. The problem which regular troops were garrisoned in the hinterland throughout the Roman period is still largely a mute point (cf. p.22 infra). In this respect, Speidel (1979) needs mentioning. Speidel discusses a stamped tile of the cohors I Thracum milliaria. The tile has been acquired in Hebron but was found in an unidentified spot in the hinterland. Unfortunately it might belong to any time from the early C2nd to 186CE and after.

10

Sidebotham’s hesitancy to accept the region of roads from Abu Sha‛ar and Leucos Limen to the Nile as a limes is over-cautious. Reference to my earlier papers might have already strengthened him in perceiving that a multi-purpose infrastructure is no objection to designating this zone a limes. See also Sidebotham 1991: n.12!

192

Mordechai Gichon: 45 years of research on the limes Palaestinae Table 2: C2nd and C3rd pre-Diocletian coin finds in the Negev and the Arava Key: + = single coin x = two or more exemplars 1 = ́Uza (Beit Arieh 1993a); 2 = Tsafit (Gichon 1974b: 23); 3 = Tamara (Gichon 1976: 90); 4 = Zafir (Cohen 1993h); 5 = R. Zafir (Cohen 1993k); 6 = Ḥazeva (Cohen-Israel 1996); 7 = Moa (Cohen 1993m); 8 = M. Neqarot (Cohen 1993f); 9 = Qazra (Cohen 1993l); 10 = Menuḥa (Cohen 1993b); 11 = Ḥ. Maḥmal (Cohen 1993e); 12 = Sh. Ramon (Cohen 1993n); 13 = Yotveta (Meshel 1989); 14 = Dafit (Cohen 1993a); 15 = ‛En Tamar (Cohen 1983). Emperor Vespasian Trajan Hadrian Ant. Pius M. Aurelius Commodus Sept. Sever. Caracalla Elagabalus Gallienus Claudius II Aurelianua Probus Carus Carinus

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

X X

X

+ X

X X

X

X X X

+ X X X

+

X

X

X

X X +

+

X + +

X +

+

X +

X

+

X

official, that travelled in the desert. He came to the 1st burgus and ate and drank there. Thence he came to the 2nd burgus ….and slept there”.

Any defence zone (as against single strongholds that do not combine into a larger tactical entity) is built around a network of latitudinal and longitudinal roads. These, together with the signal and observation posts are the very means of enabling coordinated border control and defence.

Midrash Psalms 19. 47: “…the caravan that …because of impeding darkness, approached a burgus. The burgarius said: enter, because of the robbers and the wild beasts”.

In the Negev, with its extreme topographical conditions, (mountains and deserts), roads, until recently, were limited to few, well defined traces. Also, the few sources of water add to the permanency and limitations imposed on the road infra-structure. Therefore, in this kind of terrain, detours are severely limited and even minor works will severely impede advance of large bodies of troops. Small raiding parties, though often able to move cross-country, are also in need of water for men and beast, and, besides, are apt to be enmeshed in the depth of the zone. Ergo, road defence is a main part of area defence everywhere and even more so in the barren wilderness. The limitations of movement in this type of country are apparent also from the fact that in the Negev military installations double as road stations. Mansiones and mutationes were either incorporated in the castella or built to be under their protection (Gichon 1974a). In the C6th, we have the accounts of the pilgrims such as Antoninus Martyr and Aetheria, to prove this axiom as fact (Geyer Itinera, pp.195 ff.; Perigrinatio Aetheriae). By the way, desert voyagers were housed in the castella, even after the army had evacuated them. Caravan trade, a main cog in the imperial economy, also made use of these military installations. Describing towers on the limes Palaestinae, I cited, in Gichon 1974a the following examples that need repeating. Though probably composed not before the C4th, they well illustrate also the conditions that prevailed earlier:

The first example illustrates the limes’ castella in their capacity as travel stations to aid the official travellers, the second one as aid to the commercial traveller, though not against mere brigands, as Isaac has argued (Isaac 1991). The functions of towers were manifold. In my comparative examination (Gichon 1974a), I distinguished between 8 different ones. Road defence (eg. Hieronymus Loca Sancta, p.25, ed. Klostermann) and protection from brigands (eg. CIL III.33.58) are explicitly mentioned. An important means for defining their exact purpose (s) is to establish their exact location. In the desert we need to add the further observation that robbers and brigands of rural and urban extraction were as much restricted in their movements as other non-desert dwellers. The robbers (lestes) of the deserts were of course the Saracens out for booty! The trans-Negebite roads had the special geopolitical function of connecting links between the Transjordanian outlets of the eastern trade and the Mediterranean. Strabo and Pliny both emphasized this link for their days by citing the axis Gaza-Aila pars-pro-toto also for alternative routes less known to them (Strabo Geog. XVI.4.27; Pliny NH V.XI.65). The importance of these roads fluctuated according to conditions on the silk road and its Persian outlets, trade competition and policy between Egypt and Palaestina etc. The general flourish throughout the Roman-

Wajikra Rabba 7.4: “Once there was a king, ie. Roman 193

Limes XVIII

Byzantine period of Ascalon, Gaza and other towns on the Philistine coast, as well as of the Negebite towns, is proof that commerce never stopped to flow along these arteries. Therefore, we need accord to the Negev limes this special rôle of safeguarding the international commerce, which is particular to its geographic location and not a feature common to all limites as such.11

Sundry arrangements and foeda with tribes as reflected in the C2nd Ruwafa inscription should be considered as having been a major part of the system on both sides of the Arava.14 Mampsis has furnished epigraphic evidence for the probable C2nd presence of the legio III Cyrenaica and a cohors I Thracrum Augusta (Negev 1967: 52-53; Speidel 1977: 710 and bibliography there].

On the Philistine coast, the trans-Negebite roads link up with the so-called via maris, the great international trunk road. Together with the via nova Traiana (the Biblical ‘King’s Highway’), it forms the only land-bridge between Africa and Eurasia. This fact has always been of major geopolitical importance and has given Iudaea-Palaestina its constant strategic importance, though in times of peace this seems not to have been in the forefront of Roman imperial thinking.

An important addition to our epigraphic sources touching on the deployment of troops in Roman Iudaea is the Greek language funerary inscription, so far only preliminarily published by Y. Guvrin, hailing from Beer Sheba or nearby (Guvrin 2000: see Fig. 7).15 Although we must wait for the final publication to clear up sundry basic points and offer conclusions, I do feel that the following remarks are permissible without going into a detailed epigraphic examination: the epitaph mentions the deceased, a ίππεος (in surviving line 2) who seems to have served in an ala Phrygum (Φρυγων) and the friend, who erected the epitaph, a στρατιωτης of the legio X Fretensis (lines 6 and 7). If the Z in line 3 may be interpreted as the numeral επτα, we could bracket their common service in Iudaea to from the Bar Kokhba War (dipl. CIL XVI.87) to the period between 160CE (according to RMD 173) and 186CE (RMD 69). A vexillatio of the legion and this auxiliary regiment may thus have been stationed on the Judaean side of the provincial borders in the Negev during much of the C2nd. Consequently, even the early limes belt from Gaza to the Dead Sea shores might have been held by some regular troops, besides the village homeguards.

Nevertheless, Roman planners were capable of appreciating the importance of the Palestinian land-bridge in peace and in war, even if no ad hoc military movements were contemplated. This knowledge was not restricted to the imperial administration. In his praise of the apt functioning of the empire’s roads during the reign of Pius (?) or Marcus Aurelius, Aelius Aristides (To Rome CI) singled out the security on the Cilician Gates and the via maris to Egypt.12 A propos roads and their defence, the question of what comes first, road or fortification (Graf 1997: 124; Isaac 1988: 128) is ill-defined. If the road in a defensive belt connects between its fortifications, it is, as pointed out above, an integral part of the system and may frequently postdate the former, although this is of no fundamental importance to its being a part of the defence belt or not. If the road is an interior artery of communication, its guard stations are not directly involved and may often post-date the road’s construction - but not necessarily so. The often cited passage from the Mekhilta, Mas. Ba-hodesh (Safrai 1971: 227), mentioning in the C2nd, angaria “to repair burgi on the roads that go up to the imperial cities”, clearly refers to the latter category.13

While in Arabia, east of the Jordan and the Arava, the Severan period has been pointed out as being a revival and even initial period of frontier defence, further to the west, judging by pottery and numismatic evidence from excavation and survey alike, this period is more of a continuation, politically and culturally (Fig. 5). It was however, the golden age of affluence in pre-Byzantine Iudaea and the culmination of the improving relations between the Jewish patriarchate and rabbinical establishment and the Roman authorities. The amiable relations between them are a frequent topic in rabbinical literature. They are being described under the title of “Antoninus and Rabbi”, Antoninus being one or more of the emperors of the Severan dynasty and the rabbi the Patriarch Rabbi Jehuda Hanassi or one of his successors (Krauss 1910; cf. Avi Yonah 1976: 40-42). Clearly, some of these encounters are invented, others stem from the contacts between the Patriarch and the provincial

Hadrian then based the permanent defence structure in the Negev on its more southern parts, and the Arava. It is not clear to me what the situation was in Transjordan (Map 4). 11 Miller 1969: 135-136, etc. Warmington 1974: 12, 16, 31, 47-48, 101, 138 and Raschke 1978 emphasize the C2nd shift to Egypt as trade outlet for Arabia. Raschke (p.648) mentions the Sassanian aggressiveness as reason for the strengthening of the trade route defences. Raschke 1978: 881, 936 warns against negating the land traffic. 12 Gichon 1982 details the strategic features of the Philistine coast. Gichon 1969: maps 16-21. Connected with it, maps 22-23 and 26-36, a cartographic representation of Iudaea-Palaestina as land bridge. The importance of the land bridge was reiterated in Parker 1997: 116. 13 In the case of the via nova Traiana, the discussion must be: what was its initial purpose and how far can, in this case, civil and military functions be separated. Personally, I hold to Parker’s view of a construction ab initio “for both commercial and military reasons” (Parker 1991: 499). In this respect it follows the Biblical “King’s Highway” and its Nabataean successor. All this may be compared to the trans-Negebite highways.

14 Out of the wealth of literature dealing with the inscription, see, for our purpose and without my need to go into controversial detail: Bowersock 1975: 513-522; 1983: 96-97; Graf 1978: 9-12; Parker 1986: 124; Sartre 1982: 27-29. 15 The author is most grateful to Y. Guvrin for his permission to reproduce the inscription and relate to part of his preliminary decipherment. A comprehensive paper on this subject is being prepared, together with the discoverer and P. Varon. The possible mention of a further unit, a ̀́ιλη or εί`λη (?) – line 3, is being examined.

194

Mordechai Gichon: 45 years of research on the limes Palaestinae

authorities. Yet, all these reflect a calm, peaceful and friendly atmosphere. The quest for national independence had been relegated to the distant messianic future. The famous synagogue inscription from Qasrin in Galilee, with its dedication to Septimus Severus and his family, recalls the days when the early emperors honoured the God of the Jews by their offerings and the daily sacrifice in the Temple of Jerusalem in honour of the emperor spelled the acceptance of Roman overlordship (L. Frey C. Inscript. Iud., no.927; Roth-Gerson 1987: 125-129). Significantly, this period was the peak of the first wave of major synagogue construction (NEAEHL IV, s.v. Synagogues). The acceptance of this political reality and its sociocultural aspects are nicely evidenced in the following rabbinical saying. Commenting on the Song of Songs, where beautiful Sulamith thrice exclaims: “I charge you” (2,7, etc.), Rabbi Jossi ben Hanina states: “why these three commands (pledges)? The first- (God’s admonition) that Israel shall not man the battlements (ie. take-up arms in war), the second- that they shall not revolt against the foreign nations (ie. Rome) and the third- God’s admonishing the foreign nations (Rome) not to subjugate Israel too heavily”(Talmud Bavli, Ketuvot 111: col.a).

garrison’s main function, aside from control and guard of the borders, must be envisaged as a strategic reserve on the Palestinian land-bridge, primarily vis-a-vis Persia, even from before the Sassanians. Hadrian’s retreat from Mesopotamia and the re-establishment of Parthia on the upper Euphrates brought the Persians within 180kms striking distance of Antioch. Though the emperor’s appreciation was that Parthia would keep her peace for the time being, he must, doubtlessly, have been aware of the potential danger on the Partho-Roman borders. Imperial strategy demanded the establishment of a strategic reserve in the rear and the maintenance of the strategic roads from Egypt and the Mediterranean coast for the moving of reinforcements. This basic condition for the imperial defence remained constant throughout the principate. A propos, we must rectify the wrong notion that Persia ceased at any time to be a potential threat to Rome and that defensive preparedness in the east could be intentionally neglected. Josef Wolski’s (1976) researches draw a fair picture. To conclude my observations on the earlier periods, I need point out that, in tune with the rabbinical sources and the Metar inscription, excavation proved that, at least for some time, some sites continued to be occupied on the old limes. The excavators of Arad and Beer Sheba have noted a continuing Roman occupation into the C3rd (Aharoni 1993; Herzog 1993).

This is only one of the many rabbinical sayings of the postBar Kokhba years that command “Israel not to revolt against the government and not to try and force the liberation to come, but in God’s own good time” (eg. Rabbi Levi, ibid.; Rabbi Levi and Rabbi Helbo in Canticus Cant. Rabba, c.2.17).

That climate explains also the continuation or revival of some kind of border militia manned by Jews, who, in Severan times, kept their weapons at home, as explicitly stated in the Tosephta Eruvin (cf. p.16 supra).

On the basis of the above and further evidence, we conclude that the Jewish leadership, following the debacle of 135, relegated quest for independence to messianic days some time in the future. From Pius’ days on, commenced a, on the whole, successful policy of peaceful acceptance of pre-Byzantine Roman rule, creating conditions for a by and large acceptable coexistence, in spite of intermittent maltreatment. Safrai (1971: 224-228), while underlining the, at times negative attitude of the army towards the Jews of vanquished Iudaea, relates many instances of a tolerable and even positive co-existence that developed over the years, and draws a balanced picture as to the situation. Not so Isaac (1991), who concentrates mainly on the negative evidence. Decidedly, the designation of the exercitus Provinciae as mainly an army of occupation cannot be applied to Iudaea in the latter half of the C2nd with more justification than for any other provincia armata. To be sure, this task has always existed.16 Yet, the Judaean

From Claudius Gothicus (?) to Constantine (Fig. 6) Scope restricts this paper to few remarks on the advent of the tetrarchy and after. The coins of the second floor of the burgus of Tsafit, excavated and published by me in 1972, cover the period from Claudius Gothicus to Carus, and the pottery there definitely is of the C3rd to early C4th type. The latest vessels tally with those of the earliest phase of reestablished Tamara, which too bore Illyrian emperors’ coins. Similar pottery and coins of Gallienus were excavated at Ma’aleh Mahmal (Gichon 1976; Cohen 1993e). Altogether, ‘Illyrian’ post-Gallenian coins were found in five of the 14 sites excavated in the Negev and the Arava (cf. Table 2 above). Add to this fact that Meshel established in his soundings at Yotveta a pre-Diocletian, C3rd low level, where one Elagabalan and one Proban coin were found, and that the

16

It would be foolish to deny that any foreign rule, throughout written history, has not produced incidents and rough handling of the native population. However, we have seen in the text above, that in the case of Iudaea, after 135, both sides endeavoured successfully to create an atmosphere to bring about a viable, peaceful modus vivendi. The success of these endeavours, until Byzantine religious zeal put them to nought, is the more to be appreciated when we notice, that in the case of the Jews, as a glance at Isaac 1991 will show, and probably also in the case of the Samaritans, the incidents revolved around interference in religious affairs. Isaac’s emphasis on the inner provincial rôle of the Roman – in this case Palaestina based forces - is to be accepted. But if not overemphasized, it does not rule out the frontier defence (Isaac 1990 and epilogue there). His

supposition that “…the military occupation of the desert and the defence of the settled lands were of marginal interest to the Roman authorities” (p.420) is however totally wrong. The more taxation and other vital economic interests played a part in the army’s presence in the provinces, the more was there need to safeguard the desert borders to ensure the reaping of the empire’s benefits from the amply shielded inland. Maintaining the pax Romana meant mainly mounting the defence on the borders.

195

Limes XVIII Beersabee is for typological reasons allocated to the C4th (Fabian 1995) but might also be earlier. The excavations at Malatha have now proceeded sufficiently to make its inclusion in the Diocletian defences certain. (cf. see above). Excavations at Tamara sustain the written evidence as well as those at Eisiba.19 Three castella must be added according to their excavation: Nessana (Negev 1993b), Yotveta (Meshel 1989) and En Boqeq (Gichon 1992).20 Two sites are doubtful: Mojet Avad=Moa (?), which so far has not yielded any post-Severan evidence by excavation but is mentioned in Byzantine documents (Alt, GIPT no.2) and on the Madaba Map (Avi Yonah 1954: no.22) and Masos, which is a Byzantine monastery, but could possibly be a C4th foundation (Fritz & Kempinski 1983). As far as the excavations on the top of Tel Ira went, my identification of that site as a major visual communication centre could not be proved, since the excavated part is monastic. However, the walls around the whole of the summit were rebuilt (Biran 1999: 173; Hershkowitz 1999: 126-127).21

inscription, deciphered by Roll, speaks of alam et ostium constituerunt (Meshel 1989; Roll 1989). This means that the tetrarchs added some kind of wing or portico and an entrance structure to the existing building. But it does not say that they, respectively the provincial authorities, erected the castellum proper, as some have repeatedly argued, though neither the language of the inscription nor the archaeological evidence justify this insistence.17 Taking all above indications together, there are now sufficient signs for definite refurbishing activity of the limes defences immediately after Claudius defeated Zenobia in 272, whose disruptive activities have been recorded east of the Jordan, prior to the Diocletian reforms (Bowersock 1983: 136; Millar 1993: 172). Their importance stems from the fact that in spite of all other pressing matters, the security on the negebite borders demanded immediate actions there. Tribal communities may, either in collusion with Palmyra or by themselves, have taken the initiative for extending their grazing ground or raiding northwards.

As already mentioned, Bowersock strongly opposed the designation of the Roman Negev as a limes (1976a: 221), followed by Parker (1986: 6, 142). Unfortunately, his negations are based on two wrong assumptions: (a) the Negebite defences are no other but “a fortified road across the northern Negev, and (b) “this road is well within the Roman Empire”.

I propose to assign to this time precisely the rabbinical ruling discussed by Rabbi Elazar and Rabbi Jehonatan, both flourishing in the second-third of the C3rd CE and after, that ordained the building of town walls not only on the borders but also inland (Talmud Bavli, Baba Bathra, p.76).18 Happily the fact of the existence of the DiocletianConstantinian border defences of what was to become in 357/8CE Palaestina Salutaris is at present not any more in doubt. According to the Notitia, the following Negebite sites belong to it (cf. Gichon 1967a): Maon (Or. LXXIII.19); Birsama (Or. LXXIII.22); Beersabee (Or. LXXIII.29); Malatha (Or. LXXIV.45); Tamara (Or. LXXIV.46); Eisiba (Or. LXXIV.42); Moahila (Or. LXXIII.29, unidentified). The castra discovered at

The relation of roads to border defence is obvious (see above) In the limes, any limes, they are integral and 19 Aharoni correctly identified Ḥazeva with biblical Tamara, but according to the absence of Byzantine pottery, wrongly - with the castellum bearing this rather common name (Aharoni 1963). Hieronymus, the Tabula Peutingeriana and the Beer Sheba Edicts all mention Tamara and make the occurrence of Byzantine pottery there a must (cf. Tsafrir, di Segni & Green Tabula, p.246, sv. Thamara). My contentions (Gichon 1967a; 1976: 80-81) to Ḥazeva’s being instead Eisiba, following Alt (1935: 31), are to my mind correct since the excavation (Cohen & Israel 1996), proved beyond doubt that the fortress did not continue into the C5th (p.32). Mezad Tamar’s continuing existence until the Muslim conquest, proven by pottery and coins (not yet finally published - so far: Gichon 1976) makes it the only site, according to the Madaba map and Eusebius’ quotation of distances (Gichon 1976: 80-81 and n. 4) that fits the location of Thamara. This identification was first suggested by Alt and accepted by Avi Yonah (ibid.). 20 En Boqeq’s chronology has been challenged by J. Magness (1999). Her numismatic arguments are not relevant, since in the sealed deposits only relevant to the period coins were discovered. The same holds true for her pottery criticism. In our stratification, documented in the 19 published section drawings, we differentiate between undisturbed (usually ground floor) layers, upper storey collapse and otherwise disturbed squares. Moreover, the quotation of sources, such as Hayes and Waage, as ultimate arbiters is, according to the latter’s own warnings, nonadmissible. See for instance Hayes 1972:138-139, 141! With the fullest appreciation of Dr. Magness’ work in the Jerusalem Ceramic Chronology, I consider the stratigraphy of any site built, as is the City of David, on prominent slopes, unsuitable for definite, generally binding chronologic conclusions. 21 Ira has the best visual communication with the limes sites and the rear. To my mind, this explains its choice as a central signal relay station, both for the early defences and the Diocletian ones and after (Gichon 1967a: 59ff.; 1968: 324). The extent of the excavations so far does not rule out this possibility. Besides, even a roof or a tower in the C5th monastery might have served as a speculum.

17 A similar inscription dated to Alexander Severus from the Hadrian’s Wall fort Axelodundum (Netherby) states: basilicam equestrem exercitatorium pridem a solo coeptam aedificavit… (Daniels 1978: 312314). The fort itself is Hadrianic. Although the building of its equestrian exercise hall had been conceived (and commenced?) earlier, it was carried out only then. It might be that the addition of the Yotveta fort also was the final construction of previously scheduled components; cf. Daniel 1978: 205, the Diocletian and Maximian inscription from Birdoswald. Though Yotveta speaks of construction (constituerunt) and Birdoswald of restoration (restituerunt), the latter is one more example for commemoration of internal building activities not being related to the first construction of the same general structure. 18 To the probable destruction or temporary abandonment of Tsafit in the course of the Zenobian conquest of Palaestina and Egypt (Gichon 1974b: 24), one may possibly add the end of the fortress of ‘Uza. The latest coins are Severan, but the pottery research has not yet established a clear dating of its abandonment (Beit Arieh 1986; 1993a). While the relations between Palmyra and the Banu Tanuḥ, the main allies of Rome on the Arabian borders were strained, those with the Jews were generally very positive. One may assume that the still largely Jewish limes hinterland of southern Iudaea was not affected aversely, though it seems that towards the end of the short, official Palmyrean overrule (267-271CE), the relations deteriorated (Altheim 1943: 126-127; Bowersock 1983: 132-136). Avi Yonah (1976: 126-127) even made a case for the Palestinian Jewry’s basic loyalty to the Roman government. Excavation in the hinterland may offer a clearer picture of these turbulent years.

196

Mordechai Gichon: 45 years of research on the limes Palaestinae

necessary component of the defence system. The peculiarity of the Negev is that its roads, since Old Testament times, have served at one and the same time as trade routes, that had to be guarded, and as connecting links between the sundry fortifications. The latter were vital for the control and apprehension of the desert-fringe nomads from intruding into Iudaea.

of course the year 324CE, when the see of Aila is mentioned in the list of the council of Nicaea as belonging to Palaestina.22 Whatever the exact date of the transfer of southern Arabia from the River Arnon to the Gulf of Aeleana-Aila (Elath/Aqaba), the main reason was strategic. Since it was contrary to Diocletian’s policy of diminishing the size of provinces, he and his staff, dealing with the administrative and military rearrangements, must have felt the following circumstances keenly and pressing (for Diocletian’s reorganisations: cf. Williams 1985).

As evident from Figs. 2, 3, 5 & 6 and Gichon 1999a: 244ff. and before, from Judaean-Nabataean times to those of the tetrarchy, the communication infrastructure in the Negev was indeed built around two to four main arteries and their connecting links. This is of course a zone and provides the need for depth (Gichon 1967a; 1971: 195), which Bowersock (1976: 228) also stresses.

Rome was at war with Narses of Persia, but even without open conflict between the super-powers, it was obvious that with the disappearance of Palmyra and the formation of strong tribal groups under Sassanian influence, all of the eastern borders were prone to attack. Diocletian’s awareness of this situation is echoed by his great enterprise of defence that remained alive in the memory of later generations. Ammianus explained these fortifications as means ne vagarentur per Syriam Persae (Res Gestae XXIII.5.1-5). Malalas emphasized its extent: απο της Αιγυπτουε εως των Περσικων ορων (Chron. XII.308). Ensslin’s (1942) exemplary study on Diocletian’s eastern policy is still largely relevant and should serve with William’s 1985 as background for our deliberations.

However, even if only one road had existed or even were we speaking of the most northerly axis Gaza-Malatha-En Boqeq, to declare this road as being “well within the Roman Empire” (Bowersock ibid.) means utter disregard of the most basic geopolitical quality of the Negev, described by me ad nauseam and never disclaimed: its being the border region of Iudaea, a no-man’s land for the sedentary population, whatever the actual trace of the boundary with usually Egyptian dominated Sinai, with which it merges topographically and climatically. For all practical purposes, the Negev was Iudaea’s border region, with all its implications, and only north of Beer Sheba are we permitted to speak of ‘inside’, though not yet ‘well inside’ the borders.

Indeed, the limes in southern Jordan faced the Hedjaz as well as menace from the north-east, and the one in the Negev faced Sinai and Egypt’s Arabian Desert. However, larger tribal movements, triggered off by droughts or events in the central and southern Arabian peninsula and on the confines of Persia, had to be envisaged.23 In this event, a break through the east of the Arava defences would result in penetration into western Palestine, and feasibly in attempts to invade Egypt. The reality of the latter threat is attested in written records twice: once by the Arab queen Mavia’s troops, in about 370CE (Sozomenos Hist. Eccles. XXXVIII) and once by the Sassanian king Kavad, after 485CE (Eutychios, p.191). The newly reorganized limes Palaestinae was thus single- and Janus faced at one and the same time (Gichon 1997: 25-27; Mayerson 1980).

Byzantine sources mention the ‘limes Palaestina’ several times, often in a general sense. But what are we to make of the following funerary inscription discovered by Alt in Beer Sheba?: Ουδε λιπων λιµµιιτοιο Παλεστινης χτονα διαν ∆ωροθεος γεραων αµµορος εκ βασιληος (SEG VIII: 296; Alt 1930: 43-44; cf. Rev. Bibl. 27: 121). I translate with Alt: “Even after departing from the divine country of the Limes Palaestinae, Dorotheus was not deprived of the royal emoluments”- to which one should add: “that are due to the limitaneus” – which was then selfunderstood and therefore omitted. Moreover, if the Codex Theodosianus VII.4.30 explicitly designates the garrisons of the castra Versaminium (Bersama) and Moenaenium (Maon-Menois) as limitanei milites, why should we know better ?

The emperor evidently strove for a permanent solution of 22 The subject is beyond the scope of this paper. For my opinions, including the general solution, see Gichon 1992: 22 and notes; Tsafrir 1983; Gutwein 1981; Dan 1982 and their sources relating to the subject. Just one remark: the complications caused by the seemingly late use of Arabia as home province for places that should already have belonged to “Palaestina”, respectively “Palaestina Salutaris” may well be explained by errors and lack of update information that persists. Similarily, long after Hadrian’s adoption of this name, “Palaestina” is, more than once, mentioned by its former name “Iudaea”. Writers that attempted at exactness mentioned both names, eg. Ptolemy, Jerome, Dio Cassius (see Gichon forthc.). 23 Parker 1991: 503 correctly distinguishes, following Luttwak, between ‘low intensity’ and ‘high intensity’ conflicts. While each of the Palaestina Salutaris limes sectors was built so as to deal with the ‘low intensity’ threats on its own, ‘high intensity’ threats were to be dealt with by both sectors in their combined function.

Only if one persists in the misconceived denial of the Negev being a veritable “frontier region”, to use Mayerson’s terminology, this and further literary evidence necessitate excuses and odd arguments to negate their being references to the Cisjordanian limes (eg. Bowersock 1983: 229). I adhere to the dating of the addition of southern Arabia to the provincia Palaestina, including the transfer of the legio X Fretensis to Aila, under Diocletian. This is still the opinion held by most scholars. The terminus ante quem is 197

Limes XVIII

border security after having stemmed the immediate largescale tribal harassing, inspired or positively looked-upon by Persia. Mamertinus and possibly Malalas mention his Arab campaign: The former declares: Laurea illa de victis accolentibus Syriam nationibus….fecerunt pio grandio triumphare (Paneg.XI (II) 7.1, p.280).

The new legionary dispositions and mainly the need of the legio X to answer not only threats from the Hedjaz but to operate in the Negev and Iudaea (western Palaestina), necessitated the improvement of the roads skirting the Arava. This was a condition sine qua non, if the limites to the east and to the west of the Arava were to operate as one. But this was initially accepted by scholars only as far as the lateral east to west communications were concerned (eg. Rothenberg 1971: 214). The strategic and tactical need for a north-south road was overlooked (eg. Parker 1986: 6). As to the west side, my assumption, which was based only on the existence of the military posts (from south to north) at Tsafit, Yotveta, Beer Menuḥa, Moa, ‛Hazeva, Mezad Hakikar (Gichon 1999a: map 1-3) and possibly further ones, is now sustained by the discovery of the remains of a road and three groups of milestones pertaining to Diocletian, Constantine and Licinius (Roll et al forthc.). They skirt the west of the valley up to about 70kms from the Gulf shores. Its link-up with the via nova, the camp at Aila and the Darb el Hadj in Sinai has not yet been established. From this axis, the ascents west- and northwestwards, through Yotveta, Moyet Avad (Moa) and Ḥazeva (Eisiba?) were easy.25

The strategic circumstances thus explain the need for unified command in Palaestina, but why not continue as before with the Negev attached to Arabia? The necessity of including the Judaean hinterland to increase the effectiveness of the defence of this province was probably one of the considerations, since it was a vital rear area for mounting major offensive and defensive activities on the eastern frontier. However, the main reason was certainly logistic. The agriculturally considerably less endowed Arabia must have found it increasingly difficult to sustain the troops stationed on the limes, the more so since the upheavals following the Severan period caused the system of interprovincial supply and transport to grow more and more inefficient. When Diocletian transferred the legio X to Aila (ie. not less than 6000 men including soldiers’ families and civilian services), the logistic up-keep of the military was much easier when being based on Palaestina. There, the Judaean mountains, with their typical terrace- and valley agriculture, had for two centuries, served as the legion’s supply base (territorium).

With only the written evidence from the researches of Musil and Nelson Glueck at my command, I have not argued with equal insistence my ideas about the existence of an eastern Arava road. Today, it suffices to pursue the chapter ‘The Wadi Araba’ in Kennedy (2000: 193-204), to make this a high probability. Not only the forts, etc., but also some vestiges of roads (Kennedy 2000: 193-203) are being indicated (between the Arava castella) and they appear to be strongholds sited on three east-west roads.

If the legio VI Ferrata was transferred to Adroa/AdrouUdruh, (as assumed by many scholars following the suggestion by Speidel), parallel to the one of the X Fretensis, and since, according to the Notitia Dignitatum, the legio IV Martia was based on Betthorus-Lejjun, Arabia had to sustain three legions, in addition to its ‘home’ garrison, the legio III at Bostra (Speidel 1979: 172. n.11; Parker 1986: 142).

I would like to conclude this review with correcting an error of mine and commenting on the problem arising. Excavation has proved that both Ḥazeva and Yotveta were destroyed early (?) in the second half of the C4th. Though I myself have gathered from my survey in the 1950s that the former did not continue into Byzantine times, I have assumed the existence of the military (limitanean) presence in the western Arava at least up to Justinian. At the present state of our knowledge, I see I have evidently erred as far as the west is concerned. The survey in the east however points at a continued use of the eastern route (Kennedy 2000).

This was impossible to achieve from the local sources. The 15-20,000 mouths to be fed would have to rely on outside sources. Continuing the basing of the former Judaean legions on their existing supply bases and developing additional resources in Palaestina for the newly created legio IV was the natural answer. All this was best achieved by making the provincial authorities directly responsible.24 24 The major efforts involved in feeding the army and the reliance as much as possible on provincial resources are aptly summed up by Remesal-Rodriguez 1990 and his sources, Junkelmann 1997: 84-85, etc. and Davies 1989: 187ff. The latter stresses, as one major source, the “civilians” and as the other one “food grown on military land, the territoria and prata.” Remesal has proved that the provisioning of forces concentrated temporarily for war or the like in one region fell to the local procurator Augusti. The burden and difficulties involved must have caused the authorities to do their utmost to secure the supply for the permanent garrisons from local provincial or nearby sources. To my mind, (cf. Junkelmann 1997), the introduction of the limitanean system was to a considerable extent based on these considerations. According to the Diocletian and Constantine reforms, the responsibility for the provisionment of food for the permanent forces fell on the duces, who had no authority on the non-military food suppliers (Van Berchem 1952: 2223), which caused frequent friction (Grosse 1920: 158-159). This is

The destruction levels at Ḥazeva and Yotveta have been associated with the earthquake of 363CE (Cohen & Israel apparent also from the Beer Sheba edicts, whatever their exact meaning (Van Berchem 1952: 33). Since the saltus on the limes Palaestinae were situated within the provincia, they could serve the troops with only one civilian authority to deal with (Avi Yonah 1949: 156). The same applies to the fundi limitrophi, which must have existed here as elsewhere (Van Berchem 1952: 101). After the creation of Palaestina Salutaris (II) as a separate border province, the rank of the governor of that province was beneath the one of Palaestina I and this strengthened the duces’ position vis-a-vis that official and enabled him to appeal to the latter. 25 I take the opportunity to thank Prof. Roll for his kind permission to quote his M.S.

198

Mordechai Gichon: 45 years of research on the limes Palaestinae

1996; Meshel 1989: 238; Russell 1980. The havoc wrought at ‛En Boqeq: Gichon 1992: 52). I propose to add, for further consideration, an alternative explanation: the uprising of Mavia in the reign of Valens (cf. Meshel 1989). Reminding ourselves of Procopius’ dictum: “The Saracens are by nature incapable of conducting (regular) sieges” (Wars II. XIX.12), it could even be that those castella may not have been captured, but only their wooden ceilings etc. were burnt and that they gradually fell in, after having been evacuated or abandoned. For one or another reason, that could have happened even after peace was made and Mavia took upon herself federative service on the limes Palaestinae. From her days onward, the western Arava, much of the inner Negev and Sinai will have been under the permanent control of the Palestinian phylarch. Regular forces, always a financial burden and never sufficient, were reduced to a few east-Arava bases.26 Incidentally, copper mining and production ceased in the western Arava in Byzantine times (Gichon, apud Rothenberg, in print), while these activities continued in Phaeno (Faynan) in the east. The road(s) on that side may have served “industrial” traffic not less than military needs.

Gichon 1967a, some others in later publications. Bibliography Aharoni M. 1993 Arad. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land I. (Jerusalem): 85. Aharoni Y. 1963 Tamar and the roads to Elath. Israel Exploration Journal 13: 30-42. Aharoni Y. 1967 Forerunners of the Limes. Iron Age fortresses in the Negev. Israel Exploration Journal 17: 1-17. Aharoni Y. 1973 Beer Sheeba I. (Tel-Aviv). Alföldi A. 1941 Epigraphica IV. Arch. Ertesitö III.2: 40-45. Allg. Militär Encyclopädie 1870 Verein Deutscher Offiziere. Allg. Deutsche Militärgeschichte 1-10, 1868-73. (Leipzig). Alon D. 1986 Horvat Tsalit. Hadashot Arkheologiyot 88: 61-62 - in Hebrew. Alon D. 1987 The Nahal Yatir site. In A. Kloner & Y. Tepper (edd.) The hiding complexes in the Judean Shephela. (Tel-Aviv): 154-159 - in Hebrew. Alt A. 1921 Die Griechischen Inschriften der Palaestina Tertia, westlich der Araba (GIPT). (Berlin). Alt A. 1930 Limes Palaestinae I, II. Palästina Jahrbuch 26: 43-82 Alt A. 1931 Limes Palaestinae III. Palästina Jahrbuch 27: 7584 Alt A. 1935 Aus der ‘Araba II, III. Zeitschrift des DeutschenPaläestina Vereins 58: 1-77 Alt A. 1955 Neue Untersuchungen zum limes Palaestinae. Zeitschrift des Deutschen-Palästina Vereins 71: 82-94 Altheim F. 1943 Die Krise der Alten Welt III, Götter und Kaiser. (Berlin). Applebaum S. 1961/1962 The initial date of the limes Palaestinae. Zion 26: 1-10 - in Hebrew. Ashkenazi T. 1938 Tribus Semi-Nomades de la Palestine du Nord. ( Paris). Avi Yonah M. 1949 The historical geography of Eretz Israel from the return from the Babylonian exile to the Arabian conquest. (Jerusalem) - in Hebrew. Avi Yonah M. 1954 The Madaba Mosaic Map. (Jerusalem). Avi Yonah M. 1958 The date of the Limes Palaestinae. Eretz Israel V: 135-137 - in Hebrew. Avi Yonah M. 1976 The Jews of Palestine. (Oxford). Avni G. 1982 Har Sagi. Hadashot Arkheologiyot 80/81: 65-66. Baly D. 1957 The geography of the Bible. (New York): 65-66. Baras Z., Safrai S., Tsafrir Y. & Stern M. 1982 Eretz Israel of the 2nd Temple to the Muslim conquest. (Jerusalem) – in Hebrew. Barouch Y. 1995 The Roman castles in the Hills of Hebron. In Z.H. Ehrlich & Y. Eshel (edd.) Judea and Samaria Research Studies 4: 137-143 – in Hebrew. Barouch Y. 1997 Road studies in Judea during the 2nd Temple Period. In Y. Eshel (ed.) Judea and Samaria Research Studies 6: 125-135 – in Hebrew. Beit Arieh Y. 1986 Horvat ‘Uzza, a border fortress in the eastern Negev. Qadmoniot 73/74: 31-39 – in Hebrew. Beit Arieh Y. 1993a ‘Uzza. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land IV. (Jerusalem): 1485-1488. Beit Arieh Y. 1993b Malhata, Tel. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 936. Beit Arieh Y. 1998 The excavations at Tel Malhata – an interim report. Qadmoniot 115: 30-39 – in Hebrew. Beit Arieh Y. 1999 Tel ‘Ira – A stronghold in the Negev.

Post Scriptum Neither fortification patterns nor the limitanean establishment are being discussed in these pages. If I am right, the fact of the Roman take-over of the eastern courtyard castellum in the east, as evidenced in Iudaea ab initio and its spread westwards to become the prevalent late Roman castellum pattern (Gichon 1967a; 1967b: 182183) is not any more in doubt. Yet this phenomenon has not received the attention it deserves. The occurrence of Severan pottery on agricultural limes sites (Gichon 1967a: 184ff.) has convinced me of the fact that the SHA’s assignment of the limitanean establishment to Severus Alexander (Sev. Alex. 58.4) has a historical kernel. This is so, if we take the term limitanei to mean farmers, who, under a certain administrative arrangement within the general tactical planning of the border-area defence, were scheduled with defending sua rura. The Roman authorities might feasibly have been influenced by the long-standing proto-limitanean tradition of the Judaean confines, respectively its latest link - the village defence, that survived from Herodian into Roman times (cf. supra: 15ff.). These arrangements did precede the transformation of the entire border forces into farming militia units. Note to the maps a. Preliminary excavation reports lacking profiles and pottery drawings make the allocation of sites to ceramic transition periods frequently uncertain. This situation is indicated by question marks. b. The selected unexcavated sites, indicated, are identified by pottery from survey only. Most of them are described in 26

Mavia: Gichon 1992: 23, Mayerson 1980 and their sources.

199

Limes XVIII

(Jerusalem). Biran A. 1993 Aroer (in Iudaea). In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land I. (Jerusalem): 83-93; Herodian period: 91-93. Biran A. 1999 Tel ‘Ira, the Byzantine period. In Y. Beit Arieh Tel ‘Ira – a stronghold in the Negev. (Jerusalem): 173. Bowersock G.W. 1975 The Greek Nabatean bilingual inscription at Ruwwafa, Saudi Arabia. In J. Bingen et al (edd.) Le Monde Grec: Homages à Claire Preaux. (Brussels): 513-522. Bowersock G.W. 1976 Limes Arabicus. Harvard Studies in Classical Philology 80: 219-229. Bowersock G.W. 1983 Roman Arabia. (Cambridge, Mass.). Butzer K.W. 1957 Der Umweltfaktor in der grossen Arabischen Expansion. Saeculum 8: 359-376. Cohen R. 1983 En Tamar. Hadashot Arkheologiyot 83: 68 - in Hebrew. Cohen R. 1993a Mezad Dafit. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1135. Cohen R. 1993b Mezad Menuḥa. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1135. Cohen R. 1993c Mezad Ḥazaza. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1136. Cohen R. 1993d Mezad Yeroham. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1136-1137. Cohen R. 1993e Mezad Ma‛aleh Mahmal. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1140. Cohen R. 1993f Mezad Neqarot. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1140. Cohen R. 1993g Mezad Sayif. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1141. Cohen R. 1993h Mezad Zafir. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1143-1144. Cohen R. 1993I Horvat Zafir. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1143. Cohen R. 1993j ‛En Raḥel. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1143. Cohen R. 1993k Rogem Zafir. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1144. Cohen R. 1993 Qazra. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1144. Cohen R. 1993m Moa. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1137-1140. Cohen R. 1993n Sha‘ar Ramon. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1145. Cohen R. & Israel Y. 1996 ‛En Ḥazeva – 1990-1994. Excavations and surveys in Israel 15: 110-116. Dan Y. 1982 Palaestina Salutaris (Tertia) and its capital. Israel Exploration Journal 32: 134-137. Daniels C. 1978 Handbook to the Roman Wall. (Newcastle).

Davies R.W. 1989 Service in the Roman army. (Durham). De Haas J. 1934 History of Palestine. (New York). Elder I. & Baumgarten I. 1993 Byzantine Malhata. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 937. Ensslin W. 1942 Zur Ostpolitik des Kaisers Diokletian. (München). Eutychios Eutychios, Annales. (ed. Cheiko, Corpus Scriptorum Christianorum Orientalium III, VII). Evenari M., Shaanan L. & Tadmor L. 1971 The Negev – the challenge of a desert. (Cambridge). Fabian P. 1995 The late Roman military camp at Beer Sheba: A new discovery. In J. Humphrey (ed.) The Roman and Byzantine Near East. Some recent archaeological research. Journal of Roman Archaeology Suppl. Series 14: 235-240. Forni G. 1967 Limes. In Dizionario Epigraficodi di Antichità Romane vol. IV, fasc. 34-40. (Roma): 1959-62. Frank F. 1934 Aus der Araba I. Zeitschrift des DeutschenPalästina Vereins LVII: 191-280. Fritz V. 1973 The Roman fortress. In Y. Aharaoni (ed.) Beer Sheba I. (Tel-Aviv). Fritz V & Kempinski A. 1983 Ergebnisse der Ausgrabungen auf der Hirbet el Msas 1972-1975. (Abh. des Deutschen Palästina-Vereins). (Wiesbaden). Galestin M.C. 1997 Romans and Frisians. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 347-353. Geyer P. Itinera Itinera Hierosolymitana, Saeculi IV-8 (CSEL XXXVIII). (Prag, Wien, Leipzig 1888 = Johnson 1964). Gichon M. 1967a The limes in the Negev from its foundation to Diocletian times. (Dissertatio Hierosolymitana, H.U. Jerusalem) - in Hebrew. Gichon M. 1967b The origin of the Limes Palaestinae and the major phases in its development. In H. Schönberger (ed.) Studien zu den Militärgrenzen Roms. (Köln): 175193. Gichon M. 1967c Idumea and the Herodian Limes. Israel Exploration Journal 17: 27-42. Gichon M. 1968 Das Verteidigungssystem und die Verteidiger des flavischen Limes in Judäa. In E. Schmidt, L. Berger & P. Bürgin (edd.) Provincialia: Festschrift für Rudolf Laur-Belard. (Stuttgart): 317-334. Gichon M. 1969 Carta’s Atlas of Palestine from Bethther to Tel Hai – a military history. (Jerusalem) - in Hebrew. Gichon M. 1971 The military significance of certain aspects of the Limes Palaestinae. In S. Applebaum (ed.) Roman Frontier Studies 1967. Proceedings of the VIIth International Congress. (Tel-Aviv): 191-200. Gichon M. 1974a Towers on the Limes Palaestinae – form, purpose, terminology and comparisons. In D.M. Pippidi (ed.) Actes de IXe Congrès d’Etudes sur les Frontières Romaines, 1972. (Bucharest): 513-544. Gichon M. 1974b Migdal Tsafit, a burgus in the Negev. Saalburg Jahrbuch 30: 16-40. Gichon M. 1976 Excavations at Mezad Tamar-Tamara. Saalburg Jahrbuch 33: 80-94. Gichon M 1980 Research on the Limes Palaestinae – a stocktaking. W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies XII. British Archaeological Reports Int. Ser. 71. (Oxford); 843-864. Gichon M. 1982 The history of the Gaza Strip – A geopolitical and geostrategical perspective. In L.E. Levine (ed.) The Jerusalem Cathedra 2.

200

Mordechai Gichon: 45 years of research on the limes Palaestinae

Gichon M. 1986 Who were the enemies of Rome on the Limes Palaestinae?. In C. Unz (ed.) Studien zu den Militärgrenzen Roms III. (Stuttgart): 585-592. Gichon M. 1991 When and why did the Romans commence the defence of southern Palestine? In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989 Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 318-325. Gichon M. 1992 En Boqeq – Ausgrabungen in einer Oase am Toten Meer I: Geographie und Geschichte der Oase. Das Spätrömische Kastell. (Mainz). Gichon M. 1996a Israel Shatzman – The armies of the Hasmoneans and Herod – Review Article. Jewish Quarterly Review 86: 21-26. Gichon M. 1996b Sallies from Iotapata and their designation as “Lestrike”. In J. Mangas & J. Alvar (edd.) Homenaje a José Maria Blazquez III. (Madrid): 75-86. Gichon M. 1997 The strategic conception and the tactical functioning of the Limes Palaestinae after the Diocletian reorganisation. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 24-31. Gichon M. 1999a Developments in the research on the Limes Palaestinae during the last two decades. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman FrontierStudies 1997. (Zalău): 241-250. Gichon M. 1999b The hinterland of the Limes Palaestinae – remains and written evidence. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman FrontierStudies 1997. (Zalău): 667-688. Glueck N. 1953-60 Explorations in the Negev, I-IX. Bulletin of the American Schools of Oriental Research 131; 137; 138; 142; 145; 149; 152; 155; 159. Graf D.F. 1978 Saracens and the defence of the Arabian frontier. Bulletin of the American Schools of Oriental Research 229: 1-26. Graf D.F. 1989 Rome and the Saracens – reassessing the nomadic menace. In T. Fahd (ed.) L’Arabie Preislamique et son environnement historique et culturel, Actes Coll. Strassbourg 1978. (Leyden): 341400. Graf D.F. 1997 The via militaris and the Limes Arabicus. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 123-134. Grosse R. 1920 Römische Militärgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung. (Berlin). Gutwein K.C. 1981 Third Palestine. (Washington). Guvrin Y. 1996 Townlets of Burgin at Droma. In Y. Eshel (ed.) Judea and Samaria Research Studies 1995: 183194 – in Hebrew. Guvrin Y. 2000 A dedication inscription of a Roman soldier of the 10th legion from the northern Negev. In Y. Eshel (ed.) Judea and Samaria Research Studies 9: 121-128 – in Hebrew. Haefeli L. 1938 Die Beduinen von Beer Sheba. (trad. et ed., Luzern). Hayes J.W. 1972 Late Roman pottery. (London). Helk W. & Otto E. 1956 Kleines Wörterbuch der Aegyptologie. (Wiesbaden).

Hershkowitz M. 1999 The Hellenistic pottery. In Y. Beit Arieh Tel ‘Ira – A stronghold in the Negev. (Jerusalem): 126-127. Herzog Z. 1993 Beer Sheba. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land I. (Jerusalem): 173. Hirschfeld Y. 1979 A line of Byzantine forts along the eastern highway of the Hebron Hills. Qadmoniot XII: 78-84 – in Hebrew. Isaac B. 1984 Bandits in Judaea and Arabia. Harvard Studies in Classical Philology 88: 171-203. Isaac B. 1988 The meaning of ‘limes’ and ‘limitanei’ in ancient sources. Journal of Roman Studies 78: 125-147. Isaac B. 1990 The limits of the empire. The Roman army in the east. (Oxford). Isaac B. 1991 The Roman army in Iudaea, police duties and taxation. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies. Proceedings of the XVth Interrnational Congress of Roman Frontier Studies 1989. (Exeter): 458-462. Janon M. 1991 Remarques sur la Frontière de Numidie. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies. Proceedings of the XVth International Congress of Roman Frontier Studies 1989. (Exeter): 482-484. Junkelmann M. 1997 Panis Militaris. (Mainz). Kaminer O. 1983 Mizpeh Sayarim map survey. Hadashot Arkheologiyot 83: 79 Kennedy D. 2000 The Roman army in Jordan. (London). Klein C 1986 Fluctuations of the level of the Dead Sea and climatic fluctuations in Israel in historical times. Dissertatio Hierosolymitana. (Jerusalem). Kochavi M. 1993 Malhata, Tel. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 936. Konrad M. 1999 Frühkaiserzeitliche Kastelle in der nordsyrischen Wüstensteppe? In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 523-532. Krauss S. 1910 Antoninus und Rabbi. (Wien). Magness J.H. 1993 Jerusalem Ceramic Chronology. JSOT/ASOR Monograph Series 9. (Sheffield). Magness J.H. 1999 Redating the forts at En Boqeq, Upper Zoar and other sites in S.E. Judaea, and the implications for the nature of the Limes Palaestinae. In J.H. Humphrey (ed.) The Roman and Byzantine Near East II. Some recent arcaheological research. Journal of Roman Archaeology Suppl. Series 31: 189-206 Mayerson P. 1980 Mauia, Queen of the Saracens – a cautionary note. Israel Exploration Journal 30: 123-131. Mayerson P. 1986 The Saracens and the Limes. Bulletin of the American Schools of Oriental Research 262: 35-47. Mayerson P. 1989a Saracens and Romans. Micro-macro relationships. Bulletin of the American Schools of Oriental Research 274: 71-77. Mayerson P. 1989b The meaning of the word Limes (ΛΙΜΙΤΟΝ) in the papyri. Zeitschrift für Papyrologie und Epigraphik 77: 287-291. Meisler B./Mazar 1949/1950 The Rekem and the Heger. Zion 20: 316-319 – in Hebrew. Meshel Z. 1989 A fort at Yotveta from the time of Diocletian. Israel Exploration Journal 39: 228-238. Millar F. 1993 The Roman Near East. (Cambridge, Mass.). Miller J.I. 1969 The spice trade of the Roman empire 29 BC – AD 641. (Oxford).

201

Limes XVIII

Moodie A.E. 1947 Geography behind politics. (London). Musil A. 1907 Arabia Petraea II, Edom 1,2. (Wien). Negev A. 1967 Oboda, Mampsis and the Provincia Arabia. Israel Exploration Journal 17: 46-55. Negev A. 1993a Elusa. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land I. (Jerusalem): 379-382. Negev A. 1993b Nessana. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1145. Negev A. 1993c Oboda. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 1155-1165. Negev A. 1993d Horvat Sussia. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land IV. (Jerusalem): 1416. Negev A. 1993e Kurnub. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land III. (Jerusalem): 884-893. Oleson J.P. 1999 Preliminary report of the Al-Humayma excavation project 1995, 1996, 1998. Annual of the Dept. of Antiquities of Jordan 43: 411-447. Oren E. 1993 Sera, Tel. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land IV. (Jerusalem): 1569-1570. Parker S.T. 1986 Romans and Saracens – A history of the Arabian frontier. (Winona Lake). Parker S.T. 1991 The nature of Rome’s Arabian frontier. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies. Proceedings of the XVth Internationmal Congress of Roman Frontier Studies 1989. (Exeter): 498-505. Parker S.T. 1997 Geography and strategy on the southwestern frontier in the late Roman period. In W. Groenman-van Waateringe, B.C. Van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 115-122. Perigrinatio Aetheriae Die Pilgerreise der Aetheria. (H. Pétré ed. et comment., K. Vreska trad., Klosterneuburg 1958). Petrie W.M.F. 1930 Beth Pelet I. (London): 20-21. Prescott J.R.V. 1965 The geography of frontiers and boundaries. (London). Ptolemaio Claudius Ptolemaios, Geographia. (Müller ed., Paris 1880). Raschke M.G. 1978 New studies in Roman commerce with the east. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.9. 2. (Berlin): 546-1361. Remesal-Rodriguez J. 1990 Die Procuratori Augusti und die Versorgung des Römischen Heeres. In H. Vetters & M. Kandler (edd.) Akten des 14 Internationalen Limeskongresses 1986 in Carnuntum I. (Wien): 55-66. Roll I. 1989 A Latin inscription from the time of Diocletian, found at Yotveta. Israel Exploration Journal 39: 239260. Roth-Gerson L. 1987 The Greek inscriptions from the synagogues in Eretz Israel. (Jerusalem – in Hebrew). Rothenberg B. 1971 The ‘Arabah in Roman and Byzantine times in the light of new research. In S. Applebaum (ed.) Roman Frontier Studies 1967. Proceedings of the VIIth International Congress. (Tel-Aviv): 211-223. Russell K.W. 1980 The earthquake of May 19, AD 363. Bulletin of the American Schools of Oriental Research 238: 4764. Safrai S. 1971 Relations between the Roman army and the Jews.

In S. Applebaum (ed.) Roman Frontier Studies 1967. Proceedings of the VIIth International Congress. (TelAviv): 224-230. Safrai Z. 1994 Relations between Jews and Arabs in the southern Hebron Mountains. In Z. Safrai & Y. Levin (edd.) Settlement and security in the southern Hebron Hills in ancient times. (Ramat Gan): 30-50 – in Hebrew. Safrai Z. 2001 Jewish law in the Judean Desert documents. In H. Eshel & B. Zisru (edd.) New studies on the Bar Kokhba Revolt. (Ramat Gan): 115-118 – in Hebrew. Sartre M. 1982 Trois Etudes sur l’Arabie romaine et byzantine. Collection Latomus 178. (Bruxelles). Schleiermacher W. 1959 Der römische Limes in Deutschland. (Berlin). Schur N. 1993 The Golan dominated by the Anazeh tribes as described by travelers of the 17th-19th centuries. In I. Degani & S. Imbar (edd.) Golan Heights and Mount Hermon. (Tel-Aviv): 309-17 – in Hebrew. Shatzman I. 1983 Security problems in southern Iudaea after the Great Revolt. Cathedra 30: 3-32 - in Hebrew. Shatzman I. 1991 The armies of the Hasmoneans and Herod. (Tübingen). Sidebotham S. 1991 A limes in the Eastern Desert of Egypt: myth or reality? In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies. Proceedings of the XVth International Congress of Roman Frontier Studies 1989. (Exeter): 494-497. Speidel M. 1977 The Roman army in Arabia. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II.8. (Berlin): 687-730. Speidel M. 1979 A tile stamp of cohors I Thracum Milliaria from Hebron in Palestine. Zeitschrift für Papyrologie und Epigraphik 29: 170-172. Sperber D. 1978 Roman Palestine 200-400 – the land. (Ramat Gan). Striem H.L. 1985 Quantitative and qualitative aspects of the recent climatic fluctuations. Israel Journal of Earth Science 34: 47-48 Tsafrir Y. 1983 Why were the Negev, southern Transjordan and Sinai transferred from provincia Arabia to provincia Palaestina ? Cathedra 30: 35-56. Tsafrir Y., Di Segni L. & Green J. 1994 Tabula Imperii Romani – Iudaea-Palaestina. (Jerusalem). Van Berchem B. 1952 L’Armée de Dioclétien et la Réforme Constantinienne. (Paris). Villeneuve F. 1989 Ad-Diyatheh: Village et castellum romains et byzantines à l’est du Jebel Druze. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297. (Oxford): 119-140. Warmington E.H. 1974 The commerce between the Roman empire and India. (London). Wells C.M. 1991 The problem of desert frontiers. In V.A. Maxfield & M.J. Dobson (edd). Roman Frontier Studies. Proceedings of the XVth International Congress of Roman Frontier Studies 1989. (Exeter): 478-481. Williams S. 1985 Diocletian and the Roman recovery. (London). Wolski J. 1976 Iran und Rom. In H. Temporini & W. Haase (edd.) Aufstieg und Neidergang der Römischen Welt II.9.1. (Berlin): 195-214. Yisraeli Y. 1993 Far‘ah, Tel, south. In E. Stern et al (edd.) New encyclopedia of archaeological excavation in the Holy Land II. (Jerusalem): 443-444.

202

Mordechai Gichon: 45 years of research on the limes Palaestinae

Fig. 1. The Herodian limes.

Fig. 2. The Trajanic limes. Sites nos. 5, 10, 11, 13, 15, 17, 19, 21 and 30, and possibly more, existed in Flavian times.

203

Limes XVIII Fig. 3. Trajan’s border arrangements between Iudea and Arabia.

Fig. 4. The Hadrianic limes (after the Bar Kokba War).

204

Mordechai Gichon: 45 years of research on the limes Palaestinae

Fig. 5. The Severan limes.

205

Limes XVIII

Fig. 6. The C4th limes. Sites nos. 1, 3, 6, 19 Beersabee, Cermela (and Moahila – unidentified) are mentioned as troop bases in the Notitia Dignitatum (late C4th-C5th).

Fig. 7 206

Excavations at the Roman siege complex at Masada – 1995 Haim Goldfus & Benny Arubas In the course of summer 1995, new archaeological excavations were launched at the Roman siege works at Masada. The excavations were carried out as part of a larger development plan destined to prepare the site for tourism in the 3rd millennium. As part of the first stage, a systematic yet intentionally restricted set of archaeological excavations were carried out in the Roman headquarters (Camp F), and at the Roman siege ramp that was built against the western cliff of Masada. The excavations were directed B. Arubas, G. Foerster and H. Goldfus on behalf of the Hebrew University of Jerusalem and J. Magness of Tufts University. All in all, we unearthed 11 architecturally defined units in Camp F. These units included what seems to be the masonry foundation of the tent of Flavius Silva (the commander of the Roman siege force), a small platform (probably, the commander’s tribunal), several dwelling units (apparently, of infantry as well as cavalry officers); a row, made of several units, for ordinary legionaries, and a conicshape oven built of small field stones (one of three detected). The dwelling units of both officers and ordinary soldiers were equipped with triclinia, as well as with cooking corners. Among the special artifacts found in these units was the complete cheek-piece of a soldier’s helmet, several fine painted Nabataean bowls, luxury glass vessels, and a number of copper alloy decorated horses’ pendants. Coins found in Camp F are still under study. However, it is worth noting that unlike the numerous coins from the Jewish War found at Masada (more than 1600 coins), only one such coin was found in Camp F. Interestingly, this coin was hidden under a large sherd of a cooking pot, and was imbedded into the floor of one of the legionaries’ tents. Maybe, it was used as an amulet by one of the Roman soldiers. We performed two kinds of archaeological examinations on the Roman siege ramp: a tier trench on the south slope (digging down the incline of the ramp), and a deep vertical pit on the north slope of the ramp. It was revealed that pieces of wood were placed to form a sort of retaining wall, or as we called them ‘retaining baskets’ for rocks and earth. The soft rock quarried from the knoll at the bottom of the ramp was dropped onto the slope of the natural outcrop of the rock, upon which the ramp was built. Once a ledge had been formed by these rocks, a shallow platform of branches and some stems were laid horizontally to form a level path for workers and to support the next loads of rocks which might have otherwise collapsed.

Preface

Camp F

A development plan for tourism in the 3rd millennium at the well-known archaeological site of Masada, enabled us to conduct new excavations at parts of the Roman siege system laid out around Masada in AD73/74.1 This siege was the last substantial military campaign of the Romans against Jewish rebels during what has been called the Jewish Revolt.

This camp is one of the two larger Roman siege camps. It is located north-west of Masada. The camp has a parallelogram shape and was built on a slope moderately inclining from west to east. The camp consists of Camp F1, the original camp, and Camp F2 built within the southwestern quarter of F1 (Fig. 1). While the larger camp apparently served as the Roman headquarters during the siege of 73-74, the smaller camp, F2, was most likely built to house a small Roman garrison left at Masada following its fall. In building this later camp the builders made use of part of the existing southern and western walls of Camp F1 and its western gateway. They also dismantled parts of the building stones of Camp F1, and some of its buildings in order to build the eastern and northern walls of F2. The inner plan of the later camp was also altered and adjusted to the refurbished circuit wall and the size of the remaining force (cf. Yadin 1967: 43-45).

The Roman siege works surrounding the fortress of Masada included 8 military camps (numbered in modern literature by the letters A to H) linked by circumvallation, and a siege ramp laid against the western cliff of Mt. Masada. The only contemporary description of the siege was written by Josephus Flavius the Jewish historian of the Great Jewish Revolt (Bell. Jud. VII.304-311). Its remains were surveyed and described by several modern scholars (Brünnow & v. Domaszewski 1909: 221-235; Hawkes 1929: 195-213; Schulten 1933: 1-179; Richmond 1962: 42155; Yadin 1966: 209-231). Small-scale excavations were carried out in Camp A by S. Gutman (1965: 88-133), and in Camp F by Y. Yadin (1966: 218ff).

Bearing in mind the importance of Camp F, we decided to conduct limited excavations in both the earlier and later camps (Fig. 2). Beyond the obvious need to produce more accurate and up to date plans of some of the principle buildings in both camps, we also wanted to verify the function of the archaeological units we dug, as well as to learn more about the material culture of the Roman army

The present paper is a preliminary description of the excavations conducted at several locations in Camp F, and at two soundings on the northern and southern slopes of the siege ramp.2

of Camp F; Gabi Laron was in charge of photography. Students from Tufts University, Northwestern University, Northwestern College, Wyoming, and Bedouin workers, performed the hard labour in extreme desert condition. The work was initiated and financed by the National Parks Authorities and the Governmental Company for Tourism. Since the interpretation of the finds, and the concluding remarks, were not discussed with our co-directors, we are solely responsible for any faults or errors in this paper.

1

For this and other possible dates of the supposed year of the siege, see Cotton & Geiger 1989. 2 Benny Arubas, Gideon Foerster and Haim Goldfus on behalf of the Hebrew University, and in collaboration with Jodi Magness of Tufts University, directed the excavation carried out during June and July 1995. Doug Nelson of Northwestern College, Wyoming, supervised the dig at the siege ramp; Oded Ron inspected excavations in Units VII and XXIV

207

Limes XVIII

in this part of the world within a very well defined and datable span of time3; that is to the year 73/74. In addition we hoped to trace and study the minute changes that occurred in the camp following the departure of the major Roman force after the fall of Masada.

eastern wall of the later F2 camp directly on top of the western side of this unit. The stones of the other walls of this unit were apparently used too in the construction of that later wall. Unit XXII was located north-east of the former one. The limited preservation of its walls like the previous unit can also be attributed to its proximity to the eastern wall of the later Camp F2. It has a rectangular shape and can be divided into three distinctive parts or wings (Fig. 3). The northern part consisted of three compartments within which were found several small installations and many hearths or small cooking spots. The relatively large amount of pottery sherds, among them several painted Nabataean bowls and cooking pots, suggest that this part of the unit was the service wing. Two spacious triclinia comprised the south wing, which was obviously used for dining and sleeping. A narrow corridor separated these two wings. It led to the western, third wing of the unit which consisted a single broad space which had a sort of a raised platform in its north-west corner. The function of this part of the structure is not entirely clear. In any event, the rich small finds which included substantial amounts of fragments of colourless blown glass vessels characterize this unit as belonging to officers rather than to ordinary legionnaires. Prof. D. Barag of the Hebrew University identified these vessels as beakers that belong to the facet-cut and similar expensive luxury wares. The beakers, according to him, were not produced in Syrio-Palestine but may have originated from Alexandria. Prof. J. Price of Durham University, England, on the other hand, suggested that they originated in north-western Europe.5 These drinking vessels are among the earliest examples of such glassware from precisely datable contexts.

At this point, it would be useful to comment on the materials used for the construction of the units we dug since they were all built in the same manner. In fact, it is also has implications for the other 7 siege camps. All the structures and units were built of medium and large size fieldstones of the local hard limestone. Dry soil was added, probably to help level the courses of the fieldstones. As far as we can judge there was no bonding material used in the construction of walls. The first courses of the walls were also laid directly on top of the natural terrain without any special preparation, such as foundation trenches. In only one case were the inner walls plastered (Unit III). In most places the height of the walls did not exceed 1m. This conclusion is based on the amount of debris over the walls. Once we cleared that debris the occupation level was reached within not more than 20cms. As has been previously suggested, we are convinced too that the walls served as foundations for the tents stretched above them (Richmond 1962: 146). It is worthwhile emphasizing that the nature of dry construction on top of natural bedrock made it difficult to discern the earlier plan and constructions of Camp F1 at its south-west corner, since the construction of the later camp, on top of it, left almost no traces of the earlier camp. Camp F1 Four units were entirely excavated at the centre of this camp (Units XXI-XXIV). Towards the end of the season we began to unearth two more units in the north-western corner of the camp. These, however, are not included in this report (Units XXV-XXVI).4

In the light of the location and inner plan of this unit, and the rich finds unearthed in it we are inclined to identify this unit as the living quarters of the high functionaries of the Roman force at Masada.

Unit XXI, preserved only to its foundation courses, was located at the centre of the camp, and is shaped like the Hebrew letter ‘Het’. It was preceded by a wide and shallow stepped terrace, which led to the central open space at the centre of the camp (Fig. 3). Here, the via principalis met the via praetoria. This centrally located unit with wide walls, and of a shape which reminds one of a naos of a temple, to use Josephus’ words when he describes the tent of the Roman commander (Bell. Jud. III.82), might be identified as the tent of Flavius Silva, the commander of the Roman siege. However, since there was no triclinium in this unit, we should not ignore the possibility that this structure was indeed the camp temple, and the place for keeping the insignia of the legion. The poor state of preservation of this unit is largely due to the building of the

In close proximity to the south-east corner of this unit we cleared the debris of Unit XXIII, which consisted of a raised 3 x 3m solid structure (Fig.3). A slopped ramp at its western side (opposite the commander’s tent and next to the above mentioned, open space) indicate that this small unit was most likely the tribunal, ie. the podium upon which the commander of the Roman force stood when he addressed his troops. Unit XXIV is located to the north of the three units we have just described, and overlooked the northern entrance of Camp F1 (Fig. 4). The structure consists of two spacious triclinia built back-to-back and some cooking installations. Crescent-like stone fences extended from the corners of the southern triclinium to form semi-open courtyards around it. Since we found some horse pendants in this unit, we suspect that these semi-open spaces might have been used as stables for horses. The relatively elaborate unit is one of

3

A preliminary study of the pottery retrieved in the course of the dig, was made by J. Magness and a report delivered at the Congress. Prof. Magness plans to publish the results of her readings elsewhere. 4 The numbering of the units in Roman numerals is ours.

5

208

We would like to thank both D. Barag and J. Price for their comments.

Haim Goldfus & Benny Arubas: Excavations at the Roman siege complex at Masada – 1995

several such structures found in various parts of the camp at the end of rows of less elaborate structures, which can be identified as contubernia, the mess-units of the ordinary legionaries. Having said that, we postulate that the present structure was occupied by some of the centurions of the Roman force. Another possibility is that it was occupied by cavalry officers. However, the use of such mounted forces in a harsh terrain like that which surrounds Masada is doubtful.

Unit III is located near and to the east of Unit I. It comprises a plastered room in front of which is a semiopen courtyard (Fig. 3). This unit clearly incorporates two construction phases. While the later use of this unit can be probably assigned to the post-siege stage, in its initial stage, it was apparently part of the praetorium or principia, that is, part of the siege headquarters complex. One could clearly discern that in the first stage unparalleled attention was given to a solid construction, a quite precise square ground plan, and plastering of the inner as well as the outer face of walls of the unit. Unfortunately, we do not have any leads for the exact function of this unit in this phase. It is, however clear that the entrance to the unit was blocked, at its north-east corner, by debris and packed earth without any small finds. In the second stage the floor level was raised and another room and a forecourt was add to its north. The ceramic finds at this stage included a substantial quantity of sherds of jars and cooking pots.

Unit XXX is located near the western gate of the camp. It consisted of three small identical installations which we identified as cooking stoves. We excavated the northern one (Unit XXX/1), a nicely built, small structure of spherical shape, partially dug into the soil. A dark ashy layer found inside the structure as well as an ash layer mixed with some animal bones was found around this installation. Whether this unit served as a common kitchen of Camp F1 or F2 cannot yet be ascertained.

The last element to be excavated in Camp F2 was Unit VII (Fig. 5). Being the second largest unit in the late camp, clearly a central building at this stage, it was possibly used as a dwelling unit of the officers, or the commander, of the garrison. It was entered from the east side, opposite the possible headquarters, Unit I. A narrow corridor divided the unit into two sub-units. The northern part was probably used as a living space. It was entered from the east through a narrow space, which seems to have served as a kitchen. The southern sub-unit was, maybe, used as a storage space. To the south of this sub-unit there is a rounded closed stone fence, which could have been used for animals of burden.

Camp F2 As mentioned before, this camp was constructed within the south-western precincts of Camp F1. Here we were able to unearth entirely four architectural units. Unit I, located in the centre of the smaller camp, consisted of 10 compartments. We could clearly discern two stages of construction originating in a central core measuring 8m x 8m, two long rooms in the centre, a fore-room to the west, annexed rooms, one of which is a triclinium, benches and ledges and stone fences (Fig. 5). It is possible that in building this unit the builders incorporated some architectural elements that belonged to the construction phase of the earlier Camp F1. Because of the location of this unit in the centre of the smaller camp, its complexity, and the relatively rich finds in it, as opposed to the other units in this camp, we are inclined to identify the structure as the headquarters unit or the commander’s structure of the garrison left behind after the capture of Masada.

The siege ramp We conducted two soundings at the siege ramp in order to study its components, infrastructure, and the exact way it was laid. One of the soundings was performed at the southern slope of the ramp, and the other one in its northern slope. Unfortunately, logistic difficulties prevented us from reaching the bedrock upon which the ramp was laid. Nonetheless, we were able to learn that the Romans made use of two components when it came to building the ramp – local raw materials and common sense. The raw materials were trees growing around Masada and soft limestone quarried from the hill, which existed to the west of the ramp. The use of wood was not systematic nor was it extensive. At any rate it seems to have been used more intensively in the lower or bottom parts of the ramp. Sparse layers of Tamarisk branches and short stumps of palm trees were laid in horizontally, one above the other at irregular intervals. At times the branches were inserted diagonally or vertically to the plane of the horizontal natural layers (Fig. 7). To the best of our understanding the use of this technique was meant to secure and stabilize the mass of local limestone and soil, poured on top of the natural outcrop of the rock jutting out from the mountain of Masada.

Unit II is situated near and to the north of the headquarters building. It consisted of a line of six individual sub-units all of which had an identical plan – a triclinum preceded by a forecourt (Fig. 6). In each of the forecourts we found cooking spots and body sherds, principally of cooking pots. Finds within the triclinia were scarce. This unit comprised a series of single contubernia used as dwelling structures for the ordinary legionary. Scholars studying the siege works at Masada have taken it for granted, based on Ps. Hyginus, that each contubernium accommodated 8 men. However, based on the dimensions of any of the contubernia exposed at Masada, we cannot imagine that more than two men, less likely three, occupied a mess-unit at the same time. When assessing the capacity of such unit, one also has to take into account, the space taken by the weaponry, other equipment and personal baggage of each soldier. This fact has, of course, important implications for the size of the Roman force present at Masada.

209

Limes XVIII

Unaware of the work of the geologist, Dan Gil, (1993: 569-70) who conducted a geological examination of the ramp area, we carried out our own independent study, and came to the same conclusion: the siege ramp was, to a large extent, a natural spur. The Roman legionnaires only added boulders and meagre amounts of tree branches to the crest of the natural spur, thus forming a smooth ascent towards Masada. Whether the ramp was ever used for an assault, intended to overtake Masada, is an intriguing issue. This, however, will be treated in a separate article.

element, were abutted by another inner wall incorporating stairways and watchtowers as well. It is intriguing that the only other camp with such similar defensive arrangements is Camp A (as can be seen on Schulten’s plan). Does this phenomenon imply that Camp A was also occupied by a garrison left at Masada after its fall ? If this is true, then it strengthens our case for the existence of two, almost autonomous forces at Masada.

Summary and primary observations

Brünnow R.E. & von Domaszewski A. 1909 Provincia Arabia III. (Strassburg): 221-244. Cotton H.M. & Geiger J. 1989 The Latin and Greek Documents. In J. Aviram, G. Foerster & E. Netzer (eds.) Masada II: The Yigael Yadin excavations, 19631965. Final report. (Jerusalem): 21-23. Gil D. 1993 A natural spur at Masada. Nature 364: 569570. Gutman S. 1965 With Masada. (Tel Aviv): 88-133 - in Hebrew. Hawkes C. 1929 The Roman siege of Masada. Antiquity 3: 195-213. Richmond I.A. 1962 The Roman siege works of Masada, Israel. Journal of Roman Studies 52: 142-155. Schulten A. 1933 Masada Die Burg des Herodes und die römischen Lager. Zeitschrift des Deutschen Palästina-Vereins 56: 1-179. Yadin Y. 1966 Masada: Herod’s fortress and the Zealot’s last stand (London): 209-231. Yadin Y. 1967 Masada and the limes. Israel Exploration Journal 17/1: 42-45.

Bibliography

The excavations at Camp F enabled us to examine thoroughly several components of the Roman siege works at Masada. The first and immediate benefit of the excavation was the production, for the first time, of an accurate, and up to date set of plans of some of the more important structures of the Roman headquarter siege camp. The archaeological data gathered in the various units enabled us to assess more accurately the function of some of them, and question previous assumptions, which were based primarily on surveys and analysis of air photographs. An examination of the units at the centre of the earlier F1 camp, showed that it would be difficult to define the principia and praetorium of the Roman camp. Possibly, these two terms had little meaning in a siege camp. It is also possible that the components and functions of these units were reduced, or combined into much smaller and less defined units. A glimpse into the living spaces of the officers, and more particularly, of that of the ordinary legionnaires, brings us to suggest that the Roman siege force at Masada was probably smaller than that suggested by most scholars. Another observation, which is not necessarily based on any new finds, is that the Roman presence during the siege comprised two separate forces. We can hardly envision a possible efficient collaboration between the Roman forces stationed at the foot of the eastern and northern sides of Masada, and those forces set along the western and southern, higher terrain around Masada. The presence of two identical headquarters at the centre of both of the larger siege camps, B and F, suggest a duality of functions. The presence of two commander’s tribunals, previously identified by Schulten, supports, in our mind, such an assumption. If there was one force at Masada there seems to be no logical reason to build two headquarter camps and two tribunals. Our supposition is that Flavius Silva was the chief commander of the siege operation. However, another high imperial official or senior officer performed an independent military routine and duties at the eastern siege front of Masada. Once the siege was over, a small unit, maybe a century, was left behind at Masada. This force utilized part of the available constructions of Camp F1, and established itself in the south-west corner of that camp. To complete its defence, new eastern and northern walls were built and a gate added at the centre of the east wall. In addition, the western and southern wall, an original 210

Haim Goldfus & Benny Arubas: Excavations at the Roman siege complex at Masada – 1995

Fig. 1. General Plan of Camp F – the black lines mark the walls of F2 and the excavated units.

211

Limes XVIII

Fig. 2. General view of the excavated units of Camp F as seen from Masada.

Fig. 3. Plan of Units I (F2), XXI-XXIII (F1).

212

Haim Goldfus & Benny Arubas: Excavations at the Roman siege complex at Masada – 1995

Fig. 4. Plan of Unit XXIV.

Fig. 5. Plan of Units I & VII.

213

Limes XVIII

Fig. 6. Plan of Unit II.

Fig. 7. Photo of the construction of the siege ramp.

214

Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia Israel Roll The Jordan Valley, which makes a rather small segment of the giant graben generally known as the Syro-African Rift, stretches up to 170kms from the southern foothills of Mt. Hermon to the northern shore of the Dead Sea. This wide and deeply entrenched depression, although bounded by high cliffs and by steeply dissected hilly terrain, was not considered in historical times as a valuable military barrier nor as a suitable political border. On the other hand, it did function as a major zone of movement which permitted the channeling of a main north-south thoroughfare along it, as well as its crossing by a series of transverse roads that extended from west to east. Roman imperial road building started in the region west of the Rift - that is in the province of Judaea later named Palaestina - in early Flavian times, as attested to by the earliest dated milestone found on the Caesarea-Scythopolis transverse highway. More highways were built in later stages and the province was gradually covered with a dense road network of c.1000 Roman miles. East of the Rift, following the Trajanic annexation of the Nabataean kingdom and the formation of the provincia Arabia, a first class strategic highway was erected across it, from Bostra to Aila. In time, this road was to become the backbone of the south-eastern Roman frontier zone and of its communication system. The integration of the road networks of the two provinces into one regional traffic system became a necessity and the Romans responded to it by building 6 connecting traffic lines across the Jordan Valley. In principle, all those arteries included a bridge across the River Jordan and an engineered ascent that surmounted each one of the two fault scarps at the Rift's sides. The 6 arteries belonged to the following highways: 1. Tyrus-Paneas-Damascus; 2. Ptolemais-Julias-Bostra; 3. Ptolemais-TiberiasGadara-Bostra; 4. Caesarea-Scythopolis-Pella-Gerasa; 5. Caesarea-Neapolis-Coreae-Philadelphia; 6. Jaffa-Jerusalem-Jericho-Esbus. Some of the main visible remains of the 6 crossing arteries will be discussed in more detail in my presentation.

The giant graben, generally known as the Syro-African Rift, which extends from the southern border of Turkey to the east African state of Malawi, is one of the largest and certainly the deepest scars in the earth’s solid crust. Of its overall length, that exceeds 6500kms, our concern here is with a rather small segment of it, that is, the Jordan Valley, stretching up to 170kms, from the southern foothills of Mt. Hermon to the northern shore of the Dead Sea (Schattner 1962; Orni & Efrat 1971: 80-105; cf. Fig. 1). This unique creation of Nature continually amazes travellers who cross it and scholars that describe it (Lynch 1849; Ritter 1866 II: 160-356; MacGregor 1870; Abel 1933 I: 161-178, 474506; Glueck 1946; Baly 1974: 193-216; Kareem 2000: 527). In antiquity, Pliny the Elder (NH V.15.71) depicts the River Jordan in a rather idyllic manner as “…a delightful stream, winding about so far as the conformation of the locality allows, and putting itself at the service of the people who dwell on its banks, as though moving with reluctance towards that gloomy lake, the Dead Sea, which ultimately swallows it up”(for other sources, see: Wilkinson 1977: 162-163 s.v. “Jordan”; Tsafrir, Di Segni & Green 1994: 71 s.v. “Aulon”, 153-154 s.v. “Iordanes flumen”).

On the other hand, it was the same George Adam Smith who realized that this wide and deeply entrenched depression, bounded by high cliffs and by steeply dissected hilly terrain, was not considered in historical times as a valuable military barrier, nor as a suitable political border. That was the case at least until WWI when, apparently, developments in modern warfare and the new political order that followed, impelled to fundamental changes on the matter. But, let’s go back to antiquity. To the series of historical events listed by Smith (1966: 316-318), that emphasize the non-essential importance of the Rift Valley as a frontier, I would like to add two more examples, which may also introduce us to the subject ahead. First, the Decapolis, that league of 10 cities probably established by Pompey in 63BC; 9 of them were located east of the River Jordan, while the tenth city – that is Scythopolis, which was certainly not the lowest in importance – stood west of the river (Bietenhardt 1977; Schürer II 1979: 125-158; Graf 1992; Isaac 1998). Second, the district of Peraea, that extended further to the south and east of the River Jordan; this region belonged in the Second Temple period to the Hasmonaean state and to the Herodian kingdom, and afterwards, to the Roman province of Judaea – later named Palaestina (Avi Yonah 1977: 96-97; Schürer 1979 II: 1013, 194-195). Consequently, as the Romans got control over both sides of the Rift Valley, there was no place to make it a political border, nor a military frontier. On the other hand, the Rift did function as a major zone of movement. Its long north-south valley permitted the channeling of a main longitudinal thoroughfare along it (Thomsen 1917: 73; Dalman 1935: 233-239). The two sides of the Rift, although high and steep as they usually are, still allowed the alignment of a series of transverse routes that extended from west to east (Fast 1956; Noth 1956; see Fig. 2).

In the late C19th, George Adam Smith (1966: 301-302) was much more emotional in his expressiveness when describing the Jordan Valley. He wrote “…in its own way the Jordan is as solitary and extreme an effect of natural forces. There may be something on the surface of another planet to match the Jordan Valley: there is nothing on this. No other part on our earth, uncovered by water, sinks to 500ft below the level of the ocean. But here we have a rift more than 100 miles long and 2 to 15 broad, which falls from sea-level to as deep as 1300ft below it at the coast of the Dead Sea, while the bottom of the latter is 1300ft deeper. …Is it not true that on earth there is nothing else like this deep, this colossal ditch?”

During the early phase of their domination in the east, the Romans seem to have channeled their interest mainly

215

Limes XVIII

towards the longitudinal north-south road. In 63BC, during his campaign against the Hasmonaean state, Pompey and his army marched all along the lower Jordan Valley, on his way from Damascus to Jerusalem (Josephus Ant. Jud. XIV.3.4: 48-54; Bell. Jud. I.6.5-6: 133-139). In AD68, during the suppression of the Great Jewish Revolt, Vespasian and his army campaigned along the southern part of this road, from Coreae to Jericho (Bell. Jud. IV.8.12: 449-458). Roman imperial road building started in Judaea during the suppression of the First Jewish Revolt, as attested to by the earliest dated milestone, erected in AD69, on the Caesarea-Scythopolis transverse highway (Isaac & Roll 1976; 1982: 66). More highways were built in Judaea later on, until the whole province was covered with a dense network of c.1000 Roman miles (Roll 1983; 1994; 2000). East of the Rift Valley, and following the formation of the provincia Arabia by Trajan in 106, a first class strategic highway was erected across it, from Bostra to Aila (Brünnow & v. Domaszewski 1905 II: 19-124; Graf 1995; Bauzou 1998). In time, this artery was to become the backbone of the south-eastern Roman frontier zone and of its communication system. As a result, the imperial arteries of traffic west of the Rift, became the rear road network for the frontier zone that evolved east of it (Roll 1989: 252-260; 1999). The integration of the two networks into one unified regional traffic system became a necessity and the Romans responded to it by building six transverse connecting roads across the Jordan Valley. Their main components were an engineered paved ascent that surmounted each one of the two fault scarps at the Rift’s side, and a bridge across the River Jordan. Let us have a short overlook on those highways, and on the main visible remains related to the crossing of the Rift.

the foothills of Mt. Hermon until Damascus (Fig. 6). Of this road, only a few remains are still visible, except the bridge that spanned over the local stream south of Paneas (Wilson 1880: 114-115). The Tyrus – Paneas – Damascus highway is fully depicted on the Tabula Peutingeriana (Weber II 1976: Seg. IX; see Fig. 7), which indicates that it had a major rôle in the communication network of the Roman orient. Among the few milestones found so far along it, the one that was published at the time by Cagnat (1936), is of great importance because it mentions two of its capita viarum, that is, Tyrus and Paneas, and the distance of 5 miles from the former. During our on-going research project on the Roman roads in lower Galilee, carried out by the Israel Milestone Committee team of the Tel Aviv University (which includes Ben Isaac and myself, assisted by S. Weingarten, Y. & I. Tepper, Y. Shachar and others), we surveyed a section of an ancient west-east road. The section, first brought to our attention by Zvi Ilan (1989-90), extends for c.10kms, from the River Amud to Almagor, above the northern end of the Sea of Galilee. Its eastern part is well preserved and it shows methods of road-making similar to those we described: angled alignment of short stretches, bounding of pavement with one or two lines of curbstones, and the addition of retaining walls to keep its surface level (Fig. 8). The road crossed two small but potentially damaging streams near Almagor, on corbelled viaducts built of hewn basalt slabs, which were recorded by Yosef Stepansky (1997). The continuation of the road has not been preserved in any of its two directions and no milestones were found along it. However, as the road shows familiar Roman road building, its capita viarum might be sought for in Roman cities located much farther, to the west and to the east. The best candidates in this case are Ptolemais at one end, and Julias and even Bostra, at the other. If so, we are in the presence of a rather straight westeast transverse highway that extended from Ptolemais eastwards, along the Beth Kerem Valley and along our section, to Julias (Schumacher 1889; Schwöbel 1904: 7374; La Jonquière 1905 IV: 355-361; Israeli 1977; for a different view, see Tepper et al 2000: 109). From there it had to climb to the Golan Heights and it did so possibly along the southern cliff of River Daliot or, more likely, along the Lawiya spur, where a Roman road is well preserved (Maoz 1987: 64-65; see Fig. 9). Then, it would have continued towards the south-east, over the preserved Roman bridges that spanned the streams Rukkad and Allan (Merrill 1881: 324; Schumacher 1886: 76-79), until it reached Bostra. It is worth noting that this highway and the next one, connected the same two cities – Ptolemais and Bostra – but, the artery we described was clearly shorter in mileage.

The Tyrus – Paneas - Damascus highway: it departed from Tyre, mounting the hills in the east, settled along the range that dominates the southern bank of the River Litani and, after crossing the deep gorge of the Wadi Dubbeh over a bridge at Qanntara, it reached the watershed of Mt. Naftali near Misgav-Am (Conder & Kitchener 1881 I: 106; Schwöbel 1904: 66-67; Smith 1966: 278 n. 2). From there, the road took the shape of a series of serpentines aligned along the top of continuous descending spurs, until it reached the bottom of the Rift Valley, south of Metulla. Several segments of this engineered ascent have been preserved, and I was able to record them during a survey carried out in the area by Moti Avi ‛am (1988-89). The preserved stretches show technology of road making which is typical to the Romans: alignment on the basis of short straight segments, joining each other in an obtuse angle; road-bounding, c.7m apart, with a single or a double row of large curbstones; the terracing of steep terrain by means of supporting walls on one side and rock cuttings on the other (Figs. 3 & 4). In the valley, no remains of the road have been preserved, except one arch of an initially threearched bridge over the river Jordan’s main tributary – the Hatsbany (MacGregor 1870: 221; Wilson 1880: 103; Conder & Kitchener 1881 I: 115; see Fig. 5). From Paneas, the highway ascended to the Golan Heights along the southern hill-side of River Sa’ar and then continued along

The other highway stretched from Ptolemais towards south-east, to Sepphoris, and from there eastwards, to Tiberias. East of Sepphoris, the remains of the Roman road are still visible, until the Golani junction (Fig. 10). There, a section was cut through the road, which showed three stratified building phases: a lower one – that belonged to 216

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia its initial construction, and two upper phases – that testify to the road repair; each of the three building phases consist of two layers – roadbed and pavement (Roll 1986; 1995; see Fig. 11). As the earliest milestone found on the Ptolemais-Sepphoris road dates to the time of Hadrian (AD120; see Isaac & Roll 1979), which seems to be the date of its construction, the same might be inferred for the Sepphoris-Tiberias highway, and even for its continuation further to the east. A milestone found at the Golani junction, which includes a molding between base and pillar that is typical to Hadrianic stones, supports this date (Fig. 12). The highway seem to have descended to Tiberias along the spur leading to Mt. Berenice, and from there it ran southwards, until the outlet of the Sea of Galilee into the lower Jordan. The road crossed the river over a bridge, of which remains of up to ten arches could still be seen until the mid-C19th. It was named “the great bridge” in a late Byzantine source (Taktikon 1864: 380-381) and Umm el Qanatir (“Mother of Arches’) later on by the Arabs (MacGregor 1870: 402-404; Merrill 1881: 139; Abel 1933 I: 163, 480-481; Ben Arieh 1965). Apparently, this structure seem to have been the longest and the most important bridge over the River Jordan in Roman times, but today, no traces of it have survived. It should be added that the two bridges that crossed the River Jordan in recent times – the Jisr Benat Ya’kub north of the Sea of Galilee and the Jisr el Mujamieh at the south of it – date to medieval and later periods and show no traces of antiquity (Conder & Kitchener 1881 I: 206-207; 1882 II: 116; Abel 1933 I: 162-165). The continuation of the highway to the east has been surveyed during the 1960s by Mittmann (1970: 133-150; 1999) and most recently by Riedl (1999). According to their results, the route climbed along an engineered ascent to Gadara, and from there it continued as a ridgeway, between the basins of the River Yarmuk and the Wadi el Arab, to Capitolias (Fig. 9), and then, after crossing the Wadi Shellala, it reached Adraa and Bostra. The eastern part of this highway – from Tiberias to Bostra – appears on the Tabula Peutingeriana (Fig. 7). The entire road served as one of the two main connecting arteries between the cities of the Decapolis and the Mediterranean Sea. The other connecting artery was the next road.

Mittmann (1970: 152-159). The Caesarea – Neapolis – Philadelphia highway. This road departed from Caesarea and crossed the Sharon plain towards south-east and then, followed all along the wide river-bed of Nahal Shechem, until Neapolis (Roll 1996: 557-558). From there, until the Jordan Valley, long segments of the paved Roman road have been preserved. These include: the crossing of the plateau east of Neapolis in a straight line (Fig. 13), until the beginning of the hilly terrain; the following of the river-bed of a wadi which traverses that terrain (Fig. 14); the crossing of a series of north-south hilly ranges at a saddle; and finally the descent by means of serpentines towards Coreae below in the Rift Valley (Ilan & Damati 1974-75). The highway crossed the River Jordan over a bridge, some remains of which could still be identified in the late C19th in the medieval structure of Jisr ed-Damieh (Merrill 1881: 422-424; Steuernagel 1927: 347-349; see Fig 15). There was certainly an engineered ascent that mounted towards Philadelphia but, its alignment has not yet been identified. The Jaffa – Jerusalem – Jericho – Esbus highway. There were several connecting highways between the chief coastal cities of Roman Palestine and Jerusalem, which departed from Caesarea, Ascalon and Gaza. But, the most important among them were clearly the two arteries between Jaffa and Jerusalem, published most recently by Fischer, Isaac and myself (1996). The road from Jerusalem to Jericho, although only 18 miles in length, extended along an extremely rough and abrupt terrain in the Judean desert. Hence, a remarkable feat of Roman road engineering was needed there to cross it (Beauvery 1957; Wilkinson 1975). The artery departed from the Damascus Gate towards north-east, passed over the saddle south of Augusta Victoria and continued along the northern gradient of a descending range of low hills until its end (Fig. 16), above Qasr Aly. There, the remains of an impressive engineered ascent are still visible, which include: a paved ramp bounded by rock-cuttings at one side and a retaining wall on the other; stretches of leveled rock combined with pavement into the road’s surface; rock-cut steps, as well as built steps, distributed along the steeper curves of a running serpentine (Fig. 17). Further on, the road winded along the upper gradient of a succession of disconnected hills, until the crowning site of Tala’at ad-Damm, where the military post of Maledomnei was located in late antiquity (Tsafrir, Di Segni & Green 1994: 176). The descent from this site to the east has not been preserved, but further on, the remains of the Roman road appear again, two milestations included, along the southern ridge that bounds the Wadi Qelt, until its end above Jericho. Of the Roman artery between Jericho and Livias, and of its bridge over the River Jordan, no traces have survived. On the other hand, some segments of the engineered ascent from Livias to Esbus are rather well preserved and their remains and numerous milestones have been published most recently by Ibach (1994: 68-75) and by Piccirillo (1996; see Fig. 18).

The Caesarea – Scythopolis – Gerasa highway. The Valley of Jezreel, which crosses Judaea from west to east, certainly provides an ideal terrain for a transverse thoroughfare and the Romans indeed erected along it a major highway, from Legio to Scythopolis (Isaac & Roll 1982; Roll 1996: 557; Moorhead 1997: 130-132). East of Scythopolis, the scarp that usually crowns the Rift Valley is almost non-existent. Hence, there was no need for an engineered ascent, but just for a straight road, to Pella, and a bridge over the River Jordan. These items have not been preserved, but, the straight alignment of the road is attested by three milestations found along it. The difficult problem concerning the mounting of the road from Pella to southeast has been discussed recently by Frank Koucky (1992) of the Pella Joint Expedition. The continuation of the highway towards Gerasa and its numerous milestations were treated previously by Thomsen (1917: 65-67) and by 217

Limes XVIII

In sum, the six transverse highways that crossed the Rift Valley provided a series of engineered traffic lines, fit for all seasons, which connected between the Mediterranean coast and the south-eastern frontier zone of the Roman empire, as well as between the road networks of the two provinces – Judaea/Palaestina and Arabia.

für Papyrologie und Epigraphik 33: 149-156. Isaac B. & Roll I. 1982 Roman roads in Judaea I: The LegioScythopolis road. British Archaeological Reports Int. Ser. 141. (Oxford). Israeli A. 1977 Biq’at Beit-Kerem: an old-new lateral thoroughfare. Teva va-aretz 19: 122-129 – in Hebrew. Kareem J.M.H. 2000 The settlement patterns in the Jordan Valley in the mid- to late Islamic period. British Archaeological Reports Int. Ser. 877. (Oxford). Koucky F.L. 1992 The environs of Pella: roads, fords and occupational sites. In A.W. McNicoll et al. Pella in Jordan II. (Sydney): 199-204. La Jonquière C. de 1899-1907 L’Expedition d’Égypte (17981801). I-V. (Paris). Lynch W.F. 1849 Narrative of the United States’ expedition to the River Jordan and the Dead Sea. (London). MacGregor J. 1870 The Rob Roy on the Jordan. (New York). Maoz Z. 1987 The Golan from the Second Temple to the Roman-Byzantine periods. In M. Inbar & E. Schiller (edd.) Ramat Hagolan (Ariel 50-51). (Jerusalem): 6068 - in Hebrew. Merrill S. 1881 East of the Jordan. (London). Mittmann S. 1970 Beiträge zur Siedlungs und Territorialgeschichte des nordlichen Ostjordanlandes. (Wiesbaden). Mittmann S. 1999 Römerstrassen in Nordwestjordanien und ihr Nachleben in der Kreuzfahrzeit. Zeitschrift des Deutschen Palästina-Vereins 115: 24-44. Moorhead T. S. N. 1997 The late Roman, Byzantine and Umayyad periods at Tel Jezreel. Tel Aviv 24: 129-166. Noth M. 1956 Der Jordan in der alten Geschichte Palastinas. Zeitschrift des Deutschen Palästina-Vereins 72: 123148. Orni E. & Efrat E. 1971 Geography of Israel. (3rd ed. Jerusalem). Piccirillo M. 1996 La strada romana Esbus-Livias. Liber Annuus 46: 285-300. Riedl N. 1999 Eine neu entdeckte Meilensteingruppe in Nordwestjordanien. Zeitschrift des Deutschen Palästina-Vereins 115: 45-48. Ritter C. 1866 The comparative geography of Palestine and the Sinaitic peninsula, I-IV (Transl. by W. L. Gage: Edinburgh). Roll I. 1983 The Roman road system in Judaea. In L. I. Levine (ed.) The Jerusalem Cathedra 3: 136-161. Roll I. 1986 Walking along a Roman road in lower Galilee. In M. Yidayiah (ed.) The Western Galilee Antiquities. (Tel Aviv): 297-303 - in Hebrew. Roll I. 1989 A Latin imperial inscription from the time of Diocletian found at Yotvata. Israel Exploration Journal 39: 239-260. Roll I. 1994 Roman roads. In Y.Tsafrir, L. Di Segni & J. Green Tabula Imperii Romani: Iudaea-Palaestina. Maps and Gazeteer. (Jerusalem): 21-22 and the attached maps. Roll I. 1995 Survey of Roman roads in lower Galilee. Excavations and surveys in Israel 14: 38-40. Roll I. 1996 Roman roads to Caesarea Maritima. In A. Raban & K.G. Holum (edd). Caesarea Maritima: A retrospective after two millennia. (Leiden): 549-558. Roll I. 1999 The roads in Roman-Byzantine Palaestina and Arabia. In M. Piccirillo & E. Alliata (edd.) The Madaba Map Centenary 1897-1997. (Jerusalem): 108-113. Roll I. 2000 Ancient roads of the Land of Israel in the light of air photography. In B.Z. Kedar & A. Danin (edd.) Remote sensing. The use of aerial photography and satellite images in Israel studies. (Jerusalem): 131-141 – in Hebrew.

Bibliography Abel F.-M. 1933-1938 Géographie de la Palestine. I-II. (Paris). Avi’am M. 1988-89 Horvat Nuha – ancient road. Excavations and surveys in Israel 7-8: 143. Avi-Yonah M. 1977 The Holy Land from the Persian to the Arab conquest (536BC to AD 640). A historical geography. (Rev. edit. by A. F. Rainey: Grand Rapids). Baly D. 1974 The geography of the Bible. (New and revised edition: New York). Bauzou T. 1998 Le secteur nord de la via nova en Arabie de Bostra à Philadelphia. In T. Bauzou et al. Fouilles de Khirbet es-Samra en Jordanie I. (Turnhout): 101-255. Beauvery R. 1957 La route romaine de Jérusalem à Jéricho. Revue biblique 64: 72-101. Ben-Arieh Y. 1965 The Central Jordan Valley: a crossroads area. Bulletin of the Israel Exploration Society 29: 168-179 – in Hebrew. Bietenhardt H. 1977 Die syrische Dekapolis von Pompeius bis Traian. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang des römischen Welt II.8 (Berlin): 220261. Brünnow R.E. & Domaszewski v. A. 1905-1909 Die Provincia Arabia. I-III. (Strassburg). Cagnat R. 1936 Nouvelles archéologiques: un nouveau milliaire de Syrie. Syria 17: 99-100. Conder C.R. & Kitchener H.H. 1881-1883 The survey of western Palestine, Memoirs. I-III. (London). Dalman G. 1935 Sacred sites and ways. (New York). Fast T. 1956 Verkehrswege zwischen dem südlichen West-und Ostjordanland. Zeitschrift des Deutschen-Palästina Vereins 72: 149-151. Fischer M., Isaac B. & Roll I. 1996 Roman roads in Judaea II: The Jaffa – Jerusalem roads. British Archaeological Reports Int. Ser. 628. (Oxford). Glueck N. 1946 The River Jordan. (London). Graf D.F. 1992 Hellenisation and the Decapolis. ARAM 4: 1-48. Graf D.F. 1995 The via nova Traiana in Arabia Petraea. In J. Humphrey (ed.) The Roman and Byzantine Near East; some recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 241265. Ibach R.D. 1994 Two roads lead to Esbus. In D. Merling & L.T. Geraty (edd). Hesban after 25 years. (Berrien Springs). 65-79. Ilan Z. 1989-90 Eastern Galilee: survey of Roman roads. Excavations and surveys in Israel 9: 14-16. Ilan Z. & Damati E. 1974-75 Ancient roads in the Samarian Desert. Museum Haaretz Yearbook 17-18: 43-52 – in Hebrew, with English summary at p. 29. Isaac B. 1998 The Decapolis in Syria: a neglected inscription. In B. Isaac The Near East under Roman rule. Selected Papers. (Leiden): 313-321. Isaac B. & Roll I. 1976 A Roman milestone of AD69 from Judaea: the Elder Trajan and Vespasian. Journal of Roman Studies 66: 15-19. Isaac B. & Roll I. 1979 Legio II Traiana in Judaea. Zeitschrift

218

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia Schattner I. 1962 The lower Jordan Valley. (Scripta Hierosolymitana XI). (Jerusalem). Schürer E. 1973-1987 The history of the Jewish people in the age of Jesus Christ I-IV. (A new English version revised and edited by G. Vermes, F. Millar and others: Edinburgh). Schumacher G. 1886 Across the Jordan. (London). Schumacher G. 1889 The “Via Maris”. Palestine Exploration Fund Quarterly Statement: 78-79. Schwöbel V. 1904 Die Verkehrswege und Ansiedlungen Galiläas in ihrer Abhängigkeit von den natürlichen Bedingungen. Zeitschrift des Deutschen PalästinaVereins 27: 1-151. Smith G. A. 1966 The historical geography of the Holy Land (Fontana Library reprint of the 25th edition, 1931: London). Stepansky Y. 1997 Horvat Mishlah. Excavations and surveys in Israel 16: 30-31. Steuernagel C. 1927 Der Adschlun. (Leipzig). Taktikon 1864 Palamas, G. (ed.) Hierosolymias. (Jerusalem): 376-382. Tepper Y. et al 2000 The region of Nahal Amud. (Tel Aviv). Thomsen P. 1917 Die römischen Meilensteine der Provinzen Syria, Arabia und Palaestina. Zeitschrift des Deutschen Palästina-Vereins 40: 1-103. Tsafrir Y., Di Segni L. & Green J. 1994 Tabula Imperii Romani: Iudaea-Palaestina. Maps and gazetteer. (Jerusalem). Weber E. 1976 Tabula Peutingeriana. Codex Vindobonensis 324, I-II. (Graz). Wilkinson J. 1975 The way from Jerusalem to Jericho. Biblical Archaeologist 38: 10-24. Wilkinson J. 1977 Jerusalem pilgrims before the Crusades. (Jerusalem). Wilson C.W. 1880 Picturesque Palestine, Sinai and Egypt, II. (London).

219

Limes XVIII

Fig. 1. Oblique satellite image of the Rift Valley, looking south (courtesy of R. Cleave).

220

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

Fig. 2. Map of main Roman roads in Judaea and Arabia.

221

Limes XVIII

Fig. 3. The Tyrus – Paneas road descending towards the Rift Valley, looking east.

Fig. 4. The lower end of the descent, looking east.

222

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

Fig. 5. The bridge over the River Hatsbany, as illustrated in Wilson 1880: 103.

223

Limes XVIII

Fig. 6. Oblique satellite image of the Tyrus - Paneas (No. 5) - Damascus road, looking east (courtesy of R. Cleave).

224

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

Fig. 7. Judaea and Arabia in the Tabula Peutingeriana – a reproduction.

Fig. 8. The lateral Roman road near Almagor, looking east.

225

Limes XVIII

Fig. 9. Oblique satellite image of the area of Sea of Galilee and its crossing lateral roads, looking east (courtesy of R. Cleave).

Fig. 10. The Sepphoris - Tiberias road near the Golani junction, looking east.

226

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

Fig. 11. Section across the road in the same place.

Fig. 12. Milestone found near the Golani junction.

227

Limes XVIII

Fig. 13. The Neapolis - Philadelphia road on the plateau east of Neapolis, looking east.

Fig. 14. The crossing of hilly terrain by the same road, looking east.

228

Israel Roll: Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

Fig. 15. Jisr ed-Damieh as illustrated in Merrill 1881: 422.

Fig. 16. The Jerusalem-Jericho road above Qasr Aly, looking west.

229

Limes XVIII

Fig. 17. Carved and built steps of the engineered ascent, looking west.

Fig. 18. Oblique satellite image of the lower Jordan Valley and its crossing lateral roads (courtesy of R. Cleave).

230

Lajjun – Legio in Israel: Results of a survey* in and around the military camp area Yotam Tepper This paper will present the results of a survey which has been conducted since 1998 in the vicinity of Megiddo, at the northern outlet of Wadi 'Ara in northern Israel. The funds for the survey were provided by a research grant, sponsored by Prof. B. Isaac and the Dept. of Classical Archaeology, Megiddo Excavation Team, and the Dept. of Archaeology at the University of Tel Aviv. The survey was conducted in the region of Legio, in an area of about 13sq kms, within a radius of about 1.5 to 2.5kms around the hill of El Manach, which is located between Megiddo Junction and Kibbutz Megiddo. Professors B. Isaac and I. Roll have suggested that El Manach is the site of the camp of legio VI Ferrata. The area around this hill (a radius of about 1km) was thoroughly surveyed during the various seasons for the maximum information. In the survey material dating from the period of the British Mandate Government down to the present was collected and was analyzed within the scope of: sources of water, water-mills, aqueducts, water tunnels, roads, agricultural installations, graves, architectural elements and other ruins, etc. Within the area that was surveyed there were several sites of settlements dating from the Roman period. The Jewish village of Kefar Othnai (Capercotani) is documented in Talmudic sources. The site of the camp of the Roman legion, legio VI Ferrata, as well as the city, known as Maximianopolis, flourished during late Roman and Byzantine periods. This paper will present the data obtained from the survey in an attempt to show the connection, which exists between the data and the sites in this area: a village, a camp and a city.

between the C1st and the C7th CE: The Jewish village Kefar ‘Otnay, the Roman legionary base of the legio VI Ferrata and the Roman Byzantine city - Maximianopolis (Tsafrir, Di Segni & Green 1994: 170).

Introduction Lajjun-Legio area is located in the north part of Israel near the modern Kibbutz Megiddo, between the hills of Ramat Manash, the mountains of Samaria and Jezreel Valley, near the northern end of the Wadi ‛Ara Pass and beside an important road junction (Isaac & Roll 1982: 3-16; Tsafrir, Di Segni & Green 1994: 170, 182; Gadot 1999; Zertal & Mirkam 2000; Zertal 2000: 10-17; Gadot & Tepper forthc. a).

Several scholars passed through the area during the C19th and the C20th (Van de Velde 1854: 353; Robinson 1856: 117; Guérin 1869-1880 II: 232-238; Wilson 1880: 23-25; Conder & Kitchener 1882: 64-66; Schumacher 1903: 4-10; Nelson 1913; Alt 1914: 70; Thomsen 1917: 69-70). The first excavation in the area, which was directed by Schumacher (1908) and presented a general description of many items and archaeological sites in a well-documented study (Fig. 2), became the basis and the starting point of my research.

The survey around Lajjun-Legio was conducted within a radius of 2 to 3kms around the El Manach hill (M.R. 16772202; see Fig. 1), which is located between Kibbutz Megiddo and the modern Megiddo junction. Isaac and Roll (1982: 34-35) have suggested that El Manach hill is the site of the camp of legio VI Ferrata (Ritterling 1925: 15871596; Cotton 2000). The survey in and around the region of Lajjun-Legio covered an area of about 13 sq. kms.

Within the framework of this research vast ceramic surveys were conducted several times during various seasons and architectural items were recorded from the site as well as from private collections. Private coin collections were examined, and personal testimony provided by the senior people who live in the region was registered. Ruins, ancient roads, agricultural installations, graves, aqueducts, and water installations were also surveyed.

A few settlements existed in the area since the biblical period until today, such as Tell Megiddo (Finkelstein & Ussishkin 2000: 576-605) and Kibbutz Megiddo. Historical sources mention three settlements in this region

The Jewish Village – Kefar ‘Otnay

*

This paper discusses parts of a survey that was conducted during the years 1998-1999 within the framework of my MA thesis at the Classical Archaeology Dept., Tel Aviv University, under the supervision of Prof. Israel Roll. The funds for the survey were provided by the research fund, sponsored by Prof. B. Isaac, the Dept. of Classical Archaeology, the Megiddo Excavation project held by Prof. David Usishkin and Prof. Israel Finkelstein and the Dept. of Archaeology at Tel Aviv University. I would like to thank the owners of private collections, the Eretz Israel Museum (Qedman) and the Israel Antiquities Authority for allowing me to study their collections and publish them here. I thank all the people who participated in this survey, those who made suggestions and contributions, as well as those who diligently worked on the findings in the Archaeological Institute at Tel Aviv University and at the Hebrew University of Jerusalem. I thank them all.

The Jewish village attested as a Kefar ‘Otnay, on the southern Galilean border and once settled by the Samaritans, is mentioned in Mishnaic sources (Tosafta, Gittin 1:4, 5:7; Tosefta, Demai 5:23; Mishnah, Gittin 1:5). Its name is also mentioned in Latin inscriptions (CIL III.6814; 6816). However different scholars have identified the location of Kefar ‘Otnay with different ruins around Kibbutz Megiddo.

231

Limes XVIII

A fragment of a stone door (Fig. 3) that seems to have belonged to a monumental grave, had an inscription on it. Naveh suggested that the inscription mentioned Jewish and Samaritan names. The final results will be published later, however, the inscription testifies to the presence of a Jewish or Samaritan settlement in the region, although, unfortunately, the name of the settlement is not mentioned.

was located on the El Manach hill, measured the distances along the road to Scythopolis (Bet Shean). They surveyed the section of the road that stretches from the east of El Manach hill to Scythopolis, and found many milestones, three of which mention the distance to the military camp (Legio) as a caput viae (Isaac & Roll 1982: 55-82, 86; Isaac 1998: 63-64).

However, the main archaeological evidence for the location of Kefar ‘Otnay was stored, without being noticed, in the archives of the Antiquities Dept. of the State of Israel (No.126; Lejjun) for many years. In 1940, during the construction of the British police station (24 May - 5 June 1940), the remains of buildings and various installations were revealed (Fig. 4; Lejjun P.B. Foundation Plan, 23 May 1940). These were marked with a fine pointed pencil on the blueprints of the police station, from which we are able to draw the site plan that includes the organization of the streets and alleys, as well as the houses plan and the location of a Jewish ritual Mikveh.

In the 1940s a gate house, that had monumental lions on each of its sides (The archives of the Dept. of Antiquities in Rockefeller Museum, File No.126, Photo no. 27079), was found in the fields of El Manach hill, towards the east, on an artificial fold of ground (Fig. 5). This finding together with the suggestion of Isaac and Roll (1982: 3435) support the assumption that this was the course of the Roman road that led into the camp site (see below). In addition, both the Legio - Ptolemais (Acco) road, and the road towards Jenin and Neopolis (Nablus) appear to start from El Manach hill. Schumacher mentioned two milestones to the north of Tell Megiddo (Schumacher 1899: 339; 1908: 175), and Thomsen (1917: 69) noted two more on the road to Ptolemais (Acco) at the third-fourth mile from Legio. Three milestones were discovered on the road towards Jenin and Neopolis (Nablus), near Kubtie (No name 1902: 152; Thomsen 1917: 70).

Various findings, including ceramic lamps from the Roman, Byzantine and the early Islamic periods discovered in the area of the police station, can be found in the archives of the Rockefeller Museum, as well as in private collections (Kibbutz Givat ‘Oz Museum). According to the evidence mentioned above, we assume that the Jewish village was located to the south of Nahal Qeni (Qeni River) on the hill of the modern British police station.

Schumacher reported a road between Legio and Caesarea, which passed along the Wadi ‘Ara pass, went via the hilly ridge of Musmus and then to Legio (Schumacher 1903: 410). He also reported fragments of milestones between Megiddo and Ein Zituna. Alt saw only the milestone from Ein Zituna (Alt 1914: 80, n.2) and Thomsen saw a part of Roman road (Thomsen 1917: 69). Roll mentioned four milestones on that road west of Legio (Tsafrir, Di Segni & Green 1994; the attached map) and reported one milestone near Givat Ada dated to 162CE (Roll & Ayalon 1988: 156157). Glik (1995: 44-45) excavated a building near Ein Zituna, that probably belonged to that road. Gadot and Tepper (forthc. b) report road remains and one more milestone near Regavim as well as many Roman graves along the road south of Regavim.

The road network It is an accepted fact that throughout the empire, Roman camps were generally located at position of strategic importance, such as junctions of imperial roads, near river crossings or mountain passes. The camps were built according to accepted standards (Webster 1969: 172) and permanent camps in Europe and Africa covered an area of 180-260 dunams (Daremberg & Saglio 1962: 940-959). Although Roman camps in Judaea have not been excavated yet it seems that they were the same size and shape as legionary fortresses in the west (Parker 2000: 121-125). The identification of a Roman road and milestation may be used, as a starting point for locating a Roman camp. Isaac suggests that until Severus’ reign military key points served as capita viarum in Judaea (Isaac 1998: 66).

In the archives of the Dept. of Antiquities in the Rockefeller Museum (File No. 9 Photo no. 6920) we have found evidence of a milestone from the Wadi ‛Ara (Fig. 6). During the survey a stepped ascent (Fig. 7) was located at a spur south of Nahal Qeni. We have surveyed the outline of the southern extension along the spur of Nahal Qeni up to Musmus, as well as an northern extension towards the crossing over the “Roman bridge” (Isaac & Roll 1982: 35) at Nahal Qeni (M. R. 16735-22000; Wilson 1880: 24) near El Manach hill. All the above support the evidence from the Tabula Peutingeriana (Weber 1976. Ed. Seg. IX) and Schumacher’s (1903: 4-10) proposal of the Roman road from Legio to Caesarea.

Eusebius noted in his Onomasticon (Klostermann 1904 ed. passim) the distances that were measured from Legio (Legeon). In accordance with this, Hecker (1961: 175-186) calculated the distances along the road from Legio to Diocaesaria (Tzipori), where two Hadrianic milestones that mentioned the former (Caparcotna) as caput viae were found (Avi Yonah 1946: 97: Hecker 1961: 175-176: Isaac 1998: 63).

The Itinerarium Burdigalense (Geyer & Kuntz, ed. 1965: 586,1) reported a journey, in 332CE, from Caesarea to Maximianopolis (Legio) via Mount Syna - a distance of 21

Based on the network of Roman roads, Isaac and Roll (1982: 34-35), who have suggested that the legionary camp 232

Yotam Tepper: Lajjun – Legio in Israel: Results of a survey in and around the military camp area miles. Mount Syna and its springs were identified as Shuni and ‘Ein Tzur miraculous spring (Tsafrir, Di Segni & Green. 1994: 165, 211, 238-239; Hirschfeld 2000: 293306). Neeman found two milestones dated to 120CE near Shuni (Neeman 1989: 55). Roll suggested that the distance of 18 miles written on both stones represented the road to Legio via Geba (Roll 1996: 557). In the C6th Theodosius probably also used this road (Geyer ed. 1965: 114-125) in his journey to Diocesarea.

In addition, there are at least nine tiles, with the stamp of the legio VI (Fig. 9), that we have located in private collections (at Kibbutz Megiddo and at Kibbutz Givat ‘Oz. Israel Antiquities Authority, nos. 1996-5999/6000; 20005002/5005; see also Schumacher 1908: 175). These were found on El Manach hill and its slopes, near the British police station and in the area of Kibbutz Megiddo. Ceramic pipes like the ones that were found in and around the proposed base of legio X Fretensis in Jerusalem (Tsafrir 1984: 43; Geva 1984: 239-254), that were used for carrying water to the base, were also found at El Manach hill (see Fig. 10g; Museum K. Givat ‘Oz. Israel Antiquities Authority, nos. 1996- 5828/5830).

We have found remains of road and part of a milestone 2kms west of El Manach hill (M.R. 1655-2199). It seems that the earlier pass from Caesarea to Legio was here, via nearby Hellenistic and early Roman period sites, such as Geba (Mazar 1988) at the east and the fortress at Horbat ‘Elek (Hirschfeld 2000: 235-270) to the west.

Some Roman coins, which bear the stamp of the legio VI (LVIF), were found in the area, and can be seen today in Eretz Israel Museum (Qedman) and in private collections (Kibbutz Megiddo). Over the years many coins have been discovered in the area of Legio. It is accepted that these coins were used by the soldiers of the legio VI (Barag 1963: 117-125).

The Roman road network near the source of water - Nahal Qeni which is close to Tell Megiddo and the early Islamic khan - points to the use that was made in this area for the establishment of the Roman camp (see below). This region was also a main junction of ancient roads used before and after the Classical periods.

On nearby Moshav Ha-Yogev we have found a stone altar, which was brought there from El Manach, where it was originally found. According to the translation (Eck & Tepper 2001), it seems that one or two soldiers of the legio VI (or another legion) dedicated this altar to the Roman deity Silvanus. Another altar was found in the 1940s in nearby Horbat Muzav (= el Chazne: Avi Yonah 1946: 8991). The marble altar was dedicated to the Egyptian god Serapis by an officer from the legio VI and, it has reliefs with Roman symbols. During the past century more altars were found in this area, one of them, reused in the entrance to Kibbutz Megiddo, was previously described by Schumacher at Horbat Muzav (1908: 163-165).

The Roman legionary camp of legio VI Ferrata Schumacher (1908: 188-190) marked a camp measuring 70 x 40m on top of a hill south of Nahal Qeni (Fig. 8). He dug several graves that were in the area and suggested that this was a Roman camp - probably that of legio VI Ferrata (Schumacher. 1908: Abb. 287). In our survey we could not find anything that Schumacher described but there are still some scholars who agree with his suggestion (Ilan 1988: 71). We do think that the remains are of a camp, but that they are too small for a legion. Therefore, and according to Schumacher’s findings, we assume that the camp was used by an auxiliary unit or it may have served as a fortress in the Roman or another period.

Throughout the years and during our survey, several fragments of statues have been discovered (Schumacher 1908: 187). One of them, for example, is a marble head (archives of the Antiquities Dept. No.126; Lejjun), dated to the C3rd CE, and obviously of an important person, perhaps a governor or even one of the Caesars.

Arthur Segal (1999: 48-49) dug several test trenches on the north-eastern slope of the hill and revealed an east-west oriented paved road and a few walls. The survey that we conducted revealed the tops of walls in the south-east and south-west areas of the eastern slope of the hill, and perhaps these belong to the camp and mark its periphery.

The Arabic name of the site (El Manach), means “place of encampment” or “the campground” (Sharoni 1987: 1266). This fact together with all the above, is a good indication that the site of the Roman camp was on the El Manach hill. Ceramics survey conducted there, support our proposal that the eastern slope of the hill is probably the place of the camp.

Schumacher (1908: 176; Tafel 1) marked a gate and the course of the circuit walls to the south of the hill. Along the remains of the walls he found a burial ground with sarcophagi and burial monuments, some of which were located in our survey. Remnants of cremation burials in Roman cooking pots were found in three different places on the slope between Kibbutz Megiddo and El Manach hill (Tsuk 1988-9: 95-96; Kibbutz Megiddo Museum, Israel Antiquities Authority, no: 1998-5286/5288). These are similar to the cremation burials that were found in Jerusalem (Barkay 1984: 97-98) near the suggested location of the camp of the legio X (Geva 1984: 239-254). Such burials are accepted as characteristic of Roman legionary soldiers.

According to historical sources, a second legion was brought to Judaea at the beginning of the reign of Hadrian (Avi-Yonah 1973: 209-213; Keppie 1973: 859-864). Pflaum suggested that it even came in at the end of the reign of Trajan (Pflaum 1969: 232-233; Eck 1984: 55-67; Cotton 2000). Isaac and Roll (1998a: 208-210) claimed that this was the legio II Traiana, and that it arrived some time before 120 CE (Ibid. 1998: 203). According to Eck it 233

Limes XVIII

was the XXII Deiotariana (Eck 1999: 81). Only later did legio VI arrived and remained in Judaea.

Our team examined a 17m long tunnel that was hewn in the sub-surface rock leading from the water source up to a monumental pool. The tunnel may be dated to the C3rd CE and it might have served the camp or have been the source of the urban aqueduct. This is the same tunnel that the team of the PEF has surveyed about 120 years ago (Conder & Kitchener 1882: 64-66).

We would like to emphasize that the presence of the Roman legionary camp in the Galilee in general and in Legio in particular had influenced the course of events that took place in the region. A similar Roman camp was located in Jerusalem, hence the centrality of these two sites during that era.

According to Schumacher (1908: Tafel 1) there used to be a big ancient pool (“el-Chuwcha”; 90 x 30m), which was built of hewn stones from the Roman period (Schumacher 1908: 176). Today the sewage pools of Kibbutz Megiddo cover the site of the pool (M.R. 16765-22080).

Water sources – aqueducts, installations and mills The impressive ancient water installations of Tell Megiddo (Lamon 1935) are not the only ones that were in use in this area (Schumacher 1908: Tafel 1; Avitsur 1960: 47-51; Tsuk 1988-9: 92-97). Throughout the ages water was essential for settlement and the survey revealed a number of important systems from the Roman-Byzantine period.

Not only the Roman camp used the Qeni Springs as its source of water (see Fig. 1). The springs played a central role in the civilian habitation pattern and in the economy of the whole area in different periods, from the ancient site of Megiddo to the village of Lejjun.

Remnants from a later period were found on top of Tell Megiddo (Schumacher 1908: Tafel XLVIII; Newell 1939: 196-211; Sass 2000: 412-415; Ilan, Franklin & Halloth 2000: 101-102). The Chicago Excavation Expedition revealed a staircase leading down to the great water system dated to the “Greco-Persian period (4th century)” (Lamon 1935: 31, 36). A number of Roman coins, dated to the C1st to C4th CE, were discovered in the water system (Lamon 1935: 37-39), thus showing that a part of it continued to be in use at the Roman-Byzantine period.

The water mills in the Qeni Nahal (Schumacher 1908:185187; Avitsur 1960: 47-51) run along a 1.5kms stretch with a drop of 45m from the first to the 7th mill, which is located on the other side of the modern road. Water mills are a development of the Classical period, and are well documented in literary sources (Vitruvius De Arch. X.5.2) and in archeological findings from the C1st BC (White 1984: 65-67, 194, 196-201). The Qeni Nahal water mills have been renovated and modified throughout the ages.

During the Roman period aqueducts and water installations were signs of Roman rule (Tsafrir 1984: 54). Their development reached a peak during the C2nd and C3rd CE (Tsuk 1994: 23-24). Soldiers from the II, VI and the X legions participated in the building and repair work of the aqueducts near Caesarea (Tsafrir 1984: 55; Olami & Ringel 1975: 148-150; Negev 1964: 237-249). It is difficult to estimate whether the aqueducts were for military or for civilian use; we assume that they served both the soldiers and the civilians in the city.

We think that the fourth water mill, which is incorporated into the Roman bridge (Isaac & Roll 1982: 35) that spans the Qeni Nahal, is possibly one of the earliest example of a water mill in our region. If our assumption is correct, we may conclude that this mill served the Roman camp and the town before the establishment of the Arab village. Furthermore, the thousands of legionaries who lived in Legio camp needed a central food supply and food producing, like the milling industry at El Lejjun in Arabia (De Vries 1987: 423-424, n.51) or along the Hadrian’s Wall (White 1984: 198). The legionaries’ diet included mainly bread and porridge (Junkelmann 1997: 128-133) and it seems that the water mills at Qeni Nahal would have been the best way at the time, for grinding the grains from the Jezreel Valley.

During the Classical period there were at least two aqueducts running from Ein el Qubbi, one to the theatre and the upper part of El Manach hill, and the other to the lower part of the hill where the camp was situated (Schumacher 1908: Tafel 1).

The Roman-Byzantine City – Maximianopolis The theatre (Schumacher 1908: 173-175) measuring 45 x 75m, that was located in a small wadi to the west of El Manach (M.R. 1675-2207) could have served both the soldiers and the citizens. One of the aqueducts, which started from Ein el Qubbi and reached the theatre, might have been used for the water games or for water installations in and around the theatre.

The Pilgrim from Bordeaux (Itinerarium Burdigalense, ed. Geyer and Kuntz, 1965), passed through Maximianopolis in 332CE. Avi-Yonah (1966: 122-123) and Isaac and Roll suggested that the polis has been established in 305CE (Isaac & Roll 1982: 11). It was on the ruins of Maximianopolis that the later village of Lajjun was established in the C7th CE (Strange 1965: 492-493).

We have located a stone in Kibbutz Megiddo, which was brought there from the eastern slope of El Manach. The stone has a carving of a Gorgon head or a Lion’s head, and it used to serve as a mouth of a fountain, that received its water from the lowest aqueduct.

According to Schumacher (1908: 182; Tafel 1) two colonnaded streets were built on the hill of Dhar e-Dar (Fig. 11), today the south hill of Kibbutz Megiddo. The columns were carved of granite, which is not a local stone. 234

Yotam Tepper: Lajjun – Legio in Israel: Results of a survey in and around the military camp area This phenomenon has also been found in wealthy cities such as Beth Shean-Scythopolis and Caesarea (Tsafrir, Di Segni & Green 1994: 223-225; 94-96). Perhaps the street in Schumacher’s records was a decumanus – one of the main east-west streets in planned cities at that time, and possibly it had shops on its sides. We discovered in the archives of the Rockefeller Museum in Jerusalem a photograph that clearly shows the column street halfburied and surrounded by the mud huts of Lajjun (File no. 126. Photo no. 6921; 244/1924. Taken at Lajjun at the 19.4.1924).

this subject within the framework of this paper, however the findings in the area of Legio point to the importance of this place during the Roman and Byzantine periods. In this paper we suggest as probable the following locations: the Jewish village on the hill of the police station, the Roman camp on the eastern slope of El Manach hill, and the city of Maximianopolis on the south and east sides of the hill of Kibbutz Megiddo. There is no indication yet, as to the exact time of arrival and departure of the legio VI Ferrata to or from the region. It is attested that the Jewish village was already there by the time the Roman camp was established. However, we cannot say whether Maximianopolis was established beside the Roman camp or if it was founded after the legionaries left the area. In any event, the important fact is that three different sites occupied this limited area in the Roman-Byzantine periods.

Columns and many additional architectural items made of granite, marble, and local limestone were found in the survey. These items point to the presence of a public building, streets and additional buildings of the city in this area. One of the buildings, built of limestone strengthened by cement, has a dome. At the centre of the dome is an opening with a diameter of 80cms. We assume that this is a part of a large building; perhaps it is one of the rooms of a bathhouse, where the hot water was kept (Yegül 1992). It might have belonged to the city or to the camp.

To conclude, the preliminary survey findings presented here are intended to open new horizons for a more detailed archaeological investigation and the points suggested here may be confirmed or changed by means of archaeological excavations. The geographical and archaeological findings also contribute to the environmental study of Tell Megiddo and Lajjun.

On the southern slope of Kibbutz Meggido there is a stone drum, 80cms in diameter and 60cms high (Fig. 12), on which a wreath of victory (Heracles knot) and daggers are carved on 12 shields. If so, this is the only known fragment of an item from a monumental victory column in the region. It must have been erected in the Roman camp or in the city for either a military victory or an imperial visit.

Bibliography The Mishnah (H.D.D. Danby transl.: London 1933). The Tosefta, vol. I. (Edd. J.Zeraim, J. Neusner & R.S. Sarason: New Jersey 1986). The Tosefta, vol. III. (J.Nashim & J. Neusner (transl.), Third division: Georgia.1999). Claudii Ptolemaei Geographia. (Ed. C.F.A. Nobbe: Leipzig 1843-1845: republished Hildesheim, 1966). Corpus Inscriptionum Latinarum III. Suppl. 1-2. (Ed. T.Mommsen: Berlin. 1961). Eusebius Caesariensis episcopus. Onomasticon (Ed. E. Klostermann Das Onomastikon der biblischen Ortsnamen, Die griechischen christlichen Schriftsteller der ersten Jahrhundete, Kirchenväter Kommision der Käniglichen Preussischen Akademie der Wissenschftat: 11 I: Leipzig, 1904). Georgius Cyprius (Ed. E. Honigmann Le Synèkdemos d` Hieraklès et l’opuscule géographique de Georges de Chypre: Bruxelles, 1939). Itinerarium Burdigalense (Edd. P. Geyer & O. Kuntz Itineraria et alia geographica, Corpus Christianorum, Series Latina (CCSL) 175: Turnhout, 1965). Tabula Peutingeriana (Ed. E. Weber: Graz. 1976). Theodosius Die situ terrae sancta. (Ed. P. Geyer. In Itineraria et alia geographica, Corpus Christianorum, Series Latina (CCSL) 175: Turnhout. 1965). Vitruvius On Architecture. (Ed. T.E. Page et al The Loeb Classical Library, London. 1939).

Coins have been discovered on the eastern slopes of Kibbutz Megiddo at various times, mainly after rainy days (“Shraga collection”). These coins are from the Hellenistic to the late Muslim periods, some are C2nd CE city coins, others are Hasmonean, and some are from the days of the Jewish revolt (66-70CE). Thus, the periods in which the site was settled are expressed in this coin collection. Our survey revealed in Kibbutz Megiddo a square pedestal with inscribed crosses (Fig. 13) which testifies to the city’s Christian background (Acta Conciliorum Oecumenicorum III, In. Topographic vs. ed. Schwartz 1984: 202). Furthermore, a bishop of Maximianopolis is mentioned at the Council of Nicaea (325CE), and another is mentioned at the Council of Jerusalem (536CE; Isaac & Roll 1982: 11-13). The last mention of the polis, at the border of Galilee, is in the C7th (Georgius Cyprius l. 1034. ed. Honigmann 1939: 67; see also Hierocles, 720, 10; ed. Honigmann 1939: 43). Conclusion

Alt A.

1914 Pharao Thutmosis III in Palästina. Palästinajahrbuch des deutschen evangelischen Institutes für Altertumswissensfchaft des Heiligen Landes zu Jerusalem 10: 53-99. Avitsur S. 1960 A survey of water power installations in Eretz Israel. (Tel Aviv) - in Hebrew.

The agricultural installations found in the survey region, especially the various burial grounds (Tepper forthc.), point to the presence of various cultural entities in this area at different times: Jews, Samaritans, pagans, and Christians and civilians, as well as military. It is impossible to discuss 235

Limes XVIII

Avi-Yonah M. 1946 Newly discovered Latin and Greek inscriptions. Quarterly of the Dept. of Antiquities in Palestine 12: 84-102. Avi-Yonah M. 1966 The Holy Land from the Persian to the Arab conquest. (Jerusalem). Avi-Yonah M. 1973 When did Judea become a consular province? Israel Exploration Journal 23: 209-213. Barag D. 1963 The countermarks of the Legio Decima Fretensis. In A. Kindler (ed.) The patterns of monetary development in Phoenicia and Palestine in antiquity. International Numismatic Convention (Tel Aviv – Jerusalem): 117-125. Barkay G. 1984 Excavations on the slope of the Hinnom Valley, Jerusalem. Qadmoniot 68: 94-108 - in Hebrew. Conder R.E. & Kitchener H.H. 1882 The survey of western Palestine. Memoirs of the Topography, Orography, Hydrography and Archaeology. Vol. 2. Samaria. (London). Cotton M.H. 2000 The Legio VI Ferrata. In Y. Le Bohec (ed.) Les légions de Rome sous le Haut – Empire. Proceedings of Deuxième congrès de Lyon sur l`armée romaine. (Lyon): 351-360. Dalman G. 1935 Sacred sites and ways. Studies in the topography of the Gospels. (London). Daremberg C.H. & Saglio E.D.M. 1962 Castra. Dictionnaire des antiquités grecques et romaines III. Akademische Druck- u. Verlagsanstalt Graz/Austria. Linzuzausgabe der 1887 bei der Librairie Habnette in Paris: 940-959. De Vries B. 1987 The el-Lejjûn water system. In S.T. Parker (ed.) The Roman frontier in central Jordan, interim report on the Limes Arabicus Project, 1980-1985. Pt. i. British Archaeological Reports Int. Ser. 340: 1 (Oxford): 399-428. Eck W. 1984 Zurn konsularen status von Iudaea im fruhen 2. Jh. Bulletin of the American Society of Papyrologists 21: 55-67. Eck W. 1989 The Bar Kokhba Revolt: The Roman point of view. Journal of Roman Studies 79: 76-89. Eck W. & Tepper Y. 2001 A dedication to Silvanus near the camp of the legio VI Ferrata near Lajjun. Scripta Classica Israelica XX: 85-88. Finkelstein I. & Ussishkin D. 2000 Archaeological and historical conclusions. In I. Finkelstein, D. Ussishkin & B. Halpern (edd.) Megiddo III, The 1992-1996 seasons. (Publications in Archaeology, Tel Aviv University Monograph Series. No. 18) Vol. II: 575-605. Gadot Y. 1999 The Wadi ‘Ara Pass as an international highway during the Bronze Age, Iron Age and the Persian period, in the light of the settlement pattern. (MA at the Dept. of Archaeology & Near Eastern Studies, Tel Aviv University - in Hebrew). Gadot Y. & Tepper Y. forthc. (a) Settlement patterns at Ramat Manash region, result of an archaeological survey. Gadot Y. & Tepper Y. forthc. (b) Archaeological survey of Israel, Map of Regavim, 49. (Israel Antiquities Authority, Jerusalem) - in Hebrew & English. Geva H. 1984 The camp of the tenth legion in Jerusalem: An archaeological reconsideration. Israel Exploration Journal 34: 239-254. Glik D. 1995 Ein e-Zituna (Nahal ‘Iron Road). Hadashot Arkheologiot 103 (Israel Antiquities Authority, Jerusalem): 44-45 - in Hebrew. Guérin V. 1875 Description Géographique, Historique et Archéologique de la Palestine. II. Samarie. (Paris). Hecker M. 1961 The Roman road Legio-Zippori. Bulletin of the Israel Exploration Society 25/3: 175-186 – in

Hebrew. Hirschfeld Y. 2000 Horvat ‘Eleq: architecture and stratigraphy. In Y. Hirschfeld (ed.) Ramat Hanadiv excavations, final report of the 1984-1988 seasons. (Jerusalem): 235-370. Ilan D., Franklin N. & Halloth R.S. 2000 Area F. In I. Finkelstein, D. Ussiskin & B. Halpern (edd.) Megiddo III, The 1992-1996 seasons. (Publications in Archaeology, Tel Aviv University Monograph Series. No. 18) Vol. I: 75-103. Illan Z. 1988 Yesterdays. (Tel Aviv) - in Hebrew. Isaac B. 1998 Milestones in Judaea: from Vespasian to Constantine. In B. Isaac The Near East under Roman rule, selected papers. (Leiden): 48-68. Isaac B. & Roll I. 1982 Roman roads in Judaea I, The Legio-Scythopolis road. British Archaeological Reports Int. Ser. 141. (Oxford). Isaac B. & Roll I. 1998a Legio II Traiana in Judaea – A reply. In B. Isaac The Near East under Roman rule, selected papers. (Leiden): 208-210. Isaac B. & Roll I. 1998b Legio II Traiana in Judaea. In B. Isaac The Near East under Roman rule, selected papers. (Leiden): 198-207. Junkelmann M. 1997 Panis Militaris, Die Ernährung des römischen Soldaten oder der Grundstoff der Macht. (Mainz-am-Rhein). Keppie L. 1973 The legionary garrison of Judaea under Hadrian. Latomus 32: 859-864. Lamon R.S. 1935 The Meggido water system. (University of Chicago, Illinois). Mazar B.1988 Geva, archeological discoveries at Tell AbuShusha, Mishmar Ha-‘Emeq. (Tel Aviv-Jerusalem) - in Hebrew. Neeman Y. 1989 Stonemiles. Hadashot Arkheologiot 93 (Israel Antiquities Authority, Jerusalem): 55 - in Hebrew. Negev A. 1964 The high level aqueduct at Caesarea. Israel Exploration Journal 14: 237-249. Nelson H.H. 1913 The Battle of Megiddo. (University of Chicago, Illinois). Newell E.T. 1939 The coins. In R. Lamon & G.M. Shipton (edd.) Megiddo I. (University of Chicago, Illinois): 196-211. No name 1902 Revue Biblique 11: 152. Olami J. & Ringel J. 1975 New inscriptions of the tenth legion Fretensis from the high level aqueduct of Caesarea. Israel Exploration Journal 25: 148-159. Parker S.T. 2000 Roman legionary fortresses in the east. In R.J. Brewer (ed.) Roman fortresses and their legions. (Occasional Papers of the Society of Antiquaries of London. No. 20): 121-138. Pflaum H.G. 1969 Remarques sur le changement du statut administratif la province de Judée: A propose d’une inscription récémment découverte à Sidé de Pamphylie. Israel Exploration Journal 19: 225-233 Ritterling C. 1925 Legio VI Ferrata. RE 12.2. (Stuttgart): Cols. 1587-1596. Robinson E. 1856 Biblical Research in Palestine, and in the adjacent regions; A journal of travels in the year 1838. Vol. 3. Later Biblical Researches in Palestine and in adjacent region: A journal of travel in the year 1852. (Boston). Roll I. 1996 Roman roads to Caesarea Maritima. In A. Raban & K.G. Holum (edd.) Caesarea Maritima, a retrospective after two millennia. (Leiden): 549-558. Roll I. & Ayalon E. 1988 Highways and roads in the Sharon Plain during the Roman and Byzantine periods. In R.

236

Yotam Tepper: Lajjun – Legio in Israel: Results of a survey in and around the military camp area Zeevy (ed.) Israel People and Land 4 (1986-1987). (Tel Aviv): 146-162 - Hebrew. Sass B. 2000 The small finds. In I. Finkelstein, D. Ussiskin & B. Halpern (edd.) Megiddo III, The 1992-1996 seasons. (Publications in Archaeology, Tel Aviv University, Monograph Series. No. 18) Vol. II: 412-415. Schumacher G. 1899 Reports from Galilee. Palestine Exploration Fund. Quarterly Statement: 339. Schumacher G. 1903 Die ägyptische Hauptstrasse von der Ebene Saron bis zur Ebene Jesreel. Mittheilungen und Nachrichten des Deutschen Palaestina Vereins : 410. Schumacher G. 1908 Tell el-Mutesellim I. Band, Fundebericht. A. Text. (Leipzig). Segal A. 1999 Megiddo Junction. Hadashot Arkheologiot 109 (Israel Antiquities Authority, Jerusalem): 48-49 - in Hebrew. Sharoni A. 1987 The Comprehensive Arabic-Hebrew Dictionary. (Ministry of Defence Publishing House, University of Tel Aviv & Israel Defence Forces Intelligence Corps) - in Hebrew & Arabic. Strange G.L. 1965 Palestine under the Moslems. A description of Syria and the Holy Land from AD 650 to 1500. (Beirut). Tepper Y. forthc. Legio survey. Hadashot Arkheologiot, (Israel Antiquities Authority, Jerusalem) – in Hebrew. Thomsen P. 1917 Die Römischen Meilensteine der Provinzen Syria, Arabia und Palaestina. Zeitschrift des DeutschenPalästina Vereins 40: 1-103. Tsafrir Y. 1984 Eretz Israrel from the destruction of the Second Temple to the Muslim conquest. Vol. 2: Archaeology and Art. (Jerusalem) – in Hebrew. Tsafrir Y. Di Segni L. & Green J. 1994 Tabula Imperii Romani. Iudaea Palaestina. (Israel Academy of Sciences & Humanities, Jerusalem). Tsuk T. 1994 Water-supply Systems in Roman Palestine. In Y. Tsafrir, L. Di Segni & J. Green (edd.) Tabula Imperii Romani. Iudaea Palaestina. (Jerusalem): 23-24. Tzuk T. 1988-9 The aqueduct to Legio and the location of the VIth Roman legion. Tel Aviv 15-16: 92-97. Van de Velde C.W.M. 1854 Narrative of a journey through Syria and Palestine in 1851 and 1852. (Edinburgh). Webster G. 1969 The Roman imperial army, of the first and second centuries AD. (London). White K.D. 1984 Greek and Roman technology. (London). Wilson C.W. (ed.) 1880 Pictoresque Palestine, Sinai and Egypt I,II.. (London). Republished as C.W. Wilson The land of Galilee and the North. (Jerusalem, 1975). Yegül F. 1992 Baths and bathing in classical antiquity. (New York). Zertal A. 2000 The Iron Pass. Cathedra 97: 7-24 - in Hebrew. Zertal A. & Mirkam N. 2000 The Manasseh hill country survey, from Nahal ‘Iron to Nahal Shechem. (University of Haifa & the Ministry of Defence Publishing House) – in Hebrew.

237

Limes XVIII

Fig. 1. Map of the survey area, ElManach hill location between Kubbutz Megiddo and Megiddo Juction wl Qeni Nahal (River) makes a semi-circular curve to the south.

Fig. 2. Schumacher’s survey map of 1908: south to Tell Megiddo and around El-Manach hill.

238

Yotam Tepper: Lajjun – Legio in Israel: Results of a survey in and around the military camp area

Fig. 3. Fragment of a stone door, with a Hebrew inscription on it.

Fig. 4. Photo of the excavation during the building of the police station (Rockefeller Museum).

Fig. 5. Limestone lion sculpture, from the east slope of El-Manach hills (Rockefeller Museum).

239

Limes XVIII

Fig. 6. A milestone from Wadi Ara, 1924 (Rockefeller Museum).

Fig. 7. Stepped ascent (M.R. 16655-21941), south-west to El-Manach hill.

Fig. 8. Schumacher’s 1908 map of the Roman camp at Lajjun.

240

Yotam Tepper: Lajjun – Legio in Israel: Results of a survey in and around the military camp area

Fig. 9. Tile from Kibbutz Megiddo, bearing the stamp of the “LEG VI FE[R]”.

Fig. 10. Ceramic pipes from the east slope of El-Manach hill (Givat Oz Museum).

Fig. 11. Photo from the Arab village of Lajjun, 1924. Columns street half buried and surrounded by the huts (Rockefeller Museum).

241

Limes XVIII

Fig. 12. Half-round limestone drum (base) on which a wreath of victory (Heraklase knot) and daggers are carved on 12 shields that are hung on city’s battlements.

Fig. 13. Square pedestal with inscribed crosses at Kibbutz Megiddo.

242

Road use in late antique Palestine Susan Weingarten Late antique Palestine had a network of over 1000 miles of official Roman highways, marked by milestones. However, even when extant milestones can be dated, they only record the date of the building of the road, or its repair; they do not tell us when the road was actually in use. Excavation of roads produces very little dateable material. For this we must rely on the evidence of the literary sources and the partial evidence of sites along the route. The mediaeval copy of an ancient map known as the Tabula Peutingeriana has information for Palestine that appears to pre-date the Severan re-organisation of the mid-C3rd. This map shows a road running north-south all along the Mediterranean coast from Caesarea to Ascalon. However, the literary evidence from the C4th, from itineraria, papyri, pilgrims accounts, handbooks and other written sources, both pagan and Christian, shows that no travellers used the road from Caesarea to Ascalon at this time. I have plotted their routes on the present-day map of the known Roman roads of Palestine. They all appear to have taken the route which detoured further inland via Lydda. The coastal road between Caesarea and Ascalon was therefore presumably out of use during the C4th. The picture obtained from the literary sources is confirmed by the mounting archaeological evidence for the silting up and decline of the port of Caesarea and the lack of evidence from the C4th in the excavations at Apollonia. It is suggested that moving sands, a local problem from the time of Josephus to the present-day, may have made the road impassable. At the beginning of the C6th the emperor Anastasius is recorded to have engaged in a large-scale restoration of the port of Caesarea. Pilgrim accounts from around this time show that the coastal road south of Caesarea appears to have been back in use again, at least as far as Jaffa.

By the C4th, provincia Palaestina had a well-developed network of engineered Roman roads – over 1000 miles of them, connecting the principal towns and cities (Roll 1995: 1166-1170).1 Palestine had long served as a land bridge between Syria in the north and Egypt in the south. Thus the Roman authorities had paved the existing north-south arteries as well as built east-west roads connecting the coastal cities with the interior, and especially with Jerusalem. Figure 1 shows the road system of late antique Palestine, based on the work of Israel Roll.2 Roman road construction was so massive that much of it still remains in the field, and the map represents extant remains.3 Official Roman highways were further marked with milestone pillars. Thus the map also marks inscribed late antique milestones. To date over 550 milestones from the Roman and late antique periods together have been found in Israel and the West Bank, of which about 130 are inscribed, but not all of these are dateable, so the number of extant late antique milestones may well be much higher.

these secondary road installations may be dateable, but it is not always possible to know whether wayside buildings were or were not connected with the use of the road. So we are obliged to turn to the evidence of the written sources. Very few maps are extant from antiquity (Talbert 1990: 215f.; Lee 1993: 81-90; Brodersen 1995; Bertrand 1998). For information about Roman roads in Palestine, there is only the so-called Tabula Peutingeriana, a mediaeval copy of an ancient map (Levi 1967: 117-124; Dilke 1985: 112120). The information on the map is presented in strip format, like modern linear strip-maps (MacEachren 1986; MacEachren & Johnson 1987: 148, fig. 1), marking landmarks en route, but without the usual orientation to the north familiar from present-day topographical regional maps. The compiler of the Peutinger map used information that was often out of date – even though the presence of Constantinople shows that it was compiled no earlier than the C4th, the information used for Palestine was in places much older. This is clear from the use of names for towns which pre-date the C3rd Severan re-organisation: Betogabra (Bet Guvrin) rather than Eleutheropolis, Amavante (Emmaus) rather than Nicopolis, and Luddis (Lydda) instead of Diospolis.4 The Peutinger map for

However, even when the inscription on a milestone can be dated, this does not provide the complete answer as to when a particular road was actually in use. Roads normally continued in use for an unspecifiable time after the milestone was put up. Roman roads were originally built primarily for military and administrative needs and included provision of guard posts as well as a system of road stations mansiones and staging posts mutationes, which together made up the cursus publicus, the government post (cf. Casson 1994: 182-190). Some of

4

See the discussion of the date of the Palestinian section of the Peutinger Table in Isaac & Roll 1982: 9-10, which they summarise: “…it is clear that the Palestinian section of the Tabula Peutingeriana reflects the situation from Hadrian till Septimius Severus”. The name of Lydda, however was apparently hard to dislodge and appears in later, definitely C4th documents as well, such as the lists from the Theophanes archives (322 or 323, see below). Roberts, the editor of the archive, notes “…it is interesting to find the old Jewish name instead of Diospolis, the name given to the new town after the destruction of the old settlement in the Jewish War”. In fact Lydda became Diospolis under the Severans, but by the C3rd was ‘...a town with a considerable Jewish population perhaps a majority” (Isaac 1990: 116, 359). It appears as Lydda in the C3rd Antonine Itinerary, the Itinerarium Burdigalense (600.3) as civitas Lidda, the Expositio totius mundi (xxxi) as Lydda as well as in Jerome Epist. 108. Perhaps the large number of Jews influenced the retention of the old

1

For earlier roads and milestones, see Isaac 1978: 46-60; Graf, Isaac & Roll 1992: 782-787. A more detailed map is to be found in Tsafrir, Di Segni & Green 1994. 2 I am grateful to Prof. Roll for permission to use his map. 3 The author belongs to a team headed by Profs. B. Isaac and I. Roll which is currently investigating Roman roads and milestones in Judaea; cf. Fischer, Isaac & Roll 1996.

243

Limes XVIII

Palestine is reproduced in Fig. 2 with the west to the top. It shows Helya Capitolina antea dicta Hierusalem, other towns and landmarks, and the road system. In order to make things clearer I have transferred this information to a map of the Roman roads of Palestine (Fig. 3), based on Roll’s map (Fig. 1) marking the roads which appear on the Peutinger map, although Fig. 3 shows the C4th form of city names. It thus becomes clear that the Peutinger map shows a north-south road hugging the coast from Tyre in the north via Ptolemais and Thora (Dora), to the provincial capital of Caesarea. From here the road continues via Apolloniade,5 Ioppe (present day Jaffa) to Iamnia (Talmudic Yavneh), then to Ascalon and south to Egypt. There is also an alternative route further inland via Luddis (Lydda/Diospolis). Otherwise the important nodal points for roads, the capita viarum, are of course, Caesarea and Jerusalem.

must have used to get so far each day, Roberts and Turner, who edited the archive, concluded that he must have been entitled to the transport facilities of the cursus publicus but not to free accommodation. His route, marked on Fig. 5, is identical with the main route of Itinerarium Antonini. It too avoids the coast south of Caesarea. The anonymous traveller from Bordeaux in France, who came to Palestine in the year 333, is the first Christian pilgrim to have left a record of the journey.7 As well as a list of places along the route with the distances recorded between them – mostly wrongly – the pilgrim added various details about Christian and other sites along the way. His – or her – route is shown on Fig. 6. Once again the pilgrim does not use the western coastal road via Joppa when returning to Caesarea, but the road via Lidda/Diospolis and Antipatris. The Peutinger map and the Itinerarium Antonini give details only of cities and mansiones along the routes, but the Bordeaux pilgrim gives fuller details for the roads travelled, and often includes the smaller mutationes as well. However, the pilgrim only records mutationes and mansiones along one route in Palestine, which I have marked thickly in Fig. 6 – the coast road from the north as far as Caesarea, and then the inland road as far as Antipatris.

Although there are no other extant maps of Palestine before the C6th, information about the roads used in the province can also be found in itineraries – lists of places along the roads, prepared for travellers, noting stations of the cursus publicus. The mid-C3rd Itinerarium Antonini appears to have been compiled as an aid for tax collection.6 It includes places along main routes and secondary routes in different provinces of the empire. These roads have been marked on Fig. 4 as they appear in the Palestine section of the Itinerarium Antonini. It is clear from this that the main route here, which may be presumed to coincide with the official cursus publicus, does not follow the coast south of Caesarea as in the Peutinger map, but takes the alternative inland detour via Antipatris and Lydda/Diospolis. On this secular itinerary Jerusalem appears only in the nexus of secondary routes.

Information of a rather different sort about the roads of Byzantine Palestine can be found in the work of Eusebius, bishop of Caesarea at the beginning of the C4th, who compiled an Onomasticon, a list of biblical place names with their contemporary identifications (Thomsen 1903: 97f.; Melamed 1932; 1933; Noth 1943; Wolf 1964: 77; Barnes 1975). Eusebius, living in the Roman administrative centre for the province, seems to have had access to official documentation about the province – he notes the presence of Roman army units (Isaac 1996). He uses the road network and the mile-stations along the roads as a kind of grid reference system for locating sites all over the Holy Land. Sites are typically described as being on the road to x, n miles from y. His roads have been marked on Fig. 7 taking a minimalist approach ie. only marking roads where he uses a verb of travelling, since the supposition is that in that case there was a road to be travelled on, as opposed to the cases when he merely says that a site lies n miles from another. All the roads thus mentioned by Eusebius correspond to official highways marked by milestones but they give only a very partial picture of the roads of the province, for some of the connections of roads within the network are lacking. No road is recorded along the coast south of Caesarea (Thomsen 1903: 170).

The Itinerarium Antonini provides only a bare list of places and the distances between them. Our next document preserves a little more. In the year 322 or 323 a scholastikos called Theophanes travelled from Egypt to Antioch via Palestine, and back again (Roberts & Turner 1952; Moscadi 1970; Rees 1968; Schwabe 1954; Alt 1954; Weingarten & Fischer 2000). A whole archive of documents relating to his journey has been found in Hermopolis in Egypt, recording his stops on the way there and back, what he bought to eat at each stop, and how much he paid for it (Cadell 1989: 315-323; Bagnall 1993: 271-272). Since he records the price of his board and lodging, but has no details of any charges for the horses he Jewish name or perhaps the pilgrims at least, used it because of its NT associations (Acts 9. 32, 38), although the pilgrim from Bordeaux does not actually mention the Christian connections of the town, unlike Jerome, who does so in a rather confused manner (Epist. 108 and see below). 5 Apollonia-Arsuf has been excavated by Roll since 1977 (Roll 1989). Excavations are still on-going and Prof. Roll informs me that there is a gap in the evidence of settlement at this site between the C3rd and C6th CE. For a bibliography on the other sites mentioned, see Tsafrir, Di Segni & Green 1994. 6 Antonine Itinerary – Palestine section, see Cuntz 1990: 21, 23. The introduction by Rivet 1970 is still useful: cf. Van Berchem 1973; Reed 1978.

Eusebius’ Onomasticon was translated and up-dated by the end of the C4th by Jerome. When he first arrived in Palestine, Jerome and his friend Paula made a pilgrimage 7

Jones 1990: 832-833 points out that this is a much fuller list of the smaller mutationes than the It. Ant. and the Peutinger Table, which generally do not mention the latter, citing just cities and mansiones as a rule. See now Douglass 1996; Weingarten 1999; Limor 1998: 19-37, map p.29.

244

Susan Weingarten: Road use in late antique Palestine

around the holy places which he describes in his Epist. 108 (Wilkinson 1974: 245f.). A comparison of this letter with Jerome’s translation of the Onomasticon makes it clear that Jerome up-dated Eusebius’ information about sites he actually visited himself on his pilgrimage.8 Jerome provides the final evidence for the lack of use of the coastal road from Caesarea to Iamnia via Joppa in this period. His route as recorded in Epist. 108 is shown in Fig. 8. He writes that they set out from Antioch and then took the coastal road via Dora as far as Caesarea. From Caesarea they travelled south via Antipatris to Diospolis/Lydda, and then proceeded to Joppa, along the east-west Jerusalem-Joppa road, and then returned the same way they came to Diospolis and carried on towards Jerusalem (Epist. 108.8 repetitoque itinere). Since he writes that they had to retrace their footsteps, I think it can be safely assumed that they would have gone along the western coastal road had this been possible, but since it was not, they took the inland road like all other recorded travellers of this period.

The pilgrim Theodosius, whose information dates to some time after 518 (Theodosius de situ terrae sanctae 4 – CCSL 175: 116; Tsafrir 1986: 134-135; Limor 1998: 169f.), presents a series of itineraries all starting from Jerusalem, marked on Fig. 10. These included the coastal road from Joppa to Caesarea. The same is true of the account of the pilgrim from Piacenza around 570.10 Of course with coastal towns the road was not always the only possible means of connecting them: communication was also possible by sea. For the coast of Palestine we have evidence of this from the Mishnah, redacted at the end of the C3rd. Mishnah Nedarim (iii.6) differentiates between real seafaring and mere coastal traffic ‘between Acco (Ptolemais) and Jaffa (Joppa). Thus it is now clear that the main international north-south route from Syria to Egypt via Palestine veered inland from Caesarea via Antipatris and Diospolis/Lydda, and that only on this route there were the facilities of the cursus publicus. Christian pilgrims, unlike officials, were not bound to use this route, but nonetheless they did so. It is possible to suggest a reason for the apparently total disuse of the coastal road in the C4th? It is known that sand carried up the coast from the Nile Delta by marine currents is deposited along the Palestine coast, and this has caused problems up to the present day. Josephus writes (Ant. Jud. XV.333) that one of the reasons Herod built a new harbour at Caesarea was that the ports of Dora and Joppa were continually blocked by sand. After the building of Herod’s harbour at Caesarea there is evidence that sands were deposited south of the city. Recent archaeological work by teams from Haifa University and the Museum of London seem to confirm a gradual silting up of the artificial harbour at Caesarea in late antiquity too (Raban 1996: 657). None of the pilgrim accounts show travellers arriving in Palestine at Caesarea, the provincial capital, or Joppa, the nearest port to Jerusalem, even though sea travel was far cheaper than land travel in antiquity. The ports used by Jerome and Paula are Seleucia, the port of Antioch, in the north and Gaza in the south. Around the year 502, the emperor Anastasius undertook a full-scale restoration programme of the harbour of Caesarea because of silting up, as recorded by Procopius of Gaza.11 Blown sand also increasingly caused sand dunes which covered wide areas. During the C4th, then, the coastal road seems to have been out of use and the harbour of Caesarea little used other than for local traffic, and it may well be that this was caused by the encroachment of sand.

The vita Porphyrii by Mark the Deacon, probably written around the beginning of the C5th, describes a journey made by Porphyry from Gaza to Caesarea and back. He too obviously took the inland route as on his way back he stayed overnight at Diospolis/Lydda (Grégoire & Kugener 1930: 17). So far we have seen a general picture of the lack of use of the coastal road in the C4th. There is further parallel evidence for this picture from a C4th pagan document, the Expositio totius mundi et gentium, which gives details of cities all over the Roman empire, but concentrates on the east (Rougé 1966; Stern 1980: 495f.; Drexhage 1983). Although the Expositio mentions a string of cities along the Syro-Palestinian coast, from Seleucia, the port of Antioch, southwards through Byblos, Sidon, Sarepta and Tyre to Ptolemais and Caesarea, it does not mention any town on the coast between Caesarea and Ascalon. The picture of important inland towns, however, corresponds to the picture obtained from the pilgrim accounts – Scythopolis, Jericho, Diospolis/Lydda and Eleutheropolis are all mentioned by both the Expositio and by pilgrims. The Expositio, being a pagan source interested mainly in trade, does not mentioned Jerusalem.9 We should also note that Diospolis/Lydda appears as the important town of Lod in Talmudic sources just prior to this period too (Oppenheimer 1988; Schwartz 1991; Fischer, Isaac & Roll 1996).

From the evidence of the sources discussed above, routes By the beginning of the C6th, the road along the coast from Caesarea was back in use, at least as far as Joppa. 10

This anonymous pilgrim is sometimes wrongly referred to as Antoninus of Placentia – Itinerarium Antonini Placentini 46 (CCSL 175: 152.cf. Wilkinson 1977: 84f; Milani 1977; Limor 1998: 209-246. 11 Procopius Pan. In Emp. Anas xix (PG 87.iii.2817). It may be noted in passing that the building of the modern port at Ashdod, north of Ascalon, in the last century, has caused present-day problems with sand deposits along the coast and problems are presently forecast with the proposed expansion of the modern port of Gaza.

8 A detailed comparison of the information in Jerome’s Epist. 108 and his translation of Eusebius’ Onomasticon forms part of the author’s doctoral thesis; Weingarten 2000: 236-249. 9 This is, however, not only a characteristic of pagan sources. The C4th Itinerary from the Garden of Eden to the country of the Romans (Rougé 1966:346f) does not mention Jerusalem either !

245

Limes XVIII

used in C4th Palestine did not include the section of the coastal road from Caesarea to Iamnia and Ascalon, which does not appear in any literature between the C3rd evidence of the Peutinger map and the C6th travel documents of Theodosius and the Piacenza pilgrim. This is confirmed by what we know of the towns in this period. It is the inland towns of Neapolis, Lydda/Diospolis and Eleutheropolis which are repeatedly mentioned, and we know from Talmudic sources that Lydda rose to considerable prominence around this time. In contrast, when it comes to the towns on the coast, there is a break in the continuity – Apollonia is not mentioned at this time and the excavations there by Roll show a gap in occupation of the site between the C3rd and C6th. As we have seen, the only person we know of to visit Jaffa by road is Jerome. The fact that he records that he travelled to Jaffa from Lydda and then went back the same way as he came, is the final piece of evidence that points to the disuse of the coastal road in this period. We have suggested that archaeological evidence from the excavations in Caesarea, preliminary and incomplete as it is, makes it possible to propose that the reason for the disuse of the road was the encroachment of sand, a periodic problem along this coast from the time of the building of Caesarea. This theory is lent some weight by the C6th source which records Anastasius as clearing the harbour at Caesarea.

Münstersche Beitrage z. antiken Handelsgeschichte 2.1: 3-41. Fischer M., Isaac B. & Roll I. 1996 Roman roads II: The Jaffa-Jerusalem roads. British Archaeological Reports Int. Ser. 628. (Oxford). Graf D.F. Isaac B., Roll I. 1992 Roman roads. In D.N. Freedman (ed.) The Anchor Bible Dictionary Vol. 5. (New York): 782-787. Grégoire H. & Kugener M.-A. 1930 Marc le Diacre: Vie de Porphyre, évêque de Gaza. (Paris). Isaac B. 1978 Milestones in Judaea from Vespasian to Constantine. Palestine Exploration Quarterly 110: 4660. Isaac B. 1990 The limits of empire. The Roman empire in the east. (Oxford). Isaac B. 1996 Eusebius and the geography of Roman provinces. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 153-167 = The Near East under Roman rule: Selected papers. (Leiden 1998): 284-306, 307-309. Isaac B. & Roll I. 1982 Roman roads in Judaea I: the Legio-Scythopolis road. British Archaeological Reports Int. Ser. 141. (Oxford). Jones A.H.M. 1964 The later Roman empire. (Oxford repr. 1990). Lee A.D. 1993 Information and frontiers: Roman foreign relations in antiquity. (Cambridge). Levi A. & M. 1967Itineraria picta. Contributo allo studio della Tabula Peutingeriana. (Roma). Limor O. 1998 Holy Land travels: Christian pilgrims in antiquity. (Jerusalem) – in Hebrew. MacEachren A.M. 1986 A linear view of the world: strip maps as a unique form of cartographic representation. The American Cartographer 13: 7-25. MacEachren A.M. & Johnson G.B. 1987 The evolution, application and implications of strip format travel maps. The Cartographic Journal 24: 147-158. Melamed E. Z. 1932 Eusebius’ Onomasticon. Tarbiz 3: 314-327, 393-409 - in Hebrew. Melamed E. Z. 1933 Eusebius’ Onomasticon. Tarbiz 4: 78-96, 249-284 - in Hebrew. Milani C. 1977 Itinerarium Antonini Placentini: Un viaggio in Terra Santa del 560-570 d. C. Scienze filogiche e letteratura 7. (Milan). Moscadi A. 1970 Le lettere dell’archivo di Teofane. Aegyptus 50: 90-154. Noth M. 1943 Die topographischen Angaben im Onomasticon des Eusebius. Zeitschrift des Deutschen Palästina-Vereins 66: 32-63. Oppenheimer A. 1988 Hebrew Union College Annual 59: 115-136. Raban A. 1996 The inner harbour basin of Caesarea: archaeological evidence for its gradual demise. In A. Raban & K.G. Holum (edd.) Caesarea Maritima: a retrospective after two millenia. (Leiden): 628-666. Reed N. 1978 Patterns and purpose in the Antonine Itinerary. American Journal of Philology 99: 228-254. Rees B.R. 1968 Theophanes of Hermopolis Magna. Bulletin of the John Rylands Library 51: 164-183. Rivet A.L.F. 1970 The British section of the Antonine Itinerary. Britannia I: 34-39. Roberts C.H. & Turner E.G. 1952 Catalogue of the Greek and Latin papyri in the John Rylands Library, Manchester vol. IV. (Manchester). Roll I. 1995 Roads and transportation in the Holy Land in the

Careful reading of the written sources can thus contribute much to the archaeological picture of late antique Palestine and, as here, help to date the use and disuse of roads and the development and decline of towns. Bibliography CCSL Alt A.

Corpus Christianorum Series Latina 1954 Stationen der römischen Hauptstrasse von Ägypten nach Syrien. Zeitschrift des DeutschePalästina Vereins 70: 154-166. Bagnall R. 1993 Egypt in late antiquity. (Princeton). Barnes T. 1975 The composition of Eusebius’ Onomasticon. Journal of Theological Studies (NS) 26: 412-415. Bertrand A.C. 1998 Stumbling through Gaul: maps, intelligence and Caesar’s Bellum Gallicum. Ancient History Bulletin 11: 107-122. Brodersen K. 1995 Terra cognita: Studien zur römischen Raumerfassung. (Hildesheim). Cadell H. 1989 Les archives de Théophanes d’Hermoupolis: Documents pour l’histoire. In L. Criscuolo & G. Geraci (edd.) Egitto e storia antica dall’ellenismo all’eta araba. (Bologna). Casson L. 1994 Travel in the ancient world. (2nd edit; Baltimore). Cuntz O. (ed.) 1990 Itineraria Romana I: Itineraria Augusti.. (rev. G. Wirth. Stuttgart). Dilke O.A.W. 1985 Greek and Roman maps. (London). Douglass L. 1996 A new look at the Itinerarium Burdigalense. Journal of Early Christian Studies 4: 313-333. Drexhage H.-J. 1983 Die Expositio totius mundi et gentium, Ein Handelsgeographie aus dem 4. Jahrhundert n. Chr. eingeleitet, übersetzt und mit einführender Literatur (Kap. xxii-lxvii) versehen.

246

Susan Weingarten: Road use in late antique Palestine

early Christian and Byzantine times. Akten des XII. Internationalen Kongress für christliche Archäologie teil II (Jahrbuch für Antike und Christentum Ergänzungband 20.2 (Münster): 1166-1170. Roll I. & Ayalon E. 1989 Apollonia and southern Sharon: model of a coastal city and its hinterland. (Tel Aviv) – in Hebrew. Rougé J. (ed.) 1966 Expositio totius mundi et gentium. (Paris). Schwabe M. 1954 Documents of a journey through Palestine in the years 317-323 CE. Eretz Israel 3: 181-185 – in Hebrew. Schwartz J. 1991 Lod (Lydda) Israel: from its origins through the Byzantine period 5600 BCE – 640 CE. British Archaeological Reports Int. Ser. 571. (Oxford). Stern M. 1980 Greek and Latin authors on Jews and Judaism II. (Jerusalem). Talbert R.J.A. 1990 Rome’s empire and beyond: the spatial aspect. In E. Herman (ed.) Gouvernants et gouvernés dans l’imperium Romanum (IIIe av. J-C. – 1er ap. J.C.). Cahiers des études anciennes 26: 215-223. Thomsen P. 1903 Palästina nach dem Onomasticon des Eusebius. Zeitschrift des Deutschen- Palästina Vereins 26: 97ff. Tsafrir Y. 1986 The maps used by Theodosius: on the pilgrim maps of the Holy Land and Jerusalem in the sixth century CE. Dumbarton Oaks Papers 40: 134-145. Tsafrir Y., di Segni L. & Green J. 1994 Tabula Imperii Romani: Iudaea/Palaestina. (Jerusalem). van Berchem D. 1973 L’itinéraire Antonin et le voyage en orient de Caracalla 214-5. Comptes rendus des séances de l’académie des inscriptions et belles-lettres: 123126. Weingarten S. 1999 Was the pilgrim from Bordeaux a woman ? A reply to Laurie Douglass. Journal of Early Christian Studies 7: 291-297. Weingarten S. 2000 Jerome’s world: the evidence of his Saints’ Lives. (Unpubl. Ph.D thesis, Tel Aviv). Weingarten S. & Fischer M. 2000 Iamnia-Abella-Ibelin: a new look at the Theophanes archive. Zeitschrift des Deutschen Palästina-Vereins 116: 49-56. Wilkinson J. 1974 L’apport de Saint Jérôme à la topographie. Revue Biblique 81: 245-257. Wilkinson J. 1977 Jerusalem pilgrims before the Crusades. (Jerusalem). Wolf C. 1964 Eusebius of Caesarea and the Onomasticon. Biblical Archaeologist 27.3: 66-96.

247

Limes XVIII

Fig. 1. Roads in Palestine in late antiquity.

248

Fig. 2. Palestine on the Tabula Peutingeriana

Susan Weingarten: Road use in late antique Palestine

249

Limes XVIII

Fig. 3. Roads on the Tabula Peutingeriana.

250

Susan Weingarten: Road use in late antique Palestine

Fig. 4 Roads on the Itinerarium Antonini.

251

Limes XVIII

Fig. 5. Theophanes’ route (AD322 or 323).

252

Susan Weingarten: Road use in late antique Palestine

Fig. 6. The route of the Pilgrim of Bordeaux (AD333).

253

Limes XVIII

Fig. 7. Roads in Eusebius’ Onomasticon.

254

Susan Weingarten: Road use in late antique Palestine

Fig. 8. The route of Jerome and Paula (AD385-386).

255

Limes XVIII

Fig. 9. Cities mentioned in the Expositio totius mundi et gentium.

256

Susan Weingarten: Road use in late antique Palestine

Fig. 10. Routes in Theodosius (after AD518).

257

Limes XVIII

258

Roman military forts at Zeugma. A preliminary report Martin Hartmann & Michael A. Speidel In the early months of the year 2000 flooding of the Euphrates valley at the Birecik dam began. Large parts of the Greco-Roman city of Zeugma, all of neighbouring Apameia and many other sites of great historical interest were thereby forever destroyed. It was our scope as a joint venture of a privately sponsored Swiss team under the patronage of the universities of Bern and Lausanne together with the Gaziantep Archaeological Museum to rescue as much archaeological and historical information as possible of ancient Zeugma's great military history under the Roman empire. For this reason, the archaeological investigations we undertook in the years from 1996 to 1999 were confined to those areas which were to be flooded in early 2000 (cf. Hartmann, Speidel & Ergeç 1999; Hartmann, Rüger, Speidel & Ergeç 2000; 2001).

many decades during the C2nd and C3rd AD soldiers of legio IV Scythica worked in the quarries of Arulis. Tile stamps of that legion and military equipment found in Zeugma, as well as epigraphic and numismatic evidence have further reinforced the hypothesis that Zeugma or its surroundings were the site of IV Scythica's legionary fortress (Speidel 2000: 330ff.; 1998: 167. cf. French 1998: 134; Bishop 1998: 135-137; Ergeç & Önal 1997: 442 fig. 22; Speidel 2001: 113-115). Jörg Wagner located legio IV Scythica's permanent base within the western hills of the ancient city (Wagner 1976: 146; cf. his map II: Fig. 3) The 1996 survey and the 1997 soundings Our first step in 1996 was thus to investigate the topography of the site. It was soon clear, that the site identified by Wagner was wholly unsuitable for the construction of a legionary fortress. Although the majority of the stamped tiles of the fourth legion had been found there, this location was in the midst of hills and terraces where there is no flat area nearly large enough to have accommodated an entire legion. It remains possible, however, that this area was the site of a smaller military fortress of perhaps the C3rd or C4th AD. To the east, however, beyond the village of Belkis the upper terrace of a large plain immediately appeared very suitable for the construction of a Roman military camp. An investigation of that region revealed a very distinct field covered with pieces of Roman ceramic, fragments of tiles etc. Russian satellite photographs (KVR 1000, 22.5.1992) were acquired that showed the typical outlines of a large Roman military camp in precisely the same area (Fig. 4). A closer analysis showed the outlines to cover an area of c.11-12ha, and even appeared to indicate two super-imposed fortresses.

The Roman army at Zeugma For various reasons Zeugma played an important military rôle during the first three centuries AD (cf. Wagner 1976; Kennedy 1998). One main reason is Zeugma's location on the crossing of two major lines. The first line is the River Euphrates which for nearly 250 years was the border of the imperial Rome (Velleius II: 101; Strabo Geog. VI.4.2; Tacitus Anns. II.58; VI.37; XV.17; Suetonius Cal. 14 etc.). Zeugma was thus a frontier city. The second line is the ancient highway that linked Mesopotamia with Anatolia and the Mediterranean coast (Fig. 1). This highway is described in our sources as having been one of the Parthians favoured invasion routes (Tacitus Anns. XII.12; Frontinus Strat. I.1.6; Dio XLIX.19). It crossed the River Euphrates at Zeugma. Similar such locations at Samosata and at Melitene had permanent legionary fortresses since early in the reign of Vespasian. Philo of Alexandria (Legatio ad Caium 31. 207 and 34. 259) clearly implies that substantial parts of the Roman army of Syria were stationed on the banks of the Euphrates probably since the reign of Tiberius. Furthermore, Tacitus (Anns. XII.12.3) reports that in AD49 a temporary Roman military camp was built apud Zeugma (Fig. 2). When Zeugma became associated with a permanent legionary base is unknown, though it may have been well before AD66. For up to that year, when Vespasian made it part of his army against the Jewish rebels, legio X Fretensis guarded the River Euphrates, which probably means it was stationed at or near Zeugma (Josephus Bell. Jud. VII.1.3).

With this information at hand we decided to begin archaeological investigations in 1997 in the described area which was located directly on the building site of the dam construction (Fig. 5). We dug seven trial trenches with a total length of c.650m. The trenches were c.1m wide, between 1.8 and 2m deep and reached into sterile soil throughout. Within the described area the results from the trenches revealed two distinct phases of Roman occupation, the earlier of which marks a longer occupation than the later. Traces of internal structures in mud brick construction and of gravelled layers were detected. A ditch of the later fortress was discovered in the north end of Trench 3. The location and the orientation of the discovered features appeared to be identical with the outlines visible on the Russian satellite photographs.

Following A. v.Domaszewski, R.K. McElderry and E. Ritterling, it has been generally assumed that Zeugma was the permanent base of legio IV Scythica (v. Domaszewski 1909: 198 n.1; McElderry 1909: 48-49; Ritterling 1925: 1560). Literary and epigraphic clues led to this conclusion. Most important were perhaps the inscriptions in the quarries of Arulis (Ehnes), only c.15kms upstream from Zeugma. These inscriptions show that for

The 1998 campaign It was our objective for 1998 to confirm these results and to gain further information on the size and shape of the

259

Limes XVIII

military forts (Fig. 6). We therefore decided to uncover by machine a large field of 10 x 18m (F2/1 & 2) east of Trench 3 in which the ditch had been discovered in 1997. After the removal of another layer by hand a row of postholes on a west-east axis became visible. The distance between them varied between 3.6m and 4.2m. To the south of the row of postholes a dense layer of clay/mud with very little internal structuring was observed. There can be little doubt that this was the remains of a wall construction (Figs. 7 & 8).

northern and in the southern profile of that field. Field F5 was uncovered in order to find the place where the northern ditch of the younger fortress met the eastern ditch. Only after careful cleaning was it possible to recognise the faint traces of both ditches in this area. A number of round shaped patches of slightly different colour parallel to the eastern ditch may indicate the location of wall construction. Conclusions

At the eastern end of Field 2 a trial trench (S8) of 1m width and 50m length was dug towards the north. Just south of where it crossed trial Trench 4 the northern ditch of the younger camp was clearly visible (Fig. 9). It corresponds perfectly both with the ditch observed in trial Trench 3 during the previous campaign, as well as with the row of postholes in field 2. In the region of field 2/2 mudbricks were observed in the western profile of the trench, both in horizontal and vertical positions (Fig. 10). There were, however, no traces of plaster. At roughly 0.8m a yellowish layer of clay ran almost horizontally towards the north. It is certain that this layer belongs to the older camp.

Both the northern and the eastern fortification of two superimposed Roman military camps, including traces of the wall construction, and the north-eastern corners of both fortresses were identified. We are pleased to conclude that the outlines initially detected on the satellite photograph largely correspond with the archaeological findings in the field. Both Roman layers have been, over the centuries, badly damaged by erosion. Material from the archaeological layers has been washed away and, at the same time, erosion material from the hills above was washed in. Both phenomena made the interpretation of the findings extremely difficult. Our observations suggest that the walls were constructed in part from dried mud bricks and partly from condensed mud. The investigated internal structures, however, appear to have been constructed from dried mud bricks only. The excavated findings, though they are far from numerous, confirm the dating of both military camps to the first half of the C1st AD (Fig. 12). Both camps were occupied for a very limited period of time only.

In order to localise the northern ditch of the older camp which is clearly visible on the satellite photograph, a further trial trench (S9) of 25m length was dug east of Trench 8 in a northerly direction. In the southern region of the western profile several traces of constructions we observed. At their northern end the ditch of the older camp was clearly visible. Faint traces in the eastern profile may even belong to the corresponding wall.

It was soon obvious that we had not discovered the permanent fortress of Zeugma's legio IV Scythica. The two superimposed Roman military forts on the upper terrace of the plain east of the village Belkis were temporary camps only. It is not impossible to identify the younger camp as the one mentioned by Tacitus to have been built apud Zeugma in AD49, when the emperor Claudius tried to help a friendly Parthian prince to cross the Euphrates in order to seize power in Parthia (Tacitus Anns. XII.11ff.). A similar situation is on record for the year AD35, when Tiberius tried the same (Josephus Ant. Jud. XVIII.4.5. Tacitus Anns. VI.31ff.). The older temporary fort may, therefore, have been constructed during that Roman attempt to establish a friendly ruler in Parthia. It is, however, equally possible, that other reasons led to the construction of either of these forts (eg. Tacitus Anns. XV.3; 9). While the remains of these camps were flooded in spring 2000, legio IV Scythica's fortress still awaits discovery.1

The 1999 campaign Based on the observations made in 1998, it was our objective for 1999 to find the precise location of the eastern rampart, and to excavate the north-eastern corners both of the older and the younger fortress. In order to find the exact course of the northern ditch of the earlier fortress a trench (S10) of c.20m length running north-south was dug to the east of Trench S9. The northern ditch was visible both in the eastern and in the western profiles of Trench S10, and corresponded perfectly with the earlier findings of Trench S9 (Fig. 11). The course of the northern ditch of the earlier fortress was thereby identified, and the search for the eastern ditch began. Two trial trenches (S10 and 11) running west-east were therefore dug, c.5m and 50m south of Trench S10. After they had been cleaned both profiles revealed the eastern ditch. Their fillings contained some dried mud bricks. Our last objective was to find the north-eastern corners of both fortresses. For that purpose three large fields were uncovered (F3-5). In the northern field (F3) faint traces of the north-eastern corner of the older fortress were observed. The filling of the ditches contrasted with its surroundings not by colour but by texture only. Unfortunately, these traces could not be observed in Field F4, although they were relatively clearly visible both in the

1

The statement by Ergeç, Önal & Wagner (2000 106) that the hitherto unknown location of legio IV Scythica's fortress has been identified is misleading.

260

Martin Hartmann & Michael A. Speidel: Roman military forts at Zeugma. A preliminary report Kommagene. (Mainz): 106 French D. 1998 Miscellaneous artefacts. In D. Kennedy (ed.) The twin towns of Zeugma on the Euphrates. Rescue work and historical studies. Journal of Roman Archaeology Supplementary Series 27: 129-138. Gregory S. 1996 Roman military architecture on the eastern frontier. Vol. 2. (Amsterdam). Hartmann M., Speidel M. & R. Ergeç R. 1999 In XX. Kazi Sonuçlari Toplantasi II. Cilt. 1998 : 417ff. Hartmann M, Rüger C., Speidel M. & Ergeç R. 2000 In XXI. Kazi Sonuçlari Toplantisi II. Cilt. 1999: 337ff. Hartmann M, Rüger C., Speidel M. & Ergeç R. 2001 In XXII. Kazi Sonuçlari Toplantisi II. Cilt. 2000: 255ff. Kennedy D. (ed.) 1998 The twin towns of Zeugma on the Euphrates. Rescue work and historical studies. Journal of Roman Archaeology Supplementary Series 27. McElderry R. K. 1909 The legions on the Euphrates frontier. Classical Quarterly 3: 68-75 Ritterling E. 1925 Legio RE XII. Cols. 1186-1837. (Stuttgart) Speidel M.A. 1998 Legio IIII Scythica, its movements and men. In D. Kennedy (ed.) The twin towns of Zeugma on the Euphrates. Rescue work and historical studies. Journal of Roman Archaeology Supplementary Series 27: 163204. Speidel M.A. 2000 Legio IV Scythica. In Y. Le Bohec (ed.) Les légions de Rome sous le Haute-Empire. Actes du congrès de Lyon (17-19 septembre 1998). (Lyon): 327337. Speidel M.A. 2001 (forthc.) Legio operosa felix. In E. Dabrowa (ed.) Electrum 5: 113-115. Wagner J. 1976 Seleukeia am Euphrat/Zeugma. (Wiesbaden).

The survey of 2000 A further result of our investigations of 1996 to 1999 was that the fortress of legio IV Scythica was not situated on the banks of the Euphrates. The only area in the immediate vicinity of Zeugma suitable for such a location would have been the very plain where the temporary camps of the C1st AD had been constructed. However, no traces of any other Roman military installations were detected there. Earlier suggestions, that the fortress was located within the hills and terraces of ancient Zeugma (Wagner 1976: 146; cf. his map II: Fig. 3) or on the other side of the River Euphrates (Gregory 1996: 129-131; Kennedy 1998: 37) – ie. beyond the limits of provincia Syria and beyond the official borders of the empire – must be ruled out. In our opinion a suitable location would include the following features: the vicinity to Zeugma, a flat area of at least c.15–20ha, quick access to the city and its river-crossing, and integration into the road system. Note The archaeological campaigns were under the supervision of Dr. Rifat Ergeç of the Gaziantep Museum and under the direction of Dr. Martin Hartmann, Project leader of the Swiss Team. The team also included Mehmet Önal, Prof. Christoph Rüger, Werner Rutishauser and Dr. Michael Speidel, co-project leader. We would like to thank the following institutions and associations for their help and their support of our project: Birecik Construction Consortium, Gerda Henkel Stiftung, Goethe Stiftung für Kunst und Wissenschaft, Hochschulstiftung der Burgergemeinde Bern, Hochschulstiftung der Universität Bern, Institut für Metallurgie und Metallforschung der ETHZ, Schweizerischer Nationalfonds zur Förderung der wissenschaftlichen Forschung, UBS Kulturstiftung. We also received a great deal of help from the site managers of the Birecik consortium, in particular from Messrs. Ebner, Hölzl, Schiefer and Demir. We are truly grateful for all their support. Bibliography Bishop M. C. 1998 Miscellaneous artefacts. In D. Kennedy (ed.) The twin towns of Zeugma on the Euphrates. Rescue work and historical studies. Journal of Roman Archaeology Supplementary Series 27: 129138. Domaszewski v. A. 1909 Abhandlungen zur römischen Religion. (Leipzig). Ergeç R. & Önal M. 1997 Belkis/Zeugma Roma Hamami ve Kompleksi Kurtarma Kazizi. VIII. Müze Kurtarma Kazilari Semineri. (Ankara). Ergeç R., Önal M. & Wagner J. 2000 Seleukeia am Euphrat/Zeugma. In J. Wagner (ed.) Gottkönige am Euphrat. Neue Ausgrabungen und Forschungen in

261

Limes XVIII

Fig. 1. Eastern Mediterranean, south-east Turkey and northern Syria.

Fig. 2. The Plain beyond the village of Belkis.

262

Martin Hartmann & Michael A. Speidel: Roman military forts at Zeugma. A preliminary report

Fig. 3. Zeugma’s western hills.

Fig. 4. Satellite photograph with the outlines of military camps (Satellite photo: KVR – 1000; 22.5.1992).

263

Limes XVIII

Fig. 5. Zeugma, the position of the dam and the area of investigation.

Fig. 6. Investigation - Area with the position of the trenches and the outlines of the early Roman military camps.

264

Martin Hartmann & Michael A. Speidel: Roman military forts at Zeugma. A preliminary report

Fig. 7. Wall-postholes and ditch of the later fortress.

265

Limes XVIII

Fig. 8. Posthole of the wall in area 2.

Fig. 9. Ditch of the later fortress in Trench 8.

266

Martin Hartmann & Michael A. Speidel: Roman military forts at Zeugma. A preliminary report

Fig. 10. Shaped mudbricks in area 2/Trench 8.

Fig. 11. Trench 10: Shaped mudbricks in the filling of the northern ditch.

267

Limes XVIII

Fig. 12. Terra Sigillata from trenches 9 and 10.

Fig. 13. Satellite photograph with the outlines of the early Roman military camps and the assumed legionary fortress (Satellitephoto: KVR – 1000; 22.5.1992).

268

Preliminary report on the Roman period at Kamid el-Loz (Lebanon) Lars Petersen In 1997, the excavations of the University of Freiburg at Tell Kamid el-Loz, situated in the southern Bekaa Valley, were resumed. The preliminary results of the excavation reveal continuous settlement from the early Bronze Age into the Islamic period. The 1999 and 2000 field seasons confirmed a Roman period settlement on top of the Tell Kamid el-Loz. Architectural remains of a house and a broad glacis were found. The position of the settlement on top of the tell would have allowed to control the entire southern part of the Bekaa Valley. Did the civil settlement also have a military function? At this point, the function of the glacis as part of a fortification system is purely speculative. At present, we do not have much information about the Roman settlement in the southern Bekaa valley. The excavations will continue in summer 2001 – hopefully with results that will contribute answers to some questions about the Romans at Kamid el-Loz.

Tell Kamid el-Loz (Lebanon) was the object of regular excavations by the University of Saarbrücken under the direction of Rolf Hachmann from 1963 until 1981. In 1997 its excavation was resumed by the University of Freiburg under the direction of Marlies Heinz.

farmers” (Geog. XVI.2.18). We know very little more than this about the people who lived at Kamid el-Loz in those days, even its name in Roman times remains unknown. The 1997, 1999 and 2000 field seasons confirmed a Roman period settlement on top of the tell.1 Architectural remains of a house and a broad glacis were found. In the south-western part of the excavated area the rectangular ground plan of a house with three rooms and an L-shaped entrance was discovered. The three rooms are all 2.5m wide. (Fig. 3)

Tell Kamid el-Loz is situated in the southern Bekaa Valley, about 50kms south-east of Beirut and 10kms from the Syrian border (Fig. 1). It is one of the largest and highest settlement hills of the Bekaa Valley, though does not reach the dimensions of the large Syrian and Mesopotamian tells. From north to south the tell has a maximal length of approximately 300m and 240m from west to east. The top of the tell is nearly 950m ASL and 26m above the base of the Bekaa Valley (Fig. 2; Marfoe 1995).

In the first building period (C1st BC/AD), the house had a 10m long, rectangular ground plan with at least 2 rooms. It is not clear whether Rooms 2 and 3 were divided by a wall at this time. The wall foundations were 0.8m wide and consisted of 2 parallel rows of set undressed stones. The entrance of the house was in the south of Room 2.

The earliest known history of the settlement Kamid el-Loz begins in the early Bronze age. The first architectural evidence of settlement has been dated to the middle Bronze age. In the following late Bronze age Kamid el-Loz, then called Kumidi, grew to be a major city in the southern Bekaa. A temple, palace and workshop-area from this time were discovered by the Saarbrücken excavations (Hachmann 1989). As texts from Egypt and Kumidi itself show, the city was seat of the Egyptian administration in the Bekaa Valley (Hachmann 1982). When the Egyptians lost influence and major political changes took place in the area, Kumidi lost its important position. During the Iron age the city regressed to a small village (Heinz 2000: 364).

In the second building period (C1st - C2nd AD), the building was extended. The walls in the west and south of the building were broadened by massive stone foundations. South of Room 2 an L-shaped stone installation has been exposed, preliminarily interpreted as a wall foundation. It consists of a single-course, relatively thin row of uncut stones. The L-shaped stone row stretches 4.5m north-west – south-east, then bends to the north-west for 1.5m, meeting the massive head of Wall D. No interlock was detected. This stone installation may have been a corridorlike cul de sac or simply a wall marking an enclosure. Rooms 2 and 3 were divided by the newly erected Wall D which was 1m wide and had a foundation of 2 set rows of large undressed stones with a filling of smaller stones.

The excavations of the University of Freiburg in 1997, 1999 and 2000 revealed that Kamid el-Loz was also a settlement in the late Hellenistic and early Roman periods (Heinz 2000: 366). The fact that the site was settled during Hellenistic and Roman times sheds new light on its history compared to the results of the excavations carried out by the University of Saarbrücken. The results of the excavation reveal continuous settlement from early Bronze age into Islamic times, when the tell was used as a cemetery. Today, there is a mosque and a modern cemetery on top of the tell.

The house seems to have a domestic character, as indicated by its simple layout (Bonatz & Gilibert forthc.). The pottery found in the house confirms this interpretation. A remarkable non-architectural feature of the building seems to be worth mentioning. Along the foundations of Walls B, E and F, on their external side, three ceramic deposits were excavated, all of which included intact vessels (Fig. 4). These vessels had not been abandoned, but

The written and archaeological material of the southern Bekaa Valley in Hellenistic and Roman times is scant. Strabo writes: “The people of the Bekaa Valley are

1

Marfoe 1995: 196 noted Roman pottery at Kamid el-Loz but no architectural remains of this period.

269

Limes XVIII

carefully deposited, which explains their good condition. The deposition almost certainly took place before the foundations of the building was sealed. Even if the reasons which led to the deposition remain still obscure, they are likely to be related to foundation rituals.

luxury items to everyday commodities (Jennings 1997/8: 111). The large amount of glass-fragments and remains of melted glass indicate glass production somewhere at Kamid el-Loz.

The pottery found in the house dates to the late Hellenistic/early Roman period. Characteristic are the socalled fishplates, dating in the C2nd BC. Also typical are the red and black small plates with engobe and inwards curled rim, dating to the C1st BC. There are also fragments of Megarian bowls (Fig. 5) which are still found in the early Roman period.

The so-called ‘glacis’ (Fig. 7) excavated in 1997, 1999 and 2000 seems to be a unique example of Roman settlement architecture in the Roman Near East. The glacis stretches over the whole area for approximately 35m from the northwest to the south-east. The top of the glacis is limited by a row of larger stones which could be the substratum of a wall whose stones had already been dismantled in antiquity, or which had never been erected. From this point, it starts to drop away remarkably in a northerly direction and then runs down more regularly for about 15m without reaching its ultimate limit, following the natural slope of the area. The glacis consists of 3 or 4 layers of stones, densely packed but without a visible pattern. Of special interest is the chronological relationship between house and glacis (Bonatz & Gilibert forthc). The pottery found between the stones dates this structure to the C2nd or C3rd. The glacis has to be later than the house, as the glacis covers one wall of the house.

Fragments of Common Ware belong to the pottery assemblage found inside the house in 1997 and 1999. Characteristic of the Roman period is the tableware (terra sigillata) represented by fragments of small bowls, dating to the C1st AD. Very few examples of pottery from the C2nd AD have been found. Fragments of Brittle Ware (Red cooking-pot ware) belong to this period. Brittle Ware vessels at Kamid el-Loz are characterized by their ribbed walls, which are common in the C2nd (Hayes 1997). A small ‘industrial’ area to the north-west of the house was excavated in 1997. It seems unlikely that it possessed merely a domestic function, as there is some evidence that the area had been used for handicrafts. Thus slags, fragments of melted glass and so-called burning aids have been found. Burning aids made of clay were used in the pottery firing process and these and a fragment of an incorrectly fired vessel are indications that pottery was produced at Kamid el-Loz during the Roman period. Remains of what probably had been a furnace together with some slags was also uncovered.

The entire area was used as a cemetery in Islamic times, causing much disturbance in the glacis. The excavation area of the glacis was enlarged in 2000 in order to uncover the outer edge of the settled space. One of the results of this excavation is that the glacis turned out to be much larger than expected. Test trenches 30m to the north, 10m to the south and 30m to the east show, that the glacis has not yet been fully uncovered. The excavated area comprises approximately 3000m2 - all covered by the glacis.2 The most interesting question concerning this huge structure is its function. In 1997 the structure was considered to be the interfacing of a Roman street. During the 1999 excavation it was redefined as a broad glacis for a fortification. At this point, the function of the glacis as part of a fortification system is purely speculative. The position of the settlement on top of the tell would have allowed military control of the southern Bekaa Valley and the trade-route from Damascus to the Mediterranean coast. But is the defence construction really a defence wall and a glacis or is it a street running through or aside a large square within the settlement? Or was it a construction to prevent erosion in this area?

Since 1997 over 70 fragments of glass vessels have been retrieved from this area. The fragmentary nature of the assemblage and lack of complete vessels is typical of domestic refuse continually disturbed. It also reflects the fact that glass, along with metals, but unlike pottery, was recycled. The earliest glass at Kamid el-Loz dating to the Roman period represents a continuation of the Hellenistic tradition of making vessels by casting discs of glass over a mould. The Hellenistic and Roman cast glass found in the excavations has been divided into two groups: conical cast bowls with linear-cut grooves, dating to the second half of the C1st BC and ribbed bowls, dating to the C1st BC/AD (Fig. 6)

For now we have for the first time an example of late Hellenistic/early Roman settlement architecture in the

The linear cut bowls typically have a single fairly, shallow cut groove below the rim and one or more equally shallow grooves further down the vessel wall near the beginning of the curve of the base. The ribbed bowls have more variety in both shape and colour than the linear cut bowls. The technology of vessel production changed during this period from time-consuming, labour-intensive work to mass production; the status of glass vessels thus changed from

2

That the glacis might spread out far over the edge of the tell itself has been suggested by Frederic Husseini, Director of the Lebanese Dept. of Antiquities, while visiting Kamid el-Loz during the 2000 field season and analysing the area adjacent to the site in the east. He referred to a stretch of stones running parallel to the road that had been cut through the eastern slope of the tell. This row of stones consists of the same type of stones used for the glacis and is also set up in 3 or 4 layers.

270

Lars Petersen: Preliminary report on the Roman period at Kamid el-Loz (Lebanon) Bekaa Valley. Nonetheless further investigations are still necessary. The excavations will continue next summer hopefully with results that will contribute answers to some questions about the Hellenistic and Roman periods at Kamid el-Loz. Bibliography Bonatz D. & Gilibert A. forthc. The Roman architecture at Kamid el-Loz. Hachmann R. 1982 Die ägyptische Verwaltung in Syrien während der Amarnazeit. Zeitschrift des Deutschen Palästina-Vereins 98: 17-62. Hachmann R. 1989 Kamid el-Loz 1963-1981. Berytus 37: 9187. Hayes J. W. 1997 Handbook of Mediterranean Roman pottery. (London). Heinz M. 2000 Kamid el-Loz. Antike Welt 2000: 359-368. Jennings S. 1997/8 The Roman and early Byzantine glass from the Souks excavations. Berytus 43: 111-125. Marfoe L. 1995 Kamid el-Loz 13. The prehistoric and early historic context of the site. Saarbrücker Beiträge zur Altertumskunde Bd. 41.

271

Limes XVIII

Fig. 1. Map of the Middle East

Fig. 2. Map of Tell Kamid el-Loz (scale 1:1000) excavation areas grey scale.

272

Lars Petersen: Preliminary report on the Roman period at Kamid el-Loz (Lebanon)

Fig. 3. Ground-plan of the house in the second building period (1st/2nd century AD)

Fig. 4. Ceramic Deposit including an intact vessel (1st century BC) found next to Wall B.

273

Limes XVIII

Fig. 5. Fragments of two Megarian Bowls (1st century BC/AD)

Fig. 6. Fragments of ribbed conical cast glass bowls.

Fig. 7. The so-called Glacis to the south in 1999.

274

“. . . ostentasse Romana arma satis . . .” The military factor in Roman-Parthian relations under Augustus and Tiberius Edward Dąbrowa The significance of the military factor in Romano-Parthian relations is regularly stressed by scholars where in fact, this problem has not been discussed in details. The aim of this paper is to present the evidence on the rôle of the military factor under Augustus and Tiberius in their policy towards the Arsacids. The available evidence permits us to formulate several observations: 1) from Augustus onwards the army was a regular part of all official diplomatic contacts between Rome and Parthia; 2) in military operations against the Parthians and their allies were used mainly units of Roman army garrisoned in Syria: 3) in the period under discussion we see two different ways of proceeding in Roman policy toward the eastern neighbour: a) aggressive and offensive in Armenia, where the army was frequently used against the Arsacids and their political supporters to settle local affairs according to Roman interests and b) offensive but avoiding, if possible, any military confrontation in direct contacts with the Parthians. For the both the emperors the military factor was a very important part of their eastern policy but more as an expression of determination to protect the Roman interests than as a tool of aggression.

L. Vitellius’ attitude leaves no doubt that Tiberius’ commitment to support the Parthian pretender was far from being tantamount with Rome’s actual readiness to engage in military action across the Euphrates. The subsequent struggle for the Parthian throne confirms that it was not. When, despite initial successes, Tiridates II proved unable to retain the Arsacid throne, Rome abandoned him, confining itself to the rôle of a passive observer.

The importance of the military factor in Roman-Parthian relations during the reigns of Augustus and Tiberius has never really been questioned. Yet, no attempt has been made so far to analyse its nature and specify its rôle in relations between the two countries. A comprehensive account of Roman military involvement in relations with Parthia comes from Tacitus. He recorded the operation of AD36 under the then governor of Syria L. Vitellius, carried out in order to help Tiridates II, pretender to the throne of the Arsacids, to ascend to power. The operation ended in a surprising way. The governor’s mission ended once the troops under his command reached the Euphrates, which separated Rome from Parthian lands. From then on, Tiridates II was to be accompanied solely by his Parthian supporters (Tacitus Anns. VI.37.3) and a memory of the mighty Roman host on the Syrian bank of the Euphrates.1

The comprehensive description of L. Vitellius’ military action in 36 is particularly valuable evidence of Rome’s use of the military factor in relations with its eastern neighbour. Its importance is further emphasized by information on military action taken by the governor himself. In the light of this source, this military involvement was confined solely to the province under his administration and its mainstay was made up of legions stationed in the province and allied troops.

Any action taken by the governor of Syria must have been specified in detail by Tiberius, for whom L. Vitellius’ mission was an important part of Rome’s policy toward its eastern neighbour (cf. Tacitus Anns. VI.32.1). The course of events during the mission suggests that, from the outset, it had been carefully planned and well prepared. Consequently, it may be surmised that its final Roman demonstration of power had also been premeditated. Tacitus’ account indicates that the demonstration was intended to convince Tiridates II’s Parthian supporters that Rome was backing him and was ready to use its legions in his aid. (cf. Ziegler 1964: 61f; Dąbrowa 1983: 111).2

This action, however, is not the only example of the army being used as a diplomatic tool under the first emperors of the Julio-Claudian dynasty. Sources for the times of Augustus and Tiberius contain a number of other references to both emperors using the legions as a decisive argument in relations with Armenia and Parthia. Nonetheless, some aspects have so far passed unnoticed. This lack of interest may be explained by the nature of our sources. Although relatively many, they rarely contain information on the strength of forces involved, their composition, place of origin, march routes, etc. What is more, sometimes, it is quite difficult to find clear confirmation that the army actually played a rôle in events. Yet a precise knowledge of these facts is necessary for a fair examination of the rôle of the military factor in the eastern policy of the first emperors of the Julio-Claudian dynasty. It is also important in a study of the organization and functioning of the Roman military system in the east.

1

Tacitus Anns. VI.37.4: Vitellius ostentasse Romana arma satis ratus.... Tacitus Anns. VI.37.1: robur legionum sociorumque ripam ad Euphratis ducit. Tiridates’ supporters could be only some of those allies. Tacitus’ account suggests that those Parthian aristocrats who supported him did not rally to his side until he had crossed the Euphrates, cf. Tacitus Anns. VI.37.3. The presence among L. Vitellius’ force of troops provided by the allies was probably due to the concentration of forces he had arranged to keep in check Artabanus II – who was fighting in Armenia against Mithridates, ruler of Caucasian Iberia – by a threat to launch an attack into Mesopotamia, cf. Tacitus Anns. VI.36.1-2.

2

The first credible references to the importance of the military factor in Roman-Parthian relations refer to the events of 20BC, in which Phraates IV returned the military

275

Limes XVIII soldiers under his command, or their origin.7 All we know is that its main purpose, placement on the Armenian throne of a Roman candidate, was accomplished without too much effort.8

ensigns lost by Marcus Licinius Crassus and Marcus Antonius, released Roman prisoners-of-war, and made large political concessions regarding Armenia.3 It should be stressed, however, that the military involvement in those events resulted from a coincidence, for at that time Augustus was in the east, while Tiberius was preparing an armed intervention in Armenia to install a Roman candidate. Hence Phraates, fearing a possible assault from Rome,4 displayed a greater than usual willingness to make concessions and alleviate arguments. The Arsacid’s concessions enabled both countries to co-exist peacefully for a long time. Their foundation was recognition by both sides of the Euphrates as a border (cf. Strabo Geog. XVI.1.28 (748)).

Still, some speculations might be made as to the origin of the detachments and the route they followed. At that time, Roman legions were stationed not only in Syria, but also in Anatolia.9 It is thus quite obvious that it was from those that troops were selected for the Armenian venture. If Tiberius moved out of Anatolia, as some scholars believe,10 then, for reasons of organization,11 the core of his force must have come from that area. In that case, Tiberius commanded a modest force that could have proved insufficient to carry out the objective Augustus put before it. One more circumstance deserves to be considered. Roman possessions in Anatolia were delimited by the boundaries of Galatia annexed in 25BC. Between it and Armenia lay lands that recognized Roman domination and this fact facilitated Tiberius’ passage, but on the other hand this part of Asia Minor was relatively little known to Romans. In this light, it would be more plausible to assume that Tiberius chose Syria as his base of operations. Not only was a much greater force stationed there, but also the lands between Syria and Armenia were far more familiar to the Romans since the expeditions of M. Crassus and M. Antonius. These factors should have determined the choice of routes starting from Syria also on other occasions.

The representative of Augustus who received the ensigns from Parthian hands was probably Tiberius.5 We know no details of the ceremony of handing over the standards, nor do we know where it happened. A demonstration of Roman military might probably constituted an important – if not dominating – element of the scene. Tiberius was acting in the twin rôles of the Roman emperor’s envoy and commander in chief of the army that was to restore to the Armenian throne the representative of the local dynasty, Tigranes, who had been held in Rome as a hostage.6 Records of Tiberius’ campaign in Armenia contain no details on its duration, course of events, routes, numbers of

Another Roman military operation in Armenia occurred in the years AD2-3.12 Together with an earlier expedition by C. Caesar to Arabia,13 it was part of a large-scale mission entrusted to the young prince by Augustus. As in Tiberius’ mission, the purpose of the Armenian expedition was to place a Roman candidate on the throne.14 An important historical event happened on the way to Armenia: C. Caesar met on the Euphrates the ruler of Parthia, Phraataces (Phraates V).15 The meeting was of crucial

3

RGDA 29, 1; Livy Per. 141; Velleius II.91.1; Suetonius Aug. 21.3; Tib. 9.1; Just. XLII. 5, 11; Dio LIV.8.1; Eutropius 7.9; Orosius 6.21.29. More on this subject see: Ziegler 1964: 47ff.; Pani 1972: 29; Krämer 1973: 362-363; van der Vin 1981. See also the contribution by E. Wheeler in this volume. 4 Whether Augustus really intended to attack Parthia in case his claims were rejected is not certain, even for ancient authors, cf. Just. XLII.5.10; Dio LIV.8.1. It may be worthwhile to quote the tell-tale opinion of Justinus (XLII.5.12), who thus summed up Augustus’ successes in relations with the Parthians: plusque Caesar magnitudine nominis sui fecit, quam armis facere alius imperator potuisset. It is more an echo of official propaganda than an objective assessment of events, cf. van der Vin 1981: 130f.; Rich 1998. 5 We owe to Suetonius the information that it was Tiberius who won this privilege (Tib. 9.1). Yet some scholars question its credibility, cf. Krämer 1973: 363; Levick 1976: 234, n. 38 (ibid. a presentation of earlier opinions) (yet elsewhere in her book [p. 26], the author makes a contradictory statement: “ ... he [Tiberius] arrived on the frontier in time to receive the standards on 12 May 20 BC”). The main argument against Tiberius’ participation in the celebration is, in their opinion, that Velleius is silent about it. Without disputing this observation, it may be sufficient to note that no source suggests that the legions’ ensigns were received by Augustus himself. It is true that in his Res Gestae, Augustus boasts of the fact that he recovered the standards, but this need not be taken as meaning that he received them in person (RGDA 29: Parthos ... spolia et signa reddere mihi...). The actual contribution of Tiberius might have amounted to carrying out orders. In this context, the Velleius’ silence becomes easier to understand. 6 Velleius II.94.4; Tacitus Anns. II.3.2; Suetonius Tib. 9, 1; Dio LIV.9.4-7; Pani 1972: 17ff. Velleius erroneously suggests that Tiberius installed Artavasdes on the Armenian throne, cf. J. Hellegouar'h comm. ad loc. in Velleius Histoire romaine, Tome II, livre 2, Paris 1982, 238, n. 11; SherwinWhite 1984: 325, n. 6. Despite the claim of Sherwin-White (1984: 324), accounts of Tiberius’ expedition lack sufficient evidence to treat the regaining of the signa and the Armenian expedition as two separate events. Also cf. Pani 1972: 18f.

7 Cf. Strabo Geog. XVII.1.54 (821); Velleius II.94.2; Suetonius Tib. 9.1. Concerning the composition of Tiberius’ army, we have only one vague remark in Velleius (II.94.4: cum legionibus ingressus Armeniam). 8 The Armenian expedition of Tiberius probably involved installation of a Roman candidate on the throne of Media Atropatene (RGDA 33). Cf. Pani 1972: 70ff., 87f.; Sherwin-White 1984: 325. 9 See Syme 1933: 18-19, 24, 30; Mitchell 1976: 298-308; Sherwin-White 1984: 339 and n. 35; Keppie 1986: 412. 10 Levick 1976: 26. In their opinion, this is implied by the presence at Tiberius’ side of the king of Cappadocia, Archelaos. In itself, however, it is not a sufficiently plausible argument to support the opinion. See Pani 1972: 20 - 23. 11 Sources seem to suggest that the Armenian expedition was prepared in a hurry. If the military standards reached Augustus, then in Syria, about mid-May AD20 (cf. Ovid Fasti 5, 545-598), then with much likelihood the event can be dated at the early spring of that year. Wintertime in Anatolia with its abundant snowfall would have prevented swift passage of a large military forces. Not having too much time to get ready, Tiberius was compelled to use solely troops stationed in one of these two areas. 12 For more on problems of the chronology, see Zetzel 1970; Bowersock 1971: 227 = 1994: 111; Pani 1972: 50ff. 13 Pliny NH II.168; 6, 141, 160; Zetzel 1970: 259ff.; Bowersock 1971: 227f. = 1994: 111f.; Romer 1979: 204 ff.; Luther 1999: 171–175, esp. 175. 14 Cf. Pliny NH VI.141; Dio LV.10a. 4; Velleius II.102. 2; Florus II.32. 15 Velleius II.101.1-3, 102, 2; Dio LV.10a.4; Ziegler 1964: 53ff.; Romer 1979: 208ff.; Barzano 1985: 214, 219f.; Winter 1994: 593f.

276

Edward Dąbrowa: “. . . ostentasse Romana arma satis . . .” The military factor in Roman-Parthian relations importance for the young prince’s mission as well as for peaceful relations between the countries that had been disturbed by the Arsacid’s action against Roman interests. Thanks to the meeting, Rome managed to avert armed confrontation with Parthia.16 The subsequent course of events showed that although Phraataces recognized the provisions of the new treaty, he did not, contrary to Roman expectations, entirely give up Parthian claims to Armenia. His attitude could not have failed to influence the situation in that country and it most probably fuelled anti-Roman resistance. Although Augustus’ envoy managed to prevail at the cost of some heavy fighting, the success proved short-lived and its price disproportionate to the effects achieved.

Tacitus’ account allows a conclusion that in order to complete his mission in Armenia, Germanicus needed legions stationed in Syria, although it is difficult to prove that on his way to Armenia Germanicus indeed took any troops with him. The matter is further complicated by the historian’s ambiguous remark referring to the failure of the governor of Syria to comply with Germanicus’ order to send to Armenia some of the legions stationed in his province.20 Doubts resulting from this unclear passage boil down to the question, once he was in Armenia, why did Germanicus ask the Syrian governor to supply troops? Was it only then that he deemed it necessary to command an armed force? It seems hardly likely that this was indeed the case. It is difficult to imagine, considering for example C. Caesar’s mission, that Germanicus could have achieved anything on the strength of the emperor’s mandate alone. In our opinion, Germanicus’ order referred to a reinforcement of the troops already at his command. It may have been needed to complete some unexpected military venture planned by the emperor’s envoy. It is quite certain that the presence of the troops did not imply danger to his life or his mission. The latter had already been successfully accomplished.21

The literary evidence suggests clearly enough that C. Caesar’s base of operations in the east was Syria (Velleius II.101. 1; Dio LV.10.19; Orosius 7.3.4). Without a doubt, Syria was chosen because of its close location to the theatre of operations and the size of the force stationed there.17 The action against the Arabs probably involved only some of the force, as the remaining part was obliged to man positions in Syria in view of the tense relations with Phraataces.18 In addition to Syrian legions, the task force probably included Egyptian legions (cf. Dio LV.10a.1).

It should be stressed that Germanicus’ actions in Armenia and Syria do not indicate that he was aiming at an armed conflict with Parthia’s ruler. His attitude toward Artabanus II was in line with general eastern policy of emperor Tiberius. Although that policy aimed to avoid direct confrontation, it showed determination to achieve its purposes and it relied on much diplomatic activity directed against the Parthian ruler. It cannot be ruled out completely that in acting as he did, Germanicus deserves the credit for the laying of foundations for a permanent Roman defence system on the Euphrates (cf. Dąbrowa 1986: 96-97).

Although we have no detailed information on the force engaged in the Armenian expedition, some observations are still possible. The known chronology of events indicates that C. Caesar set out for Armenia after a meeting with Phraataces. A description of the Roman-Parthian summit suggests that it took place on the Syrian border: both men were on their own respective territory.19 It is therefore safe to conclude that the core of C. Caesar’s force was made up of Syrian legions. Serious tensions between Rome and Parthia arose in the first years of the reign of Tiberius, after the Parthian king Artabanus II installed his own son on the Armenian throne (Tacitus Anns. II.4.3; Josephus Ant. Jud. XVIII.52). In order to prevent a strengthening of Parthian influence in Armenia, Tiberius entrusted to Germanicus the task of securing Roman interests in the area. Even though Tacitus devoted a comprehensive account to Germanicus’ activities in the east in AD18-19, his Armenian venture and military involvement there remains relatively obscure.

Apart from L. Vitellius’ action for Tiridates (mentioned above), the governor has one more political-military operation to his name - the meeting with Artabanus II in the spring of 37.22 Its effect was yet another agreement on mutual relations between the two countries.23 We do not know either the composition or the strength of the force accompanying L. Vitellius, but it can be taken for granted that it included legions stationed in his province.24

16

20

It is difficult to be certain whether the orders that C. Caesar received from Augustus instructed him to take military action against the Parthian ruler, cf. Dio LV.10.21. 17 No evidence is extant to suggests that legions from elsewhere in the empire took part in C. Caesar’s Armenian expedition. If we knew in detail the military career of L. Velleius Paterculus, then his presence at the meeting between C. Caesar and Phraataces on the Euphrates (II.101.2) could cast some more light on this question. Based on his eye-witness account, it cannot be ruled out that his presence might have resulted from his service with a unit stationed in Syria (II.101.3: ... omnibus ad Orientem visis provinciis...). 18 Characteristically, Velleius makes not a single reference to the Arabian part of C. Caesar’s mission. 19 Cf. Velleius II.101.3: Prior Parthus apud Gaium in nostra ripa, posterior hic apud regem in hostili epulatus est.

Tacitus Anns. II.57.1: ... qui (C. Piso) iussus partem legionum ipse aut per filium in Armeniam ducere utrumque neglexerat. 21 Cf. Tacitus Anns. II.57.1: Cunctaque socialia prospere composita.... 22 Scholars disagree whether the meeting took place while Tiberius was still alive or after Caligula had ascended to power (for arguments by both sides, see eg. Garzetti 1956: 211ff.; Ziegler 1964: 61-62). Such details would be important for chronology, but have no great significance for political assessment of the event. Although other sources point to Caligula as the initiator of this meeting (Suetonius Cal. 14.3), in our opinion Josephus’ (Ant. Jud. XVIII.4.4 (92) suggestion that Tiberius arranged the meeting seems more credible as a logical crowning of that emperor’s policy toward the Arsacid state. 23 For more on this, see Ziegler 1964: 63-64; Dąbrowa 1983: 111 - 112. 24 Their presence on the Euphrates is attested to by Suetonius Cal. 14.3; Vit. 2.4.

277

Limes XVIII

The attendance of Syrian legions at another diplomatic ‘summit’ suggests that their presence had become an indispensable tool in implementing Rome’s eastern policy.

These episodes in Roman-Parthian relations demonstrate an unexpectedly important rôle of the army as an instrument of Roman policy in Armenia and vis-à-vis Parthia.30 Clearly perceptible is the difference with which Rome used the military factor in respect to each of its neighbours. To protect and promote Roman interests in Armenia, both Augustus and Tiberius were ready to resort to the use of troops, which Parthia formally sanctioned when it recognized Rome’s right to interfere in the politics of that land. In its policies toward the Arsacid state, the military factor served mainly to exert more or less intense pressure to complement the diplomatic protocol. The successes that both emperors managed to achieve using this tool, however, did not lull their vigilance. Although official propaganda wasted no opportunity to present Parthia as a weak, subjugated opponent, sources show that in direct relations both emperors adopted a respectful and apprehensive attitude. The equal status of both powers in diplomatic contacts, repetitious confirmation of the agreement on mutual relations as achieved in 20BC, skilful avoidance of open confrontation with Parthian rulers, and deployment of a large army in Syria further confirm this.31

For any analysis of the rôle of Syrian legions, this insight makes it possible to add more events to the list, even though sources are silent about any military presence, for example, when in 10BC Phraates IV handed over four royal sons and their families to M. Titius, governor of Syria.25 Under an agreement with Augustus, they were to stay in Rome under the emperor’s protection.26 Given the political importance of this event, the governor of Syria must have received the royal sons entrusted to his care with an escort of troops under his command. Roman military participation can be surmised in the not quite successful operation of AD6 to install on the Armenian throne Artavasdes, whom Augustus supported. The operation involved some unspecified Roman failures. Tacitus alludes to them in such a way that it may mean either the emperor’s unsuccessful political schemes or a defeat of Roman troops.27 It is difficult to fathom what the historian meant since this remark is the only record we have of that operation.

One more important observation can be drawn from the foregoing discussion. Although it did not add to our knowledge of Syria’s military organization during the reigns of Augustus and Tiberius, it casts some light on the rôle of that province and its legions in the events described. Use of Syria as a permanent base for operations in Armenia proves not only its strategic positioning in Roman policies but its strong military presence from early in Augustus’ reign. There can be no doubt that he assigned Syria a key rôle in relations with Rome’s eastern neighbour from the moment he ascended to power. Although we have only imprecise information on the size of the force he sent there (cf. Sherwin-White 1984: 328-341), the identity of units,32 and placement of their camps, there is no doubt that without this military force at their disposal, both Augustus and Tiberius could hardly have achieved their successes in their eastern policy.

It is highly probable that the legions based in Syria were involved in two other similar events. The supposition is justified by an account of Tiridates II’s ascent to power in Parthia, mentioned earlier in this paper. These events involved Rome extending its support to other pretenders to the Arsacid throne and we may surmise that their scenario did not differ significantly from the one we already know. The first was in approximately 9/10. Its hero was Vonones, one of the sons of Phraates IV residing in Rome. Asked by Parthian aristocrats, Augustus agreed to Vonones’ ascent to the Arsacid throne.28 On another occasion, in 35, in response to a similar request, Tiberius approved as king Phraates, son of Phraates IV.29

25 RGDA 32; Strabo Geog.VI.4.2 (288); XVI.1.28 (748); Velleius II.94.2; Tacitus Anns. II.1.2; Josephus Ant. Jud. XVIII.42; Suetonius Aug. 43.4; Just. XLII.5.12; Orosius 6.21.29. The king and the governor met on the initiative of the former. Although sources make no mention of the venue, it must have been on the Euphrates. 26 For more on the various aspects of the Parthian princes’ presence in Rome and the rôle they played in the Roman emperors’ eastern policies, see Pani 1972: 26-35; Dąbrowa 1983; 1989; Niedergaard 1988. 27 Tacitus Anns. II.4: Dein iussu Augusti impositus Artavasdes et non sine clade nostra deiectus. cf. Pani 1972: 25, 38ff. According to Pani (1972: 38), the task to install Artavasdes on the throne fell to the governor of Syria. However, we have no evidence for the reign of Augustus or his successor that any of governors operated with the Syrian legions stationed there outside the province. 28 RGDA 33; Josephus Ant. Jud. XVIII.2.4 (50); Tacitus Anns. II.1.1-2.4; cf. Suetonius Aug. 21.3. 29 Tacitus Anns. VI.1-2; Dio LVIII.26.2. Phraates died in Syria before he reached his country’s border. The exact place and circumstances of his death remain a mystery. It may be surmised, however, that he was escorted by Roman troops as Tiberius recognized the importance of this and the next expedition of Tiridates II a few month later.

Bibliography Barzano A. 1985 Roma e i Parti tra pace e guerra fredda nel I secolo dell’impero. In M. Sordi (ed.) La pace nel mondo antico. (Milano): 211 - 222. Bowersock G.W. 1971 A report on Arabia Provincia. Journal of Roman Studies 61: 219-242.

30 In this contribution I have not included a discussion on the Aelius Gallus’ expedition to Arabia. Arguments raised by Marek (1993) to put it into scheme of Roman-Parthian relations under Augustus are very impressive indeed, but in my opinion they are neither completely convincing nor appropriate. See also Luther 1999: 157-170. 31 The first mention containing the names the names of legions stationed in Syria refers to the years 18-19. They included legio X Fretensis (Tacitus Anns. II.57.2; AE 1993, 204), and legio VI Ferrata (Tacitus Anns. II. 79.2). 32 Cf. Dąbrowa 1986: 96f.; Keppie 1986: 413f.

278

Edward Dąbrowa: “. . . ostentasse Romana arma satis . . .” The military factor in Roman-Parthian relations Bowersock G.W. 1994 Studies on the eastern Roman empire. Social, economic and administrative history. Religion. Historiography. (Goldbach). Dąbrowa E. 1979 Les troupes auxiliares de l’armée romaine en Syrie au Ier s. de notre ère. Dialogues a l’Histoire Ancienne 5: 233-254. Dąbrowa E. 1983 La politique de l’Etat parthe à l’égard de Rome – d’Artaban II à Vologèse I (ca 11 – ca 79 de n. è. et les facteurs qui la conditionnaient. (Kraków). Dąbrowa E. 1986 The frontier in Syria in the first century AD. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. Proceedings of a Colloquium held at the University of Sheffield in April 1986. British Archaeological Reports Int. Ser. 297. (Oxford): 93-108. Dąbrowa E. 1987 Les premiers „otages” parthes à Rome. Folia Orientalia 24: 63-71. Dąbrowa E. 1989 Les héros de luttes politiques dans l’Etat parthe dans la première moitié du Ier siècle de n. è. Iranica Antiqua 24: 311-322. Garzetti A. 1956 La data dell’incontro all’Eufrate di Artabano II e L. Vitellio, legato di Siria, in Studi in onore di Aristide Calderini e Roberto Paribenmi, vol. 1. (Milano): 211-229. Krämer K. 1973 Zur Rückgabe der Feldzeichen im Jahre 20 v. Chr. Historia 22: 362-363. Keppie L. 1986 Legions in the east from Augustus to Trajan. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. Proceedings of a Colloquium held at the University of Sheffield in April 1986. British Archaeological Reports Int. Ser. 297 (Oxford): 411–429. Levick B. 1976 Tiberius the politician. (London). Luther A. 1999 Medo nescis catenas ? Die Expedition des Aelius Gallus in Rahmen der augusteischen Partherpolitik. Orbis Terrarum 5: 157-182. Marek Ch. 1993 Die Expedition des Aelius Gallus nach Arabien im Jahre 25 v. Chr. Chiron 23: 121-156. Mitchell S. 1976 Legio VII and the garrison of Augustan Galatia. Classical Quarterly 26: 298-308. Niedergaard E. 1988 The four sons of Phraates IV in Rome. Acta Hyperborea 1: 102-115. Pani M. 1972 Roma e i re d’Oriente da Augusto a Tiberio (Cappadocia, Armenia, Media Atropatene). (Bari). Rich J.W. 1998 Augustus’s Parthian honours, the Temple of Mars Ultor and the Arch in the Forum Romanum. Papers of the British School at Rome 53: 71-128. Romer F.E. 1979 Gaius Caesar’s military diplomacy in the east. Transactions of the American Philological Association 109: 199-214. Sherwin-White A.N. 1984 Roman foreign policy in the east, 168 BC to AD 1. (London). Syme R. 1933 Some notes on the legions under Augustus. Journal of Roman Studies 23: 14-33. van der Vin J.P.A. 1981 The return of Roman ensigns from Parthia. Bulletin Antike Bescharing 56: 117-139. Winter E. 1994 Die Bedeutung des Grenzraumes für den diplomatischen Verkehr: Das Imperium Romanum und seine östlichen Nachbarn. Stuttgarter Kolloquium zur historischen Geographie des Altertums 4, 1990. (Amsterdam): 589-607. Zetzel J.E.G. 1970 New light on Gaius Caesar’s eastern campaign. Greek, Roman & Byzantine Studies 11: 259266. Ziegler K.-H. 1964 Die Beziehungen zwischen Rom und dem Partherreich. Ein Beitrag zur Geschichte des Völkerrechts. (Wiesbaden).

279

Limes XVIII

280

The circumvallation at Hatra Gwyn Davies This paper will re-examine the encircling work at Hatra, the status of which has been subject to sporadic debate for nearly a century. This feature has been interpreted either as a line of circumvallation (of Roman/Sassanian origin) or as a form of advanced proteichisma comprising an integral element of the city defences. In order to assess the relative merit of these various arguments, particular reference will be made to the survey published by Andrae in 1912 which remains the most authoritative account as to the nature of this encirclement. Furthermore, the site will be assessed in the context of the author’s wider research into the structural components of Roman siege operations and comparative material will be cited to argue in favour of a Sassanian attribution for this work of investment.

The city of Hatra is located in the northern Mesopotamian desert, immediately west of the Wadi Tharthar, some 50kms to the west of the Tigris and 120kms south of the strategically important Jebel Sinjar (Fig.1). As the site of perennial springs, Hatra was the natural focus of regional caravan traffic, the revenues derived from which probably paid for its extensive public buildings and impressive double enceinte. The permanent occupation of the site ended some time after the city’s sack by Sapor I (c.240241) and before 364 when Jovian’s army passed its abandoned ruins.

has, however, focused upon the identity of the party responsible for this work. This argument will be reexamined in the course of this paper and will be put into the context of my wider research into Roman offensive siege works. Who was responsible for the construction of the circumvallation?

a. The structural evidence. In his original identification of the circumvallatory scheme at Hatra, Andrae had little doubt that this represented evidence for Sapor’s ultimately successful operation, although this conclusion was refuted by Crawford (1929: 502) for whom “the line of circumvallation…(was)…plainly Roman work”. This confident declaration was, in turn, challenged by Stein (noted in Gregory & Kennedy 1985: 67) who regarded Andrae’s original claim as “well-founded” on the basis that the considerable extent of the siege system must have “...involved protracted and heavy labours” that neither Trajan nor Severus would have had time to construct. Bradford’s commentary (1957: 74, n.3), although noncommittal, would seem to veer back towards a Roman origin, given that “...large earthworks can be raised very quickly by men under pressure as all wars prove”, a point re-emphasised by Gregory and Kennedy (1985: 398, n.67(a)). Finally, Kennedy and Riley steer an intriguing medial course by suggesting that the “...visible lines…(belong) to the final, successful Persian siege, even if they were a re-use of Roman work” (1990: 107). Unfortunately, most of these arguments would seem to have revolved around an interpretation of the aerial photographic record, when the starting point should really have been Andrae’s second published report (Andrae 1912), which remains the only substantive account of the structural composition of this encircling work.

Hatra had an eventful military history, being the target of several sieges mounted by the contending regional powers. The first recorded episode was Trajan’s attempted suppression of a Hatrene revolt, probably in 117, an unsuccessful venture which failed, according to Dio (LXVIII.31 [Xiph. 240 R St.]), because of extreme weather conditions and a plague of flies. A second Roman attack was made on the city by Severus, although our sources disagree over the exact details of this operation. According to Dio two separate attacks were mounted, in 199 and 200 respectively (LXXV.10; LXXVI.11-13), whilst Herodian records only one siege (taking place before the march on Ctesiphon), although his account (III.9.4-7) similarly notes the eventual failure of the enterprise. However, it was not just the Romans who found the Hatreni a difficult proposition, as Ardashir’s forces discovered to their cost c.230, and the city did not finally succumb to Sassanian pressure until what would appear to have been a major campaign by Sapor a decade or so later. The legacy of one or more of these military interventions is apparently visible in this famous aerial photograph of Hatra published by Bradford (1957: pl.24) where an encircling work (with a circumference of approx. 8kms) can be seen running around the city at a distance of between 350m-500m from its wall circuit. This feature was noted by Ross who re-discovered the site in 1836, albeit that he noted that this “thick rampart” was sited at a foreshortened distance of “100 to 150 paces” from the ditch fronting the urban enceinte (noted in Andrae 1912: 20). The extensive field survey of the site carried out by Andrae for the Deutschen Orient-Gesellschaft (Andrae 1908; 1912), resulted in this encircling feature being identified as a work of circumvallation, which, at that time, still stood up to 2m high and up to 4m wide (Andrae 1912: 20). Although this interpretation has not been questioned until recently (of which, more later), considerable debate

From this report, it would seem that the circumvallation was not of a homogenous character along its entire circumference. At one (unstated) location, it was reported as having a foundation course of rough, undressed stone, whilst on the western front, an unusual “form of halftimbering (made) out of stone” (Art fachwerk aus Steinen) would seem to have been employed (Andrae 1912: 20). A ditch on the Hatra-facing side was only provided in the north-west sector (Andrae 1912: 21). The brief notes published in respect of modern soundings along the line would seem to confirm this lack of consistency, with Al281

Limes XVIII

Salihi’s investigation in the east revealing a rampart composed of earth with a foundation of uncut limestone (Al-Salihi 1980: 168), whilst Ibrahim’s sections in the north and north-east, failed to identify “any special foundation” at all (Ibrahim 1981: 193).1

certainly suggests that different units, albeit working within the framework of general orders, employed their own criteria of what would have amounted to suitable defensive provision. Similar variability in construction techniques (or even in engineering competence) is also apparent from the circumvallation at Numantia (Schulten 1914-1929: Vol. III, 82-83) and from the practice works at Woden Law (Richmond & St Joseph 1982: 279).

Given this apparent failure to build a rampart in standardised form, it is tempting to follow Kennedy and Riley’s suggestion that we have evidence for a multi-phase work, where the industry of earlier siege commanders was put to good use by subsequent assailants, but this argument depends upon the implausible premise that the Hatreni would have allowed their city to remain surrounded by the vestiges of hostile works with the potential for later recycling. Therefore, if the surviving encirclement is to be thought of as the product of only the one campaign, should we be concerned as to the different forms that this structure seemingly takes along its course? Conditioned by the wellknown example of Masada, where the circumvallation displays a remarkable consistency despite the rigours of the terrain, it would be a simple matter to conclude that Roman military engineers worked to a uniform model in the elaboration of their siege works. From this, it might then be concluded that the rather haphazard arrangements displayed at Hatra would establish a prima facie case for a Persian origin, as a less methodical approach would sit easier with general preconceptions of laissez-faire ‘eastern’ attitudes. However, any such conclusion must surely be sabotaged by the realization that Roman engineering details were equally prone to variable working practices, as the recent Franco-German excavations at Alesia have readily demonstrated (Reddé 1995; 1996; 1999; Reddé & v. Schnurbein 1997; Reddé, v. Schnurbein et al 1995; v. Schnurbein et al 1991-1996).

Thus, if the failure to adhere to a regimented schema cannot be used to distinguish between Roman and Persian work, is there any other structural evidence that may serve as a pointer? I would suggest that the unusual configuration of the circumvallation in the western sector provides some such suitable indicator. As previously stated, the construction method employed here comprised a “form of half-timbering (made) out of stones”, a visual impression of which is provided by a sketch in Andrae’s report (Fig. 2) which considers that this technique would have provided extra reinforcement to the core of the earth rampart. This ‘reinforcement’ comprised a 1.5m thick stone ‘layer’ on the side facing the target and a second, 0.5m thick stone course running parallel to, but c.1m behind, the first (Andrae 1912: 20). Both were connected by 0.5m thick stone ‘bulkheads’ forming a series of casemates filled with rubble or mud-brick (Andrae 1912: 21), and it is likely that the whole was then either encased within an earthen rampart or topped out by a ‘cap’ of compacted soil. This would seem to be a particularly elaborate means of providing structural bracing to the vallum and it is difficult to think of any comparable Roman example. Perhaps the closest model would be the Scipionic circumvallation at Numantia where, on some stretches at least, three longitudinal ‘ribs’ of stone underpinned the rampart, albeit without any form of cross-bracing as appears here. Unfortunately we do not have any note as to the height of these “Steinlage” and, in consequence, it is difficult to speculate as to the degree of extra support that this sort of revetment would have provided. If the sketch provided by Andrae is correct, it would also seem that particular care was taken over the construction of the thicker, western wall course, as this is depicted as having facing stones to both front and back with a rubble core between them. It is hard to understand why such a complex mode of construction was deemed necessary, and if it was not for the fact that the resulting rampart must have been oriented to face Hatra (as the thicker of the two external ‘kerbs’ was placed on the cityfacing front), one might be sympathetic to the claim made by Ibrahim (1981: 193) that this amounted to some form of outwork built by the Hatreni.

Here, notwithstanding Caesar’s extravagant claim (Bell. Gall. VII.74) to have created a monolithic system, there are considerable variations in layout both between the lines of circumvallation and contravallation as well as between different sectors along each of these lines. In part, this variety in configuration (as evidenced, for example, in the depth and complexity of the obstacle field that fronted the main rampart) can be attributed to local security concerns. For example the vulnerable area of the plaine des Laumes was much more rigorously provided for than the substantially unthreatened montagne de Bussy, where the defences maintain an illusion of strength which is not sustained by a closer inspection (Reddé, v. Schnurbein et al 1995: 117). However, the degree of variation both in the dimensions and the profiles of the ditches that front the contravallation and the circumvallation in the plaine des Laumes (Reddé, v. Schnurbein et al 1995, Insert 7 & 10),

Given the absence of any parallel Roman example, it is tempting to ascribe this work a Persian character on the basis of the greater prevalence of casemated structures in eastern contexts (although our lack of knowledge as to the methods employed by Sassanian siege engineers makes this a speculative conjecture). One possible, distant relative of this technique may, however, be suggested for the

1

The investigations carried out by the recent Polish expedition to Hatra (Gawlikowski 1994a; 1994b), although useful for elucidating the sequential development of the urban enceinte, do not appear to have included any prospection of the encircling work. However, the identification of an early wall circuit set behind the main works, does effectively rule out a Trajanic origin for the siege system (as the circumvallation would thus have been set too far from the target to have played a fully operational rôle).

282

Gwyn Davies: The circumvallation at Hatra circumvallation at Plataea, which comprised a doublewalled structure that was roofed over and divided by internal partitions to provide quarters for the besieging army (Thucydides III.21.2). Of course, the Hatrene example was not intended to fulfil such ambitious objectives, but the incidence of a similar casemated structure may reflect the continuation of an older siege tradition than that employed by the Romans.

home (at least two breaches were effected in the enemy circuit), suggest that all efforts were directed towards an assault rather than a methodical campaign of reduction. It may also be significant that Severus’ foragers were discomfited by Arab cavalry raids, which, if they emanated from within the city itself rather than from irregular forces outside, would be a clear demonstration that no continuous work of circumvallation had been built at this time.

Certainly, the construction of this comparatively elaborate type of rampart reinforcement would seem to suggest that whoever was responsible was anticipating that operations before the city would be sufficiently prolonged to make the investment of labour worthwhile. Once again, this provides a useful hint as to the likely authors of the feature.

In view of the above, it is entirely possible that Roman forces were deployed around Hatra in one or more blockade camps and that no provision was made for a comprehensive line of investment and in such circumstances, there would be little option but to concede a Sassanian origin for the surviving siege work elements. Although we have little information as to the attack pressed home by Sapor, it is likely that his awareness of Ardashir’s failure before this target only 10 or so years earlier, would have meant that the Great King would have proceeded with a methodical operation in which chance was allowed to play little part. Of course, a prolonged Sassanian enterprise directed against Hatra was more feasible than a comparable Roman campaign, as the challenge of sustaining an army from a base on the Tigris would have been far less problematic than an attack mounted from the west or north-west. Indeed, that Sapor may have been delayed by a lengthy operation before Hatra is not unlikely, particularly when it is remembered that the local rulers had admitted a Roman cohort to assist in their defence, and prolonged resistance may well have justified the decision to provide the western circumvallation with its elaborate form.

b. The duration of Roman operations before the city. Although Gregory and Kennedy are correct in reminding us how fast Roman troops could throw up a line of circumvallation (1985: 398, n.67a), this would generally have comprised a simple earth-dump rampart or a straightforward stone wall. It is, however, unrealistic to cite Titus’ siege of Jerusalem or some of Caesar’s Gallic operations as comparanda for rapid programmes of construction, as in both cases these commanders would have had access to large armies well-practised in the methods of siege warfare. It must be doubtful whether either Trajan or Severus had access to the same resources. Certainly, Trajan’s venture has all the hallmarks of a hastily improvised operation with his deployment of cavalry (Dio LXVIII.31) suggesting that he expected the city to surrender after a simple demonstration in force, a conclusion reinforced by his apparent failure to provide any proper supply train. The failure of the first Severan campaign, in which it is laconically reported that the emperor “achieved nothing” (Dio LXXV.10.1), may also imply a similar lack of forward planning, although his subsequent operation clearly attracted a more careful approach, with the gathering together of a stock of provisions and the assembly of new siege engines. However, it is unlikely that Severus could have deployed a large body of seasoned troops before the city, not least because the local environment would have been unable to sustain a large army operating at the end of extended supply lines for very long. Also, the comment in Dio that it was only the European contingent of his army who “had the ability to do anything” (LXXVI.12.3), implies that Severus experienced a shortage of skilled manpower. Thus, even if these ‘Europeans’ were indeed the Moesian contingent that had taken part in the lengthy siege of Byzantium (194-196) as Kennedy speculates (1986: 402403), they were unlikely to have been sufficiently numerous to have allowed the completion of such a complex type of circumvallation within the allotted period of 20 days (when a mutiny forced the abandonment of the attack). In any event, it does not appear that Severus had planned to undertake a prolonged operation: his concentration upon the assembly of new engines (particularly those provided by the renowned engineer, Priscus) and the rapid way in which attacks were pressed

That the siege works before Hatra are more likely to represent the traces of Sapor’s final attack on the city should not occasion surprise as it is apparent that the Persians were capable of mounting sophisticated and protracted siege operations. The later attacks on DuraEuropos, c.252 and Amida in 359 readily demonstrate that the Sassanians regularly employed the full range of reductive techniques, from assault ramps to multi-purpose tunnelling operations. It is possible that further research into the construction methods employed by Sassanian architects may allow greater confidence to be shown in pursuing this attribution. The ‘indigenous’ alternative A further alternative explanation for the encircling work, briefly alluded to already, is that this represented a form of outwork constructed by the Hatreni themselves. This possibility, tentatively proposed by Iraqi archaeologists in the 1970s, was enthusiastically endorsed by Ibrahim in his unpublished Ph.D. thesis in 1981. However, this suggestion that the ‘circumvallation’ fulfilled a defensive function as some form of advanced proteichisma, is scarcely tenable. As we have already noted, the configuration of the rampart on the west, where the thicker of the two parallel wall-lines 283

Limes XVIII

confronts the city, suggests that the flank requiring greater reinforcement was that facing Hatra. This impression of a structure built to oppose the defenders of the city is supported by the provision in the north-western sector of a ditch in advance of the rampart which, once again, is found running along the inner, city-facing flank. Apart from these structural clues, it is extremely questionable whether the sort of outwork that has been postulated would have provided the defenders with any sort of military advantage. After all, an outer circuit positioned at such a remove from the main defensive enceinte would have been out of range of any artillery support and its garrison would have been vulnerable to isolation in the event of an enemy attack breaking through the cordon at any one point. Furthermore, the manpower that would have been required to provide this work with an adequate guard along its 8kms circumference would surely have been beyond the resources of the Hatreni, whose extensive double-walled ‘main’ circuit must have demanded the commitment of most of their available troops. Accordingly, unless this outer wall was built with purely symbolic motivation in mind, there are no substantive grounds for concluding that it was the product of Hatreni engineers.

Bibliography Caesar

1994 De Bello Gallico (H.J. Edwards (trans.). Loeb Classical Library). Cassius Dio 1914-1927 Roman History (E. Cary (trans.). Loeb Classical Library). Herodian 1969-1970 History (C.R. Whittaker (trans.). Loeb Classical Library). Thucydides 1951-1953 History of the Peloponnesian War (C.F. Smith (trans.). Loeb Classical Library).

Al-Salihi W. 1980 Al-Hadhr/Al-Tanqib fi Al-Bawabah AlShamaliyah (Hatra/Excavations in the Northern Gate). Sumer 36: 158-189. Andrae W. 1908 Nach Aufnahmen von Mitgliedern der AssurExpedition der Deutschen Orient-Gesellschaft. Vol.I. Allgemeine Beschreibung der Ruinen. Wissenschaftliche Veröffentlichung der Deutschen Orient-Gesellschaft 9. (Leipzig). Andrae W. 1912 Nach Aufnahmen von Mitgliedern der AssurExpedition der Deutschen Orient-Gesellschaft. Vol. II. Einzelbeschreibung der Ruinen. Wissenschaftliche Veröffentlichung der Deutschen Orient-Gesellschaft 21. (Leipzig). Bradford J. 1957 Ancient Landscapes. (London). Crawford O.G.S. 1929 Air photographs of the Middle East. Geographical Journal 73: 497-512. Gawlikowski M. 1994a A fortress in Mesopotamia: Hatra. In E. Dabrowa. (ed.) The Roman and Byzantine army in the east. (Krakow): 47-56. Gawlikowski M. 1994b Fortress Hatra: new evidence on ramparts and their history. Mesopotamia 29: 147-184. Gregory S. & Kennedy D. 1985 Sir Aurel Stein’s Limes Report. British Archaeological Reports Int. Series 272 – 2 vols. (Oxford). Ibrahim J.K. 1981 New evidence for settlement in the Jazirah in the pre-Islamic period: with special reference to Hatra and Jaddalah. Unpublished Ph.D. thesis, University of London. Kennedy D. 1986. ‘European’ soldiers and the Severan siege of Hatra. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297. (Oxford). Kennedy D. & Riley D. 1990 Rome’s desert frontier from the air. (London). Reddé M. 1995 Titulum et Clavicula à propos des fouilles récentes d’Alésia. Revue Archéologique de l’Est. 46.2: 349-356. Reddé M. (ed.) 1996 L’Armée Romaine en Gaule. (Paris). Reddé M. 1999 César Ante Alesia. In M. Almagro-Gorbea, J.M. Blázquez Martínez, M. Reddé, J.G. Echegaray, J.L.R. Sádaba. & E. Peralta Labrador (edd.) Las Guerras Cántabras. (Fundación Marcelino otín, Santander): 119-144.. Reddé M. & von Schnurbein S. 1997 Les nouvelles fouilles d’Alesia. In W. Groenman van Waateringe, B.L. van Beek, W.J.H. Willems & S.L .Wynia (edd.) Roman Frontier Studies 1995 Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 175-185. Reddé M., von Schnurbein S., Barral P., Bénard J., BrouquierReddé V., Goguey R., Joly M., Köhler H-J. & Petit C. 1995 Fouilles et recherches nouvelles sur les travaux de César devant Alésia (1991-1994). Berichte der Römisch-Germanischen Kommission. 76: 73-158.

In support of his contention however, Ibrahim does make the surprising observation that his ‘soundings’ (for which no plans are provided) revealed that the earth for the rampart core “...was taken from within the enclosed area of the inner side of the wall” (1981: 193) which he takes to mean that the structure was raised by the defenders as “had it been made by Roman or Sassanian attackers, they would have most likely taken the earth from their side of the barricade, that is, the outside” (ibid). However, this does not seem a particularly plausible argument unless it is also claimed that the wall also demarcated a distinct sedimentological boundary, an argument that would be difficult to sustain in view of the fact that Ibrahim’s earlier statements regarding the local geomorphology (1981: 191) contain little to support the identification of such clearly defined horizons. Summary If this ‘indigenous’ hypothesis as to the origin of the encirclement can be disregarded, then the suggestion that it represents an assailant’s siege structure is particularly persuasive. Although the subsequent identification of the circumvallation as a Roman or Sassanian work cannot be definitively proved at this time, the most obvious conclusion that it is the product of the last campaign directed against the city by Sapor would seem to be supported by circumstantial evidence, both from the limited literary accounts as to the nature of the operations conducted by the Roman commanders, and from the unusual form of construction that was adopted for at least a proportion of the rampart line. Even if future excavation at Hatra does not result in the final arbitration of this issue, it is to be hoped that a wider investigation of Sassanian siege methods may provide us with a fuller body of comparanda from which more informed conclusions might be drawn. 284

Gwyn Davies: The circumvallation at Hatra Richmond I.A. & St Joseph J.K.S. 1982 Excavations at Woden Law, 1950. Proceedings of the Society of Antiquaries of Scotland 112: 277-284. von Schnurbein S. & Parzinger H. 1991-1993 Ausgrabungen in Alesia. Berichte der Römisch-Germanischen Kommission. (1991) 72: 337; (1992) 73: 547; (1993) 74: 858-859. von Schnurbein S., Parzinger H. & Sievers S. 1994 Ausgrabungen in Alesia. Berichte der RömischGemanischen Kommission. 75: 665. von Schnurbein S. & Sievers S. 1995-1996 Ausgrabungen in Alesia. Berichte der Römisch-Germanischen Kommission. (1995) 76: 303-304; (1996) 77: 362. Schulten A. 1914-1929 Numantia. 4 vols. (Munich).

285

Limes XVIII

Fig. 1. The location of Hatra.

Fig. 2. Sketch of the circumvallation (after Andrae (1912): 20).

286

Roman treaties with Parthia: Völkerrecht or power politics? Everett L. Wheeler and occasionally punched each other. But what were the rules of the game? From the Augustan age into the C3rd the sources speak of Rome and Parthia as “rivals” or the “two greatest empires.” (eg. Strabo Geog.XI.9.1; Velleius II.101.2; Pliny NH V.88; Josephus Ant. Jud. XVIII.46; Tacitus Anns. II.56.1, 60.4; IV.5.2; XII.10.2; XV.13.2; Fronto Princ.Hist. 3, p.193; van den Hout; Dio XL.14.3; Herodian IV.10.2.). Pompeius Trogus, as found in Justin (XLI.1.1), declares a division of the world between Rome and Parthia. But is this Trogus, drawing conclusions from the Battle of Carrhae (Wolski 1993: 133-34), or Justin with maximum hindsight (Brunt 1990: 458)? Or, did Augustus himself contrive this idea to reconcile Roman claims to world domination with a reality halted at the Euphrates? On this view Parthians belonged to a world outside the oecumene (Sonnabend 1986: 202, 210; cf. Arnaud 1993; Wiesehöfer this volume). This example is typical of the many differences of opinion the sources permit. But unlike military aspects of frontier studies, where identifying propaganda and policy leads to fantasy (eg. Whittaker 1994; contra and rightly, Brunt 1990: 477; Potter 1996: 60; Nicasie 1997: 455), in international relations a Roman ideology of Weltherrschaft assumes the utmost importance: prestige and precedence are the name of the game.

Joseph Vogt (1955: 74) once remarked that Rome’s Parthian wars were like skin rashes on the empire’s body politic: often recurring but inconsequential. Tacitus (Germ. 37.2-5) had a similar view in asserting that Germans, not Parthians should be the empire’s chief strategic concern.1 Tacitus’ foresight about Germans should not obscure his statement’s contemporary polemic flavor: at the dawn of Trajan’s reign Parthian capabilities were taken seriously at Rome - and with some justification, for every generation between the reigns of Tiberius and Severus Alexander experienced a Parthian war or at least a major crisis (cf. Wheeler 1996: 235-36). Parthian relations dominated Rome’s foreign policy in the east. Recent efforts to portray Parthians as innocent victims of Roman aggression (Isaac 1990: 14-57; Whittaker 1994: 5459; Cornell 1993: 143-44; cf. Campbell 1984: 114-15, 332, 356-57, 390-93) misrepresent and oversimplify the ‘cold war’ atmosphere of Roman-Parthian affairs (eg. Campbell 1993; contra and rightly, Barzanò 1985: 211-22): the cause of every war and crisis cannot be laid at Rome’s doorstep, and every Trajan, Septimius Severus, or Caracalla seeking Roman glory at Parthian expense had his counterpart in an Artabanus II, Vologaeses I, or Vologaeses IV, probing some chink in Rome’s armor for Parthian exploitation. Parthian failure to achieve lasting territorial gains west of the Euphrates by force does not prove a lack of desire for either expansion or recovery of land, such as Armenia, to which Parthians laid claim. Black and white photos of heroes and villains make for good victimology and rather bad history (cf. Wolski 1976: 197, 212; 1993: 122; Wheeler 1993: 30-36; Kennedy 1996: 76, 79-80; Arnaud 1998: 13-14). Even if a “glory motive” of emperors for some Parthian wars be accepted, this view need not exclude other considerations such as strategy or Roman prestige, unless one wishes to sink into the historical fallacy of monocausation.

Recent evaluations of Roman-Parthian relations paint contrasting pictures. Brian Campbell (1993) outlines an edifice of Augustan diplomacy, which Trajan and other glory-hunting emperors of the C2nd and C3rd destroyed, though it is difficult to understand how he can claim that the response of Sedatius Severianus, the Cappadocian governor, to the Parthian invasion of Armenia in 161 caused the Parthian war of 161-166 (Campbell 1993: 215; likewise erroneously, Mattern 1999: 13-14, 188-89), deny the rôle of Armenia in Parthian wars (1993: 220-21), and assert that the conflict under Nero was not a Parthian war (1993: 236 with n.1). Does Mr. Campbell understand the eastern frontier at all? Edmund Frézouls’ more nuanced approach (1993: esp. 497) discerns Armenia as a circumscribed playground where the two superpowers could compete for prestige and propaganda points at reduced costs and with limited risk. Certainly Armenia held a very special place in Roman-Parthian relations, but reputations and designated future emperors (eg. the young C. Caesar dead in AD 4) could get hurt there.

Of course a lack of Parthian sources precludes evaluation of events from an Iranian viewpoint and only now, after a generation of pleas by Josef Wolski (1993 with references to earlier work; cf. Dabrowa 1983; Wiesehöfer 1998a), can we begin to situate the Parthian empire into a broader Eurasian perspective: a long overdue assessment of Parthia’s problems on her eastern and north-eastern borders has been initiated (Olbrycht 1998a; 1998b; 1998c) and traces of an apparent C2nd and C3rd Parthian limes east of Hatra in Adiabene are coming to light (Hauser 1998: 18-19).

Towering above all other discussions, Karl-Heinz Ziegler’s 1964 monograph remains the basic survey of RomanParthian relations. The work belongs to the genre of histories of international law from the pens of legally trained scholars, who exaggerate the force of international law as a real system of law (on analogy with civil law), and who play down or ignore both international law’s lack of enforcement potential for states not wishing to “play the game,” and the rôle of relationships of power in

Any new evaluation of Roman-Parthian relations provokes hesitation, given the gallons of ink already devoted to this theme. Clearly from the C1st BC to the early C3rd AD the only two “superpowers” of western Eurasia shadowboxed 1

Cf. Syme 1958: 47, 493-97 on Rome’s over-emphasis on eastern affairs.

287

Limes XVIII

international affairs. Thus, not surprisingly, Ziegler’s book reflects the theories of his Doktorvater, Wolfgang Preiser, on periods of state-systems regulated by international law (eg. Preiser 1956; 1960; cf. Ziegler 1964: 1-2; 1972: 68-71; 1982: 66 n.11). Preiser’s periods of völkerrechtliche Ordnung actually correspond to Rostovtzeff’s discernment of periods of balance of power (Rostovtzeff 1922). Although Ziegler waffles on whether Rome ever recognized Parthia as an equal (1964: 96, 117, 137; 1972: 111, 112), he believes Roman-Parthian relations laid the foundation for Roman-Sasanid relations on a basis of equal sovereign states; only the demise of the Arsacids prevented further advancement to a real völkerrechtliche Ordnung (1964: 139, 149, 150-53; 1983: 13, 15). These views, however, are not uncontested (eg. Winter 1988: 216-31).

agreement at Rhandeia (Tacitus Anns. XV.27); Hadrian’s Parthian conference of 123 was backed up by troop movements (SHA Had. 11.8; CIL XIII 1802; Ritterling 1924: 1290; cf. Birley 1997: 151, 153-54; Kennedy 1997: 71-81 ad ILS 2724 ad.); and Antoninus Pius deferred a Parthian threat c.140 by a letter and vexillationes (SHA, Pius 9.6; ILS 1076; Saxer 1967: no.52). A legalisticdiplomatic approach, though lauding the virtues of peace over hostilities, can obscure the potential use of force lurking in the background and gloss over the tension generated in a crisis. Indeed a friendly state of Roman-Parthian relations can often be deduced from rather slim evidence. Soades, a Palmyrene merchant active in trade in lower Mesopotamia in the time of Hadrian and Antoninus Pius, erected a temple to the Augusti in the Palmyrene quarter at Vologesias near Ctesiphon, the Parthian capital, probably during Pius’ reign, though the act is not precisely datable. As unusual as this dedication may seem, is it a basis for large conclusions about amiable Roman-Parthian relations? (Sic Ziegler 1964: 111; Potter 1991: 283-84; but cf. Gagé 1959: 232 n.2). This view seems to assume Parthian policing of cult practices to enforce official toleration or prohibition of religious acts - hence a presumed Parthian religious policy that sought to control dedications and foundations in a huge multicultural, multi-ethnic state. But would the Parthian King of Kings know of or even care about such a temple? John of Ephesus (HE 6.23 = Brooks 1952: 245) mentions a statue of Trajan in Ctesiphon, which apparently had stood for centuries. The Persians regarded it with fear and suspicion, until Chosroes I (531-579) had it toppled. A foreign merchant’s private act of uncertain date hardly seems sufficient proof of official Parthian attitudes, especially considering the Parthian crisis of c.140 and the Parthian offensive against Rome in 161. A statue of Trajan in the Parthian-Sasanid capital, if John is credible, was surely more odious but tolerated than a temple to the Augusti in Vologesias.

For historians skeptical of the legalistic approach to Roman-Parthian relations (and Ziegler is not the only representative), this perspective is too rigid and unrealistic. For example, one argument against the historicity of Caracalla’s alleged proposal to marry the daughter of Artabanus IV (Herodian IV.10.1-2, 5-11.1; Dio LXXVIII.1.1) contends that the story must be false because this union with a peregrina would not be a iustum matrimonium, and Augustus had forbidden marriages between members of the emperor’s family and those of client kings and foreign rulers (Timpe 1967: 487). But, if the story is true, would an emperor like Caracalla care about or feel constrained by such legalities and Augustan precedents? Personalities of major players and the possibilities or limitations on use of power must also be considered. This is not the place to rehearse arguments for and against the existence of international law in antiquity, even if believers in such a system of law, always conditioned of course by its specific historical circumstances, basically assert only a set of generally agreed upon unwritten customary rules for interstate intercourse.2 For Romans and Parthians the point, however, is not international law but relationships of power between two empires that lacked the will and/or resources to conquer the other (cf. Ziegler 1964: 84). Hence posturing, diplomacy, and limited warfare, in which Parthians rarely engaged in major field battles and often seemed to concede Mesopotamia to Roman armies, who generally exhausted their impetus before ever reaching the Iranian heartland.

Even reading the sources becomes problematic in the absence of Parthian literary and documentary texts. Romans attributed the topos of the ‘tricky barbarian’ to the Parthians (Justin XLI.3.10; Wheeler 1993: 33 n.100). Ziegler (1964: 33-34; cf. 44) presents the interesting suggestion that, for Iranians, contrary to Graeco-Roman notions of bona fides, treachery at conferences with nonIranians may have been a legitimate stratagem. The unwritten rules of ius belli in antiquity do not preclude ethnic differences of opinion on such matters, although in some cases propaganda or literary embellishment in Roman sources cannot be ruled out (cf. Wheeler 1997a: 397). If Ziegler is right, however, then there may be more to Andreas Alföldi’s notion of a “moral limes” than recent attacks on this idea, obsessed with homogenizing Romans and barbarians, concede (Alföldi 1952; contra, Whittaker 1994: 122-23, 199-200; cf. Wheeler 1993: 25 with n.69).

If diplomacy rather than hostilities became the norm, agreements often did not come without force or duress. The sources attribute the famous Parthian return of the signa in 20BC to fear (RGDA 29; Justin XLII.5.10; Dio LIV.8.1; Orosius VI.21.29); Artabanus II in 36 ceased to contest Mithridates of Iberia as king of Armenia only when L. Vitellius threatened invasion of Mesopotamia from Syria (Tacitus Anns. VI.36.1); Corbulo’s Armenian offensive of 63 forced Vologaeses I into the final

Further the only programmatic statement of Parthia’s foreign policy in the west derives from Tacitus (Anns.

2

Cf. Ziegler 1972: 68-71; followed with hesitation by Schulz 1993: 13 n.1; cf. also Rostovtzeff 1922.

288

Everett L. Wheeler: Roman treaties with Parthia: Völkerrecht or power politics? VI.31.1), who reports Artabanus II’s claim to an Arsacid inheritance of the empire of the Achaemenid Persians. Cassius Dio (LXXX.4.1) attributes a similar claim to the Sasanid Ardashir I. Hence controversy over whether this is Roman propaganda, or not only Parthians but also Sasanids really considered themselves heirs of the Achaemenids.3 However this question is decided, a real Parthian claim to Achaemenid lands west of the Euphrates would justify a wary Roman defensive posture in the east, or if the view is Roman propaganda, then Roman aggression could always be seen as strategically defensive.

border (Timpe 1975: 167-68; Sherwin-White 1983: 32325). Even Augustus’ claim (RGDA 29) that he received the signa of three armies6 is erroneous. Antonius had already secured the return of his lost standards in 33BC through his treaty with Artavasdes of Media Atropatene.7 Return of the signa and Roman prisoners was a meaningless act for Phraates IV, though it probably secured Roman recognition of his legitimacy as Parthian king against pretenders (Wolski 1993: 147; Cimma 1976: 322). Augustus and Roman poets ruthlessly exploited this diplomatic coup for maximum propagandistic effect.8

But if Roman-biased sources require caution about Parthian intentions, they encourage similar hesitation about Roman behavior. Romans celebrated even lost wars as victories. As Festus reports (Brev. 20; cf. Heil 1997: 9 n.31, 196-97): “Nero lost Armenia,” but the festive crowning of Tiridates at Rome in 66 (Chaumont 1976: 119-23; Heil 1997: 131-33) obscured both the utter abandonment of Augustan policy in yielding the Armenian throne to the Arsacids and the achievement of a longsought Parthian goal. Flavian militarization of the upper Euphrates would be the consequence (cf. Chaumont 1969: 7-8; Wolski 1976: 211 n.58; Rémy 1986: 39-55; Wheeler 1993: 31-33). Indeed any thaw in Roman-Parthian relations after the Neronian compromise lasted less than a decade, especially if Edward Dabrowa (1994) is right in associating Ulpius Traianus’ obscure Parthian victory while governor of Syria with the bellum Commagenicum of 73 (but cf. Wheeler 1996: 257 n.59). Many years later the Romans would again claim failure as victory. Macrinus in 218 could take the title Parthicus Maximus and mint Vict(oria) Part(hica) on his coins after suffering defeat at Nisibis in 217 (Kneissl 1969: 166; cf., erroneously, Mattern 1999: 198 n.122.) and agreeing to pay Artabanus IV hefty reparations (Dio LXXVIII.27.1-3; Herodian IV.15.7-9; cf. Mattern’s curious equation of these reparations with a subsidy to a client king: 1999: 159 n.157).

What a slippery slope the study of Roman-Parthian relations can become emerges more clearly in examination of the treaties. The topic is a chimera. No documentary source preserves a single text of such treaties, nor did any historian consider it worthwhile to incorporate such documents in his works, though (in fairness) Cassius Dio (LIII.19.2-6) complained about the difficulty of obtaining accurate information on foreign affairs, and notes (LXXVIII.27.3) that Macrinus did not send a full account of his peace treaty with Artabanus IV to Rome. Indeed the only Parthian treaty for which terms are specifically attested is the last one, Macrinus’ treaty, whose terms Dio as a contemporary could report - at least in part (LXXVIII.27.1-3; cf. Herodian. IV.15.7-9). Hence what was or was not contained in any treaty before 218 amounts to the reconstructions of modern scholars uncontrolled by a documentary source. As well-known, Roman law emphasized oral declaration and performance; written forms of the action were secondary. This was true even for international treaties (Gaius Inst. 3.94; Ziegler 1964: 27-28; Millar 1988: 388). Generals of the late Republic were empowered to conclude treaties and alliances when expedient - the so-called Feldherrnverträge (Ziegler 1972: 94) - many of which must have been oral. Emperors made treaties at their own discretion; publication of details in the Senate was only a courtesy (ILS 244 lines 1-2; Dio LXXI.17.1; Ziegler 1964: 48-49, 55-56; 1972: 112). Many of these, too, especially with illiterate tribes, must also have been oral. But the Parthians were hardly illiterate. Whether Parthian treaties involved the elaborate procedures described by Menander Protector for the 562 treaty of Byzantium with the Persians (6.1, pp.70-76 Blockley; cf. Schulz 1993: 154-55). remains elusive in the absence of texts. Yet it is inconceivable that the bureaucratic Romans did not keep copies of treaties and records of agreements in Rome (cf. Suetonius Vesp. 8 with Ziegler 1972: 99; Millar 1988: 361). Oral treaties need not mean an absence of records.

But of course Augustus himself had provided the model for making mountains from molehills. The much celebrated Parthian return of the signa in 20BC is one of the great non-events in Roman history: no foedus was struck with Phraates IV;4 the Parthians conceded nothing about Armenia;5 and the Euphrates was not designated as a 3

Eg. Sonnabend 1986: 283-84; Winter 1988: 26-44; Wolski 1993: 197-99 and passim; Wheeler 1993: 30-31; Strobel 1993: 287-88; Wiesehöfer 1998b: 430-32. 4 Foedus only in Orosius VI.21.29; Propertius IV.6.79. Roman-Parthian negotiations in 20BC are not attested. 5 Erroneously, Suetonius Aug. 21.3, who also incorrectly puts the delivery of Parthian “hostages” in 20BC. But Phraates IV entrusted his four sons to Augustus only in 11 or 10BC: Livy Per. 141; cf. Strabo Geog. XVI.1.28. The date depends on the beginning of M. Titius’ governorship of Syria: see Krämer 1973b; Dabrowa 1998: 131 n.32. Delivery of the signa on 12 May (CIL I2 p.229; Ehrenberg-Jones2 p.48; Ovid Fast. 5.54597) makes it unlikely that Tiberius personally received the signa (Suetonius Tib. 9.1), as does the silence on this matter of Velleius Paterculus, Tiberius’ panegyrist. See Krämer 1973a. Armenian events of 20BC seem to be distinct (at least directly) from the return of the signa. It is notable that Roman sources never charge Parthia with violation of a

foedus for Arsacid interference in or invasion of Armenia. See below on Trajan and the supposed foedus of 66. 6 Those of Crassus (53BC), Marcus Antonius (36BC), and presumably Decidius Saxa (40BC). 7 Dio XLIX.44.2; Schieber 1979: 119; cf. Suetonius Aug. 21.3: signa of Crassus and Antonius only. 8 Cf. Dio LIV.8.1-3; RGDA 29; Syme 1939: 301-302; Rich 1998; other recent bibliography in Mattern 1999: 186-87 with nn.78-79.

289

Limes XVIII

But how many Parthian treaties were struck? Ziegler (1964: 22-31, 47-50, 54-56, 58-63, 73-77, 114, 120, 132, 134, 136) posits 9, perhaps 10: Lucullus (69BC), Pompey (66BC), Augustus (20BC, AD2), Germanicus (18 or 19), L. Vitellius (37), Nero (66), Marcus Aurelius (166), Septimius Severus (199), Macrinus (218), and possibly Sulla (96BC or 92BC). References in historians to a foedus in the late Republic never come from a source contemporary with an alleged treaty and, lest we forget, the late Republic is an extremely well documented period. Only Florus (I.46.4) calls whatever Sulla did a foedus; only Orosius (VI.13.2) mentions Lucullus’ foedus; only Florus (I.40.31, 46.4), Festus (Brev. 16; cf. 3), and Orosius (6.13.2) attribute a foedus to Pompey (cf. Dio XXVI.45.3, LI.1; Arnaud 1998: 19-21 on Plutarch Crass. 16.3). All other sources refer only to amicitia and/or societas, neither of which need indicate the mutually binding force of a foedus.

In the C2nd after Trajan, Parthian wars simply ceased when Romans got tired of them and went home. Pax was proclaimed. Though aware of no hard evidence for formal peace agreements in 166 or 199, Ziegler (1964: 114, 132) still believes such occurred, even if elsewhere he asserts (1972: 97; 1989: 46, 54) that pax in no way indicates a foedus. Hence, from this perspective, Macrinus’ peace agreement with Artabanus IV in 218 following the Roman defeat at Nisibis becomes the last and the only foedus with the Parthians (Herodian IV.15.9, V.1.4: cf. Dio LXXXVIII.26.8). Macrinus’ references in negotiations with Artabanus to Caracalla’s disregard of earlier σπονδαί (foedera) must surely be, in view of what we know, the rhetorical Herodian’s literary embellishment. Nevertheless, the absence of any foedus with Parthia before 218 does not mean a lack of Roman-Parthian agreements or unwritten understandings. Frézouls’ notion of Armenia as a circumscribed playground has substance: definite rules governed that game,10 whose analysis would require a separate paper. Moreover, the Parthians - at least as Roman sources portray them - seemed to think that amicitia, societas, and foedus were synonymous (eg. Tacitus Anns. II.58.1). One wonders if the Romans did not play a similar trick on the Parthians with foedus to that used with the Aetolians in 189BC in the famous confusion of πίσις and fides (Polybius XX.9-10; Livy XXXVI.27-28; cf. Wheeler 1984: 273 with n.67). Legalistic chicanery was typically Roman.

Much the same applies to the Augustan and Julio-Claudian periods. Augustus does not refer to a Parthian foedus in the Res Gestae (eg. 29, 32). Only Orosius (VI.21.29) uses that term for the non-event of 20 BC. Sherwin-White discerns correctly (1983: 220-28) that no foedus Parthicum existed before the meeting of Phraates V and C. Caesar in AD2 (for the date: Romer 1979: 208-10), but he argues on the basis of references in Tacitus (Anns. II.58.1-2; XII.10.1; XV.1.1) for a foedus in AD2. Tacitus, however, always puts the word foedus in a Parthian, never a Roman mouth. Further, one would think that the famous Neronian compromise involved a foedus. The word is conspicuously absent, and it is not certain that the agreement of Rhandeia established any rules concerning Tiridates’ successors (Ziegler 1964: 75; Wolski 1993: 169-72; Heil 1997: 12123, 136-40 but much exaggerated), as generally assumed, though the principle of the Rhandeia agreement is clear in Parthamasiris’ expectation that Trajan would re-crown him as king of Armenia in 114 (Dio LXVIII.17.2-3). Significantly, Trajan’s response in 113 to Parthian attempts to resolve the crisis over the Armenian succession mentions φιλα (amicitia), not a foedus (Dio LXVIII.17.3). When the Parthians sent an embassy to honor Nero’s memory (perhaps during Otho’s reign), they wanted to renew a societas, not a foedus (Suetonius Nero 57.2). Tiridates I, however, crowned by Nero in 66, surely was long dead before Trajan found a casus belli against Parthia in the Armenian succession in 113. Tiridates’ immediate successor is uncertain and details of his installation are unknown.9 We would like to know more about Zalaces, an Armenian hostage at Rome early under Trajan undoubtedly an Arsacid - whose fondness for sodomy attracted Juvenal’s attention (II.164; Ferguson 1987: 249).

If the foregoing arguments present an extreme case for the non-existence of a foedus Parthicum before 218 - and granted that Roman literary sources may totally misrepresent or distort the true state of affairs - two other points may be made in passing to conclusion. First, the sources lack an explicit statement that Rome ever recognized the Euphrates as a border. The concept did not apply to the upper Euphrates opposite Armenia (cf. Wheeler 1991: 505-11), and neither of two much cited passages on this point, Strabo Geog. XVI.1.28 ad init. and Tacitus Anns. I.11.4, will prove it.11 Like foedus, this idea generally comes from Parthian mouths (eg. Plutarch Pomp. 33.6; Tacitus Anns. XV.17.3), or is inferred from ceremonial arrangements of mid-Euphrates conferences (Ziegler 1964: 54, 62), symbolic acts but hardly legally binding. Again, Roman literary sources may be deceptive, but perhaps the status of the Euphrates de facto and de iure must be distinguished.12 Rome could in practice concede 10 Eg. a prohibition of stationing legions in Armenia in the Julio-Claudian period, though apparently auxilia were tolerated; cf. Wheeler 1996: 268. 11 Tacitus does not explicitly mention the Euphrates in this passage: Augustus addideratque consilium coercendi intra terminos imperii, incertum metu an per invidiam. Strabo’s ό̀ρ̀ ιον δ̉ έ́ στι τη̃ς Παρθναίων ὰ̀ρχη̃ς ὸ Εὺφράτης καὶ ή περαία is more problematic. Is he speaking literally or figuratively? A political, legally stipulated border or a geographical reference point? From one perspective any territory east of the Euphrates and south of the Taurus could be called Parthia (cf. ILS 9200: missus in Parthiam), but Osrhoene, wandering in and out of the Roman clientele from the days of Pompey on (eg. Dio XL.20.1-2) discourages an easy answer. 12 Zeugma’s function as a tax-collection post (Philostratus VA I.20; cf.

9

It is not impossible that Tiridates survived into the C2nd, since Rabbel II ruled in Nabataean Arabia 70-106 (Marek 1993: 130 n.30), although the last undisputed proof of Tiridates on the Armenian throne comes from his escape from the Alans c.75 (Josephus Bell.Jud. VII.249-50). For the possibility that a Sanatruces (cf. Arrian Parth. fr. 47 Roos; Dio LXXV.9.6) reigned after Tiridates I and before Axidares, see the arguments summarized in Chaumont 1976: 128-30. Potter (1979; 1991: 281 with n.11) does not solve the problem of Sanatruces.

290

Everett L. Wheeler: Roman treaties with Parthia: Völkerrecht or power politics? the Euphrates as a border, but deny legal sanction to such a boundary. Second, Romans never recognized Parthia as an equal and from Trajan on made few pretenses in that direction (cf. Timpe 1967: 485-86) - at least until the treaty of 218 compelled such recognition. From this perspective Julius Africanus’ indignation at Persian claims to ισοτιµία with Rome c.230 (Cest. 1.1.4-5 Vieillefond; cf. Wheeler 1997b: 575-76) acquires a new significance: Macrinus’ treaty must have been a true foedus aequum, something the Parthians had never had before and a new recognition of equality the Sasanids wished to exploit.

antico (Milan): 211-222. Birley A.R. 1997 Hadrian: the restless emperor. (New York). Bivar A.D.H. 1983 The political history of Iran under the Arsacids. In E.Yarshater (ed.) The Cambridge History of Iran III.1 (Cambridge): 21-99. Brooks E.W. (tr.) 1952 John of Ephesus Historia Ecclesiastica, Pars Tertia, Corpus Scriptorum Christianorum Orientalium 106: Script. Syr. 55 (Louvain). Brunt P.A. 1990 Roman imperial themes. (Oxford). Campbell J.B. 1984 The emperor and the Roman army, 31 BCAD 235. (Oxford). Campbell J.B. 1993 War and diplomacy: Rome and Parthia 31 BC-AD 235. In J. Rich & G. Shipley (edd.) War and society in the Roman world. (London): 213-240. Chaumont M.-L. 1969 Recherches sur l’histoire d’Arménie. De l’avèvement des Sassanides à la conversion du royaume. (Paris). Chaumont M.-L. 1976 L’Arménie entre Rome et l’Iran I. De l’avènement d’Auguste à l’avènement de Dioclétien. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.9.1. (Berlin): 71-194. Cimma M.R. 1976 Reges Socii et Amici Populi Romani. (Milan). Cornell T. 1993 The end of Roman imperial expansion. In J. Rich & G. Shipley (edd.) War and society in the Roman world. (London): 139-170. Dabrowa E. 1983 La politique de l’état parthe à l’égard de Rome - d’Artabane II à Vologèse I (ca 11-ca 79 de n.e.) et les facteurs qui la conditionnaient. (Cracow). Dabrowa E. 1994 The Bellum Commagenicum and the ornamenta triumphalia of M. Ulpius Traianus. In E. Dabrowa (ed.) The Roman and Byzantine army in the east. (Cracow): 19-27. Dabrowa E. 1998 The governors of Roman Syria from Augustus to Septimius Severus. (Bonn). Debevoise N.C. 1938 A political history of Parthia (Chicago). Ferguson J. 1987 A prosopography to the poems of Juvenal, Collection Latomus 200. (Brussels). Frézouls E. 1993 Les relations romano-parthes avant l’époque flavienne. In E. Frézouls & A. Jacquemin (edd.) Les Relations Internationales. (Paris): 479-498. Frye R.N. 1984 The history of ancient Iran. (Munich). Gagé J. 1959 L’Empereur romain et les rois: politique et protocole. Revue historique 221: 221-260. Hauser S.R. 1998 Hatra und das Königreich der Araber. In J. Wiesehöfer Das Partherreich und seine Zeugnisse. Historia Einzelschrift 122. (Stuttgart): 493-528. Heil M. 1997 Die orientalische Aussenpolitik des Kaisers Nero. (Munich). Isaac B. 1990 The limits of empire. The Roman army in the east. (Oxford). Kennedy D.L. 1996 Parthia and Rome: eastern perspectives. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 67-90. Kennedy D.L. 1997 The special command of M. Valerius Lollianus. In E. Dabrowa (ed.) Domum Amicitiae, Electrum 1. (Cracow): 71-81. Kennedy D.L. 1998 Ancient sources for Zeugma (SeleuciaApamea). In D.L. Kennedy (ed.) The twin towns of Zeugma on the Euphrates. Journal of Roman Archaeology Supplement Series 27: 139-162. Kneissl P. 1969 Die Siegestitulatur der römischen Kaiser, Hypomnemata 23. (Göttingen). Krämer K. 1973a Zur Rückgabe der Feldzeichen in Jahr 20 v. Chr. Historia 22: 362-363. Krämer K. 1973b Zum Freundschaftsvertrag zwischen Rom und Parthien unter Augustus. Klio 55: 247-248. Luther A. 1999 Medo nectis catenas? Die Expedition des

This brief paper could only hint at the vast complexity of Roman-Parthian relations, of which the meagre sources conceal more than they reveal. A realistic history of Roman-Parthian relations emphasizing relationships of power would have to be teased out of sparse, scattered references with careful analysis of the specific historical situations as well as Parthian and Roman motives and intentions at individual chronological points - all approached with a tabula rasa free of preconceptions. Fresh reassessments of Aelius Gallus’ Arabian expedition (Marek 1993; Luther 1999), for example, demonstrating that Augustus’ Parthian policy in the 20sBC was much more aggressive and opportunistic than the ‘hands-off’, passive position that has become canonical in scholarship, reveal the possibilities available for future work. The flaw in Ziegler’s excellent monograph lies in its legalistic approach rather than his detailed analyses. But pitfalls abound. If, like Ziegler’s study of diplomatic relations, Debevoise’s general history of Parthia (1938) begs for revision,13 the acta of a conference to lay the groundwork for such a project (Wiesehöfer 1998a) suggest both how little about the Parthian empire we really still know and the limits of any single individual to master the vast amount of material now extant from this multi-ethnic state. Yes, Parthian wars were like skin rashes, which the Romans frequently scratched, but a legalistic approach to RomanParthian relations does not help us understand the malady. Bibliography Alföldy A. 1952 The moral frontier on Rhine and Danube. In E. Birley (ed.) Congress of Roman Frontier Studies 1949. (Durham): 1-16. Arnaud P. 1993 Frontière et manipulation géographique: Lucain, les Parthes et les Antipodes. In Y. Roman (ed.) La Frontière. (Lyon): 45-56. Arnaud P. 1998 Les guerres parthiques de Gabinius et de Crassus et la politique occidentale des Parthes Arsacides entre 70 et 53 av. J.-C. In E. Dabrowa (ed.) Ancient Iran and the Mediterranean world. Electrum 2. (Cracow): 13-34. Barzanò A. 1985 Roma e i Parti tra pace e guerra fredda nel I secolo dell’ impero. In M. Sordi (ed.) La pace nel mondo I.38) did not rule out Roman control of Apamea opposite Zeugma, and other territory east of the Euphrates belonged to Zeugma’s χώρα: see Kennedy 1998: 147; cf. Wheeler 1991: 506. 13 Bivar (1983) and Frye (1984) offer convenient but conventional status quaestionis without breaking new ground in understanding the political and military history of Parthia.

291

Limes XVIII

Aelius Gallus in Rahmen der augusteischen Partherpolitik. Orbis Terrarum 5: 157-182. Marek C. 1993 Die Expedition des Aelius Gallus nach Arabien im Jahre 25 v. Chr. Chiron 23: 121-156. Mattern S.P. 1999 Rome and the enemy: Imperial strategy in the Principate. (Berkeley). Millar F. 1988 Government and diplomacy in the Roman empire during the first three centuries. International History Review 10: 345-377. Nicasie M.J. 1997 The borders of the Roman empire in the fourth century. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 455-460. Olbrycht M.J. 1998a Parthia et ulteriores gentes. (Munich). Olbrycht M.J. 1998b Das Arsakidenreich zwischen der mediterranean Welt und Innerasien. In E. Dabrowa (ed.) Ancient Iran and the Mediterranean world. Electrum 2. (Cracow): 123-159. Olbrycht M.J. 1998c Die Kultur der Steppengebiete und die Beziehungen zwischen Nomaden und der seßhaften Bevölker (Der arsakidische Iran und die Nomadenvölker). In J. Wiesehöfer (ed.) Das Partherreich und seine Zeugnisse, Historia Einzelschrift 122: (Stuttgart): 11-43. Potter D. 1979 The mysterious Arbaces. American Journal of Philology 100: 541-542. Potter D. 1991 The inscriptions on the bronze Herakles from Mesene: Vologeses IV’s war with Rome and the date of Tacitus’ Annales. Zeitschrift fur Papyrologie und Epigraphik 88: 277-290. Potter D. 1996 Emperors, their borders and their neighbors: the scope of imperial mandata. In D. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 49-66. Preiser W. 1956 Die Epochen der antiken Völkerrechtsgeschichte. Juristenzeitung 11: 737-744. Preiser W. 1960 Über die Ursprünge des modernen Völkerrechts. In Internationalrechtliche und Staatsrechtliche Abhandlungen: Festschrift für Walter Schätzel. (Düsseldorf): 373-387. Rémy B. 1986 L’Évolution administrative de l’Anatolie aux trois premieres siècles de notre ère. (Lyon). Rich J.W. 1998 Augustus’ Parthian honours, the Temple of Mars Ultor and the Arch in the Forum Romanum. Papers of the British School at Rome 66: 71-128. Ritterling E. 1924 Legio. RE 12. (Stuttgart): Cols. 1186-1829. Romer F.E. 1979 Gaius Caesar’s military diplomacy in the east. Transactions of the American Philological Association 109: 199-214. Rostovtzeff M.I. 1922 International relations in the ancient world. In E.A. Walsh (ed.) The history and nature of international relations (New York): 31-65. Saxer R. 1967 Untersuchungen zu den Vexillationen des römischen Kaiserheeres von Augustus bis Diokletian. Epigraphische Studien 1. (Köln). Schieber A.S. 1979 Antony and Parthia. Rivista storica dell’ Antichtà 9: 105-124. Schulz R. 1993 Die Entwicklung des römischen Völkerrechts im vierten und fünften Jahrhundert n. Chr. Hermes Einzelschrift 61. (Stuttgart). Sherwin-White A.N. 1983 Roman foreign policy in the east 168 B.C.-A.D. 1. (Norman). Sonnabend H. 1986 Fremdenbild und Politik. Vorstellungen der Römer von Ägypten und dem Partherreich in der späten Republik und frühen Kaiserzeit. (Frankfurt a.M). Strobel K. 1993 Das Imperium Romanum im 3. Jahrhundert:

Modell einer historischen Krise? Historia Einzelschrift 75: (Stuttgart). Syme R. 1939 The Roman revolution. (Oxford). Syme R. 1958 Tacitus. (Oxford). Timpe D. 1967 Ein Heiratsplan Kaiser Caracallas. Hermes 95: 470-495. Timpe D. 1975 Zur augusteischen Partherpolitik zwischen 30 und 20 v. Chr. Würzburger Jahrbücher für die Altertumswissenschaft N.F. 1: 155-169. Vogt J. 1955 Die Tochter des Grosskönigs und Pausanias, Alexander, Caracalla. In Gesetz und Handlungsfreiheit in der Geschichte. (Stuttgart): 55-80. Wheeler E.L. 1984 Sophistic interpretations and Greek treaties. Greek, Roman & Byzantine Studies 25: 253-274. Wheeler E.L. 1991 Rethinking the Upper Euphrates frontier: Where was the western border of Armenia? In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 505-511. Wheeler E.L. 1993 Methodological limits and the mirage of Roman strategy. Journal of Military History 57: 7-41, 215240. Wheeler E.L. 1996 The laxity of Syrian legions. In D. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 229-276. Wheeler E.L. 1997a The chronology of Corbulo in Armenia. Klio 79: 383-397. Wheeler E.L. 1997b Why the Romans can’t defeat the Parthians: Julius Africanus and the strategy of magic. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S. L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVth International Congress of Roman Frontier Studies. (Oxford): 575-579. Whittaker C.R. 1994 Frontiers of the Roman empire: a social and economic study. (Baltimore). Wiesehöfer J. 1998a Partherreich und seine Zeugnisse. Historia Einzelschrift 122. (Stuttgart). Wiesehöfer J. 1998b Zeugnisse zur Geschichte und Kultur der Persis unter den Parthern. In J. Wiesehöfer (ed.) Partherreich und seine Zeugnisse. Historia Einzelschrift 122. (Stuttgart): 425-434. Winter E. 1988 Die sasanidische-römischen Friedenverträge des 3. Jahrhunderts n. Chr.—ein Beitrag zum Verständnis der außenpolitischen Beziehungen zwischen den beiden Großmächten. (Frankurt a. M.). Wolski J. 1976 Iran und Rom. Versuch einer historischen Wertung der gegenseitigen Beziehungen. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.9.1. (Berlin): 195-214. Wolski J. 1993 L’Empire des Arsacids. (Louvain). Ziegler K.-H. 1964 Die Beziehungen zwischen Rom und dem Partherreich. (Wiesbaden). Ziegler K.-H. 1972 Das Völkerrecht der römischen Republik. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt I.2. (Berlin): 68-114. Ziegler K.-H. 1982 rev. Cimma Reges Socii et Amici Populi Romani. Labeo 28: 61-67. Ziegler K.-H. 1983 Tradition und Wandel im Völkerrecht der römischen Spätantike. In A. Böhm, K. Lüdersson & K.-H. Ziegler (edd.) Idee und Realität des Rechts in der Entwicklung internationaler Beziehungen: Festgabe für Wolfgang Preiser. (Baden-Baden): 11-31. Ziegler K.-H. 1989 Friedensverträge im römischen Altertum. Archiv der Völkerrechts 27: 45-62.

292

Die ‘Sklaven des Kaisers’ und der Kopf des Crassus. Römische Bilder des Ostens und parthische Bilder des Westens in augusteischer Zeit1 Josef Wiesehöfer verschleiern wußte: Weder war 20 v. Chr. am Euphrat oder in Armenien ein Krieg gewonnen noch war der Partherkönig in die Klientel des Princeps genommen worden; tatsächlich hatten beide Seiten einen Frieden ausgehandelt und dabei Zugeständnisse gemacht, Phraates IV mit der Rückgabe von Kriegsgefangenen und Feldzeichen, Augustus mit seinem Verzicht auf Rache für die Niederlagen des Crassus bei Carrhae (53 v. Chr.) und des Antonius in Armenien (36 v. Chr.) und mit der Aufgabe des Plans einer Provinzialisierung Armeniens.4 Die politisch kluge und militärisch angeratene Akzeptanz des status quo ante an der Euphratgrenze durfte nun aber angesichts der nur allzu bekannten militärischen Leistungen eines Pompeius und eines Caesar und der öffentlichen Meinung, die vom Princeps die Rache für Carrhae über Jahre hinweg geradezu gefordert hatte, nicht als bloße Friedens-, schon gar nicht als Verzichtspolitik daherkommen; vielmehr hatte der Partherkönig als von Rom abhängiger Partner zu erscheinen, der nicht nur den Frieden regelrecht erfleht, sondern der für ihn gar noch seine Familie als Pfand gegeben hatte. Söhne und Enkel des Phraates waren in Wirklichkeit erst 10 Jahre nach dem Abkommen nach Rom geschickt worden, keineswegs als Pfand für den Frieden, sondern, um dem Sohn aus der Verbindung mit der Königin Musa die gefährliche Konkurrenz der älteren thronberechtigten Halbgeschwister und ihrer Nachkommen zu ersparen. Von Rom in der Folgezeit vergeblich als Faustpfand in römisch-parthischen Angelegenheiten eingesetzt, helfen diese parthischen Prinzen gleichsam als ‘Wanderer zwischen zwei Welten’ heute immerhin dem Verfasser, nicht nur die bis heute in Europa geschichtsmächtige augusteische bzw. römische Bestimmung der eigenen kulturellen Identität und den römischen Umgang mit der kulturellen Alterität der politisch-militärischen Gegenspieler näher zu beleuchten, sondern auch der dabei viel zu kurz gekommenen parthischen Sicht der Dinge Aufmerksamkeit zu schenken.

I. Als nach dem Tode des Augustus dessen Tatenbericht im September des Jahres 14 n.Chr. im Senat verlesen und bald darauf auch im Reich verbreitet wurde, vernahmen die patres, Bürger und Provinzialen gleichsam aus dem Munde des Princeps selbst noch einmal die ihnen längst vertraute offizielle Bewertung der Beziehungen des Römischen Reiches zu seinen Nachbarn jenseits des Euphrats, den Parthern (RGDA XXIX; XXXII): Parthos trium exercitum Romanorum spolia et signa re[ddere] mihi supplicesque amicitiam populi Romani petere coegi.

und weiter: Ad [me re]x Parthorum Phrates, Orod[i]s filius, filios suos nepot[esque omnes] misit in Italiam non bello superatu[s], sed amicitiam nostram per [libe]ror[um] suorum pignora petens.

Alle, die diese Sätze hörten oder lasen, mochten sich dabei an die kniefällig ein römisches Feldzeichen darbietenden Parther auf den Reversen römischer Silbermünzen erinnert fühlen, an die orientalischen Stütz- und Trägerfiguren etwa der augusteischen Dreifußdenkmäler oder Bauskulpturen in Rom und Athen2 oder an ähnliche Formulierungen in der letzten Ode des Horaz aus dem Jahre 13 v.Chr., in der es heißt: ...tua, Caesar, aetas Fruges et agris rettulit uberes Et signa nostro restituit Iovi Derepta Parthorum superbis Postibus et vacuum duellis ... Non qui profundum Danuvium bibunt Edicta rumpent Iulia, non Getae, Non Seres infidique Persae, Non Tanain prope flumen orti.

II. Von den parthischen Prinzen Seraspadanes, Rhodaspes, Phraates und Vonones berichtet Strabon (Geog. XVI.1.28), sie hätten mit ihren Angehörigen und Begleitern auf öffentliche Kosten gelebt und man habe “auf königliche Weise” (basilikôs) mit ihnen verkehrt. Daß diese Vorzugsbehandlung nicht uneigennützig war, sondern zum Zwecke sowohl der Informationsbeschaffung als auch der Vermittlung romfreundlicher Einstellungen gewährt wurde, gilt als ausgemacht. Interessanter sind allerdings andere Aspekte des Aufenthaltes der Prinzen in Rom, der nämlich ihrer Beurteilung durch die alten parthischen und die neuen römischen Nachbarn und der des eigenen Verhaltens in neuer und alter Umgebung.

Nur die wirklich politisch Kundigen und die Gegner des Princeps, denen allerdings - anders als Augustus - weder die ‘Macht der Bilder’ noch die ‘Macht des Wortes’ eignete3, hätten damals deutlich machen können, wie großzügig Augustus in parthischen Angelegenheiten mit der Wahrheit umging bzw. wie gekonnt er Sachverhalte zu 1

Dieser Artikel wurde angeregt durch die Lektüre eines Beitrages meines Freundes R.M. Schneider (1998). Für Hinweise und Kritik danke ich H. Schneider (Kassel) und K. Heldmann (Kiel). E. Wheeler (Duke University) sei dafür gedankt, daß er die Thesen des Beitrags (in englischer Übersetzung) in Amman – in meiner Abwesenheit – vorstellte. 2 Zu den Parthern in der Bildkunst der augusteischen Zeit s. zuletzt Schneider 1998; Schäfer 1998; Spannagel 1999: 224-255. 3 Eine überzeugende Neueinschätzung der augusteischen ‘Macht der Bilder’ findet sich bei Hölscher 1999.

Von einem bereits im Jahre 30 v.Chr. als Gefangener nach 4

293

So, zu Recht, zuletzt betont von Bleicken 1998: 359-362.

Limes XVIII

Rom gelangten parthischen Königssohn (Dio LI.18. 3) berichtet der Biograph Sueton (Aug. 43.4), Augustus habe ihn und sein Gefolge ad spectaculum durch die Arena führen und allen anschließend besondere Plätze in seiner Nähe anweisen lassen.5 Die öffentliche Zurschaustellung der Fremden, von Sueton in einem Atemzug mit der Vorstellung eines Nashorns, eines Tigers und einer Riesenschlange erwähnt, war gewiß auch als politische Demonstration gedacht, machte allerdings vor allem Sinn zur Befriedigung populären Verlangens nach Kuriositäten und exotischen Objekten. Auch die vier Phraatessöhne und ihre Begleiter dürften als Protegés des Kaisers in den Jahren nach 10/9 v.Chr. zumindest für die Römer, die mit parthischen Angelegenheiten gänzlich unvertraut waren, mit ihrer ungewöhnlichen Tracht, ihrer spezifisch parthischen luxuria und ihrem fremdartigen Verhalten ein Fascinosum gewesen sein, ähnlich vielleicht dem Partherprinzen Tiridates c.70 Jahre später, dem in Rom von Nero in einer elaborierten römisch-orientalischen Zeremonie die Krone Armeniens aufs Haupt gesetzt wurde (Dio LXIII.1-8; Suetonius Nero 13.1-2). Die literarischen Zeugnisse der frühen Kaiserzeit, die auf die äußere Erscheinung parthischer Persönlichkeiten Bezug nehmen, legen allerdings Zeugnis davon ab, daß die Kennzeichen parthischer Würde und parthischen Lebensstils vielen Römern bei all ihrer faszinierenden Pracht doch auch als typische Merkmale eines spezifisch orientalischen orbis alter erschienen sein müssen. Das bereits vorhandene literarische Konzept einer parthischen Gegenwelt im Orient, dem sich diese Anschauung verdankte (vgl. Sonnabend 1986), wurde von Augustus nach 20 v.Chr. auch auf die Ebene des politischen Denkens transponiert und zugleich ikonographisch ‘ideologisiert’: Wie die schönen parthischen Prinzen als Schützlinge des Kaisers im Theater die Zuschauer zugleich beeindruckten und ungewollt deren stereotype Vorstellungen von orientalischen Potentaten bestätigten, so waren die östlichen Luxusdiener der augusteischen Kunst, aus kostbarsten farbigen Glanzsteinen hergestellt, mit ihrer traumhaften Schönheit, ihrer märchenhaften Fremdartigkeit und ihrem sagenhaften Reichtum nicht für den flüchtigen Blick, sondern für den bleibenden Eindruck geschaffene repräsentative Vertreter des populären Typus des ganz und gar unrömischen Orientalen. Die Luxuswelt des Orients übte dabei zwar durch ihre fremdartige Pracht eine besondere Anziehungskraft auf die Römer aus, doch assoziierten viele von ihnen mit ihr auch effeminierten Lebensstil, maßlose Sexualität sowie heimtückische Grausamkeit (Schneider 1998). Über die griechischen Vorbilder solchen Denkens wird noch zu reden sein.

seine Siege über Rom gesucht und diese Erklärung im ‘skythischen’, d.h. zugleich unzivilisierten wie unverdorbenen Charakter der Parther gefunden hatten; allerdings hatte schon der römische Historiker Pompeius Trogus unmittelbar vor Augustus’ Triumph betont (Iust. 41.2.4), mit dem Eindringen orientalischen Reichtums sei die kraftvolle Herrschaft der Parther gleichsam degeneriert.6 Gefährlich, wie noch zu Zeiten eines Crassus oder eines Antonius waren die Parther, so war die vom Princeps in Wort und Bild verkündete Botschaft, dank seiner überragenden militärischen und diplomatischen Fähigkeiten und der Hilfe der Götter, nun nicht mehr: Wie Phraates seine Söhne nach Rom geschickt hatte, um sich dafür die Freundschaft des Kaisers zu erkaufen und zu sichern, so illustrierten die bunten Barbarenstatuen und der das Feldzeichen darbietende Parther der Münzreverse die Inbesitznahme des Orients durch Rom und die unumkehrbare Rollenverteilung zwischen römischen Herren und parthischen Dienern. Allerdings war der Orientale, vor allem der Parther, in augusteischer Zeit auch bereits - daran sei erinnert – „ein wesentliches Element der römischen Identität, indem er ein kulturelles Gegenbild begründete, das für die Römer schon wegen seiner schillernden Bedeutung von außergewöhnlicher Faszination war” (Schneider 1998: 118). Dies alles waren Bilder, ideologische Entwürfe, die den eigenen Untertanen als Adressaten voraussetzten; im außenpolitischen Tagesgeschäft halfen sie kaum weiter, hier waren Konzepte gefragt, die versprachen, sich auf dem diplomatischen Parkett oder auf dem Schlachtfeld zu bewähren. III. Ein mögliches, wenn auch vor Augustus noch nicht praktiziertes Mittel römischer Diplomatie war die Unterstützung oder gar Entsendung von Thronprätendenten ins Arsakidenreich. Die Parther selbst hatten es ins Spiel gebracht, als sich vor und nach Octavians Sieg über Antonius und Kleopatra (31 v.Chr.) sowohl Phraates IV als auch dessen Gegenspieler Tiridates I um Unterstützung an Rom gewandt (Dio LI.18.2-3; Iust. 42.5.4) und später der zu Augustus geflohene Tiridates ihn sogar um Rückführung in seine Heimat gebeten hatte (Iust. 42.5.8). Damals hatte der Princeps, wohl in richtiger Erkenntnis der Kräfteverhältnisse im Reich der Feinde, diese Bitte abgelehnt (Timpe 1975: 166-168); bei den in Rom weilenden Phraatessöhnen stand dagegen zu erwarten, daß sie dem Imperium noch von einigem Nutzen sein würden: Nicht nur war Phraates’ IV Sohn und Mörder Phraatakes bei Adel und Volk verhaßt, nein, er selbst forderte gar als König die Auslieferung seiner in Rom weilenden Halbbrüder bzw. den Verzicht Roms auf deren Entsendung (Dio LV.10a.4.20) und wies ihnen damit unfreiwillig die Rolle von Faustpfändern in Augustus’ Händen zu. Dieser dürfte deshalb vorbereitet gewesen sein, als im Jahre 8 n.Chr. Vertreter der parthischen Aristokratie in Rom erschienen und um die Entsendung des ältesten der Prinzen, Vonones, baten (RGDA XXXIII; Tacitus Anns.

Mit dem propagierten Erfolg des Augustus traten damals vorübergehend die Überlegungen in den Hintergrund, die nach einer Erklärung für den gleichsam unaufhaltsamen Aufstieg des Partherreiches aus kleinsten Anfängen sowie 5

Bei diesem Prinzen handelt es sich um den von Tiridates entführten und von Augustus 23 v.Chr. zurückgeschickten Phraatessohn; er ist nicht identisch mit dem in RGDA XXXII erwähnten Phrates regis Phratis filius (vgl. dazu den grundlegenden Artikel von Timpe 1975).

6

294

Zum Partherbild des Trogus s. Wickevoort Crommelin 1998.

Josef Wiesehöfer: Die ‘Sklaven des Kaisers’ und der Kopf des Crassus II.1-2; Josephus Ant. Iud. XVIII.46); Augustus wird die Zeit nach der Vertreibung des Phraatakes durch eben diesen Adel (4 n.Chr.) und der Ermordung seines gleichfalls verhaßten Nachfolgers Orodes’ III genutzt haben, um den Phraatessohn auf seine zukünftige Rolle vorzubereiten und ihm die Pflicht zur Freundschaft mit Rom einzuschärfen. Nun stattete er ihn reich aus und sandte ihn ins Partherreich, wo Vonones zum König gekrönt wurde.

A me gentes Parthorum et Medoru[m per legatos] principes earum gentium reges pet[i]tos acceperunt, Par[thi Vononem, regis Ph]ratis filium, regis Orodis nepotem...

Auch die übrigen Vorwürfe an den aus der Art geschlagenen Vonones besitzen bei aller taciteischen Überund Verzeichnung einen historischen Kern: Aus vielen Zeugnissen wissen wir, für wie wichtig Arsakidenkönige und parthischer Hochadel Jagd, Pferdezucht und Reiterei, festliche Bankette und königliche Reisen über Land hielten; doch was einem Leser des Tacitus damals wie heute als überzogene, kleinliche Kritik erscheinen mag, erhält sein richtiges Maß erst bei näherem Hinsehen: Die Jagd ist in Iran immer wichtiger Ort königlicher physischer Bewährung gewesen; und als Abkömmlinge seminomadischer Steppenvölker stand das Pferd bei den Parthern in hohen Ehren es trug als theophores Tier den iranischen Gott Mithra und beim Herrscherzeremoniell den „König der Könige”, parthische Aristokraten zogen als schwerbewaffnete Kavalleristen in den Krieg.8 Bankette dienten in Iran zwar auch der Zurschaustellung herrscherlichen Reichtums, waren darüber hinaus aber vor allem Gelegenheiten der Begegnung von König, Aristokratie und Militär sowie des königlichen redistributiven Geschenkegebens (vgl. etwa Plutarch Crassus 33). Schließlich war das iranische Königtum immer ein ausgesprochenes ‚Reisekönigtum’, d.h. der Herrscher regelte nicht von einer fernen Hauptstadt aus, sondern vor Ort die Angelegenheiten der Regionen und Provinzen, von den Untertanen zeremoniell empfangen und mit den Gütern des Landes versorgt. Die Behauptung des Tacitus, die Parther hätten, anders als der ‚degenerierte’ Vonones, Zugänglichkeit und Leutseligkeit an einem König nicht geschätzt, mag auf den ersten Blick römischer Topik entsprechen, besitzt jedoch gleichfalls einen historischen Kern: Iranische Könige legten zwar auf die Begegnung mit den Untertanen großen Wert, gestalteten diese aber zugleich so, daß das magisterminister-Verhältnis immer gewahrt blieb; so speiste der parthische Herrscher auf einer besonderen, erhöhten und räumlich von den anderen getrennten klinê (Poseidonius FGrH 87 F 5), und selbst die ranghöchsten Mitglieder der ‚Hofgesellschaft’ beachteten durch die Pflicht zur Proskynese, d.h. zur Ehrenbezeugung gegenüber dem Herrscher, den königlichen Vorrang und die herrscherliche Überhöhung gleichermaßen (Wiesehöfer 2001).

Der neue König scheint, jedenfalls behauptet das Tacitus (Anns. II.2.2-4), das Nahverhältnis zu Rom zu sehr verinnerlicht, das Leben in Rom zu sehr genossen zu haben, stießen doch seine römischen Lebensgewohnheiten die aristokratischen Gönner schon bald vor den Kopf: „Bald aber regte sich ihr [der Königsmacher] Ehrgefühl: Entartet seien die Parther (degeneravisse); erbeten habe man aus einem anderen Erdteil ( alio ex orbe) einen König, der von der Feinde Gewohnheiten angesteckt sei; schon werde der Thron der Arsakiden wie eine von den römischen Provinzen behandelt und vergeben: Wo bleibe jener Ruhm der Männer, die Crassus niedermachten, Antonius davonjagten, wenn ein Sklave des Kaisers (mancipium Caesaris), der so viele Jahre die Knechtschaft erduldet habe, über die Parther gebieten dürfe? Ihre Ablehnung steigerte Vonones selbst, indem er ganz von der Lebensweise der Vorfahren abwich (diversus a maiorum institutis): er ging selten auf die Jagd, kümmerte sich zu wenig um die Pferde, benutzte bei seinen Reisen durch die Städte eine Sänfte, lehnte die heimischen Gelage ab. Man spottete auch über sein griechisches Gefolge und über die Sitte, die billigsten Gebrauchsgegenstände unter Siegel zu verschließen; andererseits erschienen seine offene Zugänglichkeit (prompti aditus) und seine entgegenkommende Leutseligkeit (obvia comitas), den Parthern unbekannte Tugenden, als zusätzliche Schwächen, und weil seine Handlungen ihren eigenen Gebräuchen fremd waren, galt ihnen ihr Haß, gleich ob sie verwerflich waren oder ehrenhaft.” (Übers. E. Heller)

In diesem Zitat mischen sich genaue Beobachtungen mit topischen Verzerrungen, und es ist nicht ganz einfach, beide voneinander zu scheiden7: Sicher ist, daß sich die Vorstellung, daß in Rom lebende parthische Thronprätendenten degenerieren, Tacitus’ prinzipatskritischer Einstellung verdankt; m.a.W. wer zu lange in der Nähe des Kaisers gelebt hat, verhält sich knechtisch und lasziv. Ob der so verweichlichte Vonones wirklich eine Sänfte benutzt und sogar einfache Gegenstände weggeschlossen hat, ist daher fraglich. Andererseits dürfen wir dem römischen Historiker wohl vertrauen, wenn er überliefert, die parthischen Aristokraten hätten angesichts des ihren Interessen abträglichen Verhaltens des Vonones die Tatsache gegen ihn ins Feld geführt, daß er ein Geschöpf des Augustus sei; nichts anderes hatte ja auch der Kaiser selbst in seinem Tatenbericht behauptet (RGDA XXXIII):

7

Vonones war, anders als uns Tacitus glauben machen möchte, kein schwacher, verweichlichter Monarch. Er war wohl auch stolz darauf, ein Parther und Arsakide zu sein, wie seine Brüder Seraspadanes und Rhodaspes, die sich auf ihren Grabdenkmälern an der Via Appia als Phraatis 8

Zum Ausdruck kommt diese offensichtliche Wertschätzung von Jagd und Reiten in Trogus’ Diktum: Carne non nisi venatibus quaesita vescuntur. Equis omni tempore vectantur; illis bella, illis convivia, illis publica ac privata officia obeunt; super illos ire, consistere, mercari, colloqui. Hoc denique discrimen inter servos liberosque est, quod servi pedibus, liberi non nisi equis incedunt (Iust. 41.3.3-4).

Zum Partherbild des Tacitus s. Ehrhardt 1998.

295

Limes XVIII

Arsacis regum regis filii und als Parthi bezeichnen (CIL VI 1799); doch wie diese ihre Inschriften in lateinischer Sprache (und nicht etwa in Parthisch, Aramäisch oder zumindest Griechisch, den Verkehrssprachen ihres Reiches) verfaßten, so mag auch Vonones mehr römische Allüren gezeigt haben als den parthischen ‚Königsmachern’ lieb war. Zusammen mit dem Faktum seiner Entsendung und Ausstattung durch Augustus rechtfertigte dieses Verhalten in ihren Augen seine Diskreditierung als mancipium Caesaris. Hätten die parthischen Aristokraten und der königlich-arsakidische Rivale des Vonones, Artabanos II, der jenen schließlich nach Syrien vertrieb, die orientalischen Stütz- und Trägerfiguren der augusteischen Staatsdenkmäler gekannt, so hätten sie sie sicher nicht auf sich selbst – wie von Augustus bezweckt -, sondern auf ‚Römlinge’ wie Vonones bezogen.

der makedonischen Nachfolger Alexanders im Osten, der Seleukiden, achaimenidische Institutionen und Herrscherbezeichnungen, nein, sie entdecken auch das neuerworbene Parthien als ihre ‚Heimat’ und den Perserkönig Artaxerxes II als ihren Ahnherrn. Wenn – nach Tacitus (Anns. VI.31) – 21 Jahre nach Augustus’ Tod der parthische König Artabanos II, der Gegenspieler und Nachfolger unseres Vonones, in einem Schreiben an Tiberius den von Vonones den Römern hinterlassenen Staatsschatz zurückfordert und dabei auf die alten persischen und makedonischen Grenzen verweist, wenn er droht, in die einst von dem persischen Reichsgründer Kyros und später von Alexander beherrschten Gebiete einzurücken, dann darf man diese Forderung und diese Ankündigung zwar sehr wohl als überzogene ideologische Form der Diplomatie bezeichnen, sollte beides zugleich aber doch als Versuch werten, der römischen Hellenennachahmung die auch im Westen nachvollziehbare Kyrosverehrung, der römischen imitatio Alexandri die parthische Alexandernachfolge entgegenzustellen (Wiesehöfer 1986: 177-185).

IV. Der barbarisch-verschlagene, der exotischverweichlichte und auch der schöne parthische Feind, der als Angehöriger des orbis alter so ganz unrömisch ist, steht in der literarischen wie ikonographischen Tradition griechischer Barbaren-, d.h. vor allem Perserbilder.9 Es verwundert daher nicht, daß unter Augustus nach 20 v.Chr. nicht nur die griechischen Persersiege der Jahre 490 (Marathon) und 480-479 v.Chr. (Salamis und Plataiai) „systematisch aktualisiert” wurden, sondern sich „die Gleichsetzung der Parther mit den Persern ... zu einem zentralen Punkt der römischen Kaiserideologie (entwickelte), durch den die Überlegenheit des Okzidents über den Orient als einzigartige Leistung der abendländischen Geschichte herausgestellt wurde” (Schneider 1998: 110-111). Und wie schon Alexander den Griechen gegenüber seine anabasis als Rachefeldzug für Xerxes’ Hellasunternehmung ausgegeben hatte, so stellte auch der erste Princeps die römischen Rüstungen gegen Parthien seit Caesar als Vergeltungsmaßnahmen dar. Wenn die Parther von den augusteischen Dichtern immer wieder als Meder, Perser oder Achaimeniden bezeichnet werden (vgl. Wissemann 1982), so finden solche Apostrophierungen ihre Entsprechung in der achaimenidisch-persischen Kleidung der Parther in der römischen Bildkunst oder der nachgestellten Naumachie von Salamis im Rahmen der Einweihungsfeierlichkeiten des Mars-Ultor-Tempels im Jahre 2 v.Chr. (RGDA XXIII; Velleius II.100; Plinius NH XVI:190.210; Statius Silv. 4.4.7; Tacitus Anns. XIV.15; Suetonius Aug. 43.3; Tib. 7.3; Dio LV.10.7).

Mit den Bezügen auf Alexander und seine Nachfolger stellt sich die Frage nach dem p a r t h i s c h e n Verhältnis zur Kultur ihrer zahlreichen griechischen Untertanen10: In die Tradition der griechischen Sieger und Siege von Salamis und Plataiai konnten und wollten sich die Arsakiden sicherlich nicht stellen, doch folgt daraus, daß sie ein gestörtes Verhältnis zu den Griechen in ihrem Reich gehabt haben müssen? Oder ist Augustus zuzustimmen, wenn er die Parther in den orientalischbarbarischen orbis alter verweist, in dem hellenische Bildung und Kultur kaum oder nur höchst unzureichend vorstellbar sind? Nicht nur viele Römer hätten in augusteischer Zeit diese Fragen mit Ja beantwortet, auch heute noch sind solche Einschätzungen weit verbreitet. Die mangelnde Vertrautheit der Arsakidenkönige mit griechischer Kultur gilt als ausgemacht, ihre Politik durch eine feindselige Haltung gegenüber diesen unzuverlässigen, weil seleukiden- und romfreundlichen Untertanen bestimmt. Ein näherer, hier nur exemplarisch vorgenommener Blick auf die Zeugnisse erweist beide Annahmen als falsch. In seiner vita des Crassus stellt der Biograph Plutarch den Partherkönig Orodes (II) folgendermaßen vor (XXXIII.2): “Während dies [der Triumph des parthischen Feldherrn Surenas über Crassus 53. v.Chr.] geschah, hatte Orodes schon mit dem Armenier Artabazes [Artavasdes] Frieden geschlossen und dessen Schwester mit seinem Sohne Pakoros verlobt; sie gaben einander abwechselnd Gastereien und Trinkgelage, und es gab dabei auch

Dienen die griechischen Perserbezwinger den Römern in augusteischer Zeit als zugleich nachahmenswerte und tatsächlich nachgeahmte, gar übertroffene Vorbilder, so die achaimenidischen Perser den Parthern nicht nur als Vorgänger in der Herrschaft, sondern auch als Legitimation verschaffende ‚Vorfahren’: Nicht nur übernehmen sie, wohl nicht zuletzt auch durch Vermittlung

10

Der Autor hat dieses Problem ausführlich an anderer Stelle diskutiert und zahlreiche weitere Beispiele (arsakidische Würdenträger auf Delos; parthische Hofgesellschaft; arsakidischer Herrscherkult; Heiratsverbindungen und Personen bei Hofe; Apollodoros v. Artemita; ‘Philhellenismus’ auf parthischen Münzen; der Fall Syrinx; der Fall Himeros; griechische Soldaten im Heer Phraates’ II; Philhellenismus und Iranismus) angeführt; Wiesehöfer 2000; er beschränkt sich hier auf vier besonders umstrittene Fälle.

9

vgl. zusammenfassend Raeck 1981, Schmal 1994 und Hutzfeldt 1999. Zur Vielfalt der tatsächlichen griechisch-persischen Beziehungen vgl. die vorzügliche Studie von Miller 1997.

296

Josef Wiesehöfer: Die ‘Sklaven des Kaisers’ und der Kopf des Crassus der in unterschiedlicher Tracht dargestellten Personen im einzelnen geklärt ist noch der Zweck der Gebäude, die sie schmückten, so kann doch kein Zweifel daran bestehen, daß der Partherkönig Mithradates I (171-138 v.Chr.) und seine Nachfolger sich ganz bewußt für griechische Künstler und z.T. auch griechische Themen entschieden haben müssen, als sie in Nisa ihre erste große Residenz errichteten. Auch in anderen Bereichen der Hofkunst ist der Einfluß griechischer bzw. hellenistisch-griechischer Vorbilder unverkennbar. Alles weist auf die intensive Beschäftigung parthischer Könige des 2. und 1. Jh. v.Chr. mit griechischer Kunst und den sich in ihr spiegelnden griechischen Vorstellungen zum Zwecke herrscherlicher Selbstdarstellung; daß hellenistische Repräsentationskunst ohne diesbezügliches königliches Engagement undenkbar ist, hat der Blick auf die Hofkunst der Vorläufer in der Herrschaft und der Nachbarn in Ost und West bewiesen.

künstlerische Darbietungen griechischer Herkunft. Denn Orodes war der griechischen Sprache und Literatur nicht unkundig (oute phonês oute grammatôn ... Hellênikôn apeiros), und Artabazes dichtete sogar Tragödien und schrieb Reden und Geschichtswerke, von denen einige noch erhalten sind.” (Übers. K. Ziegler)

Im folgenden schildert Plutarch die berühmte Szene, in der während einer Aufführung von Euripides’ Bakkchen der Kopf des Crassus überbracht wird und der Schauspieler Iason von Tralleis die gedankliche Verbindung zwischen Pentheus und dem römischen Feldherrn herstellt. Ausdrücklich betont er an dieser Stelle b e i d e r Könige Vorliebe für griechische Schauspiele, die sie bei ihren g e g e n s e i t i g e n Einladungen aufführen ließen; und es ist gerade der Gegensatz zwischen der griechischen Bildung des Arsakiden und dem barbarischen Umgang mit der Leiche des Crassus gewesen, der Plutarch zur ausführlichen Schilderung dieser Szene bewogen oder ihr zumindest besonderen Sinngehalt verschafft hat: “Plutarch’s Parthians are dangerously violent readers of Euripides” (Zadorojniy 1997: 182).

Unser drittes und viertes Beispiel führt uns in die Niederungen der praktischen Politik und in die Zeit des Augustus zurück: Schon länger wußte man um die Thronstreitigkeiten zwischen Phraates IV und Tiridates I in den 20er Jahren des 1. Jh. v.Chr., in deren Verlauf die seleukidische Gründung Seleukeia-am-Tigris mehrfach den Besitzer wechselte und sich ganz besonders Tiridates um Rückhalt in Rom bemühte (de Callatay 1994: 55-57). Der Kampf um den Thron und Tiridates’ Anlehnung an Rom spiegeln sich auch in den in Seleukeia geprägten parthischen Tetradrachmen wider, ließ Tiridates doch Münzen mit der Legende autokratôr philorhômaios prägen, einer Formel, die von Phraates nach seinem Sieg verständlicherweise überprägt wurde (de Callatay 1994: 42-47, 58-62). Auch wenn wir über die Rolle der Bewohner Seleukeias in diesen Jahren nichts erfahren, so macht die Münzprägung doch deutlich, wie sehr diese griechisch geprägte Stadt mit ihren immerhin c.600000 Einwohnern damals in einen innerarsakidischen Konflikt verwickelt war, der zugleich große außenpolitische Bedeutung besaß und auf die Zukunft vorverwies. Immer wenn in den folgenden Jahrzehnten ein Thronprätendent die Verbindung zum römischen Osten suchte oder gar von dort ins Partherreich geschickt wurde, mußte sich wie selbstverständlich die Frage nach der Parteinahme der griechischen Städte Mesopotamiens stellen, aus strategischen, aus wirtschaftlichen, aber eben auch aus politischen Gründen.

Gerade weil sich griechische Kultur nicht im literarischsprachlichen Bereich erschöpft, ‘Philhellenismus’ sich demnach auch auf anderen Feldern äußern kann, macht angesichts der romzentrierten literarischen Überlieferung der Blick auf die archäologischen Zeugnisse besonderen Sinn, für unsere Fragestellung vor allem der auf die Kunst der königlich-arsakidischen Residenzen in iranischer bzw. nichtgriechischer Umgebung. Für eine solche Untersuchung kommt vor allem Nisa im heutigen Turkmenistan in Frage, der erste Ort, an dem eine arsakidisch-parthische Kultur gesichert ist und definiert werden kann.11 Unter den dort gefundenen Zeugnissen aus der Zeit nach 171 v.Chr. ragen vor allem zwei Fundgruppen heraus: die berühmten großen Elfenbeinrhyta und die Monumentalskulpturen aus Marmor und Ton, alle in griechischer Tradition stehend. Von den Gefäßen mit ihren Endfiguren und dionysisch-mythologischen Friesszenen weiß man heute, daß sie in griechischhellenistischen Werkstätten in Ostiran bzw. Zentralasien geschaffen worden sein müssen und die Künstler im Auftrag der arsakidischen Könige tätig geworden sind. Die Rhyta wurden wohl im Rahmen iranischer Zeremonien bei der Herrscher- und Herrschaftsmanifestation benutzt. Eine inzwischen in Nisa gefundene griechische Inschrift eröffnet die Möglichkeit, daß auch die Rhyta dort, und nicht etwa im griechisch-baktrischen Reich weiter östlich, geschaffen worden sein könnten.

Ein solcher Fall war offensichtlich während der Regierungszeit Artabanos’ II gegeben: Nicht nur verzichtete der König in einer bestimmten Situation auf das Beiwort Philellên auf seinen Münzen (Sellwood 1980: 200-202), sondern er griff auch in die Autonomie der griechischen Städte ein und bezog in den internen Auseinandersetzungen deutlich Position.12 Am Beispiel Seleukeias erweist sich dabei, daß diese Politik erneut in

Die These von den griechischen Ateliers in Nisa wird vor allem durch die monumentalen Tonstatuen aus der ersten Konstruktionsphase der Gebäude gestützt, die wegen ihrer Materialbeschaffenheit vor Ort gefertigt sein müssen. Den Stil der Tonskulpturen hat man als hellenistisch mit deutlichem Bezug zur klassisch-griechischen Tradition des 4. Jh. beschrieben. Auch wenn bislang weder die Identität

12

Im Mittelpunkt der Diskussion stehen dabei der bekannte Brief Artabanos’ II an die Magistrate von Susa (RC 75) und Tacitus’ Bericht über die Politik dieses Königs gegenüber Seleukeia (Anns. VI.42ff.). Vgl. Le Rider 1965: 408-433; Dabrowa 1983: 79ff.; 1994: 185ff. [mit der älteren Literatur]).

11 Zu Nisa s. zusammenfassend Invernizzi 1998; Wiesehöfer 2000: 709711 jeweils mit der Forschungsdiskussion

297

Limes XVIII

deutlichem Zusammenhang stand mit Thronkämpfen und der römischen Einmischung in die inneren Angelegenheiten des Partherreiches durch die Entsendung von Thronprätendenten. Artabanos suchte nämlich in dieser Stadt die Unterstützung der primores (der politischen ‘Elite’) für seine innen- wie außenpolitischen Ziele, einer Gruppe, die wegen ihrer Überschaubarkeit offensichtlich leichter zu lenken war13 und ihrerseits vom Vertrauensverhältnis zum König zu profitieren gedachte; es verwundert nicht, daß der populus (das Volk) im Gegenzug den Gegenspieler Tiridates (der vom eben erwähnten Phraatesgegenspieler gleichen Namens zu unterscheiden ist) unterstützte, der nach seinem vorübergehenden Erfolg für ein ‘demokratisches’ Stadtregiment sorgte (Tacitus Anns. VI.42.3) und wohl auch die Autonomie der Stadt zu respektieren versprach. Der spätere Aufstand Seleukeias gegen Artabanos und seinen Sohn Vardanes galt wohl, sieht man von der Angst vor Repressalien ab, der Verteidigung dieser Verfassung und dieser Autonomie, wie die erste Maßnahme des Siegers (Tacitus Anns. XI.9), die Übertragung der Macht an die boulê, den von der ‘Elite’ dominierten Rat, beweist.14 - Wenn Artabanos und Vardanes die zweifellos griechische Oberschicht der Stadt unterstützen und sich dabei in Opposition zur Mehrheit der Bevölkerung befinden, wenn Artabanos sich des Beinamens Philellên enthält, dann kann dies nichts anderes bedeuten, als daß auch der populus zu großen Teilen aus Griechen oder doch zumindest hellenisierten Nichtgriechen bestanden haben muß (vgl. Dabrowa 1983: 84ff. bzw. 1994: 195ff.). Damit erübrigt sich zum einen die Interpretation der Auseinandersetzungen in der Stadt und des Aufstandes als ethnische Auseinandersetzung, zum anderen der Versuch, Artabanos prinzipielle Griechenfeindschaft zu unterstellen. Alles, was gesagt werden kann, ist, daß sich dieser König in einer bestimmten Situation von einer die Autonomie der Griechenstädte beschneidenden Politik eigenen Vorteil versprach15, einer Politik, die offensichtlich auch im Interesse der griechischen ‘Elite’ der Stadt war.16 Es kann, das hat die Untersuchung unserer Fälle ergeben, keine Rede davon sein, daß die Arsakiden und ihr Hof nur oberflächlich hellenisiert gewesen seien. Umgeben von griechisch geprägten Nachbarn, in der Nachfolge seleukidischer Könige und Satrapen und angegangen auch

von griechischen Poleis und Personen, hatten sie Gelegenheit und Grund genug, sich mit griechischen Traditionen vertraut zu machen; dabei ließen sie sich von ihnen faszinieren, ohne darüber ihre iranischen Wurzeln zu vergessen. Ein kulturelles und kulturpolitisches EntwederOder war ihnen, wie allen Herrschern der hellenistischen Welt des Ostens, fremd. Die Arsakidenkönige vor allem des 2. und 1. Jh. v.Chr. zeigten aber nicht nur persönliches Interesse an den Errungenschaften griechischer Zivilisation, sondern sie waren zugleich darum bemüht, ihre Vorgänger und Zeitgenossen politisch-ideologisch nachzuahmen und in ihrem Kreise Anerkennung zu finden: Politische Heiraten und Kontakte dienten diesem Ziel ebenso wie die Übernahme von Einrichtungen und Ideen zur Festigung der Herrschaft nach innen und außen. Nicht nur die griechischen Untertanen im eigenen Reich sollten sich angesprochen fühlen, wenn die Könige etwa hellenistische Herrscherepitheta, Herrschaftssymbole und Schutzgottheiten auf ihre Münzen prägen ließen, sondern auch die Nachbarn; ihnen wurde signalisiert, daß man sich (auch) zum Ensemble derjenigen Könige zu zählen gedachte, die in der Nachfolge Alexanders standen und sich durch Förderung griechischer Kultur auszeichneten. Konnten hier persönlicher ‘Philhellenismus’ und solcher aus außenpolitischem Kalkül zu gegenseitigem Nutzen eine Verbindung eingehen, so vermochten dies auch das Interesse der Könige an griechischer Kultur und paideia und ihre innenpolitische ‘Griechenfreundschaft’ aus Gründen der Staatsraison zu tun. Ein gutes Verhältnis zu den Poleis und den Griechen im Reich garantierte Ruhe und Ordnung und ermöglichte zugleich die intensive Erfahrung ihrer kulturellen Errungenschaften. Eine Störung des politisch vorteilhaften Verhältnisses zwischen Herrschern und griechischen Untertanen war dabei nicht notwendigerweise mit einer Abkehr der Könige von bisherigen persönlichen Interessen verbunden. Dies kann nur der annehmen, der das Phänomen des ‘Philhellenismus’ der Parther ausschließlich dem Bereich der Herrscherideologie zuweist und politische Maßnahmen gegen die Griechenstädte zum Beweis dafür heranzieht, daß es mit der Vorliebe parthischer Herrscher für Griechisches nicht weit her gewesen sein kann. Doch so wie niemand auf die Idee käme, römische Strafmaßnahmen in Griechenland im 2. Jh. v.Chr. als Ausweis römischen Desinteresses an griechischer Kultur zu interpretieren, statt dessen die Gründe und Anlässe für solche Maßnahmen zu erläutern bemüht wäre, so sollte man auch die arsakidische Politik etwa gegenüber Seleukeia oder Susa aus dem jeweiligen historisch-politischen Zusammenhang heraus zu erklären versuchen und den kulturellen, persönlichen ‘Philhellenismus’ der Könige gedanklich davon trennen.

13

Tacitus Anns. VI.42.2 formuliert dies so: paucorum dominatio regiae libidini proprior est. 14 Dies dürfen wir aus den Münzen des Vardanes schließen, die auf der Rückseite Bildnis und Legende der boulê tragen. Vgl. Sellwood 1980: 211f. - Es sei hier daran erinnert, daß – nach Philostratus Vita Apollonii 1.32 - Vardanes der Gesprächspartner des Apollonios v. Tyana gewesen sein soll, d.h. ein griechischer Kultur aufgeschlossener Monarch. 15 Umgekehrt würde man auch Tiridates und seine politischen Ziele mißverstehen, sähe man in ihm einen vorbehaltlosen “Griechenfreund”: In seinem Bemühen um politischen Rückhalt wandte er sich an den von Artabanos enttäuschten populus von Seleukeia, buhlte aber zugleich um die Unterstützung des parthischen Hochadels. 16 Es ist, nach dieser Beurteilung, aber auch angesichts des archäologischen und epigraphischen Befundes (etwa der Herakles-Statue mit der parthisch-griechischen Bilingue), verfehlt, das Jahr 42 n.Chr. als den Beginn des unaufhaltsamen ‘Abstiegs’ oder gar der ‘Orientalisierung’ der griechischen Stadt Seleukeia anzusehen.

V. Ziehen wir ein Fazit: Der parthische orbis alter mit seiner starken skythischen, iranischen und vorderasiatischen Prägung hätte vielen Römern zu Recht als fremde, aber eben auch faszinierende ‚Außenwelt’ erscheinen können; daß er ihnen stattdessen in viel stärkerem Maße als bedrohliche oder degenerierte ‚Gegenwelt’ erschien, war Ergebnis des konkreten politischen Handelns römischer Staatsmänner: des Crassus, 298

Josef Wiesehöfer: Die ‘Sklaven des Kaisers’ und der Kopf des Crassus der die Phase vertraglicher Beziehungen zwischen Rom und Parthien um der eigenen Vorteilssuche mutwillig beendete, Antonius’ und Caesars, die die Scharte von Carrhae auszuwetzen suchten und damit nur das Feindbild und die Kriegshysterie verstärkten, des Augustus’ schließlich, der um der Stärkung der eigenen auctoritas willen den Parthern zumindest ideologisch die Anerkennung ihrer Ebenbürtigkeit versagte und römische Identität auch durch Betonung der kulturellen Alterität des Feindes zu stiften suchte.

Literatur Bleicken J. 1998 Augustus. Eine Biographie. (Berlin). Callatay F. de 1994 Les tétradrachmes d’Orodès II et de Phraate IV. Studia Iranica 14. (Paris). Dabrowa E. 1983 La politique de l’état parthe à l’égard de Rome - d’Artaban II à Vologèse I (ca. 11 - ca. 79 de n.è.) et les facteurs qui la conditionnaient. (Kraków). Dabrowa E. 1994 Dall’ autonomia alla dipendenza. Le città greche e gli Arsacidi nella prima metà del I secolo d.C. Mesopotamia 29: 85–98. Ehrhardt N. 1998 Parther und parthische Geschichte bei Tacitus. In J. Wiesehöfer (ed.) Das Partherreich und seine Zeugnisse – The Arsacid empire: sources and documentation. Historia Einzelschriften 122. (Stuttgart): 295–307. Hölscher T. 1999 Augustus und die Macht der Archäologie. In A. Giovannini (ed.) La révolution romaine après Ronald Syme. Bilans et perspectives. Entretiens sur l’Antiquité Classique XLVI. (Vandœuvres): 237–271. Hutzfeldt B. 1999 Das Bild der Perser in der griechischen Dichtung des 5. vorchristlichen Jahrhunderts. Serta Graeca 8. (Wiesbaden). Invernizzi A. 1998 Parthian Nisa. New lines of research. In J. Wiesehöfer (ed.) Das Partherreich und seine Zeugnisse – The Arsacid empire: sources and documentation. Historia Einzelschriften 122. (Stuttgart): 45–59. Le Rider G. 1965 Suse sous les Séleucides et les Parthes. (Paris). Miller M.C. 1997 Athens and Persia in the fifth century B.C. (Cambridge). Raeck W. 1981 Zum Barbarenbild in der Kunst Athens im 6. und 5. Jahrhundert v.Chr. (Bonn). Schäfer Th. 1998 Spolia et signa. Parthererfolg und augusteische Baupolitik in Ost und West. Nachrichten der Akademie der Wissenschaften in Göttingen I (Philologisch-historische Klasse, Nr. 2). (Göttingen): 45–123. Schmal S. 1994 Feindbilder bei den frühen Griechen. Untersuchungen zur Entwicklung von Fremdenbildern und Identitäten in der griechischen Literatur von Homer bis Aristophanes. Europ. Hochschulschriften III, 677. (Frankfurt). Schneider R.M. 1998 Die Faszination des Feindes: Bilder der Parther und des Orients in Rom. In J. Wiesehöfer (ed.) Das Partherreich und seine Zeugnisse – The Arsacid empire: sources and documentation. Historia Einzelschriften 122. (Stuttgart): 95–146. Sellwood D. 1980 An introduction to the coinage of Parthia. (2nd ed.; London). Sonnabend H. 1986 Fremdenbild und Politik. Vorstellungen der Römer von Ägypten und den Partherreich in der späten Republik und frühen Kaiserzeit. (Frankfurt a.M.). Spannagel M. 1999 Exemplaria Principis. Untersuchungen zu Entstehung und Ausstattung des Augustusforums. Archäologie und Geschichte 9. (Heidelberg). Timpe D. 1975 Zur augusteischen Partherpolitik zwischen 30 und 20 v.Chr. Würzburger Jahrbücher für die Altertumswissenschaft N.F. 1: 155–169. Wickevoort Crommelin B. van 1998 Die Parther und die parthische Geschichte bei Pompeius Trogus. In J. Wiesehöfer (ed.) Das Partherreich und seine Zeugnisse – The Arsacid empire: sources and documentation. Historia Einzelschriften 122. (Stuttgart): 259–277. Wiesehöfer J. 1986 Iranische Ansprüche an Rom auf ehemals achaimenidische Territorien. Archäologische

Die Parther, oft genug durch Konflikte zwischen Herrschern und Aristokratie geschwächt, boten Rom Gelegenheit genug, sich in ihre inneren Angelegenheiten einzumischen. Der Philhellenismus ihrer Könige, sowohl ihr persönliches Interesse an griechischer paideia und Kultur als auch ihre Griechenfreundschaft aus Staatsraison, war nicht die Antwort auf römische zivilisatorische Überheblichkeit, sondern Frucht vorausgegangener jahrzehntelanger Erfahrungen. Neben den in dieser Zeit übernommenen griechisch-hellenistischen Grundlagen von Herrschaftslegitimation, Herrscherrepräsentation und Reichsidee existierten im Partherreich jedoch von Anfang an die später immer wichtiger werdenden diesbezüglichen iranischen Traditionen; auf beide berief man sich in der ideologischen Auseinandersetzung mit Rom, vor allem an letzteren maß man die herrscherlichen Qualitäten der aus Rom angeforderten oder unaufgefordert gesandten Thronbewerber. Tiridates oder Vonones, die zu ihnen zählten, fanden sich in keiner der beiden – ideologischen wie konkreten – Welten zurecht, weil sie – als Spielbälle von Kaiser und parthischem Adel – zu eigener Bestimmung der Spielregeln dieser Welten nicht in der Lage waren. Auf anschauliche, aber zugleich topisch verzeichnete Art schildert in diesem Sinne Tacitus (Anns. VI.32) das Schicksal auch des letzten der vier Partherprinzen, die einst nach Rom gesandt worden waren und denen von Augustus auch dort für alle Eventualitäten ein gleichsam königlicher Status gewährt worden war. Als, schon unter Augustus’ Nachfolger Tiberius, 35 n.Chr. erneut parthischaristokratische Gesandte in Rom erschienen, um „Phraates, des Königs Phraates Sohn” heimzuholen, „da stattete er [Tiberius] Phraates aus und versah ihn mit den für die Einnahme des väterlichen Thrones nötigen Mittel, indem er an dem Grundsatz festhielt, mit klugen Überlegungen und List die auswärtige Politik zu betreiben und Waffen aus dem Spiel zu lassen. ... Phraates (jedoch) kam nur bis Syrien: nach Aufgabe der römischen Lebensweise, an die er sich so viele Jahre gewöhnt hatte, wollte er die Gebräuche der Parther wieder annehmen, war aber der heimischen Lebensart nicht mehr gewachsen und wurde durch eine Krankheit dahingerafft (dum omisso cultu Romano, cui per tot annos insueverat, instituta Parthorum sumit, patriis moribus impar morbo absumptus est).”

299

Limes XVIII

Mitteilungen aus Iran N.F. 19: 177–185. Wiesehöfer J. 2000 Denn Orodes war der griechischen Sprache und Literatur nicht unkundig ...’ Parther, Griechen und griechische Kultur. In R. Dittmann (ed.) Variatio delectat: Iran und der Westen. Gedenkschrift für P. Calmeyer. Alter Orient und Altes Testament, 272. (Altenberge). Wiesehöfer J. 2001 Proskynesis. Der Neue Pauly 10. (Stuttgart): Cols. 443-444.. Wissemann M. 1982 Die Parther in der augusteischen Dichtung. Europ. Hochschulschriften XV, 24. (Frankfurt a.M.). Zadorojniy A.V. 1997 Tragedy and epic in Plutarch’s ‘Crassus’. Hermes 125: 169-182.

300

The Cappadocian frontier: from the Julio-Claudians to Hadrian Julian Bennett Cappadocia is one of Rome's least studied territories and the evidence for its garrison and frontier is frankly quite unequal to that for most of the other Roman provinces. However, as new material steadily accrues, systematic analysis permits hypotheses to be developed about the province and its garrison during the High Imperial period. As a result of fieldwork carried out over the last 20 years, both the nature of the Cappadocian frontier and its principal bases, as well as the location and purpose of its garrison are now better understood, allowing progress towards comprehending the role of both in tactical and strategic terms. In particular, the line and purpose of the via militaris and other routes in the region, along with their development, can now be clarified, while much more can be said and inferred about the fortresses at Satala and Melitene.

continuance of Armenian neutrality, from perhaps as early as 41, and certainly by 51, a Roman garrison, apparently a regular auxiliary cohort, was stationed at Gorneae, the summer residence of the Armenian rulers (Tacitus Anns. XII.45), and similar units may have been likewise stationed at other urban centres within the kingdom.

Introduction The Cappadocian frontier was among the longest in the Roman empire. Excluding the outlying posts in the Colchis, from its northern terminus at Trapezeus, on the Pontus Euxinus, it continued overland for some 800kms to the most southerly fort in the province, Barsalium, on the Euphrates. It is also the least known or investigated. As Everett Wheeler has stressed, the Cappadocian frontier is “…the poorest of all frontiers in physical and epigraphic material” (Wheeler 1999: 222), for despite the invaluable fieldwork of James Crow (1986), David French (1983: 8494), and Timothy Mitford (1974a; 1980; 1998), secure details with which to understand the disposition and nature of the garrison remain elusive. Indeed, we remain ignorant as to the precise locations of almost all of the military stations known or believed to have existed on the frontier some of which are now irrevocably lost beneath the waters of the Keban dam - while neither of the two sites of imperial date which have been located have yet been scientifically excavated. Matters are not helped by the lack of diplomata for the province, and the almost complete absence of other epigraphic material relating to the garrison. Even so, while firm physical evidence is wanting for much of the frontier, re-examination of the available historical and archaeological sources, along with continued fieldwork, does allow for some comment on its chronology and organisation from its genesis under the Julio-Claudians to its consolidation by Hadrian.

Whatever the nature of Rome’s military presence in Cappadocia and Armenia Major under Tiberius and Gaius, disturbances in the general region towards the end of Claudius’ reign (Tacitus Anns. XII.44-51) were the catalyst for major changes. A dynastic feud in Armenia Major in 51 prompted Julius Paelignus, governor of Cappadocia (cf. Rémy 1989: 180-181), to assemble his auxiliary forces (auxiliis provincialium: Tacitus Anns. XII.49) in an attempt to bring the kingdom under direct Roman control. On the eve of his invasion, however, they deserted en masse, fearing raids by barbarians into an defenceless Cappadocia, their response to the proposed campaign suggesting they were locally raised rather than ‘imported’ regular units. Claudius, fearing a retaliatory attack from the Armenian ruler, promptly ordered Helvidius Priscus to take his legion from Syria into the kingdom, but in the event, the even greater fear of provoking war with Parthia led to his recall after he had crossed the Kurdish Taurus mountains. Encouraged by such pusillanimity, the Parthian king Vologaeses took the opportunity to impose his own candidate on the Armenian throne in 52, thereby negating the Augustan arrangements for security in the region. Such remained the political situation in the east at the time of Nero’s accession in 54. Imitating his immediate predecessors, the new princeps sought to confirm his position by instigating a war, in his case to restore the status quo ante over Armenia Major, and the following year, he gave the responsibility for the campaign to Cn. Domitius Corbulo (for this and the immediately subsequent events, see Tacitus Anns. XIII.6-9; 34-36). For the task in hand, Corbulo was assigned three full legions, two from Syria, the third from Germany, along with their auxiliaries, as well as the auxilia in Cappadocia. In the event, while some skirmishing evidently took place after his arrival (Tacitus Anns. XIII.34), active campaigning was deferred until 58.

The Julio-Claudian Period A notable feature of the early Roman occupation of Cappadocia is that after its annexation by Tiberius in AD17, it became an equestrian province with an auxiliary garrison instead of a praetorian province with one or more legions. Such clearly indicates that Rome had no fear of any change in the agreed political status of the adjoining client kingdom of Armenia Major, despite the efforts of successive Parthian kings to impose their own control over its rulers by one means or another. Consequently, while it is not impossible that auxiliary garrisons might have been stationed along the upper Euphrates from an early date, as a precaution against any threat from the east, it is more likely that under Tiberius and, probably, Gaius, control of the province was maintained by a series of garrisons stationed at suitable tactical points and in urban settlements within the province proper. That said, to ensure the

According to Tacitus (Anns. XIII.35), one reason for this three-year delay was Corbulo’s need to ‘restore’ discipline to the Syrian legions and to train their newly-inducted 301

Limes XVIII

recruits from Galatia and Cappadocia, for which purposes he enforced a programme of year-round field training (cf. ILS 9108: Wheeler 1996, passim). Given that Corbulo’s battle-plan was based on a decisive thrust against Artaxata (Tacitus Anns. XIII.39), it is a reasonable conjecture that this activity was focused on the region around the Erzincan basin (Mitford 1974a: 166; Russell 1987: 249). Another reason for the delay, however, may well have been Corbulo’s decision to secure the region with a series of auxiliary forts ‘at appropriate locations’ before his campaign began, the responsibility for this being assigned to an ex-primus pilus, Paccius Orfitius: dispositisque per idoneos locos cohortibus auxiliariis ... curam praesidiorum Paccio Orfito primi pili honore perfuncto mandat (Tacitus Anns. XIII.36).

military bases at a later date: Zimara (NH 5.20), mentioned in connection with G. Licinius Mucianus, on Corbulo’s staff in 58; and Dascusa (NH 5.20 and 6.10), identified as a known way-station from at least the time of TiberiusGaius. For neither place, however, is there any physical evidence for Roman military activity at this date, although rescue excavations at Kilise Yazisi/Pagnik, a location which can be identified as or close to Pliny’s Dascusa, revealed a fortified enclosure associated with ceramic and other material of the general period (Sedaroglu 1970: 51; 1971: 30; 1972: 33). Despite understandable speculation (eg. Sinclair 1989: 94; Wheeler 1999: 223), the irregular layout of the fortifications preclude its identification as a Roman fort of the principate (Fig. 1), and they are best interpreted as an indigenous defensive structure of immediate pre-Roman date.

Some of these praesidia were certainly positioned along Corbulo’s northern supply route, from Trapezeus and across the Pontic Alps (Tacitus Anns. XIII.39), and terminating presumably in the vicinity of Satala, the springboard for the campaign against Armenia Major. Others may well have been built on the eastern littoral of the Pontus Euxinus: at Absarus and Sebastoplis/Dioscurias (both recorded as castella by Pliny NH 6.12; 14), and possibly at Phasis, where a regular Roman garrison was certainly stationed by the early years of Hadrian’s reign (Arrian Per. 9.4l: Crow 1986: 79; Speidel 1986: but note Braund 1989: 33). If these bases were indeed Neronian foundations, they may have been intended to secure the supply of foodstuffs from the Colchis or even further afield for Corbulo’s campaign. However, on strategic grounds it seems more likely that they were founded in the closing years of Nero’s reign for his projected ‘Sarmatian’ war, apparently against the Alani (Pliny NH 6.40; Tacitus Hist. 1.6; Suetonius Nero 19; Dio LXIII.8.1: cf. Braund 1989: 31; Sherk 1980: 992-994). Whatever, they were to be retained in use throughout the history of Roman activity in the east, apparently as much to secure trade in this area as to guard against internal unrest and external attack (Braund 1989, passim).

The finer details of Corbulo’s later activities in the east do not concern us here. Suffice to say that during the period 58-66, the situation was very fluid, with a series of Roman advances and reverses (Tacitus Anns. XIII.37-41; XIV.2326; XV.1-17, 25-31: cf. Mitford 1980: 1175-1179); and that these campaigns resulted in the creation of a number of temporary and semi-permanent military stations (eg. Tacitus Anns. XV.8), none of which have yet been identified on the ground. Of greater importance is that in 59/60, vigorous Parthian opposition to Tigranes V, newly installed by Rome as king of Armenia, forced Corbulo to relinquish control of Osrhoene and the forts he had built there (again, none have yet been identified), and it was agreed to restore the Euphrates as the formal limit of Roman control in the region: castella trans Euphraten amnemque, ut olim, medium faceret (Tacitus Anns. XV.17). While the context of this passage is usually taken to refer to the middle rather than the upper Euphrates, overall tactical and political realities indicate that both sections of the river are meant and that forts now began to built along the line, if they had not been earlier. Such is indeed confirmed by Tacitus, who in connection with the events of 68, makes a specific reference to the garrisons stationed along the Armenian border (quicquid castrorum Armeniis praetenitur: Tacitus Hist. II.6). Thus, whatever military dispositions Corbulo may have decided upon in the earlier years of his eastern commission, it is clear that between 60-68, the upper Euphrates was provided with a series of auxiliary units in permanent forts. Hence, the credit for creating the basis of a permanent frontier on the upper Euphrates-Trapezeus line should be given to Corbulo and Nero, and not Vespasian, as is generally stated (eg. Mitford 1980: 1180; Mitchell 1993: 118).

It is highly likely that Corbulo also established a series of praesidia along the west bank of the upper Euphrates at this date. After all, his main objective at this stage in his operations was Artaxata, in northern Armenia, and it is frankly inconceivable that this experienced general would have left Cappadocia open to attack from the southern part of the kingdom (cf. Tacitus Anns. XII.49). As it is, we know from Tacitus that several of the local dynasts in the region were ordered to prepare for offence and defence against Armenian and Parthian attacks, and we might assume that any existing local militias in the region were reinforced by regular auxiliary garrisons along the upper Euphrates itself. Thus, it is possible that many of the military bases known here at a later date owed their origin to this preliminary phase of Corbulo’s campaigns.

Further north, the kingdom of Pontus Polemoniacus had effectively been a military region of the Roman empire from at least 55 - indeed, between 61-63, the legio V Macedonica languished there awaiting action in northern Armenia Major (Tacitus Anns. XV.9). In c.63/64, the kingdom was formally annexed in uncertain circumstances (Sullivan 1980: 930) and attached to Galatia. The

Proof to support the hypothesis is lacking. Indeed, the only possible evidence in support is Pliny’s naming of two localities along the upper Euphrates which were certainly 302

Julian Bennett: The Cappadocian frontier: from the Julio-Claudians to Hadrian Trapezountian militia was now elevated to the status of a regular cohort, its members being given Roman citizenship, weapons and armour, and Trapezeus became the headquarters of a Roman fleet, the incipient classis Pontica (Tacitus Hist. III.47). It is generally and reasonably assumed that the annexation was to secure and make permanent the existing northern supply route from Trapezeus, both to maintain the standing garrison in Cappadocia and to ensure supplies for continued campaigning in southern Armenia. Wheeler (1999: 223), however, has drawn attention to the circumstance that the route is blocked in the winter months, with the implication it was only required for the early stages of Corbulo’s campaign. This does not affect the thesis: spring and summer were the principal season for building-up stockpiles and for replenishment, allowing the garrison to build a stockpile for periods when supply could not be guaranteed (cf. Tacitus Agr. 22.1). Mitford (1980: 11781179), on the other hand, believes that the annexation was a preliminary to a projected campaign in the Caucasus, rather than a consolidation of any upper EuphratesTrapezeus frontier system, for Armenia Minor and Commagene were still client states at this date. It seems highly unlikely, however, that Nero would have proposed a Caucasian campaign at this stage in his principate, given that the situation in Armenia was far from resolution, while the continued Roman presence in Pontus Polemoniacus and in Armenia Minor and Commagene - from at least 55 reveals that Rome did not need to annex a territory in order to place garrisons within it.

of the terrain, the contemporary political reality, and the predominately semi-nomadic nature of its contemporary inhabitants. To begin with, there are many valleys running into the Euphrates which provide a means of passage over the river from east and west. It was undoubtedly these valleys which served as conduit for the trans-Euphratean raiders who were seen as a real threat to the region from at least the Hellenistic period, when the rulers of Armenia Major had found it necessary to appoint men known as vitaxa, or “generals”, responsible for guarding the kingdom’s border regions. Moreover, these valleys also provided communication between two areas of shared cultural and social affinities. As such they were likely to have been used for trading, and were almost certainly used by contemporary transhumant farmers, migrating from the drier climate of the eastern Anatolian steppe to spend the summer months in the Armenian highlands (cf. Fig. 2), an activity which continues to this day. Some form of control of this porous political boundary was evidently required, although as already indicated, there is little evidence for it on the ground. Nonetheless, the locations of possible military sites of this and/or a later date are suggested by examination of the local topography, with regard to available west-east access routes in the region, and the general requirement for garrisons to be within a day’s march of each other. On this basis, by using the admittedly much later documentary evidence for occupation sites of Roman date in the region, as supplied by Ptolemy, the Antonine Itinerary, the Peutinger Table and the Notitia Dignitatum, and other sources, a provisional schedule of possible Roman military sites on the upper Euphrates-Trapezeus line identified and used at this and in later periods can be established (Table 1 and Fig. 3).

While some scholars would deny that there was ever any need for a regular system of auxiliary forts at any period along the upper Euphrates, never mind at such at early date (eg. Hodgson 1989: 181-182), this is to ignore the nature

LOCATIONS OF ROMAN MILITARY SITES ON THE UPPER EUPHRATES-TRAPEZEUS LINE Ancient Name Trapezeus

Modern Location Trabzon

Gizenica

Hortokop

Zigana

Zigana

Longini Fossatum

?Kale area

Suissa Domana

Ptol.Geog *

Ant.It. *

Peut. Tab *

Not.Dig. *

*

*

*

*

?Kelkit area ?Kose

* *

*

* *

Satala

Sadak

*

*

*

Sisila/Ziziola

?Erzincan area

*

*

*

Charax

Bozalik area

*

Analib(l)a

Hasanova

*

*

*

*

Zimara (?= ND Aeliana) Vereuso

Sekerhan/Pingan Gerusla/Egin

* *

*

*

(?*)

Sabus/Sabbu

Cit Harabe

*

*

*

Dascusa

Pagnik

*

*

*

*

Kiakis?Chiaca

Morhaman area

*

*

*

*

*

*

*

*

Other Pliny NH. 6.11-12

Proc. Aed. 3.6.23

Melitene

Eski Malatya

Ad Aras Korne

?Pirot ?Kerar Kale

*

Claudia/Claudiopolis

?Tillo area

*

*

Barsalium/Barzalo

Kilik

*

*

Proc. Aed. 3.6.22

Pliny NH 5.20.

Pliny NH 5.20 and 6.10

* *

303

Amm. Marc. 18/7/10)

Limes XVIII

that auxiliary garrisons were entrusted with this task. With regard to the first point, the (re-)construction of masonry auxiliary forts would be unusual at this early date, but is not without parallel (cf. Lander 1984: 20-30). Indeed, it is to be expected in a region where building stone is abundant, and sufficient quantities of suitable timber scarce. For the second point, it is usually believed that masonry auxiliary forts were built by legionary craftsmen, but this is yet another exception to disprove the rule. Otherwise all that can be said about the nature of the auxiliary garrison in Cappadocia under the Flavians is that it most likely occupied sites originally established by Corbulo.

The Flavian period From at least 62, if not as early as 55, Cappadocia, while technically an equestrian province, had in practice been a consular praetorian command, with two or more legions and supporting auxiliaries stationed there on an essentially permanent basis. Nero may well have intended to regularise this anomaly as a preliminary to his proposed ‘Sarmatian’ war, but as it was, the events of 68-69 postponed any such plans, and it was left to Vespasian to formalise the situation. He began in 70-71 with the annexation of Armenia Minor to Galatia (Mitford 1980: 1180). Then, in 72, Commagene was taken in a short if spirited war, the bellum Commagenicum, for which decorations were awarded to legionary centurions (Suetonius Vesp. 8.4; ILS 9198; AE 1943, 33). By the end of 72 or early 73, Commagene had been assigned to Syria, while Galatia and Cappadocia were combined to form a single consular praetorian province, most probably with M. Ulpius Traianus pater as its first governor (Bennett 2001: 17, 216, but note Dabrowa 1998: 162, n.657).

Such speculation apart, we have slightly more evidence for the fortresses established in Cappadocia during Vespasian’s reign, even though neither have been the subject of modern excavation. In the case of Melitene (Mitford 1980: 1186; Gregory 1996: 49-53), the Roman fortress is to be sought at or close to the medieval fortifications which enclosed the Byzantine citadel of the same name and the later Turkish settlement of Eski Malatya (modern Battalgazi). This location, east of the Dermes Su at an elevation of about 800m, is almost central to the Melitene basin. Thanks to the extensive alluvial deposits which cover the basin, and the protection afforded by the surrounding low hills, it is an extremely fertile region: Strabo (Geog. XII.2.1), for one, extolled its fruittrees, wine-grapes, olives and other crops, although today it is celebrated for its apricots. Its fertility aside, regional topography makes the Melitene basin a key transit point on the principal southern west-east route from Asia Minor into Armenia Major, via a river-crossing at nearby Tomissa (Harabakayis). The area, consequently, has a long history of occupation, the important Chalcolithic and Bronze age site of Arslanteppe being only some 5kms from Eski Malatya. Hellenistic Melitene, however, was probably located further south, at Kirman Höyük/Karamildan, about 20kms to the south. This strongly-defended site occupies the summit of a steep limestone outcrop, rising some 100m above the surrounding land, and has produced plentiful evidence for intensive occupation in the Hellenistic and early Roman periods (Sevin & Derin 1989). But while a superbly defensive position, it lacks an adequate natural water supply, making it unsuitable for a major Roman military base.

The end result was the formal completion of Nero’s plans for a viable political and military framework in the east, including the creation of a stable frontier zone on the upper Euphrates-Trapezeus line. Legionary garrisons were now fixed to supervise the main west-east routes in the region: at Melitene for the XII Fulminata (Mitford 1988: 174), and at Satala, for the XVI Flavia (Mitford 1974a: 164-166; 1997: 140-141). Sections at least of the existing trunkroutes in the eastern part of Cappadocia were soon being paved and marked by milestones (CIL III 306, of AD76, naming Cn. Pompeius Collega as governor: cf. Rémy 1989: 187-188). This activity may well have continued into the reign of Domitian, for a group of milestones from the vicinity of Ancyra, all of 80-82, recording A. Caesennius Gallus as governor (eg. CIL III.312; cf. Rémy 1989: 190192), specify inter alia roadworks in Cappadocia and Armenia Minor, although there is no supporting epigraphic record from either territory (for reports on the Roman road system in the region, see especially Mitford 1980: 11831185; 1989; 1998; French 1983: 84-94). With regard to the upper Euphrates-Trapezeus line, physical evidence for any auxiliary forts of the Flavian period is virtually non-existent. Other than noting that these may have occupied the locations already listed above, on grounds of geography and spacing alone, the only evidence is the building inscription of Pompeius Collega, of 82, discovered in a reused context at Pagnik Ögreni, and erected by a cohors II[ (Harper 1974: 108; Mitford 1974a: 172, proposing the coh. III Ulpia Petraeorum, a unit not in existence until at least 106: III Aug. Cyrenaicae eq. sag, (?IIII) Ituraeorum eq. sag, or (less likely at this date) IIII Raetorum eq. are all possibilities). The inscription evidently indicates the (re-) building of the fort at Dascusa, but of greater importance is that it suggests that (a): the Flavian emperors may well have initiated a programme to (re-)build a series of stone forts along the upper Euphrates-Trapezeus line, and (b):

As others have noted, the outline of the Roman fortress is apparently preserved in the south-west corner of the medieval defences of Eski Malatya (Gabriel 1940: 196; Mitford 1980: 1186; Gregory 1996: 51-53; but cf. Crow 1986: 85-86). Field examination in 2000 indicated that the north-south road which bi-sects the medieval enceinte between the Adile and Ulu Camii delimits the eastern side of a raised platform, some 1-2m higher than the surrounding area. On this basis, one can restore the outline of a regular playing-card shape fortress in the south-west part of Eski Malatya, whose defences enclose an area approaching some 22.4ha, somewhat larger than might be expected for a Flavian period fortress (Fig. 4). 304

Julian Bennett: The Cappadocian frontier: from the Julio-Claudians to Hadrian Furthermore, it can be surmised that the fortress faced either north or south, and if the latter, the line of the original via principalis and via praetoria sinistra may well be preserved in the modern road system.

of re-building, one of them presumably to be dated to the reign of Justinian, who re-fortified the site in the early C6th (Procopius Aed. 3.4.2-5), while traces of some of the square projecting towers referred to by Biliotti are still visible in drought conditions. An early imperial date for the footings of the surviving walls, however, is suggested not simply by the limited and regular area they enclose, but also by the existence of a chamfered plinth course along the south side of the defences, and the discovery that the square projecting tower at the south-west angle was added to an earlier curved corner, a feature of early imperial date (Lightfoot 1991: 300; 1998: 278).

Despite a steady programme of building activity at the site, about half of which remains as open fields, archaeological evidence of any pre-Islamic use is almost non-existent. Indeed, the only recorded evidence for Roman period occupation is a single and undated tile-stamp of the legio XII Fulminata (Mitford 1988: 174) - there is not even any evidence of reused Roman architectural material in the standing early Islamic buildings of Eski Malatya. As Gregory (1996: 52) has indicated, this could be because the Roman occupation layers lie at some considerable depth below the modern ground surface, for the C12th Ulu Camii, due east of the presumed fortress, is now entered by a 2-3m deep descending stairway. Likewise within the fortress: in 1999, covert examination of a 3m deep cellartrench north-east of the Adile Camii revealed no evidence for any occupation layers, the nature of the visible deposits suggesting these represented the accumulated and redeposited debris from demolished mud-brick buildings.

With regard to the interior of the fortress, casual digging in c.1872 produced the twice life-size bronze head and arm from a statue of Aphrodite, both passing almost immediately into the collections of the British Museum. It is generally agreed that this statue was either of Hellenistic date, from about 150BC, or was an early Roman copy of an original of that period. That apart, Biliotti’s own excavations and investigations at Satala identified a variety of standing and demolished ancient structures, none of which need date to an early stage of the site’s Roman occupation (Mitford 1974a: 234-239). Worthy of note, however, is his observation that when digging in the centre of the site, he often reached depths of up to 2m before encountering ancient walling or other material (Mitford 1974a: 237), indicating that here, as at Melitene, the early Roman levels must lie some depth below the modern ground surface

We know a little more about Satala, thanks to the excavations in 1874 by Biliotti (Mitford 1974b), and more recent studies by Mitford (1974a: 163-167; 1977: 503; 1980: 1187) and Lightfoot (summarised in Lightfoot 1998: for the epigraphic record from the site, see French & Summerly 1987; French 1994: 43-44; Mitford 1997). It is located in a bowl-shaped depression at an altitude of about 1650m, and commands three main transit routes into Armenia Minor and Iberia (Bryer & Whitfield 1985: 33). Lightfoot (1998: 277) has drawn attention to the unusual setting of the site, at such a high altitude, in an area subject to heavy winter snowfall. But, as is often the case in semiAlpine regions, the Satala basin benefits from a natural micro-climate which permits the growing of most staple crops to meet at least subsistence demands on a year-round basis. There is archaeological evidence for activity here in the Mesolithic and Bronze ages (Lightfoot 1998: 281), but nothing to indicate any permanent occupation until the mid-C1st BC, when the place was apparently alluded to by Cicero (RE 1 (1894) 1448). The exact location of this possible immediate pre-Roman occupation remains uncertain, although it was most probably on the 45m high rock outcrop which dominates the south-east angle of the Roman fortress.

From Trajan to Hadrian The Cappadocian frontier and its garrison played a significant role in the early stages of Trajan’s Parthian War. Satala, indeed, was the assembly point for the first stage of the campaign, against Armenia Major, with perhaps as many as 8 or 9 complete legions and vexillations from about 8 others brought together here by May 114 (Bennett 2001: 192). For security reasons, however, the forts and fortresses on the upper EuphratesTrapezeus line must have retained some form of garrison for the duration of the campaign, and it would appear that one at least, Zimara, saw some building activity in the period 114-117. Such is the import of an inscription from the general location of this site, recording construction under the praetorian legate L. Catilius Severus Julius (Mitford 1974a: 171). This man had been appointed governor of the joint province of Cappadocia-Armenia Major in 114, and was presumably governor of Cappadocia alone after Armenia Major was surrendered by Trajan in 116, before taking command of Syria in the late summer of 117 (cf. Rémy 1989: 206-208). While not necessarily military in origin, the inscription may well indicate reconstruction at Zimara as part of an overall restoration of the upper Euphrates-Trapezus line in the period 116-117. Whatever, we can assume that after Hadrian formally relinquished Trajan’s eastern conquests in 117, the Cappadocian garrisons were brought back to full strength, with the XV Apollinaris replacing the XVI Flavia Firma at

The fortress defences are still quite visible for much of their length, if obscured in the west by the modern village of Sadak, and enclose an area of about 15.8-16.5ha (Fig. 5): small, but comparable to Nijmegen and Lincoln, at 16.5 and 16.8ha respectively. A plausible explanation for its relatively small area is that a sizeable part of the legion was permanently out posted, some legionaries certainly being based in Ankara (cf. Bosch 1967: nos. 110-112), other perhaps forming part of the garrison of the Colchis (cf. Arrian Per. 6; but note Speidel 1983: 16-17; 1986). Examination of the defences indicates at least two periods 305

Limes XVIII

Satala (French & Summerly 1987: 21-22). After, all while peace had been agreed between Rome, Parthia and Armenia Major, there remained a real fear of a retaliatory attack from Parthia (cf. SHA Had. 12.8, apparently referring to events in 123).

there (Gregory 1996: 11-15), together with the presence of Hadrianic coins, strongly suggest he was subsequently ordered to garrison the place (Braund 1989: 33). It is also to Arrian that we owe our most valuable source for the nature of the garrison in Cappadocia. During his tenure, the region was threatened by at least one Alanic incursion, probably in 135, and his report of the measures he took against this, the Ektaxis kat’ Alanon, provides us with our only listing of the province’s garrison during the principate (Arrian Ekt. 1-9). In all, besides the two Cappadocian legions, he lists no less than four alae (II Ulpia Auriana; I Aug. Gem. Colonorum; I Ulpia Dacorum; and II Gallorum) and nine cohortes (Apuleia c. R; I Bosporanorum m.eq; III Aug. Cyrenaicae eq.sag; I Italica eq; (?IIII) Ituraeorum eq. sag; I (?Flavia) Numidarum eq; III Ulpia Petraeorum m. eq. sag; I Raetorum eq; and IIII Raetorum eq). In addition, as supporting forces there were four units of numeri, the symachiarii Armenias Minoris sag; Colchoi; Rhizias (hastiferi) and Trapezeuntis (contrariorum). As Speidel (1983: 17) has observed, the irregular troops apart, almost all of the auxiliary units listed in the Ektaxis are of European origin. Indeed, with the obvious exception of the Dacian and the Petraean units, which cannot pre-date the reign of Trajan, it is quite conceivable that most of them had been in the province from the date of its annexation. The possibility is strengthened by the fact that none of these units are recorded in the available corpus of diplomata as having served in any other province, with the possible exception of the cohors IIII Raetorum eq. (cf. RMD 1.6, Moesia Superior, of 96).

In 129, Hadrian visited Cappadocia as part of his overall inspection of the eastern provinces (SHA Had. 13.7). A primary purpose of his visit was to review the garrisons and military dispositions and facilities of the region (cf. the exercitus Cappadocicus issues of 129: RIC 914), and an immediate result was the building of a new harbour at Trapezeus, presumably to provide better shelter for the classis Pontica (Arrian Per. 16.6). In addition, he now provided the king of Iberia with a regular quingenary cohort (SHA Had. 17.10), probably to be stationed at his capital of Harmozica, and to assist in the defence of the kingdom against Alanic attack. Given the reality of the threat from this region (cf. Arrian Per. 6-11), we might reasonably assume that like units were assigned to the kings of Albania and Armenia Major as well, just as they had been in Flavian times (cf. Crow 1986: 80; Mitchell 1993: 119-120). While in Cappadocia, Hadrian also received a body of slaves donated by the province for unspecified ‘service in the camps’ (a Cappadociubus servitia castris profurtura suscepit: SHA Had. 13.7). The terminology used is ambiguous, and while these men may have been provided for use as slaves by the Cappadocian and other garrisons, perhaps even as labourers on construction works, it could be that they were provided as military recruits. True, in normal circumstances slaves were denied entry to the regular Roman army. Augustus, however, when faced with emergencies in both Illyricum and along the Rhine, did recruit and emancipate slaves in these regions for immediate military service (Suetonius Aug. 25.2). As it is, Hadrian’s reign was seemingly characterised by a shortfall in recruitment for the legions at least, for as early as 123, he had attempted to re-introduce the levy in Hispania Tarraconensis, Trajan having exempted the province, probably at the time of the Dacian wars (SHA Had. 12.4; cf. SHA Marc. 9.7). It is possible, therefore, that in Cappadocia, Hadrian resorted to the same procedure used by Augustus as a means of bringing new and desperately needed men into the Roman army, although if this was indeed the case, these Cappadocians are more likely to have been subsumed into the auxilia or - more likely were formed into units of numeri.

It is highly improbable that Arrian stripped the province of troops for his counter-attack against the Alani. Such is in fact indicated by the presence of a vexillation only of the leg. XII Fulminata, the major part of the legion evidently being left at Melitene to secure the southern part of the Cappadocian frontier. Exactly how many other units may have remained in the province at this time is unclear, although on the basis of the epigraphic evidence, Speidel (1983: 16-17) has suggested a further three cohortes which may have been stationed in Cappadocia at this period: the I Apamenorum sag. (AE 1977: 183); I Claudia eq. (at Apsarus: AE 1977: 802 and Ch.La. 11.477); and an unnamed cohors mil.eq. (ILS 9117). Additional units are, of course, possible: an inscription of 198, for example, records the cohors I Lepidana c. R. (AE 1908, 22), and it could have been in the province long before that date. Even so, as it stands, the evidence indicates that early C2nd Cappadocia probably had an auxiliary garrison of some four alae and at least 12 cohortes, a complement of about 10,000 or so auxiliaries. This is slightly less than the number of legionaries in the province, but is close enough to conform to the general rule that (praetorian) provinces contained roughly equal numbers of both auxiliaries and legionaries each (Tacitus Anns. IV.5; see Appendix).

Of greater relevance to the history of the Cappadocian frontier is Hadrian's subsequent appointment of L. Flavius Arrianus as governor of the province in 131. His tour of the eastern littoral of the Pontus Euxinus appears to have been at the behest of Hadrian himself, and certainly resulted in the reconstruction - in brick - of the fort at Phasis (Arrian Per. 9). His report also quite possibly led to the building of a fort at Pityus: while he is specific that no garrison existed there at the time of his visit (Arrian Per. 17), the remains of a timber fort beneath the later stone fort

To conclude. Despite some three decades of near continuous fieldwork, the Cappadocian frontier remains 306

Julian Bennett: The Cappadocian frontier: from the Julio-Claudians to Hadrian the least known of all Rome’s frontiers. The situation is unlikely to change significantly in the immediate future, given the nature of the terrain, the lack of resources, and the constraints of local official procedure, which conspire together to preclude a fully comprehensive and systematic programme of fieldwork in the region, never mind excavation. Even so, the steady trickle of raw data and a re-assessment of the historical sources does allow for an increasingly more precise understanding of the frontier’s chronology, nature and organisation, which serves to rectify earlier observations, and identify where future priorities should be directed.

units in Cappadocia is apparently provided by the Notitia Dignitatum (Or. XXXVIII. 10-30: cf. Table 2). It lists 23 auxiliary units under the command of the dux Armeniae in Armenia Pontus, the successor province to Cappadocia, 11 of them being alae, 12 mixed or infantry units. Of the alae, four are those recorded by Arrian, while all the rest have titles indicating they were post-C2nd creations or newlyraised units: ala Rizena; ala Theodosiana; ala felix Theodosiana; ala constituta castellum Tablariensis; ala prima praetoria recens constituta; ala I Iovia felix; ala I felix Theodosiana. As for the cohortes, at least six of those listed in the Notitia are certainly attested in Cappadocia before the end of the C2nd: Apuleia c. R.; (I) mil. Bosporiana; I Claudia eq; I Lepidana; III Ulpia mil. Petraeorum; IIII Raetorum. A seventh may be the cohors mil. eq. of ILS 9117, for this could be the same as the cohors (?I) mil. Germanorum of the Notitia, while Arrian’s III Aug Cyrenaiciae eq. sag. and (?IIII) Ituraeorum eq. sag. are surely the two units of equites sagitarii recorded in the same list. Of the remaining three cohortes under the

Appendix: The upper Dignitatum

Euphrates

frontier

and

the

Notitia

Some support for the number of four alae and 12 or so cohortes as being the regular complement of auxiliary

THE GARRISON OF ARMENIA-PONTUS IN THE NOTITIA DIGNITATUM OR 38 ND REF

UNIT

FORT

MODERN LOCATION

11

Equites sagittarii

Sabbu

Cit Harabe

12

Equites sagittarii

Domana

?Kose

13

Leg. XV Apollinaris

Satala

Sadak

14

Leg.XII Fulminata

Melitene

Eski Malatya

16

Leg. I Pontica

Trapezeus

Trabzon

17

Ala Rizena

Aladaleariza

Golova/Agvanis

18

Ala Theodosiana

Avaxa

Avaza/Avsa

19

Ala felix Theodosiana

Silvana

?Sule

21

Ala I Augusta colonorum

Chiaca

Morhaman area

22

Ala Auriana

Dascusa

Pagnik

23

Ala I Ulpia Dacorum

Suissa

?Kelkit area

24

Ala II Gallorum

Aeliana

?

25

Ala constituta

Tablariensis

?

26

Ala I Praetoria recens constituta

?

?

27

Cohors III Ulpia mil.Petraeorum

Metita

Tillo area

28

Cohors IIII Raetorum

Analiba

Hasanova

29

Cohors mil. Bosporiana

Arauraca

?Cosburuk

30

Cohors mil. Germanorum

Sisila

?Erzincan area

31

Ala I Iovia felix

Gizenica

Hortokop

32

Ala I felix Theodosiana

Pithia

Pitsunda

33

Cohors I Theodosiana

Valentia

?

34

Cohors Apuleia c.R.

Ysiporto

?

35

Cohors I Lepidiana

Kaene Parembole Canayer

36

Cohors II Valentiana

Ziganna

Zigana

37

Cohors I Claudia eq.

Sebastopolis

Sukhumi

38

Cohors

Mochora

?

307

Limes XVIII

command of the dux Armeniae, two have late names (I Theodosiana and II Valentiana), while the third is nameless.

Gabriel A. 1940 Voyages archéologiques en Turquie Oriental. (Paris). Gregory S. 1996 Roman military architecture on the eastern frontier (Vol. 2). (Amsterdam). Harper R. 1974 Pagnik Ögreni excavations 1971. In S. Pekman (ed.) Keban Project Publications 1:4. 1971 activities. (Ankara): 107-108. Hodgson N. 1989 The east as part of the wider imperial frontier policy. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. British Archaeological Reports Int. Ser. 553. (Oxford): 177189. Hoffman D. 1970 Das spätrömische Bewegungsheer und die Notitia Dignitatum (= Epigraphische Studien 7. (Bonn). Jones A.H.M. 1964 The later Roman empire, 284-602: a social, economic and administrative survey. (Oxford). Lander J.1984 Roman stone fortifications. Variation and change from the first century A.D. to the fourth. British Archaeological Reports Int. Ser. 206. (Oxford). Lightfoot C.S. 1991 Satala yüzey arastirmasi, 1989. Arastirma SonuÁlar Toplantisi 8: 289-302). Lightfoot C.S. 1998 Survey work at Satala: a Roman legionary fortress in north-east Turkey. In R. Matthews (ed.) Ancient Anatolia (Oxford): 273-284. Mitchell S. 1993 Anatolia: land, men and gods in Asia Minor I. (Oxford). Mitford T.B. 1974a Some inscriptions from the Cappadocian limes. Journal of Roman Studies 64: 160-175. Mitford T.B. 1974b Biliotti’s excavations at Satala. Anatolian Studies 24: 221-244. Mitford T.B. 1977 The Euphrates frontier in Cappadocia. In Studien zu den Militärgrenzen Roms 2: vortrage des 10. Internationalen Limeskongress in der Germania Inferior (= Bonner Jahrbücher, Beihefte 38: Köln): 501510. Mitford T.B. 1980 Cappadocia and Armenia Minor: historical setting of the limes. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.7.2. (Berlin): 11691228. Mitford T.B. 1988 Further inscriptions from the Cappadocian limes. Zeitschrift für Papyrolgie und Epigraphik 71: 167-178. Mitford T.B. 1989 High and low level routes across the Taurus. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. British Archaeological Reports Int. Ser. 553. (Oxford): 329-333. Mitford T.B. 1997 The inscriptions of Satala (Armenia Minor). Zeitschrift für Papyrologie und Epigraphik 115: 137167. Mitford T.B. 1998 The Roman frontier on the Upper Euphrates. In R. Matthews (ed.) Ancient Anatolia (Oxford): 254272. Rémy B. 1989 Les carrières sénatoriales dans les provinces Romaines d’Anatolie au Haut-Émpire. (Istanbul-Paris). Russell J.R. 1987 Zoroastrianism in Armenia (= Harvard Iranian Studies 5). (Cambridge, Mass.). Sedaroglu V. 1970 Agin and Kalaycik excavations 1968, preliminary report. In I. Acaroglu (ed.) Keban Project Publications I.1: 1968 summer work. (Ankara): 41-56. Sedaroglu V. 1971 Agin and Kalaycik excavations 1969. In S. Pekman (ed.) Keban Project Publications 1:2 1969 activities. (Ankara): 27-34. Sedaroglu V. 1972 Agin and Kalaycik excavations 1970. In S. Pekman (ed.) Keban Project Publications I:3 1970 activities. (Ankara): 25-54. Sevin V. & Derin Z. 1989 A fortified site to the east of

The extra 7 alae of evidently late date apart, the correspondence between the names, number and type of units listed in the Notitia and those recorded in C2nd Cappadocia, is so close as to suggest the permanent garrison had remained effectively stable from at least the time of Trajan/Hadrian. Moreover, as certain of the new units on the list are named for Valentinian (1) or Theodosius (4), while two were apparently being raised at the time it was compiled/corrected (probably c.395), it seems that no substantial change was made until the joint reign of Valentinian II and Theodosius I (383-392). The enlargement of the garrison of Armenia-Pontus, therefore, as revealed by the Notitia Dignitatum, is likely to have occurred in response to the Peace of Ekeleac of c.387 (Blockley 1987; 1992: 42-44). This treaty saw the return of western Armenia Major to Roman control, and the transfer of additional units from the west as well as some of those newly raised in the east after Adrianople, to secure the region (Hoffman 1970: 516-519; cf. Jones 1964: 55-60, 207-208). The need for such a large number of cavalry suggests an intensification of mounted patrols in the newly acquired portion of Armenia Major. Bibliography Bennett J. 2001 Trajan, optimus princeps. (London). Blockley R. 1987 The division of Armenia between the Romans and the Persians at the end of the fourth century A.D. Historia 36: 222-234. Blockley R. 1992 East Roman foreign policy: formation and conduct from Diocletian to Anastasius. (Leeds). Bosch E. 1967 Quellen zur Geschichte der Stadt Ankara im Altertum. (Ankara). Braund D. 1989 Coping with the Caucasus: Roman responses to local conditions in Colchis. In D.H. French & C.S. Lighfoot (edd.) The eastern frontier of the Roman empire. British Archaeological Reports Int. Ser. 553. (Oxford): 31-43. Bryer A.A.M. & Whitfield D. 1985 The Byzantine monuments and topography of the Pontos (= Dumbarton Oaks Studies 20. (Washington)). Crow J.G. 1986 A review of the physical remains of the frontiers of Cappadocia. In P. Freeman & D. Kennedy (edd.) The defence of the Roman and Byzantine east. British Archaeological Reports Int. Ser. 297. (Oxford): 76-91. Dabrowa E. 1998 The governors of Roman Syria from Augustus to Septimius Severus. (Bonn). French D.H. 1983 New research on the Euphrates frontier: supplementary notes 1 and 2. In S. Mitchell (ed.) Armies and frontiers in Roman and Byzantine Anatolia. British Archaeological Reports Int. Ser. 156. (Oxford): 71-98. French D.H. 1994 Legio III Gallica. In E. Dabrowa (ed.) The Roman and Byzantine army in the east. (Kracow): 2946. French D.H. & Summerly J.R. 1987 Four Latin inscriptions from Satala. Anatolian Studies 37: 17-22.

308

Julian Bennett: The Cappadocian frontier: from the Julio-Claudians to Hadrian Malatya. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire. British Archaeological Reports Int. Ser. 553. (Oxford): 437460. Sherk R.K. 1980 Roman Galatia: the governors from 25 BC to AD 114. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.7.2. (Berlin): 9541052. Sinclair T.A. 1989 Eastern Turkey: an architectural and archaeological survey III. (London). Speidel M. 1983 The Roman army in Asia Minor: recent epigraphical discoveries and research. In S. Mitchell (ed.) Armies and frontiers in Roman and Byzantine Anatolia. British Archaeological Reports Int. Ser. 156. (Oxford): 7-34. Speidel M. 1986 The Caucasus frontier; second century garrisons at Apsarus, Petra and Phasis. In C.Unz (ed.) Studien zu den Militärgrenzen Roms III. (Stuttgart): 657-660. Sullivan R.D. 1980 Dynasts in Pontus. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.7.2. (Berlin): 913-930. Wheeler E.L. 1996 The laxity of the Syrian legions. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 229276. Wheeler E.L. 1999 From Pityus to Zeugma: the northern sector of the eastern frontier, 1983-1996. In N. Gudea (ed.) Roman Frontier Studies XVII. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalau): 215-230.

309

Limes XVIII

Fig. 1. The Hellenistic and Early Roman site at Kilise Yazisi (after Serdaroglu 1972).

310

Julian Bennett: The Cappadocian frontier: from the Julio-Claudians to Hadrian

Fig 2. Climatic zones in Euphrates-Tigris Region (after B.Claasz Coockson).

Fig. 3. Roman military sites (known and suspected) on the Cappadocian frontier (after B.Claasz Coockson).

311

Limes XVIII

Fig. 4. The fortress at Melitene (after B.Claasz Coockson & Mitford 1980).

Fig. 5. The fortress at Satala (after B.Claasz Coockson & Lightfoot 1998).

312

Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey): second preliminary report on the inscriptions Ergün Lafli Selef and lies near the village of Bayat6, 5kms south-west of Atabey and 18.5kms north-east of Isparta in south-west Turkey.7 The ancient site is bordered by small hills to the north and is situated at the northern end of the small plain of Kuleönü-Bozanönü.8 Thus the city stood at a junction of many roads: the ancient road climbed by way of Döşeme and Ariassus across the plain of Kestel Gölü to Sagalassus, connecting Seleuceia Sidēra and its neighbour Baris with Pamphylia (Ramsay 1890: 49). According to the ancient sources and many modern authorities this route was taken by Alexander the Great (Jones 1971: 124; Levick 1967: 41). Thus in Hellenistic-Roman times, Seleuceia Sidēra stood approximately between the Eğridir and Burdur Lakes and the most influential Pisidian cities of Sagalassus, Apollonia and Antioch as well as the neighbouring cities of Conana, Baris, Prostanna and Minassus (Jones 1971: 142; Mitchell 1976: 119). Its strategic importance combined with its fertile lands meant that it was long an important regional settlement. Other archaeological sites, at Kuleönü, İslamköy, Atabey and Tepecik as well as numerous höyüks, are evidence of the strength and continued occupation on the plain. According to numismatic evidence of the Roman imperial period, the city stood on a river called Oreandos, which might be todays dry stream which flows south from Davraz Dağı, past ancient Agrai (Atabey) and Seleuceia.9 Traces of a dry river bed can be seen today in the south-eastern district of Atabey.

Introduction A brief report on the inscriptions and fragments found at Seleuceia Sidēra (Fig. 1) in Pisidia (south-western Turkey) appeared in 1999 (Lafli 1999a). In this paper I intend to present a second preliminary view of these inscriptions with an overview of the history, archaeology and numismatic of this ancient Pisidian city in the light of earlier research. The literary and archaeological evidence will be summed up briefly. The historical implications with regard to some of these inscriptions have been discussed elsewhere;1 they therefore will be presented here in brief fashion. Seleuceia Sidēra From the early Hellenistic period about seven sites in Asia Minor and seven elsewhere were named Seleuceia.2 One of them, Seleuceia Sidēra (Fig. 1), probably the most unattractive one, was occupied at least from early Hellenistic until the late Byzantine times and lies in northern Pisidia3, in the lakes region of modern Turkey. It was first located by the German historical geographer G. Hirschfeld in 18744, at Asartepe (or Hisartepe - Fig. 1), an isolated and rocky hill (acropolis) well-suited for the purpose with its chora.5 This hill is generally known as 1

For Hellenistic Seleuceia: Lafli 2002a. I am currently preparing a book as well as four excavation and survey reports about the site. For ancient Seleuceias: various authors, Seleukeia, ́η Σελεύκεια (Seleucēa, Seleucīa). In Kroll & Wilte 1923: 1148-1205; Rostovtzeff 1954: map 4. 3 The geographical definition of Pisidia is a complex problem. It is not a region of Asia Minor which has featured significantly in earlier studies of the classical world. The settlement of the geographical region which extends from the Pamphylian plain to the lakes basins of Burdur, Eğridir und Beyşehir, was dominated in antiquity by cities. The earliest of these appear to have come into existence in the C4th BC. We currently have information about 54 Pisidian cities. The term Pisidia was, however, used in the Roman imperial period to denote administrative districts. Furthermore, the extent to which the Pisidians maintained a recognisable ethnic identity under the Roman empire and in late antiquity is another complex and difficult question. Pisidia became an independent province with a separate governor for the first time in the early C4th. This short essay is not the proper place to provide a full, or even a summary account of the history of Pisidia in antiquity. The former textual and archaeological evidence on Hellenistic, Roman Imperial and Byzantine Pisidia accompanying the following publications and their bibliography can be suggested: Olshausen & Neumann 1972: 868-870; Bracke 1993; Schwertheim 1992; Brandt 1992; Belke & Mersich 1990; von Aulock 1979; Jones 1971, 123-146; Özsait 1980; Umar 1987; Grafen Lanckoroński, Niemann & Petersen 1892. 4 For the discovery and location of Seleuceia Sidēra: Hirschfeld 1879: 312-314. 5 We have very little information about the extension of Seleceia´s chora during the Roman imperial period. One of our aims was to document whole human activity on the environs of the site: many archaeological remains from different periods were recorded (for instance an unknown aquaeduct, 4-5kms south-east of Seleuceia). Since inscriptions were lacking, it was, however, impossible to estimate which remains or settlements were dependents of Seleuceia Sidēra. Some settlements and

Today the identification of Seleuceia Sidēra with Asar Tepe is generally accepted. During the C19th Hirschfeld learnt of the name Selef and located Seleuceia Sidēra there (Hirschfeld 1879: 312). J.R.S. Sterrett, who visited the site in 1888, recorded the name Selef from the Greeks of Isparta (Sterrett 1888: 334). Although no inscriptions have yet been found at Seleuceia Sidēra which bear the name of the city, the name Selef and two inscriptions at Sagalassus (Hirschfeld 1879: 309) and at Seleuceia (below; inscription

2

monuments in the environs were recorded: Kelian: Kaya 1991: 251-259; a rock relief (?): Bean 1960, Plate XII (c); and İslamköy and Atabey: Rott 1908, 9-10. In recent years the more prominent höyük sites have been recorded and sherded by M. Özsait, but published reports refer almost exclusively to prehistoric material, except Göndürle, Çürür, Kalburnu and Findos Höyük in the surroundings of Seleuceia: Özsait 1999. 6 Bayat is a name of a Turkish tribe of Oghuzs, and also a very common place name in Anatolia. In modern Turkey there are at least 32 villages with the same name: Sümer 1980: 461; and Aydın 1984: 74. 7 For the location of Seleuceia Sidera see the following: Mitchell 2000: map 1; Kaya 1999a: 164, Abb. 1; Bracke 1993: 30; Mitchell 1993: map 5; Brandt 1992: map of Pamphylia and Pisidia. Location of the town was determined by Ptolemy: Bean 1960: 54, fig. 2 (a)-(b). According to Hirschfeld (1879: 312) a journey between Isparta and Seleuceia Sidēra takes three hours. For a general plan of the site and its in situ monuments, Lafli 2002a. 8 For the detailed description of the site and its environs topography: Lafli 2002a; Kaya 1999: 163. 9 Hill 1897: cix. He recorded Davraz Dağı as “Tounas Dagh”. For the streams on this plain, Sterrett 1888: 332.

313

Limes XVIII

no. 1), have led most scholars of the C19th and C20th to accept Hirschfeld´s identification10 (contra C. Ritter11). Today natives of Bayat Köy call the area between Asartepe (or Hisartepe) and the northern small hills as Selef Arası,12 which is confirmed by a Turkish local author.

of knowledge was gained about the history and topography of the city (Bingöl 1994). The material in this paper is largely drawn from these excavation campaigns. One of the most important results was the discovery of four fragmentary marble inscriptions as well as a terracotta pipe with an inscription in Greek.17

Since the discovery of the Greek inscription by Hirschfeld in 1874, dedicated to the emperor Claudius (below; Inscription no. 1), Seleuceia Sidēra has attracted scholars. Many ancient historians, Byzantine researchers, classical travellers and epigraphers have come to Seleuceia and reported monuments at the site.13 Prior to the 1960s the site was used as a quarry by the residents of the nearby gardens, villages and towns.14 Surveys by M. Özsait in 1980s on the Kuleönü Plain revealed that there had been several settlements around Seleuceia (Pamuklu, Göndürle, Harmanören, Findos, Karayuğ, Kızıl, Kapalıin, Bozanönü, Kanlı, Aliköy etc.) from the early Bronze age (and perhaps earlier) through to the early Hellenistic period.15 An early Bronze age cemetery near Atabey was excavated by Özsait. Despite the interest in the site that these discoveries fostered, there has never been a comprehensive investigation of the city´s topography and its extant buildings. However, efforts have been undertaken recently to redress this situation. Between 1985 and 1987 the Archaeological Museum at Isparta undertook rescue excavations, beginning in the city´s theatre. A report on the three campaigns was published by D. Kaya.16 A fourth season of excavation at Seleuceia Sidēra, again by the Isparta Museum and a team from Ankara Üniversitesi, directed by O. Bingöl was carried out in 1993. During this campaign the following locations were excavated: the southern Terrace II (a rock-structure and many late Roman buildings - Fig. 3), the southern Terrace I which lies south of it and on which a round building is situated - Fig. 4, the eastern tower on the summit of the hill which is part of an early Byzantine fortification, the eastern terrace (and its domestic housings), the theatre, the necropolis and the cistern (Fig. 5). The main results of this work were reported at a symposium, by which a considerable amount

After the death of Lysimachus in 281, Pisidia fell under the nominal control of the Seleucids. Along the northern boundary of “Pisidia” passed a road which connected Syria with Asia Minor.18 As its name indicates, Seleuceia Pisidiae was founded by either Seleucus I or Antiochus I in the early C3rd BC,19 just as other sites in southern and western Asia Minor (such as Apollonia and Antiocheia Pisidiae),20 to protect this military (and perhaps commercial) road across northern Pisidia. This Hellenistic military colony was most likely filled with the inhabitants consisting primarily of natives from the höyüks surrounding the site. The history of the colony-period of this city remains unclear:21 An honorary inscription fragment found in 1993 (below; Inscription no. 3 - Fig. 15) clearly indicates the presence of a boule in the city as early as in the C1st BC. A mention of a possible boule is very important, since we know of very few Hellenistic boules in Pisidia.22 Apart from that inscription we have found four unstratified Hellenistic pottery sherds, which reflect a valuable inventory of local Hellenistic pottery forms (one crater rim, one skyphos rim-neck, an open bowl with ivy decoration and a fusiform unguentarium).23 Although it is difficult at present to estimate the chronology of these sherds, it seems that they might belong to the early to midHellenistic period. Another notable find is a previously unpublished, possibly early Hellenistic, marble statue fragment which is a foot of an adult male with an extraordinary shoe design. The city plan of this fortified Seleucid colony has a problematic nature.24 Which buildings are of Hellenistic origin and what sort of city plan we should imagine remain unanswered. On the other hand some Hellenistic surface structures are preserved. The most important ones are the fortification walls with cyclopian stone blocks in the south-eastern part of the hill

10 Sterrett 1888: 334; Ramsay 1890: 406; Rott 1908: 9; Ruge 1923: 12041205; Bean 1976, 821; von Aulock 1979: 43; Belke & Mersich 1990: 411. 11 Ritter (1859: 482) tried to locate Seleuceia Sidēra with the modern sea town Eğridir. 12 Survival of local place names from antiquity into the Turkish period is not unusual in Pisidia. The names of 17 ancient settlements survive in the modern toponymy: Mitchell 2000: 147, n.44. 13 Chronologically: Sterrett 1888: 332-334; Rott 1908: 9-10; Ramsay 1890: 406; Robert 1955: 239-245; Bean 1976: 821; von Aulock 1979: 4345; Umar 1987. For the detailed list on the earlier research in Seleuceia, Laflı 2002a. 14 Many re-used architectural and sculptural elements from Seleuceia Sidēra and its chora can be found in Bayat, Kuleönü, İslamköy, Atabey and even in Isparta. Some of them are published; for an inscription from Isparta with Σε]λευκεϊς: Collignon 1879: 345, 26; Kaya 1995: 179-180. I attempted to collect all monuments from Seleuceia, especially those inscriptions at the famous Seljuk Türbe in Atabey, and at a forgotten mosque in the south-eastern district of Atabey. For an example, see fig. 2 (a grave stele at the wall of Türbe in Atabey). 15 Özsait 1990: 381-389 (for a map: p. 385, Harita: 1). 16 Kaya 1999; Brandt (1992: 1116, n. 1014) notes also this rescue excavation campaign.

17 This formerly unpublished pipe was found at the very end of the excavation campaign; I did not have enough time to examine it in detail. It is now stored today in a store hall of Atabey´s town municipality. Other stone inscriptions are kept at the Archaeological Museum of Isparta. 18 Bracke 1993:17, 21, n. 54; Ramsay 1890: 43. 19 Jones 1971: 127; Bracke 1993: 17, 21, n. 54; Cohen 1978: 15; Bean 1976: 821. Some authors like Imhoof-Blumer (1902: 398) believe that it was not Seleucus I Nicator who founded the city; others support the idea Seleucus or his son Antiochus I Soter as a founder: Jones 1971: 127; von Aulock 1979: 43. Probably Seleucus I Nicator can not be the founder, since during the time that Seleucids conquered Pisidia (281BC), he was already dead. 20 For Seleucid colonies in western Asia Minor, Jones 1971: 127-128. 21 For a detailed essay on the history of Seleuceia during Hellenistic period: Lafli 2002a. 22 Bracke 1993: 22. During the late Hellenistic period, at least, Sagalassus, Termessus, Ariassus and possibly also Etenna, possessed a real bouleuterion. 23 For Hellenistic pottery from Seleuceia, Lafli 2002b. 24 As a city type, the Seleucid fortified colonies of the early Hellenistic western Asia Minor were classified by Ramsay as early as the 1890s (1890: 85).

314

Ergün Lafli: Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey) (Fig. 6)25, an extent building of unknown function in the middle part of the hill (perhaps the foundation or podium of a Hellenistic temple with a later Roman structure ?: Fig. 7)26 and a cistern (Fig. 5)27, possibly of Seleucid origin, on the southern slope of Asar Tepe. Two portions of the acropolis fortification walls still stand; one portion is built of immense quadrangular blocks, while the other is of polygonal masonry. During the final excavation campaign sondages were placed near the cistern, from which no significant data for the chronology was recorded. In fact no excavated material provides us details about the site’s history after the Seleucid activity (abandonment?). Strabo (Geog. XII.7.2 (570)) cites a list of Pisidian cities compiled by his predecessor, Artemidorus of Ephesus, around 100BC: Selge, Sagalassus, Pednelissus, Adada, Tymbriada, Cremna, Tityassus, Amblada, Anabura, Isinda, Aarassus, Tarbassus and Termessus. One can accept these locations as defining the Pisidian heartland (Mitchell 1976: 119). In this list Seleuceia is missing. According to numismatic evidence, however, there is some evidence that the first three autonomous city coins were minted in the C1st BC,28 but it is not certain whether or not these coins belonged to Pisidian Seleuceia. If so, they are the only evidence for the existence of the site in the late Hellenistic period.

The boundaries of the Roman imperial province of Galatia varied greatly at different times. When Vespasian instituted the province of Lycia-Pamphylia, he must have detached a great part of Pisidia from Galatia to make the new province; and the southern frontier of Galatia then took the line indicated by Ptolemy, including the valley of Apollonia, but not that of Conana, Seleuceia and Baris.30 For this period there is some epigraphic, numismatic and archaeological evidence to indicate that the relationship between Sagalassus and Seleuceia31 continued through the late Hellenistic and Roman period, as well as that between Antioch and Claudioseleuceia, as shown in an honorific inscription set up by the inhabitants of Claudioseleuceia in Pisidian Antioch (Ramsay 1924: 197, inscription no. 26). As for other parts of the empire, the frontiers of Pisidia were undoubtedly changed in the C3rd and C4th, but in many instances this happened only after the defences had been weakened (Ramsay 1890: 398). The other three inscription fragments found at Seleuceia (below; inscription fragments nos. 4-6 - Figs. 16-18) do not help much about the history of the city during the Roman imperial period. Many buildings on the surface can be assigned to the period between the C1st and the C4th AD. The theatre is the most attractive one.32 It is situated at the north-east foot of the acropolis; most of its stones are gone, from which today only the stage building and a diazoma are visible. According to the architectural ornamentation of the stage building (Fig. 8), it must have been built during the C2nd AD. On the keystone of the diazoma was portrayed one of the most popular goddess of Roman Anatolia, Tyche (Fig. 9). In addition to this, other unknown relief figures on the casettes of the stage building were recovered (Fig. 10). The most important excavation areas of the campaign in 1993, namely southern Terraces I and II include also a great number of Roman imperial architecture. Despite the fact that most of the buildings in this area were in use during the late Roman period, among the walls and other structures, it is difficult to estimate which building belonged to the earlier periods. A certain Roman imperial structure is a building at the southern Terrace II, whose north part is carved on a rock surface (Fig. 3). This structure consists of a flattened rock surface, 11 wooden holes on the surface and a strong fragmented front-part. It was obviously a two-storeyed building of unknown (religious) function.33 The many late Roman finds from this area clearly indicate that it was in use in the late Roman period (a former pagan sanctuary with a later Christian function?). Its appearance seems to have some similarities with the Kybele Sanctuary in Antioch. Furthermore during the 1993 season 12 single-room

The years from the C1st BC to the C4th AD, during which time Seleuceia Sidēra and all Asia Minor was ruled by Roman emperors, stand at the fulcrum of a series of events that transformed small Hellenistic Seleuceia into a central Roman imperial city, and a growing late Roman city. The early Roman imperial history of Seleuceia remains, however, unclear. The emperor Claudius I, attested in the inscription found by Hirschfeld (below, Inscription no. 1), was undoubtedly an important figure in the history of the city. He left a great impression elsewhere in Asia Minor. For instance, we have evidence of road-building along the coast in Bithynia (near Amastris) and in Pamphylia (near Adalia). Cities, such as Bithynium and Neoclaudiopolis in Paphlagonia, possibly in return for benefits received, adopted the name of the emperor.29 Seleuceia Sidēra was one of these cities. At the time of the organisation of the province of Lycia-Pamphylia by Claudius, the city´s name was officially changed to Claudioseleuceia; funds for its construction may also have been supplied by the emperor (Mitchell 1993: 78). Unclear is the motivation behind the donation. The funds provided would not in themselves, have been sufficient to guarantee the complete romanization of this area. According to epigraphic evidence, Claudius was worshipped as a benefactor and as a “god made manifest” in the Hirschfeld inscription. The name Claudioseleuceia was retained on coins down to Claudius II, although in Ptolemy’s Geography it is dropped (von Aulock 1979: 43).

30

Ramsay 1890: 252. Inscriptions of the late C1st and early C2nd AD identify Pisidia as one of the constituent parts of the composite province of Galatia-Cappadocia, distinguished on the one hand from Lycia and from Pamphylia, on the other from Phrygia, Mitchell 2000: 140. 31 For an epigraphic connection, Hirschfeld 1879: 303, 309. 32 For the theatre of Seleuceia: Kaya 1999: 166-173; its pictures Tafel 31, 2-4, 32-33; 1999a: 38. 33 Bingöl 1994: 49-50; 65, pl. 2.

25

For its pictures see: Umar 1987: fig. 7; Lafli 1999a: fig. 1. Bingöl 1994: 55: “podyumlu yapı” (?). 27 Bingöl 1994: 53-55; its pictures: 70, fot. 10-11. 28 Von Aulock 1979: 43-44, 154-155, Nos. 1871-1873. 29 Activities of Claudius I in Asia Minor were studied by Magie 1950: 71 and Levick 1967: 4. 26

315

Limes XVIII

buildings on the southern Terrace II were excavated. Due to the finds from these buildings one can assume that they have been used as shops or houses.34 The first building activity on this platform is unknown, but according to the excavations in this area these buildings were at least in use during the late Roman period. On the southern Terrace I which lies south of it, a round building was excavated (Fig. 4).35 It should be a late Roman religious building, since many religious objects were recovered.36 The extensive north-western cemetery37 of the site was located mainly on the north-eastern slopes of Asar Tepe. There are numerous types of graves, but most of them have a simple and poor character in ornaments and design. According to the pottery and numismatic evidence this graveyard was in use between the C2nd and C4th AD.38 The cistern (Fig. 5), which was mentioned among the Hellenistic structures above, was also in use during the Roman imperial period. We already have as clear a picture as we could reasonably hope for the urban settlement pattern of Seleuceia in the Hellenistic and Roman imperial periods.

The name Sidēra was probably added to Seleuceia in the C6th AD (Belke & Mersich 1990: 378). The earliest ancient source, who recorded the name Seleuceia Sidēra is the Synekdemos of Hierocles (673. 8.) of the C6th. During the councils in AD692 and 787 the city´s name was recorded as Seleuceia; on the other hand at Notitiae Episcopatuum VI, VII and IX as Seleuceia Sidēra (Belke & Mersich 1990: 378). The reason for this addition remains unclear. Some scholars speculate that it was in order to differentiate it from the other Seleuceias in the late antique world. Others have suggested that Sidēra (iron) indicated the iron mines around the city which are still visible on the surface (Magie 1950: 88). However no archaeometrical research has yet been done. There is archaeological and epigraphical attestation of Christianity at the site beginning in C4th AD. According to the literary evidence, Seleuceia could boast at least two substantial C4th and C5th churches (including two possible shrines of its principal saints, Saint George and Holy Artemon – who converted his religion with the assistance of St. Paul during his visit and who was the first bishop of Seleuceia - of Seleuceia Sidēra).40 The neighbouring cities in Pisidia were attacked by the Ostrogoths in the C4th. At present we do not know if this also happened to the fortified Byzantine city of Seleuceia.41 Preliminary reports from the Sagalassus excavations indicate that the classical city, whose population is presumed to have suffered from the great plague of 542, barely recovered from heavy earthquake damage in the mid-C6th. Whether both these events occured also at Seleuceia has not yet been determined through the excavations. The division of the late Roman secular provinces is only partly reflected in the organisation of ecclesiastical provinces during the Byzantine period. According to the council lists, Seleuceia sent Eutyhcius to the Council of Nicaea (in 325)42 and Alexandros to the Council of Chalchedon (in 451). Its name is documented in 458 as epistola ad Leonem (Belke & Mersich 1990: 378). In the inscription no. 2 (see below) several early Byzantine toponyms in and around Seleuceia are attested, which cannot be identified with modern toponyms. After the C7th Seleuceia is mentioned very rarely in literary texts. But for the Notitiae Episcopatuum and the records of the major Church Councils, Pisidian settlements are very rarely mentioned in written sources of the C7th to C13th (Mitchell 2000: 142). This matches with our general knowledge of Byzantine Pisidia. Existing maps of settlements of this region, or indeed of any part of

A total of 76 of Seleuceia coins were catalogued by H. von Aulock. Aside from the three coins mentioned above, all 73 bear the heads of emperors, ranging from Hadrian to Claudius II (Gothicus) (von Aulock 1994: 44). Like Sagalassus, with which it shares a number of types, Seleuceia Sidēra seems to have been especially prosperous under Claudius II. Taken with the archaeological evidence, the coinage suggests that Seleuceia Sidēra must have been assigned at various times to Men, Tyche, Hygeia, Artemis the huntress, the emperor Hadrian, Hephaistus, Hermes, and Zeus, whose representations on the city´s imperial coins were the most common. God Men were obviously very popular in Seleuceia and in Pisidia. Within the chora of Seleuceia (especially in Bayat Köy and Kuleönü) we have found many re-used grave stones with Men representations (Fig. 11). Furthermore Men was also represented even on terracotta oinophoroi of Sagalassan origin.39 One of the most attractive coins are those with the representation of agonistic games, festivals and celebrations (von Aulock 1979: 45), which provide evidence for a unique Pisidian brand of paganism during the Roman period. The cycle of pagan communal festivals and religious ceremonies in Seleceia Sidēra waned with the advent of Christianity and the feasts of the church; it was a lengthy and uneven process in this part of the empire. Agonistic scenes are also common among the other Pisidian cities´ imperial coinage, especially in the C2nd and C3rd AD. A temple representation on a coin indicates the existence of an imperial temple at the site (the extent Hellenistic building in the middle of the acropolis?: von Aulock 1979: 45, and 165, no. 2016).

40

Ramsay 1890: 406; and Act. Sanct., 24 March 474. The main church attached to this settlement, however, appears not to have been in the city itself, but perhaps at a distance (at today´s Bayat?), since the Byzantine architectural elements are concentrated mostly in this area and we do not have any traces of any church at the site itself: a fragment of early Byzantine architectural decoration has been observed at the wall of the mosque of Bayat (for an element, see Fig. 12). 41 Ostrogoths at Sagalassos, Waelkens 2000: 177 and in late antiquity in Pisidia, Mitchell 2000. 42 At the time of the Council of Nicaea in 325 the Pisidian cities were the metropolis Antioch, Neapolis, Amblada, Metropolis, Apamea, Baris, Seleucia Sidēra, Pappa, Hadrianapolis, Iconium, Misthia and Vasada: Mitchell 2000: 140.

34

Bingöl 1994: 50-51; 66, pls. 3-4. Bingöl 1994: 51-52; 67, pls. 5-6. 36 Among other ‘religious‘ objects (oinophoroi etc.) terracotta figurines: Lafli 1998; 1999a; 1999b; 2001. 37 Bingöl 1994: 56-57; plan: 64, plan: 5 & 73 pls. 17, 74, pls. 18-19. 38 For the chronological evidence by the terracotta oil lamps: Lafli 2002c. 39 Lafli 1999b. For other religious scenes on the Sagalassus relief ware found at Seleuceia, Lafli 2001 - forthc. 35

316

Ergün Lafli: Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey) south-western Asia Minor, during the middle or later Byzantine periods after the C7th, leave us with a picture of abandoned cities (Mitchell 2000: 139). From the C8th, the name of the city was combined with Agrai (modern Atabey) in the second Notitiae Episcopatuum (10, 490 and 13, 340).43 Archaeological evidence, such as pottery, attests its existence to the end of the C11th, when it was finally abandoned entirely for unknown reasons, whereas there is the simple difficulty of any literary record, that belongs to this period.44 The city inhabitants moved to the neighbouring site of Atabey-Agrai, a settlement which became a bishopric centre in the C9th and took the ecclesiastical status of Seleuceia Sidēra in the second half of the C11th, and assumed archbishop status itself (Belke & Mersich 1990: 172). This literary (and perhaps also archaeological) evidence implies the survival of the urban settlement of Seleuceia Sidēra of late antiquity up until the time of the Turkish conquest. It spectacularly contradicts, however, the virtual silence of other literary evidence for this period and the almost complete absence of archaeological remains in Seleuceia (except pottery). During the Byzantine period there is, however, some evidence in the region for large villages in the vicinity of Seleuceia.45

shows that during the late antiquity the slopes of the town were densely populated, and there was a continuous occupation of the housing insulae at least through the C4th and C5th. At the south-western part of the hill there are also some late Roman walls of unknown function. Late Byzantine remains in Seleuceia Sidēra were recovered in the final excavation campaign at the site; they were some green glazed pottery of local origin. Another notable find is a Seljuk coin hoard in a perished purse, which was found independently in the diazoma of the theatre, just a few centimetres below the surface (Hirschfeld 1879: 313). Inscriptions 1. Greek inscription, discovered by Hirschfeld48, later revised by Sterrett49, and later still by myself50 and by D. Kaya.51 Written on white limestone block in two different letter sizes and letter types in four (b) and six (a) lines. This inscription has been observed by different authors at various locations in the village.52 It was re-used as an architectural element. Today the stone lies in a field in the village centre of Bayat. The information provided by W. Weber (with H. Rott) on the location of the inscription at the beginning of the C20th allows us to recognize that one would not have been able to transcribe fully the entire inscription since it was built into the door of the mosque, thus making certain parts unreadable. It is probable that the inscription was removed from the wall to its modern location during the restoration of the mosque. On the top of the inscription is a hole (a statue base?). At the top of the inscription is a hole which led Magie to suggest that it was part of a statue base, dedicated to the emperor Claudius (41-45). Measurements: height: 0.57m; width: 1.34m; letter size: left text 2.2cms; right text: 3.6cms. Beginning of the C1st AD.

Many of Pisidian cities existed as large centres of population in late antiquity between the C4th and C6th, during which time Seleuceia experienced a building boom. This is demonstrated not so much by written sources, such as the Synekdemos of Hierocles, as by archaeological evidence. There is no sign of a break in the occupation of Roman imperial Seleuceia down to the end of the late antiquity. The eastern tower46 on the summit of the hill which is a part of early Byzantine fortifications, is a good indication of the growing fear of the city inhabitants of the imminent invasions. Within the early Byzantine fortifications a great number of architectural and sculptural elements of the earlier periods were re-used. On the eastern terrace of the occupied hill, several domestic housings from the late Roman period were discovered.47 The northern wall of one of these buildings contained a Zeus altar of Roman imperial period which served as a reused element (Fig. 13a-c). During the late Roman period most of the above mentioned buildings on the Southern terraces were in use for various purposes, where the occupation areas are strewn with Sagalassus pottery of the later Roman periods. That all the living and work areas are found on the south and east slopes of the occupied hill

(a) ΝΟΝ Line 2: ΦΑΝΗ ΑΤΡΙΣ Line 4: ΑΥΤΟΥ ΠΙΟ∆ΩΡΟΣ Line 6: ΡΟΚΑ Γερµανι]κòν ]φαν± ]ατρις

43

48

Ό Σελευκείας τη̃ς Σιδηρα̃ς ̀ήτοι Α ̀ γρων: Darrouzès 1981: 7, 456; 9, 339; 10, 396; and 13, 403. 44 Until the C11th the name of the city was mentioned in the lists of archbishops: Belke/Mersich 1990, 378. 45 In a forthcoming article I discuss briefly the whole regional occupation pattern of the Kuleönü Plain in a wider spectrum (with settlements and monuments in the environs, Atabey, İslamköy, Kuleönü etc.). The results of our surveys in the environs of Seleuceia imply that it would be premature to abandon the search for significant traces of the middle Byzantine period in Kuleönü Plain. In the monograph in preparation I discuss this issue in more detail. 46 Bingöl 1994: 52-53; 68, pls. 7 & 69, pl. 8. 47 Bingöl 1994: 53; 69, pl. 9.

(b) ΤΙΒΕΡΙΟΝΚΛΑΥ∆ΙΟ ΚΑΙΣΑΡΑΣΕΒΑΣΤΟ ΓΕΡΜΑΝΙΚΟΝ ΘΕΟΝΕΠΙΦΑΝΗ

Τιβέριον Κλαύδιο[ν] Καίσαρα Σεβαστò[ν]

Hirschfeld 1879: 313. Sterrett 1888: 334, inscription no. 466; Cagnat 1906: 143, inscription no. 328. 50 Lafli 1999a: 60, inscription no. 1. 51 Kaya 1999: 164-165; its picture: Tafel 31,1; Kaya 1999a: 37. 52 Hirschfeld 1879: 313 - “Andere Reste sind in den Hütten des etwa 20 Minuten entfernten Dörfchens Bjad verbaut: ausser Grabsteinen spätrömischer Zeit und einer byzantinischer Inschrift besonders folgende griechische...”; Sterrett 1888: 334 - “In the wall of the Djami”; Rott 1908: 13 - “An der Westwand desselben Hauses (as here No. 2) als Türwange benutzt, teilweise im Boden” (house, mosque?); and von Aulock 1979: 43 - “...eine in der Moschee von Bayat eingemauerte Ehreninschrift für den Kaiser Claudius I....”. 49

317

Limes XVIII

]αÕτο³ ]πιόδωρος ]ροκλ[έους]

Γερµανικòν Θεòν ’Επιφαν±

]καà στεφαν´σαι αÕτ[Äν ]αγ[αθÃ]ας ñνεκεν καÃ ×ν[δρείας? έν ¿ çν προ]αιρ±ται τόπωι Õπάρχε[ιν Line 12 ]ñφοδον Øπι τÂν βουλÂν κ[αὶ τÄν β±µον καλε²σθαι] δÁ αÕτον και εÓς προε[δρίαν ]εως καà στεφα[ν ]....ΘΑ̣∆[̣ Line 10: ×νδρ-, καλοκ-, φιλ-αγαθίας.

In part (a) Line 2: ‘επι]φανη̃ Line 3: [φιλόπ]ατρι Line 5: [Σαρα]πιόδωρος, ['Ασκλη]πιόδωρος, [Ολυµ]πιόδωρος… Line 6: [`Ανδ]ροκλ[έους, [`Ιατροκλ]έυος, [Κυδ]ροκλέους ... The 5th and 6th lines can be completed as “[... ]rokles son of [....]piodoros”.

4. (Fig. 16) Fragment of a Greek inscription, found in 1993 in the southern Terrace I. Inventory number: SLV 123 Y4/93. Measurements: preserved height: 10.5cms; width: 9.5cms; letter size: 3.2cms.

2. (Fig. 14) Greek inscription, discovered by Hirschfeld (1879: 314), later revised by Sterrett53 and Weber (with Rott).54 Grégoire and Robert discussed this inscription which concerns some small localities in early Byzantine south-west Anatolia.55 The first line of the inscription is no longer visible. Measurements: height: 0.62m; width: 0.62m; letter size: 2.3cms. Probably C5th-C6th.

] .Μ .[ ]Ν̣ΟΥ[ ]Ρ̣Ο̣ . [ 5. (Fig. 17) Fragment of a Greek inscription, found in 1993 on the southern Terrace I. Inventory number: SLV 122Y3/93. Measurements: preserved height: 10cms; width: 10cms; letter size: 3.5cms.

[ƒΕτο]υς υν’ ñρ[γον χ]ριστιανόν καà ð[γιον?] το³ ×γίου Γεωργίου πρÝτ[η] συνοδία Øρ (γεπιστατήσαντος?) ’Αβραµίου καὶ Line 4: ÅπÄ προ×γοντα ’Αντίπατρο[ν] ΠορφÅρις ’Αντιφ´ν „Αρις Πα³λος ’ΗπÀκλις ΜενεµÀ[χ]ις Πα³λος ΚυριακÄς ΣτρατÄνικος ΑÕξάνων Line 8: Ζωτικός Τιµόθις „Ατταλος ’Ιωάννης Κοτόνης Γ{εσ}εννέσ[ιος?] Πα³λος Τεχνίτης Φίλιππος Κόνων Χρυσανθία ’Αλέξανδ[ρος] Line 12: Φίλιππος

]Ε . [ ]ΝΝΕ[ ]. [ 6. (Fig. 18) Fragment of a Greek inscription, found in 1993 on the southern Terrace I. Inventory number: SLV 009Y2/93. Measurements: preserved height: 14cms; width: 11cms; letter size: 3.1cms. ]ΛΙΟ . [ ]Σ vacat ]Ν[

3. (Fig. 15) Fragment of a Greek inscription, found in 1993 in the eastern tower (but not in situ). This marble inscription may also have been reused as an architectural element in the later buildings of the eastern terraces. Severely damaged and cut off from the middle part of the inscription. Inventory number: SLV 088-Y1/93. Measurements: height: preserved height 0.47m; width: 0.36m; letter size: 2cms. It might be a honorific inscription. C2nd-C1st BC (?).

It should be also noted that a terracotta pipe with an early Byzantine one-line-inscription (C5th-6th) was also found; unfortunately the single inscribed line has not yet been transcribed. Beside these inscriptions some later graffiti were documented on the later Roman pottery.

]ΡΗΣΤΙΑ[ διÀ τÂν εύ]νοιαν και διÀ τÂν [φιλίαν (?) ]εσθαι τÂν Øν ΤΟ[ Line 4: ×]ξίαις τιµα²ς ôπω[ς ç]ριστον ôντα κα[à ]αÕτÄν ζηλωτα̣[ ]Ο̣[.]ΥΣ εÓς πάντα τÀ[ Line 8: ]ωι Øπην±σθαί τε Παπ[

Bibliography von Aulock H. 1979 Münzen und Städte Pisidiens, Teil II. Deutsches Archäologisches Institut, Abteilung Istanbul, Istanbuler Mitteilungen, Beiheft 22. (Tübingen). Aydın M. 1984 Bayat, Bayat Boyu ve Oğuzlarin Tarihi. (İstanbul). Bean G.E. 1960 Notes and inscrptions from Pisidia, Part 1. Anatolian Studies 10: 43-82. Bean G.E. 1976 Seleuceia Sidera later Claudioseleuceia. In R. Stillwell (ed.) The Princeton encyclopaedia of classical sites. (Princeton): 821. Belke K. & Mersich N. 1990 Phrygien und Pisidien. In H. Hunger (ed.) Tabula Imperii Byzantini. Band 7. Österreichische Akademie der Wissenschaften,

53 Sterrett 1888, inscription no. 464. Sterrett published a very similar inscription from Isparta town centre (1888: 118, inscription no. 12). This inscription in Isparta is today in the garden of a mosque. 54 Rott 1908: inscription no. 12. 55 Robert 1955: 239-245. Theories about the content of this inscription will be discussed in my monograph on Seleuceia.

318

Ergün Lafli: Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey) Philosophisch-Historische Klasse, Denkschriften. Band 211. (Wien). Bingöl O. 1994 Seleukeia Sidera (Bayat) 1993 Yılı Arkeolojik Kazıları. In Göller Bölgesi ArkeolojikKültürel-Turistik Araştırma ve Değerlendirme Projesi. 1993 Yılı Çalışmaları (Ankara): 43-75. Bracke H. 1993 Pisidia in Hellenistic times (334-25 BC). In M. Waelkens (ed.) Sagalassos I. First general report on the survey (1986-1989) and excavations (19901991). (Leuven): 15-36. Brandt H. 1992 Gesellschaft und Wirtschaft Pamphyliens und Pisidiens im Altertum. Forschungsstelle Asia Minor im Seminar für Alte Geschichte der westfälischen Wilhelms-Universität Münster, Asia Minor Studien, Band 7 (Bonn.). Cagnat R.L. 1906 Inscriptiones Graecae ad res Romanas pertinentes. Avctoritate et impensis academiae inscriptionvm et litterarvm hvmaniorvm, collectae et editae, tomus tertivs. (Paris). Cohen G.M. 1978 The Seleucid colonies: Studies in founding, administration and organisation. Historia Einzelschriften 30. (Wiesbaden). Collignon M. 1879 Inscriptions de Pisidie et de Pamphylie, École française d´Athènes, Bulletin de correspondance hellénique/∆ελτιον Ελληνικης αλληλογραφιας, Troisiéme année, 1879 (Αθηνηει/Paris). Darrouzès J. 1981 Notitiae Episcopatuum Ecclesiae Constantinopolitanae. In Géographie ecclésiastique de l´empire byzantin I. (Paris). Grafen Lanckoroński K., Niemann G. & Petersen E. 1892 Städte Pamphyliens und Pisidiens II. Band. Pisidien. (Prag/Wien/Leipzig). Hill G.F. 1897 Catalogue of the Greek coins of Lycia, Pamphylia and Pisidia. (London). Hirschfeld G. 1879 Vorläufiger Bericht über eine Reise im südwestlichen Kleinasien. Monatsbericht der preussische Akademie der Wissenschaften zu Berlin 20. Märt. Gesammtsitzung der Akademie: 299-333. Imhoof-Blumer F. 1902 Kleinasiatische Münzen. Band II. Sonderschriften des österreischen archäologischen Institutes in Wien, Band III. [Wien]. Jones A.H.M. 1971 The cities of the eastern Roman provinces. (Oxford). Kaya D. 1991 Kelian’da Bulunan İki Yazit, T.C., Kültür Bakanliği, Anitlar ve Müzeler Genel Müdürlüğü, Türk Arkeoloji Dergisi, Sayi XXIX, 1991/The General Directorate of Monuments and Musuems, Minisytry of Culture, Turkish Review of Archaeology XXIX. (Ankara): 251-259. Kaya D. 1995 Zwei neuegefunde Grabstelen aus Atabey (Isparta). Deutsches Archäologisches Institut, Abteilung Istanbul, Istanbuler Mitteilungen Bd. 45: 179-180. Kaya D. 1999a Die Theaterausgrabung von Seleucia Sidera (Klaudioseleukeia). Forschungsstelle Asia Minor im Seminar für Alte Geschichte der Westfälischen Wilhelms-Universität Münster, Asia Minor Studien, Bd. 34: Studien zum antiken Kleinasien IV. (Bonn): 163174. Kaya D. 1999b Seleuceia Sidera. In 1. Uluslararasi Pisidia Antiocheia Sempozyumu Bildiriler Kitabi, 2-4 Temmuz 1997, Yalvaç/Isparta. (Kocaeli): 35-46. Kroll W. & Wilte K. (edd.) 1923 Paulys Real-Encyclopädie der classischen Altertumwissenschaft. Neue Bearbeitung, begonnen von Georg Wissowa. Zweite Reihe [R-Z] Zweiter Band, Sarmatia-Sila. (Stuttgart): 1148-1205. Lafli E. 1998 Les figurines romaines terre cuite de Seleucia

Sidera en Pisidie (Turquie), Orient-Express. Notes et nouvelles d´archéologie orientale, 1993/3 – décembre: 73-78. Lafli E. 1999a Greek inscriptions and inscription fragments from Seleuceia Sidera in Pisidia (south-western Turkey), Orient-Express. Notes et nouvelles d´archéologie orientale, 1999/2 – juin: 59-62. Lafli E. 1999b Sagalassos Red Slip Ware aus Seleukeia Sidera. Ein Beispiel der Verbreitung der sagalassischen Keramik im römischen Pisidien, Forum Archaeologiae. Zeitschrift für klassische Archäologie, Ausgabe 10/III/1999; which can be found on the internet at: . Lafli E. 1999c Sagalassos Roman relief wares from Seleuceia Sidera in Pisidia (Turkey). In R.F. Docter & E.M. Moormann (edd.) Proceedings of the XVth International Congress of Classical Archaeology, Amsterdam, July 12-17, 1998. Classical archaeology towards the third millenium: Reflections and perspectives. (Amsterdam): 227-229. Laflı E. 2000 Sagalassos table and common wares from Seleukeia Sidera in Pisidia (south-western Turkey) - A preliminary report. Rei Cretariae Romanae Favtorvm Acta 36: 43-47. Lafli E. 2001 Dionysiac Scenes on Sagalassos Oinophoroi from Seleuceia Sidera in Pisidia (south-western Turkey). In P.S.W. Bell & G.M. Davies (edd.) Games and festivals in classical antiquity. An interdisciplinary conference held at the University of Edinburgh, Scotland on 10-12 July 2000. British Archaeological Reports (Oxford: forthc.). Lafli E. 2002a Seleuceia in Pisidia in the Hellenistic period. Epigraphica Anatolica 2002 – forthc. Lafli E. 2002b Hellenistic pottery from Seleuceia Sidera in Pisidia (south-western Turkey). Anatolica 2002 – forthc. Lafli E. 2002c Vorläufiger Bericht über die römischkaiserzeitlichen und spätantiken Tonöllampen aus Seleukeia Sidēra in Pisidien (Südwesttürkei). In I. Thuesen (ed.) Near Eastern archaeology in the beginning of the third millenium AD. Proceedings of the second international congress on the archaeology of the ancient Near East. (Copenhagen, May 22nd-26th 2000) (Copenhagen: forthc.). Levick B. 1967 Roman colonies in southern Asia Minor. (Oxford). Magie D. 1950 Roman rule in Asia Minor to the end of the third century after Christ. Vol. I: Text. (Princeton). Mitchell S. 1976 Requistioned transport in the Roman empire. A new inscription from Pisidia. Journal of Roman Studies 66: 106-131. Mitchell S. 1993 Anatolia: Land, men and gods in Asia Minor I: The Celts in Anatolia and the impact of Roman rule. (Oxford). Mitchell S. 2000 The settlement of Pisidia in late antiquity and the Byzantine period: Methodological problems. In K. Belke, F. Hild, J. Koder & P. Soustal (edd.) Byzanz als Raum. Zu Methoden und Inhalten der historischen Geographie des östlichen Mittelmeerraumes. In H. Hunger & J. Koder (edd.) Veröffentlichungen der Kommission für die Tabula Imperii Byzantini. Band 7. Österreichische Akademie der Wissenschaften, Philosophisch-historische Klasse, Denkschriften. Band 283. (Wien): 139-152. Olshausen E. & Neumann G. 1972 Pisidien. In K. Ziegler & W.

319

Limes XVIII

Sontheimer (edd.) Der Kleine Pauly. Lexikon der Antike auf der Grundlage von Pauly´s Realencyclopädie der classichen Altertumwissenschaft. Vierter Band, Nasidius bis Scaurus. (München): 868870. Özsait M. 1980 İlk Çağ Tarihinde Pisidya I, İstanbul Üniversitesi, Edebiyat Fakültesi Yayınları. (İstanbul). Özsait M. 1990 ve 1988 Yılı Senirkent Çevresi Tarihöncei Araştırmaları. In T. C. Kültür Bakanlığı, Anıtlar ve Müzeler Genel Müdürlüğü, Araştırma Sonuçları Toplantısı, Antalya-18-23 Mayıs 1989. (Ankara) 381389. Özsait M . 1999 Yılı Isparta ve Çevresi Yüzey Araştırmaları. In T. C. Kültür Bakanlığı, Anıtlar ve Müzeler Genel Müdürlüğü, Araştırma Sonuçları Toplantısı, II. Cilt, 25-29 Mayıs 1998, Tarsus, T. C. Kültür Bakanlığı Yayınları, Yayın No: 2200, Anıtlar ve Müzeler Genel Müdürlüğü Yayınları, Yayın No: 69 (Ankara): 77-88. Ramsay W.M. 1890 The historical geography of Asia Minor. Royal Geographical Society Supplementary Papers, Vol. IV. (London ). Ramsay W.M. 1924 Studies in the Roman province of Galatia. VI. Some inscriptions of Colonia Caesarea Antiochea. Journal of Roman Studies 14: 197. Ritter C. 1859 Die Erdkunde. IX/2: Kleinasien. (Berlin). Robert L. 1955 Hellenica. Recueil d´épigraphie de numismatique et d´antiquités grecques Vol. X. (Paris): 239-245. Rostovtzeff M. 1954 Syria and the East. In S. A. Cook, F. E. Adcock & M.P. Charlesworth (edd.) The Cambridge Ancient History, Vol. VII. The Hellenistic monarchies and the rise of Rome. (Cambridge). Rott H. 1908 Kleinasiatische Denkmäler aus Pisidien, Pamphylien, Kappadokien und Lykien. In J. Ficker (ed.) Studien über christliche Denkmäler. Neue Folge der archäologischen Studien zum christlichen Altertum und Mittelalter 5./6. Heft. (Leipzig). Ruge W. 1923 Seleuekeia auf der Nordgrenze von Pisiden. In W. Kroll & K. Witte (edd.) Paulys RealEncyclopädie der classischen Altertumwissenschaft. Neue Bearbeitung, begonnen von Georg Wissowa. Zweite Reihe [R-Z] Zweiter Band, Sarmatia-Sila. (Stuttgart): 1204-1205. Schwertheim E. (ed.) 1992 Forschungen in Pisidien. Forschungsstelle Asia Minor im Seminar für Alte Geschichte der westfälischen Wilhelms-Universität Münster, Asia Minor Studien, Band 6 (Bonn). Sterrett J.R.S 1888 The Wolfe Expedition to Asia Minor. Papers of the American School of Classical Studies at Athens Vol. III 1884-1885. (Boston). Sümer F. 1980 Oğuzlar. (İstanbul). Umar B. 1987 Pisidia (Turkish/English). Ak Yayınları, Kültür ve Sanat Kitapları: 48, Yeni Dizi: 2 (İstanbul). Waelkens M. 2000 Sagalassos. In K. Belke, F. Hild, J. Koder & P. Soustal (edd.) Byzanz als Raum. Zu Methoden und Inhalten der historischen Geographie des östlichen Mittelmeerraumes. In H. Hunger & J. Koder (edd.)Veröffentlichungen der Kommission für die Tabula Imperii Byzantini, Bd. 7. Österreichische Akademie der Wissenschaften, Philosophischhistorische Klasse, Denkschriften. Bd. 283. (Wien ).

320

Ergün Lafli: Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey)

Fig. 1. Seleuceia Sidera and Asartepe (Hisartepe), from south-east.

Fig. 2. A grave stele on the wall of the Seljuk Türbe at Atabey.

321

Limes XVIII

Fig. 3. Southern Terrace II and religious rock structure.

Fig. 4. Round building at the southern Terrace I.

Fig. 5. The cistern.

322

Ergün Lafli: Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey)

Fig. 6. Hellenistic fortification walls on the south-eastern corner of Asartepe.

Fig 7. A Hellenistic building at the middle part of Asartepe.

Fig. 8. A sima block from the theatre.

323

Limes XVIII

Fig. 9. Tyche relief on the keystone of the diazoma of the theatre.

Fig. 10. A casette with some relief figures from the scene building of the theatre.

Fig. 11. A grave stele with riding Men representation (Archaeological Museum of Isparta).

Fig. 12. A Byzantine architectural element on the wall of the Camii of Bayat.

324

Ergün Lafli: Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey)

Fig. 13a-c. A re-used house altar from the eastern terrace domestic housings (a. front; b. back; c. profile).

Fig. 14. Inscription no. 2.

325

Limes XVIII

Fig. 15. Inscription fragment no. 3.

Fig. 16. Inscription fragment no. 4.

Fig. 17. Inscription fragment no. 5.

Fig. 18. Inscription fragment no. 6.

326

i(n/ a ... au)toi\ a)l / loij Bora/doi kai\ Go/tqoi gi/nwntai. Gregorios Thaumaturgos und die pontischen Beutezüge der Boranen und Goten im 3. Jh. n. Chr. 1 Jeorgios Martin Beyer In the course of the C3rd AD the Roman empire faced in some areas of its eastern borders a series of incidents that caused a temporary loss of control over parts of its provinces. The contemporary sources argue about the causes of the crisis, some citing the armed struggles at the frontiers of the empire, which culminated in the Gothic incursions of the mid-C3rd, others the Decian persecution which in turn was a reaction to the phenomena of general crisis. A primary source which integrates both aspects is the so-called ‘canonical letter’ of the Bishop Gregorius Thaumaturgus of Pontus. This letter was evidently a reaction to the incursions of some Gothic tribes, eg. the Borani, into the eastern Black Sea region in the years 254/255 and sketches the effects of the barbarian raids on the population of the province of Pontus, and of course mainly its christianized elment. The persecutions of Christians are well attested by contemporaries, but are also recorded by later Christian authors, even if tendentiously deformed. The historians focussed their interest on the incursions into the Balkan peninsula and the western parts of Asia Minor and drew their information particularly from sources like Herodian, Dexippos and Zosimos. On the other hand, the invasions of the Gothic tribes into the eastern Black Sea region are less well attested and their chronological order less certain. Therefore the ‘canonical letter’ helps us to place properly the Black Sea raids into the overall picture of the Gothic invasions. Analysis of the form and content of our primary source affords us insight into the conditions of the Christian church and community in Pontus at the furthest edge of the Roman empire. The letter names precisely the offences that happened during and after the turmoil of the raids. The scope ranges from various property offences to betrayal and collaboration with the invaders as well as manslaughter. Some Christians, for instance, acted as guides to the invaders on their expeditions of pillage and destruction. Some even murdered their fellow citizens. Thus the letter informs us about the nature of such crimes. On the other hand it marks the beginning of a process tending towards a specific system of church-legislation. Gregorius saw ‘greed’ as the cause for all offences in the course of the raids, although he did not say so by way of generalizing. Each individual offence is listed in specific canons with corresponding punitive measures. Gregorius takes into account as well whether or not the accused person admitted and regretted the offences. This division into different categories and the fixing of specific stages for penance that leads to the reintegration of the penitant into the christian community is typical of Gregorius’ way of handling the problems. The only thing missing is a specific regulation of the length of these punitive measures. Finally, comparison with later occurrences and the determinations of the Roman laws show certain analogies to Gregorius’ precise instructions. These parallels are evident in the treatment of returned captives. In most of these cases this concerns persons who were unlawfully captured by others after their liberation, so that they did not get back their property or even their social status as free Roman citizens. The records of Roman legal sources show parallels not only with the marauding infringements on liberated captives, which Gregorius describes. This state of affairs reminds us of incidents mentioned by Ammianus Marcellinus in the C4th. It might be noted that in these cases the provincials also guided the barbarians to rich villages and hamlets, storage rooms and human hideouts, with the same results. The inhabitants were captured as slaves or murdered and the entire region devastated. In these cases the problems were regulated by official acts of legislation. As Pontus had been extensively christianized from the mid-C3rd, it is reasonable to assume that at first the church leaders had to react to assure or at least roughly re-establish public order. The church thereby assumed, for a time, the duties of the Roman government and its central administration.

römischen Herrschaft durch die Kriegszüge des Sasanidenherrschers Šapūr I. Sie spiegeln sich nicht nur in den militärischen Reaktionen, sondern auch auf religiösem Gebiet. Untrennbar mit dem allgemeinen Bewußtsein der Bedrohung und des Verfalls verbunden sind auch die decischen Christenverfolgungen, die den christlichen Teil der römischen Bevölkerung zu einer Positionsbestimmung nötigten. Sie hatten dazu geführt, daß die sich in der Entwicklung befindliche Kirche Gedanken darüber machte, wie sie die lapsi, jene Christen, die sich unter dem Druck bereitfanden, den Staatsgöttern zu opfern oder sich zumindest die nötigen Bescheinigungen zu beschaffen, wiederaufnehmen sollte. Die Fragen der Konsequenzen dieser Notsituation für die christlichen Gemeinden sind von den Zeitgenossen wie auch späteren Autoren intensiver, wenn auch tendentiös verformt, dokumentiert worden, als

Krisenzeiten bringen nicht nur Verunsicherung und Niedergang mit sich; bisweilen bieten sie auch Chancen. Sie stören eingefahrene Pfade und weisen eine hohes Potential für Veränderung und Wandel auf. Eine dieser Phasen in der antiken Geschichte ist das 3. Jh. n. Chr. Zu den Ereignissen, die in den Augen der Zeitgenossen die Krisenhaftigkeit der Epoche bestätigten, gehören besonders die Einfälle der Goten im Osten des Reiches und die Bedrohung der -------------------------------------------------1

Dieser im Rahmen des Sonderforschungsbereichs 295 („Kulturelle und sprachliche Kontakte“) der Joahnnes Gutenberg-Universität Mainz entstandene Artikel geht zurück auf eine vor einigen Jahren von meinem verehrten Lehrer Prof. Dr. Peter Herz (Regensburg) angeregte Beschäftigung mit dem Christentum an der Ostgrenze des Imperium Romanum. Danken möchte ich v.a. Aliki und Hans-Joachim Beyer für die intensive Diskussion der griechischen Quellen sowie S.F. Meynersen für die kritische Durchsicht des Manuskripts. Alle im Text zitierten Übersetzungen sind eigene.

327

Limes XVIII die der Raubzüge der Mitte des 3. Jhs. n. Chr.2 Eine der wenigen Quellen, die beide Aspekte vereinen, ist der sogenannte ‘kanonische Brief’ des pontischen Bischofs Gregorios Thaumaturgos. Er stellt eine direkte Reaktion auf die Barbareneinfälle der Jahre 254-256 n. Chr. im östlichen Schwarzmeerraum dar.

Die Perser waren abgezogen, hatten aber reiche Beute (und Menschen) mitgeführt. Die Bedrohung blieb trotz des Abzugs weiterhin bestehen, und es scheint beinahe so, als hätten die Boranen und Goten der nördlichen Schwarzmeerküste auf diese Situation reagiert. Ein erster Beutezug der Boranen im östlichen Schwarzmeerraum im Jahre 254 n. Chr. war nicht sehr erfolgreich. Der römische General Sucessianus verteidigte die Stadt Pityus (Picunda, Georgien) erfolgreich (Zosimus 1.32.1-3). Beinahe wären dabei die Boranen vernichtend geschlagen worden, da sie sich nur an Land setzen ließen und die Bosporaner mit ihren Schiffen abgezogen waren. Sie konnten nur durch die Benutzung römischer Schiffe wieder in ihre Heimat zurückkehren. Nach Zosimos wiegten sich auch die Bewohner von Pontos nach dieser Flucht in der trügerischen Sicherheit, daß die Barbaren nicht noch einmal einen Feldzug wagen würden (Zosimus 1.32.4). Dies erklärt auch, warum eben dieser Successianus aus Pityus abgezogen wurde, um als Helfer beim Wiederaufbau Antiochias nach den Persereinfällen zu dienen.7 Diese Maßnahme entblößte aber die Schwarzmeerregion nun völlig.

In der Forschung ist über die Chronologie dieser Invasionen z.T. kontrovers debattiert worden; die Datierungsproblematik spitzt sich im wesentlichen zu auf eine Ansetzung der Raubzüge in die Jahre 254-256 oder 255-257 n. Chr.3 Obwohl die Einordnung der Ereignisse insgesamt nicht mit allerletzter Sicherheit aus den einzelnen Fragmenten historischer Überlieferung zu bestimmen ist, läßt sich die innere Chronologie doch nach der Schilderung der Ereignisse bei Zosimos folgendermaßen rekonstruieren. Die Beutezüge der Goten zur See nahmen ihren Ausgangspunkt in dem Gebiet des ehemaligen Bosporanischen Reiches, das durch gotisch-sarmatische Stämme und innere Unruhen derart destabilisiert worden war,4 daß Zosimos die Bosporaner als a)na/cioi th=j h(gemoni/aj bezeichnet (Zosimus 1.31.3). Voraussetzung für den Beginn der Beutezüge der Goten zur See war die Schwächung dieses römischen Vasallenstaates durch den Niedergang des Handels mit Rom und die nachlassende Unterstützung durch die Römer. Als fatal erwies sich schließlich der einige Zeit zuvor erfolgte Abzug der römischen Schwarzmeerflotte von Trapezus und Sinope und ihre Verlegung nach Kyzikos, der eine enorme Lücke in die maritime Verteidigung der östlichen Schwarzmeerküste riß (Kienast 1966: 106-9; Mitchell 1993: 235). So fiel bei der Herrschaftsübernahme in diesem Gebiet den gotischen Stämmen und den Boranen5 auch die bosporanische Flotte mitsamt Besatzung in die Hände.6 Für Rom entstand durch den Untergang des Vasallenstaates also im Norden eine instabile Situation, doch auch im Osten wuchs die Gefahr. Armenien gehörte seit 252 zum Sasanidenreich, wodurch Kappadokien zum persischen Angriffsziel wurde. In Syrien war bei Barbalissos am Euphrat ein römisches Heer vernichtend geschlagen (RGDA (gr.) Zeile 10-11), Antiochia ebenso wie viele andere Städte zerstört worden.

Am erneuten Zug im folgenden Jahr (255 n. Chr.) waren neben den Boranen nun auch ein Teil der auf der Krim ansässigen Goten beteiligt.8 Im Besitz der von ihnen requirierten Flotte der Bosporaner führte der neuerliche Raubzug zunächst in die Küstenstadt Phasis; dort wollten die Angreifer das Heiligtum der Rhea Kybele plündern. Als diese Unternehmung scheiterte, wandten sich die Boranen wieder gegen Pityus (Zosimus 1.32.1), das nun eingenommen wird, und diesmal sogar bis nach Pontos. Die Hafenstadt Trapezus wurde belagert und erobert, da die Besatzung die Stadt widerstandslos verließ (Zosimus 1.33.13). Die folgende Plünderung und Brandschatzung betraf auch die Umgebung (Zosimus 1.33.3). Auch in diesem Fall handelt es sich nicht um einen Eroberungszug, sondern lediglich um einen Beutezug; die Goten und Boranen zogen wieder ab. Unterstützt wurde der Erfolg dieser Raubzüge durch die Tatsache, daß das römische Militär durch die Persergefahr an der Ostgrenze gebunden war. Deutlich zeigten sich im folgenden Jahr die Präferenzen der militärischen Reaktion; der erneute Beutezug nach Kleinasien betraf diesmal den Westteil des Schwarzen Meeres; Chalkedon, Nikomedeia, Nikaia, Apameia, Prusa wurden erobert und geplündert.9 Durch die Nachrichten von der erneuten Invasion in Kleinasien wurde der Kaiser Valerian veranlaßt, einen Teil seines Heeres in Richtung Bithynien zu senden. Dieser Zug wurde jedoch schon in Kappadokien abgebrochen; doch der Angriff der Sasaniden auf Dura-Europos zwang

2

Zur Rezeption der Krise in den Quellen und zum ‘Krisenbewußtsein’ allgemein s. z.B. Alföldy 1973; 1975; Kolb 1977: 277ff.; De Blois 1984: 358ff.; Strobel 1993. 3 Frühe Datierung u.a. Alföldi (1967a: 142), zuletzt etwa Bellen (1998: 218f.); die spätere Ansetzung unterstützen u.a. Wolfram (1990: 58f.) und Heather (1996: 40); Salamon (1971: 119f.) sogar um 258 n. Chr. Umfassende Zusammenstellung der Literatur bis 1959 bei Walser & Pekáry (1962: bes. 28-34); zur Chronologie der Gotenzüge in Kleinasien s. a. Alföldi (1967c: 323f.); ältere Literatur bei Salamon (1971: 109-139) und Demougeot (1969: 419ff.). Zur inschriftlichen Überlieferung s. u.a. Mitchell (1993a: 235f.). 4 Gajdukevič (1971: 464ff.); zur osteuropäischen Literatur vgl. Salamon (1971: 118 Anm. 40). Zur wirtschaftlichen Bedeutung des Handels im Bereich des Schwarzen Meeres im 2. und 3. Jh. vgl. u.a. Robert (1980: 64ff.). 5 Gregorios nennt diese in ep.can. 5 Bora/doi, während bei Zosimus I.31ff. Bora/noi erscheint. Dexippos bezeichnet alle Goten als „Skythen“. Zosimos und die Historia Augusta benutzen diesen Terminus nur für die pontischen Goten im Kontrast zu den Donaugoten; zur Namengebung und Herkunft dieser Stämme vgl. Demougeot (1969: 417f.). 6 Zosimus 1.31.3: kate/sthsen ku/rioi ploi/oij au)tou\j oi)kei/oij diabiba/santej; vgl. a. Demougeot (1969: 418).

7

Strobel (1993: 243) läßt Successianus schon 254 n. Chr. am Wiederaufbau Antiochias mitwirken; er wäre dann unmittelbar nach der Rettung von Pityus nach Antiochia beordert worden. 8 Die ausführlichste Schilderung dieser Gotenzüge bei Zosimus 1.31-33. Demougeot (1969: 419); Salamon (1971: 118f.). Vgl. auch Kettenhofen (1982: 90-6). 9 Zosimus 1.34f. vgl. a. Darstellung der Ereignisse bei Wolfram (1990: 61f.)

328

Jeorgios Martin Beyer: i(/na ... au)toi\ a)/lloij Bora/doi kai\ Go/tqoi gi/nwntai. Gregorios Thaumaturgos Valerian zur Umkehr.10 Lediglich für Byzanz beorderte er einen Emissär, der die Stadt vor dem Schicksal Chalkedons bewahren sollte (Zosimus 1.36.1). Ob dieses Manöver tatsächlich einen Abzug der Goten bewirkte, muß fraglich bleiben. Es zeigt jedoch deutlich, daß sich die römische Militärmacht einem ‘Mehrfrontenkrieg’ im Osten nicht wirksam stellen konnte und letztlich seine Prioritäten auf die Abwehr der Persergefahr legte.

vorliegenden kanonischen Brief,17 wahrscheinlich noch vor Einsetzen der valerianischen Christenverfolgung im Jahre 257 n. Chr.18 268 n. Chr. war Gregorios Thaumaturgos laut Eusebios Teilnehmer an der Synode von Antiochia gegen Paulus von Samosata.19 Da er bei der nächsten Synode nicht mehr erwähnt wird, müßte man um dieses Datum herum seinen Tod ansetzen. Wegen seiner großen Erfolge bei der Missionierung der Pontier und dabei besonders der Landbevölkerung,20 eine Tatsache, die Basileios von Kaisareia besonders hervorhebt,21 erhält er posthum den Beinamen Thaumaturgos, der Wundertäter.22 Die Informationen, die uns die erst ein Jahrhundert später verfaßte Vita des Gregor von Nyssa liefern, sind allerdings mit der nötigen Vorsicht zu genießen.23 Ausführlich hat Robin Lane Fox in seinem Buch ‘Pagans and Christians’ Gregorios’ Lebensumstände und die Rahmenbedingungen seines Wirkens dargelegt. Seine Ausführungen zum kanonischen Brief sind jedoch außer einer generellen Bewertung der Person hauptsächlich in Hinblick auf die Organisation der Kirche und die Position und Macht der Bischöfe gelesen worden.24 Wie hier wird auch in anderen Untersuchungen der Brief besonders als illustrierendes Material für die Ereignisse während der Invasionen

*** Während Zosimos besonders die äußeren Ereignisse referiert, schildert der ‘kanonische Brief’ des Bischofs von Neokaisareia, Gregorios Thaumaturgos, sehr konkret die Auswirkungen der Beutezüge in Pontos auf die Zivilbevölkerung. Im Leben des Gregorios Thaumaturgos ist kein einziges Datum ganz gesichert.11 Wahrscheinlich wurde er um 220 n. Chr.12 in Neokaisareia in der Provinz Pontus Polemoniacus geboren13 und stammte aus einer angesehenen heidnischen Familie.14 In Neokaisareia erhielt er eine weitreichende ‘enzyklopädische’ Erziehung, wurde in die Rhetorik und das Römische Recht eingeführt. Aus familiären Gründen ging er nach Caesarea in Palästina, wo er für die nächsten fünf Jahre als Schüler des Origenes weilte.15 Wahrscheinlich 238 n. Chr. verließ er diesen und wurde in noch jugendlichem Alter vom Bischof von Amaseia, Phaidimos, als Bischof nach Neokaisareia in Pontos berufen,16 wo er auch die decische Verfolgungen erlebte. Unmittelbar im Anschluß an die Raubzüge der Jahre 254-255 n. Chr. und der damit verbundenen Irritationen verfaßte der Bischof den uns

17 Wolfram (1990: 59) datiert somit auch die Entstehungszeit des kanonischen Briefes des Gregorios Thaumatourgos nach 257, während Dräseke (1881: 754) ihn fälschlich in das Jahr 254 datiert; ebenso Ste. Croix (1971: 650 Anm. 9). Migne PG 10, 967 datiert die Schrift in das Jahr 258; ebenso Ryssel (1880: 15); Phouskas (1969: 168): Ende 254/Anfang 255. 18 Eine Entstehung in Zeiten einer Verfolgung ist kaum anzunehmen, um so mehr als sich diese konkret gegen die Amtsträger richtete. Wahrscheinlicher ist eine Abfassungszeit nach den Gotenzügen und vor dem Beginn der valerianischen Verfolgung, der ja 258 auch der nordafrikanische Bischof von Karthago Cyprian zum Opfer fiel. Zum in den Acta Proconsularia Sancti Cypriani überlieferten Edikt des Valerian s. a. Schwarte (1989: 109ff.) mit Literatur. Schilderung des Verlaufes vgl. u.a. Alföldi (1967b: 285-311), Sage (1975: 337ff.). 19 Eusebius HE 7.28; zur Datierung der Synode: Millar (1971: 11). 20 Der Ausdruck pontikoi\ kai\ xristianoi/ (Greg.Thaum. ep. can. 7) könnte ein Zeichen für die weitgehend erfolgte Missionierung der pontischen Bevölkerung sein; Frend (1976: 8) dehnt die Missionstätigkeit des Gregorios auch auf Kappadokien aus; er nennt als Rahmendaten für diese Missionierung 243-272 n. Chr. Der Ausdruck könnte aber genausogut auch die gemeinsame Zugehörigkeit von Absender und Empfänger oder eine Spezifizierung der Adressaten betonen. 21 Basilius spir. 29. 70. Die Tatsache, daß Gregorios nicht nur die Stadt, sondern auch das Land bekehrte, spricht dafür, daß er die Sprache des Volkes beherrschte; Fox (1986: 533) vermutet ohne Angabe von Belegen das Gegenteil. Unterstützt wurde diese erfolgreiche Missionierung dadurch, daß Gregorios die heidnischen Festtage und Feiern in christliche umgestaltete, was den Weg für diese Praktik in den folgenden Jahrhunderten ebnete; s. a. Frend (1976: 8). Besonders die wissenschaftliche Rezeption der Arbeit von Lane Fox (1986) führte zu der sicherlich überzogenen Sicht, daß die Provinz Pontos eines der ‚most important centres’ des Christentums im Osten gewesen sei; Ball (2000: 437). 22 Um seine angeblichen Wundertaten rankt sich bald ein regelrechter Kult, Telfer (1936: 225ff.); s. a. Frend (1984: 445). 23 Greg. Nyss. PG 46.953D gibt an, daß es vorher 17 Christen und am Ende der Lebenszeit des Gregorios nur noch 17 Heiden in Pontos gab; auch schildert er in dieser Schrift die Wundertaten sehr ausführlich; vgl. a. Socrates HE 4.27. Die besondere Missionstätigkeit wurde öfter bezweifelt; kritisch zu den Bekehrungserfolgen: MacMullen (1984: 5961, 145f. mit Anm. 1); zuletzt u.a. Markschies (1997: 25), der wegen der topischen Wendungen und des zeitlichen Abstands der Hauptquelle Gregor von Nyssa Bedenken über die Auswertbarkeit äußert; vgl. allg. Abramowski (1976: 145-66). Ein Korrektiv zur Tradition bei Gregor von Nyssa bietet auch Ryssel (1894: 228ff.) und Telfer (1930: 142ff., 354ff.). 24 Lane Fox (1986: 516-43, zum kanonischen Brief 539ff.).

10

vgl. u.a. Halfmann (1986: 237). Gerade Gregors von Nyssa Panegyrikos, die schriftliche Version einer Rede, die dieser anläßlich eines jährlichen Festes zu Ehren des Heiligen Gregor des Wundertäters in einer Kirche in Pontos gehalten hat, zeigt dies deutlich; vgl. Telfer (1936: 229); zur teilweise umstrittenen Zuordnung vgl. Crouzel (1983: 782). Weitere verläßliche Quellen zum Leben des Gregorios sind einerseits seine eigene Dankrede an Origenes, zum anderen verstreute Angaben v.a. bei Basilios von Caesarea, als auch in der Kirchengeschichte des Eusebios. Die wichtigste Literatur zum Leben und Wirken: Ryssel (1880), van Dam (1902), Phouskas (1969), Mordzejewski (1971), Crouzel (1983), Lane Fox (1986: v.a. 517-43) und Slusser (1998: 13-6 zur Vita des Gregor von Nyssa). 12 Phouskas (1969: 120) nimmt in seiner Arbeit über Gregorios Thaumaturgos ein Geburtsdatum von 211/3 n. Chr. an. Die Daten bei Phouskas sind nicht immer stichhaltig belegt; ansonsten bietet diese Monographie eine gute Zusammenstellung des Handschriftenmaterials und wegen der Verwendung auch von Texten orthodoxer Tradition in theologischer Hinsicht einige interessante Gedanken. Für ein Geburtsdatum zwischen 210 und 213 n. Chr.: Crouzel (1983: 780). 13 Frend (1980: 266); ibid. 271 siedelt er Gregorios Thaumaturgos fälschlicherweise in Caesarea in Kappadokien an. 14 Ursprünglich war sein Name Theodoros (Eusebius HE 6.30), doch änderte er ihn, vielleicht anläßlich seiner Taufe, in Gregorios. 15 Eusebius HE 6.30. Nach Gregor von Nyssa (PG 46.901D) lebte er eine Zeit lang in Alexandreia inmitten von Philosophiestudenten; vgl. auch Telfer (1936: 229) und allgemein Frend (1984: 312; 332 Anm. 22) zum Verhältnis von Gregorios und Origenes („devoted pupil“). Zu den familiären Anlässen s. Lane Fox (1986: 519) und Mitchell (1993b: 53). Gleichzeitig interessierte er sich aber weiterhin für das Studium der römischen Rechtswissenschaft. Lane Fox (1986: 525); Greg.Thaum. or.prosph. 1.7. 16 Eusebius HE 6.30; sowohl er, als auch sein Bruder Athenodoros, werden als Bischöfe in Pontos erwähnt; Eusebius HE 7.14. 11

329

Limes XVIII verwandt.25

gentumsdelikte behandelt. Die beiden kano/nej 6 und 7 betreffen zusätzlich Verrat, Kollaboration und Totschlag. Kanon 10 fordert von den Büßern die uneingeschränkte Befolgung der Anordnungen.

Der vorliegende sog. kanonische Brief enthält in der hier zugrundegelegten Migne-Ausgabe 11 Einzelabschnitte, kano/nej genannt,26 wobei jedem Kanon Kommentare von Johannes Zonaras und Theodoros Balsamon aus dem 12. Jahrhundert angegliedert sind. Die Zerlegung des Briefes in einzelne kano/nej, sowie die teilweise erfolgten Kürzungen bzw. Erweiterungen stammen nicht aus der Zeit des Gregorios.27 Die Anerkennung der Rechtsverbindlichkeit durch die Synode von 692 n. Chr.28 zeugt von der Wichtigkeit der Anordnungen dieses Briefes für die Zeitgenossen und vor allem für die organisierte Kirche der folgenden Jahrhunderte.

In der Zeit der Gotenzüge haben, laut Gregorios, viele Menschen gedacht, diese Zeit der Wirren auszunutzen, um für sich selbst Nutzen zu ziehen.32 Für die Beurteilung der angesprochenen Vergehen werden im zweiten Abschnitt die Grundlagen gelegt: Schrecklich ist die Habsucht, und es ist nicht möglich, in einem einzigen Brief die göttlichen Schriften darzulegen, in denen nicht allein das Rauben als entsetzlich und grausam angeprangert wird, sondern allgemein habsüchtig zu sein und sich fremder Güter aus Habsucht zu bemächtigen; und ein jeder solcher sei aus der Kirche auszuschließen.33 Die kano/nej 2-9 beschäftigen sich also mit den Folgen der Habsucht (pleoneci/a), die als Ursache für die einzelnen Taten angesehen wird. Daß aber einige es wagten, in der Zeit des Einfalls in so großem Wehgeschrei und so großem Klagen die Zeit, die allen Verderben bringt, für eine für sie selbst gewinnbringende zu halten, ist die Haltung (ein Wesenszug) von gottlosen und Gott hassenden Menschen, und nicht von solchen, die im Übermaß "Verkehrtheit" besitzen. Daher schien es richtig, solche auszuschließen, damit niemals der Zorn über das ganze Volk komme und zwar vor allem zuerst über die Vorsteher, die ja nicht nach Gewinn streben ... Wenn aber einige wegen der früheren Habgier, die im Frieden stattfand, bestraft wurden, in der Zeit des Zornes selbst, wiederum der Habgier verfallen, indem sie aus dem Blut und Verderben von Vertriebenen oder Gefangenen, von getöteten Menschen Gewinn ziehen - was anderes muß man da erwarten, als daß sie, wenn sie den Zorn auf sich, auch auf das ganze Volk laden.34

Bevor hier auf die Vergehen eingegangen wird, muß noch einiges zur Form des Briefes hinzugefügt werden, was zusätzlich die Bedeutung dieses Briefes und auch die des Bischofs unterstreicht. In der Anrede erscheint der Titel pa/paj, in der Zeit des Gregorios die gewöhnliche Bezeichnung für einen Bischof.29 Es handelt sich also um ein Schreiben an einen Bischof und ist auf Anfrage geschrieben worden. Der Adressat erbittet die Stellungnahme des Gregorios zu konkreten Problemen. Da der Brief keine Eingangs- bzw. Schlußformel besitzt und zusätzlich von einem sugge/rwn begleitet wird,30 könnte es sein, daß es sich hier sogar um eine Art e)pistolh\ kuklikh/ handelt. Der Ausdruck a)postei/lamen würde das unterstützen.31 Schließlich könnte die Verwendung der ersten Person Plural zur Bezeichnung seiner eigenen Person ein Hinweis darauf sein, daß Gregorios, wenn nicht im Auftrag eines Kollegiums, so doch wenigstens in Übereinstimmung mit einem solchen spricht. Inhaltlich lassen sich die ersten 10 kano/nej, die sicher auf Gregorios zurückgehen, wie folgt einteilen: Kanon 1 beschäftigt sich mit den Verfehlungen, die nicht gebüßt werden müssen. Kanon 2 ist allgemeiner gehalten und beginnt mit der Formel deinh\ h(; Kanon 3 schließt sich inhaltlich daran an; die Kanones 4-5 enthalten Diskussionen von Fällen, die nicht ganz eindeutig sind (mhdei\j e)capata/tw e(auto/n); die Abschnitte 6-9 sind mit Überschriften versehen (peri/) und behandeln Einzelfälle. In den kano/nej 2-5 und 8-9 werden dabei Ei-

Es wird nicht nur Diebstahl, sondern alle Äußerungen von pleoneci/a und ai)sxrokerdi/a als verwerflich angesehen, mit der Konsequenz, daß jeder, der aus solchen Motiven handelt, aus der Kirche ausgeschlossen werden 32 Wolfram (1990: 61) nimmt an, daß der Brief auch den Zug von 257 berücksichtigt, obwohl dieser nicht nach Pontos führte, und auch von anderen Völkerschaften unternommen wurde. 33 Greg. Thaum. ep. can. 2: Deinh\ h( pleoneci/a, kai\ ou)k e)/sti di` e)pistolh=j mia=j paraqe/sqai ta\ qei=a gra/mmata, e)n oi(j = ou) to\ lhsteu/ein mo/non frikto\n kai\ frikw=dej katage/lletai, a)lla\ kaqo/lou to\ pleonektei=n kai\ a)llotri/wn e)fa/ptesqai e)pi\ ai)sxrokerdei/#! kai\ pa=j toiou=toj e)kkh/rhktoj E ) kklhsi/aj Qeou=. 34 Greg. Thaum. ep. can. 2: To\ de\ e)n kair%= th=j katadromh=j, e)n tosau/t$ oi)mog$= kai\ tosou/toij qrh/noij tolmh=sai/ tinaj to\n kairo\n ei)n= ai ke/rdouj, a)nqrw/pwn e)sti\n a)sebw=n kai\ qeostugw=n, ou)de\ u(perbolh=n a)topi/aj e)xo/ntwn. O /( qen e)d / oce tou\j toiou/touj e)kkhru/cai, mh/pote e)f`o(l / on e)l / qh to\n lao\n h( o)rgh\ kai\ e)p`au)tou\j prw=ton tou\j proestw=taj, tou\j mh\ e)pizhtou=ntaj. Fobou=mai ga\r, w(j h( Grafh\ le/gei, mh\ sunapole/s$ a)sebh\j to\n di/kaion. [Es folgt als Beleg (Eph. V.3 und 6-13) etc.] E ) a\n de\, dia\ th\n prote/ran pleoneci/an th\n e)n t$= ei)rh/n$ genome/nhn di/khn tinnu/ntej, e)n au)t%= t%= kair%= th=j o)rgh=j pa/lin pro\j th\n pleoneci/an e)ktrapw=si/ tinej, kerdai/nontej e)c ai)/matoj kai\ o)le/qrou a)nqrw/pwn a)nasta/twn genome/nwn h)\ ai)xmalw/twn, pefoneume/nwn, ti\ e(t/ eron prosdok#=n xrh\ h)\ w(j e)pagwnizome/nouj t$= pleoneci/#, e)piswreu)sai o)rgh\n kai\ e(autoi=j kai\ panti\ t%= la%=;

25

Z.B. Mitchell (1993a: 235 mit Anm. 59). Neben der Rede für Origenes und dem Bußkanon verfaßte Gregorios zahlreiche weitere Schriften, die z.T. überliefert sind; u.a. Crouzel (1983: 782ff.). 26 Eine Zusammenstellung der Handschriften bei Phouskas (1969: 127-32). Nach Phouskas (1969: 166) kommen in der griechisch orthodoxen Überlieferung auch Einteilungen in 12, bzw. 13 Kanones vor. Dräseke (1881: 730-6) wiederum bietet einen zusammenhängenden Text mit kritischem Apparat. 27 PG 10.967f. s. u. Anm. 53. 28 Dräseke (1881: 736f.); bei diesem Konzil 692 n. Chr. in Konstantinopel (PG 137.521A-C) wurde die Aufnahme dieser kano/nej in die Kanonsammlung der Ostkirche vollzogen, Phouskas (1969: 167). 29 Vgl. Eusebius HE 7.7.4; Dräseke (1881: 727f.). 30 Dräseke (1881: 728) vertritt die Meinung, daß mit sugge/rwn ein sumpresbu/teroj gemeint sei (1 Petr. 5. 1). Strobel (1993: 291): „ungenannter Amtsbruder in der Region“. 31 Vgl. auch Dräseke (1881: 728)

330

Jeorgios Martin Beyer: i(/na ... au)toi\ a)/lloij Bora/doi kai\ Go/tqoi gi/nwntai. Gregorios Thaumaturgos dies nicht erlaubt), wenn er sich im Unglück befindet, vor Feinden flieht und aus Not sein Eigentum im Stich läßt.38 Wenn nun jemand, so heißt es weiter im fünften Abschnitt, etwas verloren hat und an dessen Stelle etwas anderes als Ersatz für entstandene Schaden widerrechtlich nimmt, so verhält er sich laut Gregorios, wie die Barbaren ihm selbst gegenüber.39 Aus diesem Grunde schickt er seinen Bruder Euphrosynos, damit dieser die Strafen nach der wohl in Neokaisareia geläufigen Regel (kata\ to\n e)nqa/de tu/pon) festlegt, alle in Schuld Gefallenen anklagt, sowie alle, auf die die Anklage zutrifft, vom Gebet fernhält.40

soll. Zur Beurteilung und Einordnung der Vergehen wird im folgenden Kanon 3 noch darauf aufmerksam gemacht, daß es sich bei dem Entwendeten um geheiligtes Gut handelt: Wir müssen ein jedes (Gut), was nicht unser, sondern fremder (Besitz) ist - in dieser Zeit ein Gewinn - für gottgeweiht halten. Am Beispiel des Achan (bei Gregorios Achar) wird gezeigt (Jos. VII), daß der Habsüchtige im konkreten Fall der Gotenzüge sogar noch schlimmer handelt als der in dem Bibelzitat erwähnte Achan, der den Zorn Gottes durch seine Haltung auf das ganze Volk Israel lud. Und jener Achar nämlich nahm von dem Beutegut; und jener nahm zwar den Besitz der Feinde, die Menschen jetzt aber den Besitz der Brüder, indem sie sich verderbenbringenden Gewinn verschafften.“ 35

In Kanon 6 und 7 werden die schwerwiegendsten Verbrechen erwähnt: Dabei handelt es sich um das Verwerflichste, was Gregorios je zu Ohren gekommen ist. Christen, aber auch wohl Heiden, wie die Formulierung u(po\ a)pi/stwn kai\ a)sebw=n kai\ mh\ ei)do/twn mhde\ o)/noma Kuri/ou nahelegt, sollen aus gotischer Gefangenschaft Entflohene mit Gewalt festgehalten haben. Das scheint dem Bischof so unglaublich, daß er den Adressaten auffordert, jemanden in die betroffene Gegend zu schicken, um dies zu überprüfen.41 Daß die zur Ermittlung von Schuldigen Beauftragten auf das Land (xw/ra) gesandt werden, belegt einmal mehr, daß auch ländliche Regionen offensichtlich starke chrisltiche Bevölkerungsanteile aufwiesen (vgl. a. Mitchell 1993b: 56f.).

Die einzelnen Vergehen werden in weiteren Abschnitten aufgezählt und das Maß der Buße für sie festgelegt; dabei wird immer unterschieden, ob jemand diese Vergehen zugibt und bereut, oder nicht. In Kanon 8 geht es um diejenigen Christen, die während der Barbarenzüge in fremde Häuser eindrangen und etwas entwendeten; wenn jemand seine Schuld abstreitet, darf er nicht an der a)kro/asij teilnehmen; wenn er aber sein Vergehen bereut, wird er in die Klasse der u(po/ptwsij eingeordnet.36 Kanon 9 behandelt den Fall, daß Christen die zurückgelassenen Besitztümer von Gefangenen der Goten an sich genommen haben. Wer sich selbst anzeigt und die Güter zurückgibt, darf, wie aus dem Kommentar des Balsamon hervorgeht, wieder am Gebet teilnehmen, wer die Tat nicht zugibt, wird in die Klasse der u(po/ptwsij eingeordnet. Wenn die Buße vollständig vollzogen ist, so liest man im Kommentar des Zonaras, erhalten die Beschuldigten wieder das Recht, an allen Sakramenten teilzuhaben (me/qecij).37

Kanon 7 thematisiert schließlich Mord und Kollaboration. Diese Verbrechen sind unentschuldbar und Gregorios bestimmt in diesem Abschnitt, daß die Beschuldigten nicht einmal an der a)kro/asij teilnehmen dürfen. In diesem Kanon dreht es sich um diejenigen ‘Pontier und Christen’, die so ‘barbarisiert’ wurden, daß sie ihren Stammverwandten a)t/ opa antaten. Darunter zählte das Töten von Landsleuten durch Erschlagen oder Erdrosseln und das Aufzeigen von Wegen und Häusern, deren Lage den Barbaren unbekannt war. Dieses Verhalten wird als so verwerflich eingestuft, daß eine Zusammenkunft von betroffenen Gemeindevorstehern in einer speziellen

Die Abschnitte 4 und 5 behandeln Fälle, in denen der Beschuldigte sich im Recht glauben könnte. Wer sich z.B. die Habe eines ihm Unbekannten aneignet, die dieser weggeworfen hat, soll das Gefundene zurückgeben. Dies gilt nicht nur für die Habe von Christen, sondern auch für die der Feinde. Niemand täusche sich selbst, auch nicht als Finder; denn es ist auch nicht erlaubt, Gefundenes sich anzueignen (aus Gefundenem Gewinn zu schlagen). ... Wenn es aber im Frieden, weil ein Bruder oder Freund leichtsinnig oder nachlässig ist, und sich nicht um seinen Besitz kümmert, nicht erlaubt ist, Gewinn zu machen, wievielmehr (ist

38

Greg. Thaum. ep. can. 4: Mhdei\j e)capata/tw e(auto\n, mh/te w(j eu(rw/n! ou)t/ e ga\r eu(ro/nta kerdai/nein e)c / esti. ... [mit Stellen aus dem Deuteronomion und Exodus wird belegt, daß nicht nur das Eigentum von ‘Brüdern’, sondern auch das von Feinden zurückgegeben werden muß.] ... Ei) de\ e)n ei)rh/n$ r(#qumou=ntoj kai\ trufw=ntoj kai\ tw=n i)di/wn a)melou=ntoj a)delfou= h)\ e)xqrou=, kerda=nai ou)k e)c / estin, po/s% ma=llon dustuxou=ntoj kai\ polemi/ouj feu/gontoj, kai\ kata\ a)na/gkhn ta\ i)d / ia e)gkataleipo/ntoj; 39 Greg. Thaum. ep. can. 5: A /) lloi de\ e(autou\j e)capatw=sin, a)nti\ tw=n i)di/wn tw=n a)polome/nwn a(/ eu(r = on a)llo/tria kate/xontej! i(n/ a e)peidh\ au)toi=j Bora/doi kai\ Go/tqoi ta\ tou= pole/mou ei)rga/santo, au)toi\ a)l / loij Bora/doi kai\ Go/tqoi gi/nwntai. 40 Greg. Thaum. ep. can. 5: A ) postei/lamen ou)n= to\n a)delfo\n kai\ sugge/ronta Eu)fro/sunon dia\ tau=ta pro\j u(ma=j, i(n/ a kata\ to\n e)nqa/de tu/pon kai\ au)to\j dw/$ o(moi/wj kai\ w(n= dei= ta\j kathgori/aj prosi/esqai, kai\ ou(j / dei= e)kkurhcai tw=n eu)xw=n. 41 Greg. Thaum. ep. can. 6: A ) phgge/lqh de/ ti h(mi=n kai\ a)p / iston e)n t$= xw/r# u(mw=n geno/menon, pa/ntwj pou= u(po\ a)pi/stwn kai\ a)sebw=n kai\ mh\ ei)do/twn mhde\ o)n/ oma Kuri/ou, o(t/ i a)r / a ei)j tosou=to?n tinej w)mo/thtoj kai\ a)panqrwpi/aj proexw/rhsan, w(s / te tina\j tou\j diafugo/ntaj ai)xmalw/touj bi/# kate/xein. A ) postei/late/ tinaj ei)j th\n xw/ran, mh\ kai\ skhptoi\ pe/swsin e)pi\ tou\j ta\ toiau=ta pra/ssontaj.

35

Greg. Thaum. ep. can. 3: Ou)k i)dou\ o( A /) xar o( Zara= plhmmelei/# e)plhmme/lhsen a)po\ tou= a)naqe/matoj, kai\ e)pi\ pa=san sunagwgh\n I) srah\l e)genh/qh h( o)rgh/; Kai\ ei(j = mo/noj ou(t= oj h(m / arte, mh\ mo/noj a)pe/qanen e)n t$= a(marti/# au)tou=. H ( mi=n de\ pa=n to\ mh\ h(me/teron, a)lla\ a)llo/trion, t%= kair%= tou/t% ke/rdoj, a)na/qema nenomi/sqai prosh/kei. Ka)kei=noj me\n ta\ tw=n polemi/wn, oi( de\ nu=n ta\ tw=n a)delfw=n, kerdai/nontej o)le/qrion ke/rdoj. 36 Greg. Thaum. ep. can. 8: Tou\j de\ oi)k / oij a)llotri/oij e)pelqei=n tolmh/santaj, e)an\ me\n kathgorhqe/ntej e)legxqw=si, mhde\ th=j a)kroa/sewj a)ciw=sai! e)an\ de\ au)tou\j e)cei/pwsi kai\ a)podw=sin, e)n t$= tw=n u(postrefo/ntwn ta/cei u(popi/ptein. 37 Greg. Thaum. ep .can. 9: Tou\j de\ e)n pedi/% eu(ro/ntaj ti h)\ e)n tai=j e(autw=n oi)ki/aij kataleifqe/n u(po\ tw=n barba/rwn, e)an\ me\n kathgorhqe/ntej e)legxw=sin, o(moi/wj e)n toi=j u(popi/ptousin! e)an\ de\ e(autou\j e)cei/pwsi kai\ a)podw=si, kai\ th=j eu)xh=j a)ciw=sai.

331

Limes XVIII Zusammenkunft darüber befinden sollen.42 Ein ähnliches Verhalten wird ein Jahrhundert später für die Raubzüge der Goten auf dem Balkan von Ammianus Marcellinus berichtet.43 Hier dient das Aufzeigen von Häusern besonders der Versorgung der Plünderer mit Nahrungsmitteln. Die Kriegsgefangenen wurden dazu benutzt, Wege aufzuzeigen zu Vorratslagern von Feldfrüchten in reichen Dörfern, Schlupfwinkeln von Menschen und entlegenen Zufuchtsstätten. Die in diesen Dörfern befindlichen Menschen wurden entweder Gefangengenommen oder erschlagen.44 Auch im Falle der pontischen Raubzüge dürften die Beweggründe für die Plünderung der Dörfer in der Umgebung der Städte ähnliche gewesen sein.

Herwig Wolfram hat zurecht auf das Fehlen jeglicher Äußerung des Gregorios selbst zur decischen Christenverfolgung hingewiesen.49 In der Zeit der decischen Christenverfolgungen hatte dieser sich mit Teilen seiner Gemeinde in die Berge zurückgezogen.50 Ähnlich wie etwa Cyprian und andere Bischöfe seiner Zeit hielt er es für vordringlicher, seine Glaubensbrüder von sicherer Stelle aus zu lenken (Greg. Nyss. PG 46.947D). Während der Verfolgung fanden die Versammlungen, wie in einem berichteten Fall, wohl unter freiem Himmel statt.51 Wolfram folgert, daß Gregorios mit dem ersten Kanon in einem Schritt auch den während der decischen Christenverfolgung, die nach Gregor von Nyssa gerade in dieser Region sehr heftig wütete (Greg. Nyss. PG 46.944ff.), abgefallenen Christen die Wiedereingliederung in die Gemeinde ermöglichen wollte. Besonders die Tatsache der ausführlichen Behandlung der Frage nach dem Verzehr von Opferfleisch läßt es zumindest offen, ob es sich bei Kanon 1 daneben auch um eine Art Freibrief für alle lapsi handeln kann, denn es ist höchst wahrscheinlich, daß für die lapsi bereits Vorstellungen vor dem kanonischen Brief formuliert wurden. Dies zumindest war ein Problem der gesamten Kirche zu dem sich andere Autoritäten bereits geäußert hatten, insbesondere Bischofs Cyprian von Karthago in seiner Schrift de lapsis.52

Die Klassifizierung der einzelnen Vorfälle durch Gregorios bezieht sich ausschließlich auf solche Vergehen, die von Christen unternommen wurden. Denn die angedrohten Strafen betrafen Heiden selbstverständlich nicht. Dies zeigt sich besonders deutlich an den zwei Arten von Vergehen, die für Gregorios nicht mit einer Strafe zu belegen sind. Im ersten Abschnitt nennt er darunter den Verzehr von ‘verunreinigten’ Speisen in Gefangenschaft. Außerdem sollen vergewaltigte Frauen schuldfrei sein: Nicht die Speisen belasten uns, heiligster Vater, ob die Gefangenen das verzehrt haben, was ihnen ihre Bewacher vorsetzten, zumal da eine einzige Rede bei allen ist, daß die Barbaren, die in unsere Gegenden vorgedrungen sind, den Götzenbildern nicht geopfert haben. ... (Auch belastet uns nicht,) daß die weiblichen Gefangenen entehrt wurden, wenn die Barbaren sich an ihren Körpern vergingen (Greg. Thaum. ep. can. 1). Beides belegt er mit Bibelstellen.45 Wesentlich ist auch, daß der Bischof entweder Erkundigungen eingezogen hatte, oder daß es, wie er selbst bekennt, allgemein bekannt war, daß die Barbaren heidnischen Kultbildern nicht geopfert haben.46 Es wird deutlich, daß die Bewertung der Vorfälle dem anfragenden Bischof nicht eindeutig erscheint, betont durch die ersten Worte des kanonischen Briefes: ou) h(ma=j barei=.47 Gregorios erklärt aus diesem Grunde verhältnismäßig ausführlich, warum man sich mit den angesprochenen Problemen nicht zu befassen braucht.48

Der 11. Kanon schließlich bietet eine Zusammenstellung der fünf Bußstufen (pe/nte to/poi metanoou/ntwn) für die Teilnahme am Gottesdienst, der die Büßenden je nach Verfehlungsgrad zugeordnet werden: - pro/sklausij:

- a)kro/asij:

Die metanoou/ntej müssen außerhalb des eu)kthri/ou (oratorium) stehen; dort sollen sie die in die Kirche Gehenden anflehen, für sie zu bitten. Die metanoou/ntej dürfen innerhalb der Tür im Narthex stehen; dort dürfen sie bleiben, bis die katexou/menoi herauskommen, d.h. bis die Predigt abgeschlossen ist; sie sind aber nicht des Gebetes würdig.

der Formel deinh\ h( angerissen und ab ep.can. 4 tatsächlich fallweise abgehandelt. 49 Wolfram (1990: 60). Die Möglichkeit, daß bei der Zerstückelung des Urtextes solche Passagen, die die lapsi ansprachen, verlorengegangen sein können, sind eher unwahrscheinlich; Dräseke (1881: 730-6) macht in seiner Zusammensetzung des nicht eingeteilten Textes keine Angaben dazu. 50 Greg. Nyss. PG 46.948: o( de\ h)n= e)r / hmo/n tina katalabw=n lofi/an. Zu den Christenverfolgungen unter Decius vgl. allg. Seliger (1995). 51 Greg. Nyss. PG 46.941C sullo/gou tino\j kata\ to\ u(p / aiqron e)n/ tini tw=n kata\ th\n xw/ran to/pon gegeneme/nou. 52 Alföldy (1973: 479-502) beschäftigt sich mit der Reaktion Cyprians auf die Christenverfolgung und ihre Hintergründe; s. a. Sage (1975: 241ff.); Strobel (1993: 146-184, bes. 155 Anm. 132) widerspricht Alföldy (Krise: 301f.), daß Cyprians de lapsis aus einem Krisenbewußtsein geschrieben ist, und er sei durch die Kirchenspaltung 251 n.Chr. in eine Weltuntergangsstimmung geraten; zur Schrift selbst vgl. u.a. Sage (1975: bes. 232-240). Zum Niederschlag, den die Verfolgungen im Werk Cyprians hinterlassen haben: Saumagne (1962: 1ff.). Auch in Cyprianus de lapsis 6 wird als eine der Triebfedern die Gier und Habsucht genannt, hier allerdings in Zusammenhang mit der Klage über die Endzeit: Studebant augendo patrimonio singuli et, obliti quid credentes aut sub apostolis ante fecissent aut semper facere deberent, insatiabili cupiditatis ardore ampliandis facultatibus incubabant.

42

Greg. Thaum. ep. can. 7: Tou\j me\n ou)=n e)gkatalexqe/ntaj toi=j barba/roij kai\ met au)tw=n e)n ai)xmalwsi/# e)pelqo/ntaj, e)pilaqome/nouj, o(t/ i h)s = an Pontikoi\ kai\ Xristianoi\, e)kbarbarwqe/ntaj de\ w(j kai\ foneu/ein tou\j o(mofu/louj h)\ cu/l% h)\ a)gxo/n$, u(podeiknu/nai de\ h)\ oi)ki/aj a)gnoou=si toi=j barba/roij! kai\ th=j a)kroa/sewj a)pei=rcai dei=, me/xrij a)n\ koin$= peri\ au)tw=n ti do/c$ sunelqou=si toi=j a(gi/oij, kai\ pro\ au)tw=n t%= a(gi/% Pneu/mati. Slusser (1998: 150 Anm. 15) übersetzt cu/lon „beating with a club“, während Heather & Matthews (1991: 9) „gibbet“ übersetzen. 43 Amm. Marc. XXXI.6.5: dediticiis vel captivis vicos uberes ostendentibus, eos praecipue, ubi alimentorum repperiri satias dicebatur. Deditizier sind im Falle der Seezüge nach Pontos in der Mitte des 3. Jhs. kaum zu erwarten. 44 Amm. Marc. XXXI.6.6ff. Weitere eindringliche Schilderung der Greueltaten bei der Plünderung: Amm. Marc. XXXI.8.7f. 45 Zur Verunreinigung: 1 Kor. VI.13 und Matth. XV.11; zur Vergewaltigung: Deut. XXII.26f. 46 Greg. Thaum ep. can. 1: ei(=j lo/goj para\ pa/ntwn ... ei)dw/loij mh\ tequke/nai. 47 Für wie wichtig man das Problem der Verunreinigung durch Verzehr von Götzen geweihtem Fleisch hielt, s. Fredouille (1981: 869-877); zu Vergewaltigung s. Herter & Herrmann (1957: 1190ff.). 48 Die wesentlichen Bestimmungen werden in Greg. Thaum ep. can. 2 mit

332

Jeorgios Martin Beyer: i(/na ... au)toi\ a)/lloij Bora/doi kai\ Go/tqoi gi/nwntai. Gregorios Thaumaturgos - u(po/ptwsij:

- su/stasij:

- me/qecij:

Zur konkreten Umsetzung der Forderungen finden sich im kanonischen Brief außer der Kontrolle durch entsandte Prüfer und der Ankündigung einer Versammlung, die über die Härtefälle beraten soll, kaum Hinweise. Der Text scheint jedoch zu implizieren, daß Gregorios entwendetes bzw. verlorenes Eigentum zunächst in die Verwahrung der Kirche nehmen will.58 Wenn mit dieser Maßnahme eine Art ‘Sammelstelle’ gemeint ist, eine institutionalisierte Hilfsmaßnahme, für die Gregorios als Bischof jedoch keine rechtliche Handhabe besaß,59 so belegt auch dies die Selbstverständlichkeit, mit der die Durchsetzbarkeit der Forderungen angenommen wird. In Kanon 10 ruft Gregorios alle auf, der e)ntolh/ Folge zu leisten;60 auch sollen sie dafür keinerlei Entschädigung verlangen;61 weder eine Belohnung für die Anzeige einer Straftat (mh/nutra), noch einen entsprechenden für die Errettung aus der Gefangenschaft (sw=stra) und auch keinen Finderlohn (eu(r / htra).62 Diese Bestimmung zielt gegen alle Arten von Erpressung oder Niesnutz aus dem Unglück der Raubzüge. Gregorios scheint nicht nur aus finanziellen Gründen sondern auch prinzipiell eine scharfe Trennungslinie zu ziehen. Cyprian hingegen erweist sich in entsprechenden Notsituationen als ausgesprochen pragmatisch. So stellt er Geld für die Betreuung von Eingekerkerten bereit;63 ebenso überweist er den Bischöfen von Numidien als Unterhändlern 100.000 Sesterzen zum Freikauf von durch Berber gefangengenommene Glaubensgenossen.64 Von praktischen Hilfsmaßnahmen des Gregorios zugunsten von Gefangenen oder sogar den Freikauf aus der Hand der Boranen ist nichts bekannt,65 aber auch nicht prinzipiell auszuschließen, selbst wenn die finanzielle Situation der Kirche in Pontos nicht der afrikanischen entsprechen dürfte, zumal nach den Beutezügen. Mit der Ausführung beauftragt er andere, die seine Anweisungen auch in benachbarten Gebieten durchsetzen sollen. Gregorios übernimmt in diesem konkreten Fall also auch Aufgaben des Staates, die allerdings auf den christianisierten Teil der Gesellschaft beschränkt bleiben. Die Verfolgungen des Decius und des Valerian, wie auch die Gotenzüge, gaben den Christen die Möglichkeit, sich straffer zu organisieren.

Die Büßer dürfen ins Innere der Kirche (nao/j), müssen aber mit den katexou/menoi wieder herauskommen. Die metanoou/ntej dürfen mit den Gläubigen (pistoi/) am Gottesdienst teilnehmen und müssen nicht mit den katexou/menoi wieder heraus. Die metanoou/ntej dürfen an allen a(gia/smata (sacramenta) teilhaben.

Diese Systematik, die auf Basileios von Kaisareia zurückgeht,53 beinhaltet nicht, wie lange die Büßer zu büßen haben, d.h. wie lange sie diesen Stufen angehören sollen. In dem Kommentar Balsamons wird darauf hingewiesen, daß der Verfasser keine kairoi\ e)rgasi/aj festgeschrieben hat; diese sei erst in den kanonischen Briefen des Basilios geschehen. In den ersten 10 Abschnitten des Briefes werden nicht alle in Kanon 11 genannten Bußstufen erwähnt. In Kanon 2 wird allgemein bestimmt, daß jeder der der pleoneci/a verfällt, aus der Kirche ausgeschlossen werden.54 Spezifiziert werden durch Gregorios einzig im 5. Kanon der Ausschluß von den Gebeten bzw. Fürbitten (eu)xh/),55 im 7. Kanon die Bußstufe der Ausschluß von der a)kro/asij, in Kanon 8 erneut die Bußstufe der a)kro/asij und die Klasse der u(postre/fontej, die wiederum in der Zusammenfassung nicht in diesem Wortlaut erscheint, sowie in Kanon 9 die u(po/ptwsij und die Einbeziehung in die Gebete (kai\ th=j eu)xh=j a)ciw=sai) Die bei Gregorios etwas unscharf wirkenden Formulierungen stehen - mit Ausnahme der Bestimmungen in Kanon 8 und 9 - in deutlichem Kontrast zu der deutlichen Beschreibung der Bußstufen im angehängten 11. Abschnitt.56 Ausdrücke wie e)ntolh/ deuten jedoch darauf hin, daß sich in dieser Zeit die kirchliche Organisation auch in Fragen einer kirchlichen Rechtsprechung formiert. Der kanonische Brief steht also auch als ein Dokument mitten in einer Entwicklung hin zu einem System kirchlicher Gesetzgebung. Immerhin läßt sich feststellen, daß im Vergleich zur Situation vor den Krisen des 3. Jhs. die Bußordnung sich zunehmend verfeinerte.57

53

58 Dies könnte sich in dem Ausdruck h(mi=n de pa=n to\ mh\ h(me/teron ... a)na/qema nenomi/sqai prosh/kei andeuten. 59 Slusser (1998: 151): „This appears to refer to those, who are in charge of putting things right after the Goth’s departure, whether local leaders, an emissary like Euphrosynos, or those sent out in section 6 to sent detained captives free.“ 60 Wie konkret der Begriff e)ntolh/ gefaßt ist, geht nicht explizit aus dem Text hervor. Es handelt sich hierbei nicht um eine christliche Formel; im NT etwa wird an entsprechenden Stellen nahezu immer das Wort no/moj gebraucht. 61 Greg. Thaum. ep. can. 10: Tou\j de\ th\n e)ntolh\n plhrou=ntaj e)kto\j pa/shj ai)sxrokerdi/aj plhrou=n dei=, mh/te mh/nutra h)\ sw=stra, h)\ eu)r / etra, h)\ %(= o)no/mati tau=ta kalou=sin, a)paitou=ntaj. 62 Vgl. Sirm. 16; s.u. Anm. 70 und 87. 63 Cyprianus ep. 7; Zum Besitz Cyprians vgl. z.B. Dölger (1950: 297ff.). 64 Cyprianus ep. 62; vgl. Klingshirn (1985: 184f.). 65 Zum Freikauf von Gefangenen als ‘Pflicht’ des Bischofs in Lobreden und Panegyriken s. u.a. Klingshirn (1985: 185 Anm. 20ff.) mit Material zu späteren Jahrhunderten; dabei wurde auch Kirchenbesitz angetastet, u.a. von Augustinus (Vita August. 24) oder Hilarius von Arles (Vita Hil. 11), bis es sogar in die Gesetzeswerke als Möglichkeit einging (Cod. Iust. 1.2.21 aus dem Jahre 529 n. Chr. oder Nov. 120, 10 von 544).

Basilius can. 75 im 3. Brief des Basileios an Amphilochos über Kanones; auch die Notitia (PG 10. 968) hält ihn für unecht; Lane Fox (1986: 766 Anm. 38) hält hingegen wie Telfer (1936: 228) und andere den 11. Kanon für original, jedoch ohne nähere Begründung; desgleichen behauptet Phouskas (1969: 171f.), daß zumindest dem Inhalt nach dieser Kanon echt sei, da die gleichen Begriffe erscheinen; bedenkenswert ist jedoch die These, es handele sich hier um eine Art postscriptum. Bestätigen könnten schließlich auch ‘archäologische’ Überlegungen die Datierung des 11. Abschnitts in das 4. Jh., denn die in dem Kanon vorausgesetzten Baulichkeiten, etwa ein Narthex, sind in der Mitte des 3. Jhs. nicht nachweisbar; vgl. Mitchell (1993b: 56 Anm. 26 und 67 Anm. 82, 83 mit Literatur). 54 kai\ pa=j toiou/toj e)kkh/ruktoj E ) kklhsi/aj Qeou= (Greg. Thaum. ep. can. 2) entspricht der Formulierung in Greg. Thaum. ep.can. 5: e)kkhru/cai tw=n eu)xw=n. 55 vgl. Basilius can. 30; zur Diskussion der Termini eu)xh/ und proseuxh/ s. Grotz (1955: 401). 56 Die Stufe der u(po/ptwsij ist ebenso wie die der a)kro/asij sicher ausgebildet, die der pro/sklausij und der su/stasij noch nicht präzise identifizierbar, jedoch angedeutet; vgl. Grotz (1955: 408). 57 Vgl. u.a. Poschmann (1940: 475f.), Grotz (1955: 400ff.) und Heather (1991: 10 Anm.26).

333

Limes XVIII

Menschen, die unrechtmäßigerweise nach ihrer Befreiung in andere Hände geraten sind, bzw. ihren sozialen Status als freie Bürger nach ihrer Befreiung nicht wiedererlangt haben.71 Von den bei Gregorios Thaumaturgos angesprochenen Gefangenen wird nicht explizit gesagt, von wem sie unrechtmäßig festgehalten werden. Wahrscheinlich handelt es sich dabei entweder um Straßenräuber oder um organisierte Gruppen anderer Art. Das unrechtmäßige Festhalten von Personen durch Privatleute erinnert zudem an das Verhalten von Großgrundbesitzern, die in diesen Extremsituationen oftmals darauf verfielen, ihre Besitzungen durch Privatarmeen zu sichern.72 Als Folge daraus ist auch für das 3. Jh. zu konstatieren, daß Angehörige der lokalen Oberschicht zur Verteidigung gegen äußere und innere Feinde Selbsthilfekommandos bildeten, die auch aufgrund ihrer Ortskenntnis schnell kleine begrenzte Herrschaften ausbilden konnten (Wolfram 1990: 60f.). Im kleinasiatischen Raum gab es Privatarmeen, die im 3. Jh. Feinde des Staates angesehen wurde. Das Beispiel des Valerius Statilius Castus zeigt, daß die Banden, die die Bergregionen verunsicherten, nicht nur als Kriminelle angesehen wurden.73 Wie stark diese Umstände die Regierenden beschäftigen konnten, zeigt das Beispiel von Kremna 278 n.Chr. (Mitchell 1989: 311-328). Die Probleme, die durch die Raubzüge der Boranen und Goten aufgeworfen wurden, sind grundsätzlich denen vergleichbar, die in späteren Zeiten in der kleinasiatischen Region durch die isaurischen Straßenräuber, in Gallien und Spanien durch die organisierten Banden der Bagauden und in Afrika durch die Circumcellionen verursacht wurden.74 Konkret gegen Lokalherrscher richten sich vergleichbare Gesetze späterer Zeit z.B. gegen den Einsatz von Soldaten für private Zwecke.75 Die dort angesprochenen Nebentätigkeiten sind landwirtschaftlicher Natur, bzw. Aufgaben der Sicherung; man kann vermuten, daß auch Soldaten für die Sicherung privater Güter verwendet wurden.76

*** Wie lassen sich nun die von Gregorios Thaumaturgos angeführten Ereignisse einordnen? Während kriegerischer Auseinandersetzungen oder Barbareneinfällen standen besonders Diebstahl und Vergewaltigung an der Tagesordnung. Insofern deckt sich die Aufzählung des Gregorios mit solchen früherer wie späterer Ereignisse. Besonders kongruent erscheint dabei angesprochene Schilderung der Raubzüge des 4. Jhs. bei Ammianus Marcellinus (XXXI.6.5ff.; XXXI.8.7f.). GEM de Ste. Croix parallelisiert diese Ereignisse mit denen bei Gregorios, und erklärt dies mit der ‘hoffnungslosen’ Situation der Zivilbevölkerung in dieser Zeit.66 Es ist auffällig, daß es aus dieser Zeit keine Stellungnahme von staatlicher Seite zu den Eigentumsproblemen gibt, die sich im Zuge der Barbareneinfälle in Pontos ergaben. Eine Vorstellung von den Gegenmaßnahmen des römischen Staates bieten einige Abschnitte im Codex Iustinianus. Sie sehen für Kollaborateure, die den Barbaren durch verbrecherische Umtriebe die Möglichkeit zum Beutemachen gegen die Römer gegeben haben, bzw. mit ihnen die Beute teilen, die Verbrennungsstrafe vor.67 Auch für Übermittlung von technischem Know-How werden hohe Strafen angedroht.68 Wieder andere Gesetze unterscheiden zwischen Kollaborateuren und Nicht-Kollaborateuren.69 Ein großes Problem stellte letztlich auch die Mitnahme von Gefangenen und die Zwangsansiedlung in den Heimatgebieten dar. Sie entvölkerten nicht nur die von den Raubzügen befallenen Gebiete, sie führten auch zu einem beachtlichen Transfer von Know-How. Auch hier bietet die parallele Sicht auf die Ereignisse der Plünderungszüge der Sasaniden in Syrien, Kilikien und Kappadokien eine Vorstellung von dem Ausmaß des Schadens für die Grenzprovinzen des Imperium Romanum. Die mitgeführten Gefangenen wurden konsequent im Kernbereich des Sasanidenreiches angesiedelt (RGDS (gr.) Zeile 34-36). Besonderes Gewicht jedoch erhält bei Gregorios die Festsetzung von aus den Händen der Barbaren befreiten, entflohenen oder zurückgekehrten römischen Bürgern. So wird z.B. für entsprechende Fälle unter Diokletian das Rückkehrrecht geregelt, d.h. daß jeder, wenn er zurückkehrt, sein Eigentum und seinen Status wiedererhält.70 Dabei handelt es sich bei diesen Fällen fast ausschließlich um

71

Cod. Theod. 5.6.2 wird die Forderung nach Rückgabe der abgejagten Beute und der befreiten Gefangenen gestellt. 72 Auch war es seit dem 4. Jh. n. Chr., wie Amm. Marc. XXIX.5.13 und 25 berichtet, nicht ungewöhnlich, private Häuser zu befestigen, wogegen im 5. Jh. n. Chr. Gesetze erlassen wurden (Cod. Iust. 8.10.10). In die gleiche Richtung zielen Gesetze gegen die Haltung bewaffneter Banden (Cod. Iust. 9.12.10) und gegen private Kerker (Cod. Iust. 9.5). Eine weitere Reaktion auf die Barbarenzüge läßt sich auch anhand archäologischer Quellen erkennen. Die Städte sind nämlich in dieser Zeit neu bzw. besser befestigt worden, wobei es keine Unterschiede zwischen dem Osten und dem Westen des Reiches gab; Millar (1969: 29). Dies ist entweder im Anschluß an die Barbarenzüge oder als Präventivmaßnahme geschehen. 73 Zu Privatarmeen vgl. Mitchell (1993a: 234) mit dem Beispiel des Valerius Statilius Castus, der mir der Bezeichnung su/mmaxoj tw=n Sebastw=n erscheint; zur Identifizierung von Banditen mit Staatsfeinden s. a. Shaw (1984: 3-51). 74 Zosimus 1.69.1. Schilderung größerer Kämpfe mit den Isauriern aus den Jahren 367/8 n. Chr. bei Amm. Marc. XXVII.9.6f. und aus den Jahren 376/7 n. Chr. bei Zosimus 4.20.1f. Gesetzgeberische Maßnahmen aus dem Cod. Iust. 3.12.8 (= Cod. Theod. 9.35.7) aus dem Jahr 408 n. Chr., das von der Befragung der isaurischen Straßenräuber handelt; (Cod. Iust. 3.27.2 aus dem Jahr 403 n.Chr. erlaubt es jedem, Straßenräuber und Deserteure sofort zu töten; generell zu dem Räuberwesen in dieser Zeit: Thompson (1981: 29-47); zur marxistischen Literatur über die Bagauden: Doi (1989: 344f. Anm. 2-9). 75 Cod. Iust. 12.35.15 aus dem Jahr 458 n. Chr. 76 Umgekehrt erinnern die Bestimmungen Vergehen weitestgehend auch an die unrechtmäßigen Requirierungen von Gütern und Dienstleistungen durch

66 Zu diesen Zuständen, sowie zur Kollaboration mit den Goten: Zosimus 1.32-35. Ste. Croix (1981: 478) versteht dies im Zusammenhang mit ‘revolutionären’ Aktivitäten der Unterklassen; Lane Fox (1986: 766 Anm. 34) lehnt dies explizit ab. 67 Cod. Iust. 12.35.9 (= Cod. Theod. 7.1.1); Konstantins I. Gesetz vom 28. April 323 n. Chr., das allen Kollaborateuren (mit Barbaren) den Verbrennungstod androht, ist als Reaktion auf den Einsatz gotischer Truppen im Bürgerkrieg von Seiten des Licinius zu verstehen. 68 Cod. Iust. 9.47.25 (= Cod. Theod. 9.40.24) von 419 n. Chr. erlegt denen, die den Barbaren die Schiffbaukunst beibrachten, die Kapitalstrafe auf; s.a. Stoll (1998: 267ff. mit ausführlicher Literatur). 69 Cod. Iust. 8.50.19 (= Cod. Theod. 5.5.1) aus dem Jahre 366 n.Chr. 70 Cod. Iust. 8.50.5 aus dem Jahr 290 n. Chr. Die Rückkehr in den vorherigen Status im Falle von Nichtfreien regelt Cod. Iust. 8.50.12 aus dem Jahr 293 n. Chr., desgl. Cod. Iust. 8.50.10; zu den Erbansprüchen Cod. Iust. 8.50.15.

334

Jeorgios Martin Beyer: i(/na ... au)toi\ a)/lloij Bora/doi kai\ Go/tqoi gi/nwntai. Gregorios Thaumaturgos genüber. Keine der relevanten christlichen Quellen, erst recht nicht Gregorios, läßt eine solche Haltung erkennen. Im Gegenteil: Seine grundsätzlich positive Haltung gegenüber dem römischen Staat äußert sich auch in seiner Rede an Origenes (Greg. Thaum. or. prosph. 1.7). Als Schüler des Origenes teilt er wohl auch dessen Auffassung über die Einschließung des Römischen Reiches in den Weltenplan Gottes (Origenes c. Cels. 2.30). Ausdrücklich betont Gregorios, daß die römischen Gesetze auch für ihn Gültigkeit haben.83 Der Übergang vom Heidentum zum Christentum in der lokalen Oberschicht, der Gregorios auch angehörte, geschah, was die repräsentativen Pflichten und die Klientel angeht, sicher nicht abrupt. Die Interessen der ‘neuen’ christlichen Führung – zum großen Teil deckungsgleich mit der ‘alten’ heidnischen – waren, was den Erhalt der inneren Ordnung betrifft, kongruent mit denen des römischen Staates.84 Die von Gregorios Thaumaturgos geforderten Maßnahmen und die Regelung der Bußstufen erhöhte zusätzlich die soziale Kontrolle, da für jeden sichtbar wurde, wer sich wie weit von den christlichen Wertmaßstäben entfernt hatte. Auszuschließen ist auf jeden Fall, daß sich wie im Falle der nestorianischen Christen in späterer Zeit85 Nachbarn des Römischen Reiches sich der entstandenen Unruhe bedienten, um Christen bzw. andere Einwohner des Römischen Reiches, von denen sie wußten, daß sie in Konflikt mit der Obrigkeit standen, für ihre Zwecke einzusetzen.

Ebenso verhält es sich mit dem Problem der Kollaboration. Wolfram deutet in seinem Buch über die Geschichte der Goten an, daß sozial Schwächere die Gunst der Stunde nutzten, um ihre Situation zu verbessern und als Kollaborateure eine Art 5. Kolonne gegen die Römer bildeten.77 Die Kollaboration mit den Invasoren ist jedoch kein Zeichen für soziale Unruhen oder eine Destabilisierung innerhalb der christlichen Gemeinden nach der decischen Verfolgung, wie Frend vermutet.78 Die Tatsache, daß die Verdammung der Kollaboration formal eingerahmt wird durch Schilderung der Festsetzung der aus der Gefangenschaft Entkommenen Greg. Thaum. ep. can. 8) und die der Plünderung von verlassenen Häusern (Greg. Thaum. ep. can. 8) zeigt, daß es sich hier nicht um innere Unruhen, sondern einfach um anarchische Phänomene handelt, die auf der Zerstörung des alltäglichen Lebens gründen. Somit sind sie keine Vorboten der sozialen Spannungen, die später die CircumcellionenBewegung charakterisierten. Wie sehr die ländliche Bevölkerung zu leiden hatte, kann man an zwei Inschriften aus der Mitte des 3. Jahrhunderts ersehen.79 Die mißliche soziale Lage vieler Bewohner des Römischen Reiches, sowie ein gewisses Gefühl der Rechtlosigkeit, das sich hier ausdrückt, waren Voraussetzung für die Unterstützung, die die Goten bei ihren Zügen wohl an so manchen Orten fanden. Dies wird bestätigt durch die Tatsache, daß einige Jahre vor der Abfassung des kanonischen Briefes im Jahre 253 n. Chr., Cyprian über Mangel an Arbeitskräften, Rückgang der handwerklichen Produktion und Handel, Inflation und Unterbevölkerung klagt, und erwartet, daß der Fall Roms in unmittelbarer Zeit geschehen werde,80 Commodian die Vernichtung des Reiches durch Barbareneinbrüche und Christenverfolgungen voraussagt.81 Dennoch ist nicht davon auszugehen, daß sich Christen in stärkerem Maße als andere Teile der Bevölkerung an Vergehen dieser Art beteiligten, wie WHC Frend annahm.82

Letztendlich existierten die angesprochenen Probleme ohnehin im gesamten Randbereich des Römischen Reiches,86 nahmen aber durch die Destabilisierung in der Mitte des 3. Jhs. für die innere Sicherheit immer bedrohlichere Ausmaße an. Das Schweigen des römischen Staates zu den Zuständen in Pontos erklärt sich einzig und allein aus der Tatsache, daß die Beschäftigung mit den Sasaniden vordringlicher war. Da diese Region weitgehend christianisiert war, ist durchaus vorstellbar, daß zunächst die christliche Spitze reagieren mußte, um die öffentliche Ordnung wenigstens notdürftig zu sichern. Die Chance wurde durch die christliche Führung erkannt und um eine Lösung der organisatorischen wie auch theologischen Probleme gerungen. Damit übernahm sie zumindest zeitweise auch Teile der Pflichten der römischen

Daß also religiöse Minderheiten, wie die Christen in Zeiten der Verfolgung, die Goten unterstützten, weil auch diese sich gegen den römischen Staat wandten, ist also eher unwahrscheinlich. Wenn sie dies doch taten, so nicht aus einer prinzipiellen Ablehnung der römischen Herrschaft gerömische Soldaten. Auffälligerweise werden sie mit den Termini pleoneci/a und a)diki/a belegt. PSI IV 446 (Edikt des M. Petronius Mamertinus 133-137 n. Chr.): Z. 9f. to\ de\ stratiwtiko\n e)pi\ pleneci/# kai\ a)diki/# diaba/llesqai sunbe/bhke, vgl. dazu u.a. a. Kissel (1995: 255ff.). Daß pleoneci/a als Vorwurf sowohl in christlichem wie in heidnischem Zusammenhang in der Spätantike Verwendung fand: vgl. Leven (1988: 179 mit Anm. 28). 77 Wolfram (1990: 57) nennt das allgemein für die Gotenzüge der Mitte des Jahrhunderts, nicht nur für diese Region. Es ist anzunehmen, daß dies auch auf die pontischen Raubzüge zutrifft. 78 Frend (1967: 317) mit Bezug auf Greg. Thaum. ep.can. 7. 79 CIL III 12336; CIL III 14191. Frend (1967: 423) sieht eine Verbindung dieser Inschriften mit der Situation bei den Goteneinfällen in Pontos. 80 Cyprianus ad Demetr. 3; Zur Diskussion von Cyprians Schrift ad Demetrianum s. mit unterschiedlicher Bewertung u.a. Alföldy (1975: 116), Berthold (1988: 41ff.), Barthelink (1970: 91ff.) und Strobel (1993: 14 Anm. 22; 176 u.ö.). Für Strobel (1993: 176 Anm. 304) liegt hier kein Beleg für eine sachliche Feststellung allgemein verschlechteter Lebensbedingungen vor. Inschrift CIL III 14191 (s.o.) verwendet für diese Zeit den Ausdruck ponhri/a kai\ diaseismoi/. 81 Commod., Instr. I.41.1ff.; zu Commodian vgl. Strobel (1993: 364-9). 82 Frend (1967: 423); ablehnend Mitchell (1993a: 236 Anm. 59)

83

Greg.Thaum. or. prosph. 1.7; vgl. Millar (1992: 97); zur Diskussion Fox (1986: 764, Anm. 29). Gregorios vertritt die Meinung, daß das römische Recht dem Griechentum sehr nahe steht, Lane Fox (1986: 525f.); zur Rechtslektüre: Lane Fox (1986: 764, Anm. 30). 84 Vgl. etwa die beinahe identischen Angaben in Dig. 1.18.13 (Ulpian) zu den verschiedenen Aufgabenbereiche eines Provinzstatthalters, die der Sicherung der Ordnung innerhalb der Provinz dienen: Congruit bono et gravi praesidi curare, ut pacata atque quieta provincia sit quam regit. Quod non difficile obtinebit, si sollicite agat, ut malis hominibus provincia careat eosque conquirat: nam et sacrilegos latrones plagiarios fures conquirere debet et prout quisque deliquerit in eum animadvertere, receptoresque eorum coercere, sine quibus latro diutius latere non potest. 85 Winter (1988: 222 und Anm. 4) mit weiterführender Literatur. 86 Zu den Zuständen früherer Jahrhunderte s. a. Dyson (1975: 138ff.). Wie stark das direkte Betroffensein sich in den Quellen ausdrückt, zeigt sich auch in der Analyse der Eingaben von Provinzialen als Hilfegesuche in Krisenzeiten beim römischen Kaiser oder staatlichen Funktionsträgern; allg. Hermann (1990).

335

Limes XVIII Zentralregierung.87

Bellen H. 1998 Grundzüge der Römischen Geschichte. II: Die Kaiserzeit von Augustus bis Diocletian. (Darmstadt). Berthold H. 1988 Mundus senescens? Literatur und ihre Inhalte in der Krisenzeit. In M. Wissemann (ed.) Roma renascens. Festschrift I. Opel. (Frankfurt a. M.): 38-51. Christol M. 1975 Les règnes de Valérien et de Gallien (253-268): travaux d’ensemble, questions chronologiques. In H. Temporini & W. Haase (ed.) Aufstieg und Niedergang der Römischen Welt II 7.1. (Berlin): 803-827. Crouzel H. 1969 Grégoire le Thaumaturge. Remerciement à Origène. Sources Chrétiennes 148. (Paris). Crouzel H. 1983 s.v. Gregor I (Gregor der Wundertäter). Rivista di Archeologia Cristina 12: 782. De Blois L. 1984 The third century crisis and the Greek elite in the Roman empire. Historia 33: 358-377. Demougeot E. 1969 La formation de l’Europe et les invasions barbares. Des origines germaniques à l'avènement de Dioclétien. (Paris). Doi M. 1989 Bagaudes movement and German invasion. Klio 71: 344-352. Dölger F.J. 1950 Christliche Grundbesitzer und heidnische Landarbeiter. AChr 6: 297ff. Dräseke J. 1881 Der kanonische Brief des Gregorios von Neocäsarea. Jahrbücher für Protestantische Theologie 7: 724-756. Dyson S.L. 1975 Native revolt patterns in Roman empire. In H. Temporini & W. Haase (ed.) Aufstieg und Niedergang der Römischen Welt II.3. (Berlin): 138-175. Fredouille J.-C. 1981 s.v. Götzendienst. Rivista di Archeologia Cristiana 11: 869-877. Frend W.H.C. 1967 Martyrdom and persecution in the early church. A study of a conflict from the Maccabees to Donatus. (New York). Frend W.H.C. 1976 The missions of the early church 180-700 A.D. In W.H.C. Frend Religion popular and unpopular in the early Christian centuries. (London). Frend W.H.C. 1980 The failure of the persecutions in the Roman empire. In W.H.C. Frend Town and country in the early Christian centuries. (London): 263-287. Frend W.H.C. 1984 The rise of Christianity. (Philadelphia). Gajdukevič V.F. 1971 Das Bosporanische Reich. (Berlin). Grotz J. 1955 Die Entwicklung der Bußstufen in der vornizänischen Kirche. (Freiburg). Halfmann H. 1986 Itinera Principum. Geschichte und Typologie der Kaiserreisen im Römischen Reich. (Stuttgart). Heather P.J. 1996 The Goths. (Oxford). Heather P.J. & Matthews J. 1991 The Goths in the fourth century. (Liverpool). Hermann P. 1990 Hilferufe aus römischen Provinzen. Ein Aspekt der Krise des römischen Reiches im 3. Jhdt n. Chr., SB Jungiusgesellschaft Hamburg 8,4. (Hamburg). Herter H. & Herrmann A. 1957 s.v. Dirne. Rivista di Archeologia Cristiana 3: 1190-1212. Kettenhofen E. 1982 Die römisch persischen Kriege des 3. Jahrhunderts n. Chr. nach der Inschrift Šahpuhrs I. an der Ka‘be-ye Zartošt (ŠKZ). TAVO Beihefte B 55. (Wiesbaden). Kienast D. 1990 Römische Kaisertabelle. Grundzüge einer römischen Kaiserchronologie. (Darmstadt). Kissel T. 1995 Untersuchungen zur Logistik des römischen Heeres in den Provinzen des griechischen Ostens (27 v. Chr. - 235 n. Chr.). Pharos VI. (St. Katharinen). Klingshirn W. 1985 Charity and power: Caesarius of Arles and the ransoming of captives in sub-Roman Gaul. Journal of Roman Studies 75: 183-203.

Die anarchischen Zustände während der Einfälle haben die Handlungsfähigkeit und Struktur der Kirche nicht beeinträchtigt (Strobel 1993: 291). Es ist durchaus vorstellbar, daß Erfolge in der Bewältigung der Folgen der Goteneinfälle und die Wiedereingliederungspolitik eines Gregorios durchaus die Ausbreitung und Verfestigung des Christenverfolgung und Christentums förderten.88 Barbarenzüge lösten, verstärkt noch durch das Nachlassen der Kontrolle durch die römische Zentralmacht, auch soziale Verschiebungen aus, wie am Beispiel der Großgrundbesitzer zu sehen ist. In dieser Situation konnten so besonders für die christlichen Einwohner dieser Grenzprovinz neue Identifikationsmuster entstehen. Durch die Bedrohung von Innen und Außen wurde die Gemeinde zusammengeschweißt. Gleichzeitig fielen der sich formierenden Kirche immer mehr Bereiche des alltäglichen Lebens als moralische Instanz zu. Der kanonische Brief stellt somit – und das ist nicht nur Topos – ein gewichtiges Zeugnis auch für die stabilisierende Kraft einer sich besonders stark an den Rändern der römischen Welt entwickelnden Institution dar. Literatur Abramowski L. 1976 Das Bekenntnis des Gregor Thaumaturgus bei Gregor von Nyssa und das Problem seiner Echtheit. Zeitschrift für Kirchengeschichte 87: 145-166. Alföldi A. 1967a Die Hauptereignisse der Jahre 253-261 n.Chr. im Orient im Spiegel der Münzprägung. In A. Alföldi Studien zur Geschichte der Weltkrise des 3. Jahrhunderts nach Christus. (Darmstadt): 123-154. Alföldi A. 1967b Zu den Christenverfolgungen in der Mitte des 3. Jahrhunderts. In A. Alföldi Studien zur Geschichte der Weltkrise des 3. Jahrhunderts nach Christus. (Darmstadt): 285-311. Alföldi A. 1967c Die Bewegungen der dakischen und germanischen Völker am Pontus, an der Donau und am Rhein. In A. Alföldi Studien zur Geschichte der Weltkrise des 3. Jahrhunderts nach Christus. (Darmstadt): 312-341. Alföldy G. 1973 Der heilige Cyprian und die Krise des Römischen Reiches. Historia 22: 479-502. Alföldy G. 1975 Historisches Bewußtsein während der Krise des 3. Jahrhunderts. In G. Alföldy u.a. (ed.) Krisen in der Antike. Bewußtsein und Bewältigung. Bochumer Historische Studien 13: (Düsseldorf): 112-132. Ball W. 2000 Rome in the east. The transformation of an empire. (London). Barthelink G.J.M. 1970 Mundus senescens. Hermeneus 42: 9198. 87 Nachdem das Christentum Staatsreligion geworden war, konnte es vorkommen, daß die Kirche, und dabei speziell die Bischöfe beauftragt wurden, für die Überwachung der kaiserlichen Anordnungen Sorge zu tragen. Cod. Iust. 8.50.20 (= Cod. Theod. 5.5.2) des Jahres 408 n. Chr. betrifft sogar genau den Fall, den Gregorios in ep. can. 7 anspricht, nur daß es bei diesem Gesetz nicht nur um Christen geht. Cod. Theod. 5.7.2; vgl. auch Cod. Iust. 1.4.11. 88 Vgl. Liebeschütz (1979: 305) dafür, daß die Krisenhaftigkeit des 3. Jhs. zur Ausbreitung des Christentums geführt hat; dagegen Strobel (1993: 328 Anm. 204).

336

Jeorgios Martin Beyer: i(/na ... au)toi\ a)/lloij Bora/doi kai\ Go/tqoi gi/nwntai. Gregorios Thaumaturgos Kolb F. 1977 Wirtschaftliche und soziale Konflikte im Römischen Reich des 3. Jahrhunderts n. Chr. In Bonner Festgabe Johannes Straub. (Bonn): 277-295. Lane Fox R. 1986 Pagans and Christians. (London). Leven K.-H. 1988 Zur Polemik des Zosimos. In M. Wissemann (ed.) Roma renascens. Festschrift I. Opelt. (Frankfurt a. M.): 177-197. Liebeschütz J.H.W.G 1979 Continuity and change in Roman religion. (Oxford). MacMullen R. 1984 Christianizing the Roman empire (A.D. 100-400). (New Haven). Markschies C. 1997 Zwischen den Welten wandern. Strukturen des antiken Christentums. (Frankfurt a. M). Millar F. 1969 P. Herennius Dexippus: The Greek world and the third-century invasions. Journal of Roman Studies 59: 12-29. Millar F. 1971 Paul of Samosata, Zenobia and Aurelian: The church, local culture and political allegiance in third-century Syria. Journal of Roman Studies 61: 1-17. Millar F. 1992 The emperor in the Roman world, 31 BC - AD 337. (London). Mitchell S. 1989 The siege of Cremna A.D. 278. In D.H. French & C.S. Lightfoot (edd.) The eastern frontier of the Roman empire I. British Institute of Archaeology at Ankara Monograph No. 11, British Archaeological Reports Int. Ser. 553(i). (Oxford): 311-328. Mitchell S. 1993a Anatolia: land, men and gods in Asia Minor I: The Celts and the impact of Roman rule. (Oxford). Mitchell S. 1993b Anatolia: land, men and gods in Asia Minor II: The rise of the church. (Oxford). Mordzejewski J. 1971 Grégoire le Thaumatourge et le droit romain. À propos d’une édition récente. Revue historique de droit francaise et étrangèr 49: 313-324. Pekáry T. 1962 Bemerkungen zur Chronologie des Jahrzehnts 250-260 n.Chr. Historia 11: 123-128. Û Νεοκαισαρε²ας Phouskas K.M. 1969 Γρηγόριος Ñπίσκοπος, Û Θαυµατουργός (Ca. 211/3 - 270/5). (Diss. University of Athens). Poschmann B. 1940 Paenitentia secunda. Die kirchliche Buße im Ältesten Christentum bis Cyprian und Origenes. Eine dogmengeschichtliche Untersuchung. Theophaneia 1. (Bonn). Riedinger R. 1981 Das Bekenntnis des Gregor Thaumaturgus bei Sophronius von Jerusalem und Macarius von Antiochia. Zeitschrift für Kirchengeschichte 92: 311314. Robert L. 1980 A travers l’Asie Mineure. Poétes et prosateurs, monnaies grecques, voyageurs et géographie. Bibliothèque des Écoles Françaises d’Athènes et de Rome 239. (Paris). Ryssel V. 1880 Gregorius Thaumaturgus, Sein Leben und seine Schriften. (Leipzig). Ryssel V. 1894 Eine syrische Lebensgeschichte des Gregorius Thaumaturgus. Nach cod. Mus. Brit. syr. add. 14648 aus dem Syrischen übersetzt. Theologische Zeitschrift aus der Schweiz 11: 228-254. Sage M.M. 1975 Cyprian. Patristic Monograph Series 1. (Cambridge, Mass.). Salamon M. 1971 The chronology of the Gothic incursions into Asia Minor in the IIIrd century AD. Eos 59: 109-139. Saumagne C. 1962 La persécution de Dèce en Afrique d'après la correspondance de S. Cyprien. Byzantion 32: 1ff. Schwarte K.-H. 1989 Die Christengesetze Valerians. In W. Eck (ed.) Religion und Gesellschaft in der römischen

Kaiserzeit. (Köln): 103-163. R. 1995 Die Religionspolitik des Kaisers Decius. Anatomie einer Christenverfolgung. Europäische Hochschulschriften III 617. (Frankfurt a. M.). Shaw B.D. 1984 Bandits in the Roman empire. Past and Present 105: 3-51. Slusser M. 1998 St. Gregory Thaumaturgus. Life and Works. The Fathers of the Church. (Washington DC). Ste. Croix G.E.M. de 1981 The class struggle in the ancient Greek world. (London). Stoll O. 1998 Ordinatus Architectus. Römische Militärarchitekten und ihre bedeutung für den Technologietransfer. In L. Schumacher (ed.) Religion – Wirtschaft – Technik. Althistorische Beiträge zur Entstehung neuer Strukturmuster im historischen Raum Nordafrika/Kleinasien/Syrien. Mainzer Althistorische Studien 1. (St. Katharinen): 203-271. Strobel K. 1993 Das Imperium Romanum im '3. Jahrhundert'. Modell einer historischen Krise? Zur Frage mentaler Strukturen breiterer Bevölkerungsschichten in der Zeit von Marc Aurel bis zum Ausgang des 3. Jh. n. Chr. Historia Einzelschriften 75. (Stuttgart). Telfer W. 1930 The Latin Life of St. Gregory Thaumaturgus. Journal of Theological Studies 31: 142-155 & 354-363. Telfer W. 1936 The Cultus of St. Gregory Thaumaturgus, Harvard TheologicalReview 29: 225-344. Thompson E.A. 1981 Bauernaufstände im spätantiken römischen Gallien und Spanien. In H. Schneider (ed.) Sozial- und Wirtschaftsgeschichte der römischen Kaiserzeit. Wege der Forschung 552. (Darmstadt): 2947. Van Dam R. 1902 Hagiography and history: The Life of Gregory Thaumaturgus. Classical Antiquity 1: 272-308. Walser G. & Pekáry T. 1962 Die Krise des Römischen Reiches. Bericht über die Forschungen zur Geschichte des 3. Jahrhunderts (193-284 n. Chr.) von 1939 bis 1959. (Berlin). Winter E. 1988 Die Sasanidisch-römischen Friedensverträge des 3. Jahrhunderts n. Chr. - ein Beitrag zum Verständnis der außenpolitischen Beziehungen zwischen den beiden Großmächten. Europäische Hochschulschriften III 350. (Frankfurt a. M.). Wolfram H. 1990 Die Goten. Von den Anfängen bis zur Mitte des sechsten Jahrhunderts, Entwurf einer historischen Ethnographie. (München). Seliger

337

Limes XVIII

Fig. 1. Black Sea Region C3rd AD.

338

Excavation of a new Roman base near Balaklava in the Crimea (1992-1999) Oleg Savelya South-west and southern Crimea (Taurika), according to ancient writers, were not parts of the Roman empire. However Roman interests were represented on the coast of the Black Sea by the placing of troop contingents from Moesia. From AD86 Moesia was divided into two provinces; Upper and Lower Moesia, and control of the situation in Scythia and Taurika was assigned to the governor of Lower Moesia and his provincial garrison. Up until 1992 the established facts fixed two Roman bases in Taurika in the time of the principate: at Chersonesus (modern Sevastopol) and on the Ai-Todar cape (modern Large Yalta). In the second half of the C2nd AD, the headquarters of the Roman military contingent was placed in the so-called citadel in south-east Chersonesus. The area of the citadel was less than 0.5ha. There is also evidence of ships of the Moesian fleet in the bay at Chersonesus. This flotilla provided security in the north-western waters of the Black Sea and connected the Roman bases in Taurika with the west Pontic ports. In 1992 the remains of an obscure Roman site were opened on the northern outskirts of Balaklava (18kms south-east of Chersonesus and approximately 1.5kms north of Balaklava Bay - Sumbolon Limen - the Bay of Symbols: Strabo, Pliny and Arrian). As a result of these excavations, under deposits of the C14th - C19th, the eastern corner of a 9 roomed building of Roman date was opened. Four of the rooms (nos. 1, 3, 4 and 5), 7m x 14m, were completely excavated and the other five (nos. 2, 6 - 9) partly so. The building was apparently destroyed by a fire during repair work. The latest date from the coins found as a hoard consisted 57 denarii minted in 223. In the fallen roof more than 60 tegulae stamped V(exillatio) e(xercitus) M(oesiue) I(nferioris), 1 V(exillatio) e(xercitus) M(oesiue), 5 Le(gio) XI Cl(audia) and 2 examples with the legend CAI were collected. This building of the C2nd - early C3rd AD with traces of reconstructions and repairs obviously had an unusual history. In the wall masonry burnt tables and units from more ancient facilities were re-used. The question of the function of this building remains open. Between 1996 and 1999 more of the architecture was exposed. The character of this, because of its fragmentary nature, has remained vague but a sacred site - a temenos - with a temple dedicated to Jupiter Dolichenus was found. Thanks to a dedication on an altar of a statute to Hercules, the temple was dated to the reign of Antonius Pius (138-161), and the time of the vexillations from Lower Moesia,. Commanding Lower Moesia at this time was the military tribune of legio I Italica but the reconstruction of the temple was by a centurion of the same legion who was probably the commander of the garrison. The temple was destroyed by a fire. Its destruction can be dated to the early of the C3rd. The ruined temple was then re-used as a sanctuary which was then destroyed by a fire in the mid-C3rd AD (223+). The causes of these catastrophes are unknown and there is no evidence of crises in Taurika at these times. Likewise the nature and scale of the building work in the territory between 223 and the middle of the second half of the C3rd AD is not obvious. No doubt this was no ordinary post. In the mid-C2nd – early C3rd the garrison was commanded by a centurion of the legio I Italica Novius Ulpianus and a centurion of the XI Claudia Antonius Proculus, most likely a praepositus. Obviously a contingent of the vexillations from Lower Moesia was billeted at one more than one base in the region of modern Balaklava. It is possible that control over the approaches to Chersonesus and Balaklava Bay was protected by this garrison and those in other posts. This harbour was completely protected from Cape Chersonesus to Feodosiya.

the Moesian fleet in Chersonesus harbour. It provided security for the north-western waters of the Black Sea and connected Roman bases in Taurica with the west Pontic ports.2

The south-western and southern Crimea (Taurica to ancient authors) at the time of the principate (C1st – C3rd AD) was not incorporated into the Roman empire, but was within the sphere of its interest in the Black Sea. The defence of Roman interests was laid on contingents brought into Taurica from Moesia (Fig. 1). Beginning from AD86, after the division of Moesia into the two provinces of Lower and Upper Moesia, the situation in Scythia and Taurica was controlled by the imperial governor of Lower Moesia, who was the commander of the province’s garrison.1

Traces of a sanctuary to the Thracian gods, isolated and barely legible Latin stamps on the tiles found on the Heracleum peninsula, fragments of a tombstone with a Latin inscription from the northern side of Sebastopol, and part of some structure of the C2nd – C3rd AD with 11 stamps LEXICL uncovered at the site of Alma-Kermen allow us to suppose that Roman troops did not limit their presence in Taurica only to the Chersonesus fortress and the fortress on Ai-Todor cape. These finds allow us to speculate on: (i) the stationing of Roman troops in Taurica and about the Tauric limes in general; (ii) Roman communications in Taurica; and (iii) the territorium legionis and the organisation and supply of the occupying

Up until 1992 only two Roman strong points dating to the principate were known in Taurica: in Chersonesus (modern Sebastopol) and on Ai-Todor cape (modern Great Yalta). In the second half of the C2nd AD the headquarters of Roman troops in Taurica was housed in the so-called “citadel” in the south-eastern part of Chersonesus. The area of the citadel is less than 0.5ha. There is some evidence of 1 Зубарь В.М. Херсонес Таврический и Римская. (Киев 1994): c.57 и сл; T. Sarnovski, & O.J. Savelya Balaklava, Römische Militärstation und Heiligtum des Iupiter Dolichenus. (Warschau, 2000).

2

Зубарь В.М. Ук, Соч., с.55-57; T. Sarnovski, & O.J. Savelya Balaklava, Römische Militärstation und Heiligtum des Iupiter Dolichenus. (Warschau, 2000): 23.

339

Limes XVIII contingents and so on.3 All these hypotheses must remain speculative because the data we have is not enough to reach any convincing conclusions. For the lack of facts the scheme of distribution of Roman troops in Taurica remains unclear.

On the east it adjoins the hill where the Roman camp appears to have been perched. The temenos area was apparently entered through doorways in the north-eastern wall and close to the southern corner of the south-eastern wall of the sacred area. This area was probably originally used as a sanctuary open to the sky. This phase of its use (Phase 1) probably corresponds with two hearths near the south-western wall of its enclosure as well as with hearths and traces of bonfires under the floor level of the naos of the Jupiter Dolichenus temple and in the pronaos and court of the temenos (Fig. 3).

In 1992 a new Roman strong point was uncovered on the northern outskirts of Balaklava. It lies on a hill approximately 1.5kms north of Balaklava Bay (Fig. 6) and old Balaklava itself (Symbalon Lymen, the Bay of Symbols mentioned by Strabo, Pliny and Arrian: cf. Figs. 1b & 1c).

Abutting the western and south-western part of the enclosure the naos walls and elements of the pronaos of the Jupiter temple were erected (Phases I-IV). According to the reconstruction by the architects N.A. Andrushchencko and T. Bazhanova (Kiev) it was a templum in antis of the Ionic order, measuring inside roughly 11.8 x 5.8m with its façade facing the south-east. The reconstructed height of the temple is c.4.7m. All the temple furnishings (altars, statues, reliefs and so on) are carved from fine-grained white limestone quarried in Inkerman, 10kms north of Balaklava (Figs. 4 & 5)..

In the course of the 1992 excavations the eastern corner of a Roman building overlain by C14th-C19th cultural layers was revealed. It consisted of 9 rooms. Four of them (1, 3, 4 & 5) have been uncovered completely and five partially (Fig. 2). The building was destroyed by a fire just at a time it was being repaired. The latest coin among a hoard of 57 denarii found here was minted in AD223. From the collapsed tiled roof more than 60 specimens of tegulae with the stamps V(exillaio) e(xercitus) M(oesiue) I(nferioris), 1 - V(exillaio) e(xercitus) M(oesiue), 5 Le(gio) XI Cl(audia) and two nominative ones with a legend CAI were recovered. A structure of the C2nd, firstthird of the C3rd AD shows traces of repairs and reconstructions and was erected in place of an earlier building of the C4th/C3rd BC through the C1st AD. The slabs and blocks with traces of burning taken out of earlier buildings were included in the masonry of its walls. The details of this earlier structure remained beyond the limits of the excavation and so do not allow us to offer any suggestions concerning the functions of this structure.4

The interior walls of the Phase II temple naos were clad with plaster painted black, red and green to give the impression of marble and were decorated with cornices and plinths. The remains of the collapsed Phase II wall decorations together with the building rubbish were found in the filling of the podiums in the naos of the Phase III temple. The remains of the hearth in the centre of the naos are probably contemporary with the second phase too (Room A). In this phase there are not any traces of fire and destruction of the temple. The reasons and the time for its disuse and decline remain unknown.

Between 1996-1999, 170m north-east of the 1992 excavation part of another architectural structure was explored as well as and a sacred area – a temenos with a temple dedicated to Iovi Optimo Maximo Dolicheni. The excavations were conducted by an expedition of the National Preserve ‘Chersonesus of Taurica’ (Ukraine) in collaboration with the Institute of Archaeology of Warsaw University (Poland) (1997-1999), and with the contribution of Moscow Pedagogical University (1996) and the Academic Gymnasium of St Petersburg University (Russia; 1997-1999). The character of the first building remains unclear because of its fragmentary condition.

Judging from the dedication on the base of a statue of Heracles found in the temple (Fig. 11), the structure was restored in the reign of the emperors Antoninus Pius and Marcus Aurelius Caesar (138-161) by a vexillation from Lower Moesia under the command of a military tribune of the legio I Italica, Antonius Valentus (Phase III). This temple was equal in size to the earlier one. The renewal of the temple was directed by a centurion of the same legion, Novius Ulpianus, probably the praepositus of the vexillation and commander of the garrison.

The temenos, quadrangular in plan, is enclosed by the walls made of sandstone slabs measuring c.17 x 12 x 16.5 x 14m and 0.75-0.8m in thickness with its corners oriented to the cardinal points. For the sacred area a deep hollow levelled with an artificial earthen embankment was used.

A passage about 1.6m wide led from the naos entrance to its north-western wall where the sculpture of an eagle (Fig. 15), the Heracles statue (Fig. 16) and other dedications were put. It was flanked with two rows of stone tables 0.5m high. Some of them have the dedication I(ovi) O(ptimo) M(aximo) D(olicheni). The stone table legs (6 in each row of tables) were decorated with relief depictions of bulls heads, an eagle, a rosette inside a wreath, Minerva and one of the Castors (?). The gaps between the table rows and longitudinal naos walls were filled with tightly packed building rubbish and clay and formed pise podiumbrackets supported with tegulae placed on their ribs (Figs. 5, 8, 9, 13, 17 & 18-22).

3

Зубарь В.М. Ук, Соч., с.65 и сл; T. Sarnovski, & O.J. Savelya Balaklava, Römische Militärstation und Heiligtum des Iupiter Dolichenus. (Warschau, 2000): 23. 4 Савеля О.Я. Некоторьіе результатьі работ Севастопольсой археологической зкспедиции в округе Херсонеса в 1990-1995 гг. Херсонесский сборник, Севастополь, 1997, c.21; T. Sarnovski & O.J. Savelya Balaklava, Römische Militärstation und Heiligtum des Iupiter Dolichenus. (Warschau, 2000): 24.

340

Oleg Savelya: Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)

Naos A was entered through a pronaos (vestibulum) a twocolumned portico in antis about 5.8m wide and 2.5m deep. The portico appears to have been flanked by an altar dedicated IOM (Fig. 14 - its fragments were found to the right of the portico) and an altar dedicated to Heracles (fragments of which were found in the masonry of the wall of a later sanctuary). The style of reliefs and inscriptions shows the hands of at least three masters. These differences may reflect some chronological nuances within the second half of the C2nd AD (Phase II – Fig. 10).

The defence system and the configuration of the Roman base in Balaklava has still not been defined. But there is no doubt that it was a conventional strong point. In the midC2nd and the first-third of the C3rd AD its garrison was commanded by a centurion of the legio I Italica, Novius Ulpianus, and a centurion of the legio XI Claudia, Antonius Proculus, who most likely had the title of praepositus of the vexillation. Now it is clear that the contingent of the Lower Moesian vexillation in the Chersonesus region was quartered not only in Chersonesus itself but in at least one other base located in the Balaklava area. The garrison of this and some other strong points controlled the approaches to Heracleum peninsula through the Balaklava valley and along Balaklava Bay - Symbolon Port, the only defended harbour on the coastline from the Chersonesus cape to Theodosia.

The temple was destroyed by fire that can be dated by the latest coins recovered among the temple debris – a denarius of Septimius Severus minted in 197 and a Chersonesus dupondius of 192-211 - to the early C3rd AD. The temple ruins were roughly levelled, the north-eastern antis wall was rebuilt and from then on served as the enclosure of the sanctuary open to the sky. The part of the north-western wall of the collapsed temple, the northeastern temenos wall and south-eastern naos walls were demolished. The sacred area was extended to the north and north-east and acquired broken outlines. The south-eastern entrance of the sanctuary has survived. The sacred area could now be entered through a northern doorway as well. This entrance led to a complex of household structures that served the sanctuary (Figs. 7 & 8).

The date for the appearance of the base near Balaklava has still to be determined. The renewal of the Jupiter Dolichenus temple under Antonius Pius and Marcus Aurelius allows us to suppose that the temple and therefore the Roman base must have been founded earlier. The 1992 excavations showed that the Roman strong point had been founded on the site of Hellenistic settlement. In the area of the 1996-1999 excavations the traces of at least two fires and destructions were fixed. The first one soon followed with intensive reconstruction work and is dated to 197-211. The second one that destroyed all the structures of the Roman strong point uncovered in the 1992-1999 seasons is datable to the time soon after 223. The reasons for these catastrophes are likely to be explained by some poorly known crisis in Taurica. The excavations have not still clarified the character and the scale of the structures built on the territory between 223 and the mid-second half of the C3rd AD.

The excavations have revealed the remains of one-roomed Structure D and fragments of one more room erected over the remains of the burnt temple. These are considered to have been built to serve the needs of the sanctuary. Stone troughs, 0.3-0.5m in width, were added to the exterior side of three walls of the D structure. They were filled with animal and bird bones and oyster shells. Millstones, hearths in the yard, a pithos set into the naos podium of the temple and some other details allow to think that drinking and meals were the most important part of the ceremonies that took place in the sanctuary. The material from the temple ruins was widely reused in the sanctuary structures. It is possible to suppose that when the temple was rebuilt as a sanctuary the altar dedicated to Vulkan appeared in the niche for altars and dedications of the temple naos. It was put there by the centurion of the legio XI Claudia, Antonius Proculus, who is likely to have been the garrison commander and responsible for the sanctuary’s reconstruction (Fig. 12). The sanctuary structures were burnt no later than the beginning of the second half of the C3rd AD. The time of its destruction (223) is indirectly provided by the coins of the hoard found in 1992 not far away. With this event there are no further traces of building activity on the area until the early C14th. But judging from the find of a coin of Galerius it is likely that the temple and sanctuary ruins continued to be honoured and visited from time to time. Some building remains erected on the sanctuary place appear to date to the second half of the C3rd AD but they cannot be interpreted because of their fragmentary condition. 341

Limes XVIII

Fig. 1. The northern coast of the Black Sea: A - The main bases of the Roman vexillation; B - The Hercalean peninsula; C - The neighbourhood of Balaklava.

Fig. 2. Kadykovka. The excavation 1991-1992.

342

Oleg Savelya: Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)

Fig. 3. Kadykovka. Plan of the Dolichenum.

343

Limes XVIII

Fig. 4. The Temple of Jupiter Dolichenus. Reconstruction by N. Andruschenko

Fig. 5. Decoration inside the temple. Reconstruction by N. Andruschenko

Fig. 6. Balaklava Bay.

344

Oleg Savelya: Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)

Fig. 7. The temenos and Temple of Jupiter Dolichenus (south-eastern view).

Fig. 8. The temenos and Temple of Jupiter Dolichenus (north-western view).

Fig. 9. The interior of the Temple of Jupiter Dolichenus.

345

Limes XVIII

Fig. 10. A fragment of the altar to Hercules.

Fig. 12. The altar of Vulcanus.

Fig. 11. The base of a statue to Hercules.

Fig. 13. A fragment of the bench (kline).

346

Oleg Savelya: Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)

Fig. 14. A fragment of the altar of Jupiter.

Fig. 16. The statue of Hercules.

Fig. 15. A statue base with the remains of an eagle’s talons.

Fig. 17. Fragment of the table with images of the heads of Dionysus and the Moon.

347

Limes XVIII

Fig. 18. The leg of the tabletrapezophor with the image of a bull’s head.

Fig. 20. The leg of the tabletrapezophor.

Fig. 19. The leg of the tabletrapezophor with the image of an eagle.

Fig. 21. The leg of the tabletrapezophor with the image of Minerva.

348

Fig. 22. Fragment with relief image of a rosette in a garland.

Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis Alan Rushworth This paper focuses on the eastern end of the Severan frontier zone in Mauretania Caesariensis, and, in particular, re-examines the epigraphic evidence from two sites, Taraess (ancient Tatilti) and the city of Auzia, as a basis for some wider observations (Fig. 1).

Epigraphique as AE 1937: 156 and 157 respectively and subsequently cited in works by Pflaum (1960-1961: 703), Salama (1953: 235, 255) and Benseddik (1982: 221, no.108) for example.

The hiberna cohortis IIII Sygambrorum In the later 1930s, this eastern sector of the Severan frontier, known collectively as the nova praetentura (‘the new forward deployment’), was the object of a significant programme of archaeological and epigraphic investigation, conducted principally by Paul Massiera, a leading figure in the French colonial educational establishment (1936; 1937; 1936-1937; 1938-1940; Massiera & Megnin 1939; Christofle 1938).1 This pre-figured the more systematic field survey and epigraphic analysis conducted by Pierre Salama in the 1950s (Salama 1953 & 1955; 1954; 1959; 1966). One of the principal results of this work was the discovery of a Roman fort at Taraess, on the southern slopes of the Titteri range. This was equated with the site of Tatilti, located by the Antonine Itinerary on a very indirect route between Sitifis (Sétif) and Auzia, which looped south then west from Sitifis, around the Hodna mountain range passing through a string of sites in the Hodna basin before veering north-west to cross the Titteri mountains and reach Auzia.

However, re-examination of Massiera’s drawings and descriptions highlights the striking family-likeness of the two inscriptions. Both the form and the recorded height (c.5.5cms) of the lettering is identical on all three blocks. The thickness of the slabs is also very similar (0.34m and 0.29m). This similarity is more than coincidental. Unfortunately, it has not been possible to examine the inscriptions directly, but, by using the stated dimensions of the stones and lettering as a guide, Massiera’s illustrations can be reproduced at the same scale and arranged alongside one another (Fig. 4). This makes it abundantly clear that the two texts actually belong to a single dedication. The fragments dovetail so well that four letters which had each been split in two by the fracture of the original slab can be matched up again, in the process confirming or clarifying the reading of particular sections. The resultant reassembled text is coherent and avoids the unparalleled elements present in the previous readings. The fact that individual letters can be reunited by this method in turn bolsters confidence in the accuracy of Massiera’s representations and provides corroboration of the new reading. Massiera’s description is even more reliable and provides further corroboration. It indicates that the top of the R of hiberna (line 3) and the M of honoratum (line 5) were present in the appropriate places along the lower edge of the first slab, though these were not illustrated (1936: 469). The slight variation in the thickness of the slabs is within the normal range of variation encountered on other fort dedications of the same date (cf. Christofle 1938: 28990). The rear of the slab would simply have been keyed into the wall core and would not therefore have required an even finish. The reassembled dedication is reproduced below, with the suggested new reading.

The description of the surviving remains at Taraess is unfortunately very summary, but gives an impression of a fort, 110m square, with a curtain wall 1.40m thick, built of large ashlar blocks. Gateways were visible on the north and west sides with a third possible one on the south side. However, the bulk of Massiera’s report, published in Revue Africaine for 1936, was devoted to the two fragmentary inscriptions cut on limestone slabs which were found, both outside the east face of the fort. In the first (Fig. 2), Massiera restored a cohors IIII Fl(avia) Chalc]iden[orum eq(uitata) sagittarioru]m, a unit from Chalchis in Syria which is not otherwise attested either in Mauretania Caesariensis or beyond. Two fragments of a second inscription were recovered, the larger being found only a few metres away from the first inscription. The second fragment had been reused in the wall of the village mosque 500m to the north. It seems to mention a hiberna or ‘winter quarters’ and a well-known Severan governor, G. Octavius Pudens, whose work is also recorded at a neighbouring site, Ain Grimidi (CIL VIII 20845), and whose tenure is placed at some point between 198-209 (Fig. 3).

This results in a somewhat different text - a single building inscription commemorating the construction of the curtain wall of a fort or hiberna by Severus and Caracalla through their procurator, Pudens. The most significant difference is that IDEN now forms part of PUDENTEM, thereby eliminating any reference to a cohors IIII Flavia Chalcidenorum equitata sagittaria. This otherwise unattested regiment must therefore be removed from the roster of the army of Mauretania Caesariensis. Instead, the unit for which the winter quarters were actually destined is represented by the letters COH IIII SYG. In the case of the

The two inscriptions were duly registered in Annee 1 For the nova praetentura see Albertini 1928; Salama 1953/1955; 1977: 585-587, 595; Rushworth 1992: 147-150; 1996: 304-305.

349

Limes XVIII

Y and the G only the lower part of the letters survives, but no other restoration is feasible on the basis of Massiera’s illustration. This is immediately recognisable as an abbreviation for the cohors IV Sygambrorum, a wellattested unit of the provincial army, which features in seven other inscriptions (cf. Benseddik 1982: 64-5, 234-5, nos. 174-80). The cohort was raised from the lower German tribe, the Sygambri, and had been stationed in Mauretania Caesariensis since the very early C2nd, at least, since it is listed on the diploma of 107 from Cherchel (CIL VIII 20978 = XVI 56 = ILS 2003). Previously it was impossible to identify any site where the unit was stationed, but it is now apparent that the cohort was transferred to Tatilti during the reign of Septimius Severus, perhaps c.198 (see below), as part of the major southward redeployment. How long it remained there is uncertain, as is the location of the cohort’s earlier bases. Leveau (1977: 297; cf. 1984: 491, 497) has suggested, on the basis of its intriguing name, that Castra Germanorum, cited by Ptolemy (Geographia IV.2) on the coast west of Caesarea, might represent the unit’s station during the Trajanic era, the period to which Ptolemy’s sources belong. The identification is plausible, but no more than that, and in any case it is unclear how long the fort remained in use, as it is not mentioned in any other source and its remains, which probably lay at the mouth of the Oued Damous between Gunugu and Cartennae, have not yet been recognised and investigated.

Pflaum 1960-1961: 703-5, no. 262a; Thomasson 1960: 2: 265-266; Magioncalda 1989: 22, 133-134). He thus represents the only known example of an equestrian governor responsible for census revision within the province he was administering. His tenure falls within the period 198-211. Salama (1953: 235-7) suggested it should be placed c.198 on the basis of the similarity between the Ain Grimidi record and a dated inscription from Boghar (CIL VIII 20847), in which case Pudens’ work would mark the first stage of the Severan advance southward. Indeed it is likely that Pudens was responsible for the establishment of a whole string of forts south of Auzia, including not only Tatilti and Ain Grimidi, but also Boghar (AAA 24: 8) and Ain Touta (AAA 24: 119), where a fourth inscription of very similar type has been found (CIL VIII 9227 = 20846; cf. Salama 1953: 236; 1955: 363). Subsequently these dispositions were reinforced with the construction of a fort at Aras (Christofle 1938: 289-291) and foundation of an oppidum at Usinaza (CIL VIII 9228 = AE 1992: 1925; cf. Benseddik 1992) by P. Aelius Peregrinus, whose gubernatorial tenure in c.201-203 is marked by activity along the entire length of the nova praetentura. Massiera’s own failure to reunite all three fragments of the inscription, may be explained by ideas and debates regarding the policing of the North African frontier which were at their height at the time of the discovery, plus the sequence in which the stones were found. The rôle of oriental archers in the region was seen as especially significant in the 1930s, following highly influential articles by Carcopino (1925; 1933) and Albertini (1931; 1934). These focused attention on units in the neighbouring African-Numidian command such as the Palmyrene and Emesene numeri stationed at El Kantara and Castellum Dimmidi and, most notably in this regard, the cohors I Chalcidenorum equitata. In retrospect this work overemphasised the numerical and tactical importance of Syrian units, but at the time the reading Chalc]iden(orum) on the first of the Taraess texts would apparently have conformed to an expected pattern. Only the smaller of the two other fragments actually fits together with the first slab, and, significantly, this was only found subsequently, reused in the wall of the village mosque 500m away from the two larger stones, by which stage the interpretation of the first slab was doubtless firmly established. In these circumstances the very absence of any reference to a Chalcidenian cohort on the other two stones would explain why it was thought they must belong to a separate inscription. It thus represents an interesting example of the way in which contemporary debates can influence the interpretation of archaeologists’ findings.

The precise details of the imperial titlature on the inscription remain unclear as it is uncertain how long the text was. Severus was probably provided with a full set of honorific titles, as on the inscriptions from neighbouring Boghar, Ain Touta and Ain Grimidi (see below). There does not, however, appear to be sufficient room to fit in Geta’s name and titles after Caracalla’s, which is unusual in North Africa, where, conversely, Geta is often epigraphically accorded the status of Augustus before he formally acceded to that position in 209. It is conceivable that the honorific title, propugnatores imperii, frequently encountered on Severan dedications in Africa (cf. Pflaum 1969: 139-40; Birley 1974), was in this case inscribed over the erased name and titles of Geta, following the latter’s death. The addition of further titles to obscure Geta’s name is evident on other African inscriptions (Pflaum 1969: 13742 provides three examples for instance), although Massiera’s description gives no indication of any such erasure and reinscribing. Pudens is labelled procurator suus a censibus rather than procurator provinciae on the inscription, and was evidently charged with carrying out a census within the province, perhaps preparatory to taxing the population and resources of newly occupied territories. However he was clearly simultaneously functioning as provincial governor. His rôle in the construction of forts here and probably at Ain Grimidi (AAA 24: 155 cf. Salama 1984; 1986), 40kms west-south-west, is sufficient to demonstrate this, and he figures on dedications elsewhere as praeses and procurator Auggg(ustorum) provinciae Mauretaniae Caesariensis (cf.

Defensores provinciae at Auzia The identification of the C3rd base of the cohors IV Sygambrorum has further significant implications. The cohort is mentioned on one of a group of three inscriptions, two of which are well-known, the third less so, found 50kms to the north-west of Taraess, at Auzia. Together these inscriptions shed considerable light on the rôle of the 350

Alan Rushworth: Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

local civic élites in the maintenance of provincial security (Fig. 5).

Primianus’ charge may represent a cavalry detachment from the palatine corps, the equites itemque pedites tam seniores quam iuniores Maurorum (Speidel 1975: 219220).3 This command is listed out of sequence at the end of Primianus’ military offices. Nevertheless it must represent the earliest of those recorded, not the latest. Whilst there is nothing implausible in an equestrian notable assuming command of his local garrison during a crisis, as demonstrated by the case of Q. Gargilius Martialis below, it is inconceivable that someone already elevated to equestrian rank, who had previously commanded a cohort of vigiles at Rome, would subsequently have been content to resume service as an ex decurione alae, even in an emergency. Nor would there have been any need to take such a relatively lowly title. An ad hoc commander of this kind would simply have been designated praepositus . . . ., as was Gargilius Martialis, or, if a more formal title was required, ex tribunus IV vigilum, praepositus etc. The post was listed at this point in the inscription for emphasis, because it represented the most outstanding example of Primianus virtus (cf. Christol 1976: 72, n.4), for it was in this capacity that he earned the title defensor provinciae suae, which features next in the dedication. Thus all three elements, the cavalry decurionate, the praepositura and the provincial defensor, relate to a single outstanding episode in Primianus’ career

Auzia (modern Sour el Ghozlane; AAA 14: 105) occupied a very strategic location on a major crossroads on the route from the coast to the pre-desert steppe of the High Plateaux (Lassère 1981: 317; Fentress 1981: 201-2). It seems to have held a garrison from an early date right through to the late empire (Salama 1977: 582, 593; Rushworth 1996: 310). The settlement had acquired the status of a municipium by the reign of Marcus Aurelius - perhaps under Hadrian - and was further elevated to the rank of colonia, as the Colonia Septimia Aurelia Auziensium, by Septimius Severus (Lassère 1981: 319-320; cf. CIL VIII 9046, 9062). Fentress (1981) has provided an interesting study of the epigraphy from the site, drawing attention to the rather embattled, frontier mentality the inscriptions seem to reflect and the strong ideological attachment of the leading Auzians to their city and its specific culture, situated as it was amidst a rugged and still unruly, tribalised region. The first of the three inscriptions is a dedication erected in honour of one P. Aelius Primianus, by his daughter (probably on the initiative of his recently deceased son), in 255. The text has provoked a good deal of debate as Primianus’ career is not arranged in strict chronological order (cf. Pflaum 1960-1961: 909-912; Jarrett 1963: 217; 1972: 150-151; Speidel 1975: 216-218; Christol 1976: 72, n.4). Nevertheless the rank of the various military offices held by Primianus permits only one possible restoration of his cursus.

The date of Primianus’ military exploit is uncertain. It could well have occurred many years before the dedication was erected in his honour in 255. Some kind of unrest or disturbance - desparatissima turba et factio E[.......]mis - is attested in the region of Auzia in 227 (AE 1966: 597; cf. Pavis d’Escurac-Doisy 1966), whilst T. Licinius Hierocles was governor, and it is feasible that Primianus’ command should be connected with this episode, although a later date cannot be excluded.

Primianus’ career: CIL VIII 9045 = ILS 2766; 16 February 255 P(ublio) Ael(io), P(ublii) f(ilio), Q(uirina tribu), Primiano, | eq(uiti) R(omano), trib(uno) coh(ortis) IIII Syn|g(am)b(rorum), a mil(itiis) primop(ilo), trib(uno) | coh(ortis) IV vig(ilum), ex dec(urione) al(ae) | Thrac(um), p(rae)p(osito) vex(illationi) eq(uitum) | Mauror(um) defenso|ri prov(inciae) suae, dec(urioni) III | col(oniarum) Auz(iensis) et Rusg(uniensis) | et Equiz(etensis). P(ublius) Aeli|us Primus, dec(urio) col(oniae) | Auz(iensis), prius morte | praeventus quam | ded(icaret) pat(ri) piissimo. | Ael(ia) Audif fil(ia), pat(ri), | d(e)d(icavit) XIII Kal(endas) | Mar(tias), (anno) p(rovinciae) CCXVI.

Primianus’ next attested post is that of tribunus cohortis IIII Syngambrorum (sic.). In the light of the revision of AE 1937: 156-157 set out above, this takes on extra significance, since the revised inscription demonstrates that this cohort was stationed at Tatilti, only 50kms south-east of Auzia, at the beginning of the C3rd. If this still held true later in the century, command of the Sygambrian cohort would be the second post Primianus had exercised in the region of Auzia. At the very least it represents another military command exercised within his home province, before his elevation to the equestrian order, a militiis, then promotion to primus pilus and finally to command of the cohors IV vigilum at Rome.

Primianus’ recorded career began with service as a decurion in the ala II Thracum (the most frequently recorded unit in the provincial army). He was then given charge of a detachment of Moorish cavalry (praepositus vexillationem equitum Maurorum), which was very probably stationed at Auzia itself or in the immediate vicinity (see below). It was fairly common in C3rd North Africa for decurions to act as regimental or outpost commanders (praepositi) in this manner.2 The Moors in

Primianus’ local military offices may be compared with his curial status. Remarkably, he is said to be a decurion of no less than three coloniae, Auzia, Rusguniae and Equizeto. Moreover, two of these, Auzia and Equizeto (AAA 15: 91), 95kms to the east, are in the very region where he exercised military command. It is conceivable that

2

For example: T. Iul(ius) Germanus, dec(urio) alae Thrac(um), praep(osito) coh(ortis) Sardoru(m), AE 1932: 31, also CIL VIII 10949 = 21721, Altava; cf. CIL VIII 21720 = ILS 2607, Altava; AE 1909: 104; AE 1909: 104 = ILS 9177, Si Aoun.

3

This would perhaps explain the doubling in eqq(uitum), which might signify vexillatio equitum (seniorum atque iuniorum) Maurorum.

351

Limes XVIII

Primianus exploited the opportunities presented by the military offices to acquire landed property which carried with it membership of curial ordines, or alternatively was rewarded for his services by grateful urban communities. However it was not unparalleled for individuals to exercise civic functions in more than one municipality, as Février (1981: 147) has emphasised, and it is quite possible that Primianus’ lineage was long-established in all three cities. The fact that Gargilus Martialis was also a decurion of both Auzia and Rusguniae, on the coast, suggests there were longstanding historical or commercial links between these two communities. If so this would only emphasise the degree to which Primianus was embedded, as a substantial landowner, in the very region which he was policing, not merely as municipal gendarme, but with control over the military resources of the imperial state.

Auzia (Benseddik 1982: 63-64, 233-234 nos.169-173), suggesting the unit was probably stationed there from the time of its initial formation out of the pedites singulares Pannoniciani under Hadrian or Antoninus Pius.5 We have already seen that the vexillatio equitum Maurorum was most likely operating in the vicinity of Auzia under the command of Aelius Primianus earlier in the century. Under Martialis both units were explicitly said to be outposted in the territory of Auzia (praetendentium in territorio Auziensi). In a perceptive study comparing the commemorative dedication with two other inscriptions (CIL VIII 2615 = ILS 1194; CIL VIII 20827 = ILS 3000), Christol (1976: 7177) has shown that the crisis which prompted Martialis’ appointment occurred between 253-256, several years before his death in 260, the date traditionally applied to these events. Together the three texts illuminate the operations of the Numidian and Mauretanian armies to counter a wave of tribal unrest which spanned a wide arc of mountainous terrain stretching from Milev in north-west Numidia to Auzia in eastern Caesariensis and spilled onto the adjacent plains. The principal tribal coalitions involved were the Bavares centred on the Petite Kabylie or Babors Mountains and the Quinquegentanei in the Grande Kabylie massif. At the heart of the insurgency were a group variously described as the gentiles Fraxinenses, or Faraxen rebellis cum satellitibus suis, evidently a more disparate band united by their allegiance to a charismatic leader, Faraxen, in the classic pattern of African indigenous resistance.

Gargilius Martialis’ epitaph: CIL VIII 9047 = ILS 2767; 26 March 260 [Q(uinto) G]argilio Q(uinti) f(ilio) Q(uirina tribu) Martiali eq(uiti) R(omano), | [pr]aef(ecto) coh(ortis) I Astyrum pr(ovinciae) Britta|[n]iae, trib(uno) co(hortis) Hisp(anorum) pr(ovinciae) Maur(etaniae) Cae(sariensis), | [a] mil(itiis), praep(osito) coh(ortis) sing(ularium) et vex(illationis) | [e]qq(uitum) Mauror(um) in territorio | [A]uziensi praetendentium, | dec(urioni) duarum coll(oniarum) Auzien|sis et Rusguniensis et pat(rono) | prov(inciae), obinsignem in ci|ves amorem et singula|rem erga patriam adfec|tionem et quod eius vir|tute ac vigilantia Fa|raxen rebellis cum sa|tellitibus suis fuerit | captus et interfectus, | ordo col(oniae) Auziensis, | insidiis Bavarum de|cepto p(ecunia) p(ublica) f(ecit) d(ecreto) d(ecurionum)4 VIII kal(endas) | [A]pr(ilis) (anno) pr(ovinciae) CCXXI.

On an altar (CIL VIII 2615 = ILS 1194) set up in the Aesculapaeum at Lambaesis between 253-256, the Numidian legate, C. Macrinius Decianus, celebrated a victorious sweep westward through the mountains. His forces, presumably including legio III Augusta, newly reconstituted after its 15 year disbandment, fought a series of engagements against the Bavares, the Quinquegentanei and finally Faraxen himself. The aim was presumably to flush Faraxen and his followers out of the mountains, deprive them of their wider tribal support and drive them west towards the waiting Mauretanian forces, the anvil to Decianus’ hammer. The close involvement of the Mauretanian army is suggested by the way that key phrases on the Mauretanian dedications describing the victories clearly parallel those employed on the Lambaesis altar. Thus the altar (CIL VIII 20827 = ILS 3000) erected on 4 August 254 at Ain bou Dib (AAA 14:32), 20kms northeast of Auzia, by the governor of Caesariensis, M. Aurelius Vitalis, and one of his cavalry decurions, Ulpius Castus, was dedicated ‘ob barbaros c(a)esos ac fusos’, whilst Decianus celebrated that the ‘Bavaribus .... Quinquegentaneis gentilibus .... item gentilibus Fraxinensibus .... caesis fugatisque’. Similarly the description of Faraxen’s end on the Lambaesis altar ‘capto famosissimo duce eorum’ - is still faintly echoed

The second Auzian notable attested holding a local military command is Q. Gargilius Martialis, again recorded on a commemorative dedication, this time erected in 260 by the ordo coloniae after his death. Martialis was certainly from a prominent Auzian lineage. His father, a veteran, served as flamen perpetuus and chief civic magistrate - curator et dispunctor reipublicae (CIL VIII 20751). Martialis, himself held the required military commands to secure equestrian status a militiis, one of them as tribunus cohortis Hispanorum in his home province. It is not known where this unit was based in the C3rd (or earlier for that matter), but it is must be somewhere in the central or eastern parts of the province as we know the full sequence of regiments stationed along the western praetentura. Like Primianus, he was a decurion of Rusguniae, as well as Auzia. Having completed the normal militiae, Martialis was recalled to arms, after an interval of uncertain duration, to command the garrison at Auzia, the cohors singularium and the vexillatio equitum Maurorum. All the inscriptions mentioning the cohors I Aelia singularium derive from 4 Or f(ecit) d(e)d(icavitque), a formula common on funerary inscriptions at Auzia.

5

For the pedites singulares Pannoniciani, recorded on CIL VIII 21453 from Aquae Calidae (Hammam Righa), see Speidel 1972.

352

Alan Rushworth: Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

several years later on Martialis’ commemoration ‘Faraxen rebellis ... fuerit captus et interfectus’.

North Africa in 297-299. However, there is nothing to specifically connect Capito’s dedication with any of these events and it is quite possible that the appointment of a local notable to military command within his own region was no longer a response to exceptional circumstances, but now a much more routine matter.

This final denouement probably occurred in the vicinity of Auzia, with Martialis and his forces, praetendentium in territorio Auziensi, playing a decisive rôle in Faraxen’s capture. For this Martialis was greatly honoured by his city. Yet he himself was to die some years later, insidiis Bavarum decepto, a reflection of the continuing unrest which afflicted the region, and which eventually led to the appointment c.258-260 of a general with extraordinary powers, M. Cornelius Octavianus, dux per Africam, Numidiam Mauretaniamque.

Discussion and conclusions The identification of the Hiberna cohortis IIII Sygambrorum further amplifies a picture of the municipal elite in one part of Caesariensis taking over local military command during the C3rd. It suggests that, crisis appointments aside, notables such as Primianus were routinely holding military offices rather closer to their home city than might be anticipated. It would be interesting to know where exactly the cohors Hispanorum, commanded by Gargilius Martialis earlier in his career, was based, particularly given the evident breadth of his landholdings in eastern and central Caesariensis.

The final example of this group is provided by an altar dedication made by a third Auzian notable, L. Iulius Capito, in 301 (Fig. 6). The altar was actually found at Ghorfa Ouled Selama (AAA 15:37) 8kms south-east of Auzia, but, like most of the inscriptions at that site, it probably represents spolia originally robbed from Auzia in late antiquity. Capito shares many of the characteristics of his two predecessors. Again he was from a prominent local family - the elaborate verse epitaph (CIL VIII 20758) of his father was also found at the same site - and like them he commanded the local garrison. Like Primianus he had the rank of decurio alarius, but this time whilst serving as praepositus limitis - in charge of a local military district (limes). Since the district is not named it was, in all probability, that centred on Auzia itself, later attested by the Notitia Dignitatum (Occ. XXX 17 - praepositus limitis Audiensis). Capito is thus the direct counterpart of Primianus and Gargilius Martialis before him. The formalisation of these limes commands throughout the North African frontier zone was part of the tetrarchic overhaul of the military and civil administration in region (cf. Di Vita Evrard 1985a: 162-175; Rushworth 1992: 18), but the transformation was probably less dramatic than it appeared.6 The troops at Capito’s disposal very likely belonged to the same units as were commanded by Primianus and Martialis (Rushworth 1992: 13-18, 22-26).

Moreover the close involvement of local urban notables like Primianus and Martialis in the troubles of the midC3rd suggests these conflicts may have had something of the character of a civil war between different African communities and social groups - city versus tribe, Romanised urban élites versus tribal chieftains - with the imperial state intervening on behalf of the cities. The causes of these conflicts remain unclear, but it is possible that the Mauretanian cities had sought to extend their territoria following the southward advance of direct Roman authority under Septimius Severus, provoking a reaction from neighbouring tribal communities anxious to protect their social cohesion and access to land. In this context it is noteworthy that a statue dedication by the Auzienses in honour of the procurator Octavius Pudens describes him as a municeps piissimus, i.e. a fellow-citizen (CIL VIII 9049 = 20737 = ILS 1357). Whilst this might simply be a honorary title, it does raise the possibility that the advance south from Auzia, at the very end of the C2nd, was initiated by a governor who originated from that city (cf. Salama 1953: 237; Pflaum 1960-1961: 704-705). Much more fieldwork, of the kind carried out by Leveau in the hinterland of Caesarea (1984), will be required in the mountains and plains of eastern Mauretania, however, before the relationships between the urban and rural communities in that region can be fully understood.

In one respect, however, Capito’s command was different. Whereas Primianus and Gargilius Martialis both led the Auzian garrison during emergencies, there is nothing to suggest that his was anything other than a normal appointment. There had, to be sure, been severe fighting around Auzia, as throughout the province, 10 years earlier, under the energetic governor Aurelius Litua, which necessitated the restoration of a bridge, belli saevitia destructum (CIL VIII 9041 = ILS 627). The presence of the co-emperor Maximian, himself, was even required in the

For their part, the imperial authorities’ willingness to rely on such local men in the military as well as civil spheres may represent an ad hoc response to the region’s frequent troubles, but must also reflect their confidence in the extreme loyalty of Auzia’s foremost citizens. As yet it is impossible to determine whether this pattern was restricted to Auzia during the C3rd or was more widespread within the province.

6

These small frontier districts commanded by someone of equivalent rank to a regimental commander should not be confused with the larger, regional limes commands, such as the praepositus limitis Tripolitanae, which were under the authority of an officer of lesser procuratorial rank and which are attested in the mid-late C3rd (cf. Rebuffat 1985; Di Vita Evrard 1985a: 166, 168-170). Some small frontier districts, such as the limes Tentheitanus, were designated as early as the mid-C3rd, but they do not have formal commanders (praepositi) until the Diocletianic reforms (Mattingly 1991; Rushworth 1992: 15-20).

By the C4th this reliance on local magnates was particularly evident. However there was a significant difference for now it was the tribal chieftains of the 353

Limes XVIII

Mauretanian and Tripolitanian frontier zones, rather than the urban élite, who were increasingly integrated in local military command and frontier control. The construction of defensible residences, such as the tower-like gsur of the Tripolitanian pre-desert or the larger castles, for want of a better word, in western Caesariensis like Ksar el-Kaoua (Fig. 7: AAA 22: 63; Gsell 1901 I: 102-106; cf. Marchand 1895; Lacave-Laplagne 1911; CIL VIII 9725 = 21531 = ILS 6021), may in some way reflect this process. But more particularly it is marked by the award of local frontier commands or honorary titles, such as ‘tribune’, to individual chieftains, for example those commemorated on the funerary stele at Bir ed-Dreder in the Tripolitanian frontier zone (Buck et al 1983; Mattingly 1987: 85-88; Rushworth 1992: 199-204, 208-229).

tribal unrest is known during the early-mid C4th, though this may simply reflect the survival of evidence. Ultimately, however, Nubel’s death was to spark off a succession crisis which engulfed most of Caesariensis and Sitifensis, requiring the intervention of the imperial field army.

The most striking instance of this phenomenon is the rise of the house of Nubel, documented by Ammianus Marcellinus (XXIX.5) and several inscriptions (cf. Rushworth 1992: 222-229; Matthews 1976: 174-177; 1989: 372-376). This powerful figure is distinguished by his ability to operate comfortably at the highest levels in two worlds. On the one hand Nubel was an extremely wealthy landowner, possessing large estates on the Mitidja Plain around the coastal cities of Rusguniae and Icosium (Algiers). He had sufficient resources to acquire a fragment of the newly discovered True Cross in the mid-C4th and dedicate a church to house it at Rusguniae (CIL VIII 9255 = ILCV 1822, where he is styled Flavius Nuvel). The same dedication reveals he had commanded one of the regional field army units (ex praepositis equitum armigerorum iuniorum). He was probably also the builder and intended occupant of the massive mausoleum at Blad Guitoun at the eastern end of the Mitidja (Fig. 8; AAA 5:54; Gsell 1898; 1901 II: 412-417). On the other hand he is said to have been of tribal origin (Ammianus Marcellinus XXIX.5. 44), and through kinship links, alliances and the tenure of military offices, the tentacles of his influence spread throughout the tribal communities of central and eastern Caesariensis, ‘velut regulus’ in the words of Ammianus. Thus Nubel, and on a lesser scale the other chieftains given offices and titles, were valuable because of their ability to act as intermediaries between tribal communities and the military administration.

AAA = Gsell 1911. Albertini E. 1928 La route-frontière de la Maurétanie Césarienne entre Boghar et Lalla Maghnia. Bulletin de la Société de Géographie et d’Archéologie d’Oran 48: 33-48. Albertini E. 1931 Inscriptions d’El Kantara et de la région. Revue Africaine 72: 193-261. Albertini E. 1934 A propos des numeri syriens de Numidie. Revue Africaine 75: 23-42. Benseddik N. 1982 Les troupes auxiliares de l’armée romaine en Maurétanie Césarienne sous le Haut-Empire. (SNED, Algiers). Benseddik N. 1992 Usinaza (Saneg): un nouveau témoignage de l’activite de P. Aelius Peregrinus sur la praetentura sévèrienne. In A Mastino (ed.) L’Africa romana, 9. Atti del IX convegno di studio Nuoro, 13-15 dicembre 1991. (Sassari): 425-438. Birley A.R. 1974 Septimius Severus, propagator imperii. In D.M. Pippidi (ed.) Actes du IXe Congrès International d’Études sur les Frontières Romaines, Mamaia 1972. (Bucharest): 297-299. Buck D.J., Burns J.R. & Mattingly D.J. 1983 Archaeological sites of the Bir Scedua Basin: Settlements and cemeteries. In G.D.B. Jones & G.W.W. Barker (edd.) The UNESCO Libyan valleys survey IV: The 1981 season. Libyan Studies 14: 42-54. Carcopino J. 1925 La limes de Numidie et sa garde syrienne. Syria 6: 30-57 & 118-149. Carcopino J. 1933 Note complémentaire sur les numeri syriens de Numidie romaine. Syria 14: 20-55. Christofle M. 1938 Rapport sur les travaux de fouilles et consolidations effectuées en 1933, 1934, 1935, 1936 par le service des monuments historiques de l’Algérie. (Algiers). Christol M. 1976 La prosopographie de la province de Numidie de 253 à 260 et la chronologie des révoltes africaines sous le règne de Valérien et de Gallien. Antiquités Africaines 10: 69-77. Di Vita-Evrard G. 1985a L. Volusius Bassus Cerealis, légat du proconsul d’Afrique. T. Claudius Aurelius Aristobulus, et la création de la province de Tripolitaine. In A. Mastino (ed.) L’Africa romana, 2. Atti del II convegno di studio, Sassari 1984. (Sassari): 149-177. Di Vita-Evrard G. 1985b Regio Tripolitana - A reappraisal. In D.J. Buck & D.J. Mattingly (edd.) Town and country in Roman Tripolitania: Papers in honour of Olwen Hackett. British Archaeological Reports Int. Ser. 274. (Oxford): 143-163.

To conclude, the developments at Auzia in the C3rd may reveal one way in which that process originated and also go some way towards explaining why the regional military administration acquired such confidence in the efficacy of employing powerful, but apparently loyal, members of the local élite to defend and police their own province. Bibliography

Whether reliance on such intermediaries was part of a deliberate administrative policy shift is unclear. Conceivably these individuals were considered better able to deal with unrest and obtain tribal communities’ compliance with the demands of the state, without recourse to disruptive punitive action. However, it is at least equally likely that much of the initiative came from members of the tribal élite, seeking the power and prestige which military offices and honorary titles could confer. The growing restrictions placed on the curial classes (cf. Jones 1964: 740ff.), tying them to their respective city councils, may have made it increasingly difficult for decurions to exercise military commands, allowing ambitious chieftains to fill the gap. Whatever its cause, the involvement of such figures possibly had beneficial results for a time. Less 354

Alan Rushworth: Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

Donaldson G. 1985 The praesides provinciae Tripolitaniae - civil administrators and military commanders. In D.J. Buck & D.J. Mattingly (eds) Town and country in Roman Tripolitania: Papers in honour of Olwen Hackett. British Archaeological Reports Int. Ser. 274. (Oxford): 165-177. Fentress E.W.B. 1978 Dii Mauri and Dii Patri. Latomus 37: 507516. Fentress E.W.B. 1981 African building: Money, politics and crisis in Auzia. In A. King & M. Henig (edd.) The Roman west in the third century: Contributions from archaeology and history. British Archaeological Reports Int. Ser. 109 – 2 vols. (Oxford) 1: 199-208. Février P.A. 1981 A propos des troubles de Maurétanie (villes et conflits du IIIe S.). Zeitschrift für Papyrologie und Epigraphik 43: 143-148. Gsell S. 1898 Le mausolée de Blad-Guitoun (fouilles de M. Viré). Comptes-Rendus de l’Académie des Inscriptions et Belles-Lettres (1898): 481-499. Gsell S. 1901 Les monuments antiques de l’Algérie. (Paris, 2 vols). Gsell S. 1903 Observations géographiques sur la révolte de Firmus. Recueil des Notices et Mémoires de la Société archéologique du Département de Constantine 36: 2146. Gsell G. 1911 Atlas archéolgique de l’Algérie. (Paris). Jarrett M.G. 1963 The African contribution to the imperial equestrian service. Historia 12: 209-226 Jarrett M. G. 1972 An album of the equestrians from North Africa in the emperor’s service. Epigraphische Studien 9. (Bonn). Jones A.H.M. 1964 The later Roman empire 284-602: A social, economic and administrative survey. (Oxford). Lacave-Laplagne (sic) 1911 Notes sur quelques ruines romaines relevées dans la Commune-Mixte d’Ammi Moussa. Bulletin de la Société de Géographie et d’Archéologie d’Oran 31: 21-56 & pls. i-vi. Lassère J.M. 1981 La colonia Septimia Aurelia Auziensium. Ktema 6: 317-331. Le Bohec Y. 1989a La troisième légion Auguste. (Paris). Le Bohec Y. 1989b Les unites auxiliaires de l’armée romaine en Afrique Proconsulaire et Numide sous le Haut Empire. (Paris). Leveau P. 1977 Recherches historiques sur une région montagneuse de Maurétanie Césarienne: Des Tigava Castra à la mer. Mélanges d’Archéologie et d’Histoire de l’Ecole Française de Rome, Antiquité 89.1: 257-311. Leveau P. 1984 Caesarea de Maurétanie: Une ville romaine et ses campagnes. (Collections de l’Ecole Française de Rome 70. (Rome). Magioncalda A. 1989 I procuratori-governatori delle due Mauretaniae: un profilo (titolatura e carriere). In M. Christol & A. Magioncalda Studi sui procuratori delle due Mauretaniae. (Dipartimento di Storia - Università degli Studi di Sassari, Sassari): 9-154. Marchand Capt. 1895 Occupation romaine dans la circonscription d’Ammi Moussa. Bulletin de la Société de Géographie et d’Archéologie d’Oran 15: 207-220 (& fold-out map). Massiera P. 1936 La station de Tatilti. Revue Africaine 79.2: 465-476. Massiera P. 1936-1937 Bornes milliares du Hodna occidental. Bulletin Archéologique du Comité des Travaux Historiques et Scientifiques (1936-1937): 302-314. Massiera P. 1937 Le limes sous les Sévères dans le Hodna occidental. Revue Africaine 80: 503-506.

Massiera P. 1938-1940 Nouvelles bornes du Hodna occidental. Bulletin Archéologique du Comité des Travaux Historiques et Scientifiques (1938-1940): 339-344. Massiera P. & Megnin Dr. 1939 La route frontière au IIIe siècle entre Tarmount et Ain Touta. In IVe Congrès de la Fédération des Sociétés Savantes d’Afrique du Nord, Rabat, Morocco 2: 561-562. Matthews J.F. 1976 Mauretania in Ammianus and the Notitia. In R. Goodburn & P. Bartholomew (edd.) Aspects of the Notitia Dignitatum. British Archaeological Reports Int. Ser. 15. (Oxford): 157-186. Matthews J.F. 1989 The Roman empire of Ammianus. (London). Mattingly D.J. 1987 Libyans and the ‘Limes’: Culture and society in Roman Tripolitania. Antiquités Africaines 23: 71-94. Mattingly D.J. 1991 The constructor of Gasr Duib, Numisius Maximus, Trib(unus cohortis I Syrorum sagittariorum). Antiquités Africaines 27: 75-82. Mattingly D.J. 1995 Tripolitania. (London). Pavis d’Escurac-Doisy H. 1966 Un soulèvement en Maurétanie sous Sévère Alexandre. In R. Chevallier (ed.) Mélanges d’archéologie et d’histoire offerts à Andre Piganiol. (Paris): 1191-1204. Pflaum H.G. 1960-1961 Les carrières procuratoriennes équestres sous le Haut-Empire romain. 3 vols. (Paris). Pflaum H.G. 1969 Inscriptions impériales de Sila. Antiquités Africaines 3: 133-144. Rebuffat R. 1985 Le ‘limes’ de Tripolitaine. In D.J. Buck & D.J. Mattingly (edd.) Town and country in Roman Tripolitania: papers in honour of Olwen Hackett. British Archaeological Reports Int. Ser. 274. (Oxford): 127-141. Rushworth A. 1992 Soldiers and tribesmen: the Roman army and tribal society in late imperial Africa. (unpub. PhD thesis, University of Newcastle-upon-Tyne). Rushworth A. 1996 North African deserts and mountains: comparisons and insights. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 297-316. Salama P. 1953 & 1955 Nouveaux témoignages de l’oeuvre des Sévères dans la Maurétanie Césarienne. Libyca 1: 231261 and 3: 329-365. Salama P. 1954 A propos d’une inscription maurétanienne de 346 après JC. Libyca 2: 205-229. Salama P. 1959 Bornes milliaires et problèmes stratégiques du Bas-Empire en Maurétanie. Comptes-Rendus de l’Académie des Inscriptions et Belles-Lettres (1959): 346-354. Salama P. 1966 Occupation de la Maurétanie Césarienne occidentale sous le Bas-Empire romain. In R. Chevallier (ed.) Mélanges d’archéologie et d’histoire offerts à Andre Piganiol. (Paris): 1291-1311. Salama P. 1977 Les déplacements successifs du limes en Maurétanie Césarienne (Essai de synthèse). In J. Fitz (ed.) Limes: Akten des XI Internationalen Limeskongresses, Székestehérvar 1976. (Budapest): 577-595. Salama P. 1984 Masque de parade et casque d’Ain Grimidi: précisions sur le Limes de Maurétanie Césarienne centrale. Bulletin de la Societé nationale des Antiquaires de France (1984): 130-142. Salama P. 1986 Masque de parade et casque d’Ain Grimidi (Maurétanie Césarienne). In C. Unz (ed.) Studien zu den Militärgrenzen Roms III: Vortrage des 13 Internationalen Limeskongresses, Aalen 1983 (Stuttgart): 649-656.

355

Limes XVIII

Speidel M.P. 1972 The Pedites Singulares Pannoniciani in Mauretania. American Journal of Philology 93: 299305. Speidel M.P. 1975 The rise of ethnic units in the Roman imperial army. In H. Temporini (ed.) Aufstieg und Niedergang der römischen Welt II.3 (Berlin): 202-231. Thomasson B. 1960 Die Statthalter der romischen Provinzen Nordafrikas von Augustus bis Diocletianus. 2 vols. (Lund).

356

Alan Rushworth: Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

Fig. 1. Map of eastern Mauretania Caesariensis.

Fig. 2. First fragment (Massiera 1936: 468 = AE 1937: 1

[Imp(eratores) Caes(ares), L(ucius) Septimius Severus et M(arcus) | Aurelius Ant]oninus | [Aug(usti) pr]opagatores | [imperii hiber]na coh(ors or ortis) IIII | [Flavia or Flaviae Chalc]iden(orum) | [eq(uitata) sagittarioru]m |.[ . . .

Fig. 3. Second Inscription (Massiera 1936: 471 = AE 1937: 157). [Imp(erator) Caes(ar), L(ucius) Septimius Severus Pius Pertinax Aug(ustus) Arabic(us) Adiabenic(us) Parthic(us) Maximus, pont(ifex) maximus, trib(uniciae) potest(atis) . ., imp(erator) . ., co(n)s(ul) . . p(ater) p(atriae) et Imp(erator) Caes(ar) M(arcus) Aurelius Antoninus Aug(ustus) et P(ublius)] S[eptimius | Geta Caes(ar)] mur(os) h[ibe]rnac[ul(orum) . .]|s[.]g per G(aium) Octa[vi]um Pu[den]|tem Caesium Ho[n]oratu[m] | proc(uratorem) suum a cen[si]bus.

357

Limes XVIII

Fig. 4 . The reunited inscription. [Imp(erator) Caes(ar) L(ucius) Septimius Severus Pius Pertinax Aug(ustus) ......................... et Imp(erator) Caes(ar) M(arcus) Aurelius Ant]oninus | [Aug(ustus) ......]s [pr]opagatores | [imperii] mur(os) or mur(um) h[ibe]rna(e) coh(ortis) IIII | Syg(ambrorum) per G(aium) Octa[vi]um Puden|tem Caesium Ho[n]oratum | proc(uratorem) suum a cen[si]bus.

Fig. 5. The dedications to P. Aelius Primianus and Q. Gargilius Martialis.

358

Alan Rushworth: Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

Fig. 6 Iulius Capito's dedication CIL VIII 9025; AD 301. Victori|ae Aug(ustae) san|cte deae, L(ucius) | Iulius, [Luc(ii)] | f(ilius), Capito (decurio) alari(u)s, | pr(a)eposi|tus limi|tis, cum | suis o|mnibus | f(ecit) d(edicavitque), | (anno) p(rovinciae) CCLXII.

Fig. 7 Ksar el-Kaoua (after Gsell 1901).

359

Limes XVIII

Fig. 8. Reconstruction of the mausoleum of Blad Guitoun (after Gsell 1898).

360

The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs (Eastern Desert of Egypt) 1998-2000 Steven E. Sidebotham Between winter 1998 and summer 2000 the University of Delaware (USA, co-director S.E. Sidebotham) and Leiden University (The Netherlands, co-director W.Z. Wendrich) conducted six field seasons in and around Berenike and farther afield in the Eastern Desert of Egypt. The work included continued excavations at the Ptolemaic-Roman Red Sea port of Berenike, survey and limited excavations in the immediate environs of the port and survey work in the region unrelated to the Berenike Project. The ultimate goal of this on-going fieldwork is to gain as complete an understanding of the social, economic and political history of this remote area as possible. We also want to relate activities in this hyper-arid desert zone to the rest of Egypt and to the Roman empire as a whole. Three additional excavation seasons at Berenike since the report at the last Limes Congress at Zalău (Romania 1997) have shed much new light on many aspects of the ancient emporium’s physical appearance and history. Excavations have revealed more of the natural and architectural topography of the city and how it changed over the 800 year life of Berenike due to economic circumstances and local geological activities. These include radical physical transformations of the city’s coastline over time and the location and identification of the Ptolemaic industrial quarter. We also have a better understanding of the early Roman customs procedures, and the religious practices and economic activities that took place in the city throughout the Roman period. Much new evidence has come to light on the various ethnic groups - both those from within the Roman empire and those coming from outside - active in Berenike throughout its history. To date our research has identified at least 9 different written languages used by Berenike’s inhabitants. Indian, Sri Lankan and Axumite patterns of trade at the port, for example, are much better understood now than three years ago. Survey work in the Eastern Desert and limited excavations in some of the sites near Berenike have also revealed an elaborate infrastructure of roads, cemeteries, military road stations, water points, gold and emerald mining camps and Christian anchorite settlements. The interaction among these various sites and between these sites and Berenike indicates a vibrant and relatively wellpopulated area of the desert maintained, essentially, by an elaborate logistical support system from both the Red Sea ports and the Nile valley. This effort resulted in a substantial, but immeasurable, degree of prosperity for the local and regional economies, for both private entrepreneurs and the Roman government.

especially about the ethnicity of the inhabitants, their social, religious and economic activities and dietary habits.

Six seasons of fieldwork in south-eastern Egypt by the University of Delaware and Leiden University since the last report in Zalău, Romania in 1997 have added significantly to our knowledge about this region’s archaeology and history in the Ptolemaic-Roman period (Figs. 1-2).1 This fieldwork has included three seasons of excavation at Berenike and its environs and survey in the region during the winters 1998-2000 as well as three seasons of survey work by the University of Delaware in summers 1998-2000.

Periods of activity at Berenike The archaeological record uncovered thus far broadly reflects periods of limited, mainly industrial, activity in the early Ptolemaic period (mid-C3rd to mid-C2nd BC); there is some, but relatively little evidence for the late Ptolemaic era at the site. There followed substantial activity, the zenith of the port, in early Roman times (the JulioClaudian and Flavian era). The archaeological evidence recovered thus far suggests that there was a perceptible decline in the C2nd AD followed by further apparent decay in the C3rd and early C4th.

Excavations at Berenike Most resources have been expended in on-going excavations at Berenike (Fig. 3) and its immediate environs (Fig. 4). Three seasons of fieldwork at Berenike between 1998 and 2000 have recorded significant new data about the topography and architecture of the site, and

From the middle of the C4th and into the C5th the ceramic, numismatic and architectural evidence indicates a renaissance at Berenike with substantial construction and renovation. In many cases this renewed building activity either lay atop or greatly modified earlier structures; in a few areas, especially at the eastern edge of the site, this mid-C4th and C5th construction renaissance lay above sterile soil. The later C5th and early C6th were a final period of decline and, before the middle of the C6th, abandonment of the site.

1

S.E. Sidebotham (University of Delaware) & W.Z. Wendrich (Leiden University/UCLA) co-direct the Berenike Project. Sponsors of the 19982000 seasons were the National Geographic Society, the Netherlands Organization for Scientific Research, Foundation for History, Archaeology and Art History (NWO-SHW), the Samuel H. Kress Foundation, the American Philosophical Society, the Dorot Foundation, the Stichting Berenike Foundation and the University of Delaware. Private donors included (in alphabetical order) Ms. Millie C. Cassidy, Mr. Bruce Gould, Ms. Norma Kershaw and Ms. Carol Maltenfort. The main bibliography on the excavations at Berenike and survey of the Eastern Desert may be found in the Berenike 1994, Berenike 1995, Berenike 1996, Berenike 1997 and Berenike 1998 volumes (Leiden; 1995-2000) and the Berenike 1999 volume (Los Angeles; 2001 forthc.). Other bibliography appears at the end of this paper.

Greatly reduced commercial traffic through Berenike was undoubtedly the reason for the port’s eventual abandonment, but the causes of this decline in trade are not clear at this time and may have resulted from a variety of factors. During the later C5th and early C6th the increased 361

Limes XVIII

power of the Kingdom of Axum in the Red Sea-Indian Ocean trade, and hostilities between Axum and one or more states in south-western Arabia would not have been conducive to Roman commerce and would have led to a decline in the importance of Roman trade in the region (Munro-Hay 1996: 410-412). On-going silting of Berenike’s harbour may also have contributed to the decline in trade through and ultimate abandonment of the port (Harrell 1996: 99-105, 112-125; Sidebotham 1999d: 91-94; Sidebotham & Wendrich 1999b: 447).

only Indian coin (of the second half of the C4th AD) found in excavations anywhere on the site to date both derive from this Christian edifice.2 This numismatic evidence may indicate something of the international character of the Christian community at Berenike in late antiquity. Several city blocks west of the harbour area and between the Christian ecclesiastical edifice and the Serapis temple, excavations have thus far uncovered a number of late Roman structures. Here there was a city quarter replete with multiple storied structures with walls made of coral heads (Fig. 7). Interior walls often preserve an unusual type of construction that the excavators term opus Berenikeum. This technique, a variation of opus Africanum (cf. Adam 1994: 120-122), has a boxed arrangement of ashlars made of locally quarried gypsum/anhydrite enclosing coral heads; such a construction scheme may have provided added strength to the walls. There is no evidence that opus Berenikeum was used for the exterior walls of buildings.

Paradoxically, identifiable and attributable coins excavated from the site do not post-date the late C4th to early C5th (Sidebotham 1995; 1998a; 1999c; 2000a; 200la forthc.; Sidebotham & Seeger 1996). Yet other data (especially ceramic with associated floral and textile remains) indicate that the C5th was a period of robust trade through Berenike. One can only speculate on this apparent dichotomy between numismatic and other evidence. It may indicate an increased reliance on barter in both local, regional and ‘international’ commerce or, for reasons that are not understood at this time, continued use of coinage from earlier periods during the C5th boom.

These buildings also preserve staircases and niches lined with wooden shelves (Fig. 7). Pottery and coins date activity in these edifices to the late C4th to C5th.

Topography of Berenike These buildings appear to be, on the ground floors, centres of commercial or governmental activity. Heavy wear and multiple rebuildings of thresholds at main entrances into the ground floors of the structures indicate a sizeable amount of traffic. Remains suggest a concern for security and close monitoring of visitors. These late Roman buildings have single narrow entrances. A number of copper alloy locking mechanisms found in these edifices indicates heightened security concerns. The recovery of small delicate weighing scales and small weights suggests that items of small size were handled in these areas. Given the security concerns, one might speculate that items of diminutive proportions but relatively great value (eg. gem stones) were handled in these locations.

Excavations and geological coring suggest that the port in the Ptolemaic period was farther west and north, and throughout the early and late Roman periods farther east and south relative to the coastline (Fig. 3; Harrell 1996: 99105; 112-125; Sidebotham 1999d: 91-94; Sidebotham & Wendrich 1999b: 447). This was due to silting of the harbours over the centuries that led to a diminishing area for the movement of ships. This is manifest by a complete lack of Ptolemaic structures beneath the Roman edifices nearest to the eastern coastline of the site (which suggests that the shore at that time was farther west and north than in the Roman period), and the silting-up of earlier Roman sea walls and the extension of later Roman buildings farther east towards the sea than the earlier Roman structures.

Remains in the ground floor courtyard of one of these buildings included large numbers of ashlars and three inscriptions that had clearly been removed from earlier structures including some temples. One of the inscriptions, fragmentary and dedicated to Isis, dates to the reign of either Trajan (98-117) or Hadrian (117-138; Fig. 8); two others, duplicates of one another, were dedications made by a woman named Philotera to Jupiter. These were probably carved sometime during the reign of Nero (5468) though the critical portions for dating the text undoubtedly the emperor’s name and titulature - have been deliberately destroyed, probably the result of damnatio memoriae (Figs. 9-10; Bagnall in prep.). Portions of reused

This discussion will progress from the eastern to the western parts of the site. Aside from a huge early Roman pottery dump redeposited sometime in the C4th AD that spread the length of the eastern edge of Berenike, there is little evidence of early Roman activity at the eastern-most edge of the site. This is undoubtedly due to the siltation process described above. Buildings closest to the harbours on the eastern edge of Berenike in the later periods of occupation were sea walls, warehouses and edifices of as yet unknown function. There was also a Christian ecclesiastical complex. Most edifices along the eastern edge of the site preserve evidence of squatters in their latest occupation phases.

2 The Axumite coin is an aes issue of King Aphilas (c.270/290 - before 330), the last pre-Christian king of Axum: see Munro-Hay and JuelJensen 1995: 95-96 (copper type 13) for a parallel. The Indian coin is a silver issue of Rudrasena III (Svami Rudrasena) of the dynasty of Kshatrapas of Western India (348-390): M. Bates (American Numismatic Society in New York City) & A. Sedov (Russian Academy of Sciences in Moscow) pers. comm.

Divided into a church area on the south and kitchen/domestic facilities on the north (Figs. 5-6), the Christian ecclesiastical building is of special interest. The single Axumite coin (of the late C3rd/early C4th AD) and 362

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

columns and column capitals, that may or may not be from the same buildings as the ashlars and inscriptions stacked in the courtyard, flank a staircase leading from the courtyard to an adjoining room (Fig. 11).

recovered a small bronze statuette possibly of the goddess Isis here. That and the proximity of the Serapis temple, only about 90m south of this shrine - both Isis and Serapis were closely associated with Mithras (Witt 1975; Lease 1986: 123-124 & nn. 38-39; Meiggs 1973: 372) - reinforce the identification of this structure as a mithraeum.

The upper floors of these buildings in this late Roman quarter were seemingly reserved for private purposes. Debris from the fallen upper stories comprises predominantly cooking and table wares that would suggest domestic residential activities in those locations. Staircases leading to upper floors are made, for the most part, of reused gypsum-anhydrite ashlars exhibiting little wear (Fig. 7) suggesting that most visitors to the buildings confined their activities to the ground floors. Such an arrangement of commercial activities on the ground floor and residential on the floor above is common throughout the Roman period in many urban settings.

The Serapis temple and its related courtyard, the shrine of the Palmyrenes and a later cultic centre atop that of the Palmyrenes dominated the central portion of Berenike. A large trench, under excavation since 1996, immediately north of and adjacent to the Serapis temple, has produced a relatively complete stratigraphic and ceramic chronology for the site as a whole from the early Roman through late Roman periods. At one point in the early Roman era this area served as the courtyard for the Temple of Serapis. The recovery of wooden bowls containing burnt offerings similar to ones also recovered in the putative mithraeum noted above and in the cult centre of the Palmyrenes strongly suggests that this courtyard was, indeed, closely associated with the early Roman phase of the Serapis temple. In addition, one of two large Nile-silt dolia found imbedded in the courtyard surface (Fig. 14) contained 7.55kg of black peppercorns from India (the largest single concentration found to date anywhere in the Roman world) that may also have been destined for ritual purposes in the Serapis temple.

Excavation of a C5th trash dump in this late Roman commercial and residential quarter provides abundant evidence of on-going contacts with India and Sri Lanka at that time in the form of Indian resist dyed textiles (cf. Sidebotham & Wendrich 1998: 91, fig. 11; Wild & Wild 2000: 269-270, 272), coconut and pepper, and beads from India and Sri Lanka (Sidebotham 2000b:120-134; Francis 2000: 221-223; Cappers 1998; 1999a; 2000). Graffiti in Hebrew and late Palmyrene from this C5th dump indicate the presence of a Semitic population at Berenike in this late period (Schmitz 2000).

There is abundant evidence for late Roman activity in strata above the Serapis temple courtyard. This takes the form of numerous walls, trash dumps and small scale industrial activities (metal and glass slag, bead working and crucibles). It is uncertain if the glass, bead and metal working that took place here was associated with the Serapis temple or independent of it (cf. Finnestad 1997: 190) as it has not yet been established when the Serapis temple ceased to function as a religious facility in the city.

One structure in the area immediately north of the late Roman commercial-residential quarter and aforementioned trash dump served as a religious centre for a nonChristian/non-Jewish cult at the turn of the C4th-C5th (Sidebotham 2000b: 134-144; 2001b). Excavations in this small enclosed shrine - that appears to have been barrel vaulted - produced numerous terracotta oil lamps and about 50 amphora toes that seem to have been used as torches (Sidebotham & Wendrich 1998: 96, Figs. 18-19). This suggests that some type of mystery cult with a limited clientele restricted to initiates took place here. The cult cannot be identified with a high degree of certainty; the limited evidence, however, suggests that it may be that of Mithras.

Excavations in the late Roman levels above the Serapis temple courtyard recovered kiln fired bricks in a secondary use capacity. In the Eastern Desert such bricks are found almost exclusively in hydraulic contexts. In addition there were fragments of glass vessels usually found associated with bathing activities (Nicholson 2000: 203-205). The bricks and glass suggest the presence, somewhere in the centre of the city and not far from the Serapis temple, of a bath building that by the late Roman period may have been in a state of decay.

While few mithraea have been identified in Egypt (Lease 1986; Harris 1996) and the Berenike structure lacks a blood pool indicative of one of the central cult rites of Mithras, a mithraeum in an international port of Berenike’s calibre would be appropriate (Daniels 1975). Lack of a blood pool does not necessarily preclude an identification of a mithraeum; numerous mithraea in Britain lack such pools. Furthermore, the small size and plan - in the earliest of two major phases, lines of benches parallel the long north and south walls of the building (Fig. 12); one of the benches has a seat that was clearly a reused relief carved on local gypsum/anhydrite most likely of religious significance (Fig. 13) - of the Berenike structure closely resemble mithraea found in Britain (Harris & Harris 1965: 51, 54; Liversidge 1968: 450, 452, 454). Excavations

The northern part of the site preserves a major trash dump from the early Roman period. Here hundreds of texts (predominantly ostraca, some papyri and a few documents written on wood, stone and eggshell) mainly from the mid to late C1st AD attest activities of the Roman customs house. The ostraca are predominately permits to allow goods to pass through the customs gate to outfit ships: food, drink, ships’ rigging etc., or as cargo destined for India or other eastern ports (eg. various wines, including Aminaean and Laodicean, that are also independently 363

Limes XVIII

attested in the Periplus Maris Erythraei 6, 49) as well as Kolophonian, Rhodian and Ephesian wines (Casson 1989: 113; Bagnall, Helms & Verhoogt 2000: 14-21). The best preserved papyrus yet found on site records the sale of a white male donkey for 160 drachmai during the reign of Nero (R. Bagnall & C. Helms pers. comm.).

representing a sizeable proportion of all beads found in late Roman levels at Berenike; Harrell 1999: 114; Francis 2000: 222-223). There was also commercial exchange with other regions of the Indian Ocean (coconut; Cappers 1998, 1999a, 2000), the Red Sea kingdom of Axum (pottery, at least one coin), the Persian Gulf (pottery, perhaps pearls) and southern Arabia (abrus precatorius seeds from the Barygaza area of India that indirectly point to the incense trade as these seeds were mixed with frankincense and myrrh; Sidebotham & Wendrich 1998: 91; Cappers 1999a: 302).

Other finds from the early Roman trash deposit include a wide range of organic materials: floral and faunal remains, textiles - including Indian sail cloth (Wild & Wild 2000: 265-269) basketry, matting and cordage, and wood or bone brailing rings (Sidebotham & Wendrich 1998: 91, Fig. 12). There is also evidence for the burial of a pet dog or cat complete with a collar made of blue faience beads. Inorganic finds include coins, beads, plaster jar stoppers, some beautiful painted and cameo and millefiore glass. A broken statuette of Venus/Aphrodite made from local anhydrite gypsum (Fig. 15) suggests the presence of a local sculpture school in Berenike.

The presence of lapis lazuli at Berenike indicates contacts, most likely indirect, with Afghanistan, the only known source of this stone in antiquity (Harrell 1999: 111). Though additional research must be conducted, it is likely that a bead found on the surface during the 2000 season originates from eastern Java and that two others have a provenance of either Thailand or Vietnam (P. Francis pers. comm.). Contacts between Berenike and the Mediterranean in the early Roman period were wide-ranging, less so in late Roman times when trade seems to have been limited mainly to the eastern Mediterranean.

The Ptolemaic areas of use and occupation discovered thus far at Berenike are confined to the northern (beneath the early Roman trash dump; Sidebotham 1999d: 46-57), central (beneath the Palmyrene religious complex: Sidebotham 2000b: 45-47) and especially the western (Ptolemaic industrial) parts of the site. Few texts from this period have been documented though one late Ptolemaic ostacon in Demotic from beneath the early Roman trash dump has been recovered. It appears to be an account, probably of wine and one other unidentified substance (U. Kaplony-Heckel pers. comm.).

The textual evidence is extensive in the form of large and small inscriptions on stone, a terra-cotta oil lamp (Fig. 16), egg shell, wood, ostraca, papyri and a coin. At least 10 different languages are now attested at the site, an extraordinary diversity, including Latin, Greek, Demotic, Coptic, Tamil-Brahmi, Hebrew, Palmyrene, SanskritPrakrit hybrid probably in Brahmi script and two as yet unidentified scripts.

The recovery of a fragment of an elephant’s tooth from the early Ptolemaic industrial zone provides the first archaeological evidence from Berenike for the presence of pachyderms (W. Van Neer pers. comm.). Papyri and literary sources record the passage of elephants through Berenike transported from ports farther down the Red Sea coast for use by the Ptolemaic military during the C3rd and C2nd BC (Scullard 1974: 123-145; Wilcken 1912: 533535). Some human remains, both adult and infant, have also been recovered in the Ptolemaic industrial zone that appear to have been Roman burials made after abandonment of the area in the Ptolemaic period (Barnard 1998: 389-396).

The bulk of the texts are commercial (Roman customs house archives on ostraca noted above) or religious. The latter include an inscription dated September 215, and its accompanying bronze statue, made by a Palmyrene auxiliary soldier and dedicated to the Roman imperial cult of Caracalla and Julia Domna (Verhoogt 1998: 193-196), a bilingual Greek-Palmyrene inscription of the later C2nd to early C3rd AD dedicated to the Palmyrene deity Hierobol/Yarhibol (Dijkstra & Verhoogt 1999; Sidebotham & Wendrich 1998: 93-94 and Figs. 15-16), a pair of C1st AD dedications in Greek to Jupiter (Figs. 9-10), a C2nd AD inscription addressed to Isis (Fig. 8), and a C5th Coptic Christian aphorism on a terracotta lamp that exhorts: “Jesus forgive me.” (Fig. 16; R. Bagnall pers. comm.).

Trade items and ethnicity of Berenike’s inhabitants Analysis mainly of organic, ceramic and, to a lesser extent numismatic evidence indicates extensive trade contacts with India (pepper, sorghum, perhaps rice, bamboo, teak wood, textiles, pottery, woven mats and baskets, semiprecious stones, perhaps agate cameo blanks, at least one coin, job’s tear - an Indian grass seed used as a bead, black myrobalan and sail cloth; cf. Cappers 1998; 1999a; 2000; Vermeeren 1998; 1999a; 2000; Harrell 1998: 143-145; 1999: 109-115, 118-121; Wild & Wild 2000: 265-269).

Faunal remains (mainly sheep, goat, camel and their byproducts) suggest in the Ptolemaic period that Egyptian and desert dwelling inhabitants of Berenike predominated. This changed in early Roman times when an increase in pig and fish bones, supplemented by fine - mainly terra sigillata - pottery from the western and eastern Mediterranean, and quantities of iron weapons (Hense 1999: 225, 230) suggests more of a Roman military and romanized Mediterranean population at the port. The abundance of Indian coarse, cooking and fine wares, predominantly of early Roman date, suggests a south Asian

Commerce with Sri Lanka was also extensive especially in the late Roman period (sapphires and beads, the latter 364

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

population at Berenike at that time and, to a lesser extent, later (Sidebotham & Wendrich 1998: 89; Begley & Tomber 1999; Tomber 2000; Tomber & Begley 2000).

AD stone statuette of Aphrodite (Fig. 15) suggest other cults observed by the port’s inhabitants. Excavations in the environs of Berenike

From the second half of the C4th AD on there was, once again, a predominance of goat, sheep and camel bones and their by-products, an increase in the frequency of Nubian/Blemmye-style pottery, a weaving pattern evident in textiles and matting typical of Nubia and pottery predominantly from the eastern Mediterranean. These material remains might indicate a different perhaps less diverse population group at Berenike than in the early Roman period, one that derived mainly from Egypt (especially Lower Nubia and Upper Egypt) and the eastern Mediterranean (Sidebotham & Wendrich 1999b: 452-453; 2000: 419).

Excavations in the environs of Berenike were conducted at the late C4th-C6th AD site of Shenshef (23˚ 44.25’ N/35˚ 22.72’ E) about 21.3kms south-west of Berenike in 1996 and 1997 and are not reported here (cf. Aldsworth & Barnard 1998; Gould 1999; Aldsworth 1999; Cappers 1999b; Vermeeren 1999b; Van Neer & Ervynck 1999). More recently the project conducted excavations at the large (23˚ 51.84’ N/35˚ 24.46’ E) and small hydreumata (23˚ 52.54’ N/34˚ 24.97’ E) in Wadi Kalalat about 8.5 and 8kms south-west of Berenike respectively and at the small hydreuma at Siket (at 23˚ 55.88’ N/35˚ 24.46’ E), approximately 7kms west-north-west of Berenike (Fig. 4; Sidebotham & Wendrich 1998: 91, Fig. 10; Sidebotham, Barnard, Pearce & Price 2000).

There was, however, apparently, continued Indian presence at Berenike in this late period and the appearance - most evident from the ceramic remains and the find of one coin of Axumites on site at this time.

Excavations in three trenches in the large hydreuma in Wadi Kalalat conducted in 1997 and 1998 revealed two gates, part of the curtain wall and over 30 fragments of at least two inscriptions related to the installation, one of which is from the period of Trajan (Van Neer & Ervynck 1999: 443-444; Sidebotham, Barnard, Pearce & Price 2000: 383, 387, 394; Bagnall 2000). Excavations in the smaller hydreuma in Wadi Kalalat produced no texts, but ceramic evidence for occupation that ran later than that at the large hydreuma (Haeckl forthc.).

The excavation of a number of luxury items points to elements of the population that were relatively affluent during both the early and late Roman periods. These include finds such as escargot from the Mediterranean, fine jewellery (including the sapphire mentioned above and a gold and pearl earring; Sidebotham & Wendrich 1998: 90), gold glass and gold foil beads; Nicholson 1999: 235-236; Francis 2000: 216-217), elaborate wall hangings or carpets (some from India; Wild & Wild 2000: 271-273), and marble wall or floor revetment from the Proconessus in Asia Minor (Harrell 1996: 111; 1998: 142).

The small hydreuma at Siket had surface pottery suggesting activity there from the C1st-C3rd AD (Sidebotham 2000c: 359-365). A massive triangularshaped Latin inscription 2.46m long x 1.07m high x 0.17m thick carved on local gypsum/anhydrite stone found at the main gate during the excavations in the winter 1999-2000 season chronicles the installation’s activities in the 9th year of Vespasian (76/77) when M. Trebonius Valens was praefect of Mons Berenicides, Lucius Julius Ursus was governor of Egypt and Q. Valerius, possibly, camp commandant. The inscription records that the installation at Siket was a “...HYDRERUMA...PRAESIDIUM ET LACVS.” (Figs. 17-21; Bagnall in prep.).

Skeletal remains mainly from the late Roman period and small finds suggest that men, women and children resided at the port (Sidebotham 1998b: 51, 84-85, 107-108; Barnard 1998). Some had sufficient wealth to afford pet animals (Sidebotham 2001b: forthc.). Religious beliefs of the inhabitants were varied as might be expected in such an international emporium. Epigraphic evidence attests cults of Serapis, Isis, Jupiter, Hierobol/Yarhibol (Sidebotham & Wendrich 1998: 94-95, Figs. 15-17), the Roman imperial cult (Caracalla, Julia Domna), Christianity and, though evidence is minimal at this point (ostraca with ‘Jewish delicacies’ (C.C. Helms pers. comm.), and a C5th graffito in Hebrew: Schmitz 2000: 183-186), Judaism.

Undoubtedly these outlying forts plus 7 others (10 in all) provided a ring of defense for Berenike in the early and, to a lesser extent, in the late Roman periods and those nearest the city (eg. those in Wadi Kalalat and Siket) also provided drinking water to the port (Fig. 4).3

Other religions were practiced or probably were in light of the near certain presence of Indians (Buddhism) and others at Berenike, but specifics elude us at this point. Though not attested epigraphically, the cult of Mithras (noted above) seems to have been present at Berenike in the late Roman period. A small stone head of Harpocrates found in the same shrine as the Roman imperial cult and that of Hierobol/Yarhibol, and a late Roman wooden pyxis lid depicting Aphrodite (Haeckl 1999) together with a C1st

3

The 10 forts are from north to south the fort in Wadi Lahma (Laham, at 24˚ 09.92’ N/35˚ 21.81’ E), the 5 forts at Wadi Abu Greiya (Vetus Hydreuma, at 24˚ 03.44’ N/35˚ 17.02’ E; 24˚ 03.68’ N/35˚ 17.23’ E; 24˚ 03.64’ N/35˚ 17.90’ E; 24˚ 03.73’ N/35˚ 17.95’ E; 24˚ 03.79’ N/35˚ 18.03’ E], the hydreuma at Siket, the two hydreumata in Wadi Kalalat and the hill top fort in Wadi Shenshef [at 23˚ 44.12’ N/35˚ 22.98’ E]. See Sidebotham 1999b; 2000c.

365

Limes XVIII

Survey of the hinterland

Conclusion

An integral part of the Berenike project is the local and regional survey that seeks to locate, identify, pinpoint (using the global positioning system - GPS), date and determine the functions and inter-relationships of roads and settlements in the region to each other and to Berenike itself (see the various Berenike volumes).

The long-term objectives of excavations at Berenike and environs and of the Eastern Desert survey are to understand in an integrated way the overall rôle the entire region played in local, regional, provincial, empire-wide and extra-imperial history (economic, social, religious and political) in this part of the ancient world in the almost 1,000 years spanning the beginning of the Ptolemaic to the end of the Roman occupation of Egypt.

In addition to surveys linked directly to the excavations at Berenike, there have been three surveys conducted in the summers of 1998, 1999 and 2000 by the University of Delaware. These latter projects have concentrated specifically on detailed examination of road systems in the Eastern Desert.

The first seven seasons of fieldwork at and around Berenike between 1994 and 2000 have provided a general overview of activity in the entire region of the southeastern desert of Egypt. There is ample archaeological evidence for early Ptolemaic activity at Berenike predominantly at this point from the industrial point of view. Little from the late Ptolemaic age has come to light in the excavations to date. And, aside from the find of a fragment of an elephant tooth, there is no indication of a sustained long-distance maritime trade with other areas of the Red Sea-Indian Ocean in the Ptolemaic period.

Over the years scores of previously unrecorded sites (including putative Christian and other settlements, gold mines, emerald mines, cemeteries, military installations and hydreumata, road stations and wells) and roads have been found and recorded. Some of these have been drawn in measured plan and have been published in the annual reports or as separate publications.

Of the three major periods of activity known at Berenike (early Ptolemaic, early Roman and late Roman) the Ptolemaic is the least represented at sites recorded by the survey within a radius of 45kms of Berenike; only two sites clearly preserve surface evidence of Ptolemaic activity. Those are the small fort in Wadi Lahma (Lahami) and at least two of the forts in Wadi Abu Greiya (Fig. 4).

Not all can be mentioned here, but the more noteworthy of these sites found or otherwise studied by the surveys in some detail between 1998-2000 includes the recording of the hydreuma at Siket noted above, the discovery of the putative Christian anchorite settlement of Gariya Mustafa Amr Gama in Wadi Umm Atlee (100-110 buildings; at 23˚ 36.77’ N/35˚ 23.43’ E) dating to the C5th to C6th (Fig. 4; Sidebotham 2000c: 367-372; Sidebotham, Barnard & Pyke: in prep.) and drawing a detailed plan of the large C1st to C5th AD emerald mining settlement at Sikait (ancient Senskis or Senskete) in the Mons Smaragdus region (about 300 buildings, rock cut temples; at 24˚ 37.92’ N/34˚ 47.82’ E; Rivard & Foster in prep.). The surveys have also recorded, more or less entirely, the route of the via Hadriana (about 800kms in length) and drawn plans of many sites along or associated with it (Fig. 22; Sidebotham & Zitterkopf 1997 & 1998; Sidebotham, Zitterkopf & Helms 2000; Sidebotham, Zitterkopf & Tomber forthc.). Most recently there was a detailed survey of sites along or near the Marsa Nakari (Nechesia?)-Nile road (discovered by the survey in 1997; Fig. 23; Bagnall, Manning, Sidebotham & Zitterkopf 1996; Sidebotham 1997: 388-390; 1999f: 364-368; Seeger in prep.)4 and the location of scores of sites throughout the Eastern Desert which had previously been unrecorded (cf. Sidebotham & Wendrich 2001).

Farther afield along the road leading from Berenike to the Nile as well as in other regions of the Eastern Desert other sites preserving evidence of Ptolemaic activity are known. These were, perhaps, originally part of the building project of Ptolemy II Philadelphus, noted in various ancient literary sources, whose construction works in the Eastern Desert and along the Red Sea coast prompted subsequent Ptolemaic and early Roman penetration into the region. The early (C1st AD) and late Roman (mid-C4th, into the C5th to C6th) periods are abundantly documented in the vicinity of Berenike. Ceramic, epigraphic and numismatic evidence indicates that all 10 forts shielding Berenike at the time - some of which also provided drinking water to the emporium - from Lahma (Lahami) in the north-west to the hilltop fort of Shenshef in the south-west (see n.3 above), functioned in the early Roman period (C1st into the early C2nd; Fig. 4). In addition, there are cemeteries near Berenike and numerous sites farther afield on or near the main road between Berenike and the Nile that preserve ceramic evidence indicating extensive and intensive activity in the early Roman period. The via Hadriana was also functioning at the end of the early Roman period joining Berenike to other ports along the Red Sea coast to the north and, ultimately, to Antinoë/Antinoopolis on the Nile in Middle Egypt.

4

Dr. J.A. Seeger of Northern Arizona University conducted excavations at the port of Marsa Nakari (perhaps ancient Nechesia) in summer 1999. Excavations are scheduled to continue there in summer 2001. The University of Delaware conducted a detailed survey (drawing plans of sites) along the Marsa Nakari-Nile road in summer 2000. S.E. Sidebotham, G. Compton, B. Tratsaert & L.A. Pintozzi are preparing the final publication of the Marsa Nakari (Nechesia ?)-Edfu (Apollinopolis Magna) road survey.

Surprisingly, the late Roman period, from the mid-C4th AD on, was one of robust activity in the environs of 366

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

Berenike in settlements with uncertain functions (eg. Shenshef, Hitan Rayan, Gariya Mustafa Amr Gama, the latter two perhaps Christian laura settlements) as well as in at least three of the ten forts that had guarded approaches to the city in the early Roman period (Wadi Lahma (Lahami) and two in Wadi Abu Greiya/Vetus Hydreuma, which, interestingly, were the only two sites also active in the Ptolemaic period). Of note were on-going operations at a number of the forts on the road between Berenike and the Nile at Koptos and several of the emerald and gold mining operations on or near that trans-desert thoroughfare. In addition, the road linking Marsa Nakari (perhaps the ancient Nechesia) on the Red Sea coast and Edfu (Apollinopolis Magna) on the Nile, that appears to have been used if not actually built in early Ptolemaic times, continued in operation into the late Roman period (Fig. 23). At least portions, if not all, of the via Hadriana also continued in use into the C5th and C6th AD (Fig. 22).

Bagnall R.S., Manning J.G., Sidebotham S.E. & Zitterkopf R.E. 1996 A Ptolemaic inscription from Bir ‘Iayyan. CE LXXI, fasc. 142: 317-330. Barnard H. 1998 Human bones and burials. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 389-401. Begley V. & Tomber R.S. 1999 Indian pottery sherds. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 161-181. Cappers R.T.J. 1998 Archaeobotanical remains. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 289-330. Cappers R.T.J. 1999a The archaeobotanical remains. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 299-305. Cappers R.T.J. 1999b Archaeobotanical remains from Shenshef. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 419-426. Cappers R.T.J. 2000 The archaeobotanical remains. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 305-310. Casson L. 1989 The Periplus Maris Erythraei text with introduction, translation and commentary. (Princeton). Daniels C.M. 1975 The role of the Roman army in the spread and practice of Mithraism. In J.R. Hinnells (ed.) Mithraic studies. Proceedings of the First International Congress of Mithraic Studies, Vol. II (Manchester): 249-274. Dijkstra M. & Verhoogt A.M.F.W. 1999 The Greek-Palmyrene inscription. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 207-218. Finnestad R.B. 1997 Temples of the Ptolemaic and Roman periods: Ancient traditions in new contexts. In B.E. Shafer (ed.) Temples of ancient Egypt. (Cornell): 185237. Francis P. jnr. 2000 Human ornaments. In S.E. Sidebotham & W.Z. Wendrich, (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 211-225. Gould D.A. 1999 Excavations at Shenshef. In S.E. Sidebotham & W.Z. Wendrich, (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 371-383. Haeckl A.E. 1999 The wooden ‘Aphrodite’ panel. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 243-255.

The archaeological evidence from sites examined within a 45kms radius of Berenike strongly suggests that most if not all were in regular contact with the port for a variety of purposes, primarily economic. Farther afield along the roads leading to the Nile at Edfu and Koptos some sites preserve indications of the importation of Red Sea fauna that most likely came from Berenike. It is evident that Berenike played an important rôle as an economic centre or transhipment point for goods not only at the ‘international’ level, but also at the local and regional levels. Berenike’s rôle in the local and regional economies is not as well understood as its function in the wider ‘international’ trade. Continued surveying and excavation in the region around Berenike should document better this facet of the ancient port’s relationship with the Eastern Desert hinterland. Bibliography Adam, J.-P. 1994 Roman building materials and techniques. (Bloomington). Aldsworth F.G. 1999 The buildings at Shenshef. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 385-418. Aldsworth F.G. & Barnard H. 1998 Survey of Shenshef. In S.E. Sidebotham & W.Z. Wendrich (edd). Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 427-443. Bagnall R.S. 2000 Inscriptions from the hydreuma in Wadi Kalalat. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 403-412. Bagnall R.S., Helms C.C. & Verhoogt A.M.F.W. 2000 Documents from Berenike. Volume I Greek ostraka from the 1996-1998 seasons. Papyrologica Bruxellensia 31. (Brussels).

367

Limes XVIII

Harrell J.A. 1996 Geology. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1995. Preliminary report of the 1995 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 99-126. Harrell J.A. 1998 Geology. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 121-148. Harrell J.A. 1999 Geology. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 107-121. Harris E. & Harris J.R. 1965 The oriental Cults in Roman Britain. (Leiden). Harris J.R. 1996 Mithras at Hermopolis and Memphis. In D.M. Bailey (ed.) Archaeological research in Roman Egypt. The Proceedings of the Seventeenth Classical Colloquium of the Dept. of Greek & Roman Antiquities, British Museum, held on 1-4 December, 1993. Journal of Roman Archaeology Supplementary Series 19: 169-176. Hense A.M. 1999 The metal finds. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 219-241. Lease G. 1986 Mithra in Egypt. In B.A. Pearson & J.E. Goehring (edd.) The roots of Egyptian Christianity. (Philadelphia): 114-129. Liversidge J. 1968 Britain in the Roman empire. (New York). Meiggs R. 1973 Roman Ostia. (2nd edit.: Oxford). Munro-Hay S.C.H. 1996 Aksumite overseas interests. In J. Reade (ed.) The Indian Ocean in antiquity (London): 403-416. Munro-Hay S. & Juel-Jensen B. 1995 Aksumite coinage. (Revised & enlarged edit.; London). Nicholson P.T. 1999 The glass. In S.E. Sidebotham & W.Z. Wendrich (edd.). Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 231-241. Nicholson P.T. 2000 The glass. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 203-209. Rivard J.-L. & Foster B.C. in prep. Survey of Sikait. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 2000. Report of the 2000 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat and Siket. Schmitz P.C. 2000 Semitic graffiti. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 183-189. Scullard H.H. 1974 The elephant in the Greek and Roman world. (Ithaca). Sidebotham S.E. 1995 Coins. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1994. Preliminary report of the 1994 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 47-48.

Sidebotham S.E. 1996 Overview of archaeological work in the Eastern Desert and along the Red Sea coast of Egypt by the University of Delaware-Leiden University, 1987-1995. Topoi 6/2: 773-783. Sidebotham S.E. 1997 Caravans across the Eastern Desert of Egypt: Recent discoveries on the BerenikeApollinopolis Magna-Coptos roads. In A. Avanzini (ed.) Profumi d’Arabia Atti del Convegno. Saggi di Storia antica 11. (Rome): 385-393. Sidebotham S.E. 1998a The coins. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 181192. Sidebotham S.E. 1998b The excavations. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 11-120. Sidebotham S.E. 1999a Berenike. In K.A. Bard (ed.) Encyclopedia of the archaeology of ancient Egypt. (London): 170-172. Sidebotham S.E. 1999b Berenike, trade and the military: Investigations at a Red Sea port in Egypt. In N. Gudea (ed.) Roman frontier studies 1997. Proceedings of the XVIIth International Congress of Roman frontier studies 1997. (Zalău): 251-257. Sidebotham S.E. 1999c The coins. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 183-199. Sidebotham S.E. 1999d The excavations. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 3-94. Sidebotham S.E. 1999e News of members. The Explorers Club Newsletter 31/3 (July-September): 3-4. Sidebotham S.E. 1999f Survey of the hinterland. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 349-369. Sidebotham S.E. 2000a Coins. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 169-178. Sidebotham S.E. 2000b The excavations. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 3-147. Sidebotham S.E. 2000c Survey of the hinterland. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 355-377. Sidebotham S.E. 2001a The coins. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1999. Report of the 1999 excavations at Berenike. (Leiden): forthc. Sidebotham S.E. 2001b The excavations. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1999. Report of the 1999 excavations at Berenike. (Leiden): forthc.

368

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

Sidebotham S.E., Barnard H. & Pyke G. in prep. Sidebotham S.E., Barnard H., Pearce D.K. & Price J.A. 2000 Excavations in the hydreuma in Wadi Kalalat. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 379-402. Sidebotham S.E. & Seeger J.A. 1996 The coins. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1995. Preliminary report of the 1995 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 179-196. Sidebotham S.E. & Wendrich W.Z. 1996 Berenike: Roman Egypt’s maritime gateway to Arabia and India. Egyptian Archaeology 8: 15-18. Sidebotham S.E. & Wendrich W.Z. 1996-1997 Fieldwork at Berenike (Red Sea coast), Egypt 1994-1996. Archaeological News 21-22: 34-40 & 117-131. Sidebotham S.E. & Wendrich W.Z. 1998 Berenike: Archaeological fieldwork at a Ptolemaic-Roman port on the Red Sea coast of Egypt: 1994-1998. Sahara 10: 85-96. Sidebotham S.E. & Wendrich W.Z. 1999a Berenike: an ancient emporium at the cross-roads of the Mediterranean-Red Sea-Indian Ocean trade. The Indian Ocean Review 12/3 (December): 16. Sidebotham S.E. & Wendrich W.Z. 1999b Interpretative summary and conclusion. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 445-456. Sidebotham S.E. & Wendrich W.Z. 2000 Interpretative summary and conclusion. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Desert, including excavations in Wadi Kalalat. (Leiden): 413420. Sidebotham S.E. & Wendrich W.Z. 2001 Berenike: archaeological fieldwork at a Ptolemaic-Roman port on the Red Sea coast of Egypt 1999-2001. Sahara 13: forthc. Sidebotham S.E. & Zitterkopf R.E. 1995 Routes through the Eastern Desert of Egypt. Expedition 37/2: 39-52. Sidebotham S.E. & Zitterkopf R.E. 1997 Survey of the via Hadriana by the University of Delaware: the 1996 Season. BIFAO 97: 221-237. Sidebotham S.E. & Zitterkopf R.E. 1998 Survey of the via Hadriana: the 1997 season. BIFAO 98: 353-365. Sidebotham S.E., Zitterkopf R.E. & Helms C.C. 2000 Survey of the via Hadriana: The 1998 season. Journal of the American Research Centre in Egypt 37: 115-126. Sidebotham S.E., Zitterkopf R.E. & Tomber R.S. in prep. Survey of the via Hadriana: The final report. Tomber R.S. 2000 Indo-Roman trade: The ceramic evidence from Egypt. Antiquity 74: 624-631. Tomber R.S. & Begley V. 2000 Indian pottery sherds. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 149-167. Van Neer W. & Ervynck A.M.H 1999 The faunal remains from Shenshef and Kalalat. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the

Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 431-444. Verhoogt A.M.F.W. 1998 Greek and Latin textual material. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 193-198. Vermeeren C.E. 1998 Wood and charcoal. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1996. Report of the 1996 excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden): 331-348. Vermeeren, C.E. 1999a Wood and charcoal. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 307-324. Vermeeren C.E. 1999b Wood and charcoal from Shenshef. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden): 427-429. Vermeeren C.E. 2000 Wood and charcoal. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 311-342. Wendrich W.Z. & Sidebotham S.E. 1995 Port of elephants and pearls. Egypt Today 16/12: 144-145. Wilcken, U. 1912 Grundzüge und Chrestomathie der Papyruskunde. Erster Band, Historischer Teil, Zweiter Hälfte. Chrestomathie. (Leipzig). Wild J.P. & Wild F.C. 2000 The textiles. In S.E. Sidebotham & W.Z. Wendrich (edd.) Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations in Wadi Kalalat. (Leiden): 251-274. Witt, R.E. 1975 Some thoughts on Isis in relation to Mithras. In J.R. Hinnells (ed.) Mithraic studies. Proceedings of the First International Congress of Mithraic Studies. Vol. II. (Manchester): 479-493.

369

Limes XVIII

Fig. 1. Map of the Eastern Mediterranean, Red Sea and north-western Indian Ocean showing location of Berenike.

Fig. 2a. Map of northern end of the Red Sea.

370

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

Fig. 2b. Map of northern end of the Red Sea.

Fig. 3. Plan of Berenike indicating location of trenches (numbered) and of magnetometric survey (grey outline) of a portion of the Ptolemaic industrial area. Drawn by F.G. Aldsworth, H. Barnard & A.M. Hense.

371

Limes XVIII

Fig. 4. Map of the environs of Berenike. Drawn by A. Hoseth.

Fig. 5. Plan of the ecclesiastical complex at Berenike. Drawn by A.M. Hense.

372

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

Fig. 6. The central eastern end of the ecclesiastical complex at Berenike. To left (area with stairs) is Trench BE99/00-30 (part of the domestic/residential complex). To right is Trench BE98/99-22 (ashlar floor and benches) is the north-eastern part of the church. Scale = 1m. Photo by S.E.Sidebotham.

Fig. 7. Trenches BE99-28 and BE00-38 in the late Roman commercial/residential quarter looking towards the south-east. Scale = 1m. Photo by S.E. Sidebotham.

Fig. 8. Inscription dedicated to Isis during the reign of either Trajan or Hadrian from Trench BE00-37 in the late Roman commercial/residential quarter. Each black and white segment of the scale = 20cms. Photo by S.E. Sidebotham.

373

Limes XVIII

Fig. 9. The better preserved one of a pair of inscriptions dedicated to Jupiter by Philotera probably during the reign of Nero from Trench BE00-37 in the late Roman commercial/residential quarter. Each black and white segment of the scale = 20cms. Photo by S.E. Sidebotham.

Fig. 10. The less well preserved one of a pair of inscriptions dedicated to Jupiter by Philotera probably during the reign of Nero from Trench BE00-37 in the late Roman commercial/residential quarter. Each black and white segment of the scale = 20cms. Photo by S.E. Sidebotham.

Fig. 11. Staircase with reused portions of columns and column capitals in Trench BE00-37 in the late Roman commercial/residential quarter looking southeast. Scale = 50cms. Photo by S.E. Sidebotham.

374

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

Fig. 12a. Western end of possible Mithraeum in Trench BE98/99-23 in its earliest phase looking north-west. Scale = 50cms. Photo by S.E. Sidebotham.

Fig. 12b. Western end of possible Mithraeum in Trench BE98/99-23 in its earliest phase looking north-west. Scale = 50cms. Photo by S.E. Sidebotham.

Fig. 13. Broken relief fragment depicting religious scene found as part of seating for northern benches in the possible Mithraeum in Trench BE98/99-23. Scale = 10cms. Photo by S.E. Sidebotham.

375

Limes XVIII

Fig. 14. Two dolia embedded in the temple courtyard adjacent to and immediately north of the Serapis temple looking east. The one in the foreground (western-most one with extant rim) contained 7.55kg of black Indian peppercorns. Scale = 50cms. Photo by S.E. Sidebotham.

Fig. 15. Gypsum /anhydrite statuette of Venus /Aphrodite from Trench BE99-33 in the early Roman trash dump at the northern end of the site. Scale = 10cms. Photo by S.E. Sidebotham.

Fig. 16. Terracotta oil lamp from Trench BE98-22 in the Christian ecclesiastical complex. The aphorism translates “Jesus forgive me!” Scale = 5cms. Photo by S.E. Sidebotham.

Fig. 17a. Hydreuma at Siket with view of trench excavated at the main gate, threshold, inscription (as found), part of north wall and north-west corner looking west. Scale = 1m. Photo by S.E. Sidebotham.

Fig. 17b. Hydreuma at Siket with view of trench excavated at the main gate, threshold, inscription (as found), part of fort interior looking southwest. Scale = 50cms. Photo by S.E. Sidebotham.

376

Steven E. Sidebotham: The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs

Fig. 18. Detail of Figure 17a. Hydreuma at Siket with view of trench at the main gate, threshold and inscription as found looking west. Scale = 50cms. Photo by S.E. Sidebotham.

Fig. 19. Latin inscription from the hydreuma at Siket. Dimensions are 2.46m long x 1.07m high x 0.17m thick. Scale = 1m. Photo by S.E. Sidebotham.

Fig. 20. Hydreuma at Siket. Artist’s reconstruction of gate and inscription. Drawn by A.M. Hense.

377

Limes XVIII

Fig. 22. Map of the Eastern Desert and northern end of the Red Sea depicting route of the Via Hadriana and other major road networks. Drawn by R.E. Zitterkopf.

Fig. 23. Map of the Marsa Nakari (Nechesia?)-Edfu (Apollinopolis Magna) route in relation to the BerenikeNile roads and the Quseir alQadim-Qift road. Drawn by R.E. Zitterkopf.

378

Merchants, pilgrims and soldiers on the Red Sea route Gabriele Ziethen & Erik Klingenberg In the ancient past, as well as today, the Red Sea belonged to one of the most broadly defined maritime regions and trade routes of the Near East. Its status, described by several Sea Laws, remains ambivalent: either defined as a ‘closed ocean’ or a ‘semi-closed ocean’ by Anglo-American, European and International Law, a further definition, seen from a Russian perspective, interprets the Red Sea as border zone (so-called ‘Doktrin der Randmeere’). The reason for this perspective might be seen in the guarantee of free access to the Suez Canal and the trade centres of the Near and Far East, the Mediterranean and Africa. This perspective contains a high risk of potential conflict as new concepts of geo-strategy are focused on the triangle between the northern harbours, the Arabian peninsula and the unspecified interests of the African neighbours in these modern times. In searching for a period in history where similar problems had to be mastered, the C3rd to the early C6th might offer some examples from the world of trade and the Roman to early Byzantine military presence on the coasts of the Red Sea. The history of Graeco-Roman trade in the Red Sea region can be reconstructed by interpreting literary, papyrological, juridical and numismatic sources. However the subject of interest is often focused on one special period or thematic perspective, mostly concerning the C1st and C2nd, because the Periplus Maris Erythraei, by an unknown author, and the Naturalis Historia, written by Pliny the Elder, are very detailed records. In addition to Heliodor’s Aithiopica, which was written in the early C3rd and which contains some information concerning trade, the descriptio totius orbis in the C4th and the records of the early Christian church fathers can provide further information. After the emperor Diocletian, Constantine and his successors’ trade and tax policies were criticized by their contemporaries. Modern historical research has often focused less on this subject and more on the history of early Christianization in the east. The perspective of historic interpretation has changed. However, the records contain some signs of a closer coherence between the changing culturalreligious world and the distribution and management of wealth. There the trade centres along the Red Sea route became of some importance. The Christian church fathers (eg. Clemens Alexandrinus, Hieronymus) and the authors of pilgrimage itineraries were well informed about trade and products along the Red Sea route between Alexandria in Egypt, Clysma, Adulis (at the coast of Eritrea) and also concerning the way to India. Additional notes on luxury goods and admonitions for ‘Christian’ merchants show the main position of the theological records: to maximize ecclesiastical property by minimizing individual investment in the local markets. In order to look for new marketplaces outside the borders of the Roman empire and the Roman tax administration, the merchants developed an interest in the exchange of goods with the kingdom of Aksum in the highlands of Ethiopia via the harbour of Adulis (eg. Kosmas Indikopleustes, early C6th). At the beginning of the C4th Christian pilgrims also became passengers of the merchant ships. They made use of the Roman military garrisons along the Red Sea coasts, eg. in the area of Clysma/Suez, and asked for military escorts during their excursions to the desert. On-board of one of these ships two young men, Frumentius and Aedesius, reached Aksum after they had been taken prisoner when the ship had anchored near a fresh water reservoir. A short time after their imprisonment Frumentius and Aedesius were welcomed at the Royal Court of Aksum where they began their work of converting the inhabitants to the Christian religion. The legend of the Christianization of Aksum contains the key to understanding the events of the following 150 years. The kings of Aksum developed a special interest in establishing influence along the Red Sea trade routes. The Aksumite presence can be found in the harbours of Yemen and within the protection of Christian merchant families in the Arabian city of Najran. The early Aksumite monetary system was developed from elements of the Roman coin system. The result of the rising influence of this new economic power was that Byzantium decided to accept this new competition in the Red Sea. However, after the Jewish citizens in Yemen and Najran perceived an attack on their own economic interests and trade network, the situation changed for the worse. Aksum decided to protect the Christian interests and King Kaleb sent a fleet to Arabia. There is even the possibility that monks were manning the warships’ oars. The final result was that envy of the trade situation led to a religious war. The changing situation of political influence after 524 and the Persian conquest of Yemen between 570-572 produced two losers: Aksum lost Yemen and the Byzantine fleet lost its control over the Red Sea routes. A new period of oriental naval power anticipated the way for early Muslim traders some decades later.

the Red Sea 1946: 61, 139ff., 183, 607ff.). The periplusliterature of antiquity, written pleas for help from stranded Muslim sailors in the middle ages and the still interesting and applicable marine handbooks of the C20th contain many examples of cultural coherence. It is perhaps interesting to note that information about the area was in some ways more extensive in the past than today.2

The coherence between culture, economy and different religions, which have linked the neighbours of the Red Sea region since ancient times, offers a unique perspective for modern interpreters. On the one side we see a vacation paradise described in detail and on the other the specifics of maritime law1 shrouded in foggy and murky legal waters. Or we consider navigation, which today as then is complicated: the transit of the passage at Bab-el-Mandeb is still a difficult task for marine pilots (Western Arabia and

Almost completely absent from the historic perspective of interpretation is the context of the rise of early Aksum as an integral part of Christian cultural history. From

* A detailed study is published in Ziethen (2000). Thanks to Prof. Dr. W. Hahn (University of Wien), Prof. Dr. S.T. Parker (Univ. of North Carolina) and to Prof. Dr. S.E. Sidebotham (Univ. of Delaware). 1 Brown 1994: 87ff.: Beckert & Breuer 1991: 6ff., 29ff.: Dupuy & Vignes 1985: 224-227; Colombos 1963: 159-168. Freerich 1982: 65-70: EI² 1 (Leiden/London 1960): 930-938.

2

For example the auction of the Ethiopian military fleet in 1996 was only mentioned by the Deutsche Presseagentur and in a few obscure newspaper articles in Europe.

379

Limes XVIII

antiquity to early modern times, the Red Sea region was an integrated part of an earlier perspective of the world. This view was defined by a positive assessment of the region and its cultures in the epic poetry of Homer, one of the earliest references to this area. A similar attitude was furthered in the trade activities of ancient Egypt, the dynasty of the Ptolemies and played an important rôle in Roman commercial policy (Vasiliev 1933; Raunig 1971: 137). For Jews and Christians, the Torah and the Bible contained aspects identifiable with the region, too.

juridical and encyclopedic records of the early Roman imperial era (C1st-C2nd), but can be also found in the publications of the early Christian Church fathers.4 Between the C2nd and C4th, the Christian Church fathers were critical in their view of the world of trade. Bishop Tertullian of Carthage criticized trade practices, because his city was a port of destination for Indian goods. In the Egyptian city of Alexandria, Clemens Alexandrinus described, from a negative point of view, the luxury of private households in detail. Born of a wealthy family, his book titled paidagogos-the educator and his essay “Who of the rich will achieve salvation?” embodies his special knowledge (Ziethen 2000: 21).

The results of recent excavations in Yemen, Ethiopia, the Sudan and the Eastern Desert of Egypt demonstrate the methods by which this cultural heritage was applied during periods of change (Ziethen 2000: 20). The examples of trade, military activities and Christian missions during late antiquity demonstrate some detailed aspects.

Clemens admonished Christians not to use trade methods like those the adherents of the ambivalent god Mercury were thought to prefer. This meant that Christians should not advertise their goods by shouting loudly in the market place, thus avoiding any suspicion of swindle or deception. From a juridical point of view, this could be interpreted as an allusion to the common practice of buying and selling goods by auction and the resulting possibility for fraud (Clemens Alexandrinus Paidagogos 3.79.1: Jakab 1997: 27-31).

With the rise of the Aksumite kingdom as a Christian neighbour of the Roman empire during the middle of the C4th AD an era of change began. Until the C6th AD the political power and influence of Aksum developed as a serious rival of Roman and Byzantine economic interests (Ahrweiler 1966: 8). With differing details, but incorporating one identical nucleus of their subject, Rufinus (Hist. Eccl. 1.9), Sozomenos (Eccl. Hist. 24), Theodoretos (Eccl. Hist 1.22), Gelasios (Eccl. Hist. 3.9) and Sokrates (Eccl. Hist. 1.19) summarized the history of religious conversion and discussed the mission to Aksumite aristocracy of Frumentius and Aedesius. Although at first this event was only relevant to Aksumite history, the following decades revealed the implications for the Aksumite neighbours, too.

Many other examples of the trade of luxury goods and the regulations imposed upon ‘Christian’ merchants reveal the main position of the theological records: to maximize ecclesiastical property by minimizing individual risk and investment in the local markets in order to obtain spiritual wealth. Examples from Palestine in late antiquity show the Christian church’s keen interest in owning farmland and estates. Finally the church was powerful enough to wield influence in changing local economic systems as H.-P. Kuhnen and at present P. Baumann have discussed.5

From the perspective of Christian ecclesiastical history, research was primarily focused on the differing interpretations of the theological roots of this mission and their links to the bishops’ (Athanasius and Arrian) quarrel concerning the real nature of Jesus Christ. As a result, the academic discussion on the mission’s history did not address the aspects of the period’s ancient daily life.3

The list of goods mentioned by Clemens Alexandrinus contains articles such as precious stones, valuable household goods, expensive cosmetics and drugs, wigs, woven materials and silk robes with golden threads.6 These products do not differ from the list of goods described by the Periplus Maris Erythraei in the C1st AD (Casson 1989), the trade laws of the late C2nd/early C3rd (esp. the Severan period) and Diocletian’s price edict of 301. And there are enough examples from the C4th, too.7

The success of the Christian mission in Aksum was not only due to the initiative of ecclesiastical institutions, but depended in equal manner on other conditions too.

At the beginning of the C6th Kosmas Indikopleustes also

In a letter, the Church father Hieronymus (125.3-4) described the nautical dangers for the sailors navigating the Red Sea route. He characterized India as the nautical destination and he described the luxury goods of the Far Eastern markets. In this context, Hieronymus contrasted the merchants’ and entrepreneurs’ pursuit of profit with the practice of real Christian values.

4

For example, although the 1994 published Sabaean wood inscriptions of the C2nd/C3rd mention musk/Moschus (‘mskn), the earliest Latin reference as muscus was used by Hieronymus in his correspondence c.390; Ibrahim 1991: 228ff; Ryckmans, Müller & Abdallah 1994: 65ff.; Müller 1997: 209. 5 Kuhnen 1994: 49ff.; Baumann 1999; 16-19; Ziethen 2000; 21ff. Cf. Clemens Alexandrinus Τίς ό σoζόµευος πλούσιος (“Who of the rich will achieve salvation?”) 14.1; 16. 6 Clemens Alexandrinus Paidagogos 2 pass. - Clemens Alexandrinus Τίς ό σoζόµευος πλούσιος (“Who of the rich will achieve salvation?”) 14.1. 7 Dig. 39.4.16.7; Lauffer 1971: 100ff., 114f., 148, 160, 167f., 194f., 197ff.; Sedov 1997: 375 with. ann. 19. - cf. Ziethen 1999b; Sidebotham & Wendrich 2001: 257.

Trade and merchandise is not only mentioned in the 3

Vasiliev 1933; Müller 1967; Raunig 1971: 137; Heyer 1971: 217ff.; Maraval 1996a: 1007-1029, 1091ff.; Maraval 1996c; Salamito 1996; Oesterle 1993; Troupeau 1994: 435-452.

380

Gabriele Ziethen & Erik Klingenberg: Merchants, pilgrims and soldiers on the Red Sea route mentioned many of these goods (Anastos 1946; WolskaConus 1968; Garzya 1992). It can be concluded from this fact that there was no major interruption of trade between the early Roman principate and late antiquity. However, the conditions of the market had changed by the C4th AD.8

As we have seen the economic situation of the Roman empire at the beginning of the C4th, primarily through Diocletian’s edict, was one of change. After Diocletian’s administrative reforms (297) the praefectus praetorio per Orientem was responsible for Egypt, the Libyan Pentapolis, the orient up-to the borders of Mesopotamia, huge parts of Asia Minor (including Cilicia, Cappadocia, Armenia, from Pamphylia to Trapzunt), Cyprus, the Cycladic islands (except the isles of Lemnos, Imbros and Samothrake) and Thracia. Following this reform, Constantine gave the order that the troops of the border regions had to be garrisoned in the cities. This decision imposed a high financial burden on the cities and reduced the military protection of the rural inhabitants. The final consequence was the abandonment of rural farmland and a certain dissatisfaction within the army (cf. Kuhnen 1994: 42). This led to the emperor’s mother, Helena, sometime after 326 having to placate the army by granting donations (donativa: Drijvers 1992: 65ff.). The final result was that the financial policy and the rising inflation survived Constantine’s death (Ermatinger 1996: 46f, 111f.). Zosimus (2.38) summarised the alarming situation.

At this time two young boys – Frumentius and Aedesius – partook in a voyage with an unexpected destination. Grillmeier and the Dombrowskis reviewed the written records and proposed a readable reconstruction of the boys’ itinerary and chronology at the beginning of the C4th. Frumentinus and Aedesius were accompanied by Meropius, an educator or member of the family. They started their voyage shortly before or approximately during the time of the change between the first and the second tetrarchy: that means c.305 or shortly after (Dombrowski & Dombrowski 1984; Grillmeier 1990: 304-314, esp. 307 ann. 28). The emperor Diocletian had just retired from his duties as a ruler or was preparing his withdrawal from politics.9 The tetrarchs succeeding him, especially Constantine, were fighting for nearly half a decade to establish their rule. Political and religious symbols – the vision of the true cross and the definition of the ruler’s legacy – changed step-by-step after Constantine became the victor at the Milvian Bridge (Weiss 1993; cf. Portmann 1989; Chantraine 1993). However, Diocletian’s administrative and economic reforms still were alive and followed during the following decades. It is obvious that Diocletian’s price edict from 301 had a negative influence on the trade of luxury goods. This meant that foreign trade with India, Arabia and Aksum dealing with rice, Indian turtles, drugs, ivory and the furs of leopards and lions was affected.10

The Expositio totius mundi showed the necessity to explore new sources of income and market. The text was written in the reign of Constantius II by a non-Christian author in Tyros - it is worth noting: the home of Aedesius. The Expositio totius mundi emphasised the wealth, the military strength and the efficiency of the Aksumites (Brakmann 1994: 21 on Expositio 17 (SC 124, 152); Drexhage 1983: 3-41). Heliodor (Aithiopika 4.16.6) mentions that in the early C3rd ships from Tyros transported, in addition to normal cargo, goods from India and the country of the Ethiopians across the Mediterranean to Carthago (Tertullian’s bishopric: Schöllgen 1984: Ziethen 1999a Anhang: Wirtschaftsleben (3); Ziethen 1999c.)

What was the result of such a policy of the maintenance and stimulation of the economy in the Red Sea region and effect upon the financiers of the merchants’ voyages? It can be assumed that these entrepreneurs were looking for new markets and the possibility of establishing trade stations outside the Roman customs and price area in order to avoid Roman administrators and retail trade tariffs (Antoniadis-Bibicou 1963: 71, 86ff., 91, 177, 194). During the years before Roman edicts of tolerance, an expansion of the religious mission outside the Roman borders seemed to be a tempting undertaking, as the example of the Manichaeans from Palmyra demonstrates. Manichaean merchants successfully organized trade along the route to India and were engaged by foreign rulers for administrative duties. Kosmas Indikopleustes describes a conversation between an Indian ruler and two Roman and Persian merchants.11

With this historical background we can review the journey of Frumentius and Aedesius. After their ship was captured the two boys were the sole survivors. They then spent more than 20 years living at the royal court of Aksum. When eventually the royal family offered them the chance to return home, Frumentius decided to start an ecclesiastical career in Alexandria after he was ordained a priest and later a bishop by Athanasius. Aedesius returned to Tyros, where he lived as a deacon.12 A review of their original travel route before they were imprisoned reveals some interesting facts. Originally the travellers decided to make their way via Egypt (eis

A-Thousand-And-One-Nights: Sindbad the Sailor was a descendent of a wealthy merchant family, he was a ship-owner, a shipwrecked sailor, an excellent administrator in foreign services and he possessed first-hand knowledge of the harbours along the sea-route from Arabia to India; Tubach 1995; Hendy 1985: 276ff.; Harl 1996: 290ff., 308, 314; Maraval 1996a: 1007ff.: Nath 1995; Allibert 1996: pass; Arabian Night’s Entertainments (1932): 367-433. 12 Dombrowski & Dombrowski 1984; Maraval 1996a: 1014-1019; for the legend of Fremenatos, see Heyer 1998: 23f.

8

Antoniadis-Bibicou 1963: 71, 86f., 91, 177, 194; Tubach 1995; Hendy 1985: 276ff.; Harl 1996: 290ff., 308, 314. 9 Diocletian’s antipathy against Christianity was a well-known fact. 10 Eusebius (v.C.IV.7) mentioned that Constantine initiated the public presentation of the diplomatic gifts brought by Indian, Blemmyan and Ethiopian envoys. cf. Patyal 1995; Laufer 1971: 100f., 130f., 148; Ziethen 1994: 166. 11 These close contacts were preserved as literary elements in the tale of

381

Limes XVIII

Aigypton). Here were to be found many markets and trade centres, where the goods from the Red Sea harbours were exchanged. There were also many opportunities to negotiate with private bank offices (Bogaert 1973: 249f., 259; 1997: 90, 123f.). Especially in late antiquity there existed many specialized opportunities for foreign exchange transactions and saving deposits. Although they were criticized by their colleagues, some Christian bishops, too, made use of these possibilities to invest capital.13

seen during the spread of early Islam, perhaps due to the fact that the Prophet Muhammad was a learned merchant, too.15 The results of the Berenike project show the existence and alteration of the harbours settlement, including early Christian buildings in the C5th/C6th. The transfer of goods from India, Arabia and the Sahara region was still alive at this time. The diversified traffic system of the wadis in the Eastern Desert of Egypt, including the ways to Edfu (Apollinopolis Magna) and Koptos (Qift) on the River Nile, demonstrates a similar situation (Sidebotham 1997: 385-393: Sidebotham & Wendrich 1996: 3, 441-452: 1998: 415-426; 1999; 2000: 414-419).

With Frumentius’s return to Aksum around 330, the real mission and christianization of Aksum began (Brakmann 1994: 28; Munro-Hay 1997: 61ff.), perhaps with Bible lectures in Greek and their translation into Ge’ez, similar to the pilgrimage practice in the Holy Land (Mulzer 1996). King Ezana’s decision to accept Christianity can be interpreted as a reaction to the trade policy of his Roman neighbours and as an attempt to dominate the control of the trade routes. And Ezana successfully made use of wellaccepted links with the Mediterranean.

The rise of pilgrimages in a Christian sense during the C4th required secure travel methods and accommodation: around during the same time the spread of religious legends began (Kötting 1950; Mulder 1993; cf. Schmidt 1989). The legend of the finding of the True Cross during Helena’s pilgrimage to the Holy Land between 312 and 325 (cf. Bieberstein 1993: 153ff.) was one such tale which became of extraordinary importance for the Aksumite church. As J.W. Drijvers argued, this legend, too, was spread for anti-Semitic propaganda and to distract from the political motivated assassinations in the Constantine family (Perrin 1996: 683-687; Drijvers 1992: 55-72, 142-145; Holum 1977/78).

Because of the independence of Aksum from Rome and Constantinople, another symbol of cultural and economic connection had to be created, which would be accepted by all trade partners. The high gold and silver content of the Aksumite currency was combined with a change of the written coin legends into bilingual texts (Greek and Ge’ez) and the replacement of the Semitic astral symbols by various types of Christian cross. This sophisticated mixture presented a currency of wide acceptance, which was guaranteed by a Christian ruler and his god. In addition the Aksumite king might have made use of the Roman example of currency. In this he was possibly encouraged by Christian traders. Following this example he could confirm his royal status, bring the currency into line with the Christian symbols in order to be accepted by the foreign exchange dealers and balance the precious metal currency in the case of a Roman ban on gold exports to the barbarian nations.14

The Christian pilgrimage itineraries contain information about the important harbour and garrison of Clysma (Kleopatris/Suez) and the military escorts during the pilgrims’ excursions to the holy places.16 Goods too from India and coconuts were available in Clysma (Mayerson 1995; 1996; Honigman & Ebied 1986: pass). Coconuts were found in Berenike, too (Sidebotham & Wendrich 1998: 314ff.; 1999: 303, 305). The fruit was a healthy food for the digestive system and useful in the prevention of scurvy in the sailors.17 During the C10th (C4th of the Higra), the Muslims estimated the value of the sand from Clysma because of its gold content.18

Finally we have to look to the harbours, which would be used by the merchants and captains. The literary and archaeological records show more details for the period of late antiquity. The mooring time in the harbours inside the Roman empire can be reconstructed approximately by the known tax customs. First a certain time passed from the unloading and the public announcement of goods; secondly tax exemptions are known, if the mooring time was less than 20 or 30 days (Jakab 1997: 25; Engelmann & Knibbe 1989). Merchants, who made every effort to ensure that their customers were well looked-after visited them at their homes. This was a very successful tactic of sales management. Missionaries, too, could have made use of a like strategy to win their subjects. Similar behaviour can be

Christian entrepreneurs, such as Kosmas Indikopleustes, who preferred the harbour of Adulis (Eritrea), and the wealthy Christian families of the Arabian city of Najran became the beneficiaries of these markets. They obtained control of the trade routes to south Arabia and Yemen, 15 In addition to this, the Arabic letters from Quseir (C13th/C7th AH) show the organization of this trade and the structure of merchants’ families; Guo 1999; Cf. Labib 1965: 226ff.; Patyal 1995. 16 Examples of the inner structure of a pilgrimage centre are known from Abu Mena near Alexandria in Egypt; Mulder 1993; Ziethen 2000: 31 ann. 57f. 17 Itin. Anton. 41: Illic accepimus nuces plenas virides, quae de India veniunt, quas de paradiso credunt homines esse.Cuius gratia talis est: quanticumque gustaverint, satiantur. Lauffer 1971: 114 Nr. 50; Majupuria & Joshi 1989: 132-137; Zoller & Nordwig 1997: 244f.; Ziethen 2000: 24, 31. 18 Amar 1997. The business letters from Quseir on the Egyptian Red Sea coast in the C13th/C7th of the Higra show a similar tradition of former trade centres; Guo 1999.

13

Cyprianus De lapsis (“About the fallen believers”) 6; Altaner & Stuiber 1978/1993: 172f., 176. 14 Hahn 1993; 1999; 2000a/b/c); Howgego 1990: 21; 23 on Codex Theosianus 9.23.1 (353 or 354) and CJ 4.63.2 (374/5); Philippson 1998: 71-74.

382

Gabriele Ziethen & Erik Klingenberg: Merchants, pilgrims and soldiers on the Red Sea route where Aksum also focused its economic ambitions.19 Russian investigations document the stay of foreign merchants in Qani (from graffiti) and the existence of Aksumite hand made pottery (Sedov 1997: 375ff.; Sidebotham & Wendrich 2000: 4-19).

Antoniadis-Bibicou H. 1963 Recherches sur les Douanes à Byzances. L’“octava”, le “kommercion” et les commerciaires. (Paris). Arabian Night’s Entertainments 1932 The Arabian Night’s Entertainments or the Book of a Thousand Nights and a Night. A selection of the most famous and representative of these tales from the plain and literal translation by Richard F. Burton. The stories have been chosen and arranged by B. A. Cerf and are reprinted complete and unabridged with many of Burton’s notes (New York, The Modern Library) 367-433 (= Burton vol 6/1). Arethas = ΜΑΡΤΥΡΙΟΝ ΤΟΥ ΑΓΙΟΣ ΑΡΕΘΑ ΚΑΙ ΤΗΣ ΣΥΝΟ∆ΙΑΣ ΑΥΤΟΥ ΕΝ ΝΕΓΡΑ ΤΗ ΠΟΛΕΙ: Anekdota Graeca e codicibus regiis descripsit annotatione illustravit J. Fr. Boissonade 5 (Nachdr. Hildesheim 1962). Baumann P. 1999 Spätantike Stifter im Heiligen Land. Darstellungen und Inschriften auf Bodenmosaiken in Kirchen, Synagogen und Privathäusern. In B. Brenk, J.G. Deckers, A. Effenberger, L. Kötzsche (edd.) Spätantike - Frühes Christentum - Byzanz. Kunst im Ersten Jahrtausend. Studien und Perspektiven, Bd. 5. (Wiesbaden). Beckert E. & Breuer G. 1991 Öffentliches Seerecht. (Berlin). Bieberstein K. 1993 Der Gesandtenaustausch zwischen Karl dem Großen und Hārūn ar Raŝîd und seine Bedeutung für die Kirchen Jerusalems. Zeitschrift des Deutschen Palästina-Verein 109: 152-173. Bogaert R. 1973 Changeurs banquiers chez les Pères de l’Église. Ancient Society 4: 239-270. Bogaert R. 1997 La Banque en Égypte Byzantine. Zeitschrift für Papyrologie und Epigraphik 116: 85-140. Brakmann H. 1994 ΤΟ ΠΑΡΑ ΤΟΙΣ ΒΑΡΒΑΡΟΙΣ ΕΡΓΟΝ ΘΕΙΟΝ. Die Einwurzelung der Kirche im spätantiken Reich von Aksum (Bonn). Brown S. 1994 The international law of the sea 2. Documents, cases and tables. (Aldershot, USA). Casson L. 1989 The Periplus Maris Erythraei. Text with introduction, translation, and commentary. (Princeton). Chantraine H. 1993 Die Kreuzesvision von 351 - Fakten und Probleme. Byzantin. Zeitschrift 94: 430-441. Clemens Alexandrinus Paidagaogos = Übersetzung: Bibliothek der Kirchenväter 2.8.2 (München 1934). Colombos C.J. 1963 Internationales Seerecht. (München). Cyprian De lapis = Übersetzung: Bibliothek der Kirchenväter 34 (München 1918). Dig. = The Digest of Justinian. Latin text edited by T. Mommsen with P. Krueger, English translation edited by Alan Watson, 4 vols. (Philadelphia 1985). Dombrowski B.W.W. & Dombrowski F. 1984 Frumentius/Abbā Salāmā: Zu den Nachrichten über die Anfä nge des Christentums in Äthiopien. Oriens Christianus 68: 114169. Drexhage H.-J. 1983 Die ‘Expositio totius mundi et gentium’. Eine Handelsgeographie aus dem 4. Jahrhundert n. Chr. eingeleitet, übersetzt und mit einführender Literatur (Kap. XXII-LXVII) versehen. Münstersche Beitr. Antike Handelsgesch. 2/1: 3-41. Drijvers J.W. 1992 Helena Augusta. The Mother of Constantine the Great and the legend of her finding of the True Cross. (Leiden). Dupuy R.J. & /Vignes D. 1985 Traité du Nouveau Droit de la Mer. (Paris). EI = The Encyclopedia of Islam. (1st & 2nd edit). Engelmann H. & Knibbe D. 1986 Das Zollgesetz der Provinz

It seems possible that by the end of the C5th or the beginning of the C6th the Jewish communities felt surrounded by the Christian centres in the north and the Aksumite controlled harbours in the south of the Arabian peninsula (Maigret 1997: 324; Miller 1969; Faure 1990; Paszthory 1990; Pfeifer 1997). So the persecution in Najran might have been motivated by an attempt to eliminate wealthy Christian high society and to occupy the precious metal and money depots which were the securities for trade credits (Arethas; Pigulewskaya 1969). The wealth of Ruhaima and her husband in Najran are one example for the lifestyle of such Christian families (Shahîd 1971: 149f.; 1993; Heyer 1998; 33-37; Pigulewskaya 1969: 175-307). King Caleb from Aksum felt encouraged and accepted by the government in Constantinople to undertake the protection of Arabian Christians in the first half of the C6th. A well-equipped fleet, recruited from the important harbours of the Red Sea region set sail to Arabia. There is even some possibility that monks manned the warships’ oars. The final result was that the envy of the trade situation led to a religious war (Shahîd 1964; 1993; Grillmeier 1990: 314-330; cf. Kaegi 1981: 148f.). The changing situation of political influence after 524 and the Persian conquest of Yemen between 570-572 produced two losers: Aksum lost Yemen and the Byzantine fleet lost its control over the Red Sea routes. A new period of oriental naval power anticipated the way for early Muslim traders some decades later.20 Bibliography Ahrweiler H. 1966 Byzance et la mer. La marine de guerre, la politique et les institutions maritimes de Byzance aux VIIe - XVe siècles. (Paris). Allibert C. 1996 Voyage aux “confins des mers du monde”. Le commerce arabe avec l’Asie du sud-est à travers les sept voyages de Sindbad le marin. Topoi 6/2: 841-852. Altaner B. & Stuiber A. 1978 Patrologie. Leben, Schriften und Lehre der Kirchenväter. (Freiburg i. Br. = Nachdr. 1993). Amar Z. 1997 Gold production in the ‛Arabah Valley in the tenth century. Israel Exploration Journal 47/1-2: 100103. Anastos M.V. 1946 The Alexandrian origin of the Christian topography of Cosmas Indicopleustes. Dumbarton Oaks Papers 3: 73-80. 19

Kosmas Indikopleustes 2.48f.; 54; 56; 63 and 3.66 (Adulis). Hahn 2000c; Pigulewskaya 1969; Maraval 1996b; 1093. 20 Very late traces of the Persian interest in Yemen can be found in the modern penal law of the Islamic Republic of Iran (the revised version from AD1991): the expiation for the crime of manslaughter can be paid with 200 faultless Yemenite robes (“zweihundert fehlerfreien jemenitischen Gewändern” in German translation by Tellenbach 1995: 96).

383

Limes XVIII

Asia. Eine Inschrift aus Ephesos. Epigraphica Anatolica 15: 1989, pass. Ermatinger J.W. 1996 The economic reforms of Diocletian. Pharos 7. (St. Katharinen). Faure P. 1990 Magie der Düfte. Eine Kulturgeschichte der Wohlgerüche – Von den Pharaonen zu den Römern. (München/Zürich). Freerich G. 1982 Die lautlose Eroberung der Meere, ed. Deutsches Marine Institut. (München). Garzya (1992) = Cosma Indicopleusta, Topographia Cristiana, libri 1-5, a cura di A. Garzya (Napoli). Grillmeier A. 1990 Jesus der Christus im Glauben der Kirche 2,4: Die Kirche von Alexandrien mit Nubien und Äthiopien nach 451. (Freiburg i. Br.). Gruszka P. 1981 Die Stellungnahme der Kirchenväter Kappadoziens zu der Gier nach Gold, Silber und anderen Luxuswaren im täglichen Leben der Oberschichten des 4. Jahrhunderts. Klio 63: 661-668. Guo Li 1999 Arabic documents from the Red Sea port of Quseir in the seventh/thirteenth century, Part 1: Business letters. Journal of Near Eastern Studies 58/3: 161-190. Hahn W. 1983 Die Münzprägung des axumitischen Reiches. Litterae Numismaticae Vindobonensis 2 (Wien): 113180. Hahn W. 1994-99 Touto Arese Te Chora - St. Cyril’s Holy Cross cult in Jerusalem and Aksumite coin typology. The Israel Numismatic Journal 13: 103-117. Hahn W. 2000a Diener des Kreuzes. Zur christlichen Münztypologie der Könige von Abessinien in spätantiker Zeit. Money Trend 32/6: 58-63. Hahn W. 2000b Von der Münze zum Schmuck und zurück Montierung und Demontierung von Henkeln an spätrömischen und axumitischen Beispielen. Money Trend 32/9: 56-58. Hahn W. 2000c Eine numismatische Spurensuche im alten Jemen - vom axumitischen Okelis zum türkischen Scheich Said. Money Trend 32/9: 58-63. Harl K.W. 1996 Coinage in the Roman economy, 300B.C. to AD700. (Baltimore). Hendy M.F. 1985 Studies in the Byzantine monetary economy c.300-1450. (Cambridge). Heyer F. 1971 Die Kirche Äthiopiens. Eine Bestandsaufnahme. (Berlin). Heyer F. 1998 Die Heiligen der Äthiopischen Erde. Oikonomia 37. (Erlangen). Hieron. ep. = des Heiligen Kirchenvaters Eusebius Hieronymus ausgewählte Briefe; aus dem lateinischen übersetzt von L. Schade. Bibliothek der Kirchenväter Rh. 2, Bd. 16 (München 1936). Holum K.G. 1977/78 Pulcheria’s Crusade AD 421-22 and the ideology of imperial victory. Greek, Roman & Byzantine Studies 18: 153-172. Honigman E. & Ebied R.Y. 1986 al-Ķulzum. In EI² 5 (Leiden): 367-369. Howgego C. J. 1990 Why did ancient states strike coins ? The Numismatic Chronicle 150: 1-25. Ibrahim J. 1991 Kulturgeschichtliche Wortforschung. Persisches Lehngut in europäischen Sprachen. (Wiesbaden). Jakab E. 1997 Praedicare und cavere beim Marktkauf. Sachmängel im griechischen und römischen Recht. Münchner Beitr. Papyrusforsch. u. antike Rechtsgesch. 87. (München). Kaegi jnr. W.E. 1981 Byzantine military unrest 471-843. An interpretation. (Amsterdam). Kötting B. 1950 Peregrinatio Religiosa. Wallfahrten in der

Antike und das Pilgerwesen in der alten Kirche. Forschungen zur Volkskunde: 33-35. Kuhnen H.-P. 1994 Kirche, Landwirtschaft und Flüchtlingssilber. Zur wirtschaftlichen Entwicklung Palästinas in der Spätantike. Zeitschrift der Deutschen Palästina-Verein 110: 36-50. Labib S.Y. 1965 Handelsgeschichte Ägyptens im Spätmittelalter (1171-1571). Beih. Vierteljahresschr. Sozial. u. Wirtschaftsgesch. 46. (Wiesbaden). Lauffer S. 1971 Diokletians Preisedikt. Texte u. Kommentare 5. (Berlin). Maigret A. de 1997 The frankincense road from Najrān to Ma’ān: A hypothetical itinery. In Profumi d’Arabia. Attil del convegno, a cura di Alessandra Avanzini. Saggi di storia antica 11. (Roma): 315-331. Majupuria T.C. 1989 Religious & useful plants of Nepal & India. Medicinal plants and flowers as mentioned in religious myths and legends of Hinduism and Buddhism. (Rev. by D. P. Joshi (Bangkok). Maraval P. 1996a Alexandrien und Ägypten. In J.M. Mayeur et al (Hrsg.) Die Geschichte des Christentums. Religion. Politik. Kultur 2. (Freiburg i. Br.): 1007-1029. Maraval P. 1996b Die neuen Grenzen. In J.M. Mayeur et al (Hrsg.) Die Geschichte des Christentums. Religion. Politik. Kultur 2. (Freiburg i. Br.): 1076-1095. Maraval P. 1996c Das Mönchtum und der Osten. In J.M. Mayeur et al (Hrsg.) Die Geschichte des Christentums. Religion. Politik. Kultur 2. (Freiburg i. Br.): 816-847. Mayerson P. 1995 Aelius Gallus at Cleopatris (Suez) and the Red Sea. Greek, Roman & Byzantine Studies 31: 17-24. Mayerson P. 1996 The port of Clysma (Suez) in transition from Roman to Arab rule. Journal of Near Eastern Studies 55/2: 119-126. Miller J.I. 1969 The spice trade of the Roman empire 29 BC to AD 641. (Oxford). Mulder N.F. 1993 Abu Mena: How the early Christian pilgrimage was given shape. Boreas 16: 149-164. Müller C.D.G. 1967 Kirche und Mission unter den Arabern in vorislamischer Zeit. Sammlung gemeinverständlicher Vorträge und Schriften aus dem Gebiet der Theologie und Religionsgeschichte 249. (Tübingen). Müller W.W. 1997 Namen von Aromata im antiken Südarabien. In Profumi d’Arabia. Attil del convegno, a cura di Alessandra Avanzini. Saggi di storia antica 11. (Roma): 193-210. Mulzer M. 1996 Mit der Bibel in der Hand ? Egeria und ihr ‘Codex’. Zeitschrift der Deutschen Palästina-Verein 112: 156-164. Munro-Hay S.C. 1997 Ethiopia and Alexandria. The Metropolitan Episcopacy and Ethiopia. Bibliotheca nubica et aethiopica 5. (Warszawa). Nath A. 1995 Antiquities of Graeco-Roman affinity from Adam. An island mart of central India. East and West 45: 149-171. Oesterle H.J. 1993 Antonius von Koma und die Ursprünge des Mönchtums. Archiv Kirchengesch. 75: 1-18. Paszthory E. 1990 Salben, Schminken und Parfüme im Altertum. Herstellungsmethoden und Anwendungsbeispiele im östlichen Mediterraneum. Antike Welt 21 Sondernr. (Mainz). Patyal H. C. 1995 Tortoise in mythology and ritual. East and West 45: 97-107. Perrin M.-Y. 1996 Die neue Form der Missionierung. In J.M. Mayeur et al (Hrsg.) Die Geschichte des Christentums. Religion. Politik. Kultur 2. (Freiburg i. Br.): 667-704.

384

Gabriele Ziethen & Erik Klingenberg: Merchants, pilgrims and soldiers on the Red Sea route Pfeifer M. 1997 Der Weihrauch - Geschichte, Bedeutung, Verwendung (Regensburg). Philippson D.W. 1998 Ancient Ethiopia. Aksum: its antecedents and sucessors. (London). Pigulewskaja N. 1969 Byzanz auf den Wegen nach Indien. Aus der Geschichte des byzantinischen Handels mit dem Orient vom 4. bis 6. Jahrhundert. (Berlin). Portmann W. 1989 Die 59. Rede des Libanios und das Datum der Schlacht von Singara. Byzantin. Zeitschrift 82: 1-18. Raunig W. 1971 Bernstein - Weihrauch - Seide, Waren und Wege der antiken Welt. (Wien). Ryckmans J., Müller W.W. & Abdallah Y.M. 1994 Textes du Yémen antique inscrits sur bois. Publications de l’Institut Orientaliste de Louvain 43. (Louvain-laNeuve). Salamito J.-M. 1996 Christianisierung und Neuordnung des gesellschaftlichen Lebens. In J.M. Mayeur et al (Hrsg.) Die Geschichte des Christentums. Religion. Politik. Kultur 2. (Freiburg i. Br.): 768-815. Schmidt N. 1989 Sinai und Rotes Meer. Reise-Handbuch, bearb. von P. O. Scholz. (Köln2). Schöllgen G. 1984 Ecclesia sordida? Zur Frage der sozialen Schichtung frühchristlicher Gemeinden am Beispiel Karthagos zur zeit Tertullians. Jahrb. Antike u. Christentum Ergbd. 12. (Münster). Sedov A.V. 1997 Sea-trade of the Hadramwt kingdom from the 1st to the 6th centuries AD. In Profumi d’Arabia. Attil del convegno, a cura di Alessandra Avanzini. Saggi di storia antica 11. (Roma): 365-383. Shahîd I. 1964 Byzantino-Arabica. The conference of Ramla, AD524. Journal of Near Eastern Studies 23: 115-131. Shahîd I. 1971 The martyrs of Najrân. New documents. Subsidia Hagiographica 49 (Bruxelles). Shahîd I. 1993 Nadjrān. In EI² 7 (Leiden): 871-872. Sidebotham S.E. 1997 Caravans across the Eastern Desert of Egypt: Recent discoveries on the BerenikeApollonopolis-Magna-Coptos roads. In Profumi d’Arabia. Attil del convegno, a cura di Alessandra Avanzini. Saggi di storia antica 11. (Roma): 385-393. Sidebotham S. & Wendrich W. 1996 Berenike ‘95. Preliminary report of the excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden). Sidebotham S. & Wendrich W. 1998 Berenike ‘96. Report of the excavations at Berenike (Egyptian Red Sea coast) and the survey of the Eastern Desert. (Leiden). Sidebotham S.E. & Wendrich W.Z. 1999 Berenike 1997. Report of the 1997 excavations at Berenike and the survey of the Egyptian Eastern Desert, including excavations at Shenshef. (Leiden). Sidebotham S. E. & Wendrich W.Z. 2000 Berenike 1998. Report of the 1998 excavations at Berenike and the survey of the Egyptian Eastern desert, including excavations in Wadi Kalalat. (Ledien). Sidebotham S.E. & Wendrich W. Z. 2001 Berenike, Roms Tor am Roten Meer nach Arabien und Indien. Antike Welt 32/3: 251-263. Tellenbach S. 1996 Strafgesetze der Islamischen Republik Iran. Sammlung außerdeutscher Strafgesetzbücher in deutscher Übersetzung 156. (Berlin). Troupeau G. 1994 Kirchen und Christen im muslimischen Orient. In J.M. Mayeur et al (Hrsg.) Die Geschichte des Christentums. Religion. Politik. Kultur 4. (Freiburg i. Br.): 391-472.

Tubach J. 1995 Mani und der palmyrenische Kaufmann aus Forat. Zeitschrift für Papyrologie und Eprigraphik 106: 165-169. Vasiliev A. 1933 Justin I. (518-527) and Abyssinia. Byzantin. Zeitschrift. 33: 67-77. Weiss P. 1993 Die Vision Constantins. In J. Bleicken (ed.) Colloquium aus Anlaß des 80. Geburtstages von Alfred Heuss. Frankfurter Althist. Stud. (Kallmünz): 143-169. Western Arabia and the Red Sea (1946) = Western Arabia and the Red Sea. Geographical Handbook Series, ed. by. Naval Intelligence Division [British Navy]. (Oxford 1946). Wolska-Conus W. 1968 La topographie chrétienne des Cosmas Indicopleustes. Sources chrétiennes 141, 159, 197. (Paris). Ziethen G. 1994 Legationes Externae in der frührömischen Kaiserzeit: INDI-ΑΙΘΙΟΠΕΣ-ΣΗΡΕΣ. Nubica 3/1: 141197. Ziethen G. 1999a Heliodor’s Aithiopika und die Gesandtschaften zu den Aithiopen. Klio 81/2: 490-525. Ziethen G. 1999b Nicht nur Weihrauch … In Von Ananas bis Zimt - Tropische Nutzpflanzen. Ausstellungskat. Palmengarten Frankfurt a. M., 4.6.-26.9. 1999 (Frankfurt a.M.): 11-19; 97-99. Ziethen G. 1999c Günstige Winde, sichere Häfen - Zur Darstellung des Handels in Heliodors Aithiopika. In Äthiopien und seine Nachbarn/Ethiopia and its Neighbours. 3. Wissenschaftliche Tagung des Orbis Aethiopicus, Gniew 25.-29.9.1997, hrsg. vom Muzeum Archeologicne w Gdańsku und Orbis Aethiopicus. (Gdańsk/Frankfurt a.M): 93-120. Ziethen G. 2000 Fromme Passagiere - Handel und Mission im Raum des Roten Meeres. Mare Erythraeum 3 (München): 19-32. Zoller A. & Nordwig H. 1997 Heilpflanzen der Ayurvedischen Medizin. Ein praktisches Handbuch. (Heidelberg).

385

Limes XVIII

386

The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL) Harry van Enckevort1 excavated areas on the south-east part of the plateau. Most of the results are briefly published in a long list of short archaeological contributions, especially in the annual reports of the ROB.3 In 1997-1999 the Bureau of Archaeology of the Municipality of Nijmegen commenced excavations on the plateau. The results of the excavations of both archaeological authorities are combined in the following brief overview of the eastern part of the canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen. After that I will go into further into detail regarding the forum.

The town of Nijmegen is situated at the transition from the holocene river delta to the pleistocene valleys of the Rhine and Meuse. The strategic location of the Hunerberg and the Kops Plateau, two plateaux on the high ice-pushed ridge in the eastern part of the town overlooking the river area, together with the easy connections to the south, have especially contributed to the importance of Nijmegen. Around 16BC the Roman army built a legionary camp on the Hunerberg. This camp, with a surface area of 42ha, seems to have been deserted 5 or 6 years later. At that time a new and smaller fort was built on the nearby Kops Plateau, to the east of the Hunerberg. This fort was rebuilt several times, its maximum size being 4.5ha. In the earliest phases it was a base-camp of a high ranking officer, possibly Drusus himself (and later on his successors). Features and finds from Claudian-Neronian times indicate the presence of cavalry, probably the ala Batavorum. Shortly before or during the Batavian Revolt in 69-70, the plateau was abandoned. After defeating the rebels, legio X built a new camp on the Hunerberg with a surface area of 16ha. Soon the castra formed the centre of a large agglomeration (Fig. 1) to which the surrounding canabae legionis and the Roman town in the western part of Nijmegen also belong. The originally wooden castra was replaced in the last decade of the first century by a stone one. Shortly after the renewal of the castra, about AD104, legio X left for the Balkans. Until the middle of the C2nd the vexillatio Britannica and units of the legio VIIII Hispana and the legio XXX Ulpia Victrix were successively stationed on the Hunerberg.2

The canabae legionis is partially built over the older structures of pre-Flavian times. The earliest large-scale excavations were undertaken here in 1972 by Tom Bloemers, at that time working for the ROB. The excavators found (on a building site which was called ‘terrein Schuller’ in the north-west corner of the plateau) six smaller buildings and a horreum 5 x 10m in size (Fig. 2D) from the Flavian period, aside from prehistoric and early Roman traces. They also found a kind of wooden cistern (140 x 100 x 30cms). The seams between the planks and the corners were sealed by lead slabs and the exterior was made watertight by covering it with clay. The cistern was connected to a pit by a lead pipe. The function of this structure is unknown but it seems that it was used for artisanal activities (Bloemers 1972; 1973; 1979: 53, fig. 54). Indications for these kinds of activities were also found in other places in the vicinity. In 1973 the excavators found, just north of the forum (Fig. 2C), a cellar-like structure built up of 17 layers of tegulae which were divided in length by halves. Over 250 pieces with stamps of the legio X Gemina and the vexillatio Britannica indicated that the soldiers were involved in several activities in the canabae. Inside the structure, a thick black ash layer with pieces of stone was found which suggests this was an oven, used to smoke meat (Bloemers 1979: 53, fig. 53; Willems 1990b: 55-56). A year before a large quantity of broken glass was found in a few pits. The sherds, scorries and wasters indicate that a craftsman worked in a glasshouse in the neighbourhood (Bloemers 1972; Isings 1980). Besides indications for other artisinal activities and several buildings found in 1981, indications of a small amphitheatre (Fig. 2B) were found on the eastern slope of the small valley through which the Beekmansdalseweg leads from the plateau to the river valley. This valley is also the north-western border of the Kops Plateau (Bloemers 1983: 31-32; 1985: 31-32).

The Romans started to build houses, shops and workshops in the neighbourhood of the castra shortly after 70. In the following years the surface of the canabae legionis grew to at least 40ha. Between 1987 and 1997, archaeologists of the Catholic University of Nijmegen worked extensively in the western part of the canabae (Haalebos 1995; 1998a). We await the appearance of detailed studies with the final results of these excavations in the next few years. Concerning the southern part of the canabae legionis, we know only the location of the amphitheatre (Bloemers 1980b: 45-46; 1980c: 473). Regarding the eastern part, on the Kops Plateau, extensive areas were excavated between 1972 and 1995 by the ROB (Rijksdienst voor het Oudheidkundig Bodemonderzoek - The Dutch State Archaeological Service). Figure 2 gives an overview of the 1

I am indebted to Tilmann Bechert (Abteilung Stadtarchäologie Stadt Duisburg) for the photo, Roel Hoek (Bureau Archeologie Gemeente Nijmegen) for ideas about the forum, Rob Mols (Bureau Archeologie Gemeente Nijmegen) for the drawings and Glenn Storey (University of Iowa, Iowa City, USA) for the text corrections. 2 For more information about both the Augustan and Flavian castra and canabae legionis on the nearby Hunerberg and the early-Roman fortress on the Kops Plateau, I refer to Haalebos 1995; 1998b; van Enckevort 1997; van Enckevort, Haalebos & Thijssen 2000.

The layout of the canabae was in broad outline determined by the main east-west road on the plateau, the continuation of the via principalis of the castra. Two phases of the road 3

See literature list for the publications of Bloemers 1972 etc; 1973 etc.; Willems 1987 etc; 1990a etc; van Enckevort 1993 etc; Zee 1998.

387

Limes XVIII

were found. The oldest had a width of 12m. The most recent, which was very well preserved where the road surface of gravel was still intact in several places, was 15m wide.4 In a small area in the middle of the plateau south of the road, many pits and fences were found but no indications of buildings.5 While excavating the early Roman fortress on the eastern part of the Kops Plateau between 1986 and 1995, directed by Willem Willems of the ROB, again parts of the eastern canabae legionis were excavated. Traces of at least six buildings from the Flavian period were identified among many thousands of features. Five of them are concentrated on both sides of the road on the plateau leading from the castra to the east. One building, probably a roadstation or stabulum (Fig. 2F), was also found just north of the Berg en Dalseweg, a modern road with Roman or even prehistoric roots. I am certain there were more Flavian buildings on the plateau, but we had not the time to work out their locations from the excavations drawings. When we look to the distribution map of the Flavian coins (Fig. 3) we see that coins in particular were lost on both sides of the roads. Comparing Figs. 2 and 3, we see that this distribution corresponds broadly speaking with the locations of the buildings and the cemeteries.

especially the western part which was excavated in 19971999, were heavily damaged in the past (Fig. 5). The excavation in those years was situated on a building site. Until 1997, houses of a working class neighbourhood called ‘Het Rooie Dorp’ (the “Red Village”) stood on the location. These houses were built in the 1820 and the housing association ‘De Gezonde Woning’ thought it was time to replace them. So between 1997 and 1999, the Bureau of Archaeology was forced to undertake large-scale excavations after the demolition of the old houses. Besides some Neolithic and Bronze age graves we found many traces of the canabae legionis. The investigations were handicapped by extensive disturbances which were over 100 years old.9 At that time, large quarries were dug for sand and pebbles. Some years later, just before the building of the neighbourhood, these quarries were filled up with sand from the close nearby. This meant that over a large surrounding area at least the upper 50cms of the topsoil was removed for incorporation into the old quarries. In addition, the building of the houses and activities of the inhabitants of the district caused a lot of damage to the archaeological features. Altogether we can say that, besides the oldest features, probably also the majority of the expected finds (coins, militaria, sherds, etc.) which were thrown away or lost in Roman times and which could give us some clues about the function of the forum are missing. Moreover, we lack also sufficient information on the internal construction of the forum and its courtyard.

During the investigations of the ROB in 1973-1979 behind the State Institute ‘De Hunerberg’6 the excavators found the 1m wide robber trenches (Fig. 4) of the broken up foundations of a huge building south of the road and in the vicinity of the eastern gate of the castra, the porta principalis dextra.7 Bloemers thought he had found the eastern part of a market hall, a forum. Because one of the foundation trenches in the northern wing bends inward (Figs. 5 & 6), Bloemers thought at that time that this was the location of the main entrance of the forum and he estimated the dimensions of the building as 100 x 174m (Bloemers 1978: 34). These measurements were used for reconstruction drawings in several publications.8 During a visit to Nijmegen, Bloemers showed the features of the eastern wing of the forum to the participants of the Xth Roman Frontier Studies Congress in Xanten in September 1974 (Fig. 4). Thanks to recent excavations by the Bureau of Archaeology of the Municipality of Nijmegen in 19971999, we now know that the forum was significantly larger than Bloemers thought.

The dimensions of the rectangular complex are 166.25 by 132.75m. These measurements immediately suggest that the Roman architects used the pes Drusianus (one foot is 33.319cms) when they set out the outlines of the building because the length is about 500ft and the width is about 400ft (Fig. 6). It is the largest building we know of from Roman times in The Netherlands. Due to the fact that the ruins of the building in the late Roman period and in the later middle ages were used as a stone quarry, and the aforementioned digging activities around 1900, almost no building material was left. Only a few parts of the so-called “Grauwacke Schiefer” were found in situ, which indicates that this kind of stone was used for the foundations. The central courtyard is on all sides bordered by three parallel foundations. That suggests we are dealing with long rows of chambers with an open portico facing the courtyard. The depth of the western and eastern wings, chambers, and porticos, is about 13.25m (40ft). The ratio of porticos to chambers is 2 to 3, which calls to mind the ratio Vitruvius prescribes for the width and length of for a (Vitruvius V.1.2). The depth of the front and back wing is about 11m (33ft). Only in the back wing do we have traces of the dividing walls of these chambers. The absence of internal partitioning and structures in the eastern, western, and northern wing is probably due to the already mentioned turn of the C20th digging activities and the shallow foundation construction techniques. In the chambers in the eastern wing four rows of heavy postholes

The excavations made it clear that large areas of the forum, 4

Bloemers 1974: 180; 1975: 16; 1975c: 21; 1978: 18. See also van Enckevort & Zee 1996: 61 for a photo of the intact road surface of gravel of the same road on the eastern part of the plateau. 5 Bloemers 1974: 180-181; 1975: 17; 1975c: 22-24; 1977: 20. In the same area in the middle of Fig. 2 and south of the extension of the via principalis and on both sides of the north-south road which leads to the Berg en Dalseweg, Holwerda 1943 thought he had found the so-called Batavian village. 6 Now called De Hunnerberg. 7 Bloemers 1974: 180; 1975a; 1975b: 17; 1975c: 23-24; 1976: 25-26 1979: 51, fig. 50.3; 1980b: 48; 1985: 31). See also Willems 1990b: 54-55. 8 See for instance Bloemers, Louwe Kooimans & Sarfatij 1981: 88; Willems 1990b: 49; van Enckevort & Zee 1996: 70 for different reconstructions.

9

388

See Fig. 5. Only the largest and deepest disturbances are drawn.

Harry van Enckevort: The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL) survived (Figs. 4 & 6).10 The distance between the rows is about 1.6m (5ft). To the extent that they can be securely attributed to the building, they functioned, according to Bloemers, as the grounding supports for bearing the construction of a wooden floor in the chambers. But he also suggests that the small deviation in the direction of the rows (Fig. 6), compared to the walls, suggests that they belong to an older period. The portico columns may have rested upon low sleeper walls. During the excavations we found some limestone fragments of the portico columns.

In two places of the courtyard we found interesting brick foundations, both precisely 100ft from the inner wall of the porticos. Although we found no direct evidence for this, we presume that huge bronze statues once stood on these foundations. We know very well that these kinds of lifesize statues were erected in the environs of Nijmegen from different excavations in the town of Ulpia Noviomagus and from the surrounding Batavian settlements in the vicinity. At the moment we have about 30 relatively small fragments of such statues in the collections of the Bureau Archaeology.

There are no indications of the main entrance along the front of the forum, but it seems certain that the opening of the complex lay in the northern wing, onto the road coming out of the porta principalis dextra. The fact that the middle trench is bent inward was for Bloemers in the 1970s the reason to suggest that the entrance was once located on this spot. However, we now know from the recent excavations in the “Red Village” that this location is not in the middle of the front side of the forum. There are no other indications for an entrance.

Behind the back wing of the forum we found a broken up foundation extending over the whole width of the forum, which indicates that there was a large hall at a depth of just over 13m (40ft), the same depth as the eastern and western wing of the forum. Because of the digging activities in the recent past, no indications for the internal structures of this back hall survive. In general we can compare the forum from the Kops Plateau with the forum found just outside the castra in Carnuntum and Vindonissa (Kandler & Vetters 1986: 226; Hartmann 1986: 103-105). With its length of 138m and width of 122m, the latter is slightly smaller than the Nijmegen forum. The courtyard in the forum of Vindonissa was also surrounded by a portico, but only the front and backside had a row of chambers behind the portico. The forum in Carnuntum is on the other hand larger (225.6 by 182m).

During the excavations in the 1970s, Bloemers found over 20 rows of small postholes with a length of almost 105m in the eastern part of the courtyard. He suggests they belonged to the foundation of extraordinarily long storehouses or horrea. (Bloemers 1976: 26; 1980b: 473). During the excavations of 1997-1999, we discovered heavily disturbed traces of these rows also in the western part of the courtyard. We have the impression that the distance between the primary rows is, again, about 1.6m (5ft), but secondary postholes are found almost everywhere between these rows. Although only a relatively few postholes survived in the western part of the courtyard, we now know that an area of 10,179sq m in the centre of the courtyard (Fig. 6) was once covered with over 50,000 posts. The distance between the rows with postholes and the foundation of the western and eastern porticos is about 5m (15ft); the distance between the postholes and the foundation of the northern and southern porticos is about 10m (30ft). I think that when we convert these figures to Roman measurements, it shows that these rows with postholes are contemporary with the forum, but they probably did not function as the foundation of storehouses. In my opinion, they are part of a foundation for a gigantic platform or wooden floor in the courtyard. This floor was probably laid level to prevent erosional problems that might arise due to the weak slope on which the forum was built. Besides, this feature of the floor prevented problems with muddy shoes. Maybe the small zone around the floor or platform was not provided with a wooden floor because this was the dripping zone off the roof of the portico. Although we found no indication for it, it is possible that this zone was covered with gravel or something else for the convenience of the people who wanted to visit the shopkeepers and the craftsmen in their chambers directly from the courtyard.

10

We think that the forum on the plateau probably was built shortly after AD90, a period when large parts of the castra were rebuilt in stone. The size and the location of the forum suggests that it had an important function for the soldiers in the castra and the inhabitants of the canabae legionis. It is even possible that the forum was meant as the marketplace for the inhabitants of the Roman town in the western part of Nijmegen or even for all Batavians living in the civitas Batavorum. When Trajan became emperor in 98 he stayed in Germania Inferior. There are several indications he reformed the limes zone and the structure of the civitas Batavorum to integrate the Batavians into the Roman empire (van Enckevort, Haalebos & Thijssen 2000: 71-73; van Enckevort & Thijssen 2001; in press). After the reforms were completed, legio X was withdrawn from Nijmegen and sent to Pannonia. Simultaneously large parts of the canabae were deserted. To prevent econonomic disasters for the Batavian civitas, he gave the almost 30 year old Roman town in the western part of Nijmegen market rights, municipal rights (Haalebos 2000: 35-39), and a new name, Ulpia Noviomagus Batavorum, the Ulpian Newmarket in the Land of the Batavians. Notable in this name is “Newmarket.” That implies there was an old market in Nijmegen or in the civitas Batavorum. There are no direct indications for it from the Kops Plateau, but it seems obvious to us that the forum on the Kops Plateau is the “Oldmarket!”

Each stick on the photo indicates the location of a posthole.

389

Limes XVIII

legerplaats in Nijmegen. (Amersfoort). Haalebos J.K. 1995 Castra und Canabae. Ausgrabungen auf dem Hunerberg in Nijmegen 19871994. Libelli Noviomagenses 3. (Nijmegen). Haalebos J.K. 1998a Die Canabae der Legio X Gemina in Nijmegen Jahresbericht Gesellschaft Pro Vindonissa Jahresbericht 1997: 33-39. Haalebos J.K. 1998b Centuriae onder Centuriae Hof. Opgravingen achter het hoofdgebouw van het voormalige Canisiuscollege te Nijmegen 1995-1997. Libelli Noviomagenses 5. (Nijmegen). Haalebos J.K. 2000 Mosterd na de maaltijd. Een vergeten jubileum: Traianus en het jaar 98 na Chr. in Nijmegen. Numaga Jaarboek XLVII: 5-41. Hartmann M. 1986 Vindonissa. Oppidum Legionslager - Castrum. (Gemeinde Windisch, Windisch). Holwerda J.H. 1943 Een Bataafs dorpje bij het Oppidum Batavorum uit de jaren voor Chr. geb. Oudheidkundige Mededelingen uit het Rijksmuseum van Oudheden te Leiden 24: 35-58. Isings C. with a contribution by J.H.F. Bloemers 1980 Glass from the canabae legionis at Nijmegen. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 30: 283-346. Kandler M. & Vetters H. 1986 Der Römische Limes in Östereich. Ein Führer. (Wien). Willems W.J.H. 1987-1988 Romeinse militaire en civiele bewoning te Nijmegen. In Jaarverslag Rijksdienst voor het Oudheidkundig Bodemonderzoek 1986 (Rijswijk): 42-49; 1988 (1987): 36-39; 1989 (1988): 34-35. Willems W.J.H. 1990a-1992 Romeinse militaire en civiele bewoning te Nijmegen. In Jaarverslag Rijksdienst voor het Oudheidkundig Bodemonderzoek 1989 (Amersfoort): 43-47; 1991 (1990): 42-44; 1992 (1991): 33-36. Willems W.J.H. 1990b Romeins Nijmegen. Vier eeuwen stad en centrum aan de Waal. Historische Reeks Nijmegen 2. (Utrecht). Zee K. 1998 Romeinse militaire en civiele bewoning te Nijmegen. In Jaarverslag Rijksdienst voor het Oudheidkundig Bodemonderzoek 1995/1996. (Amersfoort): 187-189.

Bibliography Bloemers J.H.F. 1972-1985 Nijmegen. Bulletin Koninklijke Nederlandse Oudheidkundige Bond 71: 130-131; 1974 73: 180-181; 1975a 74: 161-163; 1980a 79: 32-36; 1985 84: 30-37. Bloemers J.H.F. 1973-1981 Nijmegen. In Jaarverslag Rijksdienst voor het Oudheidkundig Bodemonderzoek 1972 (Ministerie van Cultuur, Recreatie en Maatschappelijk Werk, Rijswijk): 17-19; 1975b (1973): 16-18; 1976 (1974): 23-27; 1977a (1975): 18-20; 1977b (1976): 17-18; 1978 (1977): 32-34; 1980b (1979): 4350; 1983 (1981): 30-33. Bloemers J.H.F. 1975c Bodemonderzoek Kopse Hof en oostzijde Trajanusplein. Numaga XXII: 21-26. Bloemers J.H.F. 1977c Die augusteischen Lager und die flavisch-trajanischen Canabae Legionis in Nijmegen. In D. Haupt & H.G. Horn (edd.) Studien zu den Militärgrenzen Roms, II, Vorträge des 10. Internationalen Limeskongresses in der Germania Inferior, Xanten-Nijmegen 1974. Bonner Jahrbücher, Beiheft 38. (Bonn): 87-91. Bloemers J.H.F. 1979 Het gebied rondom de legerplaatsen op de Hunerberg. In J.H.F. Bloemers, J.E. Bogaers, J.K. Haalebos & S. Wynia Noviomagus. Op het spoor van de Romeinen (Rijksmuseum G.M. Kam, Nijmegen): 51-56. Bloemers J.H.F. 1980c Nijmegen: ROB excavations 19741979 in the Roman forts, cemeteries, and canabae legionis. In W.S. Hanson & L.J.F. Keppie (edd.) Roman frontier studies 1979. Papers represented to the 12th international congress of Roman frontier studies (British Archaeological Reports Int. Ser. 71. (Oxford): 471-474. Bloemers J.H.F., Louwe Kooijmans L.P. & Sarfatij H. 1981 Verleden Land. Archeologische opgravingen in Nederland. (Meulenhof Informatief, Amsterdam). Enckevort van H. 1993-1995 Romeinse militaire en civiele bewoning te Nijmegen. In Jaarverslag Rijksdienst voor het Oudheidkundig Bodemonderzoek 1992 (Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort): 32-35; 1994 (1993): 3539; 1995 (1994): 151-153. Enckevort van H. 1997 Die Belegung des frührömischen Lagers auf dem Kops Plateau. Römer, Gallier, Bataven und Keltiberiker in Nijmegen. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman frontier studies 1995. Proceedings of the XVIth International Congress of Roman frontier studies (Oxford): 555-564. Enckevort van H., Haalebos J.K. & Thijssen J. 2000 Nijmegen. Legerplaats en stad in het achterland van de Romeinse limes. Archeologische Berichten Nijmegen 3. (Nijmegen). Enckevort van H. & Thijssen J. 2001 Der Hauptort der Bataver in Nijmegen im 1. Jahrhundert n.Chr. Von Batavodurum und Oppidum Batavorum nach Ulpia Noviomagus. In G. Precht (ed.) Genese, Struktur und Entwicklung römischer Städte im 1. Jahrhundert n. Chr. in Nieder- und Obergermanien. Xantener Berichte 9: (Mainz): 87-110. Enckevort van H. & Thijssen J. in press Nijmegen - A Roman town in the frontier zone of Germania Inferior. In P. Wilson (ed.) The archaeology of Roman towns. Enckevort van H. & Zee K. 1996 Het Kops Plateau. Prehistorische grafheuvels en een Romeinse

390

Harry van Enckevort: The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL)

Fig. 1. Overview of Nijmegen in the middle Roman period (AD70-260/270). Ditches and roads are reproduced in black (dotted = reconstructed) lines. The grey lines (roads) represent parts of the modern topography. 1 – Fortress of Legio X; 2 – Canabae legionis; 3 – Settlement on the banks of the Waal; 4 – Ulpia Noviomagus Batavorum; 5-7 – Cemetery: A – forum B – principia; C – amphitheatre; D – mansio; E and G – temple; F – baths; H – bridge. (Drawing Rob Mols).

Fig. 2. The excavation trenches on the Kops Plateau with the main Flavian structures. 1 – Ditch of the castra of Legio X; 2 – Building; 3 – Graves; 4 – Road; A – forum; B – amphitheatre; C – oven; D and E – horrea; F – road-station. (Drawing Rob Mols).

391

Limes XVIII

Fig. 3. Distribution of Flavian coins on the Kops Plateau. Inside the two outlined areas, the ground was only extensively searched for metal. Outside the metal detector was intensively used. (Drawing Rob Mols).

Fig. 4. The eastern wing of the forum (S-N view). In dark the robber trenches of the foundations. Between the two trenches at the right side the rows with postholes. Tilmann Bechert took this photograph as a participant on the Xth Roman Frontier Studies Congress in Xanten, September 1974 .

392

Harry van Enckevort: The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL)

Fig. 5. Ground plan of the forum. In white the excavated area. In the dark screen the disturbances of the last 150 years. (Drawing Rob Mols).

Fig. 6. Reconstruction of the ground plan of the forum. The distances are in Roman feet (pes drusianus). The screened area was filled with over 50,000 small potsholes. (Drawing Rob Mols).

393

Limes XVIII

394

Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes: Topographie, Siedlungsstruktur, Nutzungszonen, Grundrisstypen sowie belegbare Nutzungen. M.N. Filgis Lage und Topographie

Siedlungsstruktur (Grabung 1983-87)

Zur Sicherung der römischen Reichsgrenze in Obergermanien wurden um 100 n. Chr. entlang des linken Neckarufers (Abb. 1) die Kastelle Stuttgart-Bad Cannsatt, Benningen, Walheim, Heilbronn-Böckingen und Wimpfen im Tal als Teil des Neckar-Odenwald-Limes errichtet und durch Staßen miteinander verbunden. Der Neckar diente als natürliche Grenze zum freien Germanien und als Transportweg für schwere Güter. Sein Hochwasser könnte in römischer Zeit auch zur Bedrohung von Kastell und Vicus in Wimpfen (Abb. 2) geführt haben. Nach 150 n. Chr. wurde die römische Reichsgrenze um etwa 30kms nach Osten vorgeschoben, der obergermanische Limes errichtet und das Militär von Wimpfen nach Jagsthausen verlegt.

kastellzeitlicher

Holzbauten

Im folgenden wird die kastellzeitliche Holzbebauung vorgestellt, die nur in Restflächen zwischen der Steinbebauung untersucht werden konnte, weil die Steinbauten für eine spätere Präsentation erhalten werden sollten. Für die Interpretation einzelner Gebäudegrundrisse und möglicher Nutzungen bedurfte es daher erheblicher hypothetischer Ergänzungen (Abb. 5). Das Grabungsgebiet (L. c.150m, B c.65m) erstreckte sich zwischen der Heilbronner Straße, unter der die Römerstraße verläuft, und einem nordöstlich davon durch Bohrungen nachgewiesenen Altneckarlauf. Die mehrfach angeschnittene Römerstraße wurde von ihrem frühesten bis zu ihrem spätesten Nutzungsniveau durch wiederholte Bekiesung um c.1m aufgehöht. Im Bereich der Gebäude ist eine ähnlich mächtige Niveauanhebung durch Bauschutt und Erde erkennbar. So liegen die Fußböden in den spätesten Steingebäuden teilweise über 1m höher als die in den frühen kastellzeitlichen Holzbauten, deren Laufhorizonte bei c.146m ü. NN angetroffen wurden.

Das Wimpfener Römerkastell, im Bereich des heutigen Teilortes Wimpfen im Tal angelegt, ist bisher nicht näher untersucht (Schumacher 1900: 2f, Taf 11, 2, 5.1), seine Lage, Größe und Orientierung sind daher nur näherungsweise bestimmbar. Der dendrochronologische Nachweis einer römischen Brücke (Frey 1991: 112, mit Anm. 23, 24) über den Neckar, in nördlicher Fortsetzung der heutigen Fischergasse, läßt die Lage des Kastells wohl zwischen Stifts- und Brunnengasse (Abb. 3) vermuten. Die Größe des in Abb. 3 hypothetisch eingezeichneten Kohortenkastells entspricht mit c.2.1ha der Größe des etwa 25kms südlich von Wimpfen angelegten Kastells in Walheim, wobei unklar ist, ob die für die Wimpfen durch Ziegelstempel (Regia Wimpina 1988: 65f., Abb. 57, 1-3) belegte cohors II Hispanorum eine equitata war und daher eventuell ein größeres Kastellareal anzunehmen wäre. Für die Orientierung des Kastells wurde die porta Praetoria – wie an den übrigen Neckarkastellen – zu Fluß und Brücke hin angelegt.

Die zur Straße giebelständigen Holzbauten nehmen meist die gesamte Parzellenbreite ein, die zwischen 6–11.5m liegt. Ein Gebäude mit etwa 14m Breite könnte auf eine Zusammenlegung von zwei Gebäuden von c.6m und 8m Breite hindeuten. Häufig bestand zwischen zwei benachbarten Gebäuden nur eine tragende Wand, eigenständige Gebäudelängswände mit dazwischen liegender Traufgasse sind nur vereinzelt zu belegen, einmal fand sich ein bekiester Weg von c.2m Breite, der sich am Ende der beiden Gebäude auf c.1m Breite verengt und wohl weiter bis zum Altneckarufer führte. Nutzungszonen der Parzellen

Eine römische Straße, die durch die Grabungen von 196971 erfaßt wurde (Filtzinger, Planck & Cämmerer 1986: 220, Abb.65), führte wohl südlich am Kastell vorbei (Abb. 3), eine weitere, während der Grabungen 1983-87 angeschnitten straβe, führte wohl zum Osttor des Kastells. Entlang dieser Straße konnte 1983-87 im östlichen Teil des Kastellvicus eine mehrperiodige Straßenrandbebauung aufgedeckt werden, die interesssante Einblicke in die Bauund Siedlungsstruktur sowohl der kastellzeitlichen Holzbebauung als auch der Steinbebauung der später umwehrten Zivilsiedlung gibt (Abb. 4).

Das Siedlungsgebiet zwischen Straße und Altneckarlauf läßt deutlich mehrere Nutzungszonen erkennen. Parallel zur Straße verlief wohl eine Baulinie, auf der die Gebäudefronten zu errichten waren. Zwischen diesen und dem Straßenrand verblieb ein c.5m breiter Streifen, eine von Portiken überdeckte Zone wohl überwiegend gewerblicher Nutzung. Unter den Portiken lagen vereinzelt Holz- oder Steinkeller, die vom jeweiligen Gebäude aus zugänglich waren. Die zweite Zone ist von Gebäuden überbaut, die etwa 20-25m der Parzellentiefe einnehmen, dahinter folgen in der dritten Zone die Höfe und Gärten, in denen sich neben zahlreichen Gruben (in Abb. 5 nicht dargestellt) Überreste handwerklicher Nutzung (Töpferöfen, Kalbrennerei) sowie zahlreich Brunnen fanden. Als vierte Zone kann der Uferstreifen zum

395

Limes XVIII

Altneckarlauf bezeichnet werden, der zur Anlandung von Booten und zum Entladen schwerer Güter – wie beispielsweise der Kalbruchsteine zum Brennen im Kalkofen – gedient haben könnte. Reste einer Abgrenzung der Grundstücksparzellen zum Neckar hin konnten nur in den beiden am östlichen Rand der Grabung freigelegten Parzellen festgestellt werden, sodass die übrigen Grundstücke zum Neckar hin wahrscheinlich gar keine Abzäunung besaßen.

eine neue Parzellenabgrenzung (Zaungräbchen) sichtbar wurde. Als Typus 2, hier nur zweimal belegt, ist ein zur Straße gelegener, die gesamte Parzellenbreite einnehmender großer Raum zu bezeichnen, dem im Nordosten drei etwa gleich große Räume oder zwei Räume und ein Mittelflur anfgefügt sind. Ausgehend von der nachgewiesenen dichten ‘Streifenhaus’-Bebauung stellt sich die Frage nach der optimalen Nutzung der Grundstücksparzellen und einer möglichen Zweigeschossigkeit der Holzfachwerkbauten. Zweigeschossige kastellzeitliche Holzfachwerkbauten ließen sich in Wimpfen bisher nicht nachweisen. Es fehlen eindeutige Baubefunde wie Treppenschwellen oder Überreste von anderen Treppenansätzen, die auf eine Erschließung des Obergeschosses hindeuten. Leitern für den Einstieg in den Dachraum wird es immer gegeben haben.

Nutzungszonen innerhalb der Gebäude Ähnlich wie auf den Parzellen zeichnen sich auch innerhalb der Gebäude gewisse nutzungsorientierte Zonen ab. Viele der im ersten Drittel der Gebäude liegenden und zur Straße geöffneten Räume werden dem Handel, dem Verkauf selbst produzierter Produkte oder auch der Gastronomie gedient haben, wobei die Portiken eine wichtige Vorzone bildeten. Im mittleren Drittel der Gebäude heben sich besonders die zahlreichen Ofenresete und mehrere Grillkanäle (Abb. 5) hervor, die zum einem der Metallverarbeitung, zum anderen gewerblichen Küchen zugerechnet werden. Im letzten Drittel fanden sich meist im Anschluß an die Räume die ältesten Holz- und Steinkeller. Die Erlaubnis, Keller unter der Portikus anzulegen, scheint hier jüngeren Datums zu sein.

Zu Nutzung der ‘Streifenhäuser’ Das giebelständige ‘Streifenhaus’ zählt in Straßenvici zum gebräuchlichsten Gebäudetypus, der mit seinem relativ schmalen Giebel (etwas 1/2 bis 1/3 der Gebäudelänge) bei dichter Reihung einer Vielzahl von Parzellenbesitzern Anteil am öffentlichen Straßen- und Handelsraum in nicht zu großer Entfernung vom Kastell gewährte – eine Grundvoraussetzung für rasche Entwicklung von Handel und Gewerbe wie für gemeinschaftliches Leben. Die Anbindung an den öffentlichen Raum ermöglichte innerhalb der langrechteckigen ‘Streifenhäuser’ sehr unterschiedliche Nutzungen, die sich jedoch nur selten anhand des während der Ausgrabungen geborgenen Fundmaterials zweifelsfrei nachweisen lassen. Bessere Anhaltpunkte zur Gebäudenutzung bieten bauliche Überreste von gewerblichen Einrichtungen wie beispielsweise verschiedene Arten von Feuerstellen, Herden und Öfen.

Erschließungstypen der ‘Streifenhäuser’ Die entlang der Straße dicht nebeneinander errichteten giebelständigen ‘Streifenhäuser’ besaßen ihren Hauseingang zwangsweise an der Giebelseite, dahinter folgte ein Mittelflur (Abb. 6, Typus 1), der die zu beiden Seiten des Flures liegenden Räume erschloß und einen Ausgang zu Hof und Garten besaß. In den Abwandlungen dieses Grundtypus besteht der Mittelflur ansatzweise weiter, indem auf der einen Seite des Flures die Raumflucht beibehalten, auf der anderen, sei es beim Eingang an der Straße (Abb. 6, Typus 1a) oder beim Ausgang zum Hof (Abb. 6, Typus 1b) die Raumflucht auf einen abgegrenzten Raum reduziert wird und des Rest des Erschließungsflures in einen großen, vielfach nutzbaren Raum übergeht.

Töpferei Am eindeutigsten belegen Überreste von Keramikbrennöfen (Abb. 7) im Hofbereich eines c.10.5 x 28m großen Holzfachwerkbaus (einschließlich Portikus) die Existenz einer Töpferei1, obwohl sich innerhalb des Gebäudes von der Werkstatt des Töpfers keine Spuren mehr erhalten haben (Abb. 5 und 6.1). Vier der elf Räume waren mit kleinen Feuerstellen beheizbar.

Die Gebäudeerschließung mit durchgehendem Flur und nur auf einer Flurseite gelegenen Räumen (Abb. 6, Typus 1c) scheint aus der Aufteilung einer Normalparzelle von etwa 8.5m Breite und einem ‘Streifenhaus’ Typus I enstanden zu sein, indem der Nachbar einen Parzellenstreifen von knapp 3.5m hinzu erwarb und damit seine Parzelle auf c.14m verbreiterte, während der Flur mit einseitiger Raumflucht als funktionsfähiges, jedoch nur etwa 5.5m breites Gebäude erhalten blieb. Unklar ist, ob die Dachkonstruktion der beiden ‘Streifenhäuser’ den Grundrißveränderungen angeglichen wurde – dann hätten zwei sehr ungleiche Gebäude nebeneinander gestanden – oder ob die Raumumnutzung unter Beibehaltung bestehender Dächer und Fassaden erfolgte und lediglich im Hof- und Gartenbereich die Nutzungsverlagerung durch

Die Fußböden des Gebäudes bestanden allgemein aus Stampflehm, nur im nordöstlichsten Raum fanden sich Reste eines sorgfältig geglätteten Kalkestrichs mit Ziegelsplittbeimengung, der nach Zerstörung dieses Holzfachwerkbaus beim Bau eines jüngeren Töpferofens 1

Die am Hofausgang des Fachwerkbaus beginnende Kiesfläche überlagerte den Fuchs des kleineren der beiden an einer Schürgrube liegenden Töpferöfen.

396

M.N. Filgis: Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes (vg. Abb. 4) durchschlagen wurde. Möglicherweise befand sich in diesem besonders ausgestatteten und wohl auch ausreichend belichteten Raum die Werkstatt des Töpfers. In den beiden zur Straße gelegenen Räumen sowie unter der Portikus könnte die selbst produzierte Töpferware verkauft worden sein, wenn nicht ein Großteil der Keramikproduktion von Händlern direkt übernommen und weiter verhandelt wurde. Ob dieser zu den großen Gebäuden zählende Holzfachwerkbau ein Obergeschoß besaß, ist ungewiß, denkbar wäre jedoch im dritten, nur c.2m breiten Raum an der linken Seite des Mittelflures eine zweiläufige Treppe, die wiederum zu einem Mittelflur im Obergeschoß führte.

zum Warmhalten von Speisen schließen. M. Kokabi und S. Frey stellten an Tierknochenfunden der Grabung Wimpfen Brandspuren fest, die beim Grillen von nicht entbeintem Fleisch entstanden sein dürften (Kokabi & Frey 1988: 4344). Vermutlich siedelten in diesen drei ‘Streifenhäusern’ Gastwirte, die in ihren tabernae auf offenen Herdstellen oder solchen Grills (Abb. 10) Essen für ihre Gäste zubereiteten, das wohl unter der Portikus oder in dem anschließenden, etwas größeren Raum eingenommnen wurde. Ein Gastwirt hatte vor seinem Haus zwei kleinere Grills in den Boden unter der Portikus eingetieft und wird dort Fleischstückchen am Spieß oder ähnliches gegrillte und Passanten angeboten haben.

Metallverarbeitung

Bäckerei

Im westlich anschließenden c.11m x 23m großen Holzfachwerkbau (einschließlich Portikus) (Abb. 5 u. 6,2) fand sich ein 1.1m x 1.1m großer Ofen aus in Lehm verlegten Kalkbruchsteinen mit kreisförmiger Feuerplatte von etwa 0.5m Durchmesser und 0.2m breiten Schüröffnung, der an derselben Stelle mehrfach grundlegend erneuert wurde. Die an den Ofen angrenzenden Fußbodenaufhöhungen aus Stampflehm enthielten stark korrodierte Blechreste, anhand derer U. Zwicker dennoch nachweisen konnte, dass hier Buntmetall geschmolzen und verarbeitet wurde.2 Vermutlich war dieser hallenartige, c.10.5m x 14m große Raum mit weiterer Ausstattung die Werkstatt eines Metallhandwerkers, der wohl nicht nur Buntmetall verarbeitete. Öfen vergleichbarer Konstruktion, gelegentlich auch als Doppelöfen ausgeführt, sind im Ausgrabungsgebiet mehrfach belegt. Die ungewöhnlich starke Verziegelung des unter den Öfen liegenden Lehmbodens läßt auf höhe Temperaturen (wohl um 10000 C) über längere Zeit schließen. Diese Öfen finden sich auch in kleineren Räumen, meist jedoch im mittleren Drittel, also im weniger gut belichteten Teil der ‘Streifenhäuser’, vielleicht um beim Glühen oder Schmelzen von Metall anhand der sich verändernden Rotfärbung die Temperatur des zu verarbeitenden Metalls besser einschätzen zu können. Im freigelegten Siedlungsareal dürfen also in weiteren ‘Streifenhäusern’ kleinere Werkplätze von Metallhandwerkern angenommen werden.

Ein innerhalb des zweiten Gebäudes östlich der Töpferei nahe der Portikus gelegener Backofen (Abb. 5) mit c.1.1m x 1.3 x (1.43m2) große Backfläche darf wegen seiner Lage und Größe als gewerblicher Ofen – vermutlich einer Bäckerei – gedeutet werden. Im rückwärtigen Teil dieses Gebäudes muß sich in dem mit Holzboden ausgestatteten Raum vor dem Keller ein Getreidelager befunden haben, da sich an dieser Stelle im Brandschutt des Hauses eine große Zahl verbrannter Getreidekörner, vor allem von Spelzgerste und Dinkel, fand. Spätestens nach einer Generation dürfte sich eine Entwicklung angebahnt haben, die eine Veränderung der ursprünglich angelegten Parzelleneinteilung und möglicherweise auch der Gebäudenenutzung einleitete: Einzelne Grundstücke und Gebäude werden frei geworden, aufgeteilt und die Gebäudestruktur neuen Bedürfnissen angepasst worden sein. Zusammenfassung Hochwasser der Jagst und des Neckars bedeuteten für Kastell und Vicus eine latente Gefahr. Die um 200 n. Chr. errichtete Vicusmauer gewährte einen gewissen Schutz. Hochwasserfreie römische Siedlungsgebiete befanden sich westlich und östlich der Jagst. Das ‘Streifenhaus’ war ein vielfach variierbarer Gebäudetypus, der im Straßenvicus durch Giebelständigkeit einer möglichst großen Zahl von Gebäuden Anteil am Straßen- und Handelsraum gewährte. Anhand des Fundmaterials läßt sich die Nutzung eines Streifenhauses kaum bestimmen, Überreste gewerblicher Einrichtungen geben dafür deutlichere Hinweise. Die ursprüngliche Parzelleneinteilung für die Holzbauten wurde wohl bereits im mittleren Drittel des 2. Jhs. n. Chr. durch Teilverkauf verändert, in der Steinbebauung konnten dann wohl mehrere Parzellen erworben und größere zusammenhängende Bauten errichtet werden.

Gastronomie Im westlichen Teil Grabungsgebietes kamen in mehreren ‘Streifenhäusern’ etwa 0.3-0.4m breite, 1.5-3.5m lange und bis zu 0.4m tief in das anstehende Erdreich eingelassene Feuergruben zum Vorschein (Abb. 5 u. 9). Ihre Sohle war mit Asche oder Holzkohleresten bedeckt, Boden und Wände der Gruben wiesen nur geringe Verziegelung auf. Ihre große Zahl, ihre Abmessungen und ihre Konzentration auf einen begrenzten Bereich lassen auf gewerbliche Einrichtungen zum Grillen von Fleisch oder Fisch oder 2

U. Zwicker, Universität Erlangen-Nürnberg, Untersuchungsbericht 399/85 vom 4.3.1985 und 414/85 vom 13.8.1985.

397

Limes XVIII

Literatur Filtzinger

P., Planck D. & Cämmerer B. (Hrsg.) 1986 Die Römer in Baden-Württemberg. (Stuttgart). Kokabi M & Frey S. 1988 Neue Funde…, Bad Wimpfen, Kreis Heilbronn. Archäologie im Deutschland 3: 43-44. Frey S. 1991 Bad Wimpfen I. Osteologische Untersuchungen an Schlacht - und Siedlungsabfällen aus dem römischen Vicus von Bad Wimpfen. Forsch. u. Ber. zur Vor- u. Frügeschichtliche in Baden-Württemberg 39. (Stuttgart). Schumacher K. 1900 Kastell und Vicus bei Wimpfen. Der Obergermanisch-räetischer Limes des Römerreiches B. Nr. 54 und 55. (Heidelberg). Regia Wimpina 1988 Regia Wimpina. Beiträge zur Wimpfener Geschichte 5, Das römische Wimpfen. (Bad Wimpfen).

398

M.N. Filgis: Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes Abb.1. Das Limesgebiet zwischen Rhein, Main und Schwäbischer Alb mit der Lage von Wimpfen am Neckar-Odenwald-Limes. Nach H. Schönberger Das Römerkastell Öhringen-West (Bürgkastell). 53 Ber. RGK (1972): 233ff. abb.1, teilweise uberarbeitet. Abb. 2. Bad Wimpfen in Tal. Überschwemmung des antiken Siedlungsareals durch Hochwasser der Jagst (im (Hintergrund) und des Neckars am 4.3 (1956). (Aufn,. R. Helfrich).

Abb. 3. Bad Wimpfen in Tal. Übersichtsplan über das römische Wimpfen mit Eintragung der bisher archäologisch untersuchten Flächen und der hypothetischen Lage des Kastells.

399

Limes XVIII

Abb. 4. Bad Wimpfen in Tal. Grabung 1983-1987. Vorläufiger Übersichtsplan der Holz- und Steinbebauung.

400

Abb. 5. Bad Wimpfen in Tal. Grabung 1983-1987. Vorläufiger ergänzter Übersichtsplan der kastellzeitlichen Holzbauung.

M.N. Filgis: Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes

401

Limes XVIII

Abb. 6. Bad Wimpfen kastellzeitlicher Holzbauten.

in

Tal.

Grundrisstypologie

Abb. 7. Bad Wimpfen in Tal. Zwei an gemeinsamer Bedienungsgrube gelegene Töpferöfen.

Abb. 9. Bad Wimpfen in Tal. Zwei sich überlagernde, in die Erde eingetiefte Grillkanäle mit Holzkohleresten.

Abb. 8. Bad Wimpfen in Tal. Mehrphasiger Metallschmelzofen aus in Lehm verlegten Kalkbruchsteinen. Außenabmessungen 1.1 x 1.1m, Innendurchmesser 0.5m.

Abb. 10. Bad Wimpfen in Tal. Küche einer taberna mit Herd und Grillkanal. Rekonstruktion J. Haas.

402

Die Früheste Belegung des Hunerberges in Nijmegen J.K. Haalebos† verschiedenen Schnitten zwischen 40 und 58 Grad. Die Berme kann um 1.8m (6 Fuß) breit gewesen sein. Die Schichten in der Verfüllung waren gesenkt. Die untere Spitze scheint, nach der veröffentlichten Foto, mit relativ hellem Sand zugeschwemmt gewesen zu sein. Der äußere Graben war kaum weniger tief. Die Böschung war einmal etwas flacher. Der Graben war offenbar auf gleiche Weise zugeschüttet worden.

Einleitung Südlich des Waals befinden sich auf einer für die Niederlande außerordentlich hohen Moräne (bis um 60m ü. NN) über eine Länge von fast 5kms Spuren aus der Römerzeit. Diese Höhe bildet vom Süden aus die letzte hohe und trockene Stelle am linken Rheinufer. Die strategische Bedeutung wird noch vergrößert, da hier verschiedene römische Fernstraßen entlang des Rheins und der Maas zusammentrafen. Dank diesen Gegebenheiten wurde hier im Rahmen der augusteischen Eroberung des Rheinlandes ein vieleckiges Lager errichtet, das sich in die Reihe der frühen groβen Lager wie Oberaden und Neuss einfügt und als erster Beleg für römische Militärpräsenz in den Niederlanden gelten darf. Zwischen 1987 und 1997 wurden in der westlichen Hälfte dieses Lagers gröβere Ausgrabungen durchgeführt, die die Kenntnisse wesentlich förderten. Die Spuren wurden in der Nordost-Ecke des Lagers von der flavischen Festung der legio X Gemina überdeckt (Abb. 1).

Die westlichen Wehrgräben des Lagers wurden in den Jahren 1992-1996 ausgegraben. Sie scheinen in römischer Zeit lange offengestanden und als Abfluß für das Regenwasser gedient zu haben. Vom äußeren Graben wurden um die 100m ausgegraben, der innere konnte über eine Länge von mehr als das doppelte untersucht werden. Wahrscheinlich waren beide Gräben vor dem Westtor durch eine fast 14m breite Erdbrücke für die aus dem Lager herausführende Hauptstraße, die Via principalis, unterbrochen. Auf der Innenseite des Innengrabens befand sich ein schmaler Graben, der den Durchgang abschloß und an die Gräbchen zwischen den Grabenenden vor dem Ost- und Nordtor in Haltern erinnert. Die beiden Grabenende wurden geradlinig abgeschlossen.

Die Form und Lage des Lagers Bei der Errichtung des 42ha großen Lagers hat man, soweit möglich, den polygonalen Plan dem Gelände angepaßt. Die Nordseite bestimmte der Steilhang; die beiden nördlichen Festungsecken wurden von zwei kleinen Tälern - dem Beekmandalseweg im Osten und dem Vrouwendal im Westen - begrenzt. Die südlichen Wehrgräben markieren den Rand der sanft nach Süden abfallenden Lagerfläche bzw. den Übergang zu einer steilen Böschung, die trotz neuzeitlicher Bebauung noch heute im Stadtbild sichtbar ist. Auch die Stelle der westlichen Wehranlagen wird von einer leichten Geländesenke bestimmt.

Das Gebiet der Gräben wurde in flavischer Zeit offenbar zur Müllablagerung verwendet. Zahlreiche Abfallgruben und von Überschwemmungen verursachte Erosionslöcher waren zu beobachten; die Grabenböschungen konnten daher nur in grösserer Tiefe erkannt werden. Der äußere Graben war südlich des Westtores teilweise mit sehr schwerem Lehm verfüllt. Die höheren Schichten bestanden aus gelbem und graubraunem Sand und reichten bis an den Rand des inneren Grabens. Das ganze erweckt den Eindruck, als ob hier oft eine Schlammpfütze gestanden hätte. Für den inneren Graben konnten eine Maximalbreite von 6m und eine Tiefe von 2.3m nachgewiesen werden. Die Böschungen dieses Spitzgrabens zeigten einen Neigungswinkel von weniger als 40 Grad. Bis in große Tiefe wurden flavische Scherben und Ziegelfragmente gefunden, die darauf hinweisen, daß der Graben hier erst in flavischer Zeit einplaniert wurde. Nur in der Spitze wurde verhältnismäßig reiner Sand, der während der Benutzungszeit des großen Lagers auf natürliche Weise in den Graben geraten sein kann, vorgefunden. Auf der lagerseitigen Grabenböschung befand sich eine humose Schicht, die eine Grube abdeckte; sie konnte eigentlich nur in nachaugusteischer Zeit, als das Lager verlassen war, ausgehoben worden sein. Die Grabenverfüllungen wurden von Abfallschichten und im nördlichen Teil von einer Brandschicht aus der Zeit der Canabae der 10. Legion überdeckt.

Die Verteidigungsanlagen Die Wehrgräben. Die beiden Wehrgräben des ältesten Lagers wurden von Brunsting durch Schnitte vor dem Osttor freigelegt. Ihre Datierung in augusteische Zeit ergab sich aus den in der Verfüllung gefundenen Scherben und zwei Gräbern, die um die Mitte des 1. Jahrhunderts in dem zugeschütteten inneren Graben angelegt wurden. Der äußere Graben war vor dem Tor unterbrochen und zeigte einen 12.5m breiten Durchgang. Beim inneren wurden keine Hinweise auf eine Erdbrücke gefunden, doch kann diese beim Ausheben des flavischen Wehrgrabens der Periode 4, offenbar beim Osttor nicht richtig beobachtet, abgetragen worden sein. Die Breite des inneren Grabens hat maximal 7.5m betragen, doch scheint er im Durchschnitt nicht größer als 5m gewesen zu sein. In der Tiefe erreichte er meist kaum 2m. Nur an einer Stelle, wo die Innenseite des Grabens bedeutend höher als die äußere lag, ist mit einer Tiefe bis zu 3m zu rechnen. Die Neigungswinkel der Böschungen variieren in den

Die Porta principalis dextra. Das Osttor (Abb. 2, 19) wurde von H. Brunsting 1960 ausgegraben. Es ist zangenförmig angelegt und besteht aus zwei rechteckigen Türmen von je

403

Limes XVIII

acht Pfosten, die durch eine von vier Pfosten getragene Brücke verbunden waren. Auf der Frontseite wurden die Türme außenseitig durch Pfostenpaare flankiert, welche die Verbindung mit dem Wall herstellten. Aus einem der Pfostenlöcher wurde eine arretinische Scherbe des Services II zu Tage gefördert und aus der harten Abdeckschicht einer der anderen Gruben kamen Scherben, die nach Brunsting unverkennbar aus den Jahren um den Beginn unserer Zeitrechnung, der Regierung des Augustus stammten.

augusteischen Zeit stammen müssen. Es liessen sich bisher folgende Holzbauten erkennen (Abb. 2): 1. Die Principia (Abb. 2, 1). Für eine gesicherte Interpretation ist zu wenig von diesem Bau ausgegraben worden. Doch lassen die gefundenen Spuren - ein Fundamentgraben und eine dazugehörige Reihe von Pfostengraben - vermuten, daß der Peristylhof des Stabsgebäudes angeschnitten worden ist. Der Bau wäre dann etwas östlich aus der Mitte des Lagers verschoben. Das Fehlen von Nebenräumen an den Seiten des Hofes scheint ein Charakteristikum der augusteischen Zentralgebäude zu sein. Die Gröβe läβt sich nicht genau bestimmen, kann aber ohne Schwierigkeit um 75m x 75m betragen haben. Der Bau wurde von Baracken des flavischen Lagers von Periode 4 und 5 überlagert. Es gibt keine Funde, die eine engere Datierung erlaubten.

Die Porta principalis sinistra. Das Westtor (Abb. 2, 20; Abb. 3) wurde 1918 von J.H. Holwerda angeschnitten und 19921994 vollständig freigelegt. Es hat die gleiche Form und Abmessungen (25.2m x 9m) wie das Osttor. In den beiden seitlichen Durchgängen wurden jeweils Pfostenlöcher gefunden, die als Teil des eigentlichen Türverschlusses interpretiert werden können und die zeigen, daß die Anlage über zwei Durchgänge verfügt hat. Zwischen den mächtigen Pfostengruben der ursprünglichen Konstruktion gab es auch weniger tiefe Gruben, die von einer Reparatur herstammen müssen. Auf der Innenseite der Türme wurden die Pfostengruben von Straßengräbchen angeschnitten. Hieraus ist zu schließen, daß die Via principalis auch nach der Schleifung des Tores weiter verwendet wurde. Aus einer der Pfostengruben oder einem Pfostenloch wurde die einzige Lyoner Altarmünze geborgen, die sich - mit einiger Mühe mit dem Lager der Periode 1 verbinden läßt.

2. Ein Praetorium? (Abb. 2,2) Der größte bisher ausgegrabene Baukörper hat eine Länge von fast 60m oder 200 Fuß und besteht aus einem nahezu quadratischen Haus von 35.5m x 36.5m mit einem im Süden anschließenden Peristylhof (c.24m x 6.5m). Die Wohnräume sind um einen zweiten Hof (c.14.5m x 12m) und einen kleineren Innenhof (c.5.5m x 5.5m) angelegt worden. Der Haupteingang scheint an der Westseite gelegen zu haben. Kennzeichnend für den Grundriß sind nicht nur die beiden Innenhöfe, sondern auch die langen Korridore, welche die verschiedenen Raumgruppen von einander trennen.

Die beiden Tore gehören zum Typus der Zangentore, der in augusteischer Zeit üblich ist; sie besitzen aber nicht die in die Tiefe greifende Form wie in Oberaden, die auf republikanische Vorgänger wie in Renieblas und Pompeji zurückgeht.

In der Größe stimmt der Bau etwa mit den Wohnungen der Legionstribunen überein, die - soweit wir wissen - nie über ein neben dem Haus gelegenes Peristyl verfügt haben. Die Einteilung läßt sich mit jener des kleineren Praetoriums von Oberstimm vergleichen, wo in claudischer Zeit eine Auxiliareinheit lag. Für ein Legionspraetorium scheint der Bau in Nijmegen jedoch zu klein zu sein, wenn man die Lager aus der mittleren Kaiserzeit zum Vergleich heranzieht. Wenn man das Peristyl zum Gebäude hinzurechnet, läßt sich die Größe mit den neuentdeckten Praetoria von Oberaden, Anreppen und Marktbreit vergleichen. Diese Gebäude sind dem Nijmegener Bau in manchen Einzelheiten – zu vergleichen ist etwa der Eingang des Praetoriums von Oberaden - sehr ähnlich.

Vom Wall wurde so gut wie nichts gefunden. Nur nördlich der Porta principalis sinistra wurde ein Fundament aus nebeneinander gelegten Ästen beobachtet. Weiter war der Verlauf des Walles nur an Hand der Türme zu verfolgen. Auf der Westseite der Legionsfestung lagen diese knapp 24m (= 80 Fuß) von einander entfernt. Die ungefähr quadratischen (3 x 3.6m) Türme wurden je von vier Pfosten gebildet. Die Pfostengruben waren annähernd 1m tief und zeigten oft die Verfärbung des Pfostens, der bisweilen mit Feldsteinen gefestigt war. Die Innenbebauung Die Spuren der Innenbebauung werden von vielen späteren Bauten und Gruben der flavischen Lager und Canabae überdeckt. Die Rekonstruktion der Einteilung des augusteischen Lagers ist also recht schwierig. Die hier dargelegte Übersicht über die westliche Hälfte zeigt aber eine so große Kohärenz, daß das Ergebnis einigermassen befriedigt. Für die östliche Hälfte fehlt die Befundauswertung noch weitgehend, doch darf man erwarten, daβ die im Winter 2000 angefangenen Ausgrabungen der Gemeinde Nijmegen auch hier Aufschlüsse für die Lagereinteilung bringen werden. Aus den älteren Grabungen wurden bisher nur zwei kleine Bauten eingetragen, die so nahe an der Festungsmauer des flavischen Steinlagers liegen, daß sie eigentlich aus der

Der Bau wurde von den Wehrgräben der flavischen Legionslager (Periode 4 und 5) sowie von dem älteren Abwasser- oder Wehrgraben X geschnitten. Die Balkengräbchen ergaben nur unbedeutende Funde. In Anbetracht der Größe des Lagers darf man in Nijmegen drei Legionskommandanten und vielleicht auch noch einen Oberbefehlshaber erwarten, die jeder ihre Unterkunft gehabt haben müssen. Damit läßt sich die exzentrische Lage in Bezug auf die Principia erklären. 3-6. Tribunenhäuser. An der Südseite der Via principalis und westlich der Principia wurden vier quadratische oder 404

J.K. Haalebos: Die Früheste Belegung des Hunerberges in Nijmegen rechteckige Häuser gefunden, deren Flächen zwischen 240 und 540m2 variierten. Es scheinen Wohnhäuser gewesen zu sein und man darf dabei wohl an Wohnungen der Tribuni militum denken, obwohl in späterer Zeit ihre Unterkünfte normalerweise an der Praetorialseite der Via principalis liegen. Polybius stellte aber die Zelte der Tribuni vor und neben das Zelt des Feldherrn, was der Nijmegener Situation entspricht.

8-17. Mannschaftsunterkünfte (Abb. 4). Die meisten übrigen Bauten können, mit Ausnahme von einigen spärlichen Resten, die sich einer Interpretation entziehen, wohl als Mannschaftsbaracken betrachtet werden. Am deutlichsten ist die Situation am westlichen Wall. Dort wurden die Kasernen von zwei einander gegenüber liegenden Cohorten gefunden, die mit ihren fast quadratischen Kopfbauten (um 9 x 9m) an eine 18m breite Straße grenzten. Es wurde für jede Cohorte fünf Centurionenhäuser entdeckt, die jeweils auf eine Einzelbaracke und zwei Doppelbaracken hinweisen. Die fehlende Baracke der sechsten Centuria läßt sich leicht im nicht untersuchten Gelände östlich der Ausgrabung ergänzen. Die Cohorten beanspruchten je eine Fläche von mehr als 64 x 60m, d.h. im Prinzip 200 x 200 Fuß.

Der Raum reicht, um die Reihe der hier aufgeführten Häuser mit acht weiteren zu ergänzen und so sechs Tribunen einer Legion hier unterzubringen, doch scheinen die zwischen den Häusern 3 und 4 gefundenen dürftigen Spuren nicht für eine solche Ergänzung zu sprechen, wenn diese wirklich aus augusteischer Zeit stammen. Wenn hier wirklich zehn Häuser gestanden haben, könnten sie vielleicht über die Tribunen (46 und noch nicht gefundene Häuser südlich davon) und andere höhe Offiziere (3 + ?), wie den Praefectus castrorum oder einen Primipilus, verteilt gewesen sein.

Verschiedene Häuser verfügten an der Straßenseite über Latrinengruben. Hinter den Offizierswohnungen schlossen große Räume an, in denen man wohl Höfe sehen darf. Bei mindestens einer Baracke waren Mannschaftsunterkünfte an einem Fundamentgraben für eine Trennungsmauer zwischen den beiden Barackenhälften erkennbar.

3. Ein rechteckiges Gebäude, dessen Länge nicht abgesteckt werden konnte, aber mehr als 18.5m betragen haben muß. Die Breite betrug 20.4m. Die Zimmer scheinen um einen Hof gelegen zu haben. Die Südseite des Gebäudes konnte nicht ausgegraben werden. Das Haus wurde vom Wall und der Wehrmauer der Lager 4 und 5 überdeckt.

Von den übrigen Mannschaftsbaracken wurden meist nur die Spuren der Offizierswohnungen aufgedeckt. Die Länge der Baracke kann man bisweilen nach den sie begleitenden Reihen von Gruben erschliessen.

4. Von diesem fast vollständig ausgegrabenen Haus mit einer Fläche von mehr als 22 x 20m sind nur zwei Außenseiten lokalisiert worden. Die Nordseite fehlt, obwohl diese gesucht wurde. In der Ost-Westausdehnung kann dieses Haus kaum gröβer gewesen sein, da anders zu wenig Raum für das Nachbarhaus frei bleibt. Das Gebäude wird von dem ausgebrochenen Steinkanal der Periode 5, Abwassergräbchen aus spätflavischer Zeit (Periode 4 und 5) und Pfostengruben einer Palisade aus dem 2. Jahrhundert geschnitten. Neben der nordöstlichen Ecke und im hier gelegenen Eckraum lagen Gruben mit augusteischer Keramik. Vielleicht darf man daraus nicht nur auf die Datierung des Gebäudes in augusteische Zeit schließen, sondern mit diesen Befunden auch die Vermutung begründen, dass der Eckraum eine Latrine gewesen sei. Der Plan erinnert an das sogenannte Praetorium im Halterner Hauptlager und weist einen atriumartigen Innenhof auf.

Die Einteilung Obwohl im westlichen Teil des Lagers viele Überreste von Innenbauten bekannt sind, ist es noch immer nicht möglich, die genaue Einteilung zu umreissen. Auffällig ist die Lage der Principia, die aus der Mitte des Lagers nach rechts verschoben wurden. Auch zieht die weit nach Norden verschobene Via principalis die Aufmerksamkeit auf sich. Wenn die durch die Baracken am westlichen Wall ermittelte Einteilung in Blöcke von annähernd 60 x 60m der weiteren Einteilung zugrunde gelegt wird, so können westlich der Principia vier Reihen solcher Blöcke, östlich davon drei rekonstruiert werden. In Nord-Süd-Richtung können sich südlich der Porta principalis sinistra entlang der Wehrmauer fünf Scamna erstreckt haben, die einer halben Legion Unterkunft geboten hätten. Die Praetentura nördlich der Via principalis bot genau Platz für zwei Reihen von Scamna. Diese kleine Übung – unzuverlässig wie sie ist – zeigt klar, daβ drei Legionen ohne Schwierigkeiten im Lager untergebracht werden können und daβ ihre Mannschaftsbaracken alle am Wall gelegen haben können, wie es Hyginus vorschreibt und in Oberaden auch der Fall gewesen zu sein scheint. Andererseits ist es sehr verlockend sich vorzustellen, daβ am Westwall in zwei Reihen von fünf Cohortes eine vollständige Legion gelegen hätte, wie in Abb. 5 eingetragen wurde.

5. Es wurde nur die fast 16m lange westliche Wand dieses Baus mit einigen Querwänden freigelegt. Das Haus kann in der Ost-West Richtung maximal 20m breit gewesen sein. Auf der Nordseite lagen Gruben mit augusteischer Keramik und Metallresten, die auf den Betrieb eines Bronzegieβers hinweisen. 6. Die erhaltenen Reste messen 18 x 30m und ermöglichen die Rekonstruktion eines Hauses von 60 x 100 Fuß. Der westliche Abschluß fehlt aber. Das Haus kann einen Innenhof gehabt haben. An der Nordseite gab es zwei Reihen mit augusteischen Gruben, die bei der Ostwand endeten. Der Bau wurde von dem Steinkanal des spätflavischen Lagers und anderen, noch jüngeren Spuren geschnitten.

Belegung Epigraphische Hinweise fehlen fast ganz. Die Größe des Lagers und die aufgefundenen Baracken weisen auf die Anwesenheit von Legionen hin. Legionssoldaten können 405

Limes XVIII

der schon lange bekannte Helm aus der Nordostecke des Lagers und ein gebogener trapezförmiger Schild (0.61m x 1.06m) mit spindelförmigem Umbo zugeordnet werden, der an Hand der in einer Grube gefundenen Überreste rekonstruiert werden konnte (Abb. 6-7). In dieser frühen Zeit werden Legionarii den aus späterer Zeit bekannte Schienenpanzer, die lorica segmentata, kaum getragen haben, sondern den einfacheren Ringpanzer, die lorica hamata. Von solchen Panzern wurden mehrere Schlieβhaken gefunden. Einer dieser Haken trägt den Namen seines ehemaligen Eigentümers, Q. Antonius, der der centuria des Mar(cellus?) angehörte (Abb. 8).

ungesicherten Umständen - nur eine halbierte Lyoner Altarmünze gefunden. Ob man auf Grund dieser Funde die Errichtung des Lagers schon früh im zweiten vorchristlichen Dezennium ansetzen will oder eher mit dem ersten Auftreten des Drusus am Nieder-Rhein im Jahre 12 vor Chr., hängt von der Beurteilung des Berichtes des Cassius Dio (LIV.32) ab, der den Eindruck erweckt, daβ diese Strecke der Grenze von Drusus zum ersten Mal mit römischen Truppen belegt worden war. Ein Teil der frühen Terra Sigillata-Stempel (Abb. 10 und Tabelle 1) stammt von Töpfern, deren Erzeugnisse auch in Dangstetten und Neuss vorliegen; sie können als Argument für eine Frühdatierung des groβen Lagers gelten. Zusammen mit den vielen Copia- und ViennaMünzen, die besonders in der Zeit vor der Belegung Oberadens zirkuliert haben, zeigen diese Stempel, daβ Nijmegen schon in der Zeit der Clades Lolliana als Basis gedient haben kann. Die Frage, ob der Lagerausbau vor oder nach dieser Niederlage stattfand, bleibt vorläufig dahingestellt.

Pfeilspitzen (Abb. 9) und leider nicht in stratigraphischem Zusammenhang gefundene Schleuderbleie könnten von Auxiliarsoldaten verwendet worden sein, ein Beweis für ihre Präsenz ist dies aber nicht. Die große Fläche (42ha) zeigt, daß das Lager als ein Bereitstellungslager zu betrachten ist, das während eines Offensivkrieges eine Rolle gespielt haben muß und wenigsten 10 000 Mann Unterkunft geboten hat. Die eben erwähnten Reste eines Schildes (Abb. 7) wurden in einer Grube südlich der Via principalis gefunden, die u.a. einen vollständig erhaltenen einheimischen Topf enthielt. Der Schild war zur Hälfte erhalten und lag mit der Vorderseite nach unten gegen den Grubenrand. Der eiserne, spindelförmige Schildknopf wurde zuerst geborgen und seine Position in Bezug zu den Resten des bronzenen Randbeschlages läβt sich infolge dessen nur ungefähr bestimmen. Der bronzene Rand zeichnete sich in seiner ganzen Länge im Boden ab. Eine Ecke konnte noch geborgen werden, die zweite Ecke war völlig zerkrümelt. Zu diesen Bronzeresten gehört ein Ring, dessen Lage nicht gesichert ist und der wohl als Aufhängevorrichtung des Schildes oder zur Befestigung eines Tragriemens diente. Wenn man davon ausgeht, daβ die im Felde aufgenommenen Umrisslinien des Bronzebeschlages nicht allzu stark durch den Druck des Bodens verformt sind, läβt sich ein auf den Schmalseiten abgerundeter Schild rekonstruieren, dessen Längsseiten nicht ganz parallel verlaufen, sondern sich nach unten verjüngen. Die Form des Schildes ist damit den Schilden der Soldaten auf einem republikanischen Relief aus Estepa in Spanien sehr ähnlich. Die rekonstruierten Maβe entsprechen fast genau dem ledernen Schildüberzug aus Oberaden (c.1.10m x 62cm), dessen Seiten aber parallel verlaufen.

Das Ende des groβen Lagers scheint weniger umstritten. Spätestens der Tod des Drusus im Jahre 9 vor Chr. hat die Aufgabe des großen Lagers veranlaßt, wenn die Hauptmasse der Truppen nicht schon während der Germanien-Offensive verlegt worden ist, z.B in das 11 vor Chr. gegründete Lager von Oberaden. Bereits früher war ein Unterschied zwischen der westlichen Hälfte und dem östlichen Teil des Lagers aufgefallen: Die Stempel der Ateius-Gruppe stammen nur aus dem östlichen Teil, wo auch die gröβte Menge der Scherben des Service II gefunden wurde. Als Erklärung für diesen Unterschied wurde auf die mögliche Existenz eines spätaugusteischen kleineren Lagers in der nordöstlichen Ecke des Lagers oder auf die Möglichkeit von Müllablagerung vom Kops Plateau hingewiesen. Die neuesten Grabungen der Gemeinde Nijmegen im Bereich der Via principalis haben eine Grube mit einem Teller des Services II, gestempelt Crestus geliefert, was zeigt, daβ mit einer spätaugusteischen Belegung zu rechnen ist. Die westliche Grenze eines spätaugusteischen Lagers könnte ebenfalls bei den gleichen Grabungen geschnitten worden sein. Dieser unter den Steinfundamenten des flavischen Lagers gefundene Graben ist aber verhältnismäβig schmal und die wenigen daraus geborgenen Funde sind noch nicht gereinigt. Das vermutete kleine Lager scheint die frühtiberische Zeit nicht überdauert zu haben.

Datierung Die Keramik zeigt, daß das große Lager in der Zeit Oberadens oder einige Jahre früher belegt gewesen sein muß. Kennzeichnend sind besonders die vielen Aco-Becher, Rippenbecher und Becher mit innen gekehltem Rand, sowie frühe Terra sigillata-Gefäße, wie z.B. Schrägrandteller. Unter den Stempeln auf arretinischer Terra sigillata befinden sich neben der Ateius-Gruppe Stücke, die in den frühesten Neusser und Dangstettener Horizont gehören. Das Münzbild wird hauptsächlich von Copia- und Vienna-Münzen, sowie von Nemausus-Assen der 1. Serie bestimmt. Es wurde - unter

406

J.K. Haalebos: Die Früheste Belegung des Hunerberges in Nijmegen

Nijmegen

Haltern Oberaden

Neuss

Dangstetten

Lyon

Italien etc.

Olus Albius

=

x

Princeps (Naevii?)

=

x

Sphaerus

x

x

G. Tullius

x

x

P. Hertorius

x

x

x

x

L. Tetius Pamphylus

=

x

x

x

L. Tarquitius

x

x

x

x

x

L. Tettius [Samia?]

=

x

x

x

x

x

Sextus Varius (Niger)

=

x

x

x

Tab. 1. Nijmegen. Die Namen der ältesten arretinischen Töpfer, die das groβe Lager beliefert haben und ihr Erscheinen in einigen mittelaugusteischen Lagern, in Lyon und Italien. Zweizeilige Stempel sind in Nijmegen mit = markiert.

Literatur Bockius R. 1989 Ein römisches Scutum aus Urmitz, Kreis Mayen-Koblenz. Zur Herkunft und Verbreitung spindelförmiger Schildbuckelbeschläge im Gebiet nördlich der Alpen. Archäologisches Korrespondenzblatt 19: 269-282. Brunsting H. 1961 De Nijmeegse castra. Resultaten van de opgraving in 1960. Numaga 8: 49-67. Haalebos J.K. 1991 Das große augusteische Lager auf dem Hunerberg in Nijmegen. In R. Asskamp & S. Berke (Hrsg.) Die römische Okkupation nördlich der Alpen zur Zeit des Augustus. Bodenaltertümer Westfalens 26. (Münster): 97-107. Haalebos J.K. u.a. 1995 Castra und Canabae. Ausgrabungen auf dem Hunerberg in Nijmegen, 1987-1994 Libelli Noviomagenses 3. (Nijmegen): 9-26. Haalebos J.K. 1999 Das groβe Lager auf dem Hunerberg in Nijmegen (NL). In W. Schlüter & R. Wiegels (Hrsg.) Rom, Germanien und die Ausgrabungen von Kalkriese.Osnabrücker Forschungen zu Altertum und Antike-Rezeption 1. (Osnabrück) 381-399.

407

Limes XVIII

Abb. 1. Nijmegen-Hunerberg. Übersichtsplan über das große augusteische Lager mit dem Steinlager der Legio X Gemina in der nordostecke. Westlich der Legionsfestung sind die Steinbauten der flavischen Canabae eingetragen. Maßstab 1: 5000.

408

J.K. Haalebos: Die Früheste Belegung des Hunerberges in Nijmegen

Abb. 2. Nijmegen-Hunerberg. Übersichtsplan über das große augusteische Lager. Maßstab 1: 5000. 1 – Rand des Stabsgebäudes; 2 – Praetorium; 3-6 – Tribunenhäuser; 7-17 – Baracken; 18 – nordöstliche Ecktürme (2 Perioden); 19 – Osttor, Porta principalis dextra; 20 – Westtor, porta principalis sinistra mit anschließenden Türmen des Verteidigungswalles; 21-23 – Türme.

409

Limes XVIII

Abb. 3. Nijmegen-Hunerberg. Das Westtor des großen augusteischen Lagers. Maßstab 1: 500. Zeichn. E.J. Ponten. 60 – Augusteisches Tor mit Pfostenlöchern von einer Reparatur in den Türmen; 66 – ausgebrochener spätflavischer Steinkanal; 67-68 – in späterer Zeit ausgehobene Straßengrabchen entlang der augusteischen Via principalis; 69 – flavischer Holzbau; 70 – Teile einer augusteischen Baracke; 98 - Innengraben des augusteischen Lagers.

410

J.K. Haalebos: Die Früheste Belegung des Hunerberges in Nijmegen

Abb. 4. Nijmegen-Hunerberg. Die Lagerflächen von zwei Cohortes an der Innenseite des westlichen Verteidigungswalles des großen augusteischen Lagers. Vgl. Abb. 2, 11-12. Maßstab 1: 1000. Zeichn. E.J. Ponten.

411

Limes XVIII

Abb. 5. Nijmegen-Hunerberg. Übersichtsplan über das große augusteische Lager: grau gerastert die Lagerflächen der Cohortes, die Stelle der Principia und die westlich anschliessenden Häuser der Tribuni militum. Maßstab 1: 5000.

Abb. 6. Nijmegen-Hunerberg. Der an Hand der in einer augusteischen Grube gefundenen Metallreste rekonstruierte Schild mit spindelförmigen Umbo. Maßstab 1: 10. Zeichn. E.J. Ponten.

412

J.K. Haalebos: Die Früheste Belegung des Hunerberges in Nijmegen

Abb. 7. Nijmegen-Hunerberg. Eiserner spindelförmiger Umbo und Teile des bronzenen Randbeschlages eines Schildes, vgl. Abb. 5. Maßstab 1: 3. Zeichn. E.J. Ponten.

413

Limes XVIII

Abb. 8. Nijmegen-Hunerberg. Bronzener Schließhaken eines augusteischen Ringpanzers mit Punzinschrift auf der Rückseite: >(=centuria) MAN I.C.ANTONI. Maßstab 1: 1.

Abb. 9. Nijmegen-Hunerberg. Eiserne Pfeilspitze aus augusteischen Gruben. Maßstab 2: 3. Zeichn. R.P. Reijnen.

Abb. 10. Nijmegen-Hunerberg. Arretinische Stempel aus dem großen Lager. Maßstab 1:1 (Gefäße 1: 3). Zeichn. R.P. Reijnen.

414

Reduktion von numerus- und Kleinkastellen des obergermanischen Limes im 3. Jahrhundert Marcus Jae & Markus Scholz Der wiederholte Abzug von Truppen für die Perserkriege der 1. Hälfte des 3. Jhs. fürhrte am obergermanischen Limes zu Veränderungen. Der offenbar für längere Zeit absehbare Truppenmangel verlanlaßte die Militärführung vielleicht zu einer Neuverteilung der zurückgebliebenen Verbände. Wahrend wenige Standorte aufgegeben wurden, scheint man andere Kastelle verkleinert zu haben. Eindeutige Hinweise auf solche Reduktionen erbrachten jüngste Ausgrabungen im Numeruskastell Miltenberg-Ost. Sie werfen ein neues Licht auf seit längem bekannte Befunde Strukturen im Limeskastell Kapersburg (Taunus). In beiden Fällen scheint die Restbesatzung burgus-artige Befestigungen in einer Kastellecken errichtet zu haben. Die ursprünglichen Umwehrungen blieben jedoch bestehen - als Refugien für die Zivilbevölkerung nach den Erfahrungen des Germaneneinfalls von 233 n. Chr. ? Die planvoll wirkenden Umbauten kšnnten Bestandteile eines 'Wiederaufbauprogramms' in der Wetterau unter Maximinus Thrax oder Konsequenz eines massiven Truppenabzugs Gordians III. im Jahre 242 n. Chr. gewesen sein. ******* Troops from Upper Germany repeatedly participated in the Persian wars of the first half of the C3rd. The withdrawal of soldiers required reorganisation of the limes in Upper Germany. Few of the forts were abandoned but others seem to have been reduced. Clear evidence of C3rd reduction was furnished by the latest excavation in the numerus fort at Miltenberg-Ost. This evidence helps to reevaluate similar structures in the fort at Kapersburg (Taunus). It seems that in both cases the remaining garrison erected burgus-like fortifications within a corner of the fort. The original defences survived, perhaps as a 'shelter' for the civilian population during the Germanic raids in AD233. These rebuilding appear to have been well-planned and might have either been part of a reconstruction programme by Maximinus Thrax or as a reaction to a massive withdrawal of troops by Gordian III in AD242.

Rückkehrer oder Neuaushebungen ausgeglichen wurden, ist zu bezweifeln. Parallel dazu scheinen Teile der verbliebenen (Reiter-)Truppen in Städte des Hinterlandes zurückgenommen worden zu sein (Nuber 1997).

1. Etappen des Niedergangs Aus heutiger Sicht stellt sich das Ende des Limes als ein jahrzehntelanger, komplexer Prozeß des kontinuierlichen Niedergangs dar, der schließlich durch die sich in der Mitte des 3. Jhs. überschlagenden Ereignisse beschleunigt wurde und heute noch lange nicht in all seinen Schattierungen bekannt ist (Nuber 1990; Kuhnen 1992; Strobel 1999; Steidl 2000).

253 n. Chr. zog Valerian im Auftrag des Kaisers Trebonianus Gallus in Raetien Truppen zusammen, die ihn zum Gegenkaiser ausriefen. Mit diesem Heer, in das offensichtlich das Gros der Limestruppen eingereiht wurde, marschierte er zur Durchsetzung seines Machtanspruchs an die mittlere Donau, im Jahr darauf zog er gegen die Perser nach Kleinasien. Daß überlebende Soldaten noch einmal in ihre Limeskastelle zurückkehrten, ist sehr unwahrscheinlich. Germanen nutzten die Grenzentblößung zu Plünderungszügen. Spätestens die Ereignisse dieser Jahre müssen den im Limesgebiet verbliebenen Menschen vor Augen geführt haben, daß die Sicherheit der Grenzregionen nicht mehr zu gewährleisten war (Kortüm 1998: 59; Strobel 1999; Steidl 2000: 77). Der endgültige Zusammenbruch zumindest des obergermanischen Grenzschutzes dürfte eine Folge der Usurpation des Postumus 260 n. Chr. und seines unentschieden gebliebenen Machtkampfes mit Gallienus gewesen sein, in dessen Verlauf das Limesgebiet gewissermaßen zum „Niemandsland“ geriet (Nuber 1990: 52).

Als Hauptursache wird der wiederholte Abzug umfangreicher Kontingente der in den Limeskastellen stationierten Truppen für Feldzüge gegen die Perser, die die Ostprovinzen massiv bedrohten, sowie für Bürgerkriege und Abwehrkämpfe in anderen Reichsteilen herausgestellt. Als Kaiser Severus Alexander auch unter Einsatz obergermanischer und raetischer Truppen einen groß angelegten Feldzug gegen die Perser führte, nutzten Germanen 233 n. Chr. die Schwächung des Limes zu verheerenden Plünderungszügen (Reuter 1999: 533f.; Steidl 2000: 75). Die offenbar überraschende Katastrophe leitete für die am schwersten betroffenen Gebiete im nördlichen Obergermanien und in Westraetien trotz des anschließenden Wiederaufbaus unter Maximinus Thrax den Niedergang ein. Der „Raubbau“ an militärischen Resourcen wurde 236 n. Chr. durch die Expedition des Maximinus Thrax nach Pannonien sowie durch den Perserfeldzug Gordians III. ab 241/42 n. Chr. (Reuter 1999: 535; Okamura 1996: 33f.) fortgesetzt und stellte die lückenlose Grenzüberwachung sowie den wirksamen Schutz der rechtsrheinischen Provinzgebiete zunehmend in Frage. Ob die entstandenen Personallücken durch

2. Definition später Umbauten und Fragestellung Die Verringerung der Grenztruppen hatte auch Auswirkungen auf den baulichen Bestand der Limesanlagen, die von den wenigen Soldaten nur noch

415

Limes XVIII

teilweise besetzt und wiederaufgebaut bzw. instandgehalten werden konnten. Nutzungsänderungen bestimmter Funktionsgebäude, Zumauerungen von Kastelltoren, Verkleinerungen von Kastellbädern, Errichtung atypischer Gebäude über älteren Kastellinnenbauten sowie Verkleinerungen von Wehranlagen zeigen mancherorts eine Auflösung der Kastellstrukturen an. Derartige Spuren sind wiederholt angesprochen worden (Schönberger 1985: 418f.; Nuber 1990: 63; 1997; Reuter 1996; 1999; Steidl 2000: 77), doch stößt deren Analyse auf beträchtliche Schwierigkeiten, was eine umfassende Darstellung der Problematik bisher verhindert hat. Zum einen ist die große Mehrzahl dieser oft eher unscheinbaren Strukturen bei Altgrabungen zutage gekommen. Damaligen Grabungsmethoden und Wissensstand entsprechend, wurden sie in ihrer möglichen Bedeutung oft nicht erkannt und aus heutiger Sicht allzu kursorisch dokumentiert. Dies betrifft vor allem Maueranschlüsse und stratigraphische Überlagerungen älterer Bauten. Zum anderen fehlt fast immer chronologisch signifikantes Fundmaterial bzw. ist den Befunden nicht mehr zuzuordnen, so daß diese nur relativchronologisch als „spät“ eingestuft werden können. Indizien liefern die Überlagerung „etablierter“ Innenbauten sowie die Verwendung von Spolien.

nicht zur ursprünglichen Konzeption des Militärlagers gehört haben, doch gilt eine römische Datierung als sicher. Dies deutet zusammen mit dem wehrhaften Charakter dieser Mauer auf eine Abtrennung des nördlichen Kastellteils in einer späteren Nutzungsphase hin (Jae 1998: 80f.). Zu betonen ist, daß lediglich die Fundamentstückung dieser Quermauer erhalten war, die aus trocken gesetzten Sandsteinbruchstücken und einzelnen Lesesteinen bestand; Spolien fand man nicht. Trotz unterschiedlicher Konstruktionsweisen im Westen (durchgehende Rollierung) und im Osten (Streifenfundamente) beweisen gleichartiges Baumaterial, Bauflucht und stratigraphische Lage die gleichzeitige Entstehung und Zusammengehörigkeit der Teilstrukturen. Auf ihrer Nordseite schlossen sich in der Mauerflucht Pfostenstellungen und Reste schmaler Trockenmauerfundamente an, die auf die Existenz eines Wehrgangs, vielleicht auch auf kasemattenartige Anbauten schließen lassen (Jae 2000: 119-123). Im südöstlichen retentura-Bereich wurde über den Balkengräbchen und dem einplanierten Schutt abgebrannter Holzbaracken ein steinernes Badegebäude errichtet (Abb. 1, 4). Dieser stratigraphisch jüngste Befund ist durch die Überschneidung einer Grube, deren Inhalt sich gut mit dem Händlerdepot von Langenhain vergleichen läßt, in nachseverische Zeit datiert (Jae 2000: 138-143).

In diesem Beitrag sollen Baubefunde in Kastellen des obergermanischen Limes besprochen werden, die massive Verkleinerungen während der Spätzeit erkennen lassen. Mit diesen Reduktionen verbindet sich die Frage, ob sie auf provisorische „Notbehelfe“ gemäß dem örtlichen Bedarf der Restbesatzungen zurückgehen oder ob in ihnen auch weitergehende Versuche der Militärführung durchscheinen, das Funktionieren des Überwachungssystems trotz verminderter Kräfte aufrecht zu erhalten.

Die Ausgrabungsergebnisse lassen also den Schluß zu, daß das Kastellareal nach erheblichen Brandzerstörungen und Verringerung der Besatzungsstärke völlig neu strukturiert wurde. Die Besatzung zog sich offenbar in einen kleinen Teil der ursprünglichen Anlage zurück, vom Rest durch eine starke Wehrmauer abgegrenzt (Jae 1998: 82). Offen bleibt zunächst, ob die Reduktion die gesamte praetentura oder nur deren nordöstliches Viertel umfaßte, was wahrscheinlicher ist, falls die Interpretation der Rollierungsunterbrechungen als Toranlage zutrifft. Die bauliche Abtrennung des NO-Viertels liefe auf ein Kleinkastell von c.34 x 35m Ausdehung heraus. Diese Größenordnung entspricht dem 32.6 x 32.6m (0.1ha) messenden centenarium bei Burgsalach am rätischen Limes, dem bis heute einzigen bekannten Limeskastell, das während des 2. Drittels des 3. Jhs. neu errichtet wurde (Schleiermacher 1962: 195). Möglicherweise eignet sich dieser Vergleich auch als Anhaltspunkt für die maximale denkbare Besatzungsstärke, die wahrscheinlich aber weit weniger als 100 Mann betrug. Gegenwärtig ist das libysche Kastell Bu Njem u. W. das einzige Grenzkastell, dessen Größe und Sollstärke ebenso bekannt sind wie – durch den glücklichen Fund der Ostraka – dessen tatsächliche Besatzungsstärke in den 250er Jahren. Bu Njem (1.2ha) war für eine bis zu 200 Mann starke Garnison, bestehend aus einem numerus sowie einer Legionsvexillation, ausgelegt. Aus den täglichen Stärkemeldungen der Garnison geht hervor, daß deren Mannschaftsstärke in den Jahren 254-259 n. Chr. auf einem erschreckend niedrigen Niveau zwischen c.45 und 60 Mann stagnierte (Marichal 1992: 63-66). Ob für diesen enormen Schwund dieselben

3. Numerus-Kastell Miltenberg-Ost (Abb. 1-2) Die umfangreichsten späten Umbauten in einem Kastell am obergermanischen Limes lassen sich gegenwärtig in Miltenberg-Ost nachweisen. Flächige Ausgrabungen erfaßten 1998 erstmals in größerem Umfang Innenbauten des 0.6ha großen Lagers. Untersucht wurde rund ein Sechstel der Innenfläche in der Osthälfte des Kastells inklusive der porta principalis dextra (Jae 1998: 80; 2000: 104). Die vorgefundenen Baustrukturen gewähren bemerkenswerte Einblicke in das Schicksal des Kastells während des 2. Drittels des 3. Jhs. Hervorzuheben ist eine an die Innenseite der nördlichen Wange der porta principalis dextra angelehnte, über der via principalis verlaufende Steinrollierung (Abb. 1, 2-3). Dadurch wurden Tordurchfahrt und Kastellstraße zumindest teilweise unbrauchbar. Aufgrund ihrer Massivität mit durchschnittlich 1.70m Breite wird man der ursprünglich darauf stehenden Mauer einen fortifikatorischen Charakter nicht absprechen können, eine größere Unterbrechung etwa in Flächenmitte erweckt den Eindruck einer Toranlage. Der beschriebene Befund kann 416

Marcus Jae & Markus Scholz: Reduktion von numerus- und Kleinkastellen des obergermanischen Limes Ursachen verantwortlich waren wie für entsprechende Entwicklungen in den Nordprovinzen (z. B. Truppenabzüge für Valerians Perserkrieg) oder ob sich hier gesonderte Verhältnisse am afrikanischen Limes bemerkbar machen, kann hier nicht ausdiskutiert werden. Dieses Beispiel veranschaulicht jedoch, wie weit Standorte jener Zeit unter ihre Sollstärke geraten konnten.

Die übrigen Steinbauten stammen aus der Spätzeit, von denen hier nur die wichtigsten kurz erläutert werden sollen. So ist das Praetoriumbad (Abb. 3, 5) entweder ganz oder teilweise abgerissen worden, da über dessen südlichem Vorraum ein kleiner Wohnbau mit L-förmiger Hypokaustheizung (Abb. 3, 6) errichtet wurde. Neben anderen Spolien waren hier Trümmer eines Fortuna-Altars vermauert (CIL XIII 7440 = ORL B Nr. 12, 32 Nr. 3), der möglicherweise zuvor in dem abgebrochenen Vorraum des Praetoriumbades oder – wahrscheinlicher – im Kastellbad extra muros aufgestellt war.

Die Neuerrichtung eines Bades intra muros ist für ein Kastell am obergermanischen Limes höchst ungewöhnlich und findet – abgesehen von dem in mancherlei Hinsicht eine Ausnahmestellung beanspruchenden Kastell Niederbieber (Reuter & Steidl 1997: 215-234, bes. 228) – seine besten Vergleiche erst in der Festungsarchitektur der Spätantike. Die Weiternutzung bzw. –bewohnung des übrigen Kastellareals läßt sich in Miltenberg-Ost ferner anhand stratifizierter Gruben und Feuerstellen belegen. Auch die Umwehrung des numerus-Kastells scheint weiterhin bestanden zu haben, worauf die Zusetzung der porta decumana mit einer Trockenmauer als zusätzliche Sicherungsmaßnahme hinweist (ORL B Nr. 38a, 2). Der Torweg der porta principalis dextra war durch die Quermauer zur Hälfte blockiert, durch die freigehaltene Passage führte der Abwasserkanal des Badegebäudes. Für den Hauptverkehr standen also nur noch die porta praetoria und allenfalls die in Teilen ergrabene porta principalis sinistra zur Verfügung.

Aufgrund seiner Lage und Gestalt kann der schiefwinklige „Wallbau“ (Abb. 3, 4) nicht Bestandteil der ursprünglichen Lagerkonzeption gewesen sein, da er das praefurnium des Praetoriumbades (Abb. 3, 5) stört und nachträglich in den Erdwall der severischen Umwehrung eingebaut wurde (Außenmaße: 13.35/13 x 7.58m). Dieser Bau entbehrte eines ebenerdigen Eingangs, seine Mauerstärke von 1.351.5m entspricht gängigen Maßen römischer Fortifikationen. Ob der Zugang über hölzerne Außentreppen/-leitern und/oder über eine Türöffnung zur Wallkrone erfolgte, bleibt fraglich. Das zweischalige Mauerwerk besteht aus den üblichen groben Taunusschiefer-Handquadern unter Verwendung vereinzelter Spolien (Ziegel- und Sandsteinbruchstücke). Aus seinem Inneren stammt ein Teil der relativchronologisch jüngsten Funde des Standortes, darunter Fragmente Rheinzaberner Reliefsigillata des Pervincus und Ware mit E 31 (Pervincus II), von Drag. 41, Niederbieber 19 sowie „Urmitzer Ware“. Sie zeigen an, daß der Bau bis zum Schluß genutzt wurde. Unserer Ansicht nach ist der „Wallbau“ am zutreffendsten als ein spätes Refugium zu betrachten, das einer kleinen Mannschaft Zuflucht bot (Scholz 2001: 70-73).

4. numerus-Kastell Kapersburg (Abb. 3) Die jüngsten Grabungsergebnisse im Kastell MiltenbergOst werfen ein neues Licht auf die seit langem bekannte Konzentration von Steingebäuden im NO-Viertel des Taunuskastells Kapersburg (ORL B Nr. 12, 12-14 u. Taf. 2). Zwar liegen nur knappe stratigraphische Angaben vor, doch ist offensichtlich, daß all diese Bauten unmöglich gleichzeitig bestanden haben können. Mit Ausnahme der principia gab es im übrigen Kastellareal keine steineren Innenbauten. Die in nachseverischer Zeit fortlaufende Münzreihe zeigt, daß das Kastell bis zur endgültigen Aufgabe des Limes Bestand hatte (Kortüm 1998: 36). Die jüngste Münze aus dem Kastellareal wurde 256 n. Chr. geprägt, ein Antoninian des Postumus (266 n. Chr.) stammt aus dem Kastellbad. Dank abgeschiedener Lage im Wald bestehen gute Erhaltungsbedingungen für jüngste Strukturen.

Das Ensemble von Steingebäuden wird nach Westen durch eine 0.7m starke Quermauer (Abb. 3, 8) begrenzt, die entlang der via praetoria bis zu den principia verläuft. An ihr südliches Ende wurde später ein turmartiges Bauwerk (Abb. 3, 3) von 6.5 x 5.45m mit wehrhaften 1.20-1.25m starken Mauern angefügt. Der 0.80m breite Durchlaß mit Flankenmauern bezeugt vermutlich einen nicht ausgegrabenen Kellerabgang. Für die Chronologie dieses Bauwerks ist wichtig, daß es über der hölzernen Vorhalle der principia liegt, die folglich bei dessen Errichtung nicht mehr bestand. Es gibt weitere Anzeichen dafür, daß die principia in der Spätzeit zurückgebaut, vielleicht sogar ganz aufgegeben wurden (Scholz 2001: 66). Mit aller gebotenen Vorsicht neigen wir dazu, den Bau als Turm einer späten Befestigung oder zumindest als Wohnbau wehrhaften Charakters zu betrachten. An seinem römischen Ursprung bestehen keine Zweifel, doch bezeugt seine Plazierung an zentraler Stelle die Auflösung der alten Lagereinteilung. Möglicherweise rühren diese Baubefunde von dem Versuch her, das NO-Viertel des Kastells durch eine Befestigung abzutrennen. Ein in Analogie zu Miltenberg-Ost zu postulierender Mauerfortsatz nach Osten bis zur porta principalis dextra harrt des Nachweises, falls die von uns vermutete

Der Ausbau des jüngeren Steinkastells auf 1.6ha Größe erfolgte um 200 n. Chr. (Scholz 2001: 56-60), dessen Garnison ein n(umerus) N(...) sowie veredarii (Meldereiter) bildeten (CIL XIII 7441; 7439 = ORL B Nr. 12, 32 Nr. 4; 32 Nr. 2). Sicher in das Jahr 209 n. Chr. (CIL XIII 7441) ist das horreum datiert (Abb. 3, 1). Es liegt über Fundamenten eines älteren Komplexes (Abb. 3, 2), der vermutlich zusammen mit dem Apsidenbau (Abb. 3, 5) als Bestandteil des praetorium anzusehen ist, das im frühen 3. Jh. teilweise verändert wurde. Sein Westteil scheint aber weiterbestanden zu haben, worauf eine fünfarmige Kanalheizung mit zentraler Verteilerkammer hinweist, ein für das 3. Jh. charakteristischer Heizungstyp. 417

Limes XVIII

Reduktionsbefestigung je vollendet wurde.

vermitteln die Reduktionsbauten in der Kapersburg, die nicht zerstört worden war, eher den Eindruck eines mehrperiodigen Stückwerks. Die Errichtung eines Rückzugskastells wird vermutet, bedarf jedoch des Beweises durch neue Ausgrabungen. Kartiert man die oben besprochenen Reduktionsbefunde, so fällt auf, daß sie sich beim Stand der gegenwärtigen Forschung auf den nördlichen obergermanischen Limes konzentrieren, insbesondere auf die nordwestlichen Strecken 1-2 (Abb. 5).

Östlich der Steingebäude fand man Partien von Trockenmauern (Abb. 3, 9), die aus groben, teilweise wuchtigen Bruchsteinen bestehen. Sie sind, von Gestrüpp überwuchert, teilweise noch heute im Gelände auszumachen. Aufgrund ihrer oberflächlichen Lage darf man sie für die mithin spätesten Baumaßnahmen im Kastell in Anspruch nehmen. Auf ähnliche Trockenmauern stieß man in der Saalburg. Auch dort entstanden sie vermutlich erst in der Schlußphase des Kastells (ORL B Nr. 11, Taf. 2, p, q). Wahrscheinlich handelte es sich um Schwellbalkenunterlagen von Fachwerkbauten.

Am Wetteraulimes gewinnt in jüngerer Zeit ein anderes, dazu durchaus passendes Phänomen an Konturen, nämlich die offenbar vorzeitige Auflassung einiger Kleinkastelle (unter 0.6ha Fläche). Dies trifft auf den Vorposten Butzbach-Degerfeld zu, der wahrscheinlich 233 n. Chr. aufgelassen wurde (Simon & Köhler 1992: 104). Das Kleinkastell Holzheimer Unterwald hat in dieser Zeit Rückbaumaßnahmen hinnehmen müssen und war kaum bis zum Limesende besetzt (Seitz 1999). Spätestens 233 n. Chr. wurde das Kleinkastell Echzell-Haselheck geräumt (Mitteilung B. Steidl). Um dieselbe Zeit könnte auch das westlich von Butzbach gelegene „Zwischenkastell“ Hunnenkirchhof zerstört worden sein. Der knappe Ausgrabungsbericht läßt zumindest auf niedergebrannte Holzbauten schließen (ORL A Strecke 4-5, 79). Ob sich ein Wiederaufbau anschloß, bleibt ungewiß. Mehr oder weniger deutliche Zerstörungsspuren sind in den Stützpunkten Hainhaus (ORL A Strecke 4-5, 108), Feldheimer Wald (Wp 69) Lochberg (Wp 89) und Staden (Wp 94) aufgefallen (ORL A Strecke 4-5 S. 125, 139, 141).

Die Befunde der Kastelle Miltenberg-Ost und Kapersburg bezeugen keine Einzelschicksale, sondern scheinen an den nördlichen obergermanischen Limesstrecken 1 und 2 in den befestigten Einbauten der Kleinkastelle Anhausen, Hillscheid und vielleicht auch Dörsterberg Parallelen zu finden, auf die bereits Reuter (1996: 76-79) hingewiesen hat (vgl. Abb. 4). Während in Anhausen, wo verhältnismäßig gründliche Untersuchungen stattfanden, tatsächlich eine (undatierte) Reduktion nachgewiesen wurde (ORL A Strecke 1, 56f.), fanden im Kastell Hillscheid nur flüchtige Aufdeckungen statt. Der im ORL publizierte Plan läßt keine Unterscheidung zwischen nachgewiesenen und ergänzten Strukturen zu, er wirkt geschönt. Die Ansicht der Ausgräber, das befestigte „Reduits“ sei gleichzeitig mit der Gesamtanlage entstanden, mag man daher bezweifeln (ORL A Strecke 1, 124). Die nebulöse Dokumentationslage begründet jedoch Skepsis gegenüber dem gesamten Befund, ebenso wie im Fall des Kleinkastells Dörsterberg. Bei allen drei Anlagen fehlen auswertbare Angaben zu Innenbebauung und Datierung, außer daß es sich jeweils um die jüngsten Aktivitäten am Ort handelt.

Einzuschränken bleibt, daß nirgendwo das jeweilige Ausmaß der Brände noch deren Datierung gesichert sind. Einzig die stets erwähnten Brocken verbrannten Hüttenlehms begründen den Bezug auf zerstörte Fachwerkgebäude, und nicht auf sonstige Feuerstellen. Das von jedem dieser Kleinkastelle gemeldete Fundmaterial ist leider nie ausgewertet worden. Das betrifft natürlich auch die Auflassungsdaten dieser Zwischenposten. Gegenwärtig eignet sich die Feststellung, daß im Gegensatz zur Taunuslinie in den meisten Wetterau-Kleinkastellen Brandspuren beobachtet wurden, lediglich als Ausgangspunkt künftiger Untersuchungen. Dabei wäre zu prüfen, ob sich diese Brandspuren mit dem Katastrophenhorizont von 233 n. Chr. korrelieren lassen. In den skizzierten Rahmen dürfte sich ferner die schon vor längerem beobachtete Schließung von Limesübergängen, so bei Butzbach-Degerfeld und Holzheim, vielleicht sogar die Schaffung des Wall-Graben-Systems am obergermanischen Limes fügen.

Dasselbe Schema liegt ferner den spätantiken Reduktionskastellen von Eining und Dormagen zugrunde (Mackensen 1994: 479-513; Gechter 1997: 93f.), die ebenfalls in den Ecken mittelkaiserzeitlicher Kastelle unter Verwendung noch vorhandenen Mauerwerks errichtet worden waren. In Miltenberg, Eining und Dormagen war offensichtlich die Nähe zum jeweiligen Flußufer für die Wahl der Ecke ausschlaggebend. In allen genannten Parallelfällen blieb die ursprüngliche Umwehrung weiterhin bestehen. Dies trifft auch auf die spätzeitliche Kapersburg zu, dokumentiert durch die gute Erhaltung der Kastellmauern. Hier wählte man das am höchsten gelegene Kastellareal, das zugleich Standort steinerner Innenbauten war, als Refugium.

Im Gegensatz zu der sich im Bereich von Taunus und Wetterau andeutenden vorzeitigen Auflassung vieler Kleinkastelle hatten solche am vorderen Limes offenbar bis zum Schluß Bestand, worauf Münzen bis Gallienus aus den Anlagen Haselburg und Rötelsee hindeuten (Schönberger 1985: 492 f.).

5. Reorganisationsprogramm oder lokale Notmaßnahmen und Behelfskonstruktionen? In der Zusammenschau hinterlassen all diese ausschnitthaften und oft unsicheren Befunde zunächst ein heterogenes Bild. Während in Miltenberg-Ost die Neuorganisation des Kastellinneren wohl als einheitliche Maßnahme nach einer Zerstörung (233 n. Chr.?) erfolgte,

Die 418

Summe

dieser

Veränderungen

läßt

auf

eine

Marcus Jae & Markus Scholz: Reduktion von numerus- und Kleinkastellen des obergermanischen Limes Neukonzeption zumindest der Taunus- und Wetteraulinie schließen, die wohl infolge des Germaneneinfalls von 233 n. Chr. notwendig geworden war. Die Reduktionsbauten können, müssen aber nicht mit den anschließenden Wiederaufbaumaßnahmen unter Maximinus Thrax in Zusammenhang stehen (Reuter 1999), zumal vorausgegangene Zerstörungen nur in Miltenberg-Ost – hier ist ein solcher Kontext wahrscheinlich – nicht aber in den Taunuskastellen nachgewiesen sind. Keiner der Befunde ist absolut datiert; es läßt sich nur sagen, daß man hier die jeweils letzten nachweisbaren römischen Bauaktivitäten am Ort faßt. Im Falle der Kastelle, für die aufgrund ausgedehnter Brandschichten Zerstörungen 233 n. Chr. angenommen werden (Butzbach, Arnsburg, Altenstadt, Echzell, Oberflorstadt; zum Ausmaß der Zerstörungen: Schönberger 1985: 418; Reuter 1999: 534) sind befestigte Reduktionsbauten nicht nachgewiesen. In Echzell deutet der vollständige Wiederaufbau der principia vielmehr darauf hin, daß man zunächst den Wiedereinzug einer gleichstarken Besatzung wie zuvor plante (Baatz 1965: 146).

kaum bedeutender Standort war die Kapersburg. Nicht alle aus den Limeskastellen abgezogenen Truppen müssen dabei in das Feldheer für auswärtige Expeditionen eingegliedert worden sein. Berittene Teileinheiten, so z. B. die in der Kapersburg stationierten veredarii, könnten als mobile Melde- und Eingreifreserven auch Operationsbasen im Hinterland besetzt haben (Nuber 1997: 155). Vergleichbare Abkommandierungen könnten die turmae des im Kastell Miltenberg-Ost stationierten numerus exploratorum Seiopensium betroffen haben. Ein gutes Beispiel für die postulierte Umschichtung am Limes verbliebener (Fuß-) Truppen bietet das Kastell Niederbieber, dessen letzte Besatzung, aus dem bekannten Fund von Signumteilen zu schließen, u. a. aus einer Abordnung der cohors VII Raetorum des nachbarten Kastells Niederberg bestanden zu haben scheint (Nuber 1997: 153). An den nordwestlichen Abschnitten des obergermanischen Limes (Strecken 1-3), an denen die reduzierten Kleinkastelle Anhausen, Hillscheid und Dörsterberg liegen, wurden wiederholt verbarrikadierte Tore in numerus-Kastellen beobachtet, so in Holzhausen, Arzbach, Kemel und vielleicht auch in Bad Ems und Hunzel (vgl. die Ausführungen der jeweiligen ORL-Bände). Die von den Ausgräbern des Kastells Holzhausen getroffene Aussage, daß die Verbarrikadierung der Tore Zeugnis einer Belagerung und anschließender Eroberung sei, in der das Lager zugrunde ging, gilt weiterhin (ORL B Nr. 6, 13). In den übrigen Fällen könnte sich hierin auch eine prophylaktische Befestigungsverstärkung widerspiegeln, durch die man vielleicht auch Wachpersonal einsparen mochte. Reger Verkehr durch die Kastelltore scheint jedenfalls nicht mehr geherrscht zu haben.

Umgekehrt setzen bauliche Reduktionen voraus, daß man die Hoffnung auf Wiedererlangung vormaliger Sollstärke aufgegeben hatte. Eine solche Situation ist am ehesten nach den wiederholten Truppenabzügen der Folgejahre denkbar. Dabei darf die mögliche psychologische Wirkung auf die zurückgebliebenen Soldaten (und Zivilisten) nicht unterschätzt werden: die nur schwer wieder rückgängig zu machenden Verkleinerungen konnten als Signal aufgefaßt werden, daß eine nachhaltige Verbesserung der Lage nicht mehr möglich war oder „von oben“ gar nicht mehr angestrebt wurde. Eine Anordnung zu Rückbauten von übergeordneter Stelle läßt sich nicht erkennen, da es nicht gelingt, die an einzelnen Standorten beobachteten Maßnahmen chronologisch zu parallelisieren. Es fehlen aber auch – kaum zufällig – nachseverische Bauinschriften, die diese Annahme untermauern könnten.

Der erst mit Verzögerung ab der 2. Hälfte/Ende des 2. Jhs. einsetzende Ausbau der Strecken 1-3 und ihre im Vergleich zur Wetterau-, Main- und vorderen Limeslinie schwache Besetzung mit numerus-Kastellen lassen erahnen, daß diese Abschnitte von Beginn an als eine weniger akut gefährdete Grenzzone eingestuft wurden. Die Gegenkartierung der bis heute bekannten, im Vorfeld des Limes gelegenen germanischen Siedlungszentren unterstützt diese Einschätzung (v. Schnurbein 1992: 73). Es besteht also durchaus Anlaß zu vermuten, daß zumindest die Strecken 1-2 frühzeitiger Truppenausdünnungen hinnehmen mußten als etwa die Wetterau-, Main- oder vordere Linie. Zu betonen ist aber, daß offenbar kein Kastell ab numerus-Größe ganz wegfiel.

Ohne Beweise dafür anbringen zu können, wäre folgender Ablauf immerhin vorstellbar: Der unmittelbar an das Katastrophenjahr 233 n. Chr. anschließende Wiederaufbau verfolgte die schnelle Wiederinbetriebnahme bestehender Einrichtungen, wahrscheinlich verbunden mit einem ersten Schrumpfungsprozeß in Gestalt der Aufgabe von Kleinkastellen. An strategisch wichtigen Positionen, zumal die im Brennpunkt der Ereignisse von 233 n. Chr. gelegenen Kastelle, wie z. B. Echzell, wurden angesichts der offenbarten Feindbedrohung in möglichst schlagkräftigem Umfang belassen bzw. wiederhergestellt. Die zunächst wieder stabilisierte Lage verschlechterte sich nachhaltig durch nicht vorhersehbare Truppenabzüge 236 und/oder 241/42 n. Chr., als erkennbar werden mußte, daß das vormalige Limeskonzept auf absehbare Zeit nicht mehr greifen würde. Spätestens jetzt war man gehalten, die Besatzungen abgelegener Kastelle bzw. nicht unmittelbar gefährdeter Linien neu zu verteilen bzw. an strategischen Brennpunkten (nördliche und östliche Wetteraulinie, Mainlinie, größere vici des Limeshinterlandes) zu konzentrieren. Ein solch abgelegener, in dieser Zeit wohl

Für Kapersburg und Miltenberg-Ost ist eine Minimalmannschaft (s. o.) zu erwarten, die zwar kaum Schutzfunktionen, wohl aber die lineare Überwachung aufrecht zu erhalten vermochte, auf die offensichtlich noch immer Wert gelegt wurde. Die alten Kastellareale konnten von den wenigen Menschen nicht mehr gepflegt, geschweige denn verteidigt werden. Nach 233 n. Chr. wird in der Wetterau ein deutlicher Schrumpfungsprozeß der Kastellvici faßbar, so in Langenhain, Butzbach, Echzell und Oberflorstadt (Steidl 1996: 24). Bewohnten nun 419

Limes XVIII

zurückgebliebene bzw. überlebende Zivilisten die frei gewordenen Kastellflächen (Nuber 1990: 63)? Frauenspezifisches Fundmaterial der Spätzeit, z. B. Haarnadeln, aus den betreffenden Arealen der Kastelle Kapersburg und Miltenberg-Ost verlocken zu dieser Überlegung. Randbereiche des Kastellvicus von Miltenberg-Ost mit Töpferöfen wurden 1998 ergraben (Jütting, Lebeda & Wernard 2000). Das in Auswahl publizierte Fundmaterial belegt den Fortbestand dieses Siedlungsareals im 2. Drittel des 3. Jhs. nicht, doch bleibt die Auswertung dieser Grabung abzuwarten. Der Kastellvicus der Kapersburg ist unerforscht, so daß die Möglichkeit einer „Gegenprobe“ entfällt.

(ed.) Roman Frontier Studies 1997 Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 533-537. Reuter M. & Steidl B. 1997 Eine neue Statuenbasis für Septimius Severus aus dem Kastell Niederbieber. Neue Aspekte zum Gründungsdatum des Lagers. In H.-H. Wegner (Hrsg.) Bericht der Arch. Mittelrhein u. Mosel 5 (Trier). Schleiermacher W. 1962 Centenaria am rätischen Limes. In: J. Werner (Hrsg.), Aus Bayerns Frühzeit. Festschrift Friedrich Wagner. (München ): 195-204. Scholz M. 2001 Die Keramik des Limeskastells Kapersburg. (ungedr. Diss Freiburg). Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Bericht der Römische-Germanische Kommission 66: 321-495. Schnurbein v. S. 1992 Perspektiven der Limesforschung. In Der römische Limes in Deutschland. Arch. Deutschland Sonderh. Seitz G. 1999 Das Kastell Holzheimer Unterwald. Arch. Denkmäler Hessen 133. (Wiesbaden). Simon H.-G. & Köhler H-J. 1992 Ein Geschirrdepot des 3. Jahrhunderts. Grabungen im Lagerdorf des Kastells Langenhain. Mat. Röm.-Germ. Keramik 11. Steidl B. 1996 Vom römischen Provinzterritorium zum Siedlungsgebiet der alamannischen Bucinobanten. Die Wetterau im 3. Jahrhundert n. Chr. In E. Schallmayer (Hrsg.) Niederbieber, Postumus und der Limesfall. Stationen eines politischen Prozesses. Saalburg-Schr. 3. (Bad Homburg ): 22-30. Steidl B. 2000 Der Verlust der obergermanisch-raetischen Limesgebiete. In L. Wamser, Chr. Flügel & B. Ziegaus (Hrsg.) Die Römer zwischen Alpen und Nordmeer. Zivilisatorisches Erbe einer europäischen Militärmacht. (Mainz): 75-79. Strobel K. 1999 Pseudophänomene der römischen Militärund Provinzgeschichte am Beispiel des „Falles“ des obergermanisch-raetischen Limes. Neue Ansätze zu einer Geschichte der Jahrzehnte nach 253 n. Chr. an Rhein und oberer Donau. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 9-33.

Literatur Baatz D. 1965 Limeskastell Echzell. Kurzbericht über die Grabungen 1963 und 1964. Saalburg-Jahrbuch 22: 139-157. Gechter M. 1997 Das spätantike Kastell Dormagen. Archäologie im Rheinland 1996: 93f. Jae M. 1998 Eine Flächengrabung im Numeruskastell Miltenberg-Ost. Das Archäologie Jahr in Bayern 1998: 80-82. Jae M. 2000 Eine Flächengrabung im Numeruskastell Miltenberg-Ost auf der Gemarkung der Marktgemeinde Bürgstadt im Jahre 1998. Erste Einblicke in die innere Struktur des Kastells. Mainfränkische Studien 67: 103146. Jütting I., Lebeda A. & Wernard J. 2000 Der neu entdeckte Vicus von Miltenberg, Lkr. Miltenberg. Ergebnisse und ausgewählte Fundgattungen der Grabung 1998. Mainfränkische Studien. 67: 147-163. Kortüm K. 1998 Zur Datierung der römischen Militäranlagen im obergermanisch-rätischen Limesgebiet. SaalburgJahrbuch 49: 5-65. Kuhnen H.-P. (Hrsg.) 1992 Gestürmt – Geräumt – Vergessen? Der Limesfall und das Ende der Römerherrschaft in Südwestdeutschland. (Stuttgart). Mackensen M. 1994 Die Innenbebauung und der Nordvorbau des spätrömischen Kastells Abusina/Eining. Germania 72/2: 479-513. Marichal R. 1992 Les Ostraca de Bu Njem. Libya Antiqua suppl. 7. (Tripolis). Nuber H.U. 1990 Das Ende des Obergermanisch-Raetischen Limes - eine Forschungsaufgabe. In K. Schmid, H. Steuer & Th. Zotz (Hrsg.) Archäologie und Geschichte des ersten Jahrtausends in Südwestdeutschland. (Sigmaringen): 51-68. Nuber H.U. 1997 Späte Reitertruppen in Obergermanien. In W. Groenmann-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 151-158. Okamura L 1996 Hoards lost during third-century „Wirren“. In E. Schallmayer (Hrsg.) Niederbieber, Postumus und der Limesfall. Stationen eines politischen Prozesses. Saalburg-Schr. 3 (Bad Homburg): 31-37. Reuter M. 1996 Der Limesfall im Spiegel ausgewählter Befunde in Kleinkastellen und Wachttürmen. In E. Schallmayer (Hrsg.), Niederbieber, Postumus und der Limesfall. Stationen eines politischen Prozesses. Saalburg-Schr. 3. (Bad Homburg): 76-83. Reuter M. 1999 Der Wiederaufbau des obergermanischraetischen Limes unter Maximinus Thrax. In N. Gudea

420

Marcus Jae & Markus Scholz: Reduktion von numerus- und Kleinkastellen des obergermanischen Limes

Abb. 1. Kastell Miltenberg-Ost. Gesamtplan der Grabung 1998. 1 - Kastellmauer; 2 - porta principalis dextra; 3 - jüngere Quermauer; 4 - Steingebäude (Bad); 5 - Abwasserkanal (des Bades).

421

Limes XVIII

Abb. 2. Kastell Miltenberg-Ost mit später Innenbauung (Grabung 1998) und zugemauerter porta decumana. M 1 : 2000.

Abb. 3. Kastell Kapersburg (Plan nach D. Baatz, RiH). Schwarz: Steinbauten des jüngeren Numeruskastells; Numerierung: vgl. Text. M 1 : 2000.

422

Abb. 4. Vermutlich in der Spätzeit reduzierte Kleinkastelle: 1 Anhausen; 2 Hillscheid; 3 Dörsterberg (nach ORL und Reuter 1996). M 1: 1000.

Marcus Jae & Markus Scholz: Reduktion von numerus- und Kleinkastellen des obergermanischen Limes

423

Limes XVIII

Abb. 5. Obergermanischer Limes: reduzierte Kastelle (‫ ;)ٱ‬blockierte Kastelltore (×).

424

Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr. Stefan Neu und Matthias Riedel Over the last few decades a series of excavations and test-pits have been carried out along the waterfront of Colonia Claudia Ara Agrippenensium. This has led to the publication of a number of academic articles. The most important result to emerge shows that the harbour within the arm of the Rhine was only in use during the C1st. Then, through a combined process of natural silting and the depositing of domestic waste, the area was in-filled. The harbour was subsequently moved to the banks of the open river. In this lecture we will try to present a more detailed picture of the in-filling of the Rhine than has been previously shown. Furthermore new theories relating to the eastern side of the river - the site of the pre-colonial town (the oppidum Ubiorum) - will be briefly touched upon.

Das Kölner Rheinufer (Abb. 1) ist ein Abschnitt des niedergermanischen Limes. Es erscheint daher angebracht, angesichts neuerer Ausgrabungen und Sondierungen sowie Aufsätze, in denen dieses Thema eine Rolle spielt, einige Aspekte dieses Grenzabschnittes des Imperium Romanum hier zu behandeln. Während der Vorbereitung dieses Referates wurde das Ergebnis, nämlich die Tatsache, dass der vor der Stadt fließende Rheinarm bereits im ersten Jahrhundert n. Chr. zusandete und systematisch verfüllt wurde, in drei neuen Artikeln betont (Schütte 1999: 2ff.; Schütte & Gechter 2000: 78ff.; Knörzer & Neu 1998: 445ff.). Hier sei der Vorgang der Verlandung detaillierter dargestellt.

Komplex ausgegliedert und in einem Neubau in der Nähe des Rathauses neu eröffnet) eine größere Fläche des Ufers, des davor liegenden Flussarmes und der Insel freigelegt und – partiell – archäologisch untersucht (Neu 1989: Beilage). Als erste Baumaßnahme wurde direkt östlich des Römisch-Germanischen Museums (Abb. 1: c), also noch westlich der römischen Stadtmauer und damit im Bereich der ummauerten Stadt, eine Ausschachtung für ein elektrisches Umspannwerk baubegleitend dokumentiert, die - wie die beschriebene an der Ostseite der Stadtmauer mächtige Schichten dunklen Erdreichs mit Keramik- und organischen Resten ans Licht brachte (Abb. 2). Diese lagen nicht waagrecht wie Schwemmschichten, sondern folgten in einem Winkel von ca. 20o (1: 5) dem sich gegen den Rheinarm neigenden Gelände. Ihre Mächtigkeit betrug durchschnittlich 3 - 4m. Die tiefste erfasste Stelle lag bei 37.2 NN; also etwas höher als die Sohle der Doppelfeldschen Untersuchung. Die in diesen Schichten gefundenen Artefakte entsprechen ihrer Zusammensetzung nach durchaus den „Hafenfunden vom Alten markt“. Sie sind in die zweite Hälfte des ersten und das frühe 2. Jahrhundert zu datieren.

Nach H.J. Lückgers 1936 veröffentlichten Beobachtungen (Lückger 1936), welche die bis dahin oft angezweifelte Existenz der Rheininsel bewiesen, waren es O. Doppelfelds 1953 publizierte „Hafenfunde vom Alten markt“ (Doppelfeld 1953: 102ff.), die - wenn auch nur an einem bestimmten Punkt – die Verfüllschichten des Rheinarmes darstellten. Damals wurde für die Gründung eines Geschäftshauses ein Schacht von 3m x 3m an der Westseite des Rheinarmes nahe der Stadtmauer - Ostseite und c.250m südlich der NO – Ecke der Stadtmauer (Abb. 1: a) bis auf den gewachsenen Kies gegraben. Darin konnte Doppelfeld - nach 3 Höhenniveaus unterschieden zahlreiche Artefakte aus einem Bereich zwischen 38 und 36.5m NN bergen. Es handelte sich dabei natürlich überwiegend um Keramik-, aber auch Metallfragmente. In dem feuchten Boden waren darüber hinaus organische Reste erhalten: Leder, Holz, dazu Knochen und die überall erstaunlich häufigen Austernschalen. Die Keramikformen und die Stempel der Terra- Sigillata- Gefäße datieren die Funde in das zweite Drittel des ersten Jahrhunderts n. Chr.

Die schwarzen, schluffhaltigen Erdschichten voller Artefakte und organischer Reste reichten bis zu einer Höhe von 44m NN; d. h. sie können nicht im Wasser des Flussarmes gelegen haben, sondern weit darüber, wenn der Rheinpegel in der Antike nicht wesentlich höher lag als heute. Es gibt gute Gründe für die Annahme annähernd gleicher Wasserstände damals wie heute.1 Es hat den Anschein, dass Material aus dem Rheinarm aufgefüllt wurde, welches anfiel, als man versuchte, die Fahrrinne freizuhalten.2

Ähnlich begrenzte Einsicht in den Boden bot 1978 der Bau einer Abwasserleitung östlich des Domes (Abb. 1: b), bei dem neben zahlreichen Keramikfragmenten auch Lederund Textilreste, Fibeln und Münzen geborgen wurden, die den Beginn der Zuschüttung in die flavische Zeit datieren (Schleiermacher 1982: 211f.).

Der Raum zwischen dem Umspannwerk und der Stadtmauer wurde nur punktförmig untersucht, weil hier das damals gebaute Museum über einer Autodurchfahrt auf Betonpfählen steht. Der Aushub aus den brunnenartigen Gruben der Pfähle konnte naturgemäß bezüglich seiner Höhenlage nur ungefähr bestimmt werden. Das Fundspektrum reichte auch hier von der Keramik bis zu organischen Resten, die sich in dem stets feuchten Erdreich

1980 wurde mit der Baugrube für das Wallraf-RichartzMuseum/Museum Ludwig und die Philharmonie (das Wallraf–Richartz–Museum wurde jüngst aus dem

1

Neu 1985; auch in Xanten hat sich der Rheinpegel seit der Antike nicht verändert: Knörzer, Leichtle, Meurers – Balke & Neidhöfer 1994: 102ff. 2 Diese Deutung des Befundes verdanke ich M. Trier

425

Limes XVIII

erhalten haben. Die aus den Bohrpfählen geborgenen Artefakte sind zahlenmäßig geringer als die vorher beschriebenen, aber in dieselbe Zeit - das späte 1. Jahrhundert n. Chr.- zu datieren.

Pinienzapfens zeigen. Das recht häufige Vorkommen von Wandputz, und - etwas weniger häufig – von Ziegelfragmenten, also Bauschutt, zeigt, dass es sich aber auch um Abbruchschutt von Häusern handelte. Die Freigabe des Geländes zur Müllablagerung kann nach den keramischen Funden erst im späteren zweiten Jahrhundert erfolgt sein.

In diese Schichten wurde die Baugrube für die Stadtmauer eingetieft (Abb. 3), bis zum Niveau 39.9 NN mit steiler Böschung, darunter senkrecht bis auf den gewachsenen Kies bei 37.45 NN. Sie war mit Pfählen und Brettern verbaut. Fundament und Mauer zusammen waren hier noch über eine Höhe von bis zu 9m erhalten. An der Ostseite der Mauer stand eine Holzkonstruktion, die als Kaianlage zu deuten ist. Durch dendrochronologische Untersuchungen ist sie in die neunziger - Jahre des ersten Jahrhunderts datiert (Zerlach 1990: 64), wodurch auch der Bau der offensichtlich gleichzeitig errichteten Stadtmauer zeitlich eingeordnet ist.

Im mittleren Bereich der Ostseite der Stadt liegen keine Aufschlüsse vor, die eine Aussage über die dortigen Schichtverhältnisse erlauben. Der jüngst im Zuge der Errichtung einer Tiefgarage ausgegrabene Heumarkt (Athen et al 1998: 481ff.) liegt bereits auf dem Rücken der Rheininsel, gibt also für unseren Bereich der Westseite des Flussarmes keine Informationen. Südlich des decumanus maximus zeigte eine Ausgrabung an der Augustinerstraße (Abb. 1: f) ganz ähnliches Material aus dem ersten Jahrhundert wie das bisher Dargestellte (Thomas 1983: 252; vgl. auch Süßenbach 1984: 315).

Dass auf dem Rheinarm östlich der Stadtmauer Schiffahrt betrieben wurde, geht aus den dort gefundenen Schiffsresten hervor. Die Planke eines Rundspantschiffes, die an der Ostseite der Stadtmauer (Abb. 1: d) gefunden wurde, muss dort spätestens in den neunziger Jahren des ersten Jahrhunderts n. Chr. als Wrackteil gesunken sein, da die Pfahlreihe der Kaianlage durch diese Planke hindurch getrieben wurde. Dazu passen die keramischen Reste, die die Planke umgeben.

Genau östlich des Kapitoltempels, also im Südbereich der Stadtmauer (Abb. 1: g), bot sich ein ähnliches Bild wie im Bereich östlich des Domes: Hier wurde 1990 in der Mitte des ehemaligen Rheinarmes ein Wohnhaus ohne Keller auf Betonpfählen errichtet und der Aushub archäologisch untersucht (Knörzer & Neu 1998). Es standen mächtige schlammige Schichten, durchsetzt mit Haus - und Gewerbemüll aus der Stadt an. K.-H. Knörzer konnte anhand der pflanzlichen Makroreste nachweisen, dass es sich nicht um angeschwemmtes Material handelte, sondern um solches, das aus der Stadt hierher gebracht wurde. Die Schichten sind anhand zahlreicher TS - Stempel, aber auch durch die Keramik - Formen insgesamt in die zweite Hälfte des ersten Jahrhunderts n. Chr. zu datieren. Zwar ist hier der Rheinarm nur in seinem mittleren Teil untersucht, die Tatsache aber, dass an dieser Stelle 5m mächtige Hausmüllschichten anstehen, legt die Annahme nahe, dass er hier insgesamt verfüllt war, es sei denn, eine schmale Fahrrinne wurde künstlich offen gehalten.

Die anderen, etwa 40m weiter östlich (Abb. 1: e) in demselben Flußarm entdeckten Schiffsreste stammen wohl von einer Werft; es wurde offensichtlich an den Planken gearbeitet.3 Die Keramik aus den die Schiffsteile umgebenden Schichten (Abb. 4) ist überwiegend ins spätere erste und die erste Hälfte des zweiten Jahrhunderts, z. T. auch in das spätere zweite oder dritte Jahrh. zu datieren. Es muss aber noch eine Fahrrinne bestanden haben; sonst wäre der Schiffsbau an dieser Stelle nicht zu erklären. Aus den beobachteten Erdprofilen ist der Ort dieser Rinne jedoch nicht erkennbar. Auch oberhalb der Schiffsplanken fand sich in den Schwemmschichten Keramik des zweiten bis dritten Jahrhunderts, d. h. die für das erste Jahrhundert beobachtete Zusandung des Rheinarmes setzte sich fort. Deutlich ist im Profil die Grenze zwischen angeschwemmtem und aufgefülltem Erdreich zu erkennen. Danach reichten die höchsten Sedimentschichten bis c.41 NN, was auch eine Aussage über den Hochwasserstand in römischer Zeit ermöglicht, vorausgesetzt, dass höhere Schwemmschichten nicht durch spätere Erdbewegungen zerstört wurden. In den oberen Rand der sedimentierten Schichten greift eine flache Grube ein, deren Inhalt nicht angeschwemmt, sondern von Menschen aufgefüllt wurde. Die Verfüllung bestand zum Teil aus Hausmüll, wie die zahlreichen Fragmente von Gebrauchskeramik und eine Austernschale, aber auch seltenere Gegenstände wie ein bronzenes Baulot mit Inschrift, ein beinerner Spinnrocken, eine beinerne Haarnadel mit einen Kopf in der Form eines 3

An der Südostecke der Stadt (Abb. 1: h) wurden 1964 – 67 bei der Ausgrabung des Ubiermonuments (Abb. 5) ebenfalls große Mengen Keramik und sonstigen Materials, darunter Lederabfälle, Holz, Knochen und wiederum Austernschalen, also das bereits bekannte Materialspektrum, geborgen. Die zwischen dem Fundament des Bauwerks (um 40 NN) und dem Niveau 41.80 NN gefundene Keramik (sowie eine Münze) ist in die flavische Zeit zu datieren und entspricht damit den insgesamt längs der Ostseite der Stadt zu beobachtenden Schichten. Es gibt Anhaltspunkte dafür, dass auch hier die Schichten des ersten Jahrhunderts mächtiger waren; beim Abbruch eines Abwasserkanals im 3. Jahrhundert wurden die oberen Teile der älteren Schichten zerstört oder mit jüngerem Material vermischt (Bracker–Wester 1980: 508ff.). Das Ubiermonument selbst, dendrochronologisch 5/ 4 n. Chr. datiert (Hollstein 1980: 7f.), der älteste Steinbau Kölns und – nach dem Eichelstein in Mainz – der

Diese Deutung des Befundes verdanke ich D. Ellmers.

426

Stefan Neu und Matthias Riedel: Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr. zweitälteste Deutschlands, wurde unterschiedlichen Deutungen unterzogen (Bracker 1974; Neu 1997). Jüngst wurde er als Eckturm einer Holz - Erde – Mauer bezeichnet und weitere derartige Türme an den anderen Ecken des der Colonia vorausgehenden oppidum angenommen (Schütte & Gechter 2000: 178ff.). Gegen diese Deutung sprechen m. E. gewichtige Gründe: 1. Es gibt keine Parallele für derart monumentale steinerne Ecktürme von Holz – Erde – Mauern. Die an der Nordund Westseite des Bauwerks vorhandenen kleinen Eintiefungen, in die nach dieser Deutung die Balken einer derartigen Mauer eingelassen wären, können allenfalls nachträglich ausgemeißelt worden sein. Eine dendrochronologisch in das Jahr 24 (Hollstein 1980: 7f.) oder 9? (Leuschner & Leuschner 2000: 185) datierte Holzkonstruktion, die nördlich des Ubiermonuments ausgegraben wurde, könnte ein Rest dieser nachträglich angebauten Mauer sein. Als Eckturm einer steinernen Stadtmauer kann das Bauwerk nicht errichtet worden sein, da die Maueranschlüsse fehlen (die scheinbar nach W anschließende „Molenmauer“, auf der die Stadtmauer steht, gehört nicht zum ursprünglichen Gebäude, sondern ist später gegen dieses gesetzt (Neu 1997: 141). 2. Der für das – unvollendete Gebäude vorgesehene architektonische Schmuck (karniesverziertes Sockelprofil, zwei Säulengeschosse oder mindestens eines) ist für einen Stadtmauerturm ohne Parallele. 3. Der Umriss des anhand dieses Turmes rekonstruierten oppidum ist regelmäßiger als der aller zeitlich wie räumlich vergleichbaren Städte. Solche „Reißbrettstädte“ entstanden in spätrepublikanischer Zeit vornehmlich in Oberitalien (Lorenz 1987: 116ff .), nicht jedoch im 1. Jahrh. n. Chr. in den NO – Provinzen. Außerdem klettern die vermuteten Stadtmauern dieser rechteckigen Stadt in spitzem Winkel die Niederterrassenkante, die das Stadtplateau bildet, hinauf, was konstruktiv nicht einfach ist.

befremdet, aber nicht einzig dasteht: auch in Xanten wird der Graben an der Ostseite der Stadtmauer von Anfang an als Abfallgrube größten Ausmaßes benutzt (Petrikovits 1952: 126). Diese wenig attraktive Situation stellt auch die Ansicht, die Ostseite des römischen Köln sei ein Schaufenster der überlegenen römischen Zivilisation für die Germanen der anderen Rheinseite gewesen (Doppelfeld 1975: 729; Süßenbach 1984: 307) in Frage, zumal diese – zumindest im ersten Jahrhundert wahrscheinlich nur selten bis an das gegenüberliegende Flussufer kamen, da auch dort römische Truppen den niedergermanischen limes als „transrhenani“ schützten und wirtschaftlich nutzten (Kunow 1987). Die keramischen Funde Aus der Fülle des Fundmaterials sind hier einige charakteristische Objekte ausgewählt. Es handelt sich dabei im Wesentlichen um Terra sigillata, die durch Töpferstempel genau datierbar ist. Aber auch die feinere Ware der anderen Gattungen, also die Glanztonware und die glattwandig-tongrundige Ware wurde in exemplarischen Fragmenten zur Datierung der zur Diskussion stehenden Schichten mit herangezogen. Bei der Ausgrabung Otto Doppelfelds im Bereich „Alter Markt“ (Doppelfeld 1953: 110f.; Abb. 1: a) kam vor allem Terra sigillata mit dem Stempel des BASSVS vor. Dieser hat in La Graufesenque in claudisch-vespasianischer Zeit gearbeitet. Die Reliefsigillata des älteren Typs Drag. 29 und des jüngeren Typs Drag. 37 gehört ebenfalls in die Zeit kurz nach der Mitte des ersten Jahrhunderts. Auch aus dieser Zeit stammen die Gefäße seiner Kollegen CRESTIO, SECVNDVS und AQVITANVS. Die Glanztonkeramik aus derselben Fundstelle unterstreicht die eng geschlossene Datierung des Fundmaterials: Näpfe des Typus Hofheim 22 und Becher Hofheim 25, die wie das übrige Material in das 2. und 3. Drittel des ersten Jahrhunderts datieren. Auch die einfache, glattwandig– tongrundige Keramik ohne Überzug unterstreicht diese chronologische Hauptrichtung: es sind vor allem Einhenkelkrüge mit glattem Dreiecksrand und Zweihenkelkrüge mit abgeplatteter Wulstrandlippe.

Zusammenfassend ist festzustellen, dass der Rheinarm im ersten Jahrhundert durch Anschwemmungen des Flusses zusandete und mit Müll aus der Stadt systematisch verfüllt wurde. Die Fülle der Befunde macht es wahrscheinlich, dass er gegen Ende des ersten Jahrhunderts im Wesentlichen verlandet war. Bemerkenswert und nicht leicht zu verstehen ist die ungefähre Gleichzeitigkeit der mächtigen Verfüllschichten und der Bau der als Kaianlage gedeuteten Holzstellung an der Ostseite der Stadtmauer. Im Ostbereich des Rheinarmes wurde in den Schwemmschichten auch spätere Keramik gefunden; hier hat man vielleicht noch eine Fahrrinne offen gehalten.

Die Funde (Abb. 6) aus dem Bereich des Umspannwerks (Abb. 1: c) fügen sich mit der Terrra sigillata durch die Stempel des MONTANVS, BASSVS, AQUITANVS, VIRILIS, SECVNDVS und PRIMVS in den von Doppelfeld beobachteten Zeitrahmen des späten ersten Jahrhunderts n. Chr. ein. Dazu passt das Fragment eines marmorierten Terra sigillata – Tellers; auch der Stempel des ATEIVS aus dem frühen 1. Jahrhundert fällt nicht aus dem Rahmen. Die übrige Keramik fügt sich ebenfalls gut in das zeitliche Umfeld der Terra sigillata. Zum Ältesten zählt das Fragment einer Bildlampe und das eines Halterner Kochtopfes, beide noch in die erste Hälfte des 1. Jahrhunderts zu datieren. In die zweite Hälfte des 1. Jahrhunderts gehört das übrige Material wie die Glanztonware mit den bekannten Dekoren des späten 1. Jahrhunderts; das Gleiche gilt für die Profile.

Bisher wurden keine Gebäudereste über den Füllschichten des Rheins an der Ostseite der Stadt beobachtet. Schütte rekonstruiert auf dem Gelände des ehemaligen Rheinarmes und der Insel ein rechtwinkliges Netz basaltgepflasterter Straßen (Schütte 1999: 20). Wenn es wirklich existierte, war es wohl – wie die „Hafenstraße“ (Precht 1971: 54f.) spätantik. Aufgrund des Fehlens römischer Gebäude in dem Bereich des Rheinarmes ist anzunehmen, dass zwischen der Inselbebauung und der östlichen Stadtmauer lange Zeit eine offene Deponie lag, ein Befund, der zwar 427

Limes XVIII

Die Funde aus dem Bereich der Kaianlage vor der Stadtmauer und der westlichen Schiffsplanke (Abb.1: d) brachten unter anderem zwei Terra sigillata–Scherben, die mit den Stempeln des AMABILIS und der Sozietät FLAVIVS und GERMANVS wieder auf zwei Töpfer des späten ersten Jahrhunderts aus La Graufesenque hinweisen. Auch die zwei Krughälse fallen in dieselbe Zeit.

(Hrsg.) Stadtspuren. Bd. 26: 181-188. Lorenz Th. 1987 Römische Städte. (Darmstadt). Lückger H.–J. 1936 Agrippas Rheinhafen. (Bonn). Neu St. 1985 The Rhine frontage of Roman Cologne. In A.E. Herteig (ed.) Conference on waterfront archaeology in north European towns. Bergen 1983: 151-156. Neu St. 1989 Römische Reliefs von Kölner Rheinufer. Kölner Jahrbuch. 22: 241-364. Neu St. 1997 Zur Funktion des Kölner „Ubiermonuments“. Thetis 4: 135-145. Oswald F. 1931 Index of potter’s stamps on terra sigillata, samian ware. (Privately published) Petrikovits H. v. 1952 Kölner Untersuchungen (Rez.). Germania 30: 125-131. Precht G. 1971 Die Ausgrabungen um den Kölner Dom Vorbericht über die Untersuchungen 1969/70: Kölner Jahrbuch 12: 52-64. Schleiermacher M. 1982 Römische Leder– und Textilfunde aus Köln. Archäologisches Korrespondenzblatt 12: 205214. Schütte S. 1999 Stephan Broelmann und die Folgen. Kölner Museums – Bulletin, Heft 1: 4-26. Schütte Sv. & Gechter M. 2000 Ursprünge und Voraussetzungen des mittelalterlichen Rathauses und seiner Umgebung. In W. Geis & U. Krings (Hrsg.) Köln: Das gotische Rathaus und seine historiche Umgebung. (= Stadtspuren, Denkmäler Kölns Bd. 26):169-195. Süßenbach U. 1984 Römische und nachrömische Schichten an der rheinseitigen Stadtmauer der CCAA: eine Beobachtung in der Pipinstraße. Archäologisches Korrespondenzblatt 14: 307-317. Thomas R. 1983 Eine weitere Podiumskonstruktion in der südöstlichen CCAA. Archäologisches Korrespondenzblatt 13: 245-254. Zerlach Cl. 1990 Der römische Hafen von Köln – Untersuchungen zu den Holzfunden. (Magisterarbeit Köln - unveröffentlicht)

Die Funde (Abb. 7) aus dem Ostbereich des ehemaligen Rheinarmes (Abb. 1: e) sind unterschiedlich zu datieren: aus der Schicht mit den Schiffsplanken stammt wiederum Terra sigillata mit dem Stempel des BASSVS und dem gerillten Standring des Napfes Drag. 27 aus dem 3. Viertel des 1. Jahrhunderts, entsprechend auch die beiden Gefäßfragmente mit Goldglimmer–Überzug. Während der Einhenkelkrug noch in etwa dazu passt, sind die Hälse der Zweihenkelkrüge deutlich nicht vor dem zweiten Jahrhundert anzusetzen. Die Keramikfunde im südlichen Hafenbereich unterstreichen das bisher skizzierte Bild. Da sie zum Teil veröffentlicht sind (Bracker-Wester 1980: 508 ff.; Knörzer & Neu 1998: 451ff.), brauchen sie hier nicht erneut vorgestellt zu werden. Zusammenfassend ist also festzustellen, dass die in den Füllschichten des Rheinarmes an der Ostseite des römischen Köln gefundene Keramik mit wenigen jüngeren Ausnahmen in die zweite Hälfte des ersten Jahrhunderts zu datieren ist. Bibliographie Athen N., Frasheri Gj., Kempken Fr., & Merse M. 1998 Ausgrabungen auf dem Heumarkt in Köln. Kölner Jahrbuch. 31: 481-596. Bracker J. 1974 Neue Entdeckungen zur Topographie und frühen Geschichte des römischen Köln. Jahrbuch des Kölnischen Geschichtsvereins 45: 111-178. Bracker-Wester U. 1980 Das „Ubiermonument“ in Köln, Ein Bauwerk nach gallischem/ germanischem Maß. Gymnasium 87: 496-534. Doppelfeld O. 1953 Hafenfunde vom Alten Markt. Bonner Jahrbücher 153: 102-125. Doppelfeld O. 1975 Ubier – oppidum und Colonia Agrippinensium. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.4. (Berlin): 716782. Hollstein E. 1980 Mitteleuropäische Eichenchronologie. (Mainz). Knörzer K.H., Leichtle Th., Meurers-Balke J. & Neidhöfer R. 1994 Der römische Hafen von Xanten. Red. S. Kraus & A. Riechel Xantener Berichte: GrabungForschung-Präsentation; Sammleband 5: (Köln): 89107. Knörzer K.H. & Neu St. 1998 Archäologische Untersuchungen an der Plectrudengasse in Köln. Kölner Jahrbuch. 31: 445-480. Kunow J. 1987 Das Limesvorland der südlichen Germania inferior. Bonner Jahrbücher 187: 63-78. Leuschner H.H. & Leuschner B. 2000 Prüfung der von Hollstein (1966) vorgenommenen dendrochronologischen Datierungen „Köln, frührömischer Hafenturm oder Quaderbau“ sowie „Köln, Uferbefestigung“. In W. Geis & U. Krings

428

Stefan Neu und Matthias Riedel: Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr.

Fig. 1. Die Ostseite des römischen Köln.

429

Limes XVIII

Fig. 2. Schnitt durch das umspannwerk am Dom.

Fig. 3. Schnitt durch die römische Stadtmauer.

Fig. 4. Schwemm- und Füllschichten im östlichen Bereichs der Römischen Rheinarmes.

430

Stefan Neu und Matthias Riedel: Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr.

Fig. 5. Das Ubiermonument und die römische Stadtmauer von W (nach Bracker – Wester).

Fig. 6. Terra sigillata aus dem Bereich des Umspannwerks M. 1: 2 (Zeichnung S. Haase).

431

Limes XVIII

Fig. 7. Keramik vom Ostrand des Rheinarmes. M. 1 : 2 (zeichnung S. Haase).

432

Die spätaugusteische Siedlung in Lahnau-Waldgirmes - Zusammenfassende Bemerkungen zum Stand der Fundauswertung Gabriele Rasbach bestimmt werden.1 Vergleicht man die Münzreihen von Waldgirmes mit anderen augusteischen/tiberischen Fundplätzen, so fällt die Ähnlichkeit zu den Münzreihen von Haltern und von Kalkriese auf (Abb. 1). Sowohl in Kalkriese als auch in Waldgirmes überwiegen deutlich die Prägungen der 1. Altarserie aus Lugdunum (c.7-3 v. Chr.).2 Diese Münzen machen in Waldgirmes nahe 70% der Bronzemünzen aus, in Kalkriese sogar fast 90%.3

Die seit 1998 aufgedeckten Befunde, die Fundamente des Forums wie auch die Teiluntersuchungen an der nördlichen und südlichen Umwehrung, erbrachten nur sehr wenige Funde. Zumeist handelt es sich um einzelne Keramikscherben. Trotz der zivilen Interpretation des Platzes als Keimzelle einer Stadt, wofür die Befunde sprechen, ist wohl vorauszusetzen, daß in einer Phase, in welcher in einer eroberten Region, die zur Beherrschung notwendige Infrastruktur errichtet wurde, nur das römische Militär über die personellen und organisatorischen Voraussetzungen dafür verfügte. Nicht nur die Vermessungs- und Bautrupps sondern auch eine sicherlich noch notwendige Überwachung des Vorfeldes waren Aufgaben militärischer Einheiten, über deren Anzahl oder Herkunft und Namen jedoch die Ausgrabungen bisher keine Hinweise erbrachten. Die Anwesenheit von Militärs spiegelt sich scheinbar nur gering in den bisher geborgenen Funden. Stellt man jedoch die Funde von Werkzeugen (von Lot bis Zange) den Militaria gegenüber, zeigt sich mit einem Verhältnis von 2.5:1 die Anwesenheit von Militär deutlicher als es mit der Nennung der fünf, den Militaria zuzurechnenden Funde möglich wäre. Damit relativieren sich Aussagen wie: Militaria sind im Fundbestand von Waldgirmes völlig unterrepräsentiert.

Ähnlich vergleichbar ist die Menge der Gegenstempel auf diesen Münzen. In Waldgirmes gibt es zur Zeit 9 Münzen, die erkennbare Gegenstempel des Publius Quinctilius Varus tragen.4 Da sie nur während seiner Statthalterschaft in Germanien 7-9 n. Chr. ausgegeben worden sein können, sind diese Münzen (weiterhin) als die jüngsten nach Waldgirmes gelangten Prägungen anzusehen.5 Prägungen des Jahres 10, etwa die 2. Altarserie aus Lugdunum, fehlen bisher. Da alle gegengestempelten Bronzemünzen aus einer Schicht stammen, die der Zerstörung der Anlage zuzuweisen ist, läßt sich das Ende der römischen Besiedlung in Waldgirmes mit der Niederlage der Römer in der sogenannten „Schlacht im Teutoburger Wald“, 9 n. Chr., und dem nachfolgenden Rückzug der Römer über den Rhein verbinden. Anhaltspunkte für den Beginn der Anlage sind aus der Münzreihe, als der chronologisch sensibelsten Fundgattung, vergleichsweise nur sehr schwer zu gewinnen. Betrachtet man die Mengenverhältnisse von Nemausus-Assen (3 Ex.) und Aduatuker-Kleinbronzen (12 Ex; van Heesch 1999: 347-361) einerseits und der 1. Lyoner Altarserie (92 Ex.) andererseits, ist die Gründung der Siedlung nach dem Beginn von Haltern anzusetzen.

Mit einer noch 21.5cms lang erhaltenen Speerspitze und einer Pilumspitze ist der momentane Fundbestand an Angriffswaffen beschrieben. Die Speerspitze wird, wie bereits J.-K. Haalebos (1994) und E. Deschler-Erb (1999: 1) hervorhoben, schon aufgrund der Größe eher als Teil einer Wurfwaffe anzusehen sein (Unz & Deschler-Erb 1997: ähnlich Nr. 254). Des weiteren wurde ein eisernes Riemenscharnier eines Schienenpanzers geborgen (Vgl. Deschler-Erb 1999: Taf. 14, 210-211). Den Schienenpanzern ist außerdem noch eine kleine Eisenschnalle zuzurechnen. Das Fragment eines gestuft profilierten und gebogenen, massiven flachen Eisenteils wurde von A. Abegg-Wigg mit Vorsicht als NackenschutzFragment eines Helmes angesprochen (Wigg 1996: 147ff. v.a. 157). Dazu ist dieses Stück jedoch zu massiv; eine anderweitige Deutung steht jedoch zur Zeit noch aus.

Momentan gehen wir davon aus, daß die römische Besiedlung nur etwa 10 vielleicht sogar weniger Jahre dauerte. Anzeichen für eine über das Jahr 9 hinausreichende Besiedlung oder eine erneute römische Besetzung des Platzes sind derzeit nicht zu erkennen. Terra sigillata Diese kurze Besiedlungsdauer spiegelt sich auch in der für eine Siedlungsgrabung geringen Menge an Keramik, die aus vergleichsweise wenigen Befunden geborgen wurde, worunter wieder der Ost-West-verlaufende Straßengraben dominiert. Darunter sind natürlich die interessantesten

Andere Ausrüstungsgegenstände fehlen, sieht man von Eisenfragmenten ab, die möglicherweise als Teile von Zeltheringen anzusprechen sind. Münzen

1

Die Bestimmung der Münzen verdanken wir D. G. Wigg (Frankfurt am Main). 2 Zur Datierung zuletzt: van Heesch 1993: 535ff. 3 Die Tabellen wurden übernommen aus: Wigg 1999: 327-346. 4 Zu dem Gegenstempel VAR (in Ligatur): Berger 1996: 51f., Abb. 27. 5 An identifizierten Gegenstempeln sind zu verzeichnen: QV, VAR, Rad.

Die Münzreihe von Waldgirmes umfaßt zur Zeit 226 Stücke. Wegen der schlechten Erhaltungsbedingungen für Metallfunde konnten jedoch nur 154 Münzen näher

433

Limes XVIII

Warenarten die Terra Sigillata und die handgemachte germanische Keramik.

solche Standorte im Beginn des 1. Jahrhunderts ebenfalls die Regel gewesen sein, könnten diese Siedlungsstellen heute unter Auelehm begraben oder durch den modernen Kiesabbau verloren sein.

Unter den momentan vorhandenen Terra SigillataFragmenten, es handelt sich um Reste von etwa 100 Gefäßen6, befindet sich zur Zeit kein einziges Stück reliefverzierter Ware, jedoch immerhin 21 Stempel bzw. Stempelfragmente. 14 der 17 bestimmbaren Stempel gehören zu italischen Töpfern, deren Waren größtenteils im Bereich nördlich der Alpen sehr selten sind, auch in Haltern und anderen augusteischen Plätzen fehlen. Als letzte Stempel hinzugekommen sind ein LVCILIUS, dessen Hauptabsatzgebiet in Etrurien und Spanien lag, ein MAHES+ZOELVS und ein AVILIVS MANIVS (Oxé & Comfort 1968, Stempel 893 c, 183 k und 287 h.). Wie nicht anders zu erwarten, fehlen Teile des Services Ia ebenso wie Radialstempel, es dominieren Formen des Services II deutlich gegenüber Formen des Services Ic.

1999 konnten drei sichere germanische Brandgräber im Inneren der römischen Siedlung nachgewiesen werden. Diese Bestattungen, Urnen und einfache Brandschüttungen, lagen im Innenbereich der südöstlichen Ecke der Anlage. Nach Ausweis der germanischen Keramik datieren diese Gräber etwa in die Mitte des 1. Jahrhunderts n. Chr. Dies wäre ein Hinweis auf das mögliche Weiterbestehen einer germanischen Siedlung im Umfeld. Fibeln Im Fundmaterial sind zur Zeit 29 Fibeln bzw. Fibelfragmente zu verzeichnen. Der überwiegende Teil der Fibeln besteht aus Bronze, nur wenige sind aus Eisen. Die aggressiven Bodenbedingungen greifen vor allem die Oberflächen an und zerstören die dünnen bzw. filigranen Teile der Bronzefibeln, wodurch oft nur eine grobe Typenansprachen möglich ist. 21 Fibeln waren bestimmbar und wurden in die von Th. Völling angelegten Tabelle ergänzend eingetragen (Völling 1994: 147ff., v.a. Tab. 18). Die Aufstellung zeigt, daß das Formenspektrum trotz der kurzen Besiedlungszeit von Waldgirmes gut mit den länger belegten augusteischen Fundplätzen zu vergleichen ist.

Handgemachte Keramik7 Besonders auffallend und weiterhin ohne Vergleich in augusteischen Anlagen ist ein konstant hoher Anteil von etwa 20% an handgemachter germanischer Keramik im Fundgut. Es handelt sich überwiegend um grob gemagerte Schalen und einfache Kumpf-Formen mit senkrecht stehendem, teils verdicktem Rand, die oft mit Schlickerbewurf oder Kammstrich versehen sind (Abb. 2). Seltener sind feinere Scherben mit abgestrichenem Rand vergleichbar der Form Uslar 1. An den bisher im Fundgut vertretenen Formen lassen sich bereits verschiedene Einflüsse erkennen: keltische (Abb. 2. 13), rheinwesergermanische (Abb. 2. 9-11) sowie elbgermanische (Abb. 2. 1-3).

Auf der Ostseite der Ansiedlung, im Bereich des Tores, hingegen gibt es eine Schicht in der Verfüllung des äußeren Umwehrungsgrabens, die stark mit Holzkohle, Hüttenlehm und Funden versetzt ist. Aus dieser Schicht, aus der u.a. auch der Teil einer Bleiwasserleitung geborgen wurde, kam eine sehr gut erhaltene eingliedrige Drahtfibel zutage („Soldatenfibel“: Riha 1994: Typ 1.6.2.).

In Waldgirmes gehören die germanische und römische Keramik nicht zu unterschiedlichen Schichten der Grubenverfüllungen, womit eine chronologische Trennung von römischer und germanischer Besiedlung nicht gegeben ist. Im überwiegenden Anteil der Gruben, kommen handgemachte germanische und römische Keramik vermischt vor. So gibt es unter den seit 1996 aufgedeckten 119 Befunden mit Keramik, nur 35 aus denen allein römische Drehscheibenware zutage kam. Zu diesen Befunden ohne germanische Keramik zählen beispielsweise die 5 Gruben im Innenhof des Forums.

Handwerk Handwerkliche Tätigkeiten stellen sich vor allem in Einzelfunden von Werkzeugen selten in diesbezüglichen Befundzusammenhängen dar. Zwar streuen über das gesamte bisherige Ausgrabungsareal Eisenschlacken, aber bisher konnte weder ein sicher ansprechbarer Schmiedebereich noch eine Verhüttungsanlage nachgewiesen werden.8 Weitere Hinweise auf das Schmieden bzw. die Metallverarbeitung vor Ort, was beim Bau einer solchen Anlage nicht verwundert, sind eine Metallschere (Abb. 3.1), Gußtropfen und eine Menge von Schlacken (Schmiedeschlacken wie Luppen) ebenso wie eine große Menge Nägel.

Trotz intensiver Prospektionen (v.a. Begehungen, in Teilen auch geophysikalische Prospektionen) fehlen bisher Spuren einer germanischen, zur römischen Anlage zeitgleichen Siedlung. Wie Untersuchungen der letzten Jahre im Lahntal ergaben, liegen die germanischen Siedlungen des 2. und 3. Jahrhunderts nahe der Lahnaue (Abegg-Wigg, Biegert & Walter 2000: 55ff.). Sollten

Die Herstellung von Keramik in Waldgirmes ist durch zwei Töpferöfen belegt. Jedoch nur einem der Öfen konnte die zugehörige Brennware zugeordnet werden (Walter & Wigg 1997: 285ff.), der zweite Ofen war im Rahmen eines

6

Hierbei handelt es sich um eine Schätzung auf Grundlage einer Zählung der Randscherben. Die schlechte Erhaltung, die Scherben sind durchwegs sehr weich und die Brüche mehlig, erschwert die Suche nach anpassenden Scherben aus verschiedenen Befunden erheblich. 7 Die handgemachte Keramik und Tafel wurden von Frau Dr. Dörte Walter, Frankfurt, bearbeitet.

8

Naturwissenschaftliche Untersuchungen der Schlacken werden durch Herrn Prof. Dr. Raub, Schwäbisch-Gmünd, durchgeführt.

434

Distelfibel Aucissafibel Drahtfibel mit unterer Sehne Langton-Down-Fibel Almgren 241 Almgren 2a Omegafibel Almgren 237a Almgren 22a Almgren 19a Almgren 18a Almgren 2b gegitterte Scharnierfibel Schüsselfibel Var. Nijmegen gallische Flügelfibel Kragenfibel Typ GoeblingenAugenfibel Typ Haltern Augenfibel Almgren 45b / 47 Hülsenspiralfibel Almgren 67 Almgren 236 c

Gabriele Rasbach: Die spätaugusteische Siedlung in Lahnau-Waldgirmes

Basel-Münster (taq 15/11 v. Chr.) Dangstetten (15/12-9/7 v. Chr.) Rödgen (11-c.8 v. Chr.) Oberaden (11-7 v. Chr.) Beckinghausen (11-7 v. Chr. ?) Bad Nauheim (taq 9 n. Chr.) Waldgirmes (taq 9 n. Chr.) Haltern (taq 9 n. Chr.) Markbreit (taq 9 n. Chr.) Anreppen (taq 9 n. Chr.) Kalkriese (taq 9 n. Chr.) Bentumersiel („frühtiberisch“) Augsburg-Oberhausen (taq 15/16 n. Chr) Friedberg-Rederzhausen („spätaugusteisch-mitteltiberisch“)

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ? ●

●

●

● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ●

● ● ● ●

● ● ● ● ● ● ● ● ● ● ●

Tabelle nach Völling 1994: Bestand Waldgirmes ergänzt. Umbaus planiert worden. Bei der örtlich hergestellten Keramik handelt sich um eine in der Formgebung der „Belgischen Ware“ nahestehende Keramik, die sich aber allein vom Aussehen des Scherben durch eine grobe Schamottemagerung deutlich unterscheidet.

(Abb. 3,2), daß zum Schneiden des Leders diente.11 Diese Messer sind vor allem aus den frühen Militärlagern bekannt, was wohl mit der fehlenden Infrastruktur und der damit notwendigen Versorgung der Truppen vor Ort zu erklären ist.

Nach der chemischen Analyse der Ofenwände (1996) wurde auf Grundlage dieser Referenzgruppe begonnen, verschiedene keramische Warenarten entsprechend zu untersuchen.9 Dadurch konnten weitere in der Formgebung eng mit der „Belgischen Ware“ verwandte Keramiken als lokale Produktionen bestimmt werden. Wir werden in den kommenden Monaten versuchen weitere Produktionsorte gerade der sogenannten „Belgischen Ware“ nach Rücksprache mit den zuständigen Kollegen in diese Untersuchungen einzubeziehen.10 Wir erwarten durch diese Untersuchungen Hinweise auf die Belieferungswege zu gewinnen.

Eine Überraschung ergab sich bei der Restaurierung eines stark korrodierten Eisenfundes. Bei der Entfernung der dicken Korrosionsschicht wurde das etwa 10cms lange und 3cms breite Bruchstück einer Messerscheide sichtbar, auf der noch Reste einer ehemals roten Emailschicht erhalten waren.12 Beim Ausnehmen der Umwehrungsgräben im Bereich des Osttores wurde ein Bruchstück eines Bleirohres gefunden. Für dieses sowie für den größten Teil der übrigen Bleifunde, zumeist verschmolzene Bleiklumpen, ergaben Blei-Isotopen-Untersuchungen, daß das Metall aus den größten europäischen Bleilagerstätten um Mechernich in der Eifel stammte. Vergleichsuntersuchungen an römischen Schleuderbleien, die bei dem nahe Waldgirmes gelegenen keltischen Oppidum auf dem Dünsberg gefunden wurden (Schlott 1999: 46ff.), lassen für diese auf

Zu den Werkzeugen eines Schusters oder Sattlers gehörte das kleine lanzettförmige Messer mit tordiertem Griff

9

Die Proben werden von Herrn Dr. G. Schneider, Berlin, und Frau Dr. S. Biegert, Frankfurt, bearbeitet. 10 Unser Dank für die Bereitstellung von Scherben zur chemischen Untersuchung gilt X. Deru. In Zukunft soll nun versucht werden, diese Datenbank durch weitere Untersuchungen zu ergänzen.

11 Vgl. Harnecker 1997: Nr. 245-256; Dolenz 1998: 213-218. Erstmals und grundlegend Gansser-Burckhardt 1942: S. 18-19. 12 Zu diesen Messerscheiden zuletzt: Obmann 2000: 4ff.

435

Limes XVIII

eine Herkunft schließen.13

aus

Bleilagerstätten

im

Alpenraum

Bis jetzt sind in Waldgirmes leider keine sicher identifizierbaren Fragmente geborgen worden, die dem Reiter zuzuweisen sind. Aber in einer Zeit als die Römer Verwaltungsstrukturen in Germanien zu errichten begannen, kann es sich bei dem Reiter eigentlich nur um ein Bild des Augustus gehandelt haben. Die Statue stand sicherlich im Inneren des Forums, dem politischen Zentrum. Als weiteres Indiz dafür können ArchitekturFragmente gelten, die aus den fünf Gruben im Innenhof geborgen wurden, und zu mehreren Postameute gehörten.

Glasgemme Bei einem Schnitt durch den Ost-West-verlaufenden Wassergraben wurde 1998 eine Glasgemme gefunden (Abb. 4). Die 1.7cms große Glaspaste zeigt auf der Vorderseite das Bild einer Niobide, die einen ihrer toten Brüder im Arm hält. Die Darstellung entspricht der auf einem Relieffries aus dem 5. Jahrhundert v. Chr. aus der ehemaligen Sammlung Campana, heute in Sankt Petersburg. Glasgemmen dieser Art hatten im 1. Jahrhundert v. Chr. ihre größte Verbreitung, ab dem 2. Viertel des 1. Jahrhunderts n. Chr. werden diese gestreiften Glaspasten immer seltener.

Bei diesen Stücken handelt es sich um fein profilierte Architekturfragmente und Quader, die aus einem Muschelkalk bestehen, der im Gebiet zwischen Lahn, Rhein und Main nicht ansteht. Zur Zeit werden diese Stücke von Herrn Dr. Thomas Keller, Landesamt für archäologische und paläontologische Denkmalpflege Hessen, petrographisch untersucht und es deutet sich an, daß das Vorkommen dieses Kalksteins im Lothringer Becken zu suchen ist.

Der exzellente Erhaltungszustand des Stückes erlaubt es, einige interessante Details des Herstellungsprozesses zu erläutern.14 Die Stücke wurden in Formen hergestellt, wodurch mit einem geringeren künstlerischen Aufwand größere Mengen produziert werden konnten.15 Die Stücke wurden von der Vorderseite her aufgebaut, was an der Glasgemme gut an den Seitenflächen nachzuvollziehen ist. Zuerst kamen die zähflüssigen weißen Glasstreifen in die Form, danach wurde die Form mit hellblauem durchsichtigem Glas ausgefüllt. In das obere und das untere Drittel brachte der Handwerker eine gelbe in das mittlere Drittel eine weiße Glasschicht ein. Sowohl das Gelb wie das Weiß sind nur transluzente Glasschichten. Den Abschluß bildete eine über die ganze Rückseite aufgebrachte Schicht aus hellblauem Glas. Durch diese verschiedenen Farbschichten, die ein großes handwerkliches Geschick bei der Herstellung voraussetzen, erscheint die Gemme grün und dunkelblau.

In Waldgirmes sind nicht nur durch eine weitgehend unbebaute Fläche ideale Bedingungen für eine archäologische Untersuchung gegeben, sondern die bisher einmaligen Befunde liefern einen wichtigen Beitrag zur Antwort auf die Fragen: Wie weit die Römer Germanien unter Kontrolle hatten und wie weit die Einrichtung einer Provinz gediehen war. Die herausragende Fundzusammensetzung von dem hohen Anteil germanischer Keramik, der Existenz einer Wasserleitung aus Bleirohren bis hin zur Aufstellung einer vergoldeten Bronzestatue des Kaisers unterstreicht den Stellenwert dieser römischen Stadtgründung in Germanien für die historische Forschung. Bibliography

Reiterstatue Abegg-Wigg A., Biegert S. & Walter D. 2000 Forschungen in germanischen Siedlungen des mittleren Lahntales. KVF 5 (Bonn): 55-65. Bergemann J. 1985 Die Pferde von San Marco. Röm. Mitt. 95: 115–128. Berger F. 1996 Kalkriese 1. Die römischen Fundmünzen. Römisch-Germanische Forschungen 55. (Mainz). Deschler-Erb E. 1991 Das frühkaiserzeitliche Militärlager in der Kaiseraugster Unterstadt. Forschungen Augst. 12. Deschler-Erb E. 1999 Ad arma! Forschungen Augst 28. (Augst). Dolenz H. 1998 Eisenfunde aus der Stadt auf dem Magdalensberg. Kärntner Museumsschriften 75 (Klagenfurt): 213-218. Gansser-Burckhardt A. 1942 Das Leder und seine Verarbeitung im römischen Legionslager von Vindonissa. Veröffentlichungen der Gesellschaft Pro Vindonissa 1 (Basel). Haalebos J.-K. 1994 Bonner Jahrbuch 194: 3-705. Harnecker J. 1997 Katalog der Eisenfunde von Haltern aus den Grabungen der Jahre 1949-1994. Bodenalt. Westfalen 35. (Mainz). van Heesch J. 1993 Proposition d’une nouvelle datation des monnaies en bronze à l’autel de Lyon frappées sous Auguste. Bull. Soc. Française Num. 48.4: 535-538.

Während der Ausgrabungen der letzten Jahre kamen auch immer wieder Bruchstücke der vergoldeten, etwa lebensgroßen Bronzestatue zutage. Momentan sind es 87 Fragmente, wovon die meisten jedoch um einiges kleiner als ein Daumennagel sind. Das größte und für die Ansprache der Statue entscheidende Fragment ist ein 1.4 Kilo schweres, mit einer Ranke verzierte Bruchstück eines Balteus der Brustschirrung eines Pferdes (Abb. 5). Dieses Stück wurde bereits in Zalau gezeigt, die genaue Ansprache war jedoch damals noch nicht sicher. Als Vergleichsstücke hierfür lassen sich z. B. die Statuenfragmente aus Cartoceto (2. H. 1. Jh. v. Chr.) sowie die Pferde von San Marco in Venedig heranziehen (Pollini 1993: 423ff.; Bergemann 1985: 115ff.; Robotti 1977: 41ff.).

13

Die Blei-Isotopen-Untersuchungen verdanken wir Herrn Dr. Brauns, Institut für Geologie und Lithosphärenfoschung, Universität Gießen. 14 Zum Herstellungsprozeß dieser Gemmen s. Weiß 1999: 80ff. 15 Ein Vergleichsstück z.B. Maaskant-Kleibrink 1971: Nr. 74 (RCC Inv.Nr. 2022).

436

Gabriele Rasbach: Die spätaugusteische Siedlung in Lahnau-Waldgirmes van Heesch J. 1999 Augustan bronze coins in the north-west of Gaul. In W. Schlüter & R. Wiegels (Hrsg.) Rom, Germanien und die Ausgrabungen in Kalkriese. Osnabrücker Forsch. Altert. u. Antike-Rezeption 1. (Osnabrück): 347-361. Maaskant-Kleibrink B.M. 1971 Catalogue of the engraved gems in the Royal Coin Cabinet, The Hague. (Wiesbaden). Obmann J. 2000 Studien zu römischen Dolchscheiden des 1. Jahrhunderts n. Chr. Kölner Stud. Arch. Röm. Prov. 4. (Rahden). Oxé A. & Comfort H. 1968 Corpus vasorum arretinorum. Antiquitas 3. 4. (Bonn). Pollini J. 1993 The Cartoceto bronzes. Portraits of a Roman aristocratic family of the late first century B.C. American Journal of Archaeology 97: 423-446. Riha E. 1994 Die römischen Fibeln aus Augst und Kaiseraugst. Die Neufunde seit 1975. Forschungen Augst 18. (Augst). Robotti C. 1977 I cavalli della Basilica di San Marco a Venezia. Mus. e Gallerie d’Italie 22: 41-50. Schlott Ch. 1999 Zum Ende des spätlatènezeitlichen Oppidum auf dem Dünsberg (Gem.Biebertal-Fellingshausen, Kreis Gießen). Forschungen Dünsberg 2. (Montagnac). Unz Ch. & Deschler-Erb E. 1997 Katalog der Militaria aus Vindonissa. Veröffentlichungen der Gesellschaft Pro Vindonissa 14. (Brugg). Völling T. 1994 Studien zu Fibelformen der jüngeren vorrömischen Eisenzeit und der ältesten römischen Kaiserzeit. Berichte der Römisch-Germanischen Kommission 75: 147-282. Walter D. & Wigg A. 1997 Ein Töpferofen im augusteischen Militärlager Lahnau-Waldgirmes, LahnDill-Kreis. Germania 75.1: 285-297. Weiß C. 1999 Beobachtungen an Glaskameen der Sammlung Bergau. In R. Lierke Antike Glastöpferei. (Mainz): 80-82. Wigg A. 1996 Zum Stand der archäologischen Untersuchungen im frührömischen Militärlager von Lahnau-Waldgirmes am der mittleren Lahn. Mitt. Oberhess. Geschichtsverein NF 81: 147-176. Wigg D. G. 1999 Die Rolle des Militärs bei der Münzversorgung und Münzwirtschaft am Rhein in der frühen Kaiserzeit. In In W. Schlüter & R. Wiegels (Hrsg.) Rom, Germanien und die Ausgrabungen in Kalkriese. Internat. Kongr. Osnabrück 1996. Osnabrücker Forsch. Altert. U. Antike-Rezeption 1 (Osnabrück): 327-346.

437

Limes XVIII

Abb 1. Prozentuelle Anteile verschiedener Bronzeprägungen in ausgewählten augusteischen/tiberischen Fundkomplexen. Entnommen aus: D.G. Wigg Die Rolle des Militärs bei der Münzversorgung und Münzwirtschhaft am Rhein in der frühen Kaiserzeit. In Rom, Germanien ud die Ausgrabungen in Kalkriese. Internat. Kongr. Osnabrück 1996. Osnabrücker Forsch. Altert. U. Antike-Rezeption 1 (Osnabrück 1999): 327-346, Abb. 4.

Abb. 2. Auswahl handgemachter Keramik aus Lahnau-Waldgirmes (nach D. Walter).

438

Gabriele Rasbach: Die spätaugusteische Siedlung in Lahnau-Waldgirmes

Abb. 3. Auswahl von Handwerszeugen. 1 – Metallschere; 2 – Ledemesser.

Abb. 4. Glasgemme mit Niobidendarstellung. M. etwa 1: 4. Foto: J. Bahlo, RGK, Frankfurt a. M.

439

Limes XVIII

Abb. 5. Fragment der vergoldeten Reiterstatue aus Lahnau-Waldgirmes. Bruststück des Balteus (ohne Maßstab).

440

Recent developments in south-west Germany (eastern Germania Superior - western Raetia) C. Sebastian Sommer This paper presents an update on current issues in south-west Germany. It focuses on new discoveries, changes in interpretations and questions which have come up in the past few years about the development in the conquest of that part of the Roman empire which is now Baden-Württemberg. Naturally, the paper is based on the work of many colleagues and it derives from the results at various sites. The development of greatest general impact may be the opinion that the Odenwald-Neckar-Alb limes does not go back to the reign of Domitian but that it was installed in the later years of Trajan. Almost as important are questions arising from dendro-dates of the last years of Hadrian in a settlement on the right bank of the Neckar. There is also more evidence for settlements which were abandoned before AD260.

The following paper attempts to update M. Klee’s Zum Stand der Limesforschung in der Provinz Obergermanien (Klee 1999) and my article From conquered territory to Roman province: recent discoveries and debate on the Roman occupation of SW Germany (Sommer 1999a). I will focus on new discoveries and results, changes of interpretation and questions arising in the past few years about the development of the conquest of that part of the Roman empire which is now Baden-Württemberg. Of greatest impact may be the suggestion of a Trajanic date for the Odenwald-Neckar-Alb limes. Naturally, the paper is based on the work of many colleagues and originates from results from various sites.

legionary fortress of Dangstetten (153). Originally, this seems to have been supported by the discovery of a 7.5ha camp at Untereggingen (B). However, as neither the strategically important site of Schleitheim (154) reveals any signs of a military presence or even any pre-Flavian settlement (Gassmann & Trumm 1998: J. Trumm pers. comm.), nor does Hüfingen have any strong pre-Flavian evidence, this idea has to be questioned, at least for the Claudian-Neronian period. More speaks for the possibility that Hüfingen was at this time no more than a small station along the Danube, perhaps with the special function of the last garrison before crossing the Black Forest towards the Rhine, and, as I would like to suggest, may have belonged to Raetia.

As I intend a chronological approach let me begin with the late La Tène period. Despite extensive research by G. Wieland, there is still no sign of continuity between the latest Celts and the Romans beyond the Rhine valley with the possible exception of some contacts between a small group of people seeking refuge in more or less inaccessible areas, especially caves and abris, in the valley of the upper Danube, and the Augustan army (Fig. 1, zone A; Wieland 1998; 2000). Without doubt, the conquering army of the later C1st AD did not meet any more Celts. Not only are there no indications of archaeological continuity, but a disruption in the environmental development is suggested, too, through pollen-analysis. I am referring to new studies by H.W. Smettan (2000a; 2000b). Therefore, the conclusion of K.H. Lenz in an otherwise convincing article, that the villae rusticae developed from the so-called Viereckschanzen/Fermes indigènes does not apply to south-west Germany (Lenz 1998). Instead, his Raetian model has to be considered for our area. This proposes an import of the discussed form of settlements (villae rusticae) from Gaul (or, questionably, from the southern Rhine zone). Roman finds in some of our Viereckschanzen, the latest of which is a Roman pot on top of the final layer of destruction in the ditch of the second Schanze at Nordheim (Neth 1999: 79; generally Wieland & Luik 1999: 176, 262-267), do not disprove this statement; rather, they hint at a reuse of these still standing monuments in the C2nd AD.

With the final proof of the long suspected fort at Ennetach (156; Reim 1999) a continuous chain of garrisons along the upper Danube from Claudian times onwards is almost definite. The fort was situated on a promontory and had a multi-phase defensive system (Fig. 2). This is comparable to the situation at Hüfingen further up river and Rißtissen (159) and Unterkirchberg (160) further down the Danube. From the geophysical survey it is likely that there existed another, larger fort at the same spot. These multi-period aspects again are similar to the forts at Emerkingen (158) and Rißtissen (Klein 1999), where aerial reconaissance led some years ago to the discovery of underlying or adjoining installations. A system is becoming apparent for the choice of locations along this part of the Danube: The forts were erected close to a point of change from a rather narrow to an open part of the valley (Fig. 2). In general, in difference to later forts, these early installations were never erected on the line of the main road accompanying the Danube, but lay up to 2kms north or south of it. With this in mind, we are now concentrating again on two possible sites between Ennetach and Rißtissen in our search for a missing link which was suggested long ago by Ph. Filtzinger at Ertingen (Fig. 2 and 157 on Fig. 1). Further down the Danube at Günzburg (161), just outside the defined range of this study, a most recent work on the coinage and the stamped samian ware by J. Schmid (2000) has not, to date, discovered material for a Claudian fort. Only enough material for a (Neronian) fortlet has been exposed. However, as it is possible to reconstruct a regular

Turning to the middle of the C1st AD, a strong connection between Vindonissa (Fig. 1, 139) and the fort at Hüfingen (62a) at the upper Danube has always been assumed, due, for example, to the northward orientation of the earlier

441

Limes XVIII

continuous line of forts in the upper part of the Danube, this apparent gap in the sequence of forts further down at Günzburg seems highly unlikely to me.

In respect to military construction, I would like to draw attention to Ladenburg (108). Recently we were able to restore a large piece of painted wall-plaster from a hall-like room in the praetorium of Kastell I and to install it almost at its original location. Decorating a wattle-and-daub wall, it stood up to a height of approximately 5.5m (Fig. 3). The room was erected at the end of the C1st AD, most likely after 97 (Sommer 2000a).

Concerning the Vespasian advance into the upper valley of the Neckar, we still do not completely understand all the implications of the legionary size of Kastell I at Rottweil (62) and R. Franke’s reconstruction of the garrisoning of the complete or at least almost complete legio XI there (Franke 1998 and pers. comm.). Preliminary results of chemical analysis of tiles of legio XI by G. Schneider (Berlin) and F. Giacomini (Fribourg) show that we not only have different types of tiles in the Rottweil area but that their consistency is distinctively different from the tiles at Vindonissa. This excludes the possibility of an import. Therefore, we have to ask: What was the reason for such a strong garrison? Was Vindonissa empty at that time? Who acted as a caretaker garrison there? Depending on the date of Kastell I and the not yet established relation to Planck’s Kastell III there may even be consequences for the dating of southern-Gaulish samian ware.

General phasing and the dating evidence indicate that the fort at Ladenburg was not given up before the beginning of the C2nd AD. As it was situated right in the middle of the later town of Lopodunum, its destruction at least in the central area was prerequisite for the erection of the basilica and forum. Therefore, it is most likely that the establishment of the capital of the Civitas Ulpia Sueborum Nicrensium did not occur under the governorship of Trajan in 97/98, as hitherto thought, but appoximately 10 years later (Sommer 1998a: esp. 116ff., 136f.). This later date is in accordance with the t.p. found at Nida-Heddernheim, capital of the Civitas Taunensium further north (Fischer et al 1999: esp. 9ff., 423ff.), and is relevant for the date of the construction of the Neckar-Odenwald-Alb limes and perhaps the Taunus- and Wetterau limes, too.

Generally, the idea of a legion at Rottweil is supported by G. White’s independent interpretation of the bath-complex within the fortress as legionary baths (White 2000). His starting-points were layout and size.

Different dating there was suggested in the recent study of the coinage in our area by K. Kortüm (1998). He compared differences in the relative coin-loss in relation to a standard curve and argued for a later date under Trajan for the establishment of the limites. This stands in sharp contrast to the conventional assumption of a Domitianic date for these linear installations. However, the likelihood of the later date is high, taking into consideration:

Allow me to add a detail to Rottweil’s Fort III, or rather its military vicus. Between the fort’s porta praetoria and the baths under the church of St. Pelagius, which most likely originate from the military baths, a layer of gravel and stone slabs almost 2ft thick is under excavation at present (Sommer 2000c). Though heavily disturbed by later settlement, it appears over the whole area in question. As there is not much building activity underneath, it suggests the reconstruction of a large market-place in the triangle of the long-distance road coming from the east (Straßbourg; 134) towards the porta praetoria, the fort’s south-western front and the baths. As plazas have been found in so many other sites, where large-scale excavations and prospection have taken place (Bonn, perhaps Groß-Gerau, Heddernheim, perhaps Heldenbergen, Ladenburg, Regensburg-Kumpfmühl, Saalburg, Zugmantel), I believe that we have to look for market squares within the military vici at almost every fort of some importance, most likely in front of the porta principalis dextra, or the porta praetoria, if the baths had to be installed there for topographical reasons.

a b c

that forts newly excavated and studied, like Walheim II (57), seem not to begin before the C2nd AD (Kortüm & Lauber 2000) that the forts in the new hinterland were abandoned in that period (see above and other numismatic evidence) that it is unlikely that in a time of intense preparations for the wars against the Dacians there were enough troops to man a double line of forts (there are no indications of a considerable amount of new troops in the late Domitianic to Trajanic period).

Therefore, it is likely that the troops in south-west Germany were not moved forward before the second half of the reign of Trajan and that the reorganisation of the previous military area, the new hinterland, into civitates started at that late date. The most plausible time for such a major reorganisation of large parts of upper Germany and western Raetia is the period after the ending of the Dacian wars. The goal was to turn military land into civilian hands. It was achieved by founding the known series of Trajanic civitates mostly with their capitals on top of or next to military centres in connection - as it seems - with a program to settle their surroundings with civilians mostly from Gaul (Sommer 1999b; 1999c). The impact of this

Generally, there are no doubts about the beginning of the military activity in the Rottweil area under Cn. Pinarius Cornelius Clemens in the early years of Vespasian. Under him, not only forts, but also extensive roads were constructed. Evidence for the latter, other than the wellknown milestone from Offenburg (111, M 18), is now provided by timbers from the same place, which are dendro-dated to the year 74. They may have been parts of a bridge in the valley of the Kinzig. M. Yupanqui presented the relevant report at a conference in Rosenheim recently.

442

C. Sebastian Sommer: Recent developments in south-west Germany (eastern Germania Superior - western Raetia) later date for the earlier limes lines is dramatic and not yet fully realised.

Walheim (57) at the river, on the other hand, which lacked a natural access to the heights across the river anyway, may have had only rearward functions looking or functions along the river. Perhaps the second phase of the smaller fort at Walheim with its storage and/or industrial buildings was part of this not yet fully understood system (Kortüm & Lauber 1999; 2000). Interesting and problematic is the fact that neither here, nor at the left side of the river, do we find agricultural sites beginning before the middle of the C2nd (Hüssen 2000; however, as K. Kortüm pointed out to me, the high amount of Drag. 27 and Blickweiler at the villa at Lauffen may indicate a beginning slightly earlier than the originally suggested date of after the middle of that century; see Spitzing 1988 for the finds).

For the outer limes old questions about its possible dating earlier than Antoninus Pius (Schönberger 1985: 395f.) reappeared last autumn with further excavations in the civilian settlement (not a villa rustica, as originally thought) of Mundelsheim on the right bank of the Neckar (C) and with the publication of the samian ware from the eastern fort at Welzheim (45a). Firstly, the timbers of one of the newly discovered wells at Mundelsheim can be dated dendrochronologically to precisely 136 (Stork 1999). A reused barrel made of pinewood served as a wellchamber and dates to AD 130 ±10. This coincides with the date of AD 132 ± 10 of a well found earlier (Planck 1989: 178). Secondly there is a large amount of samian ware from southern Gaul at Mundelsheim (Planck 1989: 182: I. Stork pers. comm.). Therefore, little doubt remains that this settlement started at about the date given by the wells.

In Raetia, large parts of the fort of the ala II Flavia at Heidenheim are at present under excavation (66b; G. Balle & I. Stork pers. comm.; Balle 2000). A first result is that there was only one ditch in the north. Inside, in the left praetentura the continuation of the barracks excavated in 1965 and published by J. Heiligmann (1990) is unearthed. Once again (Sommer 1995), it appears that they contained drains in the front rooms and therefore they have to be considered as stable barracks. However, additional elongated pits perhaps with drainage functions, too, were discovered in regular intervals in front of the rooms of a new barrack.

Correspondingly, H.H. Hartmann interpreted the high proportion of non-Rheinzabern samian ware at WelzheimEast as an indicator for the foundation of this fort at around 130-140 (in van Driel-Murray & Hartmann 1999, similar tendency for Aalen [66] in Hartmann 1995). Unfortunately, the publication of the finds of Welzheim-East (for the leather finds see below) suffers from the fact that the texts needed about 10 years to appear in print and remained unrevised. From the present point of research (J. Lauber & K. Kortüm pers. comm.) some of the early stamps, quoted by Hartmann, are from Rheinzabern and not from central Gaul or Blickweiler and appear at most other forts at the outer limes (compare Biegert & Lauber 1995). Of the decorated samian, most is still common in the second half of the C2nd (eg. Verecundus; Schaub 1994). As K. Kortüm’s numismatic study already cited supports the conventional date of the beginning of the outer limes after the middle of the C2nd (Kortüm 1998), the finds from Welzheim cannot be taken as an indicator for the possibility of the construction of several forts at the limes before its general installation. Instead, considering the finds from Mundelsheim, we have to ask what happened in the northern part of the Neckar limes (Fig. 4). We still do not know where the palisade ended which J. Biel discovered some years ago way south-east of the previously reconstructed end of the Odenwald limes opposite Bad Wimpfen east of Bad Friedrichshall (Fig. 5; Biel 1991). Two towers, found there already in 1962 and 1964, were originally thought to be look-out posts, but have now to be considered as part of the limes.

At this point it is perhaps worthwhile mentioning that the system of putting horses and men together can be detected at more and more places and seems to have been the general way to keep the horses in forts (eg. Wallsend and South Shields, see N. Hodgson this volume; compare Sommer 1999d for the general statement). There is less and less to support the view expressed by E. Grönke (1999). She believed in the provision of stables only for the troops who were ‘on duty’. This would mean everchanging users of the buildings and no proper responsibility for their maintenance. In this connection I would like to draw your attention to a new book and a colloquium about the Roman cavalry or rather the publication of the papers given (Kemkes & Scheuerbrandt 1998; 1999). Particularly interesting are J. Scheuerbrandt’s ideas about the strategic concept underlying the deployment of the alae, based on the assumption of a maximum daily travelling distance of about 80kms for horsemen (Scheuerbrandt in Scheuerbrandt & Kemkes 1998: 9-17). Concerning horses, too, I have to come back to my earlier suggestion, following a question of H.U. Nuber, that the military vici of the ring-type are commonly, if not exclusively, to be found with forts with a cavalry garrison (Sommer 1998b: 50ff.; 1999d: 87f.). At Buch (67), with a likely garrison of a cohors equitata, a geophysical survey by H. v. d. Osten-Woldenburg not only made stable barracks in the retentura most likely (Sommer 1995: 161 with fig. 8), but showed during the course of the last winter that the fort was surrounded by a vicus of the ring-type, too

Following a suggestion of K. Kortüm, we may have to reconsider parts of the area to the right of the river as part of the earlier frontier zone, at least under Hadrian (see Schönberger 1985: 395). Perhaps there existed an extension of the limes down to Benningen (58), and Mundelsheim was some kind of a lookout post. There seems to be the possibility of a fort here, as the site possesses an overwhelming view towards the south-east and north. The forts of Heilbronn-Böckingen (56) and 443

Limes XVIII

(Fig. 6; v. d. Osten-Woldenburg pers. comm.). The possibility of such a lay-out was suggested independently by B.A. Greiner in his Ph.D. thesis on the excavated parts (to be published by the Landesdenkmalamt BadenWürttemberg soon). According to the dendro-dates, the beginning of that vicus was around the year 161. Changes within the military vicus around 193 resulted in a different parcelling, and perhaps the new stone buildings next to the baths. Unfortunately, G. Seitz in the recent publication of her more than 10 year old study of that area was not able to find a clear interpretation for the otherwise unknown plans of these buildings (Seitz 1999). Surprising in a military vicus are two rectangular foundations at both sides of the continuation of the via praetoria where it meets the main ring-road of the vicus. They hint at a kind of a monumental arch (Greiner pers. comm.).

More and more ambiguous is the end of the Roman occupation in south-west Germany. In the hinterland there is growing evidence for individual sites which ended before 260, some as early as in the 230s. Not all found their end in a possible earthquake, as the villa rustica at Oberndorf-Bochingen (although at the time of that incident the site may have already been abandoned; south of Sulz [61a]). Here fine construction details have been revealed, among them heights of 7.5 and 8.4m for simple rectangular buildings and roofs with an angle of 32° (current excavation; Sommer 2000b). In the vicus of Walheim (57) no destruction can be detected as the cause for its end. The correspondance analysis with all the finds from a large part of the excavated vicus established an end of all activities long before for example the latest phases of Pforzheim and Holzhausen (Kortüm & Lauber 2000). Equally, the villa rustica at Wurmlingen (north of Tuttlingen [155]) does not show destruction as the reason for its abandonment in the 250s (Reuter 2000). Whether economic decline can be considered as a general reason has yet to be established.

Except for small excavations at Jagsthausen in respect to the military vicus (41; Thiel 1997), Welzheim-West, which brought to light interval towers (45; Thiel 1999), and Aalen with new knowledge about the water-supply (66; a well next to the porta principalis sinistra; Krause 1999) and about the front of the porta praetoria (Krause 1997: 154f.), the outer limes generally does not see too much rescue archaeology these days. Many forts and partially their vici are well protected and the whole limes from Neuwied at the Rhine to Eining at the Danube will be submitted in its entire length of 550kms for inclusion into the list of World Heritage Sites in 2003. The preparations are directed by A. Thiel; 4 “Streckenkommissare” will do the fieldwork, one for each participating state. Some years ago D. Müller from the Landesdenkmalamt BadenWürttemberg began to conduct detailed topographical surveys at the limes. A finished piece of work studied an interesting sector north of Mögglingen where the Raetian limes runs through an extensive area with older burial mounds (north-east of Böbingen, 65). The course of the wall here shows no respect for these mounds. A different, longterm-project aims at a better understanding of the 80kms long stretch running in a straight line from Hönehaus to the Haghof (approximately [39] to [45] on the upper German limes). The questions here are: How straight is ‘straight’ and how did the Romans do it?

On the other hand, a revision of the limes’ dendro-dates at Buch suggests the hiding of household goods in wells and the subsequent destructive fire in the military vicus in 254 (B. Greiner pers. comm.; correct the dates in Kortüm 1998: 62f.). Whether some consecutive levelling there has to be assigned to a few civilian survivors or a surviving military garrison has yet to be established. Generally, the numismatic evidence points to a continuous occupation of the limes as a whole up to the 250s and an end around 255260 (Kortüm 1998). An interesting discussion of this problem from the point of view of the hoards is provided by T. Fischer (1999). The almost immediate reoccupation of some abandoned sites, like Wurmlingen, through Germans in the 260s (Reuter 2000) lends credibility to his renewed thesis that external threats are more likely the reason for the ‘fall’ of the limes than internal problems of the Roman empire. To conclude, it has to be stated that the ideas about the sequence of the Roman occupation in the eastern part of upper Germany and in western Raetia are far from fixed. Changes and adaptations are caused by new discoveries as well as the reinterpretation of old knowledge. Some of them have large-scale consequences, like the re-dating of the Neckar-Odenwald-Alb limes; others are of more local or regional effect, like the discovery of a 4ha fort close to Neresheim at the border to Bavaria, of which I heard just some days before the conference.

In other areas, our knowledge of the limes is profiting from desk-top research and its publication. Buch and the samian ware from Welzheim were already mentioned. Very interesting are also the results which C. van Driel-Murray drew from the leather finds at Welzheim-East (in v. DrielMurray & Hartmann 1999). Not only was she able to provide a remarkable insight into the great variety of provincial Roman footwear, she also presented aspects of a group of refugees or an otherwise newly impoverished population. Admittedly, this interpretation was influenced by the actual political situation some 10 years ago, when the text was written. Now, at the time of publication, van Driel-Murray is not too sure about her conclusions any more, not least due to a certain insecurity in the dating of the finds (cf. above and 1999: 11f. and footnote).

Bibliography Balle G. 2000 Neues zum Reiterkastell von Heidenheim. Archäologische Ausgrabungen in Baden-Württemberg: 90-94. Biegert S. & Lauber J. 1995 Töpferstempel auf glatter Sigillata vom vorderen/rätischen Limes. Mit einem Beitrag von K. Kortüm. Fundberichte aus BadenWürttemberg 20: 547-667.

444

C. Sebastian Sommer: Recent developments in south-west Germany (eastern Germania Superior - western Raetia) Biel J.

1991 Siedlungsgrabungen in Bad Friedrichshall, Landkreis Heilbronn. Archäologische Ausgrabungen in Baden-Württemberg: 60-65. v. Driel-Murray C. & Hartmann H.H. 1999 Das Ostkastell von Welzheim, Rems-Murr-Kreis. Die römischen Lederfunde - Die Terra sigillata. Forschungen und Berichte zur Vor- und Frühgeschichte BadenWürttemberg 42. (Stuttgart). Fischer T. 1999 Materialhorte des 3. Jhs. in den römischen Grenzprovinzen zwischen Niedergermanien und Noricum. In Das mitteleuropäische Barbaricum und die Krise des römischen Weltreiches im 3. Jahrhundert. Spisy Archeologického Ústavu AV ČR Brno (Brno): 19-50. Fischer U., Eschbaumer P., Fasold P., Huld-Zetsche I., Rupp V. & Schubert H. 1999 Grabungen im römischen Vicus von Nida-Heddernheim 1961-1962. Schr. Frankfurter Museum Vor- u. Frühgeschichte Archäologische Museum 14. (Bonn). Franke R. 1998 Ein Lager der legio XI Claudia in Arae Flaviae/Rottweil und die Besetzung des oberen Neckargebiets. Jahresbericht 1997 Gesellschaft Pro Vindonissa. (Brugg): 25-32. Gassmann G. & Trumm J. 1998 Ein neues römisches Lager im Wutachtal ? Sondagen bei Untereggingen, Gde, Eggunggens Kreis Waldshut. Archäologische Ausgrabungen in Bad-Würrtemberg 23: 141-148. Grönke E. 1999 Grundsätzliches zur Pferdehaltung in römischen Kastellen - Die Ställe im Alenkastell in Weißenburg. In M. Kemkes & J. Scheuerbrandt (edd.) Fragen zur römischen Reiterei. Kolloquium zur Austellung “Reiter wie Statuen aus Erz. Die römische Reiterei zwischen Patrouille und Parade” im Limesmuseum Aalen am 25/26.02.1998. (Aalen/Stuttgart): 91-100. Hartmann H.H. 1995 Terra sigillata aus dem Stabsgebäude des Kastells Aalen. Fundberichte aus Baden-Württemberg 20: 667-705. Heiligmann J. 1990 Der ‘Alb-Limes’. Ein Beitrag zur römischen Besetzungsgeschichte Südwestdeutschlands. Forschungen und Berichte Vor- u. Frühgeschichte Baden-Württemberg 35 (Stuttgart). Hüssen C.-M. 2000 Die römische Besiedlung im Umland von Heilbronn. Forschungen und Berichte Vor- u. Frühgeschichte Baden-Württemberg 78. (Stuttgart). Kemkes M. & Scheuerbrandt J. 1998 Zwischen Patrouille und Parade. Die römische Reiterei am Limes. (Aalen/Stuttgart). Klee M. 1999 Zum Stand der Limesforschung in der Provinz Obergermanien. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 95-114. Klein F. 1999 Neue Untersuchungen in Rißtissen, Stadt Ehingen, Alb-Donau-Kreis. Archäologische Ausgrabungen in Baden-Württemberg: 96-99. Kortüm K. 1998 Zur Datierung der römischen Militäranlagen im obergermanisch-rätischen Limesgebiet. Chronologische Untersuchungen anhand der Münzfunde. Saalburg-Jahrbuch 49: 5-65. Kortüm K. & Lauber J. 1999 Das Kastell II von Walheim, Kr. Ludwigsburg (Baden-Württemberg). Zur Innenbebauung und Funktion umwehrter Lagerareale. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 355-359.

Kortüm K. & Lauber J. 2000 Topografie des römischen Walheim. Denkmalpfege in Baden-Württemberg 29: 32-41. Krause R. 1997 Die frühmittelalterliche Keimzelle Aalens bei der St. Johann-Kirche Aalen. Ostalbkreis. Archäologische Ausgrabungen in Baden-Württemberg: 152-159. Krause R. 1999 Zur Wasserversorgung des Reiterkastells in Aalen. Ostalbkreis. Archäologische Ausgrabungen in Baden-Württemberg: 91-93. Lenz K.H. 1998 Villae rusticae: Zur Entstehung dieser Siedlungsform in den Nordwestprovinzen des römischen Reiches. Kölner Jahrbuch 31: 49-70. Neth A. 1999 Zum Fortgang der Ausgrabung in der zweiten Viereckschanze bei Nordheim, Kreis Heilbronn. Archäologische Ausgrabungen in Baden-Württemberg: 75-79. v. d. Osten-Woldenburg H. 1999 Geophysikalische Messungen bei Mengen und Ennetach. Archäologie im Umland der Heuneburg. Vorträge des 2. Ennetacher Arbeitsgesprächs vom 18. März 1999. Archäologie Informationen Baden-Württemberg 40: 11-21. v. d. Osten-Woldenburg H. 2000 Neue geophysikalische Prospektionen im Umfeld des Kohortenkastells rainauBuch, Ostalbkreis. Archäologische Ausgrabungen in Baden-Württemberg: 87-90. Planck D. 1989 Ein römisches Mithräum bei Mundelsheim, Kreis Ludwigsburg. Archäologische Ausgrabungen in Baden-Württemberg: 177-183. Reim H. 1999 Ausgrabungen im römischen Kastell auf dem “Berg” bei Ennetach, Stadt Mengen, Kreis Sigmaringen. Archäologische Ausgrabungen in BadenWürttemberg: 87-90. Reuter M. 2000 Von der Antike ins Frühmittelalter. Denkmalpfege in. Baden-Württemberg 29: 42-44. Schaub A. 1994 Markomannenzeitliche Zerstörung in Sulz am Neckar - Ein tradierter Irrtum. Bemerkungen zur reliefverzierten Terra Sigillata vom Ende des zweiten Jahrhunderts. In H. Friesinger, J. Tejral & A. Stuppner Markommenkriege - Ursachen und Wirkungen. Spisy Archeologického Ústavu AV ČR 1 (Brno): 439-445. Schmid J. 2000 Gontia - Studien zum römischen Günzburg. Mit Beiträgen von Th. Becker, P. Berktold, M. Scholz und F. Tränkle. (München). Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Elbe. Berichte der Römische-Germanische Kommission 66: 321-497. Seitz G. 1999 Rainau-Buch I. Steinbauten im römischen Kastellvicus von Rainau-Buch (Ostalbkreis). Forschungen. und Berichte zur Vor- und Frühgeschichte Baden-Württemberg 57. (Stuttgart). Smettan H.W. 2000a Vegetationsgeschichtliche Untersuchungen am oberen Neckar im Zusammenhang mit der vor- und frühgeschichtlichen Besiedlung. Materialhefte zur Archäologie Baden-Württemberg 49: (Stuttgart). Smettan H.W. 2000b Besiedlungsschwankungen von der Latènezeit bis zum frühen Mittelalter im Spiegel südwestdeutscher Pollendiagramme. Fundberichte aus Baden-Württemberg 23: 779-808. Spitzing T. 1988 Die römische Villa von Lauffen a.N. (Kr. Heilbronn). Materialhefte zur Vor- und Frühgeschichte in Baden-Württemberg 12. (Stuttgart). Sommer C.S. 1995 “Where did they put the horses?” Überlegungen zu Aufbau und Stärke römischer Auxiliartruppen und deren Unterbringung in den Kastellen. In W. Czysz, C.-M. Hüssen, H.-P. Kuhnen,

445

Limes XVIII

C.S. Sommer & G. Weber (edd.) Provinzialrömische Forschungen Festschrift für Günter Ulbert zum 65. Geburtstag. (Espelkamp): 149-168. Sommer C.S. 1998a Vom Kastell zur Stadt. LOPODVNVM und die CIVITAS VLPIA SVEBORVM NICRENSIVM. In H.J. Probst (ed.) Ladenburg. Aus 1900 Jahren Stadtgeschichte. (Ubstadt-Weiher): 81-201 & 806-809. Sommer C.S. 1998b Kastellvicus und Kastell - Modell für die Canabae legionis? Jahresbericht 1997 Gesellschaft Pro Vindonissa. (Brugg): 41-52. Sommer C.S. 1999a From conquered territory to Roman province: recent discoveries and debate on the Roman occupation of SW Germany. In J.D. Creighton & R.J.A. Wilson (edd.) Roman Germany. Studies in cultural interaction. Journal of Roman Archaeology Supplementary Series 32: 159-198. Sommer C.S. 1999b Trajan und die Einrichtung der Civitas Ulpia Sueborum Nicrensium. In E. Schallmayer (ed.) Trajan in Germanien - Trajan im Reich. Bericht des 3. Saalburgkolloquiums. Saalburg-Schr. 5: 241246. Sommer C.S. 1999c Unterschiedliche Bauelemente in den Kastellvici und Vici. Hinweise auf die Herkunft der Bevölkerung in Obergermanien. In N. Gudea (ed.) Roman Frontier Studies. Proceedings of the XVIIth International Congress of Roman Frontier Studies. (Zaláu): 611-621. Sommer C.S. 1999d Wohin mit den Pferden? Stallbaracken sowie Aufmarsch- und Übungsplätze in römischer Zeit. In M. Kemkes & J. Scheuerbrandt (edd.) Fragen zur römischen Reiterei. Kolloquium zur Austellung “Reiter wie Statuen aus Erz. Die römische Reiterei zwischen Patrouille und Parade” im Limesmuseum Aalen am 25/26.02.1998 (Aalen/Stuttgart): 84-90. Sommer C.S. 2000a Eine 5.5m hohe, bemalte Wand aus dem Kastell I in LOPODVNVM - Ladenburg am Neckar - Bergung, Restaurierung und Rekonstruktion. Fundberichte Baden-Württemberg 23: 157-192. Sommer C.S. 2000b Ein großes landwirtscaftlides Nebengebäude in Oberndorf-Bochingen. Aspekte der römischen Architektur. Archäologische Ausgrabungen in Baden-Württemberg. 117-121. Sommer C.S. 2000c Römischen Graßbau, Platz und mittelalterliche Siedlung um St. Pelaglus in Rottweil. Archäologische Ausgrabungen in Baden-Württemberg: 102-105. Stork I. 1999 Bemerkenswerte Funde aus dem römischen Vicus auf der “Ottmarsheimer Höhe”, Gemarkung Mundelsheim und Besigheim-Ottmarsheim, Kreis Ludwigsburg. Archäologische Ausgrabungen in BadenWürttemberg: 116-119. Thiel A. 1997 Weitere archäologische Aufschlüsse im Kastellvicus von Jagsthausen, Kreis Heilbronn. Archäologische Ausgrabungen in Baden-Württemberg: 109-111. Thiel A. 1999 Die Wehrtürme des Westkastells von Welzheim, Rems-Murr- Kreis. Archäologische Ausgrabungen in Baden-Württemberg: 94-96. White G. 2000 Das römische Badegebäude auf dem Nikolausfeld in Rottweil am Neckar - Studien zur Baugeschichte und historischen Bedeutung. Fundberichte aus Baden-Württemberg 23: 193-242. Wieland G. 1998 Überlegungen zur Höhlennutzung im oberen Donautal während der Spätbronze- und Eisenzeit. In H.

Küster, A. Lang & P. Schauer (edd.) Archäologische Forschungen in urgeschichtlichen Siedlungslandschaften. Festschrift G. Kossack zum 75. Geburtstag. Regensburger Beitrage z. prähistorische Archäologie 5 (Regensburg): 389-418. Wieland G. 2000 Späte Kelten in Südwestdeutschland. Forschungen zum Siedelwesen der Spätlatènezeit zwischen Neckar und Oberer Donau. Denkmalpflege Baden-Württemberg 29: 26-31. Wieland G. & Luik M. 1999 Die keltischen Viereckschanzen von Fellbach-Schmiden (Rems-Murr-Kreis) und Ehningen (Kreis Böblingen). Mit Beiträgen von M. Luik und anderen. Forschungen und Berichte Vor- und Frühgeschichte Baden-Württemberg 80. (Stuttgart).

446

C. Sebastian Sommer: Recent developments in south-west Germany (eastern Germania Superior - western Raetia)

Fig. 1. Military installations and civilian settlements in south-west Germany. The signs of the sites mentioned in the text are enlarged. For the phasing of the sites see Sommer 1999: fig. 6.2.

447

Fig. 2. Forts along the Upper Danube with road system. 1 – Ennetach; 2 - Ertingen?; 3 – Emerkingen; 4 - Rißtissen; 5 - Unterkirchberg. Hatching: (military) vicus. Scale 1:150.000.

Limes XVIII

448

C. Sebastian Sommer: Recent developments in south-west Germany (eastern Germania Superior - western Raetia)

Fig. 3. Painted wall-plaster from a wattle-and-daub wall of a hall in the praetorium of the latest phase of Kastell I at Ladenburg, approximately 5.5m high.

449

Limes XVIII

Fig. 4. Installations at the southern end of the Odenwald-Limes and the northern part of the Neckar-Limes. 1 - forts at Neckarburken: 2 - southern-most confirmed towers of the Odenwald-Limes WP 10/75 and 10/76: 3 - fort at Bad Wimpfen: 4 - fort at Kochendorf: 5 - tower at Kochendorf: 6 - tower at Bad Friedrichshall; 7 - fort at Heilbronn-Böckingen: 8 - forts at Walheim: 9 vicus at Mundelsheim: 10 - fort at Benningen: 11 - vicus at Murr. Continuous line: Limes, palisade/course proven: Hatched line: Limes, reconstructed course: Dotted line: old reconstruction of the Limes. Scale 1: 150.000.

450

C. Sebastian Sommer: Recent developments in south-west Germany (eastern Germania Superior - western Raetia)

Fig. 5. Trajanic (?) palisade (1) with timber (2, circular ditch) and stone watch-tower (3) east of Bad Friedrichshall-Kochendorf. LBNr. L 6720/069B-04, 1802/08, 7.6.1991 (O. Braasch, Landesdenkmalamt Baden-Württemberg).

451

Limes XVIII

Fig. 6. Fort and military vicus at Buch. Geomagnetic survey by H. v. d. Osten-Woldenburg.

452

The late Roman fort at the Valkhof in Nijmegen. Nijmegen at the transition from the Roman period to the middle ages J. Thyssen

In the past, most publications about Roman Nijmegen generally only paid little attention to the late Roman period (Willems 1990: 73-85; Wynia 1979). The most important reason for this was the lack of a clear overview, which affected adversely the possibilities to lay any links between the various finds discovered in the town centre over a relatively long period. Archaeological research carried out in 1996-1997 by the Bureau Archeologie of the Municipality of Nijmegen at the construction site of the new Valkhof Museum, and during the re-development of the inner town in 1998-1999 has yielded a lot of new data that significantly clarify the late Roman and early medieval topography. A significant proportion of these data constitute additions and corrections to the survey published in 1990, for which information with regard to the continuity of habitation in the period 400-750 had been gathered systematically for the first time (Bloemers & Thijssen 1990; for recent developments see Van Enckevort & Thijssen 1996; 2001; Van Enckevort, Haalebos & Thijssen 2001: 97-109). Central to this article were the developments in the late Roman castellum in the Valkhof area and the adjacent late Roman cemetery in the inner town of Nijmegen. The rôles that the Roman road system and the former town of Ulpia Noviomagus Batavorum played in this development were not considered there. Nor were those elements of the settlement whose activities really remained limited to the late Roman period: the settlement on the bank of the River Waal, the Frankish habitation south-east of the castellum and the late Roman inhumation-cemetery in the east of Nijmegen. The Merovingian settlement, as a new element, in the west of the lower town did receive ample attention, however. The following survey however, will deal with all these elements.

and the temple complex on the Maasplein were burnt down. Indications suggest that in the C3rd, life in the town continued on a more modest scale. The continuous waves of Germanic attacks shortly after the middle of the C3rd brought about a definite change in this. From that time onwards, only a handful of people in and between the ruins of the town could find sufficient possibilities to survive (for a recent overview: Van Enckevort, Haalebos & Thijssen 2001: 64-84). The archaeological research carried out in the past years in the Roman town area has revealed that several activities were definitely still taking place in the C4th. About 10% of the Roman coins found in excavations date from that time, and coins from the period after 360 are no exception. In the course of time, the town area and its immediate vicinity also yielded other objects, such as Argonne ware with rouletting and crossbow fibulas (Zwiebelknopffibel), which suggest some form of habitation in the late Roman period. The ruins of the town were definitely used as a quarry in this period, but in particular the large public buildings in the centre of the town may well still have offered possibilities for habitation (Van Enckevort, Haalebos & Thijssen 2001: 105). The late Roman period: the historical framework (Fig. 1) From the end of the C2nd, the rapid succession of a number of military emperors who rather lacked decisiveness was a limiting factor to the influence of the central authority in Rome on the events in the border areas of the Roman empire. The Gallic emperors Postumus (259-268), Victorinus (268-270) and Tetricus I and II (270-273) still managed to maintain a semblance of authority in the northwestern part of the empire, though they were unable to halt the Germanic attacks. In the years 258/259 and 270, large groups of Germans, usually called Franks in the sources from that period, penetrated from the north into the deep south of Gaul. The continuing disturbances resulted in a decrease of the population, not only in the town but also, more particularly, in the countryside.

Nijmegen in the middle Roman period For Nijmegen, the granting of town privileges and elevation to municipium at the beginning of the reign of Trajan (98-117) meant the start of a heyday for the town Ulpia Noviomagus Batavorum, which would last about a century. The construction of the town wall during the reign of Marcus Aurelius (161-180) had to secure that situation, but also constituted an indication that everything must come to an end. The construction of this wall is probably related to the threat formed by the first serious Germanic attack by the Chauci, who scourged the Dutch and Belgian coastal areas in the year 172. From that time onward, the unrest at the borders of the Roman empire gradually began to constitute a serious threat to the layout and infrastructure of the Roman town. Already during the first skirmishes, the low-lying town at the Waal turned out to be extremely vulnerable. Large public buildings, such as the bathhouse

The decisiveness of the emperor pair Diocletian (284-305) and Maximinianus Herculius (286-305) and vice-emperor Constantius Chlorus (305-306), father to Constantine the Great (306-337), imbued the central authority with renewed prestige. They restored the military infrastructure, in which the connecting roads to the surrounding countryside were most important. The rigid defence line along the border was exchanged for a more intensive defence, with troops, particularly cavalry, stationed also in the surrounding countryside, the more effectively to be 453

Limes XVIII

able to intercept any intruders. Through roads were protected with guard posts and river crossings and town were protected and fortified. New fortifications were built in strategic places. In addition to this, the Franks were used as laeti and foederati for the defence of the northern border of the empire. All in all, up to the middle of the C4th, this was a reasonably successful strategy whenever the struggle for the throne would flare up and mainly border area troops were employed. The Franks realised their opportunities and benefited immediately and on a massive scale. With great difficulty and the use of high bribes, peace was restored to the lower Rhine area once more, under Julian (360-363) and in the beginning of the reign of Valentinian I (364-375), but this peaceful period would turn out to be brief. From then onward, the army leadership and border troops for the largest part were made up of Franks. As a result of this, the power in the lower Rhine area was really in Frankish hands. The fall of Mainz (406) and Cologne (shortly after 450) opened up the northern part of Gaul to the formation of an own Frankish kingdom. Eventually, Clovis (481-511) managed to unite the various principalities into a single kingdom.

Beside data with regard to its development, the investigation at the Valkhof Museum construction site also resulted in the first section through the second defence ring of the late Roman castellum (Fig. 3). In combination with the finds, this offers enough information to be able to sketch its development in broad outlines. The second defence zone around the castellum on the Valkhof consisted of two ditches of different widths and an earthen wall with stone facing. The outer, most southern ditch is 12 to 13m wide and at least 4m deep. The inner ditch measures over 10m, with a depth of more than 3.5m. The outer edge of the ditch system is about 45m from a soil mark that contains a lot of small lumps of tuff and mortar, and which undoubtedly indicates the site of the knockeddown castellum wall. Between the ditches, there is a verge of some 15m. The sections of the ditches show that different phases may be distinguished in their filling. There are indications that the fortifications were carried out in wood in the first phase and in stone in a second phase. If they were, the development of this fortification would be very similar to the well-known late Roman castellum at Cuijk (Bogaers 1967), which was strategically situated at the bridge across the Meuse investigated in 1992-1993 (Goudswaard, Kroes & Van der Beek 2000/2001).

The late Roman castellum at Nijmegen The discovery of a late-Roman ditch1 at the construction site of the ‘De Lindenberg’ cultural centre in 1969 first confirmed the presence of a long-suspected military settlement on the Valkhof, in the centre of Nijmegen (Bogaers 1969). This suspicion had been based mainly on finds discovered by the town architect Weve (1910-1911), and later by the Leiden archaeologist Brunsting (1946; cf. Bloemers & Thijssen 1990: 140-143). Weve’s main contribution was the discovery of skeletal remains of six individuals during his excavation of the so-called Barbarossa ruins of the Valkhof. In 1921, the then Nijmegen town archivist Daniëls published these finds as the mortal remains of Frankish soldiers who had died during fights for the possession of the fortification in the C4th (Daniëls 1921).

At a larger distance from the castellum, the remains of a first defence ring around the castellum were found. Two ditches were discovered also here, during archaeological research carried out by the Rijksdienst voor het Oudheidkundig Bodemonderzoek (ROB). The few finds do show that these ditches must have functioned in at least part of the C4th. During the investigation in the 1990s, about 3,000 coins were found. The great majority of these are found in the ditches of a second, inner defence ring. A few hundred have been dated to the beginning of the Christian era and belong to the commodities of the early Roman settlement Oppidum Batavorum. The other coins include a number from the beginning of the reign of Constantine the Great and his predecessors, including Diocletian. The images presented by the coins found in the two ditches are rather dissimilar. The filling of the outer ditch, up to a considerable depth and in combination with a whole range of other C4th coins, contained coins of Theodosius and Arcadius (388-394). This is also true of the inner ditch, but the layers that contain these coins are considerably less deep there. Among the coins from the bottom layers of the filling of the inner ditch, the period 364-378 (Gratian, Valens and Valentinian I) is remarkably well represented; much more so than in the outer ditch, where coins dominate that stem from the second quarter of the C4th in particular. All in all this does suggest that the ditches still looked like ditches around the year 400.

In the subsequent 30 years, the ditch that was discovered in 1969 was re-examined several times. Renewed research of the old finds, in combination with that of finds from recent excavations, revealed that the ditch, and therefore also the fortification on the Valkhof, had definitely been in use up to the C5th. The upper layers yielded numerous pottery sherds from subsequent periods. In the upper layers of the ditch, recent excavations at the building site of the Valkhof Museum have even uncovered sherds that cover the entire post-Roman period into the C13th. These clearly show that up to that time, this ditch must have constituted a clear topographic element in the landscape. The closing date of the find material is undoubtedly related to the fortification of the Valkhof by the counts of Gelre and the connected restructuring of the environment in or soon after 1250 (Fig. 2).

The civilian settlement on the east bank of the Waal The most important data concerning the walled-in settlement on the east bank of the River Waal were obtained during an investigation of the ROB between 1985

1

Later research has shown that this is the outer ditch of the second defence ring.

454

J. Thyssen: The late Roman Fort at the Valkhof in Nijmegen and 1987, at the building site of the Holland Casino (Sarfatij 1986; Van Enckevort & Thijssen 1996: 90-91; Van Enckevort, Haalebos & Thijssen 2001: 103). The excavations revealed heavy tuff walls, part of which had already been uncovered in 1941, and another part in 19541955 by Brunsting. On those occasions, its significance and date had remained unclear. The later research showed that the walls had been built in various phases. Part of them probably belonged to a villa-type structure constructed against the hill of the plateau around the year 100. On the basis of the few finds, a trade settlement is assumed to have been present here during the C2nd and C3rd. In the C4th, parts of the earlier constructions were incorporated in the walls of the new settlement. Research has shown that at the time, there must have been a number of buildings on the river side of the wall. Habitation on the Waal bank however seems to have remained limited to the first half of the C4th.

graves were studied, the great majority of which turned out to be inhumation graves, into which the dead are almost always buried in coffins. Cremation graves scarcely occurred. A number of the graves investigated could be dated to the C4th on the basis of the interment gifts contained in them. A large part of these gifts had been placed in a niche dug out into the sides of the graves. The custom of grave gifts with the dead decreased considerably in the middle of the C4th. It is remarkable that this change seems to have been accompanied by a change of the direction of the graves; instead of interments with the head to the east (east-west graves), the later graves are turned to the west (west-east graves). This is likely to have resulted from the increasing Christian influence in the lower Rhine area, possibly related to the Milan Decree of 313. Brunsting found no burial gifts in most of the 631 graves. Most of these graves were east-west oriented. Very rarely, iron objects such as knives had been included, but these were insufficient for adequate dating of those graves at the time. This was also true of a group of graves with fibulae and belt mounts. Such objects are currently known to have belonged to the standard fittings of German officers and soldiers, and to be typical of the late C4th and early C5th. It is tempting to link this group of graves to the second period of the castellum under emperor Valentinian I, which is why the question of continuity, which was so often brought forward in the 1950s and 1960s, has only too often been answered with considerable reservation (Brunsting 1955).

Frankish habitation Archaeological research carried out in 1990 on the corner of the Berg en Dalseweg and the St. Canisiussingel, southeast of the castellum, for the first time in Nijmegen yielded traces of Frankish habitation. Though partly cut by a late medieval ditch and more recent disturbances, a large part of a sunken hut measuring over 4 by 2.5m, with a residual depth of about 1m, could be investigated. Parts of carbonised standards and horizontally placed wallboard could be recovered of the construction, the basis of which is likely to have been weakened by fire and collapsed and buried by the pressure of the surrounding sand. The entrance was on the south side (also the street side?), to judge from the hinges and lock found there. On the bottom, the burnt remains of a possible (?bread) oven were discovered. In the immediate vicinity of the sunken hut, a few dozen coins from the second and third quarters of the C4th were found. The area showed additional indications of habitation, which definitely seems to have extended also in a westerly direction. Municipal research on the Gerard Noodtstraat yielded a few hundred late Roman coins, which cover the entire C4th. In addition to this, a number of postholes may possibly be attributed to this period (Van Enckevort & Thijssen 1996: 91-92; Van Enckevort, Haalebos & Thijssen 2001: 103-105).

Earlier indications to prompt less reserve with regard to the issue of continuity had been found however. During work carried out just after the war, on the Grutberg and environs, a number of complete pots were found, which must stem from graves and may be dated to about 700 (Thijssen 1988; Van Enckevort & Thijssen 1996: 98; Van Enckevort, Haalebos & Thijssen 2001: 108; cf. Daniëls 1955: 239240). The municipal excavations in the inner town of Nijmegen in the 1990s increased the number of known graves from the late Roman period and early middle ages to over 1300. The boundaries of the cemetery still have not been reached, although they may be indicated roughly. To the west, the cemetery connects to a concentration of earlier cremation graves on both sides of the old Roman road between the castra of the tenth legion and Ulpia Noviomagus Batavorum. This road constitutes the northern boundary of the cemetery. Another road from the middle Roman period forms its southern boundary; the eastern boundary of the cemetery is not yet known.

Immediately east of this settlement, in the grounds of the Margriet nursing home, is the accompanying cemetery, which was used mainly during the C4th. It attained a final size of some 850 burials (Bloemers 1983: 193-198; Van Enckevort & Thijssen 1996: 93; Van Enckevort, Haalebos & Thijssen 2001: 107-108).

A new and important contribution to the issue of continuity is the recent find of a number of graves along the Burchtstraat. This group is somewhat different, on the one hand because of its slightly more isolated position in relation to the other graves, and on the other for its remarkable interment gifts. The presence of iron weapons such as axes, a lance head and even a shield boss is

The cemetery in the inner town of Nijmegen After World War II, Brunsting used the opportunities that resulted from the destructions inflicted on the centre of Nijmegen. Between 1947 and 1966, 10 excavation campaigns were carried out in the area of the town hall (Brunsting 1952; 1983, in particular 41). Hundreds of 455

Limes XVIII

notable, and warrants dating of this group to the second half of the C5th and early C6th. Such dateable early medieval graves had not been found in the cemetery in the inner town of Nijmegen before. Moreover, there are other finds that also indicate the special character of these graves: remains of gold brocade were found on one skeleton, while pieces of amber were found in the eye sockets of another (Van Enckevort & Thijssen 2001).

ditches and bent in westerly direction, toward the former Burchtpoort, in front of the row of houses on the south side of the Kelfkensbos, is also significant. The filling yielded only few finds, roughly dateable to the C14th. The construction seems to have been connected to the earliest layout of the town. On the eastern edge of this ditch, centrally between the two Roman ditches, the remains of a stone cellar were uncovered. The walls contained a lot of Roman debris. The cellar itself was filled with demolition debris. The pottery in a hearth area on the floor warrants dating of this cellar to the C13th. In addition to this, a pair of sunken huts from the same period was found between the Roman ditches. The use of stone for house construction in medieval Nijmegen was limited to the second half of the C13th. The demolished house must have been important. As far as is known, it is the earliest stone house in Nijmegen. Its location is close to the parish church, outside the castellum wall, but inside the outer Roman ditch. This is undoubtedly the presbytery that was demolished in 1249 at the construction of a new ditch (novum fossatum) to fortify the castle on the Valkhof.2 At that occasion, the parish church that must have been situated just north of the Valkhof, in the place of the Gertrudis chapel ruins, was demolished as well.

Earlier grave finds stem from the Valkhof itself. In 1980, the aforementioned old finds discovered by Weve in the Barbarossa ruins on the Valkhof could be studied anew (Thijssen 1980; Ypey 1983). On this occasion, the weapons that had originally been found near two of the skeletons were immediately recognised as the remains of saxes (Fig. 4). Such finds really only ever occur in graves. Closer inspection of both saxes – knives at the back of the sheaths, as well as a number of copper rivets and two decorative mounts on the sheath of the shorter sax – resulted in a reliable date (c.620-680), which makes it impossible for Daniëls’ earlier dating to have been correct. Merovingian habitation in the west of the lower town From 1978, sherds were repeatedly discovered in the area to the west of the Priemstraat in find situations that could only indicate an early medieval settlement, which had been ploughed over and built on in later times. In several profiles under the medieval habitation layers, again and again an elevated tillage layer was observed, of a thickness between c.0.50-1.50m. The top layer sometimes clearly showed the furrows of ploughing. The date of the Merovingian pottery found in the tillage layer is C6th/C7th. In a pit, some pottery was discovered that was made in nearby Ubbergen, about 1.5kms east of the Valkhof. Apart from domestic waste from the period 400-700, production waste was found from a Merovingian pottery in which red pottery was manufactured around 700. This suggests that the basis for the pre-urban development of medieval Nijmegen should be sought here, in the Waal riverbank area. Later construction obliterated nearly all traces of habitation in this period. Only occasional finds of pottery sherds and metal objects show that there were once houses here (Fig. 5; cf. Thijssen 1988; Bloemers & Thijssen 1990: 143-144; Van Enckevort, Haalebos & Thijssen 2001: 108109).

These finds indicate that the eastern part of the late Roman castellum is likely to have been in the hands of the ecclesiastical elite, starting with the Bishop of Cologne, for about a thousand years from the Merovingian period onwards. From the Roman period onwards, the western part of the castellum was in the hands of the secular authorities. Historical mention of a palatium (Paltz) set up here by Charlemagne is an indication of this. To summarize, the Valkhof may be concluded to have been in continuous use from the late Roman period onwards until 1795 (Fig. 7). Bibliography Bloemers J.H.F. 1983 Acculturation in the Rhine/Meuse basin in the Roman period: a preliminary survey. In R. Brandt & J. Slofstra (edd.) Roman and native in the Low Countries. Spheres of interaction British Archaeological Reports Int. Ser. 184. (Oxford): 159209. Bloemers J.H.F. & Thijssen J.R.A.M. 1990 Facts and reflections on the continuity of settlement at Nijmegen between AD400 and 750. In J.C. Besteman, J.M. Bos & H.A. Heidinga (edd.) Medieval Archaeology in the Netherlands. Studies presented to H.H. van Regteren Altena. (Assen/Maastricht): 133-150.

The Valkhof in the early and late middle ages North of the broken out late Roman castellum wall, and thus inside the fort, on the construction site of the Valkhof Museum, skeletons and skeletal remains were found that may be attributed to the ‘Alde Kirkhof’ mentioned in historical sources. The castellum wall formed the boundary of the grave finds. This is likely to be helpful for a more precise ascertainment of the place of the parish church (Fig. 6).

2

Thijssen 1980 13; Bloemers & Thijssen 1990. On the basis of current knowledge, the place of the parish church should be sought in the immediate vicinity of the Gertrudis chapel. In 1249, both the church and the chapel had to make way for the construction of a novum fossatum: the medieval ditch that was found was situated at the western edge of the construction site of the Valkhof Museum (cf. Sloet van Beele 1870: 67). The current Voerweg was not dug before 1400, and the traced-out road in that case remained a little to the north of the old church. Gorissen's topography (1956: 79, pl. 35) is not correct in this aspect.

The discovery on the west side of the Valkhof Museum building site of a medieval ditch, which cut the Roman 456

J. Thyssen: The late Roman Fort at the Valkhof in Nijmegen Bogaers J.E. 1967 Enige opmerkingen over het Nederlandse gedeelte van de limes van Germania Inferior (Germania Secunda). Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 17: 99-114. Bogaers J.E. 1969 Ontdekkingen op de Lindenberg. Numaga XVI, 1: 1-4. Brunsting H. 1952 Laat-Romeins grafveld onder Middeleeuwse kerk. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 3, 1: 9-12. Brunsting H. 1955 Continuiteit van bewoning in Nijmegen. Numaga (gedenknummer): 59-64. Brunsting H. 1983 Nijmegen, Valkhof, 1805. In M. AddinkSamplonius (ed.) Urnen delven. Het opgravingsbedrijf artistiek bekeken. (De Bataafsche Leeuw, Dieren): 3741. Daniëls M.P.M. 1921 Romeinsch Nijmegen. Het Valkhof. Oudheidkundige Mededelingen uit het Rijksmuseum van Oudheden te Leiden. N.R. 2: 6-37. Daniëls M.P.M. 1955 Romeins Nijmegen. (Nijmegen). Enckevort H. van, Haalebos J.K. & Thijssen J. 2001 Nijmegen. Legerplaats en stad in het achterland van de Romeinse limes. (Archeologische Berichten Nijmegen 3: Uitgeverij Uniepers/Bureau Archeologie Gemeente Nijmegen, Abcoude/Nijmegen). Enckevort H. van & J. Thijssen 1996 Graven met beleid. Gemeentelijk archeologisch onderzoek in Nijmegen 1989-1995. (2nd ed. 1998; Uitgeverij Uniepers/Bureau Archeologie Gemeente Nijmegen, Abcoude/Nijmegen). Enckevort H. van & Thijssen J. 2001 Graven op Mariënburg. Archeologisch onderzoek in het centrum van Nijmegen. (Archeologische Berichten Nijmegen 4: Bureau Archeologie Gemeente Nijmegen, Nijmegen). Geelkercken N. 1639 Vogelvluchtkaart van Nijmegen. Gorissen F. 1956 Stede-atlas van Nijmegen. (Gelre. Vereniging tot beoefening van Gelderse Geschiedenis en Recht, Arnhem). Goudswaard R., Kroes R.A.C. & van der Beek H.S.M. (edd.) 2000/2001 The Late Roman bridge at Cuijk. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 44, (Amersfoort): 439-560. Sarfatij H. 1986 Nijmegen. Opgraving stadskern (Waalkade). In Jaarverslag Rijksdienst voor het Oudheidkundig Bodemonderzoek 1985. (Ministerie van Welzijn, Volksgezondheid en Cultuur, Rijswijk): 52-55. Sloet van de Beele L.A.J.W. (ed.) 1870 Wilhelmus de Berchen. De nobili principatu Gelrie et eius origini. (Hagae Comitum - Den Haag). Thijssen J. 1980 Graven op het Valkhof. In G. Lemmens (ed.) Het Valkhof te Nijmegen. (Catalogi van het kunstbezit van de Gemeente Nijmegen 3: Commanderie van St. Jan, Nijmegen): 10-13. Thijssen J. 1988 400-750. Het gat gedicht. De bewoningsgeschiedenis van Nijmegen. Nijmeegs Katern. Nieuwsbrief voor de geschiedenis van stad en streek 2, nr. 4: 2-3. Willems W.J.H. 1990 Romeins Nijmegen. Vier eeuwen stad en centrum aan de Waal, (Historische Reeks Nijmegen 2. Stichting Matrijs, Utrecht). Wynia S.L. 1979 De laat-Romeinse tijd (ca. 260/270-400). In J.H.F. Bloemers, J.E. Bogaers, J.K. Haalebos & S.L. Wynia Noviomagus. Op het spoor der Romeinen in Nijmegen. (2nd ed. 1988; Rijksmuseum G.M. Kam, Nijmegen): 64-68. Ypey J. 1983 Twee saxen uit frankische graven op het Valkhof te Nijmegen (c.620-680). Numaga 30: 32-34.

457

Limes XVIII

Fig. 1. Late Roman Nijmegen (drawing by R. Mols and & H. van Enckevort, Gemeente Nijmegen). Key: 1 – The castellum; 2 – The Frankish settlement; 3 – The civil settlement on the eastern bank of the Waal; 4 - Civilian habitation on the southern edge of the former town of Ulpia Noviomagus Batavorum; 5 – The inner town cemetery; 6 – The Margriet Nursing Home cemetery.

Fig. 2. Late Roman ditch. Inner ditch of the second defence ring from the west. Profile at the foot of the rampart behind the town wall. Practically clean sand in the tip and dark latest phase. Photo: Rob Mols, Bureau, Archeologie Gemeente Nijmegen.

Fig. 3. Late Roman ditch. Outer ditch of the second defence rings from the east. Profile on the area of the former miniature golf course. Here also, clean filling in the tip and dark upper filling. Photo: Rob Mols, Bureau, Archeologie Gemeente Nijmegen.

458

J. Thyssen: The late Roman Fort at the Valkhof in Nijmegen

Fig. 5. Red pottery (so-called Ubbergen Ware) from the inner town. Probably c. 700. Photo: Rob Mols, Bureau, Archeologie Gemeente Nijmegen.

Fig. 4. Saxes from Graves III and V in de Barbarossa ruins of the Valkhof. According to Ypey (1982).

Fig. 6. Situation. Valkhof environment based on archaeological research. After Gorissen (1956): 77 pl. 33 with some changes. The two parallel ditches are Roman, between them the Presbytery. The ditch dates to the C13th-C14th.

459

Limes XVIII

Fig. 7. The Valkhof according to Van Geelkerchen 1639.

460

Die Ausgrabungen in Lahnau-Waldgirmes Armin Becker fand sich dort ein 7m breiter Zwischenturm auf 6 Pfosten. Auch der Turm in der Südostecke besaß ursprünglich 6 Pfosten, von denen nur die Standspuren der tiefer eingegrabenen hinteren Pfosten nachweisbar waren. Zu beiden Seiten des Eckturmes standen in 11 bzw. 16m Abstand 3.5m breite Zwischentürme auf 4 Pfosten.

Bei Geländebegehungen war in Waldgirmes auf einer hochwasserfreien Terrasse etwa 1300m nördlich der Lahn seit 1990 germanische und römische Keramik aus der Zeit um Christi Geburt geborgen worden. Im Spätherbst 1993 führte die Römisch Germanische Kommission (RGK) des Deutschen Archäologischen Instituts in Frankfurt daraufhin eine Sondagegrabung durch, in deren Verlauf erste Gruben aufgedeckt wurden. Eine geophysikalische Prospektion erbrachte im folgenden Jahr den leicht trapezoiden Umriß einer etwa 7.7ha großen Befestigungsanlage mit abgerundeten Ecken. 1994 und 1995 unternahm die RGK im Rahmen eines Schwerpunktprogrammes der Deutschen Forschungsgemeinschaf (DFG) weitere Grabungen.1 Da die Gemeinde Lahnau auf dem Areal der Anlage ein Gewerbegebiet plant, werden die Ausgrabungen seit 1996 gemeinsam von der RGK und dem Landesamt für Denkmalpflege Hessen, Abteilung Archäologische und Paläontologische Denkmalpflege fortgeführt.2 1999 und 2000 gewährte die DFG erneut Zuschüsse.

Straßen, Wasserversorgung Von den beiden nachgewiesenen Straßen verlief eine vom Osttor nach Westen und durchquerte den gesamten Innenraum. Im Zentrum zweigte eine weitere Straße nach Süden ab. In der Mitte beider Straßen verliefen bis zu 1.2m tiefe und 1.7m breite Wassergräben, die durch einen flachen Überlauf miteinander verbunden waren. Reste von Bleirohren belegen darüber hinaus die Existenz einer Frischwasserleitung. Der bis jetzt einzige, wohl nie fertiggestellte Brunnen lag im Innenhof des Zentralgebäudes 6. Hier befand sich ursprünglich eine Reihe von 5 rechteckigen, 2 mal 3.5m großen Gruben, deren mittlere aufgegeben wurde. An ihre Stelle trat der 3.8m tiefe Brunnen mit mit unregelmäßiger Sohle. Die Trennung von Baugrube und eigentlichem Brunnenschacht gelang nicht. Die kiesige bis lehmige Verfüllung enthielt ebenso wie die benachbarten Gruben eine große Anzahl von Kalkbruchsteinen unterschiedlicher Größe. Außerdem fanden sich einige kleine Bruchstücke einer vergoldeten Reiterstatue sowie mehrere sorgfältig bearbeitete Architekturteile aus Kalkstein.

Im Sommer 2000 fand sich östlich des bisher untersuchten Gebiets die Südwestecke einer weiteren Grabenanlage (Graben C). Es handelt sich um einen steril verfüllten, 1.10m breiten und 0.6m tiefen Spitzgraben, der auf einer Länge von 50m erfaßt wurde. Im Nordosten verlief ein weiteres, flaches Grabenstück mit unregelmäßiger Sohle in 2.5m Abstand parallel zu dem Spitzgraben, das vermutlich der Materialentnahme für die anzunehmende Wallschüttung diente. Der Verlauf der beiden Gräben konnte im Nordosten durch geophysikalische Prospektion bis zu einer 1998 unbeobachtet errichteten landwirtschaftlichen Gerätehalle verfolgt werden.

Innenbebauung Die Bebauung im Innern der Anlage unterscheidet sich sowohl in der Anordnung der Gebäude als auch in den einzelnen Gebäudetypen wesentlich von der Innenbebauung augusteischer Militärlager. In Waldgirmes fand sich eine straßenbegleitende Bebauung mit vorgelagerten Laubengängen und dahinter gelegenen Freiflächen. Erfaßt wurden mindestens 10 Gebäude, die mit einer Ausnahme durch Brand zerstört wurden. Die aufgedeckten Grundrisse legen für die unmittelbar an den Straßen errichteten Gebäude 2, 3, 4, und vermutlich auch für den nur ganz randlich angegrabenen Bau 10 wirtschaftliche Funktionen nahe. Der nördlich der Befundkomplexe 7 bis 9 gelegene Töpferofen T 1 weist für diesen Bereich auf eine ähnliche Interpretation, wobei auf Grund erheblicher Bodenerosion offen bleiben muß, ob die dort angetroffenen Befunde zu kleineren Einzelgebäuden oder zu einem langrechteckigen Bauwerk ähnlich Gebäude 2 zu ergänzen sind.

Umwehrung Die Umwehrung bestand aus einer 3.2m breiten HolzErde-Mauer mit zwei vorgelagerten, zusammen etwa 8m breiten Spitzgräben. Die rechteckigen bis quadratischen Pfosten saßen in einem Abstand von 0.9m jeweils an den Außenkanten der Pfostengräbchen.Untersucht wurden bisher größere Abschnitte im Norden und Südosten, darunter das Osttor. Die Gesamtanlage des Osttores war 11.8m breit und besaß zwei hinter die Holz-Erde-Mauer zurückspringende Seitentürme. Die Breite der Tordurchfahrt betrug zwischen 3.7 und 4m. An Stelle des im Verlauf der Nord-Süd-Achse vermuteten Nordtores 1

Zu den Grabungen bis 1995: von Schnurbein, Wigg & Wigg 1995; A. Wigg 1996a; 1996b 1997; Walter & Wigg 1997; 1999. 2 Becker, Köhler & Rasbach 1996; Becker & Rasbach 1998; Becker, Köhler & Rasbach 1999; Becker & Rasbach 1999; Becker 1999; von Schnurbein 2000a; Rasbach & von Schnurbein 2000: von Schnurbein 2000b; Becker & Rasbach 2000. Leider wurde bei diesem Beitrag durch ein redaktionelles Versehen nicht der aktuelle Grabungsplan abgebildet. Abb. 30 zeigt den Grabungsstand im Dezember 1998! Becker 2000: 601606.

Bei den Gebäuden 1a-1c handelt es sich wohl nicht, wie zunächst angenommen, um ein einziges Bauwerk, sondern um drei in Größe und Grundriß identische Häuser von etwa 12 mal 15.4m Größe. An das Gebäude 1a schloß im Süden 461

Limes XVIII

nach einem Zwischenraum von knapp 4m ein weiteres Pfostengräbchen an. Es könnte sich dabei um die Südostecke einer Hofbegrenzung handeln, der im Süden an Gebäude 1a anschloß. Das Gebäude selbst ist in Größe und Grundriß nahezu identisch mit Gebäude 5, sobald man die dort vorgelagerten Portiken außer Betracht läßt. Bei beiden Gebäuden befand sich der Eingang zwischen zwei rechteckigen, gleich großen Räumen. Er führte in einen großen, annähernd quadratischen Raum im Zentrum des Hauses, der von zwei schmäleren Seitenräumen flankiert wurde. Eine weitere Raumgruppe schloß sich an. Vom Gebäudetyp vergleichbare Grundrisse liegen aus Nimwegen und Oberaden vor. Der westliche Teil des Prätoriums auf dem Kops Plateau (van Enckevort 1995: 49 Abb. 13) zeigt eine ähnliche, im einzelnen jedoch komplizierter gestaltete Aufteilung. Gleiches gilt für ein Atriumhaus mit angeschlossenenm Peristylhof südlich der Principia von Oberaden.3 Beide Bauwerke werden als Unterkünfte hochrangiger Offiziere gedeutet. Die vergleichsweise geringe Größe und einfache Raumaufteilung sowie das Fehlen einer auf der Hofinnenseite gelegenen Portikus widersprechen meines Erachtens einer solchen Deutung für Waldgirmes. Hier wäre eher an einfache Wohnhäuser mit vielleicht überdachtem Atrium und nach Süden anschließenden Wirtschaftshöfen oder Nutzgärten zu denken.

zerstört. An der gegenüberliegenden, westlichen Straßenseite fand sich eine Reihe von fünf Pfosten, die den Laubengang eines weiteren Gebäudes (10) bildeten. Der im Sommer 2000 östlich vor der Umwehrung aufgedeckte, aus 4 Einzelpfosten bestehnde Bau 12 gehört seiner Ausrichtung nach ebenfalls zur römischen Bebauung. Das Forum Im Zentrum der Siedlung wurden zwei Gebäude nachgewiesen. Das ältere, 12 mal 15 Meter große Atriumhaus 5 wurde niedergelegt und durch das 2200 Quadratmeter große Zentralgebäude 6 ersetzt. Die Fundamente des insgesamt 54 mal 45m großen Bauwerks bestanden aus 40 bis 45 cm breiten Steinmauern, die teilweise bis zu drei Lagen hoch erhalten waren. Partiell waren Reste der ursprünglichen Vermörtelung nachweisbar. Auf diesen Sockelmauern war das Gebäude in Fachwerk errichtet. Vergleichbare Bauweisen konnten bislang im augusteischen Germanien nicht beobachtet werden, hier sind ausschließlich Holz- bzw. reine Fachwerkbauten bekannt. Drei je 6m breite Gebäudeflügel umgaben im Osten, Süden und Westen einen 32 mal 24m großen Innenhof. In der Außenwand des Ostflügels saßen in regelmäßigen Abständen von 3.8m etwa 30 mal 40cms große Pfosten, deren Standspuren noch bis zu 1m tief erhalten waren. Das Steinfundament überlagerte die Pfostengruben und sparte nur den Raum für die Pfosten selbst aus. Die Mauer war im Süden auf Grund der Bodenerosion wesentlich schlechter erhalten als im Norden. Dort konnten auf ihrer Oberseite längliche Streifen aus Holzkohlepartikeln beobachtet werden, die als Reste aufliegender Schwellbalken anzusehen sind. Gleichartige Holzkohlespuren fanden sich auch auf der 24m langen Fundamentmauer der Innenwand des Ostflügels. Am Südende dieser Mauer zweigte ein noch 2m langes Stück der Innenwand des Südflügels rechtwinklig nach Westen ab. Ansonsten waren vom Südflügel nur vereinzelte Steinreste nachweisbar, die sich exakt in den anzunehmenden Mauerverlauf einfügten. Mit der 1999 gezielt angelegten Sondage wurde die oben geschilderte Bauweise auch für den Westflügel belegt.

Das vollständig aufgedeckte Gebäude 2 war 60m lang und 12m breit. Es besaß an beiden Straßenseiten eine vorgelagerte Portikus. Teilweise waren die hinter den Portiken gelegenen Räume zur Straße hin offen. Im Westen nimmt die Raumaufteilung deutlich auf den dort angelegten Töpferofen T 2 Rücksicht. Eindeutige Wohnräume waren im Grundriß von Gebäude 2 nicht zu identifizieren, auf Grund der Fundamentierung erscheint die Annahme eines weiteren Stockwerks jedoch nicht ausgeschlossen. Das aus Einzelpfosten bestehende Gebäude 3 lag gleichfalls an der ost-west-verlaufenden Straße. Weiter östlich gelegenen Pfostenspuren waren zunächst als eigenständiges Gebäude interpretiert worden. Ein im Jahr 2000 in diesem Bereich angelegtes zweites Planum erbrachte jedoch weitere Befunde, so daß die Pfostengruben und -standspuren in diesem Areal nun zu einem einzigen, 12.4 mal 25.6 – 27.2m großen Bauwerk zusammengefaßt werden.

Im Norden wurde der Innenhof durch eine 45 mal 12m große Halle abgeschlossen. In ihrer Längsachse verlief eine Reihe von ursprünglich 10 Pfosten. 4 Pfosten saßen in Pfostengruben. Sie hatten eine Kantenlänge von 50cms und waren als Standspur noch bis zu 1.92m tief erhalten. Zwischen den eingetieften Pfosten fanden sich 50 mal 50cms große Steinsockel für die übrigen Pfosten. Drei dieser Fundamentierungen konnten im Ost- und Mittelteil der Halle nachgewiesen werden, so daß sich der Takt der restlichen sicher erschließen läßt. An der Nordseite lagen drei Annexbauten, wobei der mittlere, 10 mal 10m große Saal von zwei weiteren, je 6m breiten Räumen mit halbrundem Abschluß flankiert wurde. Die Anlage der beiden äußeren Apsiden war anscheinend zunächst anders geplant. Hinweise darauf gab ein braungrau schluffig und steril verfülltes Gräbchen, das an der Nordostseite der Querhalle in deren Fundamentgräbchen einband und

Weitere Bebauungsspuren fanden sich beiderseits der Nord-Südverlaufenden Straße. Gebäude 4 war 12 Meter breit und nur durch einen knapp drei Meter breiten Korridor von Bau 2 getrennt. Es besaß ebenfalls zur Straße hin offene Räume, jedoch keinen vorgelagerten Laubengang. Der Südteil ist durch moderne Überbauung 3

Kühlborn 1995: 112, Abb. 7. Auch die Gebäude 7a-7g in Haltern sind hinsichtlich ihrer Größe (durchschnittlich 15 mal 30m) und der Aufteilung in einen Wohn und einen Hofteil vergleichbar. Die Gliederung des Wohnteils ist insgesamt jedoch weniger einheitlich und seine Erschließung erfolgt zumindest teilweise über Korridore. von Schnurbein 1981: 51-53.

462

Armin Becker: Die Ausgrabungen in Lahnau-Waldgirmes bogenförmig nach Westen verlief. Auf Grund seiner Verfüllungsmerkmale ist dieses Gräbchen mit den Fundamentgräbchen des Gebäudes als gleichzeitig anzusehen. Die Raumtiefe dieses nicht ausgeführten Anbaus betrug mit 3m jedoch nur die Hälfte des westlich davon errichteten Gebäudeteils.

Literatur Balty J.-Ch. 1991 Curia ordinis. Recherches d’architecture et d’urbanisme antiques sur les curies provinciales du monde romain. Mém. Acad. roy. Belg., Cl. Beaux-Arts, 2e sér., XV.2. (Brüssel). Becker A. 1998 Zur Logistik der augusteischen Germanienfeldzüge. In P. Kneissl & V. Losemann (Hrsg.) Imperium Romanum. Festschrift Karl Christ. (Stuttgart): 41-50. Becker A. 1999 Die Ausgrabungen in Lahnau-Waldgirmes 1999. Eine colonia nova aus der Zeit des Kaisers Augustus im Lahntal? Denkmalpfl. & Kulturgeschichte 2: 60-65. Becker A. 2000 Eine römische Stadt an der Lahn? Die Ausgrabungen in Lahnau-Waldgirmes. Antike Welt 31/6: 601-606. Becker A., Köhler H.-J. & Rasbach G. 1996 Das spätaugusteische Militärlager in Lahnau-Waldgirmes. Denkmalpfl. Hessen 1 & 2: 27-29. Becker A., Köhler H.-J. & Rasbach G. 1999 Der römische Stützpunkt von Waldgirmes. Die Ausgrabungen in der spätaugusteischen Anlage in Lahnau-Waldgirmes, Lahn-Dill-Kreis. Arch. Denkm. Hessen 148. (Wiesbaden). Becker A. & Rasbach G. 1998 Der spätaugusteische Stützpunkt Lahnau-Waldgirmes. Vorbericht über die Ausgrabungen 1996-1997. Germania 76: 673-692. Becker A. & Rasbach G. 1999 Die Ausgrabungen im spätaugusteischen Militärlager Lahnau-Waldgirmes in den Jahren 1996-1997. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 349-354. Becker A. & Rasbach G. 2000 Neue Forschungs-ergebnisse der Grabungen in Lahnau-Waldgirmes in Hessen. In L. Wamser (Hrsg.) Die Römer zwischen Alpen und Nordmeer. Zivilisatorisches Erbe einer europäischen Militärmacht. Katalog-Handbuch zur Landesausstellung des Freistaates Bayern, Rosenheim 2000. Schriftenreihe der Archäologischen Staatssammlung 1. (Mainz): 38-40. Bridel P. 1994 Le programme architectural du forum de Nyon (Colonia Julia Equestris) et les étapes de son développement. In La ciudad en el mundo romano. Actes XIV Congrés Internacional d’Arquelogia Clasica. (Tarragona): 137-151. van Enckevort H. 1995 Das Lager auf dem Kops Plateau. In J.-S. Kühlborn Germaniam pacavi - Germanien habe ich befriedet. Archäologische Stätten augusteischer Okkupation (Münster): 42-58. Drack W. & Fellmann R. 1988 Die Römer in der Schweiz. (Stuttgart). Goetz H.-W. & K.-W. Welwei K.-W. (Hrsg.) 1995 Altes Germanien. Ausgewählte Quellen zur deutschen Geschichte des Mittelalters, Bd. 1 a. (Darmstadt). Gros P. 1996 L’Architecture Romaine au début du IIIe siècle av. J.-C. à la fin du Haut-Empire. Bd. 1: Les monuments publics. (Paris). Kühlborn J.-S. 1995 Das Römerlager Oberaden. In J.-S. Kühlborn Germaniam pacavi - Germanien habe ich befriedet. Archäologische Stätten augusteischer Okkupation. (Münster): 103-124. Lehmann-Hartleben K. 1929 Städtebau. RE III A2: Cols. 20172124. (Stuttgart).

Die Annexbauten sind für die Interpretation des gesamten Baukomplexes von zentraler Bedeutung. Sie verbinden das Zentralgebäude von Waldgirmes mit Forumsanlagen der ausgehenden Republik und der frühen Kaiserzeit. Gemeinsame Kennzeichen dieser Gruppe sind Annexbauten an der Basilika und das Fehlen eines Tempels auf dem eigentlichen Forumsareal (NünnerichAsmus 1994: 74-77; 1999/2000: 40-66; vgl. LehmannHartleben 1929: 2117; Balty 1991: 256-357; Gros 1996: 225f.). Auch für die zweischiffige Halle finden sich unter den zeitgleichen Forumsbasiliken Vergleiche.4 Zusammenfassung Die römischen Truppen fanden in den Gebieten östlich des Rheins völlig andere Bedingungen vor als in Gallien. Politische, administrative und wirtschaftliche Zentren auf die sich die römische Herrschaft stützen konnte existierten bis auf ganz wenige Ausnahmen nicht mehr. Statt dessen herrschte in Germanien eine bäuerliche Subsistenzwirschaft vor, die kaum ausgereicht haben dürfte, um die Versorgung der Legionen aus dem Lande selbst sicher zu stellen. Wollte man das eroberte Germanien auch tatsächlich beherrschen, war der Aufbau einer eigenen Infrastruktur unumgänglich (Becker 1998). Anzeichen dafür finden sich während der Statthalterschaft des Varus. An relevanten Schriftquellen stehen für diese Problematik nur wenige Sätze des griechischen Historikers Cassius Dio und eine kurze Notiz in den Annalen des Tacitus (Anns. I.59.6) zur Verfügung. Insbesondere die Interpretation der Dio-Stelle wird kontrovers diskutiert: „Ihre Truppen überwinterten dort [in Germanien] und gründeten Städte, und die Barbaren paßten sich an ihre Ordnung an, gewöhnten sich an Märkte und trafen sich in friedlichen Versammlungen“ (Dio LVI.18.2. Zitiert nach: Goetz & Welwei 1995: 53). Von althistorischer Seite wird die Gründung regelrechter Städte mangels eindeutiger Beweise meist bestritten. Lediglich für das Lager von Haltern konnte S. von Schnurbein auf Grund von Baubefunden eine Verlagerung der Funktion hin zu administrativen und logistischen Aufgaben wahrscheinlich machen (von Schnurbein 1981: 89-97; 1989: 100-103; 1999-2000). In Lahnau-Waldgirmes spricht die Anlage eines Forums sowie die übrige Bebauung zumindest für den Versuch einer Stadtgründung. Nur die möglichst vollständige Ausgrabung des Platzes wird zeigen, wie weit dieser Versuch im Jahre 9 n.Chr. bereits gediehen war.

4

Erwähnt seien die Basiliken von Zuglio, Glanum, Nyon, Conimbriga, Valeria und Lausanne. Nünnerich-Asmus 1994: 106, 217f., 245; Roth 1987; Bridel 1994: 140-145; Drack & Fellmann 1988: 109 Anm. 7; Balty 1991: 389f.

463

Limes XVIII

Nünnerich-Asmus A. 1994 Basilika und Portikus. Die Architektur der Säulenhallen als Ausdruck der gewandelten Urbanität in später Republik und früher Kaiserzeit. (Köln). Nünnerich-Asmus A 1999/2000 Leitmuster städtischer Architektur in den nordwestlichen Provinzen - zu den Mechanismen römischer Urbanisierungspolitik. In Jahrbuch des Heimat- und Altertumsverein Heidenheim an der Brenz 8: 40-66. Rasbach G. & von Schnurbein S. 2000 Was hat Augustus in Germanien erreicht? In Deutsches Archäologisches Institut (Hrsg.) Archäologische Entdeckungen. Die Forschungen des Deutschen Archäologischen Institusts im 20. Jahrhundert. Antike Welt, Sonderbd. 1999, Bd. 1. (Mainz): 39-41. Roth Congés A. 1987 L’hypothese d’ une basilique a deux nefs à Conimbriga et les transformations du forum. Mél. École Francaise Rome 99: 711-751. von Schnurbein S. 1981 Untersuchungen zur Geschichte der römischen Militärlager an der Lippe. Berichte der Römische Germanischen Kommission 62: 5-101. von Schnurbein S. 1989 Die Archäologie der augusteischen Feldzüge nach Germanien. Altertum 35: 95-103. von Schnurbein S. 1999/2000 Polis und Agora in Germanien unter Varus. In Jahrbuch des Heimat- und Altertumsverein Heidenheim an der Brenz 8: 96-104. von Schnurbein S. 2000a Die Zeit um Christi Geburt. Archäologie im Deutschland 1: 26-29. von Schnurbein S. 2000b Die augusteischen Stützpunkte in Mainfranken und Hessen. In L. Wamser (Hrsg.) Die Römer zwischen Alpen und Nordmeer. Zivilisatorisches Erbe einer europäischen Militärmacht. KatalogHandbuch zur Landesausstellung des Freistaates Bayern, Rosenheim 2000. Schriftenreihe der Archäologischen Staatssammlung 1. (Mainz): 34-37. von Schnurbein S., Wigg A. & Wigg D.G. 1995 Ein spätaugusteisches Militärlager in Lahnau-Waldgirmes (Hessen). Bericht über die Grabungen 1993-1994. Germania 73: 337-367. Walter D. & Wigg A. 1997 Ein Töpferofen im augusteischen Militärlager Lahnau-Waldgirmes (Hessen). Germania 75: 285-297. Wigg A. 1996a Roman and native in the Lahn Valley between Wetzlar and Gießen. In J. Metzler u.a. (Hrsg.) Integration in the early Roman west. The role of culture and ideology. Doss. Arch. Mus. Nat. Hist. et Art 4. (Luxembourg): 169-181. Wigg A. 1996b Zum Stand der archäologischen Untersuchungen im frührömischen Militärlager Lahnau-Waldgirmes an der mittleren Lahn. Mitt. Oberhess. Geschver. Gießen N.F. 81: 147-176. Wigg A. 1997 Im Schatten des Wetteraulimes. Neueste Forschungen im Lahntal. In W. Groenmann-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995: Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 217-233. Wigg A. 1999 Neu entdeckte halternzeitliche Militärlager in Mittelhessen. In W. Schlüter & R. Wiegels (Hrsg.) Rom, Germanien und die Ausgrabungen von Kalkriese. Internationaler Kongress der Universität Osnabrück und des Landschaftsverbandes Osnabrücker Land e.V. vom 2. bis 5. September 1996. (Osnabrück): 419-436.

464

Armin Becker: Die Ausgrabungen in Lahnau-Waldgirmes

Abb. 1. Lahnau-Waldrirmes. Gesamtplan der römischen Befunde. Stand: August 2000.

465

Limes XVIII

466

Timber baths in Augustan and Tiberian fortresses Paul Bidwell The earliest known fortress baths of conventional type at Vindonissa and Exeter are of Claudio-Neronian date. Two earlier fortress baths at Vindonissa are of a type which until recently has seemed to be unique: the Tiberian Holzthermen were built of timber and clay with a lead-lined hot bath and open-air pool; their later Tiberian successor had the same plan and fittings but was built with stone walls. Neither of the baths had hypocausts; the warm rooms were presumably heated with braziers. Despite their unusual construction and heating arrangements, the plans of these baths correspond in all important respects to public baths of the late Republic and early empire. At the Augustan military sites of Mušov and Marktbreit are timber buildings which in several respects resemble the early baths at Vindonissa. They have been tentatively identified as fabricae but examination of their plans and internal features strongly suggests that they are baths.

something to say about the earlier Roman development of bath architecture, a field where much of our knowledge depends on sketchily recorded and poorly dated buildings in the Mediterranean regions.

Introduction Excavations in the fortress of Vindonissa by Laur-Belart and Simonett in the 1920s and 1930s recovered the largely complete plans of three baths and fragments of one or possibly two others. Most immediately remarkable are the huge Claudian baths: they represent the earliest complete plan as yet known of baths in a new style, architecturally elaborate and to be repeated on an ever larger scale, which in the cities of the Mediterranean area and sometimes further afield came to supersede the simple layouts best seen at Pompeii.1 The two other baths at Vindonissa with complete plans were of Tiberian date. The earlier of these (the Holzthermen) is in its own way also remarkable, for it was built of timber and clay; until recently it was the only example of baths not built in masonry.

Vindonissa: the early baths east of the via principalis The fortress at Vindonissa was established in c.AD17 and occupied by legio XIII Gemina until c.45. On the east side of the via principalis, almost opposite the site of the Claudian baths, are the remains of two or possibly three earlier baths dating to between c.25 and 45. Their excavation was undertaken by Laur-Belart and Simonett from 1932 to 1935 who were able to recover the greater part of the plans of two of the baths.3 The Holzthermen (Fig. 1A): the only large-scale published plan (Simonett 1936: Abb. 1) shows, superimposed, the remains of the six successive buildings which had occupied the site throughout the life of the fortress and consequently it is difficult to determine whether some of the post-holes which are illustrated belong to the Holzthermen, or to later structures, although the overall plan of the Holzthermen is easy enough to make out. The baths consisted of a block of three rooms measuring 33.7m by 22.5m, the walls of which were formed by timber uprights with a clay infilling. The north-eastern room measured 21.5m by 11m. Its interior had a clay surface cut by slots spaced about a metre apart and running along almost the entire length of the room. The excavation in 1933 uncovered carbonised wood in the slots and also many iron nails still set upright at intervals of between 0.12m and 0.7m (Simonett 1934: 79-80. Abb. 1; cf. 1936: 168). This suggests that the timbers served as joists for a planked floor which had been laid immediately above the clay packing between the joists. The planked floor had been at least partly burnt when the baths were demolished, the remainder decaying in situ. Simonett thought that the floors of the two northern rooms had been covered with lead sheeting (Simonett 1936: 168). This is hardly consistent with the frequent survival of nails in their original positions on the lines of the timber joists, which otherwise would have been ripped out along with the lead

From a technical point of view the excavation of the Holzthermen was difficult, its remains being covered and in places destroyed by two later periods of building. Its publication, however, was by the standards of the time exemplary. The interpretation of the remains was perceptive but confined to identifying the building as baths. Perhaps because the building was of a type otherwise unknown in timber, no attempt was made to analyse its plan and to determine whether it resembled baths of conventional construction which were broadly of the same period. The Holzthermen have remained something of a little-known curiosity, noted in von Petrikovits’ detailed study of legionary fortresses,2 but never mentioned in studies of the development of baths architecture. This paper shows that in the wake of the recent excavations of Augustan fortresses in Germany and elsewhere, the Holzthermen at Vindonissa are no longer an isolated curiosity but find their place in a series of military baths built in timber. These buildings contribute to our understanding of the purpose of Augustan and Tiberian fortresses and of the relationship between civilian and military architecture in those periods. They also have 1

For the early development of a more sophisticated baths architecture than that found in the Pompeian type, see Nielsen 1990: 43-9; Yegül 1992: 130-9. 2 von Petrikovits 1975: 102; their plan was omitted from the figure (Bild 28) otherwise collecting together all the plans of legionary baths no matter how fragmentary.

3

Laur-Belart 1933 described the open-air pool associated with the ummauerte baths; Simonett 1934: 73-88, the western part of the Holzthermen and ummauerte baths; Simonett 1936, the remainder of these buildings.

467

Limes XVIII

sheeting.

accommodated in a separate building but no traces of it were seen. The silt in the drain at the north end of the swimming-pool contained several intaglios lost by bathers.

The floor covering of lead sheeting appears to have been confined to an area at the north end of the room. The slots for joists running down the length of the room terminated 2.5m short of the north wall; the remaining space was occupied by another set of timbers c.0.25m wider and more closely spaced at intervals of c.0.75m. Originally, lead plates measuring l.4m by 0.45m by 2mm had been nailed to the upper surfaces of these timbers, one still surviving in position. Immediately adjacent, outside the building, was found the base of a furnace. These features point to the presence of a bath lined with lead sheets, the water for which must have been heated in a boiler placed over the furnace.

The Holzthermen were certainly built in the Tiberian period but their exact date is uncertain. They do not appear to belong to the first period of the fortress (17-25/30), but rather to the enlargement of the fortress which was undertaken in the third or fourth decade of the C1st (Meyer-Freuler 1989: 38-43, Abb. 19-20). The ummauerte baths (Fig. 1B) measured 38.2m by 27.6m overall and occupied the site of the Holzthermen, which they superseded, although they were aligned east-west rather than north-south as their predecessor. The plan of the baths closely resembled that of the Holzthermen, but the walls, represented only by robber trenches, had been of stone. The interior surfaces of all three rooms were cut by parallel settings for timbers, which as in the Holzthermen will have held joists for planked floors. The slots in the eastern and southern rooms cut through mortar surfaces which were perhaps constructional deposits associated with the stone walls rather than floors.

A masonry apse with external buttresses opened off the east side of the north-east room (Simonett 1936: 170-1, Abb. 11-12). In its final state the interior of the apse had slots for timbers set 0.9m apart and running at 900 to those in the main room. A little to the west of the centre of the apse was a masonry foundation 2m square; it almost certainly supported the base of a shallow granite labrum 2m in diameter found a few metres to the south. A wooden water-pipe (not shown on the published plans) was found running towards the masonry foundation. In plan the apse represented two-thirds rather than half the circumference of a circle. Simonett explained that this was because the apse was a reduced version of a circular laconicum which had been associated with baths earlier than the Holzthermen. When the later baths were built, the west side of the laconicum was demolished, converting it into an apse. It had a circular foundation at its centre, underlying the square foundation associated with the timber settings. On the other hand, there were no remains of baths earlier than the Holzthermen elsewhere on the site; it is conceivable that the masonry structure was built as an apse and might represent an addition to the main room to the west.

The eastern room measured 25m by 11m. The only feature of note was at the north end of the room where the timber settings stopped short at a distance of c.l.5m from the wall. At the north end a wall of brick-mortar about 0.3m wide survived, its position coinciding with the termination of the timber settings; a single narrow step ran along the front of the wall. The wall presumably formed one side of a rather narrow bath, perhaps matched by another at the southern end of the room, most of which lay beyond the limits of the excavation. Both these baths could have been supplied with hot water from furnaces which lay outside the excavated area. Little can be said about the other two rooms. Simonett noted that the mortar surface in the southern room stopped along a rather ragged edge, some 6.75m from the west wall of the room. He suggested that the intervening space was occupied by a bath but if the mortar was associated with the construction work there is no reason why this should be so.

The other two rooms of the Holzthermen, which each measured about 19m by 10m, were almost featureless.4 Both had floor of clay with timber settings spaced about a metre apart as in the room to the north-east. An open-air pool was found 12.5m east of the baths. It measured 15.8m by 11.8m and was 1m deep; its sides were lined with clay and also its base, which had timber settings to which a lining of lead sheets had been secured, as a few scraps still nailed to the surviving oak baulks demonstrated.5 Simonett suggested that the pool was

There was a separate structure measuring 22m by 20m overall at a distance of 3.75m to the west of the main building. The gap between the two buildings could well have been roofed over to create a passage or vestibule. The structure to the west contained on one side a large pool measuring 18m by 6.75m and at least 1.3m deep. It was lined with masonry and on its east side were two flights of steps 0.97m wide, each with five treads. The bottom of the pool had a quarter-round moulding at the junction with the

4

In a small reconstruction of the overall plan Simonett (1936: Abb.7) shows a bath at the north end of the north-western room identical in form to that in the caldarium. The large-scale plan (1936: Abb. 1) shows no trace of it and no evidence for its existence is mentioned in the text. If, as is suggested below, this room was a tepidarium or apodyterium, there would be no need for the presence of a bath. 5 Simonett mentions a similar structure found at Bregenz where there was Tiberian military occupation (Hild 1930: 130-31). It was a tank or pool measuring 8.5m by 5.5m and lined with clay. The bottom was covered

with pine planks set on timber baulks which were imbedded in the clay lining. It was filled with material containing objects datable to the Tiberio-Claudian period. This may well represent another swimming-pool of the type found at Vindonissa; no contemporary structures were found in the immediate vicinity.

468

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses walls and there was an outlet-pipe on the north side which emptied into a plank-lined drain. The walls which surrounded the pool were only about 0.9m wide and could hardly have supported a roof with a span of over 18m. They must have served to enclose a courtyard.

long axis of the caldarium at right angles to those of the tepidarium and apodyterium. The baths at Cales (Calvi Risorta) in Campania, originally built in c.90-70BC but restored in c.AD100, has a caldarium with its long axis at right angles to those of the apodyterium and tepidarium (Fig. 3A; cf. Nielsen 1990: fig. 70); the same arrangement can be seen in the late Augustan South Baths at Conimbriga in Portugal (Fig. 3B) and also in the Tiberian Thermenhaus at Cambodunum (Fig. 3C; cf. Yegül 1992: fig. 89; Kleiss 1962: 13-25, taf. 13-14).

The ummauerte baths were associated with the second period of the fortress, beginning in the third or fourth decade of the C1st, and were demolished in the mid-40s (Meyer-Freuler 1989: 43, Abb. 20). The Holzthermen also belonged to the earlier part of this period, the maximum duration of which was 20 years. Wooden baths will not have been long-lived.

Despite the differences between the arrangement of the rooms in the Vindonissa baths and in the Pompeian type, there is an exact correspondence between the method of bathing they represent and that occurring in many other early baths. The three main rooms in the Pompeian type served as a caldarium, tepidarium and apodyterium. In some of the larger baths, for example the Stabian Baths at Pompeii (Fig. 2), cold bathing took place in an open-air pool in the palaestra; open-air pools are sometimes even found next to domestic baths, as at the House of the Silver Wedding in Pompeii.7 A trend towards the provision of covered cold baths began in the late C1st BC when the circular laconica opening off the apodyteria of the Stabian and Forum Baths at Pompeii were converted into unheated pools. However, frigidaria of the normal later type, which adjoined the tepidarium and included a large, usually rectangular bath, are scarcely known before the mid-C1st AD.

Masonry remains of early baths under the Flavian basilica thermarum should also be noted (Laur-Belart 1928: 1827). They consisted of a circular room 5.4m in diameter contained within one corner of a rectangular building and an adjoining room which measured at least 6m by 4m. Laur-Belart described four structural phases: the original building was demolished to make way for the basilica, but following the latter’s demolition the remains of the bath were dug out and its walls rebuilt, the new building then undergoing further alterations. As a whole this sequence is scarcely credible, but the circular room was cut through by one of the basilica piers and is thus earlier than the Flavian period. It is also likely to be earlier than the Claudian baths to which the basilica had been added, for the walls of the rectangular building of which it formed part were on a slightly different alignment to that of the baths.

The two baths at Vindonissa both had large open-air pools. In each block of three rooms, only one room was equipped with a bath or baths, which in the case of the Holzthermen was certainly heated. They thus correspond to the Pompeian type, the three rooms in each of the baths serving as an apodyterium, tepidarium and caldarium, with emplacements for hot and cold bathing provided respectively by a heated bath or baths in the caldarium and an open-air pool.

Comparisons with other early Roman baths Once their plans were fully revealed, Simonett did not hesitate to identify the buildings at Vindonissa as baths, despite their unorthodox methods of construction which are otherwise unknown in baths architecture, at least until very recently. Decisive finds were the granite labrum, which could only have been associated with bathing, the intaglios from the drain of the Holzthermen pool, loosened by immersion from their ring-settings,6 and the steps down into the later pool which showed that it too was used for bathing. These details sufficed for Simonett, but the overall arrangement of rooms, and indeed their likely means of heating, can be paralleled amongst other early Roman baths.

The translation of a stone building-type into timber construction: technical aspects of the early Vindonissa baths One of the greatest challenges facing the builders of early fortresses was the adaptation of stone building-types such as basilicas and palatial houses, the models for legionary praetoria, for construction in timber. In the Holzthermen the bath in the caldarium was a lead-lined tank, but the room also contained a labrum housed in an apse. In the caldaria of conventional baths labra were often supplied by a pipe running up the inside of the pedestal so that the water bubbled out at the centre of the bowl and flowed over the edge of the labrum and onto the hot floor of the room where its evaporation contributed to the steamy atmosphere. Even if the labrum in the Holzthermen was drained separately, water splashing from it and from the

The larger baths at Pompeii and Herculaneum, for example the Stabian Baths (Fig. 2), are characterised by a loose grouping of three main rooms of approximately equal size arranged with their long axes running parallel. Most of the few baths recorded elsewhere which date to before the mid-C1st AD are generally similar to the Pompeian type, but there are exceptions which have rooms arranged in a pattern similar to that of the two Vindonissa baths, with the 6

The drain of the frigidarium in the fortress baths at Caerleon produced many gemstones: see Zienkiewicz 1986: 118, also arguing that the large numbers of gemstones from the conduit at Bath were not a ritual deposit but were lost by bathers.

7 Yegül 1992; fig. 46. He also notes that ‘frigidaria with pools are rare’ in Pompeian domestic baths.

469

Limes XVIII

bath would have soon made flooring of earth or clay as muddy as a farmyard. A planked floor was thus provided which would nevertheless have been slippery when wet and might have been scattered with sand.8 Planked floors were also provided in the other two rooms of the Holzthermen presumably because they too would have been gradually saturated by dripping bathers entering from the caldarium or open-air pool.

to have been a fabrica because of its resemblance to the Marktbreit building, even though it had no furnace (Fig. 4B; cf. Bálek & Šedo 1996). Neither building produced debris from metal-working or from any other industrial activity. Two points of resemblance between the buildings at Marktbreit and Mušov are immediately apparent: their plans are dominated by blocks of three rooms in which the long axes of two rooms are at right angles to that of the third, as in the Vindonissa baths, and peristyle courtyards are attached to the main blocks. They differ in detail. The interiors of the three main rooms at Marktbreit had at one period parallel slots for floor joists and against the outside of the end wall of the north-east room there was a large furnace, both as at Vindonissa. Neither of these features was present in the Mušov building which, however, might have been unfinished: the excavators certainly considered that the large well outside the north-east corner of the main block, presumably intended as the water supply for the baths, was never completed. The site as a whole produced remarkably few finds apart from burnt daub and nails.

The presence of wooden floors in the ummauerte baths is puzzling because mortar was presumably used in the walls and brick-mortar was certainly employed to form the bath in the caldarium and to line the open-air pool. An opus signinum floor would certainly have been more serviceable than one of planks. For the baths to function properly the tepidarium and caldarium would have required heating. By the Tiberian period hypocausts built of brick or sometimes of other materials were widespread. However, more primitive forms of heating remained in use in the earlier C1st AD. For example, the baths at Aguntum in the eastern Tyrol (Alzinger 1977), on the Magdalensberg (Piccottini 1977), and at Pompeii in Regio VIII, Insula V, nr. 36 (Maiuri 1950), all had comparatively crude stone-built hypocausts of the channel type, which were less efficient than the type which employed pila-bricks or other individual supports. At Pompeii the men’s tepidarium in the Forum Baths had neither a hypocaust nor hollow walls at the time of the eruption in 79 but was heated by a large bronze brazier which still remains in its original position at the southern end of the room (Mau 1904: 204-5). The absence of a hypocaust in the women’s tepidarium of the Forum Baths at Herculaneum, built in the Julio-Claudian period, is particularly remarkable.9 It is clear that in the Tiberian period not all the heated rooms in baths had hypocausts; the use of braziers to heat the Vindonissa baths would not have seemed archaic.

Peristyle courtyards were lacking at the Vindonissa baths, although the pool associated with the ummauerte baths was in an enclosed, probably unroofed space. Elsewhere in baths of this period palaestrae or exercise courtyards often occur: at Glanum (dating to the second half of the C1st BC) a peristyle courtyard separates the baths from the pool (Rolland 1946: 49-65) and in the late Augustan South Baths at Conimbriga there was a courtyard without colonnades separate from the enclosure that contained the pool (Fig. 3B). The palaestrae in the Campanian baths are rather larger than these two examples. If the peristyle courtyards at Marktbreit and Mušov were exercise courtyards, one of the two adjacent rooms in the main blocks would have been the apodyterium where bathers would have entered the baths suite. The room furthest from the courtyard would then probably have been the caldarium. At Marktbreit the furnace was placed outside one of the end walls of the room in this position and thus could have supplied hot water to a bath. This room also had a small square room opening off its southeast side which might have housed a labrum. At Mušov the room furthest from the courtyard had an apse on its northeast side which is the more usual method of accommodating a labrum in baths of this period. Apses were rarely attempted in timber construction (the apse of the Holzthermen at Vindonissa was of stone), but the example at Mušov was a faithful imitation of a masonry exemplar of some sophistication: the apse opening is partly masked by the stubs of the wall it abuts, throwing the interior into partial shadow and thus emphasising the form of the apse.

Other early military baths of timber construction Two Augustan military sites, Marktbreit, north-east of the later limes in Germany, and Mušov, in the Czech Republic and well beyond the Danube, have produced timber buildings with complete plans similar in several telling respects to those of the early Vindonissa baths. Marktbreit, excavated in 1986-7, was tentatively identified as a fabrica, principally because a large furnace was built against one side of the main block of three rooms (Fig. 4A; cf. Wamser 1991). The building at Mušov was also thought 8 Sand was used to cover the surfaces of palaestrae (Vitruvius X.2.12). Such surfaces have been found in the palaestrae at Exeter (Bidwell 1979: 42) and Caerleon (Boon 1972: 81). 9 Maiuri 1958: 93. Some form of open heating, presumably by means of braziers, was used in the baths of Greece and other eastern provinces in the mid-C1st AD. Pliny (NH XVIII, XLIV) describes how the balneatores threw darnel seeds onto the hot coals in order to drive out unwanted clients: ‘aere granum minimum est in cortice … aiuntque in Asia at Graecia balneatores, cum velint turbam pellere carbonibus ad semen inicere’.

The buildings at Marktbreit and Mušov thus have important points of resemblance to the Vindonissa baths or, as in the case of the peristyle courtyards, to other baths of early date. There are one or two loose ends. Neither 470

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses Marktbreit nor Mušov has an open-air pool. Mušov might have been abandoned before the pool was dug, but the Marktbreit building was certainly completed. Perhaps there was no aqueduct at Marktbreit, rendering a pool impractical because of the effort required to fill it. Grids of posts spaced roughly 2.5-3m apart in two of the rooms at Marktbreit also require explanation. In the published account by Wamser they are described as Trocknungsanlage, with raised drying floors connected with the furnace beyond the room identified here as a caldarium. In effect Wamser was describing two hypocausts, which would have heated the tepidarium and caldarium; the third room, not supplied with a grid of posts to support a raised floor, would have been the apodyterium, an unheated room. The timber framework of the hypocaust might have needed a fire-resistant covering such as a coating of clay. But in all likelihood, sooner or later, the timbers would have caught fire. However, if the hypocaust was built of a hardwood with a high ignition temperature such as oak, and the furnace was properly controlled, a protective coating might not have been necessary. At some stage the grids of posts were replaced by parallel joists of the type seen in the Vindonissa baths. In an annexe of the Tiberian fortress at Velsen a grid of post-holes in what appears to have been a shallow subbasement has been interpreted as a hypocaust (Bosman 1999). Its remains consisted of six rows of four post-holes, all spaced c.1.1m apart; the posts had been driven into the ground to depths of up to 0.45m. The harbour of the fortress has produced a number of wooden baths slippers and there have been other finds of unguent flasks in pottery and glass and of metal strigils, a fragmentary example being found in the immediate vicinity of the supposed hypocaust. These objects certainly suggest that there were facilities for bathing at Velsen, but because of post-Roman erosion of the site building plans scarcely survive.

(Fig. 5B; cf. Fingerlin 1970-71: 207-8, Beil. 28). It had no internal divisions and the presence of a furnace outside the room on its long axis recalls the similar furnaces in the ummauerte baths at Vindonissa, at Marktbreit and at Anreppen. This resemblance might be fortuitous, and the excavator’s interpretation of the building as a fabrica correct. But the possibility remains that it represents the caldarium of a set of baths, the remainder of which had been destroyed by later activities. Finally, some comment is needed on the apparent absence of baths at Haltern. This is the most completely excavated of all the Augustan military sites, but even here the plan is not entirely known (von Schnurbein 1974: Beil. 1). The hospital occupies the northern quarter of a large area where no other buildings appear to have been excavated. At Vindonissa the early baths were immediately to the south of the recently identified hospital (Meyer-Freuler 1989: 43, Abb. 20); in later fortresses the baths and hospital were often adjacent (Vetters 1953: 53; Boon 1972: 75 and n.281). At Haltern baths might well have lain in the unexplored area next to the hospital. The significance of Augustan and Tiberian military baths The six Augustan military sites with the best-known internal plans are Haltern, Rödgen, Anreppen, Marktbreit, Dangstetten and Oberaden. Of these Marktbreit, Anreppen and perhaps Dangstetten can now be shown to have included baths amongst their internal buildings; at Haltern a likely site for the baths lay outside the excavated areas, while Rödgen, a supply base, stands apart from the other much larger fortresses; at Oberaden extensive excavations have not encountered baths. The discovery of baths at two or possibly three of these six sites, none of which has yielded a complete plan, suggests that they might have been as essential in Augustan fortresses as in those of later date. This is certainly supported by the presence of baths at Mušov and possibly at Velsen; at both sites these are the only buildings yet recorded.

Anreppen is another Augustan military site with the complete plan of a building which can be identified as baths (Fig. 5A; cf. Kühlborn 1995: 136-8, Abb. 5, Beil. 3). The building consists of four rooms measuring 26.5m by 23.5m adjoined on the south side by a peristyle courtyard 21.5m by 16m. The plan in the interim report is at a small scale, but it seems that two of the rooms had floors raised on timber supports. Hot air from a furnace outside the end of one room was circulated under the raised floors, an arrangement compared by its excavator, Kühlborn, to that at Marktbreit. The eastern of the two rooms next to the peristyle courtyard is identifiable as the tepidarium because of its raised floor and the western room presumably served as the apodyterium. The room to their north which was adjoined by the furnace would have been the caldarium; it was possibly sub-divided at some stage. Just beyond the north-west corner of the enclosure around the baths is what appears to be a well, perhaps, as apparently intended at Mušov, the water-supply for the baths.

The only evidence which might indicate for whom the baths were supplied is their size. The areas of their caldaria range from 105sq m at Mušov to 112sq m at Marktbreit (Table 1). This is about a third of the area of the caldarium in the Claudian baths at Vindonissa (Fig. 6A) and half the area of the caldarium in the Neronian baths at Exeter (Fig. 6B), both situated in fortresses built for single legions. The Augustan fortresses accommodated groups of units with a total strength far exceeding that of a single legion; Marktbreit, for example, covers an area of 37ha as opposed to the 22ha of Vindonissa. The baths at Marktbreit, three times smaller than the Vindonissa baths but in a fortress almost twice as large, could not have been provided for the general use of all those present. But the units accommodated in the larger Augustan fortresses were probably a combination of legionaries and auxiliaries (von Schnurbein 2000: 36-7). It is doubtful whether the latter would have been considered when baths were supplied.

It is also worth noting a rectangular building at Dangstetten measuring 16m by 9.5m with unusually deep post-trenches 471

Limes XVIII

Baths Stabian Baths, Pompeii Glanum Forum Baths, Herculaneum Forum Baths, Pompeii Augustan Baths, Conimbriga Marktbreit Mušov Thermenhaus, Cambodunum Holzthermen, Vindonissa Ummauerte Baths, Vindonissa Claudian Baths, Vindonissa Fortress Baths, Exeter

Date C1st BC C1st BC Augustan?

Area of tepidarium 11m x 6.2m (68.2sq m) 9.75m x 6.5m (63sq m) 12m x 6m (72sq m)

Area of caldarium 13m x 7m (91sq m) 13.8m x 6.8m (94 sq m) 12m x 7m (84sq m)

Augustan Augustan

10.7m x 5.6m (60sq m) 9m x 7m (68sq m)

13m x 5m (65sq m) 12m x 7m (84 sq. m)

Augustan Augustan Tiberian

8m x 7m (56sq m) 14m x 7m (98sq m) 7.9m x 5.25m (41sq m)

14m x 8m (112sq m) 14m x 7.5m (105sq m) 8.5m x 5.4m (46sq m)

Tiberian

19m x 10m (190sq m)

21.5m x 11m (236sq m)

Tiberian

23m x 13m (299sq m)

25m x 11m (275sq m)

Claudian

24m x 10m (240sq m)

25m x 12m (300sq m)

Neronian

?

22.3m x 9.75m (217sq m)

Table 1. Comparison of the size of tepidaria and caldaria in baths dating from the C1st BC to the Neronian period. Overall comparisons of the size of baths are difficult because in many cases their full extent, including palaestrae, is uncertain. Dimensions are of the main body of rooms excluding apses and recesses. Note that in the case of the Stabian and Forum Baths at Pompeii and the Forum Baths at Herculaneum there were double suites for men and women; the dimensions of rooms in the larger men’s suites are given.

Baths were not built at auxiliary forts until the end of the C1st AD (Atkinson 1942: 336). Presumably it was not until then that the auxiliaries, for the most part recruited from the newer, less Romanised provinces, came to regard bathing as an essential part of everyday life. Also, comparison of the sizes of early and later C1st baths can be misleading when trying to estimate the numbers of bathers for which they were devised. The development of luxurious thermae meant that bathers could now spend more time in baths which were intended for long and selfindulgent visits.

Schnurbein 2000: 35-6). The addition of baths to the plan of these fortresses completes the complement of buildings regarded as essential in later fortresses. Senior officers were given peristyle houses where much of the usual routine of a wealthy Roman household could be replicated; legionaries, it seems, could use the baths, one of the amenities of ordinary urban life. Förtsch has seen resemblances between early fortresses and Republican coloniae marittimae such as Ostia and Minturnae (Förtsch 1995: 618). He also cites a passage from Livy (XLIV.39.25) which describes a fortified camp as a ‘patria altera’ for the soldiers and their tents a ‘domus ac penates’. His argument is that fortresses mirrored many aspects of urban life, an argument which is certainly reinforced by the inclusion of baths amongst their buildings. Recruitment to the legions in the Augustan period was predominantly from Italy and the older provinces where public baths were widespread (for the growth of the bathing habit in Italy, see Fagan 1999: 40-74, 350-56).

With these two factors in mind it is possible to regard the Augustan fortress baths as adequate for the use of all the legionaries rather than a privileged group such as the senior officers. The Tiberian baths at Vindonissa, where the caldaria had floor areas of 236sq m (Holzthermen) and 275sq m (ummauerte baths), must have been intended for the use of the entire legion; they were larger than the fortress baths at Exeter where the caldarium had an area of 217sq m.

The military baths built of timber and clay are a unique building type; the ummauerte baths at Vindonissa, with stone walls but with timber floors and no hypocausts are clearly a continuation of the type.10 No examples are known from civilian contexts, although Martial refers to wooden baths in one of his epigrams (IX.75). The subject was Tucca who built a balneum ‘with the wreckage of

Despite the temporary nature of the Augustan fortresses the plan of Haltern, which is the most fully known, included most of the main building types that are found in later, permanent fortresses: principia, praetorium, tribunes’ houses and hospital, in addition to barracks. Apart from hospitals, these buildings are also known at the other less extensively excavated Augustan fortresses, even though at Marktbreit, Oberaden and Dangstetten the men might still have been accommodated in tents (von

10 At some auxiliary forts of the C2nd and C3rd in Britain and Germany there were timber apodyteria or exercise halls (Baatz 1973; to his list can be added Bearsden on the Antonine Wall, Breeze 1984: 48-55). These are not comparable to wooden baths: their main suites were built in masonry.

472

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses woods and a framework of pine’ and also thermae ‘of every marble that Carystus discovered’. Tucca is told by the poet to ‘put the balneum under the thermae’, in other words to use the wooden baths as fuel for the thermae.11 This casual reference to wooden baths built by a private individual implies that in the later C1st AD, when Martial was writing, they were still current in Italy. Whether Tucca’s wooden baths were a long-established building type, as yet unknown to archaeology, that had served as the model for those in Augustan and Tiberian fortresses is questionable. They might have been little more than a small wash-house where hot water and a sweating room were available for the poor.12 That might well be deduced from Martial’s epigram, where he contrasts the shoddiness of the wooden balneum with the sumptuous decoration of the thermae. It is also doubtful whether wooden hypocausts existed in civilian baths. They were perhaps no more than a technological curiosity, only brought into being when the specifications of Roman baths in masonry were adapted for timber construction.

in Raetia. The dominance of the Pompeian type in the late Republican and early Imperial period now seems less complete. By the end of the Tiberian period communal bathing had become a well-established part of fortress life. But wooden baths must have offered little opportunity for the physical indulgence possible in contemporary baths of conventional type. Braziers probably could not have achieved high temperatures in the heated rooms and wooden hypocausts were unlikely to have been very effective. Nothing is known of the decoration or lighting of wooden baths but it would have been at best utilitarian, a dreary contrast to the growing ostentation of public baths. When the fortress at Vindonissa was rebuilt in c.45-6 the ummauerte baths were demolished and replaced on a different site by thermae built in stone.14 The new building (Fig. 6A) provided the greatest possible contrast to its predecessor. Of far greater size, its elaborate architecture was complemented by decoration with mosaics, wall-paintings and Jura marble. This sudden appearance in a frontier zone of thermae of the most advanced design coincided with the widespread beginning of stone construction in the fortresses of the north-western provinces. Only in Britain was stone construction delayed until later in the C1st. There it began in the Flavian period and only became general in the Trajanic period: the turf and timber defences of Caerleon, York and Chester, for example, were not rebuilt in stone until several decades after the foundation of these fortresses, although at Inchtuthil work had begun on a stone curtain wall when the fortress was abandoned after only a few years. Exeter and Usk are exceptions, although the Neronian baths apparently were the only stone buildings in those fortresses.15 This suggests that the appearance of thermae in fortresses was not necessarily a result of a general trend towards construction in stone. They represent a continuation of the idea that fortresses should reflect aspects of urban life together with an acceptance that the timber baths of earlier fortresses were too primitive to meet the expectations of modern bathers. The building of the Exeter and Usk baths might have represented the completion of all the essential components of the fortresses rather than the beginning of their comprehensive rebuilding in stone.

The origins of building-types in Augustan and Tiberian fortresses appear to have depended on their functions. Some were sui generis: the plans of barracks were based on tent lines in temporary camps and valetudinaria were an exclusively military building-type.13 The relationship between principia and civil fora is complex: the Augustan principia were probably derived from the plans of contemporary fora, but it is possible that some later fora, especially in the north-western provinces, were in turn inspired by legionary principia (von Petrikovits 1975: 1404). The praetoria reflected contemporary ‘Luxusarchitektur’, providing for an aristocratic life-style little different in its essentials whether in camp or town (Förtsch 1995). The plans of tribunes’ houses were likewise based on Mediterranean peristyle houses. It has been shown that almost every detail of the Augustan and Tiberian fortress baths can be paralleled in civilian baths of conventional masonry construction. They can thus be numbered amongst the building-types in fortresses which were adopted directly from urban architecture. Of great interest is the correspondence of their plans not with the Pompeian type with its loosely-arranged series of parallel rooms, but with a less common type where the long axes of two rooms are at 900 to that of a third room (see above). The fortress baths add another five examples to the three civilian baths of this type at Cales in Campania, at Conimbriga in Portugal and at Cambodunum

The wider architectural significance of the baths at Vindonissa, Exeter and Caerleon has been much debated. Their excavators have seen in their axially-symmetrical plans a clear link with the Imperial type (Laur-Belart 1931: 29; Bidwell 1979: 43-50; Zienkiewicz 1986: 115-23; see

11 Epig. IX.75. The translation of Martial’s epigram and the determination of his meaning follow Fagan (1999: 16). 12 At the Claudian fort of Hod Hill in Dorset, which is thought to have accommodated legionaries and auxiliaries, a combined latrine and ablution block was identified by the southern defences (Richmond 1968: 86-7, fig. 46 B). In addition to the latrine there were two rooms each measuring 6.6m by 3m that formed the supposed ablutions block. If their identification (which depends entirely on their proximity to the latrine) is correct, they might have contained some basic adjuncts of bathing such as braziers for heating and a boiler for hot water. 13 For the background to the development of valetudinaria, see Künzl 1991: 197-202.

14 Laur-Belart 1928; 1930; 1931. For earlier work on the site, see Hauser 1901; 1904. 15 For Usk, see Manning & Scott 1989: 134, fig. 59; only the corner of a plunge-bath has been seen. At Exeter the baths were built on a reserved site in c.60-65, some 10 years after the foundation of the fortress (Bidwell 1979: 56-7). The only other stone construction known in the fortress is the aqueduct, the main purpose of which would have been to supply the baths (Henderson 1988: 100-1). That the fortress remained in occupation until c.75 (op. cit., 15-17; Holbrook & Bidwell 1991: 6-8) is a view that has sometimes been challenged in general reviews of legionary dispositions in Britain.

473

Limes XVIII

also comments by Schwartz 1969: 66 on the Flavian baths at Avenches). This has not been accepted in recent general surveys of baths architecture (DeLaine 1988: 13-4). Yegül commented that ‘this hypothesis only gains strength if it can be shown that the architects of the ambitious imperial projects of Rome were the same ones who travelled with the legions in the provinces’ (Yegül 1992: 75). This paper cannot discuss the development of the imperial type, but it has shown something of the relationship between military and urban architecture in the early imperial period. From the reign of Augustus onwards fortresses provided some of the amenities of urban life, not least the baths. It is quite credible that such buildings were designed by architects familiar with metropolitan architecture, the forerunners of Apollodorus of Damascus who is known alike for his bridge over the Danube, a military work built for the Dacian campaigns of Trajan, and for Trajan’s Forum in Rome.16

provisional interpretation. In J. DeLaine J. & D.E. Johnston (edd.) Roman baths and bathing. Journal of Roman Archaeology Supplementary Series 37: 245250. Breeze D.J. 1984 The Roman fort on the Antonine Wall at Bearsden. In D.J. Breeze (ed.) Studies in Scottish antiquity presented to Stewart Cruden. (Edinburgh): 32-68. DeLaine J. 1988 Recent research on Roman baths. Journal of Roman Archaeology 1: 11-32. Fagan G. G. 1999 Bathing in public in the Roman world. (Ann Arbor). Fingerlin G. 1970-1 Dangstetten, ein augusteischen Legionslager am Hochrhein: Vorbericht über die Grabungen 19671969. Bericht der Römisch-Germanischen Kommission 51-52: 197-232. Förtsch R. 1995 Villa und praetorium. Zur Luxusarchitektur in frühkaiserzeitlichen Legionslagern. Kölner Jahrbuch 28: 617-634. Hauser O. 1901 Die Arbeiten der antiquarischen Gesellschaft von Brugg im Jahre 1900. Anzeiger für Schweizerische Altertumskunde 3: 31-35. Hauser O. 1904 Vindonissa, das Standquartier römischer Legionen. (Zürich). Henderson C. 1988 Exeter (Isca Dumnoniorum). In G. Webster (ed.) Fortress into city, the consolidation of Roman Britain, first century AD. (London): 91-119. Hild A. 1930 Archäologische Forschungen in Bregenz. Jahreshefte des Österreichischen Archäologischen Instituts. 26: 155-176. Holbrook N. & Bidwell P. 1991 Roman finds from Exeter. Exeter Archaeological Reports 4. (Exeter). Kleiss W. 1962 Die öffentlichen Bauten von Cambodunum: Baubeschreibung und Rekonstruction. Materialhefte zur Bayerischen Vorgeschichte, Heft 18. Kühlborn J.-S. 1995 Das Römerlager bei Anreppen. In J.-S. Kühlborn (ed.) Germaniam Pacavi – Germanien habe ich befriedet: Archäologische Stätten augusteischer Okkupation. (Rothenburg): 130-144. Künzl E. 1991 Die medizinische Versorgung der römischen Armee zur Zeit des Kaisers Augustus und die Reaktion der Römer auf die Situation bei den Kelten und Germanen. In B. Trier (ed.) Die römische Okkupation nördlich der Alpen zur Zeit des Augustus, Kolloquium Bergkamen 1989 Vorträge. (Münster): 185-202. Laur-Belart R. 1928 Grabungen der Gesellschaft Pro Vindonissa in den Jahren 1926/27. Anzeiger für Schweizerische Altertumskunde 30: 18-27. Laur-Belart R. 1930 Grabungen der Gesellschaft Pro Vindonissa im Jahre 1929. Anzeiger für Schweizerische Altertumskunde 32: 65-89. Laur-Belart R. 1931 Grabungen der Gesellschaft Pro Vindonissa im Jahre 1930. Anzeiger für Schweizerische Altertumskunde 33: 203-236. Laur-Belart R. 1933 Grabungen der Gesellschaft Pro Vindonissa im Jahre 1932: An der Via Principalis (K.-P. 1446). Anzeiger für Schweizerische Altertumskunde 35: 1-16. Maiuri A. 1950 Scoperta di un edificio termale nella Regio VIII, Insula 5, nr 36. Notizie degli Scavi di Antichita 66: 116-36. Maiuri A. 1958 Ercolano, I Nuovi Scavi. (Rome). Manning W.H. & Scott I. 1989 Report on the excavations at Usk 1965-1976: The fortress excavations 1972-1974. (Cardiff). Mau A. 1904 Pompeii, its life and art (2nd edit. transl. F.W.Kelsey, London).

Acknowledgements I am grateful to Dr N. Hodgson for commenting on a draft of this text. The illustrations were prepared by David Whitworth. Bibliography Alzinger W. 1977 Das Municipium Claudium Aguntum: vom keltischen Oppidum zu frühchristlichen Bischofssitz. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der römischen Welt II.6 (Berlin): 380-413. Atkinson D. 1942 Report on the excavations at Wroxeter (the Roman city of Viroconium) in the County of Salop, 1923-1927. (Oxford). Baatz D. 1973 Römische Bäder mit hölzernen Apodyterien. Archäologisches Korrespondenzblatt. 3: 345-350. Bálek M. & Šedo O. 1996 Das frühkaiserzeitliche Lager bei Mušov – Zeugnis eines augusteischen Feldzugs ins Marchgebiet? Germania 74: 399-414. Bidwell P.T. 1979 The legionary bath-house and basilica and forum at Exeter. Exeter Archaeological Report 1. (Exeter). Boon G. C. 1972 Isca, the Roman legionary fortress at Caerleon, Mon.. (Cardiff). Bosman A.V.A.J. 1999 Possible baths at the fort of Velsen I: a 16 Fortress baths have often been excluded from general studies of bath architecture. Yegül’s comprehensive survey of Roman baths largely ignored examples in fortresses because ‘they represent a special category of investigation’ and lay beyond the Mediterranean areas which were the focus of his interest (1992: viii). For similar reasons other fortress building-types have also not featured in wider studies of Roman architecture. Förtsch’s study of the praetoria in the Augustan fortresses of Germany (1995) illustrates how their plans can contribute to the understanding of wider architectural trends. It has long been appreciated that praetoria were based on Italian prototypes, but various novelties in the plans of the Augustan examples suggested to Förtsch that they represented a stage in the development of ‘Luxusarchitektur’ not to be seen in the existing sample of Italian buildings but known in other peripheral areas such as Palestine.

474

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses Meyer-Freuler C. 1989 Das Praetorium und die Basilika von Vindonissa. Veröffentlichungen der Gesellschaft Pro Vindonissa 9. Nielsen I. 1990 Thermae et Balnea: The architecture and cultural history of Roman public baths. (Aarhus). von Petrikovits H. 1975 Die Innenbauten römischer Legionslager während der Prinzipatszeit. Abhandlungen der Rheinisch-Westfälischen Akademie der Wissenschaften, 56. Piccottini G. 1977 Die Stadt auf dem Magdalensberg: ein spätkeltisches und frührömisches Zentrum in südlichen Noricum. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der römischen Welt II.6 (Berlin): 263301. Richmond I. 1968 Hod Hill, Vol. 2: Excavations carried out between 1951 and 1958 for the Trustees of the British Museum. (London). Rolland H. 1946 Fouilles de Glanum, Saint-Rémy-deProvence. (Paris). von Schnurbein S. 1974 Die römischen Militäranlagen bei Haltern. (Münster). von Schnurbein S. 2000 The organization of the fortresses in Augustan Germany. In R. Brewer (ed.) Roman fortresses and their legions. (London): 29-39. Schwarz G.T. 1969 Die flavischen Thermen “En Perruet” in Aventicum. Bulletin de l’Association Pro Aventico 20: 59-68. Simonett C. 1934 Grabungen der Gesellscaft Pro Vindonissa: a) Auf der Breite (K.-P. 1446). Anzeiger für Schweizerische Altertumskunde 36: 73-97. Simonett C. 1936 Grabungen der Gesellschaft Pro Vindonissa in den Jahren 1934 und 1935 auf der Breite (K.-P. 1446, 1364 und 1365). Anzeiger für Schweizerische Altertumskunde 38: 161-173. Vetters H. 1953 Das Legionsbad von Lauriacum. Forschungen in Lauriacum 1: 49-53. Wamser L. 1991 Wirtschaftsgebäude. In M. Pietsch, D. Timpe & L. Wamser (edd.) Das augusteische Truppenlager Marktbreit. Bericht der Römische Germanische K ommission 72: 263-324. Yegül F. 1992 Baths and bathing in Classical antiquity. (New York). Zienkiewicz J. D. 1986 The legionary fortress baths at Caerleon: Vol. 1, The buildings; Vol. 2, The finds. (Cardiff).

475

Limes XVIII

Figure 1. A.: Holzthermen at Vindonissa; P = open-air pool (after Simonett 1936, Abb. 1 and 7). B: the ummauerte baths at Vindonissa; P = openair pool (after Simonett 1936, Abb. 1 and 4). Scale 1:500.

476

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses

Figure 2. The Stabian Baths, Pompeii (after Nielsen 1990, fig. 75). Scale 1:500.

477

Limes XVIII

Figure 3. A: the baths at Cales (Calvi Risorta) in Campania (after Nielsen 1990, fig. 70). B: the south baths at Conimbriga, Portugal (after Nielsen 1990, fig. 120). C: the Thermenhaus at Cambodunum (after Kleiss 1962, Taf. 13). Scale 1:500.

478

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses

Figure 4. A: the baths at Marktbreit (after Wamser 1991, Beil. 2). B: the baths at Mušov (after Bálek and Šedo 1996, Abb. 4). Scale 1:500.

479

Limes XVIII

Figure 5. A: the baths at Anreppen (after Kühlborn 1995b, Beil. 3). B: possible caldarium at Dangstetten (after Fingerlin 1970-71, Beil. 28). Scale 1:500.

480

Figure 6. A: the Claudian fortress baths at Vindonissa (after Laur-Belart 1931, Abb. 2). B: the Neronian fortress baths at Exeter (after Bidwell 1979, fig. 6). Scale 1:500.

Paul Bidwell: Timber baths in Augustan and Tiberian fortresses

481

Limes XVIII

482

Early Roman military activity on the Frisian coast Marjan C. Galestin The small village of Winsum located almost on the coast north of the Dutch province of Friesland, has long been known for a number of early Roman finds discovered in the late C19th. The finds, fragments of Roman amphorae, mortaria and terra sigillata, at the time led the archaeologists to interpret the site as a Roman military site. It was even tentatively identified with the unknown site of Castellum Flevum. The absence of traces pointing to a Roman military encampment and the absence of military items have led archaeologists to doubt the old interpretation as a military post on the northern shore. The site remained on some archaeological maps as a findspot of objects dating to the Tiberian period while on other maps it was completely omitted as an early Roman-period site. In 1997 an excavation was undertaken to cast some light on the problem of Winsum. The excavation (Winsum-Bruggeburen) has brought to light new finds dating to the early Roman period. Not only Roman pottery but also Roman coins and metalwork were discovered. However, neither military items nor traces of a Roman fortification were uncovered.1 The nature of these early Roman finds has made it clear that there must have been relations of some kind between the Roman military force and the local population. A comparison may be made with Bentumersiel on the River Ems (Germany) where similar Roman objects turned up, including military items, but no traces of a military campsite. In this paper the dating of the Roman finds will be discussed, as well as the significance of the presence of these Roman artefacts in this remote area of northern Europe. People on the Frisian coast are known from historical sources to have maintained relations with the Roman army, while corroborating archaeological finds are very few. For this scanty but interesting evidence it is still very difficult to find an interpretation.2

Introduction1

Tiberian period as identified by Schönberger (1985: 334, 426 A2) and by Wiegels (1993: 250-251).

The Dutch province of Friesland is known to archaeologists for its large number of dwelling mounds (Fig. 1). These so-called terpen had been raised artificially against coastal flooding and were used over a long period of time, in the course of which large amounts of archaeological material accumulated within them. When in the C19th and C20th these dwelling mounds were quarried for their fertile soil, all kinds of archaeological finds were recovered that attracted the attention of Dutch archaeologists. Foreign archaeologists too paid attention to these peculiar finds and in 1881 the Italian archaeologist Luigi Pigorini made a visit to Friesland where he noticed structures which he compared to the north Italian palafitte. The German archaeologist Karl Schuchhardt who excavated at Haltern, came to Friesland in 1906 and was accompanied to the small village of Winsum where some very interesting Roman pottery fragments had been discovered (Fig. 2). The pottery fragments turned out to be sherds of Roman amphoras and of terra sigillata (samian ware) and were dated to the period between Haltern and Hofheim. Schuchhardt even suggested that Winsum might have been Castellum Flevum (Boeles 1951: 128-129). From that time onward Winsum was placed on the archaeological map where it still is today. However, no longer as Castellum Flevum, for which Velsen is a stronger candidate, but as a findspot of Roman pottery from the

Excavation at Winsum-Bruggeburen Although most archaeologists do not regard Winsum as a military post, the site was not forgotten and when new housing was planned for this particular site, the Groningen Institute of Archaeology took the opportunity to start an excavation in 1997. The quarried terp called WinsumBruggeburen, had once been one of the largest in Friesland; it was situated south of the small village of Winsum and near the bridge across the canal (the Franekervaart). The new houses were planned on the western side of the former terp (Fig. 3). The aim of the excavation was two-fold (Bos et al 1998). The first question was whether there still were any archaeological remains in this part of the former terp, where the soil had been dug away in the late C19th; more specifically, whether the lowest occupation levels were still there or had been completely removed. The second question was whether there were any traces of a Roman military camp to which the Roman finds of 1906 could be attributed. The excavation lasted several months and 11 trenches were opened. It turned out that all occupation layers had been destroyed. No archaeological strata remained and only a large number of archaeological features, ditches and round or oval traces, were visible in the salt marsh sediment which is the natural soil in this area. Most of these features had a mixed fill, containing objects from the Iron age well into the medieval period, but some of the features could be dated to the late Iron age and early Roman period (Galestin 2000; in press). It was not (yet) possible to distinguish a groundplan and there were

1

The excavation was carried out by the Groningen Institute of Archaeology in co-operation with the Archeologysk Wurkferban of the Fryske Akademy and the municipality of Littenseradeel. The project was managed by dr J.M. Bos. The field director was drs M.J.L.Th. Niekus, while drs T.B. Volkers was the finds administrator. J.H. Zwier and K. Klaassens formed the technical staff. Many students and volunteers participated in the excavation 2 I am grateful to A. Bardet (Groningen) for correcting the English text.

483

Limes XVIII

no traces of the ditches with a V-shaped section that are characteristic of a Roman camp and so the hope of this time finding remains of a Roman encampment at Winsum remained unfulfilled.

are sherds of cups Conspectus Form 22 which in the course of the occupation of Haltern replaced Conspectus Form 14 (Conspectus 1990: 76). A second category of pottery are the amphorae. Different types of amphora could be distinguished, including fragments of Oberaden 83 and of Dressel 20, but also one handle of a Rhodian amphora and a number of fragments of flat-bottomed amphorae. For the type Oberaden 83 an Augustan date seems appropriate, but the type Dressel 20 may be later. A number of fragments of mortaria were found with a vertical rim comparable to the form Rödgen 62A (Simon 1976: 106) which is Augustan in date. Among the fine wares fragments of Pompeian red ware and some small fragments of cups with a roughcast of quartz sand were discovered. Smooth wares consist of fragments of flagons. Some of these are difficult to date; the presence of parallels in graves at Nijmegen dated to the period AD40– 120 (Haalebos 1990: fig. 88.2) and in the latest features at Velsen, dating to the Tiberian period AD15-30 (Bosman 1997: 222) hamper a close dating of these flagons.

The Roman finds The majority of the finds consist of local native pottery but also Roman artefacts were found. For the major part they are early Roman in date, but some date to the C2nd and C3rd AD. As far as could be ascertained, the Roman finds from the later period were not discovered in the same features as the early Roman finds. Roman finds from the C2nd and C3rd are not unusual in the Frisian terpen (Galestin 1992) but early C1st Roman objects are very scarce in the northern part of the Netherlands. These early Roman finds may provide new information on the question whether the Roman army engaged in activities on the coast of Friesland and Groningen and, if so, in what periode the Romans were active. Therefore we will concentrate on these early artefacts, leaving the later Roman finds aside. The C1st Roman artefacts discovered at Winsum consist of late Republican and Augustan coins, fibulae, pottery fragments and other, more precious Roman objects. Among the pottery are different kinds of ceramics such as terra sigillata, amphoras, fine ware, smooth ware, and coarse ware. The date of the pottery is also varied. Among the earliest potsherds are three fragments of Augustan date: two fragments of a cup (Fig. 4) with a narrow hanging lip (Conspectus 14), and one very thin base fragment of a similar cup with part of a stamp, which may be attributed to L.Titius Thyrsus; Fig. 5). These fragments are the earliest terra sigillata discovered at Winsum. Similar cups are found at Haltern, where stamps of L. Thyrsus are known from the same type of cup (v. Schnurbein 1982: 113). The Winsum cup has the distinctive shape of Conspectus Form 14.1.4 which is typical of a workshop at Lyon, France. The date of Form 14 is middle and late Augustan (c.15BC–AD9) but Form 14.1 is the early phase of Form 14 (Conspectus 1990: 76-77). These early Roman cups are also known from Dangstetten, Oberaden and Rödgen (Simon 1976: 69-70, Form 3 Variante B). The form is also present in the Augustan legionary camp at Nijmegen, which may be dated to the period when Drusus was active in 12BC (Haalebos 1991: 102-105, Fig. 6.7), but this type of cup does not appear among the finds from the fortification at Velsen which is dated to the Tiberian period, c.AD15–30 (Bosman 1997: 312-313). These three fragments suggest an Augustan date for Winsum.

A special category are the so-called cork-urns. These are often called cooking pots with an inverted rim, but a corkurn discovered at Nijmegen, containing birds and more specifically breasts of song thrushes, has made it clear that cork-urns also served as containers for preserved foodstuffs (Lauwerier 1993: 18). This short survey shows that the Roman pottery discovered at Winsum covers a period from Augustus to Tiberius, from the middle of the second decennium BC to the first 40 years of the Christian era, and maybe even later. It is also clear that the Roman pottery discovered at Winsum represents a variety of forms. The different types of amphora were meant not only for wine but also for oil and fish sauce. Mortaria were typically used in Roman food preparation and the contents of cork-urns may have been delicacies imported for officers in the Roman legions. The coins discovered at Winsum provide a further indication for the date and function of the excavated site. The coins comprise two denarii, one republican and one of early Augustan date, and six asses among which four ‘altar’ coins minted at Lyon, probably all of them of the first ‘altar’ series. Some of the asses are halved and some have a countermark reading AVG (Augustus), TIB (Tiberius) or CAESR (Caesar). The date of these countermarks is late Augustan or early Tiberian. There are no Tiberian coins, so the coins provide a date in the late Augustan or early Tiberian period (Galestin 2000; in press). The halved coins and the countermarks point in the direction of soldiers who had to make do with halved coins owing to a shortage of small denominations. The countermarks point to army commanders who had these marks punched on the coins before paying the soldiers.

Five fragments of terra sigillata plates with concave vertical rims (Conspectus Form 18.2), were also discovered. This form occurs at Oberaden but the type of rim discovered at Winsum is not dissimilar to types found at Haltern. No chronological conclusions can be drawn from differences in rim form, according to von Schnurbein (1982: 41). This type of plate continues into the Tiberian period and is also among the finds discovered at Velsen (Bosman 1997: 168). Among the other sigillata fragments

Apart from coins and pottery there are also various bronze objects (Galestin in prep.). It is important to note that among the bronze objects not a single piece of military 484

Marjan C. Galestin: Early Roman military activity on the Frisian coast equipment was discovered. Among the fibulae one Aucissa fibula was discovered, which may be linked to military presence. Among the copper-alloy objects are a stopper with red inlay and the lower part of a lantern. A beautiful bronze steelyard weight turned up long after the excavation, when building activities had begun. It is in the form of a boy’s head and it may be dated to the early C1st, probably to the Augustan period (Galestin 1999a). A small stone tablet for preparing ointments is also an object indicating the presence of the Roman army. Similar objects are known from legionary contexts at Haltern and Oberaden, but also from later contexts.

the northern coastal area? The campaigns of Drusus may have been more of a voyage of discovery than of a war of conquest. He made an agreement with the Frisians and they helped him out when he was in trouble. His ships even got stranded when at low tide the water disappeared very fast, a phenomenon unknown to the Romans. This is vividly described by Cassius Dio (Dio LIV.32. 1-3). What was the function of Winsum during these activities of the Roman fleet? Winsum is not situated on the coast but this may indeed be an advantage in an area like this. The Wadden Sea is a tidal area where the force of the tides is very strong and it may have been more comfortable to have sheltered moorings further inland. Recent geological research has shown that Winsum originally was situated at the end of a funnel-shaped tidal basin. Most of the local rivers discharged into this basin, probably including the River Boorne, which provided a connection with the higher areas to the east. This was the situation around 400BC and from that period onward the funnel-shaped basin gradually silted up (Vos 1999: fig. 23 b-e). Around the beginning of the Christian era a small basin still existed but it is not known whether it was still possible for larger ships to to arrive and harbour at Winsum. It is very difficult to detect the course of the rivers dating to this period but it is highly probable that a navigable connection still existed between Winsum and the by then small tidal basin on the Wadden Sea coast (Vos 1999: 59, 63).

All the objects just mentioned: the various categories of Roman pottery, the halved asses with countermarks, the Aucissa fibula, and the bronze lantern, belong to the kind of artefacts that were used by the Roman army. Their presence at this site must be the result of some kind of interaction with the local community. To my knowledge, this combination of different Roman objects is very seldom found at a native site such as Winsum. On the other hand, it is difficult to identify Winsum as a military camp since no military equipment and no traces of a military site were found. Therefore we cannot compare Winsum with military camps like Velsen and Nijmegen. The only native site with similar objects is Bentumersiel in Germany. Bentumersiel has a similar position, it also lies on the northern coast, some 120kms east of Winsum, on the bank of the River Ems. The site is a native site with Roman artefacts and dates to the Tiberian period. The main difference from Winsum is that at Bentumersiel bronze fragments of an explicitly military character were found (Ulbert 1977). The excavation at Bentumersiel made it clear that Bentumersiel was different from the other sites on the Ems and therefore the site has been interpreted as a temporary base to be used in the summer and where the harvest could be gathered (Schmid 1984: 202-204). It may also have served as a support base for the Roman army (Schönberger 1985: 333) and it has been suggested that somewhere in the neighbourhood there may have been a naval station or a march camp of Germanicus (Schmid 1984: 204).

The Romans had army camps along the River Rhine, eg. at Nijmegen. From Vechten they sailed northwards. It is difficult to reconstruct the passage to the north because of subsequent changes in the northern part of what is now the IJsselmeer. Some archaeologists have reconstructed an inlet while others reconstruct a lake which was closed off from the Wadden Sea. However, recent investigation has shown that at the beginning of the Christian era an inlet is very likely to have existed (Vos 1999: 63). Sailing to the north and along the northern coastline, a fleet would face the problem of the tides. The water level at low tide did not permit navigation and therefore it will have been impossible to sail the Wadden Sea to the River Ems without harbouring. The journey from the Rhine to the Ems would have taken approximately four days. So the Roman ships must have made several stops on their way to the northern coast. Even the Romans could not prevail over this force of nature.

Discussion The problem is that we now have two native sites with early Roman material which are situated near the northern coast, but no military sites to connect them with. The nearest Roman military site is the fleet base at Velsen, dated to the Tiberian period (AD15-30). It is some 150kms away from Winsum and it is hardly likely that there was a close relation between Velsen and Winsum since the journey must have taken several days. Velsen is the northernmost military camp of the Romans known to us; we are informed by Roman authors that the Roman army sailed to the northern seas but we do not know anything about the organisation of support bases on land. Drusus went to the northern shores in 12BC, Tiberius in AD4 and Germanicus in AD15. Should we expect to find military camps or did the Roman army operate in a different way in

Both Winsum and Bentumersiel may have been harbouring points on the sailing route from the Rhine to the Ems and Elbe. The distance between Winsum and Bentumersiel suggests that there was at least one more anchoring point and we may think of the Lauwerszee, a tidal basin in the province of Groningen. In this area a Roman coin hoard from the Tiberian period was discovered by metal detection (Vin 1993) but until now there are no other traces of Roman presence in this early period. As was mentioned above, Bentumersiel may have been a support base for the army and Winsum too may have had this function of collecting goods and crops for the army. Winsum could 485

Limes XVIII

have been used as a small base for the collecting of tax in the form of hides, a suggestion made by Schönberger (1985: 333-334) or for exchanging raw materials. The location of Winsum is very suitable for this kind of activities; it is situated in the centre of Westergo, a densely populated area. Some 85 findspots of native pottery have been identified in an area west of Winsum measuring 10 x 20kms. The majority of these findspots are terpen, native rural settlements, with pottery dating from the period 200/150BC–AD50 (Taayke 1990: 190, fig. 61). This area was suitable for cattle-breeding and could therefore provide ox-hides that were necessary for the Roman army.

Friesland over 1998. De Vrije Fries 79: 207-208. Galestin M.C. 2000 Augusteïsche munten uit een Friese terp: vroeg-Romeinse munten uit de opgraving WinsumBruggeburen (Fr.). Paleoaktueel 11: 69-73. Galestin M.C. in press Winsum-Bruggeburen, first report of the excavation. An early Roman outpost among the Frisians? Part one: the Roman coins. Palaeohistoria 2001. Haalebos J.K. 1990 Het grafveld van Nijmegen-Hatert. Beschrijving van de verzamelingen in het Provincial Museum G.M. Kam te Nijmegen XI. (Nijmegen). Haalebos J.K. 1991 Das grosse augusteische Lager auf dem Hunerberg in Nijmegen. In B. Trier (ed.) Die römische Okkupation nördlich der Alpen zur Zeit des Augustus. (Münster): 97-107. Hiddink H.A. 1999 Germaanse samenlevingen tussen Rijn en Weser. PhD thesis, University of Amsterdam. Kolnik T. 1991 Zu den ersten Römern und Germanen an der mittleren Donau im Zusammenhang mit den geplanten römischen Angriffen gegen Marbod 6 n.Chr. In B. Trier (ed.) Die römische Okkupation nördlich der Alpen zur Zeit des Augustus. (Münster): 71-84. Lauwerier R.C.G.M. 1993 Twenty-eight bird briskets in a pot; Roman preserved food from Nijmegen. Archaeofauna 2: 15-19. Lauwerier R.C.G.M. 1995 Dertig lijsterborstjes in een pot. Geimporteerde conserven uit Romeins Nijmegen. Jaarboek Numaga 42: 7-12. Schnurbein S. von 1982 Die unverzierte terra sigillata aus Haltern. (Münster). Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. 66. Berichte Römische-Germanischen Kommission: 321-497. Schmid P. 1984 Siedlungsstrukturen. In G. Kossack, K.-E. Behre & P. Schmid (edd.) Archäologische und naturwissenschaftliche Untersuchungen an ländlichen und frühstädtischen Siedlungen im deutschen Küstengebiet vom 5. Jahrhundert v. Chr. bis zum 11. Jahrhundert n. Chr. (Weinheim): 193-244. Simon H.-G. 1976 Die Funde aus den frühkaiserzeitlichen Lagern Rödgen, Friedberg und Bad Nauheim. In H. Schönberger, & H.-G.Simon Römerlager Rödgen. Limesforschungen 15. (Berlin): 51-264. Taayke E. 1990 Die einheimische Keramik der nördlichen Niederlande, 600 v.Chr. bis 300 n.Chr., Teil I: Westergo (Friesland). Berichten Rijksdienst voor het Oudheidkundig Bodemonderzoek 40: 109-222. Ulbert G. 1977 Die römische Funde von Bentumersiel. Probleme der Küstenforschung im südlichen Nordseegebiet 12: 33-65. Vin J.P.A. van der 1993 Two new Roman hoards: Zoutkamp and Ried. Bulletin van het Anticke Beschavüng 68: 247-253. Vos P.C. 1999 The sub-atlantic evolution of the coastal area around the Wijnaldum-Tjitsma terp. In J.C. Besteman, J.M. Bos, D.A. Gerrets, H.A. Heidinga & J. de Koning The Excavations at Wijnaldum. Reports on Frisia in Roman and Medieval times. (Rotterdam): 33-72. Wiegels R. 1993 Rom und Germanien in augusteischer und frühtiberischer Zeit. In W. Schlüter (ed.) KalkrieseRömer in Osnabrücker Land. (Bramsche): 231-265.

The conclusion is that the excavation at Winsum did not provide evidence for the existence of an army camp, but the number and variety of the finds make it clear that activities took place that were connected in some way with the presence of the Roman army. The site may have been used as an anchoring point or as a base for tax collection. If Winsum was not itself an army camp, it is possible that an army camp may be found in the neighbourhood. The most likely location is northwards on the coast, at the northernmost end of the salt-marsh ridge near Beetgum. Near Dronrijp, some 5kms north of Winsum, a Roman cork-urn was discovered (Boeles 1951: 131, fig. 29). A mid-C1st terra sigillata fragment was discovered at Feins, south-west of Winsum (Galestin 1997: 349, fig. 6.19) and a Roman as with the countermark of Varus was discovered near Harlingen (Galestin 1999b). It is hoped that in the near future more finds will appear, to provide a better understanding of the connections between Frisians and Romans in the early C1st AD. Bibliography Boeles

P.C.J.A. 1951 Friesland tot de elfde eeuw. (’sGravenhage). Bos J.M., Niekus M.J.L.Th., Scheffer J. & Volkers T.B. 1998 Opgraving Winsum-Bruggeburen: Romeinen in Friesland! Paleoaktueel 9: 65-69. Bosman A.V.A.J. 1997 Het culturele vondstmateriaal van de vroeg-Romeinse versterking Velsen 1. PhD thesis, University of Amsterdam. Conspectus 1990 E. Ettlinger et al. Conspectus formarum terrae sigillatae italico modo confectae. (Bonn). Galestin M.C. 1992 Bewoningsgeschiedenis. In M.C. Galestin & T.B. Volkers Terpen en Terpvondsten (Groningen): 338. Galestin M.C. 1997 Romans and Frisians: Analysis of the strategy of the Roman army in its connections across the frontier. In W. Groenman-van Waateringe, B.L. van Beek, W.H.J. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 347-353. Galestin M.C. 1999a Littenseradiel: Winsum-Bruggeburen. In J.M. Bos & G.J. de Langen (edd.) Archeologische Kroniek van Friesland over 1998. De Vrije Fries 79: 218-219. Galestin M.C. 1999b Harlingen: Herbaijum. In J.M. Bos & G.J. de Langen (edd.) Archeologische Kroniek van

486

Marjan C. Galestin: Early Roman military activity on the Frisian coast Fig. 1. Map of the Netherlands in the Roman period (after Hiddink 1999: Fig. 1.1) showing the location of principal sites mentioned in the text. 1 - Bentumersiel. 2 Zoutkamp, 3 - Winsum, 4 - Velsen, 5 Vechten, 6 - Nijmegen. Legends 1 = clay; 2 = peat.

Fig. 2. Roman pottery fragments from Winsum (after Boekes 1951, Fig. 28).

487

Limes XVIII

Fig. 3. The modern village of Winsum with the former terp projected onto it. The excavated site is indicated by hatched lines.

Fig. 4. Rim fragment of cup, Conspectus Form 14.

488

Fig. 5. Fragmentary stamp of Thyrsus.

Problems concerning the early Roman fortifications at Marktbreit (D) and Valkenburg ZH (NL) Thomas Grane This paper is based on observations made in my thesis on the development and strategy of the Roman fortifications in the Rhine area from Augustus to Constantine the Great. Marktbreit: The first observation concerns the headquarters of the Augustan camp at Marktbreit. As is seen often in Augustan camps, the principia is connected with the building behind it. This building was identified as a possible praetorium although the excavators themselves were not happy with this identification. Several scholars have previously suggested that the two buildings are in fact the principia and praetorium of the Polybian and Pseudo-Hygian marching camps or that the commander had moved out of the praetorium. I chose to compare these two buildings with the equivalents of the camps at Oberaden and Haltern. On the basis of this comparison I concluded that the above suggestions must be right: that the commander had moved out of the praetorium and that both buildings should be considered a whole in the sense of the headquarters with administrative facilities: and that a development can be seen from Oberaden, where the headquarters had contained both the living quarters of the commander as well as the administrative offices, to Marktbreit, where the commander had moved out and the praetorium clearly contained a number of offices as well as the aedes. Valkenburg: At Valkenburg research led to the conclusion that Period 1 must have been the most interesting as an advanced post with a garrison of mixed units. This observation is based on features like the fossae, the unique principia, the size of the praetorium and the barracks as well as the strategic purpose of the fortification and its garrison. Furthermore I believe that the change to Phase 1a and the Periods 2-3 must have taken place within a short time and probably after AD47. Otherwise this change cannot make sense. This new garrison might have contained units already present at Valkenburg.

This paper is based on some of the results of my MA thesis that was finished in the summer 2000. The thesis was an examination of the development and strategy of Roman fortifications in the Rhine area from Augustus to Constantine the Great. The thesis was based on four cases through which it was my purpose to show the various buildings used and the different types of strategic deployment. The four cases are the fortifications at Marktbreit, Valkenburg ZH, Saalburg and Köln-Deutz. Apart from some general observations I also reached a number of observations concerning the specific fortifications. I will here concentrate on my considerations on the headquarters of Marktbreit and on those concerning Period 1 and Phase 1a of the Valkenburg fort.

camp at Haltern and the somewhat earlier camp at Oberaden (Figs. 2-4: v. Schnurbein 1981: 23, 24). Both camps possess the same axiality concerning the headquarters. In all three camps the principia (Figs. 2-4.A) is connected to the building behind it (Figs. 2-4.B) by a colonnade. R. Fellmann (1983) has at several occasions discussed the evolution of the headquarters from the marching camp to the permanent camp and I believe that this evolution can be seen in these three examples. M. Pietsch (1993: 360-361, Abb. 2-3) has divided the principia-buildings into two types. Type I consists of Oberaden and Haltern (Phase 1), while Type II consists of Haltern (Phase 2) and Marktbreit (and Neuss, Lager C). As the praetorium of Haltern is connected to Phase 1, I postulate that it is slightly older that the praetorium of Marktbreit. However, as none of the two can be chronologically pinpointed, there is no way to tell. The praetoria cannot be divided in a similar manner, as they are quite different from one and other. The earliest at Oberaden (Fig. 2.B) resembles a Roman town house with a set of rooms on the main axis consisting of fauces, atrium, tablinum, peristylium and triclinium with symetrically placed flanking rooms. This is clearly built up as a residential building. Furthermore, the Oberaden praetorium is placed in the centre of the camp, just behind the via principalis. That is the position of the praetorium of the marching camp. The praetorium at Haltern is somewhat different (Fig. 3.B). The rooms on the axis are the same, only there is no triclinium at the end. The problem is that the flanking rooms seem rather to consist of six individual buildings than parts of one large building, an interpretation put forward by S. von Schnurbein (1974: 59-60). At Haltern this building is placed behind the via quintana, while the principia lies in the centre. At

Marktbreit At Marktbreit as well as other Augustan camps, the headquarters have created some problems of interpretation (Fig.1). Although the ground plan of the central buildings at Marktbreit is quite clear, only Building I can be positively identified, as the principia (Fig. 1.I). However, it contains no obvious aedes and the so-called praetorium (Fig. 1.II) is closely connected with the principia. Some specific questions arise concerning Marktbreit. Is it reasonable to think that the commander had a use for 38 uni-sized rooms in his living quarters? Or is it possible that both the administration offices and aedes not present in the principia are placed here along with the commander’s dwellings? Or has the commander moved out as has been suggested by the excavators and other scholars? I will try to illuminate the problem through a comparison with two other Augustan camps, namely the contemporary

489

Limes XVIII

Marktbreit (Fig. 4.B) there is no atrium or tablinum on the axis and no symmetry in the flanking rooms. When examining the principia and praetorium together, the Oberaden headquarters resembles the march camp praetorium perfectly, ie. the living quarters of the commander is a part of the headquarters. That must have changed at Haltern. The layout of the praetorium could not supply suitable living quarters for the commander. Still some of the old set-up remains, but the six small buildings must have been offices. At Marktbreit the step is completed. Now the praetorium consists of a large number of offices leaving nothing really suitable for living quarters. The next step would be to move the offices from the praetorium to the sides of the principia, as can be seen at a double legionary camp, Vetera I at Xanten from the time of Nero (AD54-68: Fellmann 1983: 26). This would make the rear building obsolete.

Velsen (I + II, possibly parts of one large fort; M. de Weerd pers. comm.) and Valkenburg (Bechert & Willems 1995: 25; Kunow 1987: 56; de Weerd 1977: 282). To give a short resumé, the first excavator A.E. van Giffen (1948: 307-308) suggested that the first Valkenburg fort was a supply base for the invasion of Britain in 43. In 1977, this was rejected by M.D. de Weerd (1977: 282), who argued for an initial date of 39/40. He believed the fort to have been an advanced base to control the barbarians in the area (de Weerd 1977: 255). Already in 1973 and in 1977 J.K. Haalebos suggested some relations to the campaigns of Caligula, a theory mentioned by W. Glasbergen as well. That was based on a piece of a wine barrel from Valkenburg with a stamp bearing the letters C CAE . AUG . GER, meaning: C(aius) Cae(sar) Aug(ustus) Ger(manicus): (Glasbergen 1972: Glasbergen & Groenman-van Waateringe 1974: 37, pl.11.5; Haalebos 1973: 302-303, 1977: 283-284). Later, in 1981, W.A. van Es (1981: 95-97) saw the fort as a part of the ‘military triangle’ a term introduced by Glasbergen in 1966, with the purpose of regaining control of the area of the Frisii and Chauci. H. Schönberger (1985: 345), who believed the praetorium (Fig. 5A) was a fabrica, kept the supply base theory alive. But he could not say for whom the supplies were meant. W. Groenman-van Waateringe (1977: 236; 1991: 182) argued, as de Weerd, that the function of the fort was to keep control of the Rhine basin. In 1999 S.L. Wynia as well as Haalebos and W.J.H. Willems renewed the comment made earlier by Haalebos. This was based on the find from the fort in Vechten of yet another piece of wine barrel stamped with the same name. It is suggested that the emperor most likely visited both forts on his way to Britain in 40, which supposedly ended with the gathering of seashells on the coast (Dio LIX.25; Haalebos & Willems 1999: 252-253; Wynia 1999: 145-146). It is apparently the only time that the same wine barrel stamps have been reported at different locations. The only hard facts discovered were remains from a Roman road found during excavations south of the fort, which could be dendrochronologically dated to AD39/40, thus confirming this date (Hallewas et al 1993: 17-18). Unfortunately, it later appeared that this dating was incorrect (Haalebos & Willems 1999: 251-252; Hessing 1999: 156, n.14). Thus, nothing firm has resulted from the discussion.

The placement of the aedes is problematic as well, but since it has been established that the praetorium is a part of the headquarters it should be the room at the end of the main axis. That is also supported by the fact that it is more deeply founded than the rest. A deep foundation could also indicate a two storey room, though. But where did the commander live then? Often we hear that the commander’s living quarters are situated next to the principia. At Haltern a suitable dwelling has been excavated in the right praetentura in the corner of via principalis and via praetoria. At Marktbreit such a building has yet to be discovered. One possibility is that the buildings V and VI are parts of a major residential area of the commander right next to the principia (Fig. 1.V-VI). They are both irregular with rooms on three sides of a courtyard, and could easily be part of such a complex. It could be interesting, though, if the praetorium was to be found in the right praetentura as well. Two other Augustan camps have been mentioned at this congress. How do they fit in? J.K. Haalebos reported that one building in the inner corner of the right praetentura of the Nijmegen camp might have been the commander’s living quarters, and that is the exact same place as at Haltern. Unfortunately he could not tell us what is behind the principia. S. von Schnurbein showed us the camp at Anreppen. Here, a building much similar to the Oberaden praetorium has been found. But the principia is yet to be discovered.

I believe that the purpose of the first Valkenburg fort is in fact a mix of most of those here mentioned. If Velsen II were infact founded at the same time as the first Valkenburg fort, or, considering de Weerd’s comment, perhaps a part of a large modernised base including Velsen I, which is believed to have been active as well, it could indicate that the two forts were part of one specific programme (van Es 1981: 96-96; van Dockum 1995: 8182; Hessing 1995: 99). As the date is 39-40 it must be connected with some sort of campaign initiated by Caligula such as the above, which is described by Dio (CIX.21.25). The wine barrel stamps speak for that. But with the presence of Velsen II there must have been an attempt to regain control of the lands of the Frisii and Chauci. Since the uprising of the Frisii in 28, control

However, as the Nijmegen camp is the earlier of the two it does not fit very well with my theory. Further excavations could change that of course. Could it be that this is not a matter of development, but one of style, ie. that two types of headquarters were used simultaneously? Valkenburg Period 1: Precisely when and for what reason the first fort at Valkenburg was erected has led to several theories. It is important to note that before AD47 only three forts are known to be situated in the west Dutch river area: Vechten, 490

Thomas Grane: Problems concerning the early Roman fortifications at Marktbreit (D) and Valkenburg ZH (NL) would have been reduced considerably (Tacitus Anns. IV.72-73). Already in the year following the foundation of the Valkenburg fort, the Chauci were defeated by P. Gabinius Secundus (Suetonius Claud. 24.3; Dio LX.8.7). I find it very plausible that his army departed from one or more of these three forts. Six years later, control was established. Gnaeus Domitius Corbulo, who had recently arrived to the area, quickly defeated the troublesome Chauci again and, more interestingly, had already reinstalled Roman administration of the Frisii (Tacitus Anns. XI.18).

praetoria & principalis. However, it should be noted that various plans show different building techiques. In 1986, Groenman-van Waateringe presented a ground plan of the entire Period 1 fort on which the building is marked as built with uprights in trenches. On the following plan, which is a detailed plan of the south part of the praetentura, the same building is marked as built with uprights in sleeper beams (Groenman-van Waateringe 1986: 161-162, Figs. 2, 3 ) Nevertheless, I find it hard to believe that the commander had to enter his quarters through a narrow alley from the via principalis. If my idea is correct, the entrance of the praetorium must have been situated towards the via praetoria. If there really is a difference in foundation method it is conflicting, though.

With regard to the garrison of Period 1 it is generally accepted that it was a mixed unit of both milites and equites. de Weerd and Groenman-van Waateringe have argued convincingly that at least some of the soldiers had legionary status. Since inscriptions mentioning the cohors III Gallorum equitata have been connected with this garrison, it must mean that it was the equites that belonged to this unit.

Phase 1a: Whether this building was the praetorium of the adaptional Phase 1a as well is doubtful (Fig. 6.A). The excavators, nevertheless, state that the commandant’s dwellings did not seem to have been altered (Glasbergen & Groenman-van Waateringe 1974: 17). In the light of the change of setup otherwise in Phase 1a I would think the use of this building has changed as well, especially since the central space, the peristylium, which is actually the main feature of the praetorium, seems to have been made into a couple of rooms instead.

Several architectural features as well could point to the fact that Period 1 was special. One obvious feature is of course the two sets of rooms at the front of the principia. The barracks have been discussed thoroughly as well. Arguments against the billeting of legionarii were among others that the barracks were to small. J-.M. Morel’s examination in 1989 of Augustan barracks and the development from the republican march camp to Valkenburg Period 1 showed otherwise, however (Morel 1991: 376-386). I have nothing new to add to these features except a suggestion that the 14th contubernium might have been either the quarters of the tesserarius or perhaps a scola for the unit. One feature that has never been re-investigated since the excavations of A.E. van Giffen, is the fossae. When compairing the distance from the vallum to the outer edge of the outer fossa it is seen that the distance diminishes from 12.4m in Period 1 to 10.45m and 10.17m in Periods 2 and 3. Combined with the other facts it could indicate that the importance of the fort has been reduced.

Glasbergen and Groenman-van Waateringe, who believed the garrison to be half an ala quingenaria, placed this period around 42. That would be in connection with Claudius’ British campaign in 43 (Glasbergen 1972: 148149: Glasbergen & Groenman-van Waateringe 1974: 1315, 17-20). This is repeated by Hessing in 1995 (Hessing 1995: 96). Only de Weerd questions this date. He states that Phase 1a should be dated to after 45, thus dismissing any connection with the British campaign for this phase (de Weerd 1977: 271). There are no reliable arguments that make it clear that the adaption Phase 1a should not be dated to shortly prior to Period 2 around 47. And although nothing speaks against half an ala quingenaria being billeted at Valkenburg both before and after the change of politics in 47, I find that date a reasonable point of garrison change. All this of course based on the assumption that Corbulo’s withdrawal really was the triggering factor.

The last feature that I will bring up is the praetorium (Fig. 5.A). It has been put forward that the size of the praetorium indicated a high ranking officer as commander supporting the argument that legionarii were present. In the 1980 campaign it was established that the eastern part of the right praetentura was in fact not part of the praetorium but separate buildings. The reason was that the praetorium was built with uprights on sleeper beams while the small buildings found in the eastern side were built using uprights in trenches. Between the praetorium and the via praetoria lies an unidentified building built with sleeper beams (Fig. 5.B). The ‘space’ parting the two buildings is approximately 2m wide (Fig. 5.C). The width compared to the roads from Period 1 and the fact that drainage ditches are always present except here, indicate that this building is rather a part of the praetorium than a building in itself. Therefore, I find it quite probable that the praetorium was situated in the corner between the viae

If a date like that offered by Glasbergen and Groenmanvan Waateringe is correct, then this tactical change could indicate that the control had been regained already and that the tasks of the legionarii were fulfilled. It would also mean that Valkenburg would have had the appearance of an ordinary limes-castellum some years before the linear defense was begun, and that is a troubling thought. As the linear defense of the limes is usually connected with the withdrawal of Corbulo and his troops in 47, it would disturb the chronology of the entire northern frontier. And that is, of course, quite interesting.

491

Limes XVIII

Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Berichte der Römisch-Germanischen Kommission 66: 322-497. de Weerd M.D. 1977 The date of Valkenburg I reconsidered: the reduction of a multiple choice question. In B.L. van Beek & R.W. Brandt (edd.) Ex Horreo. (Amsterdam): 255-289. Wynia S.L. 1999 Caius was here: The emperor Caius’ preparations for the invasion of Britannia: new epigraphic evidence In H. Sarfatij, W.J.H. Verwers & P.J. Woltering (edd.) Discussion with the past. (Amersfoort): 145-147.

Bibliography Bechert T. & Willems W.J.H (edd.) 1995 De Romeinse rijksgrens tussen Moezel en Noordzeekust. (Stuttgart). van Dockum S. 1995 Het rivierengebied. In T. Bechert & W.J.H. Willems (edd.) De Romeinse rijksgrens tussen Moezel en Noordzeekust. (Stuttgart): 77-102. van Es W.A. 1981 De Romeinen in Nederland. (Haarlem). Fellmann R. 1983 Principia – Stabsgebäude. Kleine Schriften zur Kenntnis der römischen Besetzungsgeschichte Südwestdeutschlands 31. (Aalen). van Giffen A.E. 1948 De Romeinsche Castella in den dorpsheuvel te Valkenburg aan den Rijn (Z.H.) (Praetorium Agrippinae), I De opgravingen in 1941. Jaarverslag van de Vereeniging voor Terpenonderzoek 25-28, 1940-44: 1-317. Glasbergen W. 1972 De Romeinse Castella te Valkenburg Z.H. Cingula 1. (Amsterdam). Glasbergen W. & Groenman-van Waateringe W. 1974 The preFlavian garrisons of Valkenburg Z.H. Verhandelingen der Koninklijke Nederlandse Akademie van Wetenschappen, Afd. Letterkunde Nieuwe Reeks 85. (Amsterdam). Groenman-van Waateringe W. 1977 Grain storage and the supply in Valkenburg castella and Praetorium Agrippinae. In B.L. van Beek & R.W. Brandt (edd.) Ex Horreo. (Amsterdam): 226-240. Groenman-van Waateringe W. 1986 The horrea of Valkenburg Z.H. In Studien zu den Militärgrenzen Roms III. 13. Internationaler Limeskongresses, Aalen 1983, Vorträge. (Stuttgart): 159-168. Haalebos J.K. 1973 De Romeinse castella te Zwammerdam Z.H. Doctoral Thesis (Amsterdam) Haalebos J.K. 1977 Zwammerdam Nigrum Pullum. Ein Auxiliarkastell am niedergermanischen Limes. Cingula 3. (Amsterdam). Haalebos J.K. & Willems W.J.H. 1999 Recent research on the limes in the Netherlands. Journal of Roman Archaeology 12: 247-262. Hallewas D.P., van Dierendonck R.M. & Waugh K.E. 1993 The Valkenburg-Marktveld and Valkenburg - the Woerd excavations, 1985 - 1988: a preliminary report. (Amersfoort): 11-46. Hessing A.M. 1995 Het Nederlandse Kustgebied; De Romeinse rijksgrens tussen Moezel en Noordzeekust. In T. Bechert & W.J.H. Willems (edd.) De Romeinse rijksgrens tussen Moezel en Noordzeekust. (Stuttgart) : 89-103. Hessing A.M. 1999 Building programmes for the lower Rhine limes. In H. Sarfatij, W.J.H. Verwers & P.J. Woltering (edd.) Discussion with the past. (Amersfoort): 149-157. Kunow J. 1987 Die Militärgeschichte Niedergermaniens. In H.G. Horn (Hsrg.) Die Römer in Nordrhein-Westfalen. (Stuttgart): 27-109. Morel J.-M. A.W. 1991 Tents or barracks? In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 376-386. Pietsch M. 1993 Die Zentralgebäude des augustäischen Legionslagers von Marktbreit und die Principia von Haltern. Germania 71.2: 355-387. von Schnurbein S. 1974 Die Römischen Militärenlagen bei Haltern. Bodenaltertüner Westfalens 14, (Münster). von Schnurbein S. 1981 Zur Geschichte der römischen Militärlager an der Lippe. Berichte der RömischGermanische Kommission 62: 5-101.

492

Thomas Grane: Problems concerning the early Roman fortifications at Marktbreit (D) and Valkenburg ZH (NL)

Fig. 1. The central buildings at Marktbreit: I: principia. II: praetorium?. III-VI: Staff- and other buildings. 1: 1500: After M. Pietsch Die Zentral-gebäude des augustäischen Legionslagers von Marktbreit und die Principia von Haltern. Germania 71,2. (Mainz 1993): 356, Abb. 1.

493

Limes XVIII

Fig. 3. The headquarters of the Augustan camp at Haltern: A: principia. B: praetorium. 1: 1500: After S. von Schnurbein Die römischen Militäranlagen bei Haltern. Bodenaltertümer Westfalens 14. (Münster 1974): Beilage 6.

Fig. 2. The headquarters of the Augustan camp at Oberaden: A: principia. B: praetorium. 1: 1500: After T. Bechert & W.J.H. Willems (edd.) De Romeinse rijksgrens tussen Moezel en Noordzeekust. (Stuttgart 1995): 59.

494

Thomas Grane: Problems concerning the early Roman fortifications at Marktbreit (D) and Valkenburg ZH (NL)

Fig. 4. The headquarters of the Augustan camp at Marktbreit: A: principia. B: praetorium. 1: 1500: After M. Pietsch Die Zentral-gebäude des augustäischen Legionslagers von Marktbreit und die Principia von Haltern. Germania 71,2. (Mainz 1993): 356, Abb. 1.

Fig. 5. Valkenburg Z.H. Per.1. Right praetentura: A: praetorium. B: Unidentified building. C: ’Road’. 1: 500: W. Glasbergen De Romeinse Castella te Valkenburg Z.H.; Cingula I. (Amsterdam 1972): fig. 22.

495

Limes XVIII

Fig. 6. Valkenburg Z.H. Phase 1a. Right praetentura: A: peristylium of per.1 - praetorium parted in two. 1: 500: W. Groenman-van Waateringe & B.L. van Beek: De Romeinse Castella te Valkenburg Z.H.; Archeologie en Oecologie van Holland tussen Rijn en Vlie, J.H.F. Bloemers (ed.) Studies in Prae- en Protohistorie 2. (Assen 1988): 50, fig. 1.33.

496

Zur Datierung der frühkaiserzeitlichen Militärlager von Novaesium (Neuss) Norbert Hanel H. v. Petrikovits occasione diei natalis nonagesimi Bei Ausgrabungen zwischen den Jahren 1955 und 1982 wurden im Mündungsgebiet des Flusses Erft in den Rhein Spuren von 6-7 Militärlagern der augustisch-tiberischen Zeit mit insgesamt 12 - 14 Umbauphasen ihrer Umwehrungen gro§flächig freigelegt. Anhand des Fundmaterials soll die chronologische Entwicklung dieser frühkaiserzeitlichen Neusser Lagerplätze überprüft werden. Zu diesem Zweck wurden einerseits geschlossene Fundensembles untersucht, die mit den Umwehrungen (Holzerdemauern, Wehrgräben) in Zusammenhang stehen: andererseits wurde der Versuch unternommen, eine Feindatierung der sich überlagernden Innenbauten mittels gut datierarer Fundkomplexe zu gewinnen.

Der Fundort Novaesium ist eng mit der Frühzeit der römischen Okkupation Germaniens unter Kaiser Augustus verbunden (v. Petrikovits 1976: 320; Schönberger 1985: 329-330, 429). Die im Mündungsgebeit der Erft in den Rhein, durch die Ausgrabungen seit 1955 lokalisierten Militäranlagen können mit dem derzeit ältesten archäologisch faßbaren, römischen Zeithorizont nördlich der deutschen Mittelgebirgszone in Verbindung gebracht werden. Vor allem die Untersuchungen zu den Münzen und zur sogenannten ‘Arretina’ haben gewichtige Anhaltspunkte für diese Zeitstellung ergeben, die ungefähr auf die Jahre 20 – 12 vor der Zeitenwende einzugrenzen ist (Chantraine 1982: 43-44; Ettlinger 1983: 100-101, 107; Gechter 1979: 100).

Holzerdemauern einerseits oder aus den verfüllten Wehrgräben andererseits unterschieden. Während die Funde aus den Spuren der Holzerdemauern entweder bei deren Errichtung in den Boden kamen oder von einer etwaigen älteren Bebauung stammen, können Funde aus den Grabenverfüllungen – je nach ihrer Fundlage – Anhaltspunkte zur Benutzungszeit des Lagers bzw. zum Zeitpunkt seiner Auflassung geben. Falls vorhanden, wurden auch solche Funde analysiert, die im Kern der Kastenwerke lagen bzw. die aus solchen Befunden herrühren, die von den Holzerdemauern überlagert waren. Ein zweiter Schwerpunkt der Untersuchungen lag auf den Funden aus den freigelegten Innenbauten der Militärlager. Hierbei standen die Objekte im Vordergrund, die aus den Fundamentgräbchen, Pfostenlöchern und Pfostengruben dieser Fachwerk– und Holzbauten stammen. Allerdings konnten die zahllosen Funde aus solchen Gruben nicht berücksichtigt werden, die zwar in sich geschlossene Fundkomplexe bilden, aber in keinem konstruktiven Zusammenhang mit den Gebäuden stehen. Anhand dieser Vorgaben erfolgte die Sichtung des Fundmaterials, das sich mit 537 Fundnummern auf die Umwehrungen und mit 194 Fundnummern auf die Innenbauten verteilt. Im folgenden sollen die Ergebnisse der Untersuchungen zu den jeweiligen Militärkomplexen vorgestellt werden, wobei sich die neue Numerierung geringfügig von der älteren Zählung Müllers unterscheidet. Die ältere Bezeichnung wird in Klammern hinzugefügt.

Im Vordergrund der folgenden Ausführungen soll weniger das Anfangsdatum des Fundplatzes Novaesium stehen; vielmehr war zu prüfen, inwieweit das immense Fundmaterial eine zeitliche Differenzierung der insgesamt neun freigelegten Militärlager (mit 14 Bauphasen) aus der augustisch-tiberischen Zeit erlaubt. Ausgangspunkt der Untersuchungen ist das überarbeitete Manuskript von G. Müller zu den Befunden dieser frühen Neusser Anlagen und den stratigraphisch gewonnenen Lagerabfolgen.1 Weiterhin sollte geklärt werden, inwiefern Verknüpfungen zwischen literarisch überlieferten Ereignissen dieser Epoche und den archäologischen Befunden, d. h. den einzelnen Lagerausbauten, hergestellt werden können, wie dies verschiedentlich geschehen ist (Müller 1975: 384394).

Das Lager A mit zwei Umbauphasen gilt vor allem aus stratigraphischen Gründen als die älteste Anlage vor Ort (Müller 1975: 386-387). Es sind lediglich die beiden Wehrgräben erhalten, die eine Mindestfläche von 13-14ha umschließen. Die Ausdehnung nach Westen konnte nicht festgestellt werden. Hinweise auf die Wehrmauer fehlen ebenso wie Spuren der Innenbebauung; allenfalls läßt sich ein von West nach Ost verlaufender Abwassergraben diesem Lager zuordnen. Die für eine Datierung wichtigen Funde stammen überwiegend aus der Verfüllung der Wehrgräben. Die Analyse des Fundmaterials zeigt, daß keines der von E. Ettlinger für den frühen Neusser Horizont in Anspruch genommenen italischen Sigillaten und keine der in diesem Zusammenhang wichtigen Nemausus-Prägungen der Serie 1 mit einem Befund des Lagers A in Verbindung gebracht werden kann; lediglich die Randscherbe eines ‘Arretina’-Tellers wurde in dem genannten Abwassergraben gefunden (Ettlinger 1983: Taf.

Da allein die Zahl der Keramikfunde auf ungefähr 3 Millionen Scherben veranschlagt wird (Schätzung M. Gechter), galt es, die Untersuchungen auf ausgewählte Fundzusammenhänge zu konzentrieren. Zu diesem Zweck wurden anhand der Eintragungen in den Grabungszeichnungen diejenigen Funde ermittelt, die im Kontext der verschiedenen Umwehrungsabschnitte zutage kamen. Das Fundmaterial wurde nach seiner Provenienz aus den Fundamentgräbchen und Pfostenlöchern der 1

Die folgenden Ausführungen gehen auf einen Beitrag des Verf. “Ausgewählte Fundkomplexe aus den augustich-tiberischen Militärlagern von Novaesium” zurück, der in der Publikation von G. Müller Die augustisch-tiberischen Militärlager von Novaesium. Die Befunde. Novaesium 10 in den Limesforschungen erscheinen soll. An dieser Stelle werden die hier in kurzer Form dargelegten Ergebnisse ausführlich und mit Abbildungen vorgestellt.

497

Limes XVIII

23, 39). M. Gechter stellt aufgrund der Verteilung früher Funde fest, daß diese nur im Areal der Lager A und B auftreten (vgl. Ettlinger 1983: 90).

‘Halterner’ Topf, die Randscherbe einer Weinamphore Dressel 2-4 sowie der Höckelhenkel einer Weinamphore aus dem östlichen Mittelmeeraum. Weiterhin fand sich ein Kettenpanzerhaken mit Inschrift, der von der augustischen Zeit bis ungefähr zur Zeit Neros in Gebrauch war (Typ 2 nach Deschler-Erb 1991: 20). Aus einem Pfostenloch im vorderen Fundamentgräbchen der Umwehrung konnte die Wandscherbe eines glatten Sigillata-Gefäßes geborgen werden; weder der Gefäßtyp noch die Herkunft (Italien oder Südgallien) läßt sich bestimmen. Aus dem Bereich des horreum kamen nur mit jüngeren Funden vermischte Fundkomplexe zutage. Insgesamt gesehen widersprechen die Funde aus der Umwehrung der Anfangsdatierung des Lagers D in die frühtiberische Zeit durch G. Müller nicht; jedoch lassen sie eine genauere Bestimmung der Lagergründung nicht zu.

Es ist darauf hinzuweisen, daß in den Verfüllungen der Gräben des Lagers A besonders viele Fragmente einheimischer Gefäße lagen. Nach den Parallelen, die über die gesamte Niederrhein- und Mittelrheinzone bis in das Gebiet der Treveri streuen, gehören in die Stufe Latène D2 (2. Hälfte des 1. Jahrhunderts vor der Zeitenwende). Diese handgeformte oder scheibengedrehte Keramik, die am Niederrhein bis in die tiberische Zeit in den römischen Militärlagern häufig vorkommt, kann zur Feindatierung des Lagers A wenig beitragen. Die Ansicht von G. Müller, daß das Lager A um das Jahr 10 vor der Zeitenwende aufgelassen wurde (Müller 1975: 387), läßt sich aus den Funden nicht ableiten.

Zum Lager E wird von den Bearbeitern der Neusser Ausgrabungen eine ungefähr 450m lange Grabenspur gerechnet, die – von Süden her kommend – den Verlauf der jüngeren Umwehrungsspuren F 1a und b (früher G-I) mit einem einspringenden Winkel vorweggenommen hätte. Anschließend soll die Umwehrung nach einer Strecke von 250 in Ostrichtung nach Norden umbiegen. Die Innenfläche der Anlage, zu der keine Innenbauten bekannt sind, wird auf mindestens 40ha geschätzt. An dieser Stelle kann nur in aller Kürze darauf hingewiesen werden, daß parallel zum vermeintlichen Nord-Süd-Graben ein kiesgedeckter Straßenkörper und die Trasse einer hölzernen Hochwasserleitung verlaufen. M.E. muß der als Wehrgraben gedeutete Befund als westlicher Straßengraben interpretiert werden. Außerdem gibt es keinen Hinweis auf einen Vorläufer zur Umwehrung F 1b und c (G-I) mit identischer Streckenführung. Zu erwähnen ist, daß in der Verfüllung des Nord-Süd-Grabens E eine Aduatuker-Münze zusammen mit einem As der Jahre 16 vor bis 22 nach der Zeitenwende mit Gegenstempel AVG gefunden wurde. Der Umlaufhöhepunkt der AduatukerKleinerze lag im 1. Jahrzehnt nach der Zeitenwende. Die Randscherbe eines sogenannten ‘Halterner’ Topfes aus der Grabenverfüllung an der Ostfront datiert in die augustischtiberische Zeit, ohne daß das jahrgenaue Datum 17 nach der Zeitenwende für den Bau des postulierten ‘Lagers’ E auf diese Weise ermittelt werden kann.

Das Lager B mit zwei Bauphasen, dessen Umfang auf 3443ha geschätzt wird, bestand aus zwei parallel verlaufenden Wehrgräben. Überreste einer Umwehrung konnten ebenso wenig beobachtet werden wie zweifelsfrei zuweisbare Innenbauten. Im Zusammenhang mit der Umwehrung wurden 5 Fundkomplexe augustischtiberischer Zeitstellung festgestellt. Neben einer sogenannten pompeianischroten Platte mit Horizontalrand und einer Schüssel bzw. eines Gurtbechers in Terra rubraTechnik fanden sich ein halbierter Dupondius aus Nemausus der Serie 1 oder 2 sowie eine Aucissafibel mit Hohlkehle. Die von G. Müller vorgeschlagene Datierung des Lagers B in die Jahre 7 vor der Zeitenwende bis 9 nach der Zeitenwende ist in dieser Präzision aus den Funden nicht zu erschließen. Aufgrund schwieriger Ausgrabungsbedingungen konnte der Umwehrungsverlauf des Lagers C nur mit größeren Unterbrechungen freigelegt werden. Es handelt sich um einen Einzelgraben, dem auf der Innenseite ein Ausbruchsgräbchen in einem Abstand von ungefähr 5m zur Grabenspitze folgt und der die Vorderfront der Holzerdemauer markieren soll. Trotz ihrer großen Entfernung zur Lagerumwehrung von 400 - 500m werden Innenbauten wie principia etc. diesem Lagerkomplex C zugeordnet. Sowohl aus dem Bereich der Umwehrungen als auch der zugewiesenen Innenbebauung kamen keine geschlossenen, frühen Fundkomplexe zutage, so daß die Angabe von G. Müller, der Bau des Lagers C falle in den Sommer des Jahres 14 nach der Zeitenwende (Müller 1975: 390), nicht aufrechtgehalten werden kann.

Die Umwehrungsspur des Lagers F 1a (F) mit den zungenförmigen Eingrabungen an der Vorder- und Rückseite muß vom übrigen Verlauf der Anlage F getrennt werden, da eine eindeutige Überschneidung mit der Umwehrungsspur F 1b und c bzw. F2a und b nachgewiesen wurde; es muß sich um einen eigenständigen Lagerkomplex gehandelt haben. Als Größe nennen die Bearbeiter 22-30ha, wobei sie die jüngeren Lagergebiete hierzu zählen. Nur ein Fundkomplex aus dem hinteren Fundamentgräbchen der Holzerdemauer kommt für eine frühe Datierung in frage. Die Bodenscherbe eines Arretinagefäßes sowie verschiedene indifferente Amphorenbruchstücke erlauben allerdings keine feinchronologischen Aussagen. Das Jahr 25 nach der Zeitenwende ist als Zeitpunkt für die Lagererbauung aus

Vom Lager D ist lediglich der Verlauf der ungefähr 130m langen Ostfront bekannt. Diese im Vergleich zu den übrigen Neusser Lagern erheblich kleinere Anlage hatte vielleicht eine Fläche von 3.6ha Neben der Holzerdemauer bestand die Umwehrung aus 2 Spitzgräben; die Befunde lassen auf eine Zweiphasigkeit schließen. Zur Innenbebauung wird ein Speichergebäude an der nordöstlichen Lagerecke gerechnet. Aus der Verfüllung der Wehrgräben stammen ein Aduatukerkleinerz, ein Reibschüsselfragment mit Steilrand (Typ Haltern 59), ein 498

Norbert Hanel: Zur Datierung der frühkaiserzeitlichen Militärlager von Novaesium (Neuss) diesen Funde nicht abzuleiten.

gewonnenen Fundmengen ist die Zahl der Funde, die in die Frühzeit Novaesiums gehören, letztendlich relativ gering. Diese Zahl reduziert sich nochmals, läßt man alle diejenigen Funde außer acht, die aufgrund ihrer Funktion bzw. ihres Erhaltungszustandes für eine engere Datierung nicht in frage kommen: Hierunter fallen z. B. Eisennägel oder indifferente Wand- und Bodenscherben.

Als Umwehrungsspur F 1b und 1c (G) wird die Lagerfront bezeichnet, die von Westen her kommend, zweimal nach Norden umbiegt. Die Gesamtfläche wird auf 18-22ha veranschlagt, ohne daß die übrigen Umwehrungsabschnitte bekannt sind. Die Umwehrung bestand aus einer Holzerdemauer mit einem vorgelagerten Spitzgraben. Diesem Lager sind verschiedene Innengebäude auf der Nordseite zuzuweisen, die als Wirtschafts- und Verwaltungsbauten gedeutet werden.

Versucht man dagegen das Problem von der Verteilung der frühesten Funde aus anzugehen, so zeigt es sich, daß von den ungefähr 800 gestempelten und reliefierten ‘italischen’ Sigillaten keine einzige in einer Fundsituation angetroffen wurde, die für die Feindatierung der frühen Neusser Lager relevant gewesen wäre. Was die Münzen anbelangt, so kommen bei einer Gesamtzahl von rund 4000 Exemplaren 11 Stücke in Betracht, deren Aussagekraft jedoch durch unklare bzw. umstrittene Emissions zeiträume und unbekannte Umlaufszeiten eingeschränkt ist. Außerdem fehlen für die jeweiligen Lagerkomplexe ausreichend große Münzreihen, die eine Präzisierung der Belegungszeiten erlaubt hätten. Dieses Problem der relativen Fundarmut in Befunden frühkaiserzeitlicher Lager ist nicht auf Novaesium beschränkt; auch an anderen Militärplätzen läßt sich beobachten, daß zwar die Gesamtfundmenge zuweilen groß sein kann, daß aber bei genauerer Analyse der Fundumstände nur eine äußerst kleine Zahl an Funden für chronologische Fragen zur Verfügung steht. Erfahrungsgemäß kommen diese zumeist aus geschlossenen Grubeninhalten, die für sich genommen recht gut zu datieren sind, aber für die Datierung der Innenbauten zumeist nichts besagen: Dieses Phänomen ist – um nur einige zu nennen – aus Marktbreit, Vetera castra I und Nijmegen-Hunerberg bekannt.

Das zweiphasige Lager F 2a und 2b (H) entspricht weitgehend der Vorgängeranlage; es erfuhr jedoch durch die Verschiebung der nördlichen Ostfront um ungefähr 60m eine Vergrößerung um 1.1ha. Aufgrund des über lange Strecken ergrabenen Umwehrungsverlaufs konnten zahlreiche Fundkomplexe der verschiedenen Lagerausbauphasen F 1b bis F 2b geborgen werden, wobei wegen des weitgehend identischen Streckenverlaufs keine Zuweisung der Funde an die Unterphasen möglich war. Insgesamt stehen 337 Fundkomplexe mit den Umwehrungen F 1 und F 2 in Zusammenhang. Davon können 12 ein frühes Datum anzeigen: z. B. war die Randscherbe eines sogenannten ‘Halterner’ Topfes mit einem Denar des Tiberius vergesellschaftet. Dessen Emissionzeit fällt allerdings nur allgemein in die Regierungszeit des Kaisers. Aber auch ältere Funde wie z. B. der Dupondius der Nemausus-Serien 1-2 kamen aus der Verfüllung des Wehrgrabens oder ein halbierter Vienna-As des Jahres 36 vor der Zeitenwende aus dem hinteren Fundamentgräbchen der Holzerdemauer. Auch in den Bodenspuren der großflächigen Innenbebauung konnten zahlreiche Fundkomplexe beobachtet werden, ihre Zahl beläuft sich auf 113. Darunter befinden sich allerdings keine, die ausschließlich in die tiberische Zeit datiert werden könnten.

Als weitere Ursache für geringen Fundanfall in den augustisch-tiberischen Lagern wird die zeitweise Auflassung der Quartiere im Zusammenhang mit Truppenbewegungen angeführt (Gechter 1979: 100, 119, 122). Solche Truppenabzüge haben sicherlich in der frühen Okkupationszeit eine Rolle gespielt; allerdings besteht momentan keine Möglichkeit, weder die Stärke der ausrückenden Einheiten noch die Zeiträume ihrer Abwesenheit anhand des Fundmaterials einzugrenzen.

Von Lagerannex F 3 (I) wurde lediglich ein kurzer Umwehrungsabschnitt von 30m Länge, bestehend aus einem Spitzgraben und einer Holzerdemauer freigelegt, so daß Aussagen zur Fläche der Erweiterung hypothetisch bleiben. Vermutlich handelt es sich um den Rest einer Schenkelmauer, die die Hafenanlage des Lagers F abriegelte. Die Befundbearbeiter bringen diesen Annex, zu dem keine Innenbauten bekannt sind, mit Truppenansammlungen anläßlich des Britannienfeldzugs im Jahr 43 in Zusammenhang. Die Funde aus dem Umwehrungsbereich unterstützen diese präzise Datierung nicht.

(ii) Schon von Beginn der Untersuchungen an hat sich gezeigt, daß im gesamten Gebiet der frühen Neusser Militäranlagen, d. h. westlich des sogenannten Koenenlagers, die Funde vor allem in den Wehrgräben in die Mitte des 1. Jahrhunderts und jünger zu datieren sind. Zwar wurde um das Jahr 43 – nach unserem derzeitigen Kenntnisstand – der Truppenstandort geringfügig nach Osten verschoben; es ist aber davon auszugehen, daß das Areal der früheren Militärlager von den Bauten der canabae eingenommen wurde. Zu diesem Zweck mußten die älteren Anlagen mit ihren Innenbauten und Umwehrungen einplaniert werden. Darauf deuten Spuren von Steingebäuden, jüngere Bestattungen und schließlich das Keramik– und Münzspektrum mit den vermischten Fundkomplexen. Es entsteht der Eindruck, daß um die Mitte des 1. Jahrhunderts große Areale der früheren

Faßt man alle diese Einzelergebnisse zusammen, so muß festgestellt werden, daß sich alle feinchronologischen Datierungsangaben zu den Neusser Lagern – sieht man vom Anfangsdatum ab – anhand des Fundmaterials nicht bestätigen lassen. Welche Faktoren spielen eine Rolle, die dieses ‘negative’ Resultat hervorgerufen haben ? (i) Trotz großflächiger Freilegungen und der daraus 499

Limes XVIII

Militärlager eingeebnet wurden. Insgesamt gesehen hat die Analyse der Fundmaterialien aus den Umwehrungsspuren und in den Innenbauten zu keiner feinchronologischen Differenzierung der bekannten Lagerplätze geführt. Die relativchronologische Abfolge der Neusser Anlagen ist zuweilen eindeutiger als die Aussagekraft der betreffenden Funde. Alle früheren Angaben des Ausgräbers G. Müller, der die einzelnen Lagerkomplexe mit Ereignissen der historischen Überlieferung verknüpfte, werden somit hinfällig. Es kann nicht ausgeschlossen werden, daß seine Überlegungen mit den örtlichen Begebenheiten übereinstimmen; allein mit Hilfe des Fundmaterials sind diese jahrgenauen Zuweisungen nicht möglich. Trotz dieser für Novaesium negativen Bilanz muß betont werden, daß verschiedene Fundgattungen wie die frühen italischen Sigillaten, die Nemausus-Prägungen der Serie 1 und die Vienna/CopiaAsse mit ihren hohen Anteilen – für sich betrachtet – an der frühen Datierung des Platzes Neuss keinen Zweifel lassen. Literaturnachweise Chantraine H. 1982 Die antiken Fundmünzen von Neuss Gesamtkatalog der Ausgrabungen 1955-1978. Novaesium 8. Limesforsuchungen 20. (Berlin). Deschler-Erb E., Peter M. & Deschler- Erb S. 1991 Das frühkaiserzeitliche Militärlager in der Kaiseraugster Unterstadt. Forsuchungen in Augst 12. (Augst). Ettlinger E. 1983 Die italische Sigillata von Novaesium. Novaesium 9. Limesforsuchungen 21. (Berlin). Gechter M. 1979 Die Anfänge des Niedergermanischen Limes. Bonner Jahrbücher 179: 1-129. Müller G. 1975 Novaesium. Die Ausgrabungen in Neuss von 1955 bis 1972. Ausgrabungen in Deutschland gefördert von der Deutschen Forschungsgemeinschaft 19501975. Monographien des Römisch-Germanischen Zentralmuseums 1.1 (Mainz): 384-400. v. Petrikovits H. 1976 Beiträge zur römischen Geschichte und Archäologie 1931 bis 1974. Beihefte der Bonner Jahrbücher 36. (Bonn). Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Berichte der Römisch-Germanischen Kommission 66: 321-497.

500

Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit Andrea Hagendorn und Christine Meyer-Freuler1 Between 1996 and 1998, extensive excavations were carried out in the centre of the legionary camp at Vindonissa (Canton of Aargau, Switzerland). In the course of these excavations, beneath the stone constructions of the principia of the legiones XXI Rapax and XI Claudia pia fidelis, evidence was found of 7 periods of wooden buildings. Three of the later periods can be dated to the time of the legio XIII Gemina. The buildings of the first four periods, however, were erected before the arrival of the 13th legion. It is generally assumed that it was this legion which founded the first military camp in Vindonissa, around AD16/17. To the east of this early settlement, from the late Augustan period and beyond the so-called ‘Celtic ditch’, it is thought to be associated with an Augustan military base. This is based on early terra sigillata; until now, however, there have been no significant finds of militaria or remains of buildings to support the theory. During the first period of wooden buildings, the area of our excavations lay on the outskirts of a settlement. In the second period, two buildings constructed with wooden posts were erected on the site, around which various yards were grouped. These buildings were burnt down. In one of them, considerable quantities of provisions had been stored, as shown by numerous plant and animal remains. Pits for barrels and a great many fragments of dolia are also indications that provisions were kept here. In the third period, four half-timbered buildings, up to 35m long, were built on the site of the wooden post buildings. The gable ends of these buildings faced on to a street, and they were adjoined by yard areas, chiefly along the backs of the buildings. Buildings of this type are generally taken to be strip-houses. The buildings from the third period were also destroyed by fire and never rebuilt. During the fourth period, there was no further building on the area. A few pits were dug, and a partly roofed-over workshop area with a baking oven and some hearths for metal working were constructed. The finds from the first period wooden building are from a level which is just about contemporary with that of Dangstetten and Oberaden, which means that the beginning of the period must have been around the last decade BC. The pottery from the second and third periods is very similar in composition and a correlation in dating can be made with that of Haltern. Terra sigillata Services I and Services II are both equally strongly represented and come, as the stamps indicate, mostly from Italy. There is only a little material from the fourth period, but the persistence of Italian terra sigillata means it can be dated to the period from AD10 to 15. From all the layers and pits of the four periods of wooden building, vessels have been recovered which were produced in a local pottery. The forms are predominantly in the late Celtic tradition. The best parallels to this pottery are to be found in western Switzerland, which suggests cultural transfer or, possibly, migration. The buildings of the second and third periods have the appearance of a civilian settlement, even though there is also evidence in the excavated material of a military presence. The question arises as to the function of the settlements to the east and west of the ‚Celtic ditch’ and their relationship to each other. It could be that what we have come across here is the civilian quarter of the postulated Augustan military base. It is also possible, however, that extending to the east and west of the ‘Celtic ditch’ were different areas of a settlement which was supported by the army and which was inhabited by both civilians and soldiers.

1977). Es wird vermutet, dass diese Siedlung spätestens zu Beginn der Alpenfeldzüge von einer römischen Militäranlage abgelöst wurde (Hartmann 1980: 563).2

Einleitung Beim Zusammenfluss von Aare, Reuss und Limmat liegt das Legionslager von Vindonissa/Windisch, Kanton Aargau/Schweiz. Allgemein wird angenommen, dass das Lager 16/17 n. Chr. von der legio XIII Gemina gegründet wurde. Diese Legion wurde von der legio XXI Rapax (c.45-69/70 n. Chr.) abgelöst. Ihr folgte als letzte Besatzung die legio XI Claudia Pia Fidelis. Im Zuge der Dakerkriege Trajans wurde das Legionslager um 101 n. Chr. aufgelassen (Hartmann 1986). Eine nachfolgende zivile Besiedlung im Lagerareal wurde bislang nur in Ausschnitten erfasst.

Der Hof der Principia blieb nach Abzug der Legionstruppen frei von ziviler Bebauung. Deshalb haben sich dort die älteren Schichten der Lagerzeit und der Vorlagerzeit gut erhalten. Dennoch waren auch hier, wie an anderen Stellen im Lager, die Erhaltungsbedingungen aufgrund neuzeitlicher Überbauungen partiell sehr schlecht. In der Grabungsfläche wurden unter den in Steinbauweise errichteten Principia der 21. und 11. Legion sieben Holzbauperioden nachgewiesen. Drei der Holzbauperioden

In den Jahren 1996 bis 1998 wurde von der Kantonsarchäologie Aargau im ehemaligen Lagerareal eine Ausgrabung durchgeführt (Hagendorn 1998). Die c.2400 m² grosse Grabungsfläche lag im Hofbereich der Principia der 21. und 11. Legion (Abb. 1). In unmittelbarer Nähe der Ausgrabungsfläche befand sich der sog. „Keltengraben“, der eine auf dem östlichen Geländesporn gelegene spätkeltische Siedlung abriegelte (Hartmann & Lüdin

1

Einleitung, Baubefunde und Zusammenfassende Beurteilung: Andrea Hagendorn. Kommentar zur Keramik: Christine Meyer-Freuler. Die augusteische Militäranlage wurde allein aufgrund früher Terra Sigillata und Stempel vermutet. Diese Annahme kann aber weder mit Baustrukturen noch durch signifikante Militaria gestützt werden. Östlich des Keltengrabens sind bislang allerdings nur wenige grössere ungestörte Flächen untersucht worden (Hartmann & Lüdin 1977).

2

501

Limes XVIII Zurzach (Hänggi et al 1994)6 bestanden haben.

datieren in die Zeit der 13. Legion. Die Bauten der ersten vier Holzbauperioden hingegen wurden vor Ankunft der 13. Legion errichtet. Sie datieren in die augusteische Zeit (vgl. unten). Trotz der grossen Ausgrabungsfläche wurden keine spätlatènezeitlichen Gebäudestrukturen oder Fundkomplexe nachgewiesen. Die westlich des Keltengrabens zumeist in römischen Schichten gefundenen spätkeltischen Münzen (Doppler 1977) können also nach wie vor nicht mit spätlatènezeitlichen Siedlungsstrukturen in Verbindung gebracht werden.3

Die Baubefunde In der 1. Holzbauperiode war das Areal nicht flächig überbaut. Es wurden lediglich grosse Kiesabbaugruben angelegt und kleine Pfostenbauten, wohl Schuppen, errichtet. Die Gruben konzentrierten sich innerhalb eines Bereiches. Dieser war möglicherweise speziell für den Abbau von Kies ausgewiesen worden. Der umfangreiche Kiesbedarf und das Fundmaterial, das in den Gruben entsorgt wurde, weisen daraufhin, dass zu diesem Zeitpunkt auf dem Plateau bereits eine Siedlung existierte, deren Zentrum aber ausserhalb der Grabungsfläche lag.

Die sieben Holzbauperioden werden im Rahmen eines vom Schweizerischen Nationalfonds und der Kantonsarchäologie Aargau finanzierten interdiziplinären Projektes unter Mitarbeit verschiedener Autorinnen und Autoren ausgewertet.4 Im vorliegenden Beitrag werden die vorlagerzeitlichen Besiedlungsstrukturen vorgestellt, welche mit dem Arbeitsbegriff „vorlagerzeitliche Siedlung“ umschrieben werden.

Zu Beginn der 2. Holzbauperiode planierte man die südliche Hälfte des Grabungsgeländes grossflächig mit Kies und errichtete drei Gebäude in Pfostenbauweise (Abb. 3). Dies lässt auf eine Vergrösserung der ursprünglichen Siedlung schliessen. Gebäude C konnte aufgrund von Störungen nur sehr ausschnitthaft erfasst werden. Der Gebäudekomplex A/B hingegen war relativ gut erhalten, sein östlicher Teil liegt jedoch ausserhalb der Grabungsfläche. Anders als in dörflichen Siedlungen, in denen wegen der engen Parzellierung Höfe überwiegend hinter den Wohngebäuden lagen, waren hier verschiedene Hofbereiche um die Gebäude herum verteilt.

Gebäudestrukturen der Vorlagerzeit, darunter vermutlich ein Horreum, wurden erstmals 1936 im nördlichen Teil des Plateaus von Ch. Simonett dokumentiert (Simonett 1937: 82ff.; Meyer-Freuler 1998: 18).5 Es wurden zwar auch an einigen anderen Stellen kleine Ausschnitte vorlagerzeitlicher Gebäude nachgewiesen, es handelt sich jedoch um einzelne Gräbchen, die zu keinem Grundriss ergänzt werden konnten. Bekannt waren hingegen eine Töpferei und Werkplätze zur Eisenverarbeitung (vgl. Abb. 2; Meyer-Freuler 1998: 15ff.).

Der grosse Gebäudekomplex A/B bestand aus mindestens zwei Gebäuden, die über einen Korridor miteinander verbunden sind. Gebäude A umfasste eine Grundfläche von c.12m x 23m, Gebäude B von c.6m x 11m. Die beiden Gebäude umschlossen einen nach Norden offenen Hof. Im grössten Raum von Gebäude A wurden fünf Gruben angelegt, in denen insgesamt neun halbe Fässer standen. Die Fässer waren c.0.6m hoch erhalten und hatten einen Durchmesser von c.0.8m- 0.9m.

Östlich des Keltengrabens, wo in die gleiche Zeit datierende Schichten nachgewiesen wurden, konnten bislang keine grossflächigen Gebäudestrukturen erfasst werden (Hartmann & Lüdin 1977). Es stellt sich daher die Frage, welche Funktionen die Siedlungen östlich und westlich des Keltengrabens hatten und in welcher Beziehung sie zueinander standen. Daran schliesst die Frage nach ihrem Status - militärisch oder zivil - an. Geht man östlich des Keltengrabens von einer Militärstation aus, wäre westlich des Keltengrabens der zugehörige Kastellvicus zu vermuten. Es sind jedoch auch andere Modelle denkbar, die von einer Interpretation als zivile Siedlung oder Handelsstation bis zu einem militärisch unterstützten Vicus reichen (Meyer-Freuler 1998: 19ff.; Hagendorn 1998: 29ff.). Die Aufgabe eines solchen Vicus könnte vorrangig in der Versorgung der Kastelle von

Die Gebäude der 2. Holzbauperiode brannten bei einem Schadenfeuer ab. Im Brandschutt und in den in situ verbrannten Fässern fanden sich zahlreiche verkohlte botanische und zoologische Reste. Ihre Untersuchung erbrachte den Nachweis grösserer Mengen verschiedener einheimischer Nutzpflanzen, z. B. Getreide, sowie zahlreicher, in dieser frühen Zeit seltener Nahrungsmittelimporte. Nachgewiesen wurden dabei auch - erstmalig für die römische Zeit nördlich der Alpen – Kerne und Fruchtwandfragmente von Granatäpfeln Zu nennen sind auch (Punica granatum).7 Mittelmeermakrelen, die in einem solch frühen

3

Westlich des Keltengrabens sind in den letzten 70 Jahren grosse Flächen untersucht worden. Aufgrund der angewandten Grabungstechnik wurden die ältesten Holzbaustrukturen zwar nur sehr ausschnitthaft erkannt, Gruben hingegen wurden erfasst und untersucht. Auch in den ältesten Gruben war das Fundmaterial mit augusteischen italischen TS-Importen vermischt. Gruben mit rein spätlatènezeitlichem Fundmaterial wurden bislang nicht nachgewiesen. 4 Die Publikation soll im Frühjahr 2002 in der Reihe „Veröffentlichungen der Gesellschaft Pro Vindonissa“ erscheinen. 5 Zur Abfolge der Baubefunde hat Ch. Simonett widersprüchliche Angaben gemacht, doch spricht nach einer Prüfung seiner Grabungstagebücher und aller Argumente die grössere Wahrscheinlichkeit dafür, dass es sich bei dem ersten Gebäude am Platz um ein Horreum gehandelt hat.

6

In Frage kämen die Kastelle 3 und 4. Kastell 3 von Zurzach kann bislang nicht anhand von Fundmaterial datiert werden. Aufgrund historischer Überlegungen wird vermutet, dass es entweder eine Zeitlang neben dem Lager von Dangstetten bestanden hat oder in der Zeit zwischen 9 v. Chr. und 10 n. Chr. errichtet worden ist. Der Beginn des nachfolgenden Kastell 4 wird aufgrund des Keramikspektrums in die Zeit um 9/10 n. Chr. datiert. 7 Die archäobotanischen Reste wurden von Stefanie Jacomet, Institut für Ur- und Frühgeschichte, Abteilung Archäobiologie, Universität Basel ausgewertet. Ich danke ihr für die Überlassung dieser Resultate im Rahmen dieses Beitrages.

502

Andrea Hagendorn und Christine Meyer-Freuler: Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit Siedlungszusammenhang nachgewiesen wurden.8

bislang

nur

in

Dangstetten

Werkplätze angelegt: ein teilweise überdachter Werkplatz mit einem Backofen, ein Schmiedeofen sowie ein Werkplatz mit einer Feuerstelle, an der Bronze und auch Silber geschmolzen und gegossen wurden.

Die pflanzlichen und tierischen Reste sowie die Fassgruben weisen daraufhin, dass in Gebäude A Vorratshaltung betrieben wurde. In einer der Fassgruben waren zudem Fragmente von mindestens sieben Dolien, vier Amphoren für Fischsaucen aus der Baetica und zwei für Wein aus Kos verfüllt.9

Es stellt sich die Frage, warum das Areal in der 4. Holzbauperiode nicht mehr flächig überbaut wurde. Möglich wäre, dass die Siedlungsaktivitäten rückläufig waren oder sich verschoben hatten. Vielleicht war auch bereits geplant und bekannt, dass westlich des Keltengrabens ein Lager gegründet werden soll.

Der grosse Raum von Gebäude A und der südlich anschliessende Anbau waren mit Feuerstellen ausgestattet. Im östlich an den grossen Raum anschliessenden, räumlich untergliederten Gebäudeteil fanden sich hingegen weder Feuerstellen noch Vorratsgruben. Der Grundriss, die verschiedenen Funktionsbereiche und die lockere Verteilung der Vorratsgruben zeigen, dass es sich hierbei nicht um ein reines Magazingebäude, sondern um ein Wohn-Wirtschaftsgebäude handelte, in welchem Vorräte zur direkten Weiterverarbeitung zum Verbrauch oder Verkauf gelagert wurden.

Zusammenfassende Beurteilung Im ergrabenen Areal lösten sich innerhalb von 20-25 Jahren vier Holzbauperioden ab. In den einzelnen Bauperioden wurde das Gelände in unterschiedlicher Weise genutzt oder überbaut. Daneben ist auch eine periodenübergreifende Siedlungskontinuität fassbar: so errichtete man in der 1. Holzbauperiode westlich des sog. Keltengrabens eine Töpferei (Abb. 2), in welcher noch in der 3. Holzbauperiode produziert wurde (vgl. unten). Das Ausgreifen der Baustrukturen nach Norden lässt einerseits an Bevölkerungswachstum denken. Andererseits lassen die festgestellten Änderungen in der Bebauung vermuten, dass Grundstücke im Zuge einer Bevölkerungsfluktuation wieder aufgegeben bzw. von anderen Besitzern in anderer Weise genutzt wurden.

In der 3. Holzbauperiode wurde die Siedlung nach Norden erweitert. Es lassen sich nun zwei überbaute Zonen unterscheiden, zwischen denen eine c.8m breite einfache Erdstrasse verlief (Abb. 4). Die nördlich der Strasse anschliessende Bebauung entzieht sich einer eingehenderen Beurteilung, da sie durch neuzeitliche Bodeneingriffe stark gestört war. In der südlich der Strasse gelegenen Zone wurden anstelle der Pfostenbauten vier langgestreckte, bis zu 35m lange Fachwerkbauten (ohne Anbauten) errichtet.10 Die Pfosten wurden in Abständen von 0.5m und 0.7m in Gräbchen gestellt. Die Gebäude grenzten giebelseitig an die Strasse. Hofbereiche schlossen an die Längsseiten11 und an die Rückfront der Gebäude an. Diese Gebäude können als Streifenhäuser angesprochen werden. Sie finden ihre besten Vergleiche im Kastellvicus von Asciburgium (Bechert 1989: 135ff.) und auf dem Auerberg (Ulbert & Zanier 1997: 52ff.). Feuerstellen wurden nur in den rückwärtigen Gebäudeteilenangelegt. Möglicherweise wird hier eine übergeordnete Bauordnung fassbar. Auch die Gebäude der 3. Holzbauperiode wurden nach einem Schadenfeuer nicht wieder instand gesetzt.

Die Gebäude der 2. und 3. Holzbauperiode entsprechen dem Erscheinungsbild einer zivilen Siedlung, selbst wenn Hinweise aus dem Fundmaterial zeigen, dass auch in der Vorlagerzeit Militär anwesend war. Aufgrund des derzeitigen Wissensstandes zu Vindonissa ist daher zu fragen, ob die vorlagerzeitlichen Bauspuren von einem Kastellvicus stammen, der zu einer Militärstation östlich des sog. Keltengrabens gehörte. Denkbar wäre aber auch, dass sich östlich und westlich des Keltengrabens verschiedene Bereiche einer einzigen Siedlung erstreckten, in welcher ähnlich wie auf dem Auerberg (Ulbert & Zanier 1997: 121ff.) Zivilisten und Soldaten gleichermassen ansässig waren. Es würde sich dann die Frage stellen, wie sich die beiden Siedlungsbereiche zueinander verhalten haben und welche Funktion(en) sie hatten.

In der 4. Holzbauperiode wurden nur noch einzelne Gruben und eine Latrine sowie drei verschiedene

Kommentar zur Keramik in der Vorlagerzeit Im Unterschied zum Gebiet östlich des Keltengrabens gibt es im Bereich des späteren Lagers bislang noch keine Befunde, welche für eine spätlatènezeitliche Siedlung sprechen. Das Vorkommen von Münzen aus früheren Grabungen (Doppler 1977) und Amphoren Dressel 112 in der Grabung Breite wirft allerdings die Frage auf, in welchen besiedlungsgeschichtlichen Kontext diese

8

Die Kleintierknochen wurden von Heide Hüster-Plogmann, auch Institut für Ur- und Frühgeschichte, Abteilung Archäobiologie, Universität Basel ausgewertet. Ich danke ihr für die Überlassung dieses Resultats im Rahmen dieses Beitrages. 9 Die Amphoren der 2. Hp wurden von Stefanie Martin-Kilcher, Institut für Ur- und Frühgeschichte und Archäologie der Römischen Provinzen, Universität Bern ausgewertet. Ich danke ihr für die Überlassung dieser Informationen im Rahmen dieses Beitrages. 10 Innerhalb der Gebäude wurden partiell Kiesböden nachgewiesen. Deshalb ist anzunehmen, dass die Räume entweder durch Öffnungen im Dach belichtet und belüftet wurden oder dass kleine Innenhöfe vorhanden waren. 11 In der ersten Bauphase hatte Gebäude B auch an der westlichen Längsseite einen Hof. Später wurden hier Anbauten errichtet.

12 Bestimmung der Amphoren Dressel 1 durch Stefanie Martin-Kilcher (Institut für Ur-und Frühgeschichte und Archäologie der Römischen Provinzen, Universität Bern). Ich danke ihr für die Überlassung dieser Informationen im Rahmen dieses Beitrages.

503

Limes XVIII

einzuordnen sind. Es zeigt sich, dass überall, wo bis zum gewachsenen Boden gegraben wurde, Schichten und Gruben immer augusteische italische TS-Importe enthalten.

erstmals Erzeugnisse des VEPOTALVS auf, eines Produzenten, welcher in der Schweiz zu den frühesten Vertretern gehört, der TS-Imitationen hergestellt und gestempelt hat (10 von 20 Stempeln in der Grabung Breite, vgl. Drack 1945: 118; Luginbühl 2001: 306f.). Vepotalus hatte vermutlich vor allem in der Westschweiz seinen „Hauptsitz“; naturwissenschaftlichen Untersuchungen zufolge muss aber in der Schweiz mit mehreren Produktionsstätten gerechnet werden (Zanco 1999: 92). Die dünnwandige Feinkeramik ist nur in sehr kleinen Fragmenten vorhanden. Aufgrund charakteristischer Randund Wandscherben konnten Rippenbecher, Rillenbecher, Becher mit verdicktem Rand (Oberaden Typ 20) und konische Becher Typ ACO bestimmt werden, alles Formen, welche von den frühkaiserzeitlichen Militärplätzen wie Dangstetten, Rödgen, Oberaden, Neuss16 und Basel her gut bekannt sind. Auf einem ACOBecher ist der Stempel des T. CAVIVS erhalten, von dem es in Dangstetten (Fingerlin 1986: 254.62.3.7; 255. 66.16) etliche gibt und der möglicherweise seine Produktionsstätte in Lyon hatte.17 Aufgrund der Importe ist zumindest für die 2. Holzbauperiode an einen Zeithorizont zu denken, der ähnlich demjenigen der 1. Holzbauperiode Dangstetten und Oberaden noch berührt, ohne ihn ganz zu umfassen und zeitweise parallel mit Haltern läuft, während die 3. Holzbauperiode wohl etwas später um Chr. Geb. einsetzt und etwa bis zum Ende von Haltern (9 n. Chr.) dauert.

Die Befunde der 1. Holzbauperiode sind wie oben erwähnt spärlich, entsprechend wenig repräsentativ sind auch die Funde. Der Hauptteil der Keramik ist halternzeitlich, doch gibt es einige wenige ältere Funde (z. T. auch aus den nächstfolgenden Holzbauperioden 2 und 3), welche noch den Horizont Dangstetten und Oberaden berühren. Allerdings fehlen die frühesten italischen Formen wie Conspectus 1, 2, 7, 8, 10.13 Zum Spektrum gehören frühe Terra sigillata-Stempel wie SEX/ANNI (OCK 183: 1-7 s.), P.ATTI/HILARI (OCK 353: 1) - letzterer stempelgleich nur in Dangstetten, Bavai und Asciburgium vertreten - und dünnwandige Feinkeramik.14 Charakteristisch sind zudem die bemalte Keramik sowie handgeformte verzierte Kochtöpfe. Letztere sind noch so stark der SLT-Tradition verpflichtet, daß sie ohne Kontext ohne weiteres älter eingestuft werden könnten. In der ersten Siedlungsperiode wurde zudem ein Töpfereibetrieb eingerichtet, welcher sicher bis zur 3. Holzbauperiode aktiv war (Abb. 2; Koller 1990). Die Produkte daraus weisen starke spätkeltische Merkmale auf (Tonnen, Flaschen, verzierte Schultertöpfe) und haben ihre nächsten Parallelen in der Westschweiz (Meyer-Freuler 1998: 16ff.). Wichtig ist die Tatsache, daß aus dem Gebiet westlich des Keltengrabens frühe dünnwandige Feinkeramik zum Vorschein kam und sich die keramischen Funde kaum von dem östlich des Keltengrabens, wo ein augusteischer Militärposten vermutet wird, unterscheiden lassen (Hartmann & Lüdin 1977).

Aus der 4. Holzbauperiode stammt sehr wenig Material, da es sich um ein Werkareal handelt. Südgallische Terra sigillata fehlt noch fast völlig. Beim italischen TS-Service I und II zeigt sich eine - bei einer allerdings kleinen Anzahl deutliche Dominanz von Service II. An gestempelten Gefässen ist nur eines aus den frühen Ateiusbetrieben in Arezzo zu nennen (OCK 267: 31), wobei vermerkt werden muss, dass die Ateius-Produkte aus Arezzo (OCK 267) nördlich der Alpen viel weniger in Erscheinung treten als diejenigen aus Pisa (OCK 268).18 Einen besonderen Stellenwert nehmen die importierten halbkugeligen Schälchen entweder aus dunkelgrauem Ton mit rotbraunem Tonkern oder hellgrauem, hart gebranntem Ton mit weissen Einschlüssen ein. Ihre Herkunft ist noch ungeklärt, doch wird in der Literatur immer wieder Oberitalien (Ravenna, Aquileia) in Betracht gezogen (Ricci 1985: 349. Hänggi et al 1994: 107). Sie treten im Kastell Zurzach stets im Zusammenhang mit italischer TS auf und sind deshalb ein einigermassen verlässlicher Indikator für

Die Keramik der 2. und 3. Holzbauperiode ist sich in ihrer Zusammensetzung sehr ähnlich, da sich in dieser Zeitspanne keine wesentlichen Neuerungen abzeichnen. Wie in der 1. Holzbauperiode kann in der 2. und 3. Holzbauperiode immer noch ein schwaches Überwiegen von Service I zu Service II konstatiert werden. Mit Ausnahme des ALBANVS (OCK 63), der in Lyon produziert hat , stammen alle andern Töpfer aus Italien (z. B. ANNIVS, PHILEROS, SESTIVS). Sie sind stempelgleich auch in Haltern (PHILEROS OCK 1449: 4), Neuss (ANNIVS 116: 32, CRIS/PINI OCK 702:8, LEPI/DVS OCK 1024:1, SESTI OCK 1922: 5) und Mainz (ANNIVS OCK 116: 32, SESTI OCK 1922: 5) vertreten, was einen ersten Hinweis zur Datierung gibt. Die Stempeltypen des PHILA/AVILLI (OCK 382: 2) und des ANTIOCHVS (OCK 207: 2) sind einstweilen Unikate nördlich der Alpen.15 In der 2. Holzbauperiode tauchen

Vindonissa laut OCK total fünf Stempel in Korinth, Macerata Feltria, Bulla Regia und Florenz), während es vom Puteolaner Antiochus andere Stempeltypen in Mainz, Neuss, Oberaden und Xanten gibt. 16 Aufgrund der Kartierungen, welche M. Gechter für verschiedene Fundgruppen vorgenommen hat, könnte es sich dabei um die beiden frühen Lager A und B handeln. Ich danke M. Gechter für die Erlaubnis, dies hier erwähnen zu dürfen. 17 Ob sich die Werkstätte des T. Cavius tatsächlich im Rhonetal (Lyon, Vienne) befand, kann allerdings noch nicht mit Sicherheit gesagt werden. Zeugnisse aus Oberitalien lassen auch eine Werkstätte in dieser Region vermuten (Desbat et al 1996: 72, 172 pl. 69, 230ff.). 18 Diese Produktion löste nach P. Kenrick etwa um 5 v. Chr. diejenige in Arezzo ab. Kenrick 1997: 184ff.

13 Die von K. Roth-Rubi (Beitrag in diesem Band „Why Dangstetten“) auf Fig.1 mit einem Dreieck bezeichneten frühen italischen Formen beziehen sich auf eine kleine Grabung östlich des Keltengabens (Hartmann & Lüdin 1977: Taf. 3, 40.41, Consp.11, bzw. 1). Die wenigen frühen Funde gehören zeitlich alle an das Ende von Dangstetten und stammen alle aus Schichten oder Grubenverfüllungen mit mehrheitlich halternzeitlichem Material. Anhand des derzeitigen Foschungsstandes können wir eine zu Beginn von Dangstetten gehörige Siedlung noch nicht fassen. 14 Bei der einzigen erhaltenen Randscherbe dürfte es sich um einen Rippenbecher handeln. 15 Die Stempel des Töpfers Phila Avilli sind extrem selten (mit

504

Andrea Hagendorn und Christine Meyer-Freuler: Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit die spätaugusteische Zeit. Nach wie vor ist die Töpferofenware an der Scheuergasse mit dem üblichen Repertoire (verzierte Tonnen und Töpfe, Dolien) präsent. Es ist zu vermuten, daß diese Keramik in dieser Zeit entweder noch produziert wurde, oder – was noch wahrscheinlicher ist – zumindest noch weiter benützt wurde.

Ausgrabungen in der Breite 1996-1998. Mit einem Beitrag von Heide Hüster-Plogmann. Jahresbericht der Gesellschaft Pro Vindonissa 1998: 23-36. Hänggi R. et al. 1994 Die frühen römischen Kastelle und der Kastell-Vicus von Tenedo-Zurzach. Veröffentlichungen der Gesellschaft Pro Vindonissa 11. Hartmann M. 1980 Der augusteische Militärposten von Vindonissa. In W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies XII. Proceedings of the XIIth International Congress of Roman Frontier Studies. British Archaeological Reports Int. Ser. 71. (Oxford): 553-566. Hartmann M. 1986 Vindonissa, Oppidum-LegionslagerCastrum. (Windisch). Hartmann M. & Lüdin O. 1977 Zur Gründung von Vindonissa. Jahresbericht Gesellschaft Pro Vindonissa 1977: 5-36. Kenrick P.M. 1997 Cn. Ateius – the inside story. Rei Cretariae Romanae Fautorum Acta 35: 179-191. Koller H. 1990 Ein Töpferofen aus augusteischer Zeit in Vindonissa. Jahresbericht Gesellschaft Pro Vindonissa 1990: 3-41. Luginbühl T. 2001 Imitations de sigillée et potiers du HautEmpire en Suisse occidentale. Cahiers d’Archéologie Romande 83. Meyer-Freuler Ch. 1998 Gedanken zur Frühzeit von Vindonissa – Gewerbebetriebe, Getreidespeicher und andere Befunde westlich des Keltengrabens. Jahresbericht Gesellschaft Pro Vindonissa 1998: 13-22. Ricci A. 1985 Ceramica a pareti sottili. Atlante delle forme ceramiche II. Ceramica fine romana nel bacino mediterraneo (tardo ellenismo e primo impero). Enciclopedia dell’Arte Antica: 231-357. Simonett Ch. 1937 Grabungen der Gesellschaft Pro Vindonissa in den Jahren 1935 und 1936 auf der Breite (K.-P. 1446). Anzeiger für Schweizerische Altertumskunde Neue Folge 37: 81-92. Ulbert G. & Zanier W. 1997 Der Auerberg II. Münchner Beiträge zur Vor- und Frühgeschichte 46. Zanco A. 1999 Provenance and technology of galloroman terra sigillata imitations from western Switzerland. Institut de Minéralogie et de Pétrographie Université de Fribourg (Suisse) Thèse n. 1209.

Zusammenfassung 1. Die vier Holzbauperioden verteilen sich auf eine Zeitspanne vom 2./1. Jahrzehnt bis zur Mitttel des 2. Jahrzehnts n. Chr. Dabei kann aufgrund des keramischen Materials festgehalten werden, dass es sich bei der 1. und 4. Holzbauerperiode um eine kurzfristige Angelegenheit gehandelt hat , währenddem die 2. und 3. Holzbauperiode von längerer Dauer war. 2. Das spätkeltische Element ist in dieser Zeit stark spürbar, einerseits mit der Töpfereiproduktion, welche während der ganzen Vorlagerzeit aktiv war und damit auf eine Kontinuität weist, anderseits mit den handgeformten Kochgefässen, welche auf eine ansässige Bevölkerung schliessen lässt. 3. Eine kulturelle Beziehung zur Westschweiz ist offensichtlich und lässt auf kulturellen Transfer schliessen (Zuwanderung?) (Keramik aus dem Töpferofen an der Scheuergasse, bemalte Keramik, TS-Imitationen aus der officina des Vepotalus und des Villo). 4. Eine zweite ebenfalls starke Komponente sind die TSImporte aus Italien. Anhand der Stempelliste (rund 25 Stempel) lässt sich feststellen, daß nur ein einziger Töpfer aus Lyon stammt (ALBANVS). Abkürzungen Consp. = Ettlinger E. et al 1990 Conspectus formarum terrae sigillatae italico modo confectae. Materialien zur römisch-germanischen Keramik Heft 10. (Bonn). OCK = Oxé A., Comfort H. & Kenrick P. 2000 Corpus vasorum arretinorum. Antiquitas Reihe 3, 41.

Literatur Bechert T. 1989 Die Römer in Asciburgium. Duisburger Forschungen 36. Desbat A. et al 1996 Les productions des ateliers de potiers antiques de Lyon, 1ère partie: les ateliers précoces. Gallia 53: 1-249. Doppler H. 1977 Die keltischen Münzen von Vindonissa. Jahresbericht der Gesellschaft Pro Vindonissa 1977: 37-61. Drack W. 1945 Die helvetische Terra sigillata-Imitation des 1. Jahrhunderts n. Chr. Schriften des Institutes für Urund Frühgeschichte der Schweiz 2. Fingerlin G. 1986 Dangstetten I. Katalog der Funde (Fundstellen 1-603). Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 22. Hagendorn A. 1998 Neues zum Lagerzentrum von Vindonissa –

505

Limes XVIII

Abb. 1. Windisch-Vindonissa. Das Legionslager der 11. Legion. Schwarz: Die Grabungsfläche im Hofbereich der principia. M 1: 5000.

506

Andrea Hagendorn und Christine Meyer-Freuler: Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit

Abb. 2. Gebäude und Gewerbeareale der vorlagerzeitlichen Siedlung. Als Orientierungshilfe sind die späteren Umwehrungen und Strassen der Legionslager sowie die modernen Strassen eingetragen. M 1: 5000.

507

Limes XVIII

Abb. 3. Schematischer Grundrissplan der 2. Holzbauperiode (F = Feuerstellen, FG = Fassgruben, G = Gruben). M 1 : 800 (Plän: Riccardo Bellettati, Kantonsarchäologie Aargau).

Abb. 4. Schematischer Grundrissplan der 3. Holzbauperiode (F = Feuerstellen, G = Gruben). M 1 : 800 (Plän: Riccardo Bellettati, Kantonsarchäologie Aargau).

508

Why Dangstetten? Katrin Roth-Rubi Die Datierung des Militärlagers von Dangstetten (Kr. Waldshut, Baden-Württemberg) wird heute allgemein mit 15BC - 9/7BC beziffert; den Beginn stellt man mit dem von Strabo überlieferten Alpenfeldzug von 15BC in Verbindung, das Ende mit den Umstruktrierungen am Niederrhein, die auch das Ende von Oberaden bedingten. Analog zu den Offensiven am Niederrhein wird Dangstetten meist als Ausgangspunkt für die Besitznahme des rechtsrheinischen Germaniens von Süden her angesehen. Im Rahmen eines Projektes der Deutschen Forschungsgemeinschaft widme ich mich seit kurzem der Analyse des Tafelgeschirrs von Dangstetten (terra sigillata, Feine Ware). Zwei Argumente veranlassen mich, der communis opinio als Arbeithypothese gegenüberzustellen, dass Dangstetten ausschliesslich mit dem Alpenfeldzug in Verbindung zu bringen ist: • Das Fundmaterial aus den Wachtturmen am Walensee (Neuedition der Keramik durch Ref.) ist mit dem Dangstettener Fundstoff eng verknüpft • Neue Anhaltspunkte zur chronologischen Entwicklung der frühen Sigillata-Formen auf Grund stratigraphischer Aufschlüsse in Lyon (A. Desbat) deuten auf einen früheren Beginn der Dangstettner Belegung Obschon Strabo überliefert, dass der Alpenfeldzug nur einen Sommer gedauert haben soll, gibt es Hinweise, dass seine Vorbereitung längere Zeit in Anspruch genommen hat. Ich gehe davon aus, dass der Zug der beiden Prinzen das Ende der Eroberung des Alpengebietes bezeugt, mit militärischen Aktivitäten in Voralpengebiet aber, ‘ab 20BC’ zu rechnen is (so bereits C. Wells The German policy of Augustus. 1972).

If Colin Wells (1972) had known the material from Dangstetten in 1972 as it is now presented to scholars in Gerhard Fingerlin’s edition (Fingerlin 1986; 1998), and if he had been able to integrate it into his historical essay, his work would certainly have had a greater impact on research in the past 30 years. Wells believed that in 20BC, a military deployment in north-eastern Switzerland, particularly in Zürich/Lindenhof and the Walensee watch towers, prepared the way for the ‘Alpenfeldzug’, which according to Strabo (Geog. IV.6.9 (206)) took place in 15BC, with the well-known pincer attack by Tiberius and Drusus. Wells’ reasoning was acute, but his view of the events and their chronological sequence lacked the supporting evidence of the finds.

before Oberaden. In our present state of knowledge, the coins do not allow any finer chronological distinction, with the exception of the terminus ante quem of 10BC, since a critical comparison of the coin series from the numerous early military camps on the Rhine, an indispensable preliminary requirement, has not yet been undertaken. The question of Dangstetten’s strategic importance has stirred-up a certain amount of controversy. Fingerlin (1970-71) argues from its location, at the entrance to the Klettgau, that it occupies an offensive position directed towards the Danube sources, as a base for an advance wedge similar to those in the Lippe or Main valleys. There can be no doubt that an important north-south Rhine crossing lay near Zurzach/Dangstetten, a crossing destined to play a significant rôle in the C1st AD (Hänggi 1994), but whose importance in the years before Christ is difficult to assess. Hans Schönberger (1985: 435), on the other hand, is skeptical of Dangstetten’s offensive character, partly because posts of sufficient size, such as those in the Lippe region, seem to be absent from southern Germany.

In 2000, as part of a project sponsored by the Deutsche Forschungsgemeinschaft, I undertook the study of the table ware from Dangstetten, ie. the sigillata and other fine pottery (Aco beakers etc.). The aim of this undertaking is, on the one hand, to attempt to date the military installation, but also to gain a better understanding of the organisation of the ancient potters’ industry, its commercial ties, and the rôle of the army as a consumer of its products. The study is still in its early stages, although I have long been interested in the problem of early sigillata and in Dangstetten itself. In the following, I would like to present a working hypothesis.

Recent studies do not shed very much light on the question. While it is generally accepted that Dangstetten was built in connection with the ‘Alpenfeldzug’, no explanation of its strategic function is sought. The discussion seems to be stagnating (Kienast 1999: 358f.). One possible avenue for determining Dangstetten’s military function has remained unexplored: it has never been compared with the early Roman stations in northern Switzerland. For this reason, as a preliminary to the study of the pottery from Dangstetten itself, I have examined as much as possible of the early Augustan material from Swiss territory. When I use the terms ‘early Roman’ or ‘early Augustan’, I mean the period which in principle begins with the first governorship of Agrippa; its end need not concern us here. The pottery I regard as typical are the

The communis opinio on the dating of Dangstetten (Schönberger 1985: 435) is currently “founded in 15 B.C., abandoned between 9 and 7 B.C.”. The foundation date of 15BC is based on the ‘Alpenfeldzug’ just mentioned, the end is related to the well-known restructuring process in Germania, during which Oberaden and Rödgen were abandoned as well. The main point of similarity with Oberaden, on the lower Rhine, is the absence of coins of the Lyon altar series. Nevertheless, it should be pointed out that Dangstetten may very well have been relinquished 509

Limes XVIII

sigillata shapes Conspectus 1 and 2 (plates), 7 and 8 (cups), the large platters with radial stamping Conspectus 10 and 11 and the corresponding cup Conspectus 13, and Aco beakers (Desbat et al 1996). It is well established that these types are indicative of the earliest Roman presence north of the Alps (cf. Conspectus). Dangstetten offers numerous examples of the shapes Conspectus 7, 11 and 13, and several of shapes 1, 2 and 8. The large quantity of Aco beakers and related fine ware is obvious. The early shapes are found in context with canonical ‘Service I’ ware.

surrounds it on the outside, its floor consolidated by numerous amphora sherds. As to the finds, there are small deposits from the towers of Filzbach and Biberlikopf, but none at all from the third tower. In fact their number is so limited that Schönberger (1985: A 58) considered them too risky to interpret. It is true that the old publication was hardly sufficient, but our methods of classification have meanwhile evolved. It is possible to ascertain that early shapes, such as Conspectus 8, 11 and 13, as well as Aco beakers are present; the one from Biberlikopf can even be assigned to Chrysippus from Lyon (pub. Acta RCRF Lyon 2000, in prep.). The similarity between the two deposits, both in the range of shapes and in quality, indicates an identical date. Their findspots alone are sufficient proof of their military character, but military equipment was actually found with them as well. The amphorae are particularly remarkable: they come from regions as far apart as the Baetica, Tarraconensis and the Adriatic. The goods must presumably have been distributed from a supply center; no other explanation seems possible for such a large radius of proveniences. All the shapes in the two deposits have direct parallels in Dangstetten − a connection between Dangstetten and the Walensee towers would seem to be an unavoidable conclusion.

The state of research in Swiss territory for the period concerned (Roth-Rubi in press) Basle. In a large project, Eckard Deschler is currently studying the material from a canalisation trench in the Münsterhügel, where the first Roman settlement is located. His investigation will encompass earlier excavations as well. But it is hardly to be expected that his conclusions will deviate to any great extent from those of Andres Furger (Furger-Gunti 1979), based on the Münster excavation of the 1970s. Furger found that the first Roman imports, pottery and military implements, lay over a late La Tène stratum; the latter indicate an army post. So the earliest Roman presence here is military. The sigillata corresponds to that from Dangstetten; Aco beakers are also present.

To give a general view of the Swiss findspots with early Augustan material, I have devised a map showing the locations with Aco beakers, sigillata with radial stamping and the early sigillata shapes mentioned above (Fig. 1). The findspots in western Switzerland are not under discussion here; they illustrate the situation described by Wells (1972: 35ff) in connection with the opening of the Great St. Bernard pass. This route, leading from Aosta to Geneva, was a civil and not a military passage; its most important junction was Vidy, where a road branches off towards the Swiss Mittelland.

Zurich-Lindenhof. Here as well, new excavations have taken place since Emil Vogt’s work in the 1930s1. It would seem that the Lindenhof settlement was not established in early Roman times, but existed already in the later La Tène period. It is quite probable that, as in the case of BasleMünsterhügel, the earliest Roman imports are related to a military presence. I do not wish to anticipate the conclusions of the study in progress, but I do think that Vogt’s conclusions concerning the early Roman post will generally remain valid. The finds from the Lindenhof are in all respects comparable to those from Dangstetten.

The distribution area in north-eastern Switzerland has quite a different background: Basle, Vindonissa, Zürich and the Walensee towers are military posts dominated by Dangstetten. The finds from Zurzach are discussed elsewhere (Hänggi 1994: 140). Those from the colony of Augst, with a limited number of radial stamps and a few fragments of Aco beakers, are typical of a civilian settlement (see the list in Roth-Rubi in press). An isolated fragment of an Aco beaker was also found in a rural habitation near Möhlin am Rhein.3

Vindonissa. A. Hagendorn’s and Chr. Meyer-Freuler’s contribution in this volume should be consulted. Despite some new insights, the fact remains that the early periods in Vindonissa need to be thoroughly reassessed. New evidence has, however, come to light through a fresh examination of the finds from two of the three Walensee watch towers (Roth-Rubi in prep.). These towers2 are solid stone constructions situated on one end of the Walensee, well within sight of each other. Their massive walls, still preserved to a large extent, are astonishing; no parallels are known at present. The stepped base is especially remarkable. Inside the Biberlikopf tower, a water basin lined with mortar is preserved. A walled courtyard

There are two instances of Aco beakers, one from Chur (GR) (Hochuli 1991: 109, Taf 28, 1-3) and one from the Septimer Pass (GR) (Wiedemer 1966), which demand an explanation. No military presence seems indicated by the finds. The early periods of Chur are still largely in the dark; there is no early Arretine ware. As to the finds from the Septimer Pass, which Hans Rudolf Wiedemer once

1

Vogt (1948). It should be recalled that for a considerable time, the material from Zürich-Lindenhof remained the prime example of earliest Arretine ware and its formal development north of the Alps. 2 Biberlikopf: Jb SGUF 53 (1966/67): 151-156, Abb.38-42. Filzbach: Jb SGUF 48 (1960/61): 151ff.

3

Oral communication by F. Maier (Brugg). For the establishment: Jb SGUF 70 (1987): 225f.

510

Katrin Roth-Rubi: Why Dangstetten? proposed to associate with those from the Walensee towers, I was able to examine this material as well. It consists of tiny pottery fragments, impossible to classify with the exception of a small beaker fragment of Aco type in the shape of a woven basket. There is a parallel from the Solduno cemetery (Donati 1979: 96f.) south of the Alps, so it is presumably an import from a north Italian workshop. More precise chronological indications are not available. The two findspots in the Rhine valley and on the pass provide archaeological sources for an early Roman transalpine passage in Graubünden − no surprise from a historical point of view, since it means simply that the prehistoric passage endured.

necessitating fixed bases. It culminated in the glorified appearance of the two princes. Through the forced recruitment of young Raetians (Dio LIV.22.3-5), we are told, the situation was then calmed sufficiently to bring the campaign to an end. The military base became expendable and was removed. With this working hypothesis, I make a chronological distinction between Dangstetten and the military stations in Swiss territory on the one hand, and the operations in the Main and Lippe valleys on the other. The early Augustan chronology for sigillata and fine wares will have to be consolidated and refined, with evidence adduced from closed deposits. As a starting point, we have the lists of stamps made available in the expanded Oxé − Comfort and revised by Kenrick (2000), which open a whole new field of research. A first step has been accomplished with the comparison of the stamps from Oberaden (pub. in prep. Acta RCRF Lyon 2000). The results are promising.

To sum up: typical early pottery shapes appear in the Roman camp of Dangstetten, in the military stations of Basle-Münsterhügel, Zürich-Lindenhof, and in the Walensee towers. These stations are, then, contemporaneous with the camp. They lie on a west-east line beginning in Gaul and in a south-north direction which is directed towards the ‘Alpenrheintal’, the Engadin, and the Etsch valley. Dangstetten lies at the crossing of the two axis. Its position on the left bank of the Rhine guaranteed the supply line by water. I would like to emphasize again that the geographical layout I have just sketched, with Dangstetten at a central point connected with the west and the south, can be deduced from an examination of the archaeological finds.

Translation: K. Gex, Lausanne 15.11.2002 Literature Jb SGUF = Jahrbuch der Schweizerischen Gesellschaft für Urund Frühgeschichte RCRF = Rei cretariae fautores romanae Conspectus - Ettlinger E. et al 1990 Conspectus formarum terrae sigillatae italico modo confectae. Mat. zur röm.-germ. Keramik 10. Desbat A. et al 1996 Les productions des ateliers de potiers antiques de Lyon, 1ère partie: Les ateliers précoces. Gallia 53: 68-95. Donati P. 1979 Locarno. La necropoli romana di Solduno. Catalogo dei Materiali. Quaderni d’informazione 3. Fingerlin G. 1970-1971 Dangstetten, ein augusteisches Legionslager am Hochrhein. Vorbericht über die Grabungen 1967-1969. Berichte der RömischGermanischen Kommission 51-52: 197-232. Fingerlin G. 1986 Dangstetten I. Katalog der Funde (Fundstellen 1 bis 603). Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 22. (Stuttgart). Fingerlin G. 1998 Dangstetten II. Katalog der Funde (Fundstellen 604 bis 1358). Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 69. (Stuttgart). Furger-Gunti A. 1979 Die Ausgrabungen im Basler Münster I: Die spätkeltische und augusteische Zeit. Basler Beiträge zur Ur- und Frühgeschichte 6. (Basel). Hänggi R. et al 1994 Die frühen römischen Kastelle und der Kastell-Vicus von Tenedo-Zurzach. Veröffentlichungen der Gesellschaft Pro Vindonissa 11. Hochuli A. et al 1991 Chur in römischer Zeit. Band II: A. Ausgrabungen Areal Markthallenplatz. B. Historischer Überblick. Antiqua 19. Kienast D. 1999 Augustus. Princeps und Monarch. 3.erw. Auflage. (Wissenschaftliche Buchgesellschaft Darmstadt). Oxé A., Comfort H. & Kenrick P. 2000 Corpus vasorum arretinorum (2nd edit). Antiquitas Reihe 3. Roth-Rubi K. in press Zu den frührömischen Importen von Tafelgeschirr in das Gebiet der heutigen Schweiz. Eine

No one has ever really doubted that the Swiss military stations are associated with the so-called ‘Alpenfeldzug’ carried out by Augustus’ stepsons (already Vogt 1948: 35). In general, however, the accounts have attached themselves exclusively to the year 15BC, with rather positivistic confidence in a passage from Strabo, who informs us that the princes conquered the entire Alpine region of Raetia in one summer. That such a feat is impossible is obvious not only to a Swiss, which is why Wells (1972: 53ff.) would have the military occupation of the Lindenhof begin around 20BC. The appearance of the two princes, whose military credibility was to be demonstrated through this campaign, must certainly be seen as the crowning point of the whole operation − after which, says Strabo (Geog. IV.6.9 (206)), peace reigned among the Alpine populations. It is not part of my scope at present to bring the archaeological evidence into strict concordance with the historical sources, which are in any case difficult to interpret, and even less to argue for chronological revisions. But the new elements do suggest a working hypothesis which may at least enliven the discussion. I would argue that Dangstetten was built as a base station for the conquest of the Raetian Alps, in a preliminary stage of Drusus’ and Tiberius’ ‘Alpenfeldzug’. The military impulse was directed against the Alpine region; its aim was to conquer the strip of Alps connecting the eastern Transpadana and the Helvetic territory adjoining what was to become Raetia. This campaign was no lightning operation, completed in a summer, but a lengthy enterprise 511

Limes XVIII

Arbeitsskizze. In J. Poblome, R. Brulet & M. Waelkens (edd.) Early Italian Sigillata. The chronological framework and trade patterns. First International ROCT-Conference, May 1999, Leuven. Bull. Ant. Beschaving Supplements. Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Berichte der Römisch-Germanischen Kommission 66: 321-495. Vogt E. 1948 Der Lindenhof in Zürich. Zwölf Jahrhunderte Stadtgeschichte auf Grund der Ausgrabungen 1937/38. Wells C. 1972 The German policy of Augustus. An examination of the archaeological evidence. (Oxford). Wiedemer H.R. 1966 Die Walenseeroute in frührömischer Zeit. In R. Degen, W. Drack & R. Wyss (Hrsg.) Helvetia Antiqua. Festschrift Emil Vogt. (Basel): 167-172.

512

Katrin Roth-Rubi: Why Dangstetten?

Fig. 1. Map showing distribution of early Roman pottery imports in Switzerland. A = Aco beakers; circle = radial stamping on terra sigillata; triangle = early terra sigillata shapes (Conspectus 1, 2, 7, 8, 10, 11, 13).

513

Limes XVIII

Fig. 2. Possible line of deployment before the conquest of the Raetian Alps, based on military posts in northern Switzerland.

514

Kalkriese und die Varusschlacht – Archäologische Nachweise einer militärischen Auseinandersetzung zwischen Römern und Germanen Susanne Wilbers-Rost Seit 11 Jahren werden am Kalkrieser Berg nördlich von Osnabrück Ausgrabungen durchgeführt, die der Erforschung eines Schlachtfeldes zwischen Römern und Germanen dienen. Die archäologischen Untersuchungen erbrachten inzwischen immer mehr römische Münzen und Militaria aus augusteischer Zeit, die offenbar im Zusammenhang mit der Varusschlacht im Jahre 9 n.Chr. in den Boden gekommen sind. Über den Stand der Arbeiten bis 1995 wurde anlässlich einer Exkursion des Limes-Kongresses in Kerkrade berichtet. Seither sind interessante und unerwartete neue Funde und Befunde zutage gekommen. Das ausgedehnte Fundareal zwischen Wiehengebirge und Großem Moor wurde aufgrund neuer Fundstellen besser fassbar. Probegrabungen an verschiedenen Plätzen zeigten aber, dass aus einer Oberflächenfundstreuung nicht unbedingt auf einen umfangreichen Fundniederschlag und Befunde unter dem mittelalterlichen Plaggeneschauftrag geschlossen werden darf. Die Hauptfundstelle „Oberesch“ stellt noch immer in mehrfacher Hinsicht eine Besonderheit dar: neben mehr als 3000 Münzen und Militaria sind insbesondere die Befunde für die Interpretation der Kampfhandlungen von Bedeutung. Dazu zählt vor allem die von Germanen als Hinterhalt errichtete etwa 400m lange Wallanlage, von der inzwischen mehrere Abschnitte in Teilen untersucht werden konnten. Die Grabungen machten deutlich, dass es sich nicht nur, wie in den ersten Grabungsjahren vermutet, um eine einfache halbkreisförmige Rasensodenmauer handelt, sondern dass sich nach Osten mehrere bastionsartige Vorsprünge anschließen. Sie lassen auf gezielte Planungen und taktische Überlegungen durch die Germanen schließen. Im Zusammenhang mit den Befunden sind auch die Fundstreuungen auf dem Oberesch wichtig; anhand der Verteilung z.B. einzelner Sachgruppen soll in Zukunft versucht werden, nähere Einzelheiten über die Kämpfe, aber auch über die anschließenden Plünderungen zu ermitteln. Dass „Kalkriese“ nicht als eine einzige große Fundstelle, sondern als Fundareal mit unterschiedlichen Fundstellenqualitäten aufzufassen ist, ergibt sich im Vergleich z.B. zwischen dem Oberesch und einer Fundstreuung römischer Münzen und Militaria in einer etwa gleichzeitigen germanischen Siedlung. Während auf dem Oberesch eindeutig Kampfhandlungen nachgewiesen sind, ist bei den Funden aus der Siedlung zu fragen, ob es sich auch um einen Fundniederschlag aus Kämpfen an dieser Stelle handelt oder um Beute, die in die Siedlung verbracht und dort weiterverarbeitet wurde. Nicht zuletzt spielen auch die in den vergangenen Jahren auf dem Oberesch freigelegten Gruben mit Knochen von Menschen und Tieren eine wesentliche Rolle für die Erforschung des Schlachtgeschehens. Offenbar wurden diese nicht mehr im Skelettverband befindlichen Knochenreste, die längere Zeit auf der Bodenoberfläche gelegen hatten, einige Jahre nach der Schlacht mehr oder weniger sorgsam „bestattet“. Für diese Aktion kommen die Truppen des Germanicus in Betracht, doch stehen für eine endgültige Beantwortung der Frage zoologische und anthropologische Untersuchungsergebnisse noch aus. Sollte sich diese Vermutung bestätigen, wäre ein weiteres Indiz für die Identifizierung von Kalkriese als Ort der Varusschlacht gefunden. Die archäologischen Arbeiten der vergangenen Jahre haben gezeigt, dass mit den Funden und Befunden aus Kalkriese von Seiten der Archäologie ein wichtiger Beitrag zur Erforschung eines historisch überlieferten Schlachtfeldes geleistet werden kann. Darüber hinaus besteht hiermit die Chance, methodische Überlegungen für die Erforschung antiker Schlachtfelder zu erarbeiten.

Verbindung zu bringen sind. Vielmehr ist in diesem von den nächsten bekannten römischen Lagern weit entfernten Gebiet, wo in den vergangenen Jahren erstaunlich viele römische Münzen und Militaria bekannt geworden sind, von einer langgezogenen Feldschlacht auszugehen, die abseits von römischen Plätzen an einer topographisch als Engpass ausgewiesenen Stelle stattgefunden hat (Abb. 1).

Einleitung Während archäologische Untersuchungen zu römischen Grenzbefestigungen, Militäranlagen, aber auch zu Zivilsiedlungen einschließlich Gräberfeldern zum vertrauten Arbeitsfeld der provinzialrömischen Archäologie zu zählen sind, stellen Untersuchungen von Schlachtfeldern eher eine Ausnahme dar.

Auslöser für umfangreiche archäologische Untersuchungen war die Entdeckung eines römischen Münzschatzes am Rand des Großen Moores im Jahre 1987 bei Geländeprospektionen eines Amateurarchäologen mit dem Metalldetektor. Bereits in den vergangenen Jahrhunderten waren in diesem Gebiet wiederholt römische Silber- und Goldmünzen zutage gekommen. Sie waren von Bauern auf den Feldern beim Abstechen von Grassoden, die als natürlicher Dünger auf die Felder gebracht werden sollten, entdeckt worden (Berger 1996: 1f.). Diese als „Plaggenesch“ bezeichnete Wirtschaftsform war seit dem Hochmittelalter in weiten Teilen Norddeutschlands

Der nordwestliche Ausläufer des Weserberglandes, das Wiehengebirge mit seinem Übergang zur norddeutschen Tiefebene, bildet den naturräumlichen Rahmen, in dem wohl zum ersten Mal ein antikes Schlachtfeld mit archäologischen Methoden untersucht werden kann. Im Gegensatz zu Plätzen wie Alesia (Reddé & v. Schnurbein 1995) oder auch Krefeld-Gellep (Reichmann 1999) handelt es sich bei dem Fundareal von Kalkriese nördlich von Osnabrück allerdings nicht um Spuren von Kampfhandlungen, die mit Belagerung oder Angriff auf eine antike Stadtanlage bzw. befestigte Siedlung in

515

Limes XVIII

verbreitet.

Nach der Entdeckung einzelner römischer Funde auf der ehemaligen Oberfläche unter dem Esch wurden weitere Schnitte auf dem direkt benachbarten Feld angelegt. Sie lieferten nicht nur zahlreiche römische Münzen und Militaria – darunter u.a. die Gesichtsmaske eines römischen Helmes (Abb. 4; Franzius 1993: 131ff.; 1997: 16ff.) -, sondern ergaben als interessantesten Befund die spärlichen Reste einer Wallanlage (Abb. 5; Wilbers-Rost 1992; 1993; 2000b). Die anfängliche Interpretation als befestigter römischer Stützpunkt wurde nach kurzer Zeit aufgegeben, da der Wall sich als hangparallele Abschnittsbefestigung herausstellte. Zudem lag der größte Teil der römischen Funde auf der alten Oberfläche vor dem Wall. Diese Situation sowie die Fundtypen (überwiegend Teile von Militaria – Angriffs- und Verteidigungswaffen (Abb. 6), Pferdegeschirr, Wagenbeschläge, Werkzeuge, Trachtbestandteile - aus Metall, während Keramik nahezu fehlte) und die Art ihrer Fragmentierung führte zu der Überlegung, dass es sich hier um die Überreste eines Schlachtfeldes aus augusteischer Zeit handelt. Die weiträumige Fundstreuung, die sich über den Oberesch hinaus aus den Altfunden von Münzen wie auch aus der fortgesetzten Geländeprospektion ergab, ließ auf ein sehr ausgedehntes Kampfareal – mehr als 50km2 - schließen. Die Vermutung, dass hier das Areal der Varusschlacht entdeckt worden sein könnte, wurde durch die Datierung der Münzfunde nahegelegt, denn auch unter den zunehmend neben Silbermünzen (Abb. 7) entdeckten Kupfermünzen kamen keine nach dem Jahr 10 n.Chr. geprägten Münzen vor. Zugleich zeigten die Gegenstempel des Varus (VAR) auf einigen Kupfermünzen (Abb. 8) an, dass ein Verlust der Stücke nicht vor dem Jahr 7 n.Chr. denkbar war (Berger 1996: 58f.).

Kenntnis von diesen Münzfunden, die überwiegend von der ortsansässigen Adelsfamilie von Bar angekauft und in einer umfangreichen Sammlung archiviert wurden, erlangte um 1880 der Althistoriker Theodor Mommsen. Nach einer Sichtung und Bestimmung der damals auffindbaren Münzen durch den Numismatiker Julius Menadier kam Mommsen zu dem Schluss, dass die Münzen, die vor 10 n.Chr. geprägt waren, eindeutige Hinterlassenschaften aus der Varusschlacht seien. Seine Vorstellungen von der Varusschlacht „bei Barenaue“ wurden 1885 publiziert (Mommsen 1885), fanden aber wenig Rückhalt in der damaligen „Varus-Forschung“. Da eine Schlacht charakterisierende Militaria fehlten und Münzen zunächst eher als Hinweis auf Handelsverbindungen anzusehen sind, wurde die These Mommsens nicht näher überprüft, zumal nach dem Erscheinen seiner Publikation fast keine römischen Münzen mehr entdeckt wurden. Erst 1987 wurde der These Mommsens wieder verstärkt Aufmerksamkeit geschenkt, als der englische Offizier Major Clunn auf der Suche nach römischen Funden in der Nähe von Alt Barenaue einen Schatzfund von insgesamt 163 römischen Silbermünzen aufspürte. Die sofort von der Stadt- und Kreisarchäologie Osnabrück unter Leitung von Prof. Dr. Wolfgang Schlüter, mit dem Major Clunn von Anfang an zusammenarbeitete, durchgeführte Ausgrabung lieferte die Erkenntnis, dass die inzwischen leicht verpflügten Münzen gemeinsam, evtl. in einem organischen Behältnis, in den Boden gekommen sein mussten. Die Datierung durch den Numismatiker F. Berger – alle Münzen waren vor 10 n.Chr. geprägt – ließ die Vermutung aufkommen, die Münzen seien im Jahr der Varusschlacht 9 n.Chr. vergraben worden (Berger 1992: 84ff., 120f.).

In der archäologischen Forschung wurden danach zwei Wege beschritten: zum einen galt es, die Fundsituation auf dem Oberesch näher zu untersuchen und zu interpretieren, zum anderen musste die Ausdehnung des Areals anhand von Geländeprospektionen und Probegrabungen ermittelt werden.

Die Vorstellung, die in den antiken Quellen erwähnte, aber noch nie zuvor eindeutig lokalisierbare Varusschlacht, in der 9 n.Chr. drei römische Legionen von Germanen vernichtend geschlagen worden waren, könne hier in der Kalkrieser-Niewedder Senke, einem Engpass zwischen Wiehengebirge und Großem Moor (Abb. 2), stattgefunden haben, wurde zu diesem Zeitpunkt noch nicht in Erwägung gezogen. Zwar gab die Entdeckung von römischen Militaria und weiteren Münzen, darunter auch Kupfermünzen, bald zu der Überlegung Anlass, dass römische Truppen sich in diesem Gebiet aufgehalten haben mussten. Doch erst 1990, nachdem neben Geländeprospektionen mit Metalldetektoren systematische Ausgrabungen in Gang gebracht worden waren, wurde allmählich deutlich, welches umfangreiche Ereignis sich zwischen Bergland und Moor abgespielt haben musste.

Die Fundstelle „Oberesch“ Das Flurstück Oberesch, die bisher am intensivsten untersuchte, und noch immer ertragreichste Fundstelle der Forschungen in Kalkriese, umfasst ein Waldstück und einen ehemaligen Acker von insgesamt fast 100,000m2 Fläche am Hangfuß des Kalkrieser Berges. Nach Norden schließt sich an die hier knapp 100m breite trockene Hangsandzone eine erst in der Neuzeit trockengelegte Feuchtniederung an, die nach 1km in das eigentliche Moorgebiet übergeht. Neben der Hangsandzone war als zweiter passierbarer Weg durch den schmalen Engpass zwischen Kalkrieser Berg und Großem Moor bis in das 18. Jahrhundert nur ein trockener Flugsandrücken am Südrand des Moores ohne Schwierigkeiten begehbar. Hier verlief im Mittelalter der „Hellweg vor dem Santforde“ als Fernverkehrsverbindung zwischen Niederrhein und Mittelweser, und es ist anzunehmen, dass diese Ost-WestPassage zwischen Berg und Moor bereits in

Ein Waldstück auf dem sogenannten „Oberesch“ (Abb. 3) war das erste Grabungsareal, in dem geprüft werden sollte, ob römische Funde auch in primärer Lage unter dem mittelalterlichen Plaggenesch vorkamen, und wenn ja, worauf ihr Vorhandensein zurückzuführen sein könnte. 516

Susanne Wilbers-Rost: Kalkriese und die Varusschlacht

vorgeschichtlicher Zeit nutzbar war.

(Wilbers-Rost 1993: 63).

Die Lage des „Oberesches“ in der siedlungsgünstigen Hangsandzone sorgte für wiederholte Besiedlung vom Neolithikum bis in die erste Hälfte des 1. Jhs. v. Chr., wie zahlreiche Funde (Flint, Steinbeile, Keramik) und Befunde, insbesondere Vorratsgruben und Pfostenspeicher der vorrömischen Eisenzeit, belegen (Rost & Wilbers-Rost 1992; Wilbers-Rost 1991). Nach dem Auflassen der eisenzeitlichen Siedlung, die im Süden, wo der anstehende Kalkstein mit Lehm bedeckt ist, vermutlich von Wald begrenzt war, wurde wahrscheinlich zumindest ein Teil des Geländes als Weidefläche weiter genutzt.

Die ursprüngliche Sohlenbreite der Wallanlage betrug 45m; sie ist erkennbar anhand des Drainagegrabens an der Innenseite und einer dichten Streuung römischer Funde an der Vorderseite. Der fundfreie Streifen auf der ehemaligen Oberfläche zwischen diesen Befunden markiert etwa die Sohle der Wallanlage, unter der keine Funde liegen, da sie beim Verlust der Stücke, d.h. während der Kampfhandlungen, bereits stand. Ihre Höhe dürfte, soweit sich dies aus dem noch erhaltenen Wallmaterial hochrechnen lässt, nicht viel mehr als 1.5m betragen haben. Heute stellt sich diese Anlage als flache Erhebung von 10-40cms Mächtigkeit und 10-15m Breite unter dem Plaggeneschauftrag dar. Im Mittelalter lag sie vermutlich gerade außerhalb des beackerten Areals unter Wald, später wurde sie wie bereits das Gelände vor dem Wall von Esch bedeckt. Nur diese günstigen Umstände haben geholfen, Reste der ehemaligen Wallanlage bis heute zu konservieren, denn insbesondere die tief reichenden modernen Pflüge hätten sonst viele Funde und Befunde noch stärker zerstört.

Die topographisch und strategisch günstige Situation war wohl ausschlaggebend für die Auswahl dieses Platzes durch die Germanen bei ihren Plänen, römische Truppen im Jahre 9 n.Chr. in Kampfhandlungen zu verwickeln. Nach gezielter Vorplanung wurde hier offenbar kurzfristig eine Wallanlage errichtet, die aufgrund ihrer Lage an einer Waldkante relativ gut getarnt gewesen sein dürfte. Der unregelmäßige Verlauf dieser etwa 400m langen Anlage, bei der sich mehrere bastionsartige Vorsprünge abzeichnen (Abb. 3), ist noch nicht in allen Einzelheiten zu verstehen, doch dürfte die Konstruktion nicht nur der Verteidigung, sondern vor allem dem Angriff der Germanen in die Flanke der am Hang auf einem schmalen Nahverkehrsweg der Germanen entlang ziehenden Römer gedient haben. Für den Bau des Walles wurde das im näheren Umkreis vorhandene Material verwendet. So sind überwiegend Grassoden dort nachweisbar, wo Weidefläche vorhanden war, mehr Sand am östlichen Ende, weil dort möglicherweise Ackerfläche oder Baumbestand vorherrschte, eine Kombination aus Sand, Grassoden und Kalksteinen am westlichen Wallende, wo Kalksteine oberflächennah vorkommen. An der Innenseite wurden in mehreren Fällen Gruben bzw. Grabenabschnitte festgestellt, die offenbar die Aufgabe hatten, als Drainage Oberflächenwasser zu sammeln und abzuleiten (Abb. 9). Notwendig war diese Konstruktion, weil Lehm und Festgestein hinter dem Wall das Versickern von Regenwasser verhinderten, was bei starkem Regen zu einem Unterspülen und Einsturz der Wallanlage hätte führen können. Ob die Gruben mit Holz abgedeckt oder mit Reisig gefüllt waren, um das Besteigen des Walles an der Innenseite nicht zu behindern, konnte bisher nicht geklärt werden.

Abgesehen von der Wallanlage sind Befunde, die im Zusammenhang mit dem Schlachtgeschehen stehen, spärlich. Interessant und anfangs völlig unerwartet waren Knochenfunde, die in den letzten Jahren immer häufiger auf dem Oberesch freigelegt wurden (Wilbers-Rost 1999). In den ersten Grabungsschnitten waren lediglich Einzelknochen bzw. Knochenfragmente entdeckt worden, darunter Tierzähne und Teile eines menschlichen Schädels. Ihre Erhaltung im Sandboden ließ sich nur durch die Nähe zu Kalksteinen erklären. Verblüffend war 1992 die Entdeckung von Skelettresten (Schädel, Wirbel und Schulterblatt) eines Maultieres, für deren Konservierung in unmittelbarer Nähe liegende Metallteile von der Anschirrung (Eisenkette, Trense, eine große Bronzeglocke) gesorgt haben (Rost & Wilbers-Rost 1993). Möglicherweise hatte sich das Tier von einem Wagen mit Teilen seines Geschirrs losgerissen und war vor dem Wall gestürzt, wo Grassoden es vor Tierverbiss und Plünderung durch die Germanen verborgen haben. Schon bei der Bergung dieses Fundkomplexes wurde deutlich, welche außergewöhnlichen Befunde das Schlachtfeld bereit hält: in Kalkriese hat der zusammengebrochene Wall für die Erhaltung von „Momentaufnahmen“ gesorgt, wie sie sonst in der Archäologie nur selten zu entdecken sind. Mit der Freilegung einer weiteren Ansammlung von Knochen kam 1994 ein neuer Aspekt bei der Interpretation des Schlachtfeldes ins Spiel: Reste von Tier- und Menschenknochen, die ohne anatomischen verband in einer Grube vor dem Wall deponiert worden waren (Abb. 10), ließen sich als Überreste von Toten der Schlacht interpretieren, denn mehrfach wurden kleinere römische Funde (u.a. Nägel und Fibelfragmente), die den auf der alten Oberfläche im Umfeld der Grube geborgenen Funden entsprachen, auch zwischen den Knochen entdeckt.

An mehreren Stellen wurde das gesammelte Wasser mit Hilfe von Durchlässen im Wall nach außen abgeleitet. In einem Fall konnte zudem eine regelrechte Torkonstruktion mit Pfosten nachgewiesen werden. Sie war vermutlich innen zu schließen, sodass die Germanen hier eine Möglichkeit besaßen, auf das Schlachtfeld auszubrechen, sich aber auch schnell in den sicheren Schutz der Anlage zurück zu begeben. An beiden Seiten des Tores war ein Stück der Wallanlage durch eine Brustwehr verstärkt, wie Pfostenspuren, die parallel zur Wallfront verlaufen, zeigen

517

Limes XVIII

Offenbar waren diese Knochen jedoch nicht sofort nach dem Kampf vergraben worden, sondern hatten noch einige Zeit auf der Oberfläche gelegen. Die Art der Erhaltung und die Fragmentierung zeigen, dass Weichteile und Sehnen völlig vergangen gewesen sein müssen, bevor die Knochen deponiert wurden. Nach ersten anthropologischen und archäozoologischen Untersuchungen betrug der Verbleib an der Oberfläche zwischen 2 und 10 Jahren (freundliche Information von Prof. Dr. Dr. Hans-Peter Uerpmann, Universität Tübingen). Die Vermutung, dass diese Knochengruben mit den Aktivitäten der Römer unter Germanicus 16 n.Chr. in Verbindung zu bringen sind, die die Reste der Varus-Legionen nach Aussage der antiken Schriftquellen (Tacitus Anns. I.60-62) bestattet haben sollen, ist inzwischen immer wahrscheinlicher geworden.

Voraussetzungen hier überhaupt Knochen erhalten bleiben konnten, noch nicht endgültig geklärt ist, wird gerade an den Knochengruben deutlich, dass auf dem Oberesch nicht nur Aussagen zum Kampfgeschehen, sondern auch zu den Ereignissen nach der Schlacht gemacht werden können. Die Anlage dieser Knochengruben mit dem Besuch des Germanicus auf dem Schlachtfeld in Verbindung zu bringen, erscheint beim derzeitigen Bearbeitungsstand sehr viel plausibler als sie auf die Germanen zurückzuführen. Eine „Bestattung“ der Toten wäre allenfalls direkt im Anschluss an die Kampfhandlungen für die Germanen vorteilhaft gewesen, um den Verwesungsgeruch auf dem Schlachtfeld zu mildern und die siedlungsgünstigen Areale weiter nutzen zu können. Grundsätzlich darf bei der Interpretation der Spuren auf dem Schlachtfeld aber auch ein besonderer Umgang der Germanen mit dem Ort der Schlacht nicht außer Acht gelassen werden; denkbar wäre z.B. eine Sonderstellung des Platzes als „heiliger Hain“.

Eine Bestätigung dieser schon bald nach der Auffindung der ersten Grube entstandenen These ergab sich bei den weiteren Ausgrabungen auf dem Oberesch, bei denen inzwischen insgesamt fünf Knochengruben entdeckt worden sind. In Größe und Anzahl der Knochen unterschieden sie sich zwar, doch kamen in allen Gruben Tier- und Menschenknochen vermischt vor. Besonders eindrucksvoll war die Grube 5 (Abb. 11), die in einer natürlichen kleinen Mulde angelegt und auf der Sohle wie an der Wand mit Kalksteinen ausgekleidet worden war; so entstand der Eindruck einer regelrechten Grabgrube, zumal die Knochen nicht willkürlich hineingeworfen, sondern sorgfältig niedergelegt zu sein scheinen. Zwischen den Knochen befanden sich weitere Kalksteine, die für die extrem gute Erhaltung der Knochen verantwortlich sind; in der Mitte und knapp oberhalb des Grubenbodens lagen menschliche Schädel, einer davon mit einer Hiebverletzung.

Auf andere Ursachen hingegen dürfte ein Befund mit Knochen zurückzuführen sein, der im Sommer 2000 am westlichen Wallende freigelegt wurde. Hinter dem Wall, der an dieser Stelle aus Sand errichtet und an der Vorderfront offenbar mit Kalksteinen verstärkt war, fand sich das nahezu vollständige Skelett eines Maultieres (Abb. 12). Es lag ausgestreckt mit dem Kopf nach Westen und den Läufen nach Süden; am Kiefer lagen noch zwei eiserne Trensenringe, eine ursprünglich am Hals befestigte Bronzeglocke war bereits im Vorjahr bei der Geländeprospektion entdeckt worden. Das Tier war offenbar über den Wall gerannt und hatte sich beim Sturz das Genick gebrochen (Information Prof. Uerpmann). Es muss sehr bald von Wallmaterial verdeckt worden sein, sodass es weder von Wildtieren verschleppt noch die Metallteile geplündert werden konnten. Auch bei diesem Befund wird, wie bei dem ersten Maultier, die Konservierung eines Augenblicks aus den eigentlichen Kampfhandlungen deutlich; er unterscheidet sich damit grundsätzlich von den beschriebenen Knochengruben.

Mit diesen Knochengruben ergab sich eine zusätzliche Bestätigung für die Erklärung der Funde und Befunde auf dem Oberesch als Relikte eines Schlachtfeldes. Dazu tragen auch die Ergebnisse der anthropologischen und der zoologischen Untersuchungen bei. Unter den Menschenknochen sind bisher ausschließlich Knochen von gut ernährten Männern zwischen 20 und 40 Jahren vertreten. Einige der Knochen weisen Verletzungen durch Schwerthiebe auf. Das Gesamtbild entspricht somit nicht einer Normalpopulation, sondern dem, was von in Kämpfe verwickelten römischen Legionen zu erwarten ist.

Neben den konkreten Befunden und den römischen Funden gibt vor allem die Verteilung der Funde im Untersuchungsareal weitere Aufschlüsse über die Abläufe der Kampfhandlungen und die anschließenden Plünderungen. Eine erste Kartierung der etwa 3500 römischen Funde (darunter fast 300 Münzen) auf dem Oberesch zeigt ein sehr unterschiedliches Verteilungsbild. Während sich im Vorfeld des Walles eine lockere Fundstreuung abzeichnet, gibt es unmittelbar vor der Außenflanke im allgemeinen eine dichtere Konzentration. Dies dürfte darauf zurückzuführen sein, dass das Gelände vor dem Wall vor 2000 Jahren soweit vegetationsfrei war, dass die gut sichtbaren Stücke bis auf kleinere oder in den Boden getretene Teile bei den Plünderungen durch die Germanen nach der Schlacht weitgehend eingesammelt werden konnten. Funde direkt am Wallfuß hingegen wurden wahrscheinlich schon während der Kämpfe und im Anschluss daran von abrutschendem lockeren Wallmaterial überdeckt, sodass sie bei den Plünderungen übersehen

Bei den Tierknochen dominieren Maultiere, doch sind auch einzelne Pferde nachweisbar. Reste von Nahrungsabfällen (Rind oder Schwein) fehlen vollständig, sodass die Tierknochen als Hinweis auf Tross und Reiterei zu deuten sind. Auch dieses Ergebnis fügt sich gut ein in das Bild eines Schlachtfeldes, auf dem römische Truppen beteiligt waren. Für eine Verbindung mit der Varusschlacht wiederum spricht die Bestattung der Knochenreste erst einige Jahre nach dem Geschehen. Wenn auch die Bearbeitung der Knochen noch nicht abgeschlossen ist und die Frage, wie lange die Knochen auf der Oberfläche lagen oder unter welchen 518

Susanne Wilbers-Rost: Kalkriese und die Varusschlacht

wurden.

fehlende Prospektion zurückzuführen sein können. Deutlich wurde jedoch, dass sich viele Fundstellen am Hang des Kalkrieser Berges auf der trockenen Hangsandzone erstrecken. Hinzu kommt eine Kette von Fundstellen im Bereich des Flugsandrückens südlich des Großen Moores, aber auch im ehemaligen Feuchtgebiet zwischen Berg und Moor wurden Funde entdeckt. Aus diesen Fundstellen ergibt sich der wahrscheinliche Marschweg der Römer, die, aus dem Osten kommend, den Engpass nach Westen passieren wollten, um weiter Richtung Lippe und Rhein zu gelangen. Möglicherweise hat sich der langgestreckte Zug an der engsten Stelle des Geländes aufgeteilt, wobei ein Teil nach Nordwesten gezogen sein könnte (Schlüter 1999: 40ff.), doch lässt sich dies erst eindeutig beurteilen, wenn auch der Moorrand nach Osten prospektiert worden ist.

Ausnahmen von diesen „Normalverteilungen“ gibt es jedoch auch. Treten an einigen Stellen in größerer Entfernung vor dem Wall dichtere Konzentrationen von Funden auf, ist zu prüfen, ob dort stärkere Kampfhandlungen stattfanden oder aber ob andere Gründe zu derartigen Fundansammlungen geführt haben; so könnte z.B. eine Häufung von Fragmenten auch durch das Überrollen größerer Gegenstände von Trosswagen verursacht worden sein. An mehreren Durchlässen wurden auch im Drainagegraben hinter dem Wall Konzentrationen von römischen Funden entdeckt. Hier könnten Gegenstände bei den Kämpfen in den Graben gefallen und liegen geblieben sein. Nicht auszuschließen ist allerdings auch, dass Beute vom Schlachtfeld zunächst im Schutz des Walles deponiert und dann weiter verteilt und abtransportiert wurde; in diesem Zusammenhang bleibt aber noch zu klären, wie die Plünderungen vor sich gegangen sind und nach welchen Kriterien Beute eingesammelt und verteilt worden sein könnte. Möglicherweise ergeben sich aus der Auswertung der Grabungen auch dazu weitere Anhaltspunkte.

Die Probegrabungen auf verschiedenen ProspektionsFundstellen waren nicht immer von Erfolg begleitet. Gelegentlich sind, aufgrund der Veränderungen der Oberfläche durch Beackerung oder Plaggenabtrag, Funde nur in der heutigen Pflugschicht erhalten. Befunde aus dem Kontext der Schlacht, wie sie der Oberesch mit der Wallanlage und den Knochengruben ergeben hat, konnten bisher an anderen Plätzen nicht eindeutig ermittelt werden. 2kms westlich des Oberesches wurden römische Funde im Bereich einer vorgeschichtlichen Siedlung entdeckt, doch bleibt zu klären, inwieweit diese Objekte Rückschlüsse auf das Kampfgeschehen erlauben. Einige Funde stammen aus Siedlungsgruben (Harnecker 2001), und unter dem übrigen Fundmaterial dieses Platzes kommen zahlreiche offenbar der Metallverarbeitung zugeführte Stücke (Schmelzkügelchen, abgeschrotete Bronzeteile) vor. Beim derzeitigen Bearbeitungsstand sind deshalb zumindest einige dieser Fundstücke eher als Hinweis auf die Weiterverarbeitung von Beutestücken aus der Schlacht denn als Indiz für Kampfhandlungen am Fundort selbst anzusehen.

Das Areal der Schlacht Insbesondere die Ausgrabungen auf dem Oberesch haben gezeigt, dass ein Schlachtfeld völlig eigenen Überlieferungsbedingungen unterliegt, sodass geeignete Untersuchungs- und Interpretationsmethoden gefunden und teilweise im Zuge der Auswertung noch entwickelt werden müssen; die Beschreibung der Funde und Befunde allein ermöglicht keine Aussagen zum Geschehen vor 2000 Jahren. Der Oberesch ist jedoch nur eine Fundstelle in einem ausgedehnten Fundgebiet, also nur ein Teil eines sehr viel größeren Schlachtareals. Erst der großräumige Vergleich sämtlicher bisher bekannter Fundstellen in der Region Kalkriese, die Funde und gegebenenfalls Befunde aus der Schlacht ergeben haben, ermöglicht eine Bewertung der einzelnen Plätze im Gesamtgeschehen.

Bisher ist demnach der Oberesch die einzige Fundstelle mit eindeutig auf Kampfhandlungen zurückzuführenden Befunden. Da dieser Platz an der engsten Stelle zwischen Berg und Moor liegt, könnten hier von den Germanen die umfangreichsten Vorbereitungen für die Schlacht getroffen worden sein und ein wesentlicher Brennpunkt der Kampfhandlungen gelegen haben. Hier konnten aufgrund der ohnehin günstigen topographischen Situation die Voraussetzungen geschaffen werden, die vorbeiziehenden Römer erfolgreich anzugreifen.

Zahlreiche weitere Fundstellen wurden bei der Prospektion mit Metalldetektoren ermittelt (Schlüter 1993: 30ff.; 1999; 2000). In einigen Fällen wurden lediglich Einzelfunde oder Konzentrationen von Münzen – evtl. Hortfunde oder Barschaften römischer Soldaten (Berger 2001: 15ff.) entdeckt; andere Stellen lieferten neben Münzen einzelne Militaria (Teile von Waffen, Bruchstücke von Pferdegeschirr) oder Fragmente z.B. von Silbergefäßen. Auch die z.T. mit Steineinlagen verzierten silbernen Beschläge einer Schwertscheide wurden bei der Prospektion entdeckt (Franzius 1999).

Ein dichter Fundniederschlag wie auf dem Oberesch ist jedenfalls trotz zahlreicher Probegrabungen am Hang des Berges und am Rand des Moores an anderen Plätzen bisher nicht nachweisbar.

So ergaben sich im Gelände inzwischen mehrere Schwerpunkte mit Funden (Abb. 2), wobei zu berücksichtigen ist, dass die Begehungen nicht flächendeckend durchgeführt werden konnten und Lücken in der Fundstellenverteilung in einigen Gebieten auf

Um den Umfang der Kampfhandlungen und z.B. die Anzahl der beteiligten Römer und Germanen abschließend beurteilen zu können, sind daher umfassende methodische und quellenkritische Überlegungen notwendig. Ausgehend

519

Limes XVIII

von den bisher ergrabenen Funden und Befunden ist zu klären, welche Prozesse die Überlieferungsqualität beeinflusst haben und in welchem Ausmaß und auf welche Weise insbesondere die Plünderungen den Fundniederschlag nicht nur reduziert, sondern auch manipuliert haben. So lassen sich Voraussetzungen schaffen, um über die schriftlichen Überlieferungen hinaus dem historisch belegten Ereignis der Varusschlacht mit archäologischen Mitteln weiter auf die Spur zu kommen und das tatsächliche Geschehen detaillierter als bisher möglich rekonstruieren zu können.

Kommission des Deutschen Archäologischen Instituts 76: 73-158. Reichmann Ch. 1999 Archäologische Spuren der sogenannten Bataverschlacht vom November 69 n.Chr. und von Kämpfen des 3. Jahrhunderts n.Chr. im Umfeld des Kastells Gelduba (Krefeld-Gellep). In R. Wiegels & W. Schlüter (edd.) Rom, Germanien und die Ausgrabungen von Kalkriese. Akten des Internationalen Kongresses vom 2. bis 5. September 1996 an der Universität Osnabrück. Osnabrücker Forschungen zu Altertum und Antikerezeption 1. (Osnabrück): 97-115. Rost A. & Wilbers-Rost S. 1992 Die vorgeschichtliche Besiedlung am Kalkrieser Berg zwischen Engter und Schwagstorf. In W. Schlüter Archäologische Zeugnisse zur Varusschlacht? Die Untersuchungen in der Kalkrieser-Niewedder Senke bei Osnabrück. Germania 70 /2: 344-349. Rost A. & Wilbers-Rost S. 1993 Fragmente eines römischen Zugtieres mit Resten der Anschirrung. In W. Schlüter (ed.) Kalkriese - Römer im Osnabrücker Land. Archäologische Forschungen zur Varusschlacht. (Bramsche): 199-209. Schlüter W. 1992 Archäologische Zeugnisse zur Varusschlacht? Die Untersuchungen in der KalkrieserNiewedder Senke bei Osnabrück. Germania 70/2: 307402. Schlüter W. (ed.) 1993 Kalkriese - Römer im Osnabrücker Land. Archäologische Forschungen zur Varusschlacht. (Bramsche). Schlüter W. 1999 Zum Stand der archäologischen Erforschung der Kalkrieser-Niewedder Senke. In R. Wiegels & W. Schlüter (edd.) Rom, Germanien und die Ausgrabungen von Kalkriese. Akten des Internationalen Kongresses vom 2. bis 5. September 1996 an der Universität Osnabrück. Osnabrücker Forschungen zu Altertum und Antikerezeption 1. (Osnabrück): 13 - 60. Schlüter W. 2000 Varus-Schlacht: 12 Jahre archäologische Forschung in Kalkriese. In Archäologie in Niedersachsen 2000. (Oldenburg): 42-45. Wilbers-Rost S. 1991 Die Besiedlung der Kalkrieser-Niewedder Senke in vorgeschichtlicher Zeit. In W. Schlüter Römer im Osnabrücker Land. Die archäologischen Untersuchungen in der Kalkrieser-Niewedder Senke. Schriftenreihe Kulturregion Osnabrück des Landschaftsverbandes Osnabrück e.V. 4. (Bramsche): 15-17. Wilbers-Rost S. 1992 Grabungsbefunde auf dem „Oberesch“ in Kalkriese, Stadt Bramsche, Landkreis Osnabrück. In W. Schlüter Archäologische Zeugnisse zur Varusschlacht? Die Untersuchungen in der KalkrieserNiewedder Senke bei Osnabrück. Germania 70/2: 332335. Wilbers-Rost S. 1993 Geschichte und Ergebnisse der Ausgrabungen in Kalkriese. In W. Schlüter (ed.) Kalkriese - Römer im Osnabrücker Land. Archäologische Forschungen zur Varusschlacht. (Bramsche): 53-72. Wilbers-Rost S. 1999 Die Ausgrabungen auf dem »Oberesch« in Kalkriese: Deponierungen von Menschen- und Tierknochen auf dem Schlachtfeld”. In R. Wiegels & W. Schlüter (edd.) Rom, Germanien und die Ausgrabungen von Kalkriese. Akten des Internationalen Kongresses vom 2. bis 5. September 1996 an der Universität Osnabrück. Osnabrücker Forschungen zu Altertum und Antikerezeption 1. (Osnabrück): 61-89. Wilbers-Rost S. 2000a Archäologische Grabungsplätze–

Allerdings würden bei einer Reduzierung allein auf die Frage, ob es sich bei den Funden in Kalkriese um die Reste der Varusschlacht handelt, viele Chancen der Analyse eines antiken Schlachtfeldes vertan. Nicht zuletzt am Beispiel der Knochengruben wird deutlich, welche zusätzlichen Aspekte beispielsweise der Aktivitäten von Germanen und Römern nach der Schlacht durch archäologisch-naturwissenschaftliche Untersuchungen erschlossen werden können. Literatur Berger F. 1992 Untersuchungen zu römerzeitlichen Münzfunden in Nordwestdeutschland. Studien zu Fundmünzen der Antike 9. (Berlin). Berger F. 1996 Kalkriese 1. Die römischen Fundmünzen. Römisch-Germanische Forschungen 55. (Mainz). Berger F. 1999 Kalkriese: Die römischen Fundmünzen. In R. Wiegels & W. Schlüter (edd.) Rom, Germanien und die Ausgrabungen von Kalkriese. Akten des Internationalen Kongresses vom 2. bis 5. September 1996 an der Universität Osnabrück. Osnabrücker Forschungen zu Altertum und Antikerezeption 1. (Osnabrück): 271-277. Berger F. 2001 Die Münzen von Kalkriese. Neufunde und Ausblick. In R. Wiegels (ed.) Die Fundmünzen von Kalkriese und die frühkaiserzeitliche Münzprägung. Akten des wissenschaftlichen Symposions in Kalkriese, 15.-16. April 1999. Osnabrücker Forschungen zu Altertum und Antikerezeption 3. (Osnabrück): 11-45. Franzius G. 1993 Die römischen Funde aus Kalkriese. In W. Schlüter (ed.) Kalkriese - Römer im Osnabrücker Land. Archäologische Forschungen zur Varusschlacht. (Bramsche): 107-197. Franzius G. 1997 Die römischen Funde und Münzen aus Kalkriese, Ldkr. Osnabrück, Deutschland, der Jahre 1987-1996. In A. Nørgård Jørgensen & B.L. Clausen (edd.) Military aspects of Scandinavian society in a European perspective AD 1-1300. Papers from an International Research Seminar at the Danish National Museum, Copenhagen 1996. Vol. 2. (Copenhagen). 7692. Franzius G. 1999 Beschläge einer Gladiusscheide und eines cingulum aus Kalkriese, Landkreis Osnabrück. Germania 77/2: 567–608. Harnecker J. 2001 Die Suchgrabungen in Kalkriese. (Publikation in Vorbereitung). Mommsen T. 1885 Die Örtlichkeit der Varusschlacht. (Berlin). Reddé M. & von Schnurbein S. u.a. 1995 Neue Ausgrabungen und Forschungen zu den Belagerungswerken Caesars um Alesia (1991-1994). Mit einem Beitrag von Susanne Sievers. Bericht der Römisch-Germanischen

520

Susanne Wilbers-Rost: Kalkriese und die Varusschlacht

Bramsche-Kalkriese: Archäologische Kennzeichnung. In Schutz archäologischer Funde aus Metall vor immissionsbedingter Schädigung. Materialhefte zur Archäologie in Baden-Württemberg 57. (Stuttgart): 8690. Wilbers-Rost S. 2000b Varusschlacht – Untersuchungen auf dem „Oberesch“ in Kalkriese. In Archäologie in Niedersachsen 2000. (Oldenburg): 42-45. Wilbers-Rost S. 2001 – forthc. Archäologische Forschungen zur Varusschlacht in Kalkriese bei Osnabrück. Bulletin de la Société Française d`Archéologie Classique.

521

Limes XVIII

Abb. 1. Die Lage des Fundgebietes „Kalkriese“. 1 - Marsch; 2 - Moor; 3 - Geest; 4 - Mittelgebirge über 100m NN.

Abb. 2. Der Engpass der Kalkrieser-Niewedder Senke zwischen Wiehengebirge und Großem Moor. 1 - Berg- und Hügelland; 2 - Tiefland; 3 Flussauen; 4 - Moor; 5 Wege des 18. und 19. Jhs.; 6 - Fundstellen mit römischen Funden.

522

Susanne Wilbers-Rost: Kalkriese und die Varusschlacht

Abb. 3. Kalkriese-Oberesch, Übersichtsplan mit Grabungsschnitten bis 2000, Wall und Knochengruben.

Abb. 4. Oberesch, Gesichtsmaske eines römischen Helmes (Eisen, ursprünglich mit Silberblech belegt; L 16,9 cm).

523

Limes XVIII

Abb. 5. Oberesch, Grabungsschnitte 1990. Im Planum ist leicht gebogen der Wall als breites, dunkles Band zu erkennen.

Abb. 6. Oberesch, Römische Waffen: Lanzen- und Speerspitzen, Lanzenschuh, Pilumzwinge, Geschützbolzen; L der größten Lanzenspitze 20.5cms.

Abb. 7. Kalkriese, Hortfund Silbermünzen (Denaren).

524

von

Susanne Wilbers-Rost: Kalkriese und die Varusschlacht

Abb. 8. Oberesch, Kupfermünze (As) mit Gegenstempel.

Abb. 9. Oberesch, Schnitt 29, Blick auf das Ostprofil von Süden. Unter einer mächtigen Eschdecke zeichnen sich Wall und Drainagegraben (am Maßstab)ab.

Abb. 10. Oberesch, Schnitt 24, „Große Knochengrube“. Bereits die obere Knochenlage lässt ein Durcheinander von Knochen und Knochenfragmenten erkennen.

525

Limes XVIII

Abb. 11. Oberesch, Schnitt 22p, Knochengrube 5. In dieser mit Kalksteinen ausgekleideten Grube wurden die Knochen sehr sorgfältig niedergelegt. Ein menschlicher Schädel liegt in der Mitte der Grube.

Abb. 12. Oberesch, Schnitt 32. Skelett eines römischen Maultieres, das bei den Kampfhandlungen durch einen Genickbruch umkam und mit eiserner Trense und Bronzeglocke bis heute fast vollständig erhalten geblieben ist.

526

Neue Grabungen in Haltern, Oberaden und Anreppen Siegmar von Schnurbein Herr Dr. J.S. Kühlborn, der die Grabungen in den Lippelagern seit 1978 leitet, kann leider nicht zum Kongreß kommen und bat mich, ber die seit dem Kongreß 1995 erzielten Resultate in den verschiedenen Lagern zu berichten. In Haltern is die wichtigste neue Entdeckung ein offensichtlich nur kurze Zeit existierendes Lager, das etwa 1500m nordöstlich des Hauptlagers lag. Die präzise Datierung ist noch nicht möglich; bisher entsprechen die Funde dem bekannten Spektrum von Haltern. Im Gegensatz zu Haupt- und Feldlager von Haltern hat es polygonale Form, von der bisher aber nur die südliche und westliche Front erkannt sind. Das Lager war groß: In Nord- Südrichtung maß es mindestens 500m. In Oberaden sind südlich der principia weitere Wohngebäude mit Peristylhöfen untersucht worden. Besonders auffallend ist dabei ein Haus, das in die via praetoria gebaut worden ist, wodurch diese von 42 auf nur 12m Breite reduziert worden ist. -Auch in Beckinghausen wurden größere Flächen untersucht, die die alten Pläne im wesentlichen bestätigen. In Anreppen sind östlich des Südtores große Flächen untersucht worden. Dort wurde ein 56 : 68m großes Horreum freigelegt, daneben auch weitere Kasernen. Ferner ist nun auch die Lage des Osttores gesichert. Bemerkrnswert sind neben dem Ost- und dem Südtor or in der via sagularis errichtete Gebäude, ein Befund, der auch neben der porta praetoria von Oberaden begegnete. Die Funktion dieses nu_euen Gebäudetyps ist noch nicht klar. In Anreppen gefundene Keramik, die mit den Halterner Produkten übereinstimmt, beweist zwar die gleichzeitige Existenz, aber Anreppen wurde wohl sicher später errichtet und vielleicht auch schon vor 9 n. Chr. nicht mehr genutzt.

Zum Limeskongress 1995 in Kerkrade hat Johann Sebastian Kühlborn in dem handlichen Buch Germaniam pacavi – Germanien habe ich befriedet den Forschungsstand zu den augusteischen Lippelagern prägnant zusammengefaßt (Kühlborn 1995). Seither haben in Haltern, in Oberaden und Beckinghausen sowie in Anreppen neue Grabungen stattgefunden, über die hier kurz berichtet werden soll (Abb. 1). Herr Kühlborn hat mir die entsprechenden Unterlagen zur Verfügung gestellt und ich trage die Resultate hier natürlich in erster Linie in seinem Namen vor (Kühlborn 1998; 1999; 2000). Da sich in den chronologischen Etappen nichts geändert hat, beginne ich mit der Drusus-Zeit, d. h. mit Oberaden und Beckinghausen. War für mich vor 20 Jahren die Gleichzeitigkeit beider Anlagen noch nicht genügend abgesichert (v. Schnurbein 1981: 25), so ist dies dank der neuen Ausgrabungen hinreichend geklärt; beide Plätze gehören den Jahren von 11 v. bis c.8 v. Chr. an.

Atrium-Häuser aufgedeckt werden. Die Portikus vor Bau 4 im Nordwesten paßt bestens zu der Wohnhaus-Funktion und spiegelt die Absicht, städtisches Bauen zu kopieren. Die südlich gelegenen Bauten 3a und 3b bezeichnet Kühlborn mit Vorbehalt als „Legatenhäuser“. Dies ist eine berechtigte Arbeitsbezeichnung, die freilich nicht abgesichert werden kann, solange die übrigen Zentralbereiche des Lagers unerforsht sind und die Zusammensetzung der Truppe unbekannt ist (Neujahrsgruß 1999: 96). Zwar folgen die Bauten einem einheitlichen Schema mit quadratischem Wohnteil und vorgelagertem Peristylhof, sie sind aber nicht gleich groß. Bei 3a sind die Wohnteile 18 : 18m bzw c.20 : 20m groß, die Höfe 18 : 20m bzw. 13 : 17m. Nicht gedeutet werden können im großen Hof – abgesehen von dem Peristyl – die zahlreichen Gruben. Bei 3b mißt der Wohnteil 27: 23m, der Hof 25 : 24m. Eindeutig nachträglich ist westlich an 3b ein kleineres Wohngebäude gleicher Konzeption angefügt worden. Durch diesen Anbau wird die via prateroria von 27m auf 12m Breite verengt.

In Oberaden konzentrierten sich die Grabungen auf den Bereich zwischen den principia und der porta praetoria (Abb. 2). Das Areal der 94: 103m großen principia ist nun vollständig untersucht (Abb. 3): Fehlstellen im Plan beruhen auf alten Störungen, besonders im südlichen Teil. Neu sind die im Westen und Süden z. T. doppelt geführten Fundamentgräben was bisher nicht plausibel erklärt werden kann: auch die leichten Anbauten, die im Nordwesten außen angefügt sind, können in ihrer Funktion bisher nicht gedeutet werden. Bemerkenswert sind die Abwassergräben, die den Hofbereich der principia durchziehen, ebenso die Reste schmaler Wandgräbchen, die keine klaren Grundrisse ergeben. Sie müssen zu älteren Bauten gehören, die vor Beginn der Errichtung der principia hier kurze Zeit existierten, vielleicht also eine Art von Bauhütten.

Bemerkenswerterweise zieht dieser schmale Rest der via praetoria nicht direkt auf die porta praetoria, eine Unregelmäßigkeit der Gesamtkonzeption von Oberaden, die auch bei der via principalis in Bezug zu den PrincipaToren auffällt (Kühlborn & v. Schnurbein 1992: 89; v. Schnurbein 2000: 31). Der neu ausgegrabene Grundriß der porta praetoria (Abb. 4) bestätigt im wesentlichen die alten Grabungspläne von 1912 (Kühlborn & v. Schnurbein 1992: 89ff.). Freilich gibt es eine singuläre Ergänzung: Mittig zwischen den Köpfen der Holz-Erde-Mauer fand sich ein zusätzliches Pfostenpaar. Der Zugang ins Lager könnte also zwei mal überbrückt gewesen sein, es sei denn, es handelt sich um zwei Bauphasen. Dann entspräche die eine dem Typ II b nach Manning und Scott, die zweite Typ V, 1 (Manning &

Westlich und südlich der principia konnten weitere

527

Limes XVIII

Scott 1979). Wichtige Ergänzungen sind die beidseits des Tores entdeckten Pfosten für Treppen oder Rampen. Entsprechend den Distanzen der Türme an der NW-Front (Kühlborn & v. Schnurbein 1992: 23ff.) wurde in 23m Abstand östlich der porta praetoria ein Turm entdeckt, während der westliche Turm 32m vom Tor entfernt war; dieser hatte auch einen Aufgang. Dieser größere Abstand erklärt sich durch das 22m lange und 8m breite Gebäude Nr. 7, das hier in 4m Entfernung von der Holz-Erde-Mauer in die via sagularis eingebaut war; es muß wohl von Anfang an konzipiert oder gebaut gewesen sein, weil man sonst den Turm auch in der Normaldistanz von 23m hätte errichten können. Das Gebäude war nach Osten, zur via praetoria und zum Tor, hin geöffnet und hatte im Inneren einen 4 : 3m großen abgetrennten Raum. Wir kannten solche Bauwerke bisher nicht; ein ähnlicher Befund begegnet jetzt auch in Anreppen.

sprengt damit ebenso die bisher bekannten Dimensionen, wie der gewaltige Komplex der Zentralgebäude; dessen Kernbau (Nr. 1), sicher ein Praetorium mit 70 : 48m Fläche, hat Förtsch überzeugend mit der privaten Luxusarchitektur Italiens verglichen (Förtsch 1995: 628). Das principia-Gebäude muß wohl direkt östlich der schmalen Straße gelegen haben. Dafür spricht, daß sich durch die Überprüfung der frühen Grabungspläne des Jahres 1969/1970 die Lage des Osttors festlegen ließ (Nr. 9), das Kühlborn mit gutem Grund als porta praetoria bezeichnet. Südlich der Via praetoria konnten Grundrisse von drei 60m langen Kasernen (Nr. 10) entdeckt werden, die im Gegensatz zu denjenigen an der Südfront (Nr. 4) auch im Contubernien-Teil fest ausgebaut waren; typologisch gleichen sie den aus Haltern bekannten Kasernen. Bemerkenswerterweise fanden sich in der Via sagularis sowohl neben dem Osttor, als auch neben dem Südtor 24 : 6.5m bzw. 23 : 6.5m breite Bauwerke, zum Tor jeweils geöffnet (Nr. 8). Trotz des etwas verbreiterten Kopfteils entsprechen sie auf´s Beste dem Haus neben der porta praetoria von Oberaden und man darf diesen nach Lage und Grundriß bisher unbekannten Gebäudetyp als Unterkunft der Torwachen interpretieren.

Kompliziert war in Oberaden die Situation mit dem vielen Oberflächen- und Grundwasser. Westlich der porta praetoria konnte ein Abwasserkanal nachgewiesen werden, der unter der Holz-Erde-Mauer hindurch zum Lagergraben führte. Auch vom Lagergraben führte dort ein Wassergraben ins Vorgelände, ebenso, wie über die Erdbrücke quer ein Gräbchen führte, das offensichtlich demselben Zweck diente.

In Anreppen gibt es zwei Gräben, die 6-7m voneinander entfernt sind. Der innere Hauptgraben ist bis zu 5m breit, der äußere Vorgraben stark schwankend bis zu 2m. Den dazwischen liegenden Streifen könnte man in Analogie zu der Situation von Alesia als Glacis bezeichnen (Reddé & v. Schnurbein 1995; 2001). Wegen des dortigen Befundes mit den verschiedenen Annäherungshindernissen (Cippi, Lilia, Stimuli) ist dieser Streifen zwischen den beiden Gräben bei den Grabungen besonders sorgfältig beobachtet worden; er blieb frei von entsprechenden Befunden.

In dem direkt an der Lippe gelegenen Beckinghausen wurde eine größere Partie der Ostfront und eine Fläche im Innenraum untersucht (Abb. 5). Ein Osttor hat danach tatsächlich gefehlt. Die drei Gräben enthielten viel einheimische und auch größere Mengen römischer Keramik, erstaunlicherweise auch viele Ziegel. Die Funde sind noch nicht im Detail publiziert; terra sigillata und Münzen belegen aber die Gleichzeitigkeit mit Oberaden. Zusätzlich zu den beiden schon vor dem 1. Weltkrieg entdeckten Töpferöfen und unklaren Gebäudegrundrissen im Westteil (v. Schnurbein 1981: 23ff.), konnte jetzt ein 12.5m breiter Teilgrundriß freigelegt werden, dessen parallele Wandgräbchen im Inneren klar auf ein horreum weisen, entsprechend den gleichalten Grundrissen von Rödgen (Schönberger & Simon 1976: 24 ff.). Ein horreum direkt am Fluß paßt bestens, da die Lippe für den Transport des umfangreichen Nachschubs für die Truppen intensiv genutzt worden ist, wie Dutzende von Weinfässern beweisen (Kühlborn & v. Schnurbein 1992: 100 ff.).

Die Datierung von Anreppen wurde wegen der Lage am Oberlauf der Lippe schon immer gerne mit der Nachricht von Velleius Paterculus (II.105) in Verbindung gebracht, nach der Tiberius dort 4/5 n. Chr. sein Heer überwintern ließ. Die Hölzer eines Brunnens ließen sich dendrochronologisch nicht genau genug bestimmen, um dies zu erhärten. Die zahlreichen mit Haltern identischen Sigillata-Stempel (Kühlborn 1998: 115, Abb. 33) sprechen aber deutlich für eine etwa gleichzeitige Existenz, wobei Gründungs- und Enddatum durchaus verschieden sein können. Bemerkenswerterweise fehlen nämlich in Anreppen bei immerhin rund 250 Münzen bisher die Gegenstempel des Varus, die in Haltern, Waldgirmes und natürlich in Kalkriese, vielfach belegt sind (Berger 1996: 51ff.). Es ist also zu erwägen, ob Anreppen nicht schon vor 9 n. Chr. aufgelassen oder weitgehend von Truppen geräumt war.

Planmäßig fortgesetzt wurden die Grabungen in Anreppen (Abb. 6), wobei der schon 1995 angeschnittene Grundriß (Nr. 6) nordöstlich des Südtores jetzt zu einem 56: 68m großen Bauwerk ergänzt werden konnte. Zugänge führten in der Nordwest- und Südwestecke sowie der Mitte der Südfront und der Nordfront auf eine Nebenstraße bzw. die via sagularis. Parallele Reihen mit wechselnd größeren und kleineren Pfostengruben lassen ein riesiges Horreum erschließen. Da hier ein moderner Friedhof liegt, kann der Grundriß des gewaltigen Baus nicht mehr vervollständigt werden. Die Grundfläche des horreums ist mit rd. 3.700 qm beinahe dreimal so groß wie diejenigen von Rödgen (Schönberger & Simon 1976: 24f.). Dieses horreum

In Haltern fanden in den bisher bekannten Anlagen seit 1995 nur sehr begrenzte Grabungen bei Baumaßnahmen statt, die das Bild nicht wesentlich veränderten. Völlig überraschend ist jedoch die Entdeckung eines neuen Lagers, das 1.5kms nordöstlich des Hauptlagers auf einer 528

Siegmar von Schnurbein: Neue Grabungen in Haltern, Oberaden und Anreppen

Anhöhe angelegt worden ist (Abb. 7). Der c.4m breite und 2.1m tiefe Spitzgraben ist auf etwa 400m Länge festgelegt. Er läßt ein polygonales Lager erschließen, das sicher über 20ha groß war (Kühlborn 1999). Da weder eine Holz-ErdeMauer noch Wandgräbchen von Gebäuden gefunden worden sind, wird es sich um ein Marschlager handeln. Über 70 Gruben enthielten römische Funde, die denjenigen von Haltern entsprechen, weshalb die Zugehörigkeit zur Drusus-Phase ausgeschlossen ist; ob eventuell die Zeit des Germanicus nach der Varus-Niederlage zwischen 10 und 16 n. Chr. infrage kommt, läßt sich noch nicht beurteilen (Kühlborn in Neujahrsgruß 2000: 88-91).

(Mainz): 27-33. Kühlborn J.-S. & v. Schnurbein S. 1992 Das Römerlager Oberaden III. Bodenaltertümer Westfalens 27. (Münster). Manning W.H. & Scott I. 1979 Roman timber military gateways. Britannia X: 19-61. Neujahrsgruß 1998 Jahresbericht für 1997, Westfälisches Museum für Archäologie. (Münster). Neujahrsgruß 1999 Jahresbericht für 1998, Westfälisches Museum für Archäologie. (Münster). Neujahrsgruß 2000 Jahresbericht für 1999. Westfälisches Museum für Archäologie. (Münster). Reddé M. 1995 Titulum et clavicula. A propos des fouilles récentes d´Alesia. Revue Archeologique de l´Est et du Centre-Est 46: 349-356. Reddé M. & v. Schnurbein S. 1995 Neue Ausgrabungen und Forschungen zu den Belagerungswerken Caesars um Alesia. Berichte der Römisch-Germanischen Kommission 76: 73-158. Reddé M. & v. Schnurbein S. (Hrsg.) 2001 Alésia. Fouilles et Recherches Franco-Allemandes sur les Travaux Militaires Romaines autour du Mont-Auxois 19911997. Mémoires de l’Academie des Inscriptions et Belles Lettres 22. (Paris). v. Schnurbein S. 1981 Untersuchungen zur Geschichte der römischen Militärlager an der Lippe. Berichte der Römisch-Germanischen Kommission 62: 5-101. v. Schnurbein S. 2000 The organisation of the fortresses in Augustan Germany. In R.J. Brewer (ed.) Roman fortresses and their legions. Papers in honour of George C. Boon. Occasional papers of the Society of Antiquaries of London No. 20: 29-39. Schönberger H. & Simon H.-G. 1976 Römerlager Rödgen. Limesforschungen 15. (Berlin). Wells C.M. 1998 What´s new along the Lippe: Recent work in North Germany. Britannia XXI: 459-464.

Gerade die letztere Frage zu klären, ist ein altes Desiderat, denn noch immer fehlen im Lippe-Weser-Gebiet archäologische Spuren der Germanicus-Kriege 14-16 n. Chr., abgesehen von Bentumersiel an der Ems-Mündung. Eine wichtige Korrektur deutet sich für Kneblinghausen an, das mangels römischer Funde bisher wegen der Clavicula-Tore gerne den Chattenzügen des Domitian zugewiesen wurde. Da solche Tor-Formen nun aber dank Alesia bereits für caesarische Zeit belegt sind (Reddé 1995; Reddé & v. Schnurbein 1995; 2001), entfällt dieses Argument für die Spät-Datierung. Auf der 1998 entstandenen Karte (Kühlborn 1998: 78, Abb. 1) ist Kneblinghausen bereits zusammen mit den augusteischen Plätzen kartiert, was hier auf Abb. 1 übernommen worden ist; es bleibt abzuwarten, ob sich dies erhärten läßt. Nach einer Bemerkung in der Britannia von 1998 scheint man aus der Distanz den Eindruck gewinnen zu können, daß man sich in Deutschland vor allem darauf konzentriere, die bekannten Plätze weiter zu erforschen und nicht zwischen Rhein und Elbe systematisch nach archäologischen Spuren der Germanenkriege zu suchen (Wells 1998: 457). Dies ist nicht der Fall! O. Braasch, J.-S. Kühlborn und andere fliegen regelmäßig, wobei Braasch immerhin die augusteischen Anlagen von Marktbreit und Dorlar entdeckt hat. Im übrigen Raum sind gelegentlich verdächtige Spuren beobachtet worden, z. B. von St. Berke bei Werl in Westfalen, doch ließ sich bisher keine dieser Anlagen den römischen Aktivitäten zuordnen (Neujahrsgruß 1998: 82, Abb. 32). Literatur Berger

F. 1996 Kalkriese 1. Die römischen Fundmünzen. Römisch-Germanische Forschungen 55. (Mainz). Förtsch R. 1995 Villa und Praetorium. Zur Luxusarchitektur in frühkaiserzeitlichen Legionslagern. Kölner Jahrbuch 28: 617-630. Kühlborn J.-S. 1995 Germaniam pacavi - Germanien habe ich befriedet. (Münster). Kühlborn J.-S. 1998 Die römischen Militäranlagen in Westfalen. In Hinter Schloß und Riegel. Burgen und Befestigungen in Westfalen. (Münster): 77-119. Kühlborn J.-S. 1999 Antike Berichte durch Ausgrabungen bestätigt. Archäologie in Deutschland Heft 3: 6-12. Kühlborn J.-S. 2000 Schlagkraft. Die Feldzüge unter Augustus und Tiberius in Nordwestdeutschland. In L. Wamser (Hrsg.) Die Römer zwischen Alpen und Nordmeer.

529

Limes XVIII

Abb. 1. Römische Militärstationen an Niederrhein und in Westfalen aus augusteischer Zeit. Nach Kühlborn 2000.

Abb 2. Oberaden. Stand der Grabungen 1999. 1 Praetorium: 2 Principia: 3a-3b Legatenhäuser?: 4-5 Wohnbauten: 6 Centurionenhäuser: 6a Kasernenbereiche: 7 Haus für Torwachen. Nach Kühlborn 2000.

530

Siegmar von Schnurbein: Neue Grabungen in Haltern, Oberaden und Anreppen

Abb. 3. Oberaden, Principia und seitlich anschließende Bebauung. Die Bauspuren in der Mitte der Südfront des Principia-Gebäudes sind erodiert. Stand 1999.

531

Limes XVIII

Abb. 4. Oberaden. Porta Praetoria und seitlich anschließende Bebauung.

Abb. 5. Oberaden-Beckinghausen. Stand 1999.

532

Siegmar von Schnurbein: Neue Grabungen in Haltern, Oberaden und Anreppen

Abb. 6. Anreppen. 1 Praetorium: 2-3: Wohnbauten: 4 Centurionenhäuser: 4a Kasernenbereiche: 5 Porta principalis dextra: 6 Horreum: 7 Wirtschaftsbau? 8 Häuser für Torwachen: 9 Porta Praetoria: 10 Kasernen. Stand 1999. Nach: Kühlborn 2000.

533

Limes XVIII

Abb. 7. Haltern. Übersicht über die römischen Anlagen. Nach: Kühlborn 1999.

534

ISBN 9781841714639 (Volume I) paperback ISBN 9781841714646 (Volume II) paperback ISBN 9781841714653 (Volume set) paperback ISBN 9781407324784 (Volume set) e-format DOI https://doi.org/10.30861/9781841714653 A catalogue record for this book is available from the British Library

BAR

PUBLISHING

List of Contents Volume 1 Introduction

viii

Timetable of sessions, presentations and events

xvii

Formal Addresses Address by His Royal Highness Prince El Hassan bin Talal of the Hashemite Kingdom of Jordan

xxiii

Professor David Breeze, Chairman of the Congress Welcoming address

xxvii

David J. Breeze Jan Kees Haalebos†

xxix

Closing ceremony Speech given by Professor Zsolt Visy

xxx

Contributors’ addresses

xxxii

Sessions General Themes A.R.Birley Fifty years of Roman frontier studies

1

Lukas de Blois The onset of crisis: the reign of the emperor Severus Alexander (AD 222- 235)

13

Nicolae Gudea Bezüglich der Europäischen Wanderschaft der legio III Gallica zu Beginn des 3. Jh.

19

William S. Hanson Why did the Roman empire cease to expand?

25

Johann van Heesch Mints and the Roman army from Augustus to Diocletian

35

Fraser Hunter Problems in the study of Roman and native

50

Karl Strobel Die Legionen des Augustus. Probleme der römischen Heeresgeschichte nach dem Ende des Bürgerkrieges: Die Truppengeschichte Galatiens und Moesiens bis in Tiberische Zeit und das Problem der Legiones Quintae

51

Wolfgang Vetters & Heinrich Zabehlicky The northern, southern and eastern frontiers and the climate c.AD200

67

Zsolt Visy Similarities and differences in the late Roman defense system on the European and eastern frontiers

71

The Eastern Frontier S. Thomas Parker The Roman frontier in Jordan: An overview

77

i

Bill Leadbetter Galerius and the eastern frontier

85

Ariel Lewin Diocletian: Politics and limites in the Near East

91

Chaim Ben-David The Zoar ascent - a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau

103

Tali Erickson-Gini Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations

113

Zbigniew T. Fiema The military presence in the countryside of Petra in the C6th

131

George Findlater Limes Arabicus, via militaris and resource control in southern Jordan

137

David F. Graf Nomads and the Arabian frontier: the epigraphic perspective

153

Theodor Kissel A reused milestone from Imtan (southern Syria) – new evidence on the limes Arabicus in the second century AD

161

Maurice Lenoir Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes

175

Mordechai Gichon 45 years of research on the limes Palaestinae – the findings and their assessment in the light of the criticisms raised (C1st – C4th)

185

Haim Goldfus & Benny Arubas Excavations at the Roman siege complex at Masada – 1995

207

Israel Roll Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia

215

Yotam Tepper Lajjun – Legio in Israel: Results of a survey in and around the military camp area

231

Susan Weingarten Road use in late antique Palestine

243

Martin Hartmann & Michael A. Speidel Roman military forts at Zeugma. A preliminary report

259

Lars Petersen Preliminary report on the Roman period at Kamid el-Loz (Lebanon)

269

Rome and Parthia Edward Dąbrowa “. . . ostentasse Romana arma satis . . .” The military factor in Roman-Parthian relations under Augustus and Tiberius

ii

275

Gwyn Davies The circumvallation at Hatra

281

Everett L. Wheeler Roman treaties with Parthia: Völkerrecht or power politics?

287

Josef Wiesehöfer Die ‘Sklaven des Kaisers’ und der Kopf des Crassus. Römische Bilder des Ostens und parthische Bilder des Westens in augusteischer Zeit

293

The Anatolian Provinces and the Black Sea Region Julian Bennett The Cappadocian frontier: from the Julio-Claudians to Hadrian

301

Ergün Lafli Notes on the history of Seleuceia Sidēra in Pisidia (south-western Turkey): Second preliminary report on the inscriptions

313

Jeorgios Martin Beyer ίνα ... αύτοί άλλοις Βοράδοι καί Γότθοι γίνωνται. Gregorios Thaumaturgos und die pontischen Beutezüge der Boranen und Goten im 3. Jh. n. Chr.

327

Oleg Savelya Excavation of a new Roman base near Balaklava in the Crimea (1992-1999)

339

North Africa Alan Rushworth Defensores Provinciae: the militarisation of frontier zone élites in C3rd Mauretania Caesariensis

349

Steven E. Sidebotham The Roman empire’s southeastern-most frontier; recent discoveries at Berenike and environs (Eastern Desert of Egypt) 1998-2000

361

Gabriele Ziethen & Erik Klingenberg Merchants, pilgrims and soldiers on the Red Sea route

379

The Germanies Harry van Enckevort The eastern canabae legionis of the legio X Gemina on the Kops Plateau in Nijmegen (NL)

387

Meingad Filgis Holzfachwerkbauten im Kastellvicus von Wimpfen, Neckar-Odenwaldlimes: Topographie, Siedlungsstruktur, Nutzungszonen, Grundrisstypen sowie belegbare Nutzungen

395

J.K. Haalebos† Die früheste Belegung des Hunerberges in Nijmegen

403

Marcus Jae & Markus Scholz Reduktion von numerus- und Kleinkastellen des obergermanischen Limes im 3. Jahrhundert

415

Stefan Neu & Matthias Riedel Das Kölner Rheinufer im ersten und zweiten Jahrhundert n. Chr.

425

Gabriele Rasbach Die spätaugusteische Siedlung in Lahnau-Waldgirmes - Zusammenfassende Bemerkungen zum Stand der Fundauswertung

433

iii

C. Sebastian Sommer Recent developments in south-west Germany (eastern Germania Superior - western Raetia)

441

J. Thyssen The late Roman fort at the Valkhof in Nijmegen. Nijmegen at the transition from the Roman period to the middle ages

453

Early Roman Germany Armin Becker Die Ausgrabungen in Lahnau-Waldgirmes

461

Paul Bidwell Timber baths in Augustan and Tiberian fortresses

467

Marjan C. Galestin Early Roman military activity on the Frisian coast

483

Thomas Grane Problems concerning the early Roman fortifications at Marktbreit (D) and Valkenburg ZH (NL)

489

Norbert Hanel Zur Datierung der frühkaiserzeitlichen Militärlager von Novaesium (Neuss)

497

Andrea Hagendorn & Christine Meyer-Freuler Das Legionslager von Vindonissa: Neue Forschungen zur Frühzeit

501

Katrin Roth-Rubi Why Dangstetten?

509

Susanne Wilbers-Rost Kalkriese und die Varusschlacht – Archäologische Nachweise einer militärischen Auseinandersetzung zwischen Römern und Germanen

515

Siegmar von Schnurbein Neue Grabungen in Haltern, Oberaden und Anreppen

527

Volume 2 The Danubian and Balkan Provinces Claus-Michael Hüssen Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen an der Donau und im raetischen Limesgebiet

535

Markus Gschwind The Flavian timber fort at Eining (Abusina) on the Upper Danube

549

Andreas Schaub Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg

559

Stefan Groh & Helga Sedlmayer Neue Grabungen im Vicus des Auxiliarkastells Favianis (Noricum) – ein Handwerksbezirk im östlichen Kastellvicus

567

M. Kronberger & M. Mosser Vindobona – legionary fortress, canabae legionis and necropolis

573

iv

Ingrid Mader Vindobona – the Roman civil settlement at Vienna

585

Sylvia Sakl-Oberthaler & Patrizia Donat et al Die Ausgrabungen am Michaelerplatz – ein Bereich der canabae legionis von Vindobona

591

Raimund Kastler Legionslager an der Wende zur Spätantike – Ein Überblick zu Carnuntum und vergleichbaren kaiserzeitlichen Standlagern des Rhein-Donau-Raumes in einer Periode des Umbruchs

605

Eszter Istvánovits & Valéria Kulcsár The history and perspectives of the research of the Csörsz Ditch (‘Limes Sarmatiae’)

625

Snežana Golubović Decorated lead sarcophagi in Moesia Superior

629

Dorel Bondoc Artillery troops detached north of the lower Danube in the late Roman period

641

Andrzej B. Biernacki The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

649

Gerda v. Bülow Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes?

663

Sven Conrad & Dimităr Stančev Archaeological survey on the Roman frontier on the lower Danube between Novae and Sexaginta Prista. Preliminary report (1997-2000)

673

Piotr Dyczek Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae (Moesia Inferior)

685

Elena Klenina Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior)

695

Archibald Dunn Was there a militarisation of the southern Balkans during late antiquity?

705

Mirjana Sanader Tilurium, Burnum and Bigeste. A new contribution to the dating of the Delmataen limes

713

Dacia Ioana Bogdan Cătăniciu Géographie antique (Ptolémée, Tabula Peutingeriana, Ravennatus) et la stratégie impériale en Dacie

719

Cristian Gazdac Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia

737

Miroslava Mirković Deserted forts – the Moesian limes after the conquest of Dacia

757

Liviu Petculescu The military equipment of oriental archers in Roman Dacia

765

v

The Spains Martin Luik Die römischen Lager bei Renieblas, Prov. Soria (Spanien). Ergebnisse der Vermessungskampagnen 1997-2000 771 Angel Morillo & Victorino García-Marcos Twenty years of Roman military archaeology in Spain

779

Victorino García-Marcos & Angel Morillo The legionary fortress of VI Victrix at León (Spain). The new evidence (1995-2000)

791

Jorge López Quiroga & F.Germán Rodríguez Martín Les castra et les castella aux extrémités de l’Empire après la fin de la domination romaine: le Nord-ouest Ibérique et le Haut Rhin au Ve. siécle

801

C. Fernández. Ochoa, V. García Entero, F. Gil Sendino & C. Valenciano Prieto New intervention in the surroundings of the gate of the Roman wall of Gijón (10 Recoletas Street) 813

Britain Lindsay Allason-Jones The material culture of Hadrian’s Wall

821

Julian Bennett A revised programme and chronology for the building of Hadrian’s Wall

825

J.L. Davies & R.H. Jones Recent research on Roman camps in Wales

835

Alison Taylor Supplying the frontier zones: the rôle of the East Anglian Fens

843

Tony Wilmott Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999

851

Pete Wilson Cawthorn Camps – 70 years after Richmond

859

D.J.Woolliscroft The Roman Gask frontier: 1997-2000

867

The Roman Army Sara E. Phang The timing of marriage in the Roman army

873

D.B. Saddington The Roman auxilia in the east – different from the west?

879

Roman Fortifications David J. Breeze Ditches for defence: towards a study of multiple fort ditches

883

Nicholas Hodgson ‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts at Wallsend and South Shields on Hadrian’s Wall 887

vi

Birgitta Hoffmann The rampart buildings of Roman legionary fortresses

895

Fleets and Frontiers Maureen Carroll Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica)

901

Thomas Fischer Neue Grabungen an der Westseite des römischen Flottenlagers Köln – Alteburg

904

Norbert Hanel Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg (Alteburg)

913

Boris Rankov Now you see it, now you don’t. The British Fleet in Vegetius IV.37

921

Documents and Archives A.R. Birley The Roman army in the Vindolanda Tablets

925

John Pearce Food as substance and symbol in the Roman army: a case study from Vindolanda

931

Margaret Roxan Vespasianvs Velageno

945

vii

Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen an der Donau und im raetischen Limesgebiet Claus-Michael Hüssen In den vergangenen zehn Jahren sind an der Donau eine ganze Reihe neuer Erkenntnisse sowohl im Bereich des Militärs wie auch bei der ländlichen Besiedlung an der Donaugrenze und im Limesgebiet gewonnen worden. Der Beitrag soll ein Überblick über die neueren Entdeckungen und Forschungen an der römischen Nordgrenze in Raetien geben. Bei den vorgestellten Grabungen und Projekten war die Römisch-Germanische Kommission mit der Forschungsstelle Ingolstadt teils durch eigene Grabungen oder in bewährter Kooperation mit dem Bayerischen Landesamt für Denkmalpflege beteiligt.∗

Die archäologischen Quellen im Donauraum für die Zeit zwischen c.30 v. Chr. bis etwa 40 n. Chr. sind bekanntermaßen spärlich. In den letzten Jahren kamen eine Reihe von Befunden zutage, die diese „Überlieferungslücke“ etwas schließen können (Hüssen 2000b). Eine „ethnische“ Zuordnung der neuen endlatènezeitlichen Befunde ist nicht eindeutig möglich. Auch zeitlich lassen sich die Funde oft nicht exakt fassen, da die chronologischen Probleme für diese Übergangsperiode zwischen Kelten und Römern noch keineswegs gelöst sind. Es zeichnet sich jedoch ab, dass der oberbayerische Donauraum nach dem Zusammenbruch der keltischen Machtzentren, d. h. des Oppidums von Manching, nach der Mitte des 1. Jahrhunderts v. Chr. oder wenig später unter elbgermanischen Einfluss geriet.

Am Beginn des Beitrags steht ein knapper Überblick über die Zeit vor der Sicherung der Donaulinie durch die Römer. Hier stellt sich die Frage nach der vorrömischen Bevölkerung und nach den möglichen Ursachen für die zeitlich relativ späte Einrichtung einer Kastelllinie an der Donau. Neu für diese Frühzeit ist, dass es Belege für eine vorrömische Bevölkerung gibt, die hier zumindest bis in die frühaugusteische Zeit lebte. Der zweite Abschnitt beschäftigt sich mit der militärischen Sicherung der Donau in tiberisch-claudischer Zeit westlich und östlich von Oberstimm – zwischen Burgheim und Vohburg – und der ersten ländlichen Besiedlung in dieser Region. Eine Reihe ganz unterschiedlicher Militäranlagen ist an diesem etwa 60kms langen Donauabschnitt inzwischen bekannt geworden, die ein differenziertes Bild der Sicherung der Nordgrenze Vindelikiens geben. Auch eine frühe ländliche Besiedlung, etwa ab der Mitte des 1. Jahrhunderts außerhalb der Kastellorte wird jetzt fassbar. Ergebnisse neuerer Grabungen und Entdeckungen im Limesgebiet aus der Zeit der Ausdehnung des Reichsgebietes nach Norden über die Donau von flavischer Zeit bis in hadrianische Zeit werden im dritten Abschnitt vorgestellt. Neuere Untersuchungen fanden hier in den vergangenen Jahren an beinahe allen Kastellorten statt.

Im Frühjahr 2000 wurde in Ingolstadt-Oberhaunstadt, wenige Kilometer nördlich der Donau, ein endlatènezeitliches Urnengrab entdeckt. Der Bestattungsbrauch – es handelt sich um ein Brandgrab in einer abgedeckten Urne mit einem Eisenmesser als Beigabe – spricht für einen Germanen, der im ausgehenden 1. Jahrhundert v. Chr. hier bestattet wurde. Wenige Kilometer weiter östlich, in Großmehring, fand man 1998 eine Deponierung von zwei spätlatènezeitlichen Fibeln, einer Almgren 65 und einer Nauheimer Fibel, zusammen mit einer elbgermanischen Terrine (Abb. 1). Eine weitere mäanderverzierte germanische Scherbe aus dem Großromstedter Horizont wurde flußaufwärts bei Bergheim gefunden.

Die Zeit vor dem Alpenfeldzug

Diesen drei Fundplätzen mit „germanischem“ bzw. germanisch-keltischem Inventar nördlich der Donau können drei weitere Fundstellen aus der Zeit nach dem Niedergang des keltischen Oppidums gegenübergestellt werden, die südlich der Donau zutage kamen. Auch in diesen Plätzen südlich der Donau ist elbgermanischer Einfluß spürbar, ebenso aber auch keltische Traditionen.

Die Beschäftigung mit den Römern an dem hier betrachteten Donauabschnitt führt beinahe zwangsläufig zu der Frage, auf welche Bevölkerung das römische Militär bei seinen Vorstößen traf. Die militärische Sicherung der durch den Alpenfeldzug eroberten Gebiete Vindelikiens zeigt, dass bis in spättiberisch-claudische Zeit offensichtlich kein Grund bestand, eine Außengrenze militärisch zu sichern. Die Donau, die sich als Grenzfluss anbot, blieb offenbar weitgehend unbeachtet (Hüssen 2000a).

Aus einer kleinen endlatènezeitlichen Siedlung in Burgheim stammt ausschließlich handgemachte Keramik, die deutlich Verbindungen zum Formenspektrum und zur Machart keltischer Ware erkennen lässt. Das Inventar eines Brandgrabes aus Ingolstadt-Zuchering zeigt neben deutlichen „keltischen“ Beigaben auch Verbindungen in den mitteldeutschen Raum (Pauli 1995: 149ff.). Schließlich kam aus einer Grablege in GeisenfeldIlmendorf ein Trinkhorn zutage, dessen Herkunft ebenfalls im elbgermanischen Raum zu suchen ist (Hüssen 2000b). Es ist sicher noch verfrüht, auf der Basis der wenigen Befunde von einer möglichen „Kulturgrenze“ an der Donau in den Jahren vor dem Alpenfeldzug zu sprechen, hierzu bedarf es weiterer gemeinsamer Anstrengungen der archäologischen Denkmalpflege und der Forschung.

∗

Für die langjährige kollegiale Zusammenarbeit danke ich den Herren Dr. K.H. Rieder und Dr. St. Winghart, Bayerisches Landesamt für Denkmalpflege (BLfD), und ihren Mitarbeitern in Ingolstadt, Herrn Dr. B. Becker, BLfD München, für die Überlassung des Planes von Ruffenhofen, Herrn Dr. R. Koch, ehem. BLfD Nürnberg, sowie Herrn Dr. J. Weinig und den Mitarbeitern der Fa. ProArch GmbH, Ingolstadt.

535

Limes XVIII

Festzuhalten bleibt jedoch, dass die Region an der Donau in der Zeit vor und während der römischen Okkupation noch besiedelt war.

an der Donau im Westen wurde aufgrund von Militariafunden schon länger in dem 30kms entfernten Burgheim vermutet. Die Auswertung von Grabungen im heutigen Ort erbrachten Strukturen eines Vicus mit typischen frühen Holzhäusern (Abb. 3). Das noch nicht lokalisierte Kastell ist in unmittelbarer Nähe – ebenfalls im Ortsbereich – zu suchen.

Auf die Bevölkerung der dünn besiedelten Gebiete im Donauraum müssen die zahlreichen Truppenbewegungen infolge des Alpenfeldzuges und die Anwesenheit großer Truppenverbände im nur rund 80kms entfernten Augsburg Auswirkungen gehabt haben (Bakker 1994a & b). Das römische Militär zeigte vermutlich mit Patrouillen seine Präsenz schon in augusteisch-tiberischer Zeit im Donautal. Ein schwacher, aber doch wahrnehmbarer Fundniederschlag in unserer Region mit wenigen, bisher noch isolierten Fibelfunden auch nördlich der Donau, insbesondere einige frühe Doppelknopffibeln und Augenfibeln, lassen eine vorclaudische Besiedlung an der oberen Donau erahnen. Die Fundstellen mit frühem Material liegen alle fernab der bekannten Militärplätze.

Zwischen den beiden festen Militärplätzen liegen eine Reihe von kurzzeitig belegten Lagern unterschiedlicher Größe und Funktion. Donauabwärts – zwischen Burgheim und Neuburg – ist der erste vielleicht nur kurzzeitig besetzte Militärposten auf dem Stätteberg bei Unterhausen zu vermuten. Von hier stammen einige frühe Sigillaten, darunter padanische Auflagensigillata, und handgemachte Gebrauchskeramik, wie sie aus anderen Militärlagern an der Donau bekannt ist. Bei Neuburg an der Donau lagen zwei Kastelle. Auf dem Schlossberg ließ sich unter schwierigsten Bedingungen einer Stadtkerngrabung der Grundriss eines Kleinkastells mit etwa 1000qm Innenfläche ermitteln, das Platz für etwa eine Centurie bot (Abb. 4). Damit konnte hoch über dem Tal der Militärposten für nachgewiesen werden, den G. Ulbert (1965) an dieser Stelle schon vor über 35 Jahren postulierte. Die Truppe war hier von den späten 30er/40er Jahren bis in frühflavische Zeit stationiert (Hüssen 1993a). Im Tal bei Neuburg-Zell zeigt sich ein zweites Lager im Luftbild. Die etwa 1ha große Anlage an der Donausüdstraße war nur kurzfristig belegt (Braasch 1985).

Die Entdeckungen der letzten Jahre haben gezeigt, dass man in Regionen und Zeiten, die archäologisch nichts „hergeben“, nicht unbedingt auf eine reale Siedlungsleere schließen darf. Die Auffindung von Fundstellen aus einer Periode mit dünner Besiedlung hängt in entscheidendem Maße von einer konsequenten Denkmalpflege ab. Obwohl die Befundsituation rund um Ingolstadt wegen der großen Flächengrabungen auf hunderten von Hektar entlang der Donauterrassen in den letzten 10 Jahren besonders günstig ist, lässt sich angesichts der geringen Zahl von entdeckten Fundstellen aus der Endlatène- und der frühen römischen Zeit der langsame Fortschritt der Erkenntnisse ermessen.

20 kilometer weiter flussabwärts konzentrieren sich fünf Truppenlager auf einem weitgehend hochwasserfreien Niederterrassenrücken, drei in Zuchering (Braasch 1983: 70 Taf. 8; 88 Taf. 26) und zwei in Oberstimm. Den Platz des Oppidums in Manching, dessen mächtige Umwehrung zu dieser Zeit sicher noch weithin sichtbar war, mied man bewusst. Dort bot das südlich anschließende sumpfige Feilenmoos zwar Schutz, behinderte gleichzeitig aber die Bewegungsmöglichkeiten und die Anbindung der Lager am südlichen Donauufer über das Paartal an das militärische Hauptquartier in Augsburg (Czysz 1981).

Die Donaulinie Schon in tiberischer Zeit beginnt die militärische Erschließung des Donautales, die unter Claudius durch die Errichtung des Standlagers in Oberstimm abgeschlossen wurde (Schönberger 1978; Schönberger, Köhler & Simon 1989). Neben den dauerhaften Kastellen zeigt sich ein vielfältiges Bild von Militäranlagen (Abb. 2). Kleine Lager für vielleicht nur 12 Mann Besatzung stehen neben großen Lagern für Auxiliartruppen, die teils nur für wenige Monate, teils auch mehrere Jahre belegt waren. Anlagen mit kompletter Fortifikation stehen neben solchen mit Grabenanlagen, die wohl eher den Charakter einer symbolischen Einfriedung hatten und nur für einen kurzen Aufenthalt angelegt waren.

Die drei Zucheringer Lager liegen nahe beieinander an der römischen Donausüdstraße, unmittelbar an einem Altarm der Donau (Abb. 5). Alle drei dürften in der Frühzeit der militärischen Besetzung des Donautals errichtet worden sein, erfüllten aber unterschiedliche Funktionen. Die Lager I und II sind in Ausschnitten in den 90er Jahren ausgegraben worden (Hüssen 1995).

Auf dem rund 60kms langen Donauabschnitt zwischen Burgheim und Vohburg sind zehn Truppenlager der frühen und mittleren Kaiserzeit bekannt oder zu vermuten. Für eine Reihe der Plätze lässt sich die Errichtung und Belegungzeit nicht oder nur annäherungsweise bestimmen (Hüssen 1995: 95ff.). Zum militärischen Knotenpunkt der Region entwickelte sich das Gebiet unmittelbar südlich des breiten Strombetts der Donau im Bereich der heutigen Ortschaften Zuchering (Stadt Ingolstadt) und Oberstimm (Marktgemeinde Manching), wo das erste dauerhafte Standlager errichtet wurde. Das von Oberstimm aus nächstgelegene dauerhafte Kastell

Die ersten Truppenlager an diesem Donauabschnitt für militärische Erkundungsmaßnahmen könnten die beiden unmittelbar nebeneinander liegenden Lager II und III in Zuchering gewesen sein. Lager II war etwa 0.9ha, Lager III etwa 0.5ha groß. Die Umwehrung des Lagers II (Abb. 6) erfüllte zwar optisch die Kriterien einer Fortifikation, real waren die schwachen Gräben aber kaum als Schutz geeignet. Augenfällig ist die Übereinstimmung des Lagers II mit den ebenfalls nur schwach umfriedeten Lagern bei 536

Claus-Michael Hüssen: Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen

Rederzhausen in der Nähe von Augsburg (v. Schnurbein 1983). Die beiden Lager II und III entstanden wohl in der Erkundungsphase der Donaugrenze für kurze Zeit, oder sie dienten als Standorte während der Bauzeit der Donausüdstraße. Spuren einer Innenbebauung sind nicht vorhanden. Die Soldaten lebten wahrscheinlich in Zelten.

Posten überwachten Straße von Burghöfe donauabwärts waren in unserem Raum zuerst Truppen in semipermanenten Lagern mit regulärer Umwehrung oder in Lagern mit einfachen Umfriedungen wie in den drei Zucheringer Kastellen stationiert. Das Auxiliarkastell in Oberstimm ist nach den neuen Ergebnissen nicht der Ausgangspunkt der Erschließung dieses Donauabschnittes. Die Verlegung einer größeren berittenen Einheit an diesen Platz stand nicht am Beginn der militärischen Aktion, sondern bezeichnete den vorläufigen Endpunkt einer mehrjährigen Kampagne.

Wenig später errichtete man ganz in der Nähe das erste Lager mit einer massiven Rasensodenmauer. Das Lager I in Zuchering war mindestens 1ha groß und von einem breiten Doppelgraben umgeben (Abb. 7). Die Innenfläche ist nicht ausgegraben. Als Unterkünfte für die Soldaten dienten vermutlich leichte Holzbaracken. Das Lager bietet Platz für eine größere Truppe in der Stärke einer Kohorte. Das Fibelspektrum, insbesondere vier Aucissafibeln, sprechen dafür, dass dieses Lager zeitlich vor dem Oberstimmer Kastell errichtet worden war. Ein möglicher Grund für die Verlegung des Standortes war die Gefährdung des Platzes in Zuchering durch Überschwemmungen, so dass man für das wenig später errichtete Standlager in Oberstimm einen etwas höhergelegenen Platz wählte. Wahrscheinlich hatte man das Lager Zuchering I aber schon von vornherein als semipermanenten Standort geplant.

Frühe Siedlungen an der Donausüdstraße Nach der militärischen Sicherung der Donaulinie erfolgte schon in vorflavischer Zeit die zivile Aufsiedlung des südlichen Donautales östlich des Lech. Die bekannte Villenbesiedlung an der Donau mit teilweise sehr großen, repräsentativen Gehöften stammt erst aus dem 2. Jahrhundert. Im südlichen Donautal ist es in den vergangenen 10 Jahren bei ausgedehnten Flächengrabungen gelungen, eine Reihe von Gehöften freizulegen, die bereits in der Zeit der Grenzsicherung oder unmittelbar danach entstanden. Die Befunde zeigen, dass schon in dieser frühen Phase die Aufsiedlung der neu gewonnenen Gebiete begann.

In unmittelbarer Nähe des Oberstimmer Kastells liegt ebenfalls ein zweites Lager. Der Grundriss des Kleinstkastells mit Innenbebauung aus Holz ist im Luftbild erfasst (Abb. 8). Das Lager wurde an der Mündung des kleinen Flüsschens Brautlach in einen alten Donauarm errichtet. An diesem Zusammenfluss ist der Heimathafen der 1995 ausgegrabenene Patrouillenboote in Oberstimm zu suchen (Hüssen, Rieder & Schaaf 1995). Vermutlich war in diesem Kleinkastell – vergleichbar mit Nersingen (Mackensen 1987: 129ff.) – nach dem Dreikaiserjahr für einige Zeit ein Militärposten stationiert, bevor das daneben liegende Standlager wieder mit einer regulären Auxiliareinheit besetzt wurde.

Die ersten ländlichen Siedlungen an der Donausüdstraße sind in reiner Holzbauarchitektur errichtet. Bislang sind neun dieser Holzvillen ganz oder teilweise ausgegraben, so dass man für die Frühzeit der römischen Besiedlung von einem Regelbefund zumindest in dieser Donauregion ausgehen kann (Hüssen & Wegener-Hüssen 1995). Betrachtet man die Grundrisse dieser Gutshöfe, so ist nur schwer vorstellbar, dass wir hier die Vorgänger der limeszeitlichen Villen im Donauvorland vor uns haben. Es stellt sich die Frage nach der Herkunft dieser frühen Siedler. Vielleicht waren es Veteranen der raetischer Einheiten oder Siedler aus dem vindelikischen Binnenland, die sich unter dem Schutz der römischen Truppen hier niedergelassen haben.

Als Standort für eine kleine Truppe kommt östlich von Oberstimm noch der Burgberg in Vohburg in Frage. Einige vereinzelte Grabungsfunde des 1. Jahrhunderts sprechen für eine römische Präsenz auf diesem markanten Platz mit einer weiten Sicht über das Donautal. In den seit Jahrhunderten umgewälzten mittelalterlichen Schichten auf dem Berg wird seit mehreren Jahren gegraben und es sind einige Funde des ersten 1. Jahrhunderts zutage gekommen, u.a. ein prägefrischer Denar des Nero sowie die Attasche eines Bronzegefäßes. Die Chance, auf ungestörte römische Befunde zu treffen, ist hier allerdings sehr gering.

Die Versorgung der Soldaten und der Vicani mit Nahrungsmitteln musste in der Frühphase der Grenzsicherung sicherlich überwiegend aus weiteren Hinterland erfolgen. Die neueren Grabungsbefunde zeigen aber auch, dass in der insgesamt nicht besonders fruchtbaren Grenzzone am südlichen Donauufer, die durch das Donau- und Feilenmoos stark eingeengt war, bereits ab der Mitte des 1. Jahrhunderts Siedlungen an der Donausüdstraße entstanden.

Insgesamt sind auf einem nur 60kms langen Donauabschnitt in der Phase der Sicherung der Grenze jetzt zehn Militäranlagen nachgewiesen. Allein im Umfeld von Oberstimm liegen fünf Truppenlager. Wenn auch die Kenntnisse über die einzelnen Plätze noch lückenhaft sind, so wird doch die Vielschichtigkeit und Dynamik der römischen Aktivitäten am Südufer der Donau im 1. Jahrhundert n. Chr. deutlich. Nach dem Bau einer durch

Mehrere kleine Gutshöfe mit Holzbauten wurden südlich der Donau bei Ingolstadt-Zuchering ausgegraben, wo sich das Flusstal weitet. Zwei weitere lagen wenige Kilometer donauauf- bzw. donauabwärts in Weichering und Rockolding. Bei Weichering an der Donausüdstraße fand man zwei Holzpfostenbauten aus der Mitte des 1. Jahrhunderts (Hüssen 1993b). Unter den Funden sind

537

Limes XVIII

einige Militaria (Abb. 9). Vielleicht hatte sich hier ein Veteran niedergelassen, möglicherweise handelte es sich aber auch um einen frühen Straßenposten. Den nahezu vollständigen Grundriss einer Holzvilla zeigt das von einem Zaun oder einer Palisade eingefriedete Gehöft in Zuchering, Flur „Seehofer Schaffeld“, mit mindestens 5 Gebäuden in der etwa 0.25ha großen Hoffläche (Abb. 10).

nahezu vollständig ausgegraben worden und die Auswertung der Grabungen abgeschlossen. Zum ersten Mal kennen wir den kompletten Grundriss eines größeren Lagers mit 3.2ha, dessen Innenbebauung ausschließlich aus Pfostenbauten besteht. Das Kastell ist mit einem Doppelgraben umwehrt, vor den Tordurchfahrten liegen Titula. Entlang der Lagerringstraße befinden sich 15 Baracken in Holzpfostenbauweise. Die Innenfläche des Lagers, dort wo die Zentralbauten zu erwarten gewesen wären, war nicht ausgebaut (Abb. 12).

Mit der flächigen Erschließung des Landes nördlich der Donau ab 120 n. Chr. scheint die Besiedlung in den meisten Holzvillen abzubrechen. Die Gutshöfe wurden nicht in Stein ausgebaut und die meisten Standorte ganz aufgegeben. Vielleicht sind die Bewohner in das Donauvorland umgesiedelt. Die Besetzung Limesgebietes

des

Donauvorlandes

und

Die Grundrisse der Gebäude sind sehr regelmäßig und passen sich alle der ungewöhnlichen, etwas parallel verschobenen Grundfläche des Lagers an. Jede Baracke hatte an der Schmalseite sechs Pfosten und war etwa 18.5m breit. Die Länge schwankt zwischen 12m und 32m bzw. zwischen vier und zehn Pfostenreihen. Größere oder besonders ausgebaute Räumlichkeiten für die Offiziere sind nicht festzustellen. Grob eingeteilt gibt es vier große Doppelbaracken, die Platz für jeweils zwei Kohorten boten, zwei Baracken für jeweils eine Centurie und schließlich vier Baracken für jeweils eine bzw. zwei Turmen. Zwei große Bauten rechts und links des Nordtores können als Ställe interpretiert werden. In den insgesamt 15 Baracken des Lagers hatte eine 1000 Mann starke gemischte Truppe, eine cohors milliaria equitata, ausreichend Platz.

des

Für die neueren Forschungen am Limes in spätflavischer bis hadrianischer Zeit sollen beispielhaft Ergebnisse von drei Plätzen kurz aufgezeigt werden: aus dem Kastell Kösching unmittelbar nördlich der Donau, aus dem Holzerde-Kastell in Weißenburg am raetischen Limes und schließlich ein neuer Plan des Kastells Ruffenhofen. Im Bereich des heute fast vollständig überbauten Kastells in Kösching konnte in einer Bauparzelle im Ort erstmals ein Ausschnitt der Kastellbebauung untersucht werden. In der Fläche zeigte sich Ausschnitte zweiphasiger Mannschaftsbaracken (Abb. 11). Im Vicus westlich des Kastells fanden in den letzten Jahren großflächigen Ausgrabungen statt, die jedoch noch nicht komplett ausgewertet sind. Die wiederholt betonte Diskrepanz zwischen der Datierung der bislang bekannten archäologischen Funde – und hier besonders der Terra sigillata (Eschbaumer 1990: 146ff.) – und der 80 n. Chr. datierten Bauinschrift aus Kösching lässt sich nach einer ersten Durchsicht mit dem neuen Fundmaterial wahrscheinlich auflösen. Kösching kann weiterhin als das erste unter Kaiser Titus entstandene Kastell nördlich der Donau angesehen werden, und der Kastellname Germanicum unterstreicht die Bedeutung, die man diesem ersten Lager auf dem nördliche Donauufer beigemessen hatte.

In dem Lager war sehr wahrscheinlich die cohors IX Batavorum milliaria equitata exploratorum für wenige Jahre stationiert. Die Einheit ist auf einer Inschrift, die in der Weißenburger Kirche verbaut war, genannt (Czysz, Dietz, Fischer & Kellner 1995: 534ff.). Man kann davon ausgehen, dass die Bataverkohorte für einige Jahre in dem Holz-Erde-Kastell neben der ala I Hispanorum im Steinkastell Biriciana stationiert war und in der Phase der Errichtung und Sicherung der künstlichen Grenze zusätzliche Aufgaben übernahm. In der Diskussion ist noch, wohin die Bataverkohorte nach dem Abzug aus Weißenburg anschließend innerhalb Raetiens verlegt wurde. Meines Erachtens sprechen gute Gründe für die Verlegung an den neuen Standort im etwa 40kms westlich gelegenen Ruffenhofen. Das Lager in Ruffenhofen auf dem „Bürgfeld“ liegt auf einer leichten Anhöhe über dem Wörnitztal und spielte eine wichtige Rolle im Rahmen des Grenzsicherung unter den Kaisern Trajan und Hadrian. Ruffenhofen war noch vor wenigen Jahren das am wenigsten erforschte Kastell am raetischen Limes (Schönberger 1985: 471). Erstes Aufsehen machte vor einigen Jahren die Entdeckung des großen Kastellbades aus der Luft durch Otto Braasch. Von den Gebäuden des Kastells war bislang durch die Grabungen W. Kohls (ORL B 68) lediglich ein Magazinbau bekannt.

Unter den Nachfolgern Domitian und Trajan wurde die Besetzung des Donauvorlandes bis ins spätere raetische Limesgebiet hinein fortgesetzt. In Mittelfranken sind aus dieser Zeit eine Reihe temporär belegter Lager bekannt: in Weißenburg, in Burgsalach und in Theilenhofen (Hüssen 1990). Erst seit diesen Entdeckungen wissen wir, dass zwischen und neben den lange bekannten Kastellen kurzfristig Truppen stationiert waren, die die regulären Einheiten in der Phase der frühen militärischen Besetzung verstärkten bzw. beim Bau des Limes unterstützten. Für diese Truppen wurden kurzzeitig belegte Lager in reiner Holzbauweise errichtet (Hüssen 1991).

Großflächige geomagnetische Messungen durch das Bayerische Landesamt für Denkmalpflege unter Leitung von Helmut Becker haben 1999 das Kastell und den Vicus von Ruffenhofen auf einen Schlag zu einem der am besten

Das Holz-Erde-Lager in Weißenburg auf der Flur „Breitung“, unweit vom Steinkastell Biriciana entfernt, ist 538

Claus-Michael Hüssen: Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen

dokumentierten Plätze am Limes gemacht (Becker, Deinhardt & Thoma 2000). Der Magnetometerplan (Abb. 13) zeigt die komplette Innenbebauung mit Zentralbauten aus Stein und die Holzgebäude der Mannschaftsbaracken. Die zweimonatige Prospektion mit einem CäsiumMagnetometer erbrachte in kurzer Zeit einen Plan, den die Spatenforschung in mehrjährigen Kampagnen nicht hätte ergraben können.

der Präfekt in Gnotzheim selbst geweiht hat, noch eine zweite kennen, die ihm zu Ehren in Caiazzo in Italien gesetzt wurde. Durch die zweite Inschrift (CIL X 4579), die dem Q. Gavius Fulvius Proculus von seiner Heimatgemeinde für seine Verdienste gewidmet wurde, lässt sich die militärische Laufbahn des Ritters vervollständigen. Nach seiner Stationierung in Gnotzheim, die auf dem Stein in Caiazzo keine Erwähnung findet, wurde er als Tribun der 15. Kohorte freiwilliger römischer Bürger nach Leiden-Roomburg versetzt und erreichte schließlich als letzte Station seiner militärischen Laufbahn das Kommando als Legionstribun in Straßburg. Etwa zehn Jahre lang leistete er seinen Militärdienst in den drei benachbarten Nordprovinzen des römischen Reiches. Danach kehrte nach Kampanien, ins Falerner Land zurück, wo er vermutlich größere Ländereien besaß, und machte sich um seine Heimatstadt verdient.

Das Kastell ist c.190 x 197m groß. Mit einer Fläche von 3.75ha ist es nach den Lagern der ala II Flavia in Heidenheim bzw. in Aalen das zweitgrößte Auxiliarkastell am raetischen Limes. Die Umfassungsmauer ist 1.25m breit mit einem aufgeschüttetem Erdwall an der Innenseite. Zur Umwehrung gehörten mindestens vier vorgelagerte Gräben. Die Vorderfront des Kastells mit der porta principalis orientierte sich nach Nordosten zum Hesselberg hin. Hier führte eine Straße vorbei.

Literatur

Das Kastellumfeld wurde ebenfalls großflächig prospektiert, so dass wir jetzt auch den Vicus mit einer bislang nur anhand der großen Fundstreuung zu erahnenden Ausdehnung kennen. Zu sehen sind die Keller, Brunnen und Gruben der üblichen Streifenhäuser. Entlang der Ausfallstraße nach Osten erstreckte sich eine Gräberstraße mit steinernen Grabmälern.

Bakker L.1994a Kastell und Markthalle. Ausgrabungsergebnisse in Augusta Vindelicum. Das archäologische Jahr in Bayern 1993: 87-92. Bakker L.1994b Siedlungsgeschichte und archäologische Denkmalpflege in Augsburg. In B. v. Hagen & A. Wegener-Hüssen Stadt Augsburg. (Denkmäler in Bayern Bd. VII.83: München): xlv-xlvi. Becker B., Deinhardt H.-D. & Thoma H. 2000 Prospektion des Kastells Ruffenhofen mit Luftbild und Geophysik. Das archäologische Jahr in Bayern 1999: 56–59. Braasch O. 1983 Luftbildarchäologie in Süddeutschland. Spuren aus römischer Zeit. Kleine Schriften zur Kenntnis der römischen Besetzungsgeschichte Süddeutschlands 30. (Stuttgart). Braasch O. 1985 Römisches Militär an der Donausüdstraße – das Lager von Zell. Das archäologische Jahr in Bayern 1984: 95–96. Czysz W. 1981 Ein neues römisches Kastell bei Augsburg. Das archäologische Jahr in Bayern 1980: 112–113. Czysz W., Dietz K., Fischer Th. & Kellner H.-J. 1995 Die Römer in Bayern. (Stuttgart). Eschbaumer P. 1990 Das römische Nassenfels und sein Umland. (Dissertation München). Hüssen C.-M. 1990 Römische Okkupation und Besiedlung des mittelrätischen Limesgebietes. Vortrag zur Jahressitzung 1990 der Römisch-Germanischen Kommission. Bericht der Römisch-Germanischen Kommission 71: 5–22. Hüssen C.-M. 1991 Das Holzkastell auf der Breitung in Weißenburg in Bayern. In V.A. Maxfield & M.J. Dobson (ed.) Roman Frontier Studies 1989. (Exeter): 191-195. Hüssen C.-M. 1993a Die frühkaiserzeitliche Militärstation auf dem Stadtberg in Neuburg a.d. Donau. In K.H. Rieder & A. Tillmann (Hrsg.) Neuburg an der Donau. Archäologie rund um den Stadtberg. (Buch am Erlbach): 65–72. Hüssen C.-M. 1993b Eine frühkaiserzeitliche Siedlung in Weichering, Ldkr. Neuburg-Schrobenhausen. In K.H. Rieder & A. Tillmann (Hrsg.) Neuburg an der Donau. Archäologie rund um den Stadtberg. (Buch am Erlbach): 75–78. Hüssen C.-M. 1995 Römische Lager an der Donau in Ingolstadt-

Die starke Befestigung zeigt deutlich, dass das Lager vor der endgültigen Anlage von Limesposten errichtet wurde. Während die Limeslinie östlich des Hesselberges bereits in hadrianischer Zeit feststand, wird der endgültige Limesverlauf westlich von Ruffenhofen erst unter Antoninus Pius in der Mitte des 2. Jahrhunderts erreicht, als die ala II Flavia milliaria von Heidenheim aus ihre neue Garnison in Aalen bezog. Bis zu diesem Zeitpunkt hatte die Besatzung in Ruffenhofen die Hauptlast der Flankensicherung für die fruchtbaren Gebiete im Ries zu übernehmen. Aufgabe der Besatzung war die Überwachung der Täler von Wörnitz und Sulzach. Es ist durchaus vorstellbar, dass die Einheit mit der Erkundung einer günstigeren Grenzlinie im westraetischen Raum beauftragt war. Abschließend soll noch eine neue Inschrift aus dem Limesgebiet erwähnt werden, die im Jahr 1999 im Bereich des Vicus in Gnotzheim entdeckt wurde (Abb. 14). Sie lenkt den Blick auf einen Italiker, der in der zweiten Hälfte des 2. Jahrhunderts in die Ereignisse am raetischen Limes involviert war. Die bei Bauarbeiten ausgebaggerte Inschrift ist eine Weihung an die Allgöttin Diana. Sie wurde gestiftet von einem Präfekten der 3. thrakischen Kohorte mit dem Namen Gavius Fulvius Proculus (Hüssen 2001). Die Inschrift nennt zum ersten Mal den vollständigen Namen eines Präfekten der cohors III Thracum civium Romanorum equitata bis torquata, die auf einer in der Ortskirche verbauten Kaiserinschrift für Antoninus Pius aus dem Jahr 144 n. Chr. überliefert ist. Einem glücklichen Zufall ist es zu verdanken, dass wir neben der Inschrift, die

539

Limes XVIII

Zuchering. In W. Czysz, C.-M. Hüssen, H.-P. Kuhnen, C.S. Sommer & G. Weber (Hrsg.) Provinzialrömische Forschungen. Festschrift für Günter Ulbert zum 65. Geburtstag. (Espelkamp): 95–110. Hüssen C.-M. 2000a Die Donaugrenze von tiberisch-claudischer bis in frühflavische Zeit. In L. Wamser (Hrsg.) Die Römer zwischen Alpen und Nordmeer. Zivilisatorisches Erbe einer europäischen Militärmacht. (Mainz): 58–63. Hüssen C.-M. 2000b Endlatènezeitliche Fundstellen im oberbayerischen Donauraum. Bericht der RömischGermanischen Kommission 81: im Druck. Hüssen C.-M. 2001 Q. Gavius Fulvius Proculus – praefectus, tribunus, patronus. Zum Neufund einer Diana-Inschrift in Gnotzheim und zu CIL X 4579. Germania 79/2: im Druck. Hüssen C.-M., Rieder K.H. & Schaaf H. 1995 Die Römerschiffe in Oberstimm – Ausgrabung und Bergung. Das archäologische Jahr in Bayern 1994 (Stuttgart): 112– 116. Hüssen C.-M. & Wegener-Hüssen A. 1995 Römische Besetzung und Besiedlung des Donausüdufers. In K.H. Rieder & A. Tillmann (Hrsg.) Archäologie um Ingolstadt. Die archäologischen Untersuchungen beim Bau der B 16 und der Bahnverlegung. (Kipfenberg): 187-202. Mackensen M. 1987 Frühkaiserzeitliche Kleinkastelle bei Nersingen und Burlafingen. Münchner Beiträge zur Vor- und Frühgeschichte 41. (München). Pauli J. 1995 Nachbronzezeitliche Brandgräber in Zuchering, Stadt Ingolstadt. Bericht der Bayerischen Bodendenkmalpflege 34/35, 1993/1994: 146–164. Schnurbein S. v. 1983 Neu entdeckte frühkaiserzeitliche Militäranlagen bei Friedberg in Bayern. Germania 61: 529-550. Schönberger H. 1978 Kastell Oberstimm. Die Grabungen von 1968 bis 1971. Limesforschungen 18. (Berlin). Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Bericht der Römisch-Germanischen Kommission 66: 321–497. Schönberger H., Köhler H.-J. & Simon H.-G. 1989 Neue Ergebnisse zur Geschichte des Kastells Oberstimm. Bericht der Römisch-Germanischen Kommission 70: 243-319. Ulbert G. 1965 Zu den frührömischen Funden. In M. Eckstein Keltische und römische Wehranlagen auf dem Stadtberg in Neuburg a.d. Donau. Bayerische Vorgeschichtsblätter 30: 151-153.

540

Claus-Michael Hüssen: Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen

Abb. 1. Großmehring. Endlatènezeitliches Depot mi 1 elbgermanischer Terrine, 2 Almgren 65- und 3 Nauheimer Fibel. 1 M. 1:3; 2.3 M. 1:3.

Abb. 2. Karte der raetischen Donaugrenze im 1. Jahrhundrert n. Chr.

541

Limes XVIII

Abb. 3. Burgheim. Baubefunde im Vicus an der Bahnhofstraße.

Abb. 4. Neuburg a.d. Donau. Lage des Kleinkastells auf dem Stadtberg.

542

Claus-Michael Hüssen: Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen

Abb. 5. Ingolstadt-Zuchering. Topografischer Plan der Lager I–III.

543

Limes XVIII

Abb. 6. Ingolstadt-Zuchering. Befundplan des Lagers II.

Abb. 7. Ingolstadt-Zuchering. Grabenprofil 2 des Lagers I.

544

Claus-Michael Hüssen: Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen

Abb. 8. Manching-Oberstimm. Luftaufnahme des Kleinstkastells (Foto: O. Braasch).

Abb. 9. Weichering. Funde aus der frühkaiserzeitlichen Siedlung. 1 Pferdegeschirranhänger, 2–3 Fibeln, 4 Geschoßspitze, 5 Drag. 18Teller, 6–7 handgemachte Töpfe. 1–4 M. 1:2; 5–6 M. 1:3; TS-Stempel M. 1:1.

545

Limes XVIII

Abb. 10. Ingolstadt-Zuchering. Befundplan der Holzvilla, Flur „Seehofer Schaffeld“.

Abb. 11. Kösching. Befundplan der Ausgrabung im Kastell.

546

Claus-Michael Hüssen: Neue Forschungsergebnisse zu Truppenlagern und ländlichen Siedlungen

Abb. 12. Weißenburg i.Bay. Befundplan der Ausgrabungen im Breitung-Kastell (Stand 1991).

547

Limes XVIII

Abb. 13. Ruffenhofen. Magnetometerplan des Kastells und Vicus.

Abb. 14. Gnotzheim. Umzeichnung des Inschriftensteins des Q. Gavius Fulvius Proculus.

548

The Flavian timber fort at Eining (Abusina) on the Upper Danube* Markus Gschwind The purpose of this paper is to describe how it has been possible, by reconsidering the old excavation records in the light of modern knowledge, to reconstruct the whole plan of the earliest fort at Eining. In addition, a study of the stratified finds has made it possible to confirm the construction date. These results and recent reconsideration of neighbouring Flavian forts makes it possible to draw a new, slightly different picture of the building programme carried out under the Raetian governor C. Saturius.

The Roman fort of Abusina/Eining is situated on the northern frontier of the province of Raetia on the south bank of the Upper Danube, about 20 miles south-west of the later legionary base at Castra Regina/Regensburg1 (Fig. 8).

internal buildings it has been repeatedly suggested that the orientation of the fort was changed when it was rebuilt in stone.6 Knowledge of the plan of the timber fort is therefore important for the understanding of the later stone buildings, which were unearthed and consolidated by the excavators of the C19th (Figs. 1-2).

Since the excavation of the fort between 1879-1920, the remains of the stone buildings have been displayed in an archaeological park which is open to the public, well known to scholars and general visitors for over a century. Nevertheless only a few preliminary reports have ever been published about the archaeological work carried out in Eining.2 Of the huge quantity of small finds only a few exceptional pieces have been published3 and the published plans of the fort show only the stone defences and internal buildings as later rebuilt in stone (Fig. 1).

Paul Reinecke, one of the great German archaeologists of the early C20th, was fully aware of this. At the outset of his excavations, he tried to solve the question of the orientation of the Flavian fort. In 1912 Reinecke cut three sections in the south-east quarter of the fort and later enlarged them into two areas, with interesting results (Fig. 5). Karl Popp, Reinecke’s predecessor, had discovered by trenching that timber structures were to be expected in the whole praetentura. Before excavation work started, Reinecke visited Stockstadt, where members of the local antiquarian society had exposed timber structures in a Roman fort. In addition to this he placed the supervision of the excavation work at Eining in the hands of Josef Maurer, who had already dug a timber building in 1910.7

Of the first timber auxiliary fort erected at the site almost nothing has been known up to now. The 1995 edition of Die Römer in Bayern still reports that only some small finds are known from the Flavian timber fort, but no structural remains.4 The plan and dating of the first timber fort are of great interest. The dating is important for the chronology of all of the Flavian forts situated on the Raetian frontier east of Eining. For a long time the construction date of the fort depended solely on a small fragment of an inscription erected in the reign of Titus (AD79/81).5

In the excavation of 1912 Josef Maurer encountered the remains of internal buildings which had been rebuilt in timber several times (Figs. 3-4). Unable to cope with the complexities of the sequence in a modern fashion, he removed all remains of later rebuilding and uncovered the structures of the first fort erected at Eining. Considering the state of excavation techniques in the early years of the C20th the excavation of 1912 was nevertheless a very good job, especially as major parts of two buildings could be recorded. In addition to this Maurer was able to collect finds from a burnt layer which covered the remains of the first timber fort.

The location, size, orientation and plan of the fort are also of interest. Because of the unusual arrangement of the *

This paper presents an extract form my PhD thesis which was written at the Institut für Vor- und Frühgeschichte und Provinzialrömische Archäologie of the Ludwig-Maximilians-Universität München and accepted by the Philosophische Fakultät in summer-semester 1999. As a whole my Masters and PhD thesis will be published as Gschwind (forthc.). I owe much to N. Hodgson, who improved the English of this paper substantially. Nevertheless for the views put forward here and for any errors I am entirely and solely responsible. 1 Itineraria Antionini 250: 20 mp; Tabula Peutingeriana: 22 mp. 2 For an exception see Mackensen 1994, who recently discussed the structural remains of the late Roman fort in detail. 3 For an exception, see Faber 1993 who published the southern Gaulish samian ware of the site. 4 Czysz et al 1995: 104. Faber 1993: 116 summarized the former state of research as follows: “Die Strukturen des ältesten Kohortenkastells der Coh. IV Gallorum, einer Truppe mit 500 Mann Sollstärke, sind unbekannt, da noch kein Versuch unternommen wurde, die Befunde in einem Plan zusammenzustellen. Weder die exakte Lage auf dem Plattenkalk-Hochufer der Donau noch seine Größe, Ausrichtung, Bauten und deren Bauweise sind zum jetzigen Zeitpunkt geklärt”. 5 Reinecke 1914: 18, n. 2; Vollmer 1915: no. 331; Schönberger 1985: 365; Faber 1993: 114-116; 1994: 30-31 & 105; Dietz in Czysz et al 1995: 104.

Reinecke knew that the structures thus revealed were the key to the question of the orientation of the earliest fort. As the excavation and interpretation of timber structures was still being pioneered in 1912, he was initially mistaken about which way the fort faced.8 After having worked on Eining for decades, he deduced, in the end, the true orientation of the earliest fort; but as he failed to state his reasons, most scholars did not take notice of his conclusions.9 Although Reinecke was fully aware of the 6

Reinecke 1914: 17-18; Sommer 1988: 537-539, fig. 16 no. 166 & figs. 20-22. 7 Ulbert 1994: 52-55, fig. 22. 8 Reinecke 1914: 17-18. 9 Reinecke 1957: 7. For the reception see Fischer & Spindler 1984: 26;

549

Limes XVIII

importance of the excavations of 1912, he did not include them in the draft he wrote on the fort of Eining, an indication of how poor was Josef Maurer’s documentation.

In the area of these specialists’ quarters and the granary north of them (Fig. 7, horreum), a ditch was dug and a mound built in the C4th to protect the late Roman fort. The granary was almost completely destroyed by the strengthening of the C4th defences. In contrast to this the specialists’ quarters were covered and preserved by the mound. This explains why Reinecke was able to record the whole building sequence of the fort from the C1st to the C3rd at this point. In addition to this the section revealed that the layout of the first period was not altered during the principate. Details recorded during Reinecke’s other excavations show that this applies to the internal buildings as well as to the defences of the fort.

After Reinecke’s death, none of the scholars who tried to publish the work done at Eining considered the results of the excavations carried out in 1912, because the drawings of the ground plans and of the sections were stored in different towns. Having reunited the surviving parts of the documentation I have finally been able to interpret the structural remains uncovered in 1912. The results are as follows (Fig. 5). At the southern end of the long section cut from north to south, parts of the core of the rampart, the foundation trench of its rear timber revetment and the metalling of the via sagularis of the first fort could be revealed. North of this, parts of five barracks were recorded which enabled me to reconstruct the whole plan of the south-eastern quarter of the fort.

The excavations of 1912 in the south-east quarter of the fort showed clearly that the headquarters building of the Flavian timber fort was not located south of the later via praetoria (Fig. 5). Therefore the principia of the Flavian fort is to be sought below the headquarters building of the stone fort.

In contrast to the standard plan of Roman forts the officer’s houses were located on the via principalis. Another interesting fact is that in the front rooms of barrack 1 elongated pits were revealed (Figs. 3 & 5). As cavalry equipment was found in the debris of the first period of the building, it was undoubtedly a cavalry barrack.10 Barrack 2, which shows similar dimensions, was probably also a cavalry barrack. In contrast, barracks 3 and 4 probably housed two centuries of infantry (Figs. 4 & 5). Barrack 5 does not seem to have a projecting officer’s house, like a similar building in the Flavian fort of Moos-Burgstall.11

As the area of the headquarters was cleared rather than carefully excavated, a reconstruction of the history of the building can only be based on the surviving masonry. The walls of the nucleus of the building are of exactly the same thickness as the walls of the building in the south-west corner of the fort. Stratigraphic observations prove that the latter was rebuilt in stone in the second period of the fort. As it is hard to imagine that a workshop or store should be rebuilt in masonry earlier than the headquarters, the nucleus of the principia can be dated to the second period of the fort (hatched in Fig. 6).

For the reconstruction of the plan of the north-eastern quarter only three clay floors are available (Fig. 7, B.B1., B2). There were probably another four barracks, running north-south, their officer’s houses adjoining the via sagularis (Fig. 7).

Thus the thin walls of an earlier headquarters building erected in a combined timber-and-stone building technique are a feature of the first period of the fort (Figs. 2 & 6). Apart from that, the exceptional technique of filling the space between the timber uprights with masonry was, in the case of Eining, only applied in parts of the commandant’s house. Probably these parts of the building also go back to the first period of the fort. The reconstruction presented here (Fig. 7) can be supported by a comparison with the contemporary fort of Pen Llystyn in Wales, as in both forts the spacing of the principia and praetorium is very similar.13

In the western half of the fort many buildings were later rebuilt in stone (Fig. 1). In the south-western quarter Reinecke cut a series of small sections, while in the northwestern part he only recorded the stone walls excavated and conserved by his predecessors. In the court of the late Roman fort he was able to reveal a part of a large building of unusual layout (Fig. 7, building SW). As there are neither diagnostic features nor small finds, it is not possible to decide whether it housed workshops or was used as a stores-building. The soldiers working in this building probably lived in the two buildings to the north, each consisting only of a single row of rooms12 (Fig. 7, building α and building β-ε).

Although no excavation work has been carried out in the Roman fort of Eining since 1920, it is now possible to present a complete plan of the internal building structure of the Flavian fort (Fig. 7). There are several conclusions that can be drawn from the new plan of the Flavian fort of Eining. First, in spite of many earlier suggestions that the fort was orientated to the north during the first period, in Flavian times the orientation of the fort was already as unusual as it remained during the whole principate. The modification of

Sommer 1988: 537. An exception is Kellner 1971a: 64, n. 72. 10 Sommer 1995: 158-168, fig. 4-7; Hodgson in Bidwell 1999: 84-88, fig. 14-16. 11 Schönberger 1982: 203-204, fig. 12, Beilage 5. 12 Johnson 1987: 212-213, figs. 145 & 191,4; cf. 1983: 192–193; Schönberger 1978: 67-70 & 138, figs. 30, 32 & 65,3.

13

Hogg 1968: 109, 128-130, figs. 5 & 19.

550

Markus Gschwind: The Flavian timber fort at Eining (Abusina) on the Upper Danube the plan that caused the unusual arrangement of the internal buildings must have taken place at a very early stage: after the Roman army had started to build the fortifications but before they began to put up the internal buildings.

of Eining in the way that it would in a wholly timber fort. A contemporary inscription with an almost identical formula is known from Kösching.19 The Roman military road, which runs north of the Danube from the river crossing at Steppberg (Fig. 8, A) via Nassenfels (Fig. 8,6) and Kösching (Fig. 8,7) to Eining (Fig. 8,2), is therefore usually believed to have been built under the governor C. Saturius. The foundation of forts at Kösching and Eining is seen as a consequence of pushing ahead across the Danube.20

Secondly, the cavalry and infantry barracks uncovered in the south-east quarter of the fort and both stratified and unstratified finds of Flavian cavalry equipment, show that cohors IV Gallorum, the first unit stationed at Eining, was a cohors quingenaria equitata.

As mentioned the construction date of Eining is now secure. For Kösching, a new analysis of the numismatic evidence as well as the samian ware points to a date at the turn of the century or later. Therefore the inscription from the year 80 is likely to have been transferred from a fort on the south bank of the Danube such as Oberstimm (Fig. 8, 1).21 As the fort of Nassenfels (Fig. 8,7) seems to be contemporary with Kösching22 (Fig. 8,7), the military road from Steppberg to Eining was probably built some time after the foundation of the fort at Eining.

It is commonly believed that six centuries and four turmae made up a part-mounted quingenary cohors.14 As one barrack could only house one century or one turma, not all of the soldiers could be accommodated in the nine barracks of the praetentura. The remaining soldiers probably lived in the two specialists’ quarters in the south-west quarter of the fort. As demonstrated, it was possible to house a cohors equitata in the fort of Eining, although with only 1.5ha internal space it was much smaller than many forts later built at the Upper German and Raetian limes. Therefore there is no need to deduce the garrisioning of a vexillation of cohors II Tungrorum milliaria from the small size of the stone fort as postulated by Aladar Radnóti.15

The pushing forward of the border line across the Danube was seemingly unconnected with the reorganisation of the defence of the Raetian frontier under the governor C. Saturius. His reorganisation affected only the rebuilding of the Claudian forts south of the Danube and the construction of new auxiliary forts east of Oberstimm (Fig. 8,1), which are Eining, Regensburg-Kumpfmühl, Straubing and Moos-Burgstall (Fig. 8,2–5). The topographic setting chosen by he Roman army for the new fort at Eining was not influenced by the nearby river crossing of the later military road running west towards Germanicum/Kösching (Fig. 8,6).

It is more likely that cohors IV Gallorum was replaced by cohors III Britannorum as suggested by Andrea Faber.16 The latter unit was also partly mounted, which fits perfectly with the observation that the plan of the fort remained unaltered during the whole principate. The date of the construction of the Flavian fort can now be considered to be reliable, which is of great importance for the chronology of the eastern part of the Raetian limes. Small finds and pottery that stratigraphically belong to the first period of the fort confirm the Flavian date, which was already suggested on the basis of the fragment of an inscription mentioned above (AD79/81) and southern Gaulish samian ware.17 For the destruction of the first period, a coin of Hadrian gives a terminus post quem of 125.

Bibliography Bidwell

P. 1999 Hadrian’s Wall 1989-1999. A summary of recent excavations and research. (Carlisle). Braun R. et al 1992 Der römische Limes in Bayern. 100 Jahre Limesforschung. Ausstellungskataloge der Prähistorischen Staatssammlung 22. (München). Czysz W. et al 1995 Die Römer in Bayern. (Stuttgart). Davies R.W. 1967 A note on a recently discovered inscription from Carrawburgh. Epigraphische Studien 4. Beihefte Bonner Jahrbücher 25. (Köln & Graz.) Eschbaumer P. 1986 Nassenfels in römischer Zeit. Nassenfels. Beiträge zur Natur- und Kulturgeschichte des mittleren Schuttertales (Kipfenberg): 107-140. Faber A. 1993 Die südgallische Terra Sigillata aus Kastell und Vicus Eining. Zum Beginn des Militärstützpunktes. Bayerische Vorgeschichtsblätter 58: 99-122. Faber A. 1994 Das römische Auxiliarkastell und der Vicus von Regensburg-Kumpfmühl. Münchner Beiträge zur

In addition, it has to be pointed out that it is no longer a problem18 to interpret the fragment of a stone dedication as a building inscription of the Flavian timber fort, as the headquarters building and the commandant’s house were built in a combined timber-stone technique. A stone inscription does not constitute a surprise in the Flavian fort 14

Schönberger 1975: 110; Johnson 1987: 34-36 & tab. 1; Sommer 1995. Radnóti 1961: 102-103, 115. His considerations were commonly accepted: Davies 1967: 109; Kellner 1971b: 214; Ulbert & Fischer 1983: 108; Fischer & Spindler 1984: 26; Schönberger 1985: 457; Fischer in Braun et al 1992: 52; Fischer in Czysz et al 1995: 434; Nouwen 1997: 462. 16 Faber 1994: 108. 17 Faber 1993. 18 Faber 1993: 116, n. 21-25; Kortüm 1998: 41-42, ns. 159–160. 15

19

Vollmer 1915: no. 257. Schönberger 1985: 364; Faber 1993: 114-117; Dietz in Czysz et al 1995: 103. 21 Pferdehirt 1986: 253 & 292–293; Faber 1994: 105; Kemkes 1996: 19; Kortüm 1998: 42-43, fig. 89. 22 Eschbaumer in Czysz et al 1995: 485; 1986; Faber 1994: 105, n.235. 20

551

Limes XVIII

Vor- und Frühgeschichte 49. (München). Fischer H.T. 1980 Neue Untersuchungen in Kastell und Vicus von Eining/Abusina, Ldkr. Kelheim, Niederbayern. Bayerische Vorgeschichtsblätter 45: 125-153. Fischer T. & Spindler K. 1984 Das römische Grenzkastell Abusina-Eining. (Stuttgart). Gschwind M. forthc. Das römische Auxiliarkastell Abusina/ Eining an der raetischen Donau vom 1. bis 5. Jahrhundert n. Chr. Münchner Beiträge zur Vor- und Frühgeschichte 53. (München). Hogg A.H.A. 1968 Pen Llystyn: a Roman fort and other remains. Archaeological Journal 125: 101-192. Johnson A. 1983 Roman forts of the 1st and 2nd centuries AD in Britain and the German provinces. (London). Johnson A. 1987 Römische Kastelle des 1. und 2. Jahrhunderts n. Chr. in Britannien und in den germanischen Provinzen des Römerreiches. Kulturgeschichte der Antiken Welt 37. (Mainz). Kellner H.-J. 1971a Die Römer in Bayern. (München). Kellner H.-J. 1971b Exercitus raeticus. Truppenteile und Standorte im 1.–3. Jahrhundert n. Chr. Bayerische Vorgeschichtsblätter 36: 207-215. Kemkes M. 1996 Das Kastell Rißtissen und die militärische Sicherung der Donau im 1. Jahrhundert. Römer an Donau und Iller. Neue Forschungen und Funde. (Sigmaringen) 8-21. Kortüm K. 1998 Zur Datierung der römischen Militäranlagen im obergermanisch-rätischen Limesgebiet. Chronologische Untersuchungen anhand der Münzfunde. Saalburg-Jahrbuch 49: 5-65. Mackensen M. 1994 Die Innenbebauung und der Nordvorbau des spätrömischen Kastells Abusina/Eining. Germania 72: 479-513. Nouwen R. 1997 The vexillationes of the cohortes Tungrorum during the second century. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H.Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. (Oxford): 461-465. Pferdehirt B. 1986 Die römische Okkupation Germaniens und Rätiens von der Zeit des Tiberius bis zum Tode Trajans. Untersuchungen zur Chronologie südgallischer Reliefsigillata. Jahrbuch des Römisch-Germanischen Zentralmuseums Mainz 33: 221-320. Radnóti A. 1961 Neue rätische Militärdiplome aus Straubing und Eining. Germania 39: 93-117. Reinecke P. 1914 Neue Grabungen im Kastell Eining. RömischGermanisches Korrespondenzblatt 7: 17-21. Reinecke P. 1957 Das römische Grenzkastell Abusina bei Eining/Donau. Verhandlungen des Historischen Vereins für Niederbayern 83: 7–10. Schönberger H. 1975 Kastell Künzing-Quintana. Limesforschungen 13. (Berlin). Schönberger H. 1978 Kastell Oberstimm. Limesforschungen 18. (Berlin). Schönberger H. 1982 Moos-Burgstall: Ein neues Römerkastell. Berichte der Römisch-Germanischen Kommission 63: 179-279. Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Berichte der Römisch-Germanischen Kommission 66: 321-497. Sommer C.S. 1988 Kastellvicus und Kastell. Fundberichte Baden-Württemberg 13: 457-707. Sommer C.S. 1995 “Where did they put the horses?” Überlegungen zu Aufbau und Stärke römischer Auxiliartruppen und deren Unterbringung in den

Kastellen. In W. Czysz et al (edd.) Provinzialrömische Forschungen. Festschrift für Günter Ulbert zum 65. Geburtstag. (Espelkamp): 149-168. Ulbert G. 1994 Der Auerberg I. Münchner Beiträge zur Vor- und Frühgeschichte 45. (München). Ulbert G. & Fischer T. 1983 Der Limes in Bayern. Von Dinkelsbühl bis Eining. (Stuttgart). Vollmer F. 1915 Inscriptiones Baivariae Romanae. (München).

552

Markus Gschwind: The Flavian timber fort at Eining (Abusina) on the Upper Danube

Fig. 1. Abusina/Eining. Plan of the Roman auxiliary fort. Mackensen (1994): 480 abb. 1. and after Fischer (1980): 127, fig. 2. Scale 1: 1000.

553

Limes XVIII

Fig. 2. Abusina/Eining. The principia of the Roman auxiliary fort seen from the south with remains of the Flavian headquarters building in front of the wooden shelter (early C20th photograph, Bayerisches Landesamt für Denkmalpflege).

Fig. 3. Abusina/Eining. Excavation of 1912 in the south-eastern quarter of the Roman fort. Foundation trenches and stable pits of barrack 1 seen from the east-north-east. (Bayerisches Landesamt für Denkmalpflege).

Fig. 4. Abusina/Eining. Excavation of 1912 in the south-eastern quarter of the Roman fort. Foundation trenches of barrack 4 seen from the north-east. (Bayerisches Landesamt für Denkmalpflege).

554

Markus Gschwind: The Flavian timber fort at Eining (Abusina) on the Upper Danube

Fig. 5. Abusina/Eining. Plan of the south-west quarter of the Roman fort with barracks excavated in 1912 and the defences of the later stone-fort. Scale 1: 500. Plans produced by M. Gschwind and redrawn by N. Djamali (Deutsches Archäologisches Institut, Orient-Abteilung, Außenstelle Damaskus).

555

Limes XVIII

Fig. 6. Abusina/Eining. Plan of the principia of the Roman fort. Remains of the first period (blackened), principia of the second period (hatched) and later additions. Scale 1: 250. Plans produced by M. Gschwind and redrawn by N. Djamali (Deutsches Archäologisches Institut, Orient-Abteilung, Außenstelle Damaskus).

556

Markus Gschwind: The Flavian timber fort at Eining (Abusina) on the Upper Danube

Fig. 7. Abusina/Eining. Plan of the Roman fort. Internal buildings of the Flavian fort and defences as later rebuilt in stone. As relevant for the reconstruction, at barracks 6-8 and buildings α and β-ε also later structures are shown. Ditches after the largely hypothetical reconstruction of Fischer (1980): 127, fig. 2. Scale 1: 1000. Plans produced by M. Gschwind and redrawn by N. Djamali (Deutsches Archäologisches Institut, Orient-Abteilung, Außenstelle Damaskus).

557

Limes XVIII

Fig. 8. Roman forts in north-east Raetia: a) Claudian; b) founded c. AD80; c) founded at the end of the C1st or the early C2nd AD 1 Oberstimm; 2 Eining; 3 Regensburg-Kumpfmühl; 4 Straubing; 5 Moos-Burgstall; 6 Nassenfels; 7 Kösching; 8 Pförring; 9 Alkofen; 10 Mangolding-Mintraching (military occupation probable); 11 Steinkirchen; 12 Künzing; 13 Passau; 14 Passau-Innstadt; A river crossing at Steppberg. Scale 1: 1250000. Plans produced by M. Gschwind and redrawn by N. Djamali (Deutsches Archäologisches Institut, Orient-Abteilung, Außenstelle Damaskus).

558

Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg Andreas Schaub Zwei kleinere Untersuchungen im Westen der Römerstadt erbrachten 1998 (Grabung 'Volkhartstraße') und im Frühjahr 2000 (Grabung 'Auf dem Kreuz') erstmals für Augsburg Befunde und Funde der jüngeren La Tène Zeit. Beiden Grabungen gemeinsam ist ferner das vereinzelte Vorkommen frühkaiserzeitlicher Keramik - vielleicht der bislang ältesten im Stadtgebiet. In der Volkhartstraße fanden sich außerdem zwei Wehrgräben. Bei dem älteren der beiden Gräben ist nicht auszuschließen, daß er zu einer Anlage gehörte, die zeitgleich mit dem bekannten 'Waffenplatz' in Augsburg-Oberhausen bestanden hat. Der jüngere Graben, zu dem auch Reste einer Holz-Erdemauer nachgewiesen werden konnten, wurde in claudisch-neronischer Zeit verfüllt. Über den Zeitpunkt seiner Errichtung ist hingegen nichts bekannt. Obgleich die Zahl der auswertbaren Funde äußerst spärlich ist, zeichnen sich durch diese Hinweise weitere Fragen zur militärischen Situation im frühkaiserzeitlichen Raetien ab. So bestand zumindest das jüngere Lager einige Zeit neben dem bereits bekannten Vexillationslager im Osten der Altstadt. Der Bereich in dem beide o. g. Grabungen liegen, war bislang nur selten von Grabungen betroffen. Eine für das Jahr 2001 geplante Baumaßnahme - zwischen den beiden Fundpunkten gelegen - läßt vor diesem Hintergrund besonders darauf hoffen, durch die vorher stattfindenden Grabungen mehr Klarheit in diesen sensiblen Fragenkomplex zu bringen.

Die militärischen Aspekte des römischen Augsburg haben seit der Entdeckung des „Waffenplatzes“ von Oberhausen 1912/13 erst wieder ab 1993 neue Beachtung erfahren. Damals wurde das längst vermutete „Nachfolgelager“ in der östlichen Altstadt auf der Hochfläche zwischen Lech und Wertach lokalisiert. Mit seiner Größe von mehr als 10 ha beherrschte es zwischen c.5/15 und 70 n. Chr. das Gebiet der späteren raetischen Provinzhauptstadt und war während dieser Zeit auch das größte, bislang bekannte Militärlager der Provinz (Abb. 1: Zu Oberhausen - v. Schnurbein 1985: Zum Altstadtlager - Schaub 1999).

Färbung des ansonsten gelbbraunen, anstehenden Lößlehms auffallend. Eine erste bodenkundliche Autopsie ergab, dass es sich wahrscheinlich um einen frühen landwirtschaftlich genutzten Boden handelte. Darin zeichneten sich dann die frühesten Strukturen ab (Abb. 2). In Schnitt 1 handelte es sich dabei zunächst um einen c.2.5m breiten und max. 1m tiefen Sohlgraben, der mit völlig sterilem Lehm verfüllt war. Er war im Nordwesten von einem 0.3m breiten und 0.4m tiefen Gräbchen begleitet. Darin konnten zwar keine Pfostenstellungen erkannt werden, doch könnte es sich um ein Zaungräbchen handeln. Aus ihm stammt lediglich eine winzige Wandscherbe handgemachter Keramik. In Schnitt 1 konnten diesen Befunden eine Grube sowie ein kleiner Pfosten aus stratigraphischen Gründen zur Seite gestellt werden.

Seine dominierende Rolle für die topographische Genese Augusta Vindelicums zeigt sich in der Übernahme der meisten Kastellstraßen sowie in der mutmaßlichen Platzkontinuität von Stabsgebäude zum zivilen Stadtforum (Schaub 2001).

Da diese ältesten Befunde weitgehend fundleer waren, schien es legitim, ihre Nutzungszeit mit dem ältesten Material des Platzes in Verbindung zu bringen. Es handelt sich dabei um eine gewisse Anzahl von Keramik „vorgeschichtlicher Machart“, die weder charakteristische Formen noch Verzierungen aufwiesen. Unter ihnen befanden sich allerdings mehrere Fragmente sogenannter Graphittonware, die eindeutig in die jüngere Latènezeit zu datieren ist. Somit handelt es sich um den allerersten Nachweis spätkeltischer Siedlungstätigkeit auf dem Gebiet der Augusta Vindelicum.

Das die Frühzeit Augsburgs aber in facettenreicherem Licht zu sehen ist, zeigten jüngst zwei kleine Sondagen im Westen der Altstadt, deren vorläufige Ergebnisse im Folgenden vorgestellt werden.1 Volkhartstraße 18a (Abb. 1.2) Die 1998 im Vorfeld geplanter Baumaßnahmen erfolgten Grabungen umfassten zwei kleine Schnitte bei einer Gesamtfläche von c.120m2. Die mittlere Kaiserzeit hatte nur geringe Spuren an dieser Stelle hinterlassen, was mit der Lage außerhalb der Stadtmauer zu erklären ist. In spätrömischer Zeit zählte das Gelände offenbar zum großen Gräberfeld an der südwestlichen Fernstraße (nach Cambodunum/Kempten), wie zwei Körperbestattungen des 4. Jahrhunderts nahe legen.

Im Abstand von nur 1m nordwestlich des Sohlgrabens verlief, diesem parallel, ein c.2m breiter und 1m tiefer Spitzgraben. In dessen Sohlenbereich fand sich die gleiche Bodenkonsistenz wie bei dem oben beschriebene vorgeschichtliche Ackerboden. Darüber folgte eine einheitlich graubraune, lehmige Verfüllung, die nur wenige Funde enthielt. Die einzige Münze, ein abgegriffener Lugdunum-As (RIC2230), weist allerdings in die

Was die Frühzeit betrifft, so war zunächst die braunrote 1

Beide Grabungen wurden von der Stadtarchäologie Augsburg unter Letung des Verf. durchgeführt. Mein Dank gilt an dieser Stelle der gesamten Grabungsmannschaft, allen voran R. Pfennig.

559

Limes XVIII augusteische Zeit.2 Diesem numismatischen Ansatz, der ja zunächst nur als terminus postquem für den Zeitpunkt der Grabenverfüllung beweiskräftig ist, widerspricht das vergesellschaftete Keramikspektrum nicht. Es handelt sich um wenige Scherben freihandgeformter, teils kammstrichverzierter Töpfe und einiger helltoniger Krüge sowie um die Wandscherbe einer Flasche mit Resten roter Streifenbemalung (Abb. 3). Aussagekräftiger scheint jedoch der Boden einer grautonigen, „latènoiden“ Flasche mit geringen Resten eines schwarzglänzenden Überzuges (Abb. 4.1). Zumindest fehlt dieser Gefäßtyp im bislang bearbeiteten Bestand des Altstadtlagers und darf deshalb m. E. chronologisch interpretiert werden. Diesem Stück ist möglicherweise die Randscherbe einer Schüssel zur Seite zu stellen, die ebenfalls formale Wurzeln in latènezeitlicher Keramik hat und keine Parallelen im Altstadtlager besitzt, aber aus der Verfüllung eines zweiten, jüngeren Grabens (s.u.) stammt (Abb. 4.2).

auch eine relativ große Menge von Keramik und Tierknochen sprechen könnte. Die großen Pfostengruben der Holzerdekonstruktion weisen im Nordosten von Schnitt 2 eine Zweiperiodigkeit auf. Dieser Befund zeigt eine längere Belegungsdauer an, weshalb wir es wohl mit einem Standlager zu tun haben. Für den Beginn dieser Anlage kann eine abgegriffene Nemaususprägung der 3. Serie (10-14 n. Chr.) heran gezogen werden, die im Sohlenbereich der Holzerdemauer gefunden wurde. Ebenfalls aus dem Bereich der Holzerdemauer stammen zwei abgegriffene Prägungen von Caius-Caligula und/oder Claudius (RIC 57 und 58) wobei in diesem Fall aber nicht zu klären war, ob sie bei Errichtung oder Schleifung der Anlage in den Boden gelangten. Eine erste Durchsicht der Keramik aus den Fundkomplexen dieser zweiten Wehranlage bestätigt den numismatischen Ansatz von etwa tiberischer bis claudischneronischer Zeit. So fehlen sicher die Sigillataformen Drag. 35/36 und 37, denen allgemein ein verstärktes Auftreten in frühflavischer Zeit zugebilligt wird, aus gesicherten Befunden dieses Lagers. Seine Belegungszeit wäre in etwa mit den Perioden 3/4 des Altstadtlagers zu parallelisieren (Zur Periodisierung des Altstadtlagers, cf. Schaub 1999). Ein niellierter Pferdegeschirranhänger sowie der bronzene Ohrschutz eines Helmes waren die einzigen Funde militärischer Ausrüstung.

Es handelt sich bei diesem kleinen Spitzgraben (Graben 1) zweifellos um einen frühkaiserzeitlichen Befund, der am ehesten mit einem Lager in Verbindung zu bringen ist, das gleichzeitig mit dem „Waffenplatz“ von Oberhausen bestanden haben könnte. Wenngleich es aus Gründen der großen Entfernung (c.1.5 - 2kms) zu dieser Fundstelle keinesfalls das Lager sein kann, aus dem die dortigen Funde stammen. Die Verfüllung des Grabens erfolgte mit großer Wahrscheinlichkeit noch in augusteischer Zeit, während wir für den Zeitpunkt der Errichtung keinerlei Anhaltspunkte besitzen. Ebenso ungewiss ist, auf welcher Seite des Grabens das Innere des Lagers zu suchen ist, wenngleich die nachfolgende Anlage darauf hindeutet, es müsste sich südöstlich davon befunden haben.

Trotz der Kleinräumigkeit der Untersuchung in der Volkhartstraße, ergaben sich völlig neue Aspekte für die Frühzeit Augsburgs. Auf den zwar spärlichen, aber nicht zu leugnenden ersten Hinweis keltisch-vindelikischer Siedlungstätigkeit auf dem Boden von Augusta Vindelicum folgten zwei verschiedene Militäranlagen: die erste wohl noch augusteisch, die zweite tiberisch claudisch/neronisch.

Wie bereits angedeutet, wurde dieser Platz auch in der Folgezeit von einem Lager eingenommen. Der wohl keltische Sohlgraben wurde im Südosten von einem weiteren, c.4.5m breiten, muldenförmigen Graben (Graben 2) von 1m Tiefe geschnitten. Seine Verfüllung war stark sandig und zeugte von Einschwemmungen, die bisweilen stärkere Holzkohleanteile aufwiesen. Die Situation in Schnitt 2 deutet möglicherweise auf einen Durchgang, der sich unglücklicherweise in dem nicht zu untersuchenden Bereich zwischen den beiden Grabungsflächen befunden haben müsste. Das dort flache Grabenprofil sowie eine Reihe kleiner Pföstchen in der südöstlichen Grabenböschung könnten damit in Zusammenhang stehen. Wenn diese Pföstchenreihe nicht von älteren Bauspuren herrührt – was nicht gänzlich auszuschließen war – könnte es sich um die Reste eines Grabenübergangs handeln.

Für den Beweis einer echten Siedlungskontinuität von spätkeltischer zu frührömischer Zeit reichen die spärlichen Funde bislang nicht aus. Auch deshalb wäre es wünschenswert, z. B. die unmittelbar benachbarte Grabung L. Ohlenroths von 1953 (Ottmarsgässchen 7) auszuwerten, stammen doch von dort immerhin eine Nemaususprägung der Serie 1 oder 2 sowie eine gewisse Anzahl bemalter Keramik, die ansonsten in Augsburg keineswegs häufig anzutreffen ist (FMRD 1962: 26, Nr. 11). Auf dem Kreuz 17 Bei einer dreiwöchigen Notgrabung wurden im Frühjahr 2000 c.20m2 untersucht, wobei die dort angetroffene Situation die Befunde der Volkhartstraße nicht mehr isoliert stehen lassen.

Südöstlich des Grabens fanden sich die Reste einer Holzerdekonstruktion (Torturm ?). Ein im Südwesten von Schnitt 2 verlaufendes unregelmäßiges Gräbchen könnte als weiteres Indiz einer Tordurchfahrt herangezogen werden, wenn man es als Abwassergräbchen interpretieren möchte, wofür neben dem leichten Gefälle zum Graben hin

Unter Schichten der mittleren und späten Kaiserzeit waren vier Holzbauperioden zu trennen, wobei hier die beiden ältesten besondere Beachtung verdienen (Abb. 5). Eine funktionale Deutung dieser Strukturen war aufgrund des kleinen Grabungsausschnittes nicht zu gewinnen, doch fällt zumindest die durchgehend gleichbleibende Ausrichtung

2

Für die Bestimmung der Münzen danke ich Frau M. Overbeck (München).

560

Andreas Schaub: Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg der Gebäude auf. Die ältesten Funde stammen aus einer fast sterilen Lehmschicht über dem anstehenden Lößlehm (Schicht (73), siehe Katalog im Anhang). Die Verteilung der Stücke macht wahrscheinlich, dass sie mit der ersten Phase (1a) zu verbinden sind. Es handelt sich wiederum um Keramik „vorgeschichtlicher Machart“ unter denen ebenfalls die spätkeltische Graphittonware als chronologischer Indikator fungierte (Abb. 6.1). Damit vergesellschaftet war allerdings auch die Randscherbe eines Tonbalsamariums (Abb. 6.3) sowie erneut eine Prägung der 1. Lyoner Altarserie.3 Aus einer großen Lehmentnahmegrube (Grube (85), siehe Katalog im Anhang) stammt neben einer schlecht erhaltenen (und noch nicht bestimmten) Aes-Prägung eine weitere Graphittonscherbe (Abb. 7.1), mehrere Fragmente bemalter Drehscheibenkeramik (u.a. Abb. 7.2, 3) sowie zwei weitere Fragmente von Tonbalsamaria (Abb. 7.4, 5.). Bei dem größeren Fragment (Abb. 7.4) handelt es sich um den Hals sowie den Ansatz des Gefäßkörpers. Üblicherweise sind diese Gefäße innen meist mit einem Überzug versehen. Wenn ein solcher auch außen angebracht wurde beschränkt er sich in der Regel auf den Rand und Hals. In vorliegendem Fall scheint ein Überzug innen und außen, auch den Gefäßkörper zu bedecken. Erste Recherchen haben nun ergeben, dass diese Beschaffenheit offenbar chronologisch zu werten ist. So sind z. B. immerhin 4 von 45 Exemplaren aus Dangstetten innen und außen am Gefäßkörper überzogen, während bei allen 13 Exemplaren des tiberischen Auerberges der äußere Überzug fehlt (Fingerlin 1986, 1998; Flügel 1999). Unabhängig einer abschließenden Prüfung der chronologischen Relevanz dieses Umstandes, weist auch der übrige Keramikbestand in die frühe Kaiserzeit.4 Und obwohl besser zu datierende Feinkeramik fehlt, scheint die o. g. Lyoner Altarprägung einen realistischen chronologischen Rahmen der entsprechenden Befunde anzuzeigen. Leider kann ein militärischer Charakter dieser wenigen Siedlungsreste weder bestätigt noch widerlegt werden.

Stammen doch eine Nemaususprägung der 1. Serie, eine der Serien 1 oder 2, eine der Serien 1, 2 oder 3, 2 der Serie 2 und eine der Serie 3 von diesem Gelände (FMRD 1962: Nr. 8-10, 12-14). Alle drei Plätze liegen in der Peripherie des römischen Stadtzentrums und somit in Arealen, die bislang äußerst selten Ziel archäologischer Grabungen waren. Es bestehen deshalb durchaus berechtigte Hoffnungen, dieses noch schüttere Bild der frühesten Siedlungsphase Augsburgs in Zukunft besser beurteilen zu können. Es zeichnet sich aber schon jetzt ein Bild ab, dass weder der „Waffenplatz“ Oberhausen, noch das zeitlich anschließende Altstadtlager isoliert zu sehen sind. Vielmehr scheint Augsburg, ähnlich der Plätze Haltern, Mainz oder Xanten, einen militärischen Schwerpunkt der frühen Kaiserzeit zu bilden, bei dem neben einem größeren Standlager auch immer wieder mit temporären, unterschiedlichen Truppenkonzentrationen zu rechnen ist. Vor diesem Hintergrund sind auch weitere – hier nicht berücksichtigte – Spitzgrabenstücke zu sehen, die aber ebenfalls noch ihrer abschließenden Auswertung harren. Die keltisch-vindelikische Vorgängersiedlung, deren Kontinuität in die Okkupationszeit, wie gesagt, bislang nicht gesichert ist, könnte durchaus eine beachtliche Größe besessen haben. Zumindest liegen die beiden geschilderten Grabungen c.450m auseinander und es wäre wohl ein großer Zufall, hätte man mit den kleinräumigen Sondagen zwei isolierte Kleinsiedlungen angeschnitten. Anhang Fundkatalog der Befunde (73) und (85) der Grabung „Auf dem Kreuz 17“ Schicht (73) Bis zu c.0.1m mächtige lehmige Schicht auf dem anstehenden Lößlehm. Kaum von letzterem zu trennen, lediglich geringfügig hellgrau. Durchsetzt mit wenigen Holzkohle- und verbrannten Lehmpartikeln.

Die Gemeinsamkeit beider geschilderten Grabungen liegt zum einen im Vorkommen jüngerlatènezeitlicher Keramik zum anderen aber auch in der jeweiligen Vergesellschaftung mit der offenbar ältesten bislang bekannten römischen Keramik der Augsburger Altstadt. Die beiden stratifizierten Lugdunumprägungen deuten für das Einsetzen der römischen Siedlungstätigkeit vage auf den Horizont Haltern-Oberhausen, doch ist für weiterführende Analysen der Fundbestand noch zu gering. Angemerkt sei noch, dass bei den laufenden Grabungen „Am Pfannenstiel“, nördlich der Altstadt, inzwischen die erste Nauheimer Fibel aus Augusta Vindelicum zu tage gekommen ist.5 Und auch hier lässt ein Blick in den Altbestand der Augsburger Fundmünzen aufhorchen.

Abb.6: 1: WS, Graphitton. Schwarzgrau, weich. Im Streiflicht schwach erkennbare, senkrechte Rillen (feiner Kammstrich ?). Wandungsstärke 0.6cms. 2: WS, dunkelgrau, sandig gemagert (Korngröße bis 2mm); hart. Freihandgeformt mit grobem, tiefem Kammstrich. Wandungsstärke 0.85cms. 3: RS, Tonbalsamarium. Ton hellrot, fein geschlämmt, ohne Einschlüsse; hart. Innen und außen rötlich-brauner Überzug. 4: RS, Schälchen. Ton hellbeige, fein geschlämmt, glimmerhaltig; hart. Oberfläche außen geglättet. 5: BS, Krug. Ton hellbeige, fein geschlämmt, glimmerhaltig; weich. Oberfläche außen geglättet.

3

Die Bestimmung dieser Münze erfolgte im Zuge der Grabung, vor der Restaurierung, durch den Verf. 4 Neben der bemalten Keramik sind hier vor allem die beiden Schultertöpfe mit gerillter Schulter zu nennen. 5 Die örtliche Letung dieser Grabung liegt in den Händen von V. Babucke (Friedberg), dem ich für die Mitteilung und Diskussion herzlich danke.

Ohne Abb.: 6: WS, Flasche ? Verbrannt, glimmerhaltig; weich. Oberfläche außen geglättet, mit Resten flächiger 561

Limes XVIII

7: 8: 9: 10: 11:

roter Bemalung. Wandungsstärke 0.65cms. WS, Flasche ? Ton hellorange, glimmerhaltig; weich. Oberfläche außen geglättet, mit Resten flächiger hellroter Bemalung. WS, Amphore. Ton hellorange, feinsandig. Oberfläche außen hellbeige; hart. Wandungsstärke 1 – 1.3cms. Neun WS, Krüge ? Helltonig; weich. Drei davon verbrannt. WS ? Grautonig, abgerollt. Tierknochen, n. n. bestimmt.

17: 18: 19:

Scheibenförmiges Eisenfragment, Dm c.2.8cms, Stärke 0.6cms (n. n. restauriert). Tierknochen, n. n. bestimmt. Ae Münze. Stark korrodiert, n.n. bestimmt und restauriert.

Literatur Fingerlin G. 1986 Dangstetten I. Katalog der Funde (Fundstellen 1 bis 603). Forschungen und Berichte Vor- und Frühgeschichte Baden-Württemberg 22. (Stuttgart). Fingerlin G. 1998 Dangstetten II. Katalog der Funde (Fundstellen 604 bis 1358). Forschungen und Berichte Vor- und Frühgeschichte Baden-Württemberg 69. (Stuttgart). Flügel C. 1999 Der Auerberg III. Die römische Keramik. Münchner Beiträge Vor- und Frühgeschichte 47. (München). FMRD 1962 Die Fundmünzen der Römischen Zeit in Deutschland. H. Gebhart & K. Kraft (Hsrg.) Abt. I Bayern, Band 7: Schwaben. (Berlin). Schaub A. 1999 Das frührömische Militärlager im Stadtgebiet von Augsburg. Neue Überlegungen zur Militärgeschichte Raetiens im 1. Jh. n. Chr. In N. Gudea (Hrsg.) Roman Frontier Studies 1997. Proceedings of the XVI th International Congress of Roman Frontier Studies 1997. (Zalău): 365–372. Schaub A. 2001 - im Druck Zur Lokalisierung des Forums von Augusta Vindelicum. Archäologische Aspekte zur Diskussion über die Provinzhauptstadt Raetiens. Augsburger Beiträge Archäologie 3. Schnurbein v. S. 1985 Die Funde von Augsburg-Oberhausen und die Besetzung des Alpenvorlandes durch die Römer. In J. Bellot, W. Czysz & H. Krahe (Hrsg.) Forschungen zur provinzialrömischen Archäologie in Bayer. Schwäbische. Schwäben Geschichtsquellen und Forschungen 14. (Augsburg): 15-43.

Grube (85) Grube mit unregelmäßigem Rand und Sohle. Wurde nicht vollständig ausgegraben. Verfüllung: Kompakter, rötlichbrauner Lehm mit einer kleinen Holzkohlekonzentration. Abb.7: 1: WS, Graphitton. Schwarzgrau; mäßig hart. Reste einer Kammstrichverzierung. 2: WS, Flasche? Ton hellorange, feinsandig, geringer Glimmeranteil; mäßig hart. Oberfläche außen geglättet, mit Resten hellrot-brauner Bemalung. 3: WS, Flasche? Ton hellorange; weich. Oberfläche außen hellbeige, geglättet, mit Resten dunkelbraun-roter Bemalung. 4: WS, Tonbalsamarium. Hals mit Ansatz des Gefäßkörpers. Ton hellbeige, fein geschlämmt, ohne Einschlüsse; hart. Innen und außen flächiger hellrotbrauner Überzug. 5: WS, Tonbalsamarium. Ton hellocker, fein geschlämmt, ohne Einschlüsse; hart. Oberfläche außen ungewöhnlich fein geglättet (poliert ?). Innen flächiger schwarzbrauner Überzug. 6: RS, Schultertopf. Ton rotbraun, sandig, glimmerhaltig; hart. Reste von Rußspuren am Rand (Gebrauch ?). 7: Zwei WS, BS, Schultertopf. Ton hellgrau, sandig; hart. An dickeren Partien im Kern graubraun. 8: WS, Schüssel mit Wandknick. Ton hellorange, feinsandig, glimmerhaltig; hart. 9: BS, Krug. Verbrannt. Ton fein geschlämmt, glimmerhaltig; hart. Oberfläche außen geglättet. ohne Abb.: 10: WS, Flasche? Ton hellorange, fein geschlämmt, glimmerhaltig; mäßig hart. Oberfläche außen geglättet, mit Resten dunkelrot-brauner Bemalung. 11: 18 WS, Krüge. Helltonig; weich. Drei davon verbrannt. 12: Vier WS, freihandgeformte Töpfe/Schüsseln. Grautonig, sandig; hart. Einmal mit Kammstrich. 13: WS, Amphore/Reibschüssel ? Rottonig; weich. Wandungsstärke 1.3cms. 14: Sechs fragmentierte Eisennägel. 15: Kleines Stück Eisenschlacke. 16: Mehrere verbrannte Lehmstückchen (ohne Rutenabdrücke!). 562

Andreas Schaub: Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg

Abb. 1. Augsburg, vorläufiger Gesamtplan der römischen Steinbauten mit Lage der römischen Wehrgräben des 1. Jahrhunderts. 1. Altstadtlager, Umwehrungsperiode 2 (tiberisch-neronisch), 2. Grabung Volkhartstraße 18 a, 3. Grabung Auf dem Kreuz 17, 4. Fundstelle der Nauheimer Fibel (Am Pfannenstiel).

563

Limes XVIII

Abb. 2. Augsburg, Volkhartstraße 18 a. Ergänzter Gesamtplan der spätkeltischen und frühkaiserzeitlichen Befunde.

Abb. 3. Augsburg. Volkhartstrasße 18 a. Keramik aus dem frühkaiserzeitlichen Spitzgtaben 1.

564

Andreas Schaub: Neue Befunde zur Spätlatène - und frühen Kaiserzeit in Augsburg

Abb. 4. Volkhartstraße 18 a. Frühkaiserzeitliche Keramik „latènoider“ Formen. 1. Flasche aus Spitzgraben 1,2. Schüssel aus Graben 2.

Abb. 5. Augsburg, Auf dem Kreuz 17. Gesamtplan der römischen Befunde unter Hervorhebung der frühen Perioden und Verteilung spätkeltischer und frühkaiserzeitlicher Einzelfunde.

565

Limes XVIII

Abb. 6. Augsburg. Auf dem Kreutz 17. Spätkeltische und frühkaiserzeitliche Keramik aus Schicht (73).

Abb. 7. Augsburg, Auf dem Kreuz 17. Spätkeltische und frühkaiserzeitliche Keramik aus Grube (85).

566

Neue Grabungen im Vicus des Auxiliarkastells Favianis (Noricum) – ein Handwerksbezirk im östlichen Kastellvicus Stefan Groh & Helga Sedlmayer Bei neueren Grabungen des Österreichischen Archäologischen Instituts in Mautern an der Donau (Niederösterreich) wurden in den Jahren 1996–1999 sowohl Flächen im Inneren des Kastells (Groh 1999a; 2000) als auch im südlichen und östlichen Lagervicus (Groh 1998; 1999b; 2001) untersucht (Abb. 1). Das Auxiliarkastell war vom ausgehenden 1. Jh. n. Chr. bis in das 5. Jh. n. Chr. besetzt, die neueren Untersuchungen, insbesondere im Lagerinneren, ermöglichten eine verfeinerte Periodisierung und Aussagen über das jeweilige ökologische und ökonomische Umfeld (Groh & Sedlmayer in Vorbereitung).

Das im ersten Jahr eines Forschungsprojektes dokumentierte Fundmaterial der Grabungen Mautern Vicus Ost 1997–1999 läßt erste Rückschlüsse auf die Siedlungsentwicklung und die Bewohner dieses östlichen Abschnitts des Lagervicus zu. Das zahlenmäßig wenig bedeutende, jedoch in seiner Zusammensetzung und charakteristischen Typenverteilung sehr aussagekräftige Fibelspektrum (Abb. 3) erlaubt einen durch hauptsächlich in die zweite Hälfte des 1. Jhs. n. Chr. datierte Westformen (Britannien: Fowler A3; Gallien/Obergermanien: Riha 5.16.2; Riha 5.17.5) geprägten Horizont von einem ab dem dritten Viertel des 1. Jhs. n. Chr. (Almgren 68/69; Almgren 73, eingliedrig) bzw. jünger (Almgren 15/Jobst 9B; Almgren 70) datierten Horizont zu differenzieren, der eine Konsolidierung im mitteldonauländischen und speziell im ufernorischen Milieu dokumentiert. Diese Konsolidierungsphase fällt in die wirtschaftlich prosperierendste Phase der Siedlungsentwicklung während der letzten Jahrzehnte des 1. Jhs. n. Chr. und der ersten Jahrzehnte des 2. Jhs. n. Chr. Das wenig differenzierte Spektrum dieses Horizonts ist gänzlich in den norischen Formenkreis eingebunden. Akkulturationen von Fremdformen sind in dieser Phase nicht zu fassen. Ein neuerliches überregionales Korrespondieren mit Trachtregionen, im speziellen mit den Fibelkreisen der obergermanischen und der pannonischen Provinzen (Jobst 13A; Böhme 46c; Jobst 31A), ist zwischen den mittleren Jahrzehnten des 2. Jhs. und dem Beginn des 3. Jhs. zu konstatieren. Anhand der jüngsten Typen des späteren 2./frühen 3. Jhs. n. Chr. sind somit Evidenzen eines möglichen Zuzugs aus Obergermanien und Pannonien parallel mit einer ausschließlich für die norische Limesregion typischen Fibelform (Jobst 10B) zu fassen.

1997–1999 konzentrierte sich die Grabungstätigkeit auf die Untersuchung einer 7000m2 großen Fläche im östlichen Teil des Lagervicus. Das mit seiner Nordfront an einer natürlichen Geländekante (Flußterrasse) situierte Kastell umgab im Westen, Süden und Norden eine c.20ha großen Siedlung zivilen Dorfcharakters. Bei Anlage der Bauten im östlichen Vicus parzellierte man das Bauland in streifenförmige 10m x 40m große Flächen, die man in der Kaiserzeit mit Holzgebäuden verbaute (Streifenhäuser bzw. Komplexbauten). Die schmalen, langrechteckigen, analog dem Verlauf der Kastellmauern in nordwestsüdöstlicher Richtung orientierten Parzellen nahmen Bezug auf eine parallel zur Limesstraße verlaufende Ost-WestStraße in Richtung Augustianis-Traismauer (Abb. 2). Auf dem Grabungsareal konnten insgesamt elf Parzellen untersucht werden. Die innerhalb einer Parzelle vorgefundenen Baustrukturen erlauben Rückschlüsse auf die spezifische Funktion des jeweiligen Hauses. Als wichtigste Bauelemente der Streifenhäuser sind einfache Grubenhütten mit Doppelpfostenstellung (vgl. Gabler 1982; Varsik 1999) und komplexere, in ihrem Grundriß langrechteckige Mehrraumhäuser anzuführen, die in einer relativchronologisch jüngeren Phase errichtet wurden. Als letzte Baumaßnahme sind im untersuchten Areal einfache Kleinhäuser mit Steinfundamenten und Schlauchheizungen erfaßbar (Groh & Lindinger 1999).

Die in Abb. 4 ersichtliche Verteilung der Fundmünzen aus dem östlichen Lagervicus bestätigt die bei der Analyse der Fibelformen gewonnenen Datierungsansätze. Im Diagramm wurden die stratifizierten Fundmünzen aus dem Grabungsbefund (schwarz) jenen aus der obersten umgelagerten Humusschicht (grau) gegenübergestellt, wodurch sich ein stringentes Bild von der Bebauungsgeschichte des Platzes zeichnen läßt. Die stratifizierten Fundmünzen bestätigen, mit geringen vorflavischen Prägungen, einen Horizont vom ausgehenden 1. Jh. n. Chr. bis gegen die Mitte des 2. Jhs. n. Chr., die wenigen Prägungen des späten 3. und 4. Jhs. n. Chr. reflektieren die äußerst schlecht erhaltenen Baustrukturen dieses Zeithorizontes. Anhand der Verteilung der Oberflächenfunde und der Münzen aus dem Humus kann nämlich auf eine Besiedlung bzw. intensive Nutzung des Areals auch in der Spätantike geschlossen werden.

Analoge Verbauungsschemata sind auch für den Lagervicus Süd bezeugt, wo bei Grabungen der Jahre 1988–1991 einzelne durch Gräbchen voneinander getrennte Parzellen, Grubenhütten und Erdkeller einer oder mehrerer Holzbauphasen und Steinfundamente von spätrömischen Häusern mit Fußbodenheizungen (Schlauchheizungen) dokumentiert wurden. In einigen der Parzellen des südlichen Vicus richteten sich, wie auch im Vicus Ost, Töpferbetriebe ein (Stangelmaier & Zabehlicky 1989; Zabehlicky 1990; 1991/92). Eine wenn auch bis dato punktuelle Chronologie der Verbauung im Vicus Süd konnte anhand einer kleinflächigen Grabung des Jahres 1998 gewonnen werden (Groh 2001).

567

Limes XVIII

An wirtschaftlichen Grundlagen lassen sich anhand signifikanter Funde aus diesem Vicusabschnitt das Töpferhandwerk, die Metall- und Bein-/Hornverarbeitung sowie die Textilerzeugung belegen. Direkte Befunde des Töpferhandwerks sind durch vier Keramikbrennöfen gegeben. Die Töpferwerkstätten lagen c.250m vom Kastell entfernt im Südostteil des Vicus. Es wurden zwei Töpferbetriebe mit jeweils zwei Brennöfen freigelegt; im Zuge dieser Untersuchungen wurde auch ein im Jahre 1930 freigelegter Töpferofen (Riedl 1941: 16, „Töpferofen 1“; Risy & Zabehlicky 1998: 56, Nr.1 Taf. 1,2; Taf.12,1) exakt verortet und in den allgemeinen Kontext der Werkstätten gestellt (Ertel 1998: 94 ff. Abb. 1, mit unzutreffender Kartierung, und Abb. 5). Im Umfeld der Brennöfen ließen sich auch infrastrukturelle Einrichtungen untersuchen, woraus Rückschlüsse auf die innere Organisation der Werkstätten gezogen werden können. Im straßennahen Bereich gelegene einfache überdachte Grubenhütten dürften als Werk- und Wohnräume genutzt worden sein, die Töpferöfen und zahlreiche, wohl im engen Kontext mit der Aufbereitung des Tones und der Produktion der Gefäße zu sehende Gruben befanden sich im Mittelteil der Parzellen. Der hintere Teil beherbergte Latrinen und Abfallgruben.

/Vorratsgeschirr charakterisierte, alle funktionalen Bereiche bedienende Produktpalette läßt sich mit weiterem Töpfereiabfall aus Altgrabungen vergleichen (Gassner 1997: 212 Taf. 23,2; Risy & Zabehlicky 1998: 56f. Taf. 12,1; Ertel 1998: 97 mit Anm. 32). Über dem Kernbereich des zwar zerdrückten, jedoch in großen Teilen rekonstruierbaren Brennguts lagen Verfüllstraten kleinteilig zerscherbten keramischen Materials, welches typologisch keine Differenzierung vom Brenngut des Kernbereichs erkennen läßt. Anstelle der Lochtenne lag nur noch eine kompakte Schicht graugrün gebrannten Lehms vor, der mit vitrifizierter Schlacke und Keramik versetzt war. In der verstürzten Heizkammer wurden neben der architektonischen Unterkonstruktion der Tenne auch mehrfach gebrannte und somit stark vitrifizierte Gefäße als Stützen beobachtet. Der Vergleich mit den datierten Funden aus den Kastellgrabungen 1996–1997 sowie die Fundvergesellschaftung spezifischer Analogien in den zahlreichen Abfallgruben dieses Töpferbetriebs läßt ein Parallelisierung mit den Funden der Kastellperioden 2 bzw. der frühen Kastellperiode 3 erkennen und somit auf eine Datierung des Werkstattbetriebs in die erste Hälfte des 2. Jhs. n. Chr. schließen. Über diese Primärbelege des Töpferhandwerks hinaus sind aus dem östlichen Vicusbereich in sekundärer Fundlage auch spezifische Funde der Metallverarbeitung von besonderer Bedeutung. Für eine intensive Eisenindustrie sprechen die mehr als 30kgs Schlacken, darunter typische kalottenförmige Ausheiz- bzw. Schmiedeschlacken (SennLuder 1997: 29ff. Abb. 12), sowie vier Düsenziegel der Herd- bzw. Essenwand (Senn-Luder 1997: 28ff. Abb. 1011.13; Serneels 1999: 208, Abb. 169; Orengo u.a. 2000). Ein Steckamboß sowie mehrere Schleifsteine sind gleichfalls als Evidenzen des Schmiedehandwerks zu nennen. Quantitativ überwiegen unter den aus Eisen gefertigten Fundobjekten dieses dicht besiedelten Areals naturgemäß die Baubeschläge, das Spektrum der signifikanten Militaria-, Geräte- und Werkzeugformen ist wenig differenziert (Abb. 8).

Die vier Keramikbrennöfen (Ofen 1–4) entsprechen dem in der römischen Keramikproduktion am häufigsten vertretenen Typ des Ofens mit runder Brennkammer, Lochtenne und Bedienungsgrube (Henning 1977; Duhamel 1978/79; Gassner, Jilek & Sauer 1997). Der Durchmesser der Brennkammern reicht von 0.8 – 2m, drei der vier Öfen verband ein kurzer, c.0.7m langer Schürkanal mit einer im Grundriß rechteckigen Bedienungsgrube. In Ofen 1, mit einem c.1.5m langen, seitlich gemauerten Schürkanal, trugen neun radiale Stützmauern die nur noch fragmentarisch erhaltene Lochtenne. Die Heizkammer war mit großen Mengen teils nur grob zerbrochener Gebrauchskeramik, darunter auch verglaste und deformierte Fehlbrände, verfüllt (Abb. 5). Dieser Ofen besaß die größte Brennkammer mit 2m Durchmesser. Die kleineren Öfen 2, 3 und 4 (Durchmesser der Brennkammern c.1.7m bzw. 0.8m) zeigten unterschiedliche Stützvorrichtungen für die Lochtenne: Das Spektrum reicht von vier mittig in der Heizkammer situierten, frei stehenden, über seitlich an die Ofenwand gesetzten Pfeilern bis hin zu Bogenkonstruktionen.

Die Fundlage und Vergesellschaftung von Düsenziegeln und Kalottenschlacken lassen darauf schließen, daß ihre Herstellungsbzw. Gebrauchszeit der Hauptproduktionsphase der Töpfereien vorangeht. Eine exakte Differenzierung der durch unterschiedliche Handwerkszweige charakterisierten Produktionsphasen im östlichen Lagervicus des Kastells Favianis-Mautern an der Donau wird nach Sichtung und Auswertung aller für diese Fragestellung relevanten Fundzusammenhänge als ein wichtiges Ergebnis der Projektarbeit angestrebt.

Entscheidend ist in diesem Zusammenhang der Nachweis des samt des letzten Fehlbrands aufgegebenen Brennofens (Ofen 1). Eine relative Quantifizierung des in situ vorgefundenen Brennguts war insofern möglich, als der Kernbereich mit c.200 Gefäßen (Abb. 5) über der offensichtlich eingestürzten Heizkammer ohne sekundäre Störung zu erfassen war (Abb. 6). An Leitformen der produktionstechnisch einheitlichen, weißgrau gebrannten Grobkeramik sind Töpfe mit ausgebogenem, kantigem Rand, Töpfe mit Flachrand, Becher mit ausgebogenem, kantigem Rand, Einhenkelkannen, Knickwandschüsseln, Teller mit eingebogenem Rand und konische Deckel zu nennen (Abb. 7). Die durch Trink-, Eß-, Koch-

Literatur Almgren O. 1923 Studien über nordeuropäische Fibelformen. Mannus-Bibliothek 32: (Leipzig). Böhme A. 1972 Die Fibeln der Kastelle Saalburg und Zugmantel. Saalburg-Jahrbuch 29: 5-112. Duhamel P. 1978/79 Morphologie et évolution de fours céramiques en Europe Occidentale – protohistoire,

568

Stefan Groh & Helga Sedlmayer: Neue Grabungen im Vicus des Auxiliarkastells Favianis (Noricum) monde celtique et Gaule romaine. Acta praehistorica et archaeologica 9/10: 49-76. Ertel C. 1998 Grabungen von Alexander Gaheis in Favianis/Mautern (1930). Fundberichte aus Österreich 37: 91-94. Fowler E. 1960 The origins and development of the penannular brooch in Europe. Proceedings of the Prehistoric Society 26: 149-177. Gabler D. 1982 Aspects of the development of late Iron age settlements in Transdanubia into the Roman period (Evidence based upon the excavations at Szakály, in southern Hungary). In D. Gabler, E. Patek, & I. Vőrős Studies in the Iron age of Hungary. British Archaeological Reports Int. Ser. 144. (Oxford): 57-127. Gassner V. 1997 Mautern – Favianis. In H. Friesinger & F. Krinzinger (edd.) Der römische Limes in Österreich (Wien): 208-215. Gassner V., Jilek S. & Sauer R. 1997 Der Töpferofen von Carnuntum. In H. Stiglitz (ed.) Das Auxiliarkastell Carnuntum 1. Forschungen 1977–1988. Österreichisches Archäologisches Institut Sonderschriften 29: (Wien): 179-256. Groh S. 1998 Mautern. Fundberichte aus Österreich 36: 847850. Groh S. 1999a Neue Ausgrabungen im Kastell von MauternFavianis. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 375-377. Groh S. 1999b Ein Töpferbezirk im östlichen Vicus des Kastells Mautern/Favianis am norischen Limes. Instrumentum 10: 16. Groh S. 2000 Die Grabungen des Jahres 1996 im römischen Kastell Favianis (Mautern an der Donau) – Ein Vorbericht. In V. Gassner u.a. Das Kastell Mautern – Favianis. Der römische Limes in Österreich 39: (Wien): 123-130. Groh S. (ed.) 2001 Die Grabung 1998 im Kastellvicus Süd von Mautern an der Donau/Favianis. Ergänzungshefte zu den Jahresheften des Österreichischen Archäologischen Institutes in Wien 1. (Wien). Groh S. & Lindinger V. 1999 Die Grabungen 1999 im östlichen Lagervicus von Mautern/Favianis. Archäologie Österreichs 10/2: 22-24. Groh S. & Sedlmayer H. nd Vorbereitung Forschungen im Kastell Mautern-Favianis. Die Grabungen der Jahre 1996 und 1997. Der römische Limes in Österreich. (Wien). Henning J. 1977 Entwicklungstendenzen der Keramikproduktion an der mittleren und unteren Donau im 1. Jahrtausend unserer Zeitrechnung. Zeitschrift für Archäologie 11: 181-206. Jobst W. 1975 Die römischen Fibeln aus Lauriacum. Forschungen in Lauriacum 10: (Oberösterreichisches Landesmuseum, Linz). Orengo L., Bonnon J.M. & Bevilacqua D. 2000 L´emploi des blocs-tuyères dans les forges antiques du centre de la Gaule (Avergne, Lyonnais et Forez au Deuxième âge du Fer et à l´époque romaine). Découvertes archéologiques et expérimentation. In M. Feugère & M. Guštin (edd.) Iron, blacksmiths and toils. Ancient European crafts. Acts of the Instrumentum conference at Podsreda (Slovenia) in April 1999. (Monographies instrumentum 12: Editions Monique Mergoil, Montagnac): 121-136. Riedl H. 1941 Mautern zur Römerzeit. Niederdonau/Natur und Kultur 9: Museum des Reichsgaues Niederdonau.

(Wien). Riha E. 1979 Die römischen Fibeln aus Augst und Kaiseraugst. Forschungen in Augst 3: Römermuseum Augst. Risy R. & Zabehlicky H. 1998 Handwerk im römischen Mautern. In Gemeinde Mautern (ed.) Römermuseum Favianis – St. Severin, Mautern/Donau. (Mautern): 5457. Senn-Luder M. 1997 Die Weiterverarbeitung des Eisens. In Vereinigung des Archäologisch-technischen Grabungspersonals der Schweiz (ed.) Technique des fouilles – Grabungstechnik. Minerai, scories, fer – Erze, Schlacken, Eisen. (Basel): 29-37. Serneels V. 1999 Les vestiges sidérurgiques. In P. ChadronPicault & M. Pernot (edd.) Un quatier antique d´artisanat métallurgique à Autun (Saône-et-Loire). (Documents d´archéologie française 76: Èdition de la Maison des sciences de l´ Homme, Paris): 201-213. Stangelmaier U. & Zabehlicky H. 1989 Mautern. Grabungen 1988. Jahreshefte des Österreichischen Archäologischen Institutes in Wien 59: 40-41. Varsik V. 1999 Ländliche Besiedlung im Hinterland des Kastells Gerulata. Siedlung der autochthonen Bevölkerung in Rusovce. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies. (Zalău): 629-642. Zabehlicky H. 1990 Mautern. Grabungen 1989. Jahreshefte des Österreichischen Archäologischen Institutes in Wien 60: 56-58. Zabehlicky H. 1991/92 Mautern. Grabungen 1990-1991. Jahreshefte des Österreichischen Archäologischen Institutes in Wien 61: 40-44.

569

Limes XVIII

Abb. 1. Noricum im 2.Jh. n. Chr.

Abb. 2. Blick von Süden auf die Grabungsfläche 1997-1999 im östlichen Kastellvicus von Favianis-Mautern an der Donau (Photo: S. Groh, ÖAI).

570

Stefan Groh & Helga Sedlmayer: Neue Grabungen im Vicus des Auxiliarkastells Favianis (Noricum)

Abb. 3. Fibelspektrum des östlichen Kastellvicus von Favianis-Mautern an der Donau, Grabungen 1997– 1999 (n = 23).

Abb. 4. Fundmünzen auf den Parz. 800/1, 800/4, 800/7, 800/9 und 805/1 im östlichen Kastellvicus von Favianis-Mautern an der Donau, Grabungen 19971999 (n = 51).

Abb. 5. Töpferofen 1 mit verfüllter Brennkammer (Photo: S. Groh, ÖAI).

Abb. 6. Relative Verteilung der Gefäßformen im Brenngut von Töpferofen 1, östlicher Kastellvicus von FavianisMautern a. d. Donau, Grabung 1998 (n = 204).

571

Limes XVIII

Abb. 7. Typenspektrum aus Töpferofen 1 im östlichen Kastellvicus von Favianis-Mautern an der Donau (Graphik: H. Sedlmayer).

Abb. 8. Typenspektrum aus Töpferofen 1 im östlichen Kastellvicus von Favianis-Mautern an der Donau (Graphik: H. Sedlmayer).

572

Vindobona – legionary fortress, canabae legionis and necropolis Michaela Kronberger & Martin Mosser Research into Roman Vienna during the last few years has generated a lot of new results concerning the structure and chronology of ancient Vindobona These results are in contrast to the conclusions drawn in the past. The content of our paper will be a summary of the new results to create a new picture of Roman Vindobona. The issues are as follows: • excavations in the retentura and principia of the legionary fortress ended with the conclusion that we can no longer maintain the theory of a stone-built fortress in the first period. Two periods of wooden structures and adobe walls are evident. • with respect to the beginning of Roman settlement in the area of Vindobona, this can be placed in the first half of the C1st AD. But there are no structures evidently belonging to these findings. The settlement of this time has to be seen in a military context. • the main period of the legionary fortress seems to be the C4th. Paved roads and courtyards, heating systems and stone walls, which were not customary before this time, can be seen in this late Roman period. • outside of the fortress, there was a different development: A great part of the area of the C2nd and early C3rd canabae turned out to be a inhumation necropolis in the late Roman period.

Introduction

The legionary fortress (Fig. 1)

In Roman times Vienna, ancient Vindobona, was a legionary fortress with a canabae legionis and a civil settlement situated on the Danubian limes in Pannonia Superior, close to the more famous fortress at Carnuntum. It was located at a very favourable site, surrounded by the River Danube to the north and by two brooks to the west and the east. The expanse of the fortress was very soon recognised1 and a Trajanic building inscription led to the assumption that the beginning of construction of the fortress was at the end of the C1st AD and that it was built completely in stone.2 An earlier settlement was suggested by a tombstone of a miles of legio XV (Fig. 2: CIL III 4570; Neumann 1961/62: 19 Nr. 37), tilestamps3, a bronze inscription mentioning Galba4 and tombstones of auxiliary soldiers of the ala I Flavia Britannica milliaria c. R. in the time of Domitian.5 But until now firm evidence was lacking.6

Current research The following circumstances have enabled a different view on the chronology and the topography of Vindobona: • A systematic and AutoCAD supported collection and mapping of all the excavations ever made in the area of Vienna (Mosser 1998: 74–88). • Excavations in the principia of the legionary fortress from 1994 to 1995 (Harl & Süss 1995: 726-727; Gaisbauer & Mosser 2001). • The very conscientious excavations at the Judenplatz, in the inner city of Vienna, situated in the retentura of the fortress between 1995 and 1998 (Helgert 1997: 880–884; Chmelar 1998: 829–830; Chmelar & Helgert 1998: 20–26). • The scientific work of the excavations at the Michaelerplatz in the area of the canabae legionis (for example, Donat 1999a: 32–46; 1999b: 210–215; cf. Sakl-Oberthaler & Donat, this volume). • The current research on the Roman cemeteries of Vindobona (Kronberger & Mosser 2001).

Certain periods of Roman Vindobona have been determined, summarised as pre- and post-Marcomannic war periods and a late Roman period ending at the beginning of the C5th (Genser 1986: 512–515).

Collecting and mapping the results of the older excavations has led to a provisional plan of the legionary fortress of approximately the mid-C2ndAD (Fig. 1) and to the rediscovery of the porta principalis dextra, which was excavated but misinterpreted in 1843 (Mosser 2001). Comparing the results of the Judenplatz excavations with the earlier excavations at the same site has enabled the definite identification and localisation of the praetorium west of the principia and east of the barracks excavated at the Judenplatz and the almost certain dimensions of the principia (87 x 100m – total area of the fortress: 229,500m2 - Gaisbauer & Mosser 2001: 114-157). When reconstructed as a rectangular fortress, the percentage of the area of the principia in proportion to the total size of the fortress (3.8%) would make it one of the largest along

1

Polaschek 1944: 91–121, Taf.; Harl 1978a: 84 - 89; Kandler 1978b: 90100; Genser 1986: 472-501. 2 CIL III 1435922; Neumann 1961/62: 9 Nr. 5; Kandler 1978b: 91. Wooden beams at the lowest level of the stratification excavated by H. Ladenbauer-Orel have not been accepted by previous scholars, because they were not mapped on any plan. See Ladenbauer-Orel 1996: 61-65. 3 Neumann 1973: 96, 118, 122 (tilestamps of legio XV Apollinaris) 102 (tilestamps of Pansiana - Nero and Galba); Harl 1978b: 142. 4 See Vindobona–Katalog 1978: 251, M 15; Pohanka 1997: 100 Fig. bottom right. 5 CIL III 4575, 4576, 15197. Neumann 1961/62: 18-19, Nr. 33-35; Neumann 1967b: 25–27 Nr. 30–32 Taf. XXX–XXXII; Vindobona– Katalog 1978: 192–195, S 78–S 80. 6 To this discussion, see Speidel 1978: 20–23; Genser 1986: 509–512; Müller 2000: 80 Anm. 25; 94 - 95.

573

Limes XVIII

the limes (cf. Pitts & St. Joseph 1985: 86). This could be an indication that the northern part of the fortress was swept away by a flood of the Danube, probably in Roman times. An altar dedicated to Acaunos, the genius of the present day Wienfluß, on the occasion of its regulation (between 233 and 279) and 9 milestones in the nearby area of Vindobona recording the destruction of roads and bridges and their re-erection between 239 and 258, should be seen in this context.7 Tower structures at the level of the Danube serve as arguments against an original rectangular fortress belonging to the docks of the late Roman fleet base of the classis Histrica.8

the cognomen and the stipendia, but it was apparently erected much earlier.12 Other evidence for a settlement dating from Augustan to Flavian times are a fibula, pottery (Fig. 3), tile stamps and coins. They are mapped on the plan of the inner city of Vienna (Fig. 4). But there are no structures related to these artefacts. The stratigraphy within the courtyard of the principia indicates early, probably Augustan structures (Fig. 5). A wooden beam under an adobe tile at the lowest level of the stratification covered by a sterile layer, which lies under the floor of the earliest legionary fortress points to an early and previously unknown settlement. Wooden structures without related finds are known from several locations in Vindobona, especially from the area of the later canabae.13 It is debatable whether this evidence combined with early Roman artefacts are sufficient to postulate an Augustan supply base, maybe one connected with the expedition of Sentius Saturninus and Tiberius against Maroboduus in AD6 (Velleius II.109), as we have similar camps in Germany east of the Rhine, as for example at Marktbreit (v. Schnurbein 2000: 34–37). Such Augustan camps are of larger size than the later legionary fortresses and are of irregular ground-plan. Their task was the supply of the troops during their campaigns (Kühlborn 2000; v. Schnurbein 2000). The short-term character of such a supply base could be an explanation for the lack of structures and finds. The C. Atius tombstone could therefore be an indication of the presence of legio XV Apollinaris around AD6 in Vindobona and its participation in the expedition of Tiberius14 or its presence in a preClaudian base. Actually a base of this legion in late Augustan and Tiberian times is unknown (Šašel-Kos 1995), but there is definitely a permanent one in Carnuntum which was founded in Claudian times (Kandler 1974; 1978a).

The excavations in the retentura and principia resulted in the conclusion that we can no longer maintain the theory of a stone built fortress in the first period of occupation. Almost two periods of timber structures and adobe walls are evident (Chmelar & Helgert 1998: 20-21), although some of the main buildings could have been originally built in stone. The timber periods seem to start towards the end of the C1st. Stone barracks were not built before the mid-C2nd AD (Chmelar & Helgert 1998: 22-23). No interruption or break in the layers could be identified during the Marcomannic wars. Surprisingly the main period of activity in the legionary fortress falls into the C4th, maybe starting at the end of the C3rd (Chmelar & Helgert 1998: 24-25). Paved roads (Kenner 1904: 138, fig.109; 1905: 171, fig. 324) and courtyards, heating systems and stone walls, which were not customary in the previous phases, could be verified for this late Roman period.9 For example the paved courtyard of the principia (Fig. 5: Gaisbauer & Mosser 2001: 131, 141, 144, Abb. 27) is not evident in earlier periods. In addition a very great part of the finds, especially the pottery, dates from the C4th. The last Roman period extended far into the C5th with buildings, rooms and roads oriented differently to those of the earlier fortress (Chmelar & Helgert 1998: 25–26). This period has to be seen in a different context to the function of the earlier fortress.10

Summary (Fig. 6) Early artefacts within the area of Vindobona indicate some form of military occupation beginning after the failed expedition of Tiberius against Maroboduus, or in any case in the late Augustan period. The evidence of an early tombstone leads to the assumption that the legio XV was stationed there.15 From Claudian to Flavian times, while the legion built its permanent fortress at Carnuntum, it left behind a detachment in Vienna. As a consequence of the German wars of Domitian legio XIII built, c.89, a new

An early legionary fortress? A settlement at Vindobona before the beginning of the well-documented legionary fortress is evident in the tombstone of C. Atius, a soldier of legio XV Apollinaris11 (Fig. 2). Its latest date is probably AD50, because it lacks 7

See Neumann 1961/62: 15–16 Nr. 25; 1967b: 20–21 Nr. 18; Weber 1968-71: 129-141 Nr. 7–10, 12–14, 17, 19; Vindobona–Katalog 1978: 184–185, S 52. 8 See Kenner 1904: 128–132, Fig. 102–104; Ladenbauer-Orel 1984: 75 79; Soproni 1985: 81; Genser 1986: 515; Mosser 1999b: 4; Not. Dign. Occ. XXXIV 28. 9 On the structure of the barracks in the C4th, see R. Kastler this volume. 10 This context could be the installation of a fleet base and /or the settlement of the civil population in parts of the previous fortress. Soproni 1985: 80–81, 105–106. Comparable docks and bridgeheads of late Roman fleet bases at the Pannonian Danubian limes are for example Tahitótfalu Balhavár, Szigetmonostor - Horány, Veröce - Dunamezö dülö, Szentendre - Dera patak. See Soproni 1985: 69, 77–78, 99. 11 CIL III 4570; Neumann 1961/62: 19 Nr. 37; Vindobona–Katalog 1978: 192, S77 Taf. 5.

12 The more or less full and very individual formula of the inscription indicates a pre-Claudian dating. See Boppert 1992: 33 and the Iulian tombstones Nr. 20, 27, 60-61, 65-66, 77, 116, 118, 132 and Nr. 159. 13 Later canabae: excavations on Freyung, Palais Harrach (Freyung 3), Herrengasse 23, Michaelerplatz; inside the fortress: Ruprechtsplatz, Sterngasse, Judenplatz, Tuchlauben 17; See Harl & Süss 1995: 726; Ladenbauer-Orel 1996: 61–65; Süss & Bauer 1997: 871; Müller 1997: 876-881 Abb. 886; Helgert 1997: 880–882; Gaisbauer & Mosser 2001: 134-135. 14 Together with 6 other legions: XX, XVI, XXI, XIII, VIII, XIIII: see Keppie 1984: 163, 210. 15 Veterani of legio XV Apollinaris of this period are documented on stone monuments of Scarbantia: CIL III 4229, 4235, 4247: AE 1914/6-7; RIU Nr. 162, 185, 194, 197, 213.

574

Michaela Kronberger & Martin Mosser: Vindobona – legionary fortress, canabae legionis and necropolis

obtained by the more recent excavations.19 It is especially in the area of Michaelerplatz that a firmer picture of the ancient settlement can be formed, caused by the advanced stage of the evaluation of the site discovered in the early 1990s (cf. Donat & Sakl-Oberthaler, this volume).

military base. This legion was accompanied by an auxiliary unit, the ala I Britannica. The tombstones of its soldiers can be dated 93 to 101. From the Dacian wars onwards, legio XIII was relieved by legio XIV until 114 or 118.16 As legio XIV was then moved to Carnuntum, the legio X Gemina became the permanent garrison at Vindobona from Hadrian onwards (Gómez-Pantoja 2000: 188–190). The change in the military organisation in the late Roman period, from the end of the C3rd, not yet recorded exactly, must be seen in context with the development of the canabae and its later use as an inhumation necropolis. The Notitia Dignitatum (Occ. XXXIV 25 and 28) mentions Vindobona as the fortress of legio X Gemina and as a base of the classis Histrica in the province of Pannonia prima. We cannot exclude the possibility of a settlement for the remaining civil population within parts of the C4th and C5th legionary fortress, and a reduced military station in the south-western part of the former fortress (Soproni 1985: 81), where the residence of the Babenberger was in medieval times.

However, this excavation will only be briefly summarised here as the focus of this paper is on the initial results of the processing of the data from the earlier excavations. They are based on the new interpretation of the documentation which accumulated in particular during the great building boom at the turn of the C19th and during the rebuilding of Vienna after the destructions of WWII. The settlement structures and cemeteries recorded on such plans have been mapped on a digital map of Vienna by means of CAD software To identify no longer visible street alignments today, this map can be overlaid with older card plans – again rendered into a digital form. After completion of this task it should be possible to form at least an idea of the structural substance of the canabae legionis. The latest results of research in the area of the legionary fortress of Vindobona, where the same process has been chosen, shows that the results are promising.

The canabae legionis of Vindobona (Fig. 7) Although excavations in the area of the canabae legionis of the fortress of Vindobona have been carried out since the early C19th, little is known about its extent, its structures or its history. The most important municipal facilities, such as the forum, temple, theatre and thermal springs have remained virtually unknown until now. Only the last phase of the use of the area in the vicinity of the legionary fortress in the Roman period, could be appreciated relatively well in recent times, these are the extensive cemeteries which were laid out from the late C3rd in the area of the earlier canabae legionis. Since very little stone material came to light in the excavations and only a few wall lines could be made out,17 discussion in the late C19th and early C20th assumed that no settlement existed and that the burial districts had been used as such throughout the entire Roman period.

To work out a chronological framework for the settlement, the material finds which had come through the confusion of the time has to be evaluated. To be able to delimit the occupation period of the canabae legionis, the stratigraphic horizon which is latest and best documented at the same time had to be established.20 This turned out to be the inhumation necropolis that was laid out in the area of the earlier canabae legionis sometime in the second half of the C3rd.21 From the composition and dating of the graves and their contents, its use can be placed between the late C3rd and the C4th. Graves without burial goods suggest a far longer use, into the C5th. If one removes this culture layer, one gets at least for those parts of the canabae legionis a chronological framework, of which enough finds for the evaluation have been preserved.

Timber building phases, as they have been excavated in the area of the canabae legionis at the Michaelerplatz or on the Freyung as well as in the legionary fortress18, were not perceived by earlier researchers. Convinced of phenomena assigned to the daily life, such as kilns which therefore served to produce grave ceramic, they were squeezed into this scheme (Kenner 1900: 60f.).

Expanse and chronology of the settlement (Fig. 7) The extent of the canabae legionis can be stated only roughly in the area of the 1st Viennese district. Large areas were destroyed in the middle ages by the construction of the town’s fortifications.22 Roman finds were made in the course of time when places were redesigned or houses rebuilt. As a result of these activities, it is particularly difficult to obtain a picture of the ancient settlement in

Since no pioneering discoveries which could have been interpreted as a municipal infrastructure of the canabae area were made within the following decades, the research concentrated on the better comprehensible legionary fortress and the later necropolis. The first insights into parts of the districts outside the legionary fortress were

19 1987 excavations Freyung; 1990–1992 excavations Michaelerplatz; 1992 excavations Freyung (Palais Harrach), 1995–1996: Herrengasse 23 (Palais Porcia), 1990: Renngasse. The results of the excavations are just evaluated. 20 These have just been evaluated as part of a project on the town archaeology of Vienna by M. Kronberger & M. Mosser. The first part, the area around the Neuer Markt is complete: cf.Kronberger & Mosser 2001. 21 Valuable groundwork has been done by H. Zabehlicky & O. Harl, for which we are grateful. 22 Historischer Atlas von Wien. Wachstumsphasen. 1. Lieferung 1981, Blatt 1. 2. The first defence system was erected by the Babenberger around 1200, the construction of the second started in 1530.

16 For the problems of dating the movements see Gabler & Lörincz 1977; Lörincz 1981; Fitz 1986; Strobel 1988; Fitz 1993: 372. 17 See n. 13. 18 See n. 13.

575

Limes XVIII

those areas at the centre of Vienna where structures of the baroque period are still preserved. However with mapping of the known Roman locations and stamped tiles23 one can show that the canabae legionis grew, starting in the Trajanic era to the mid-C3rd, to a size of at least 50ha (Fig. 7). If the areas interpreted as uncertain are added, the fortress suburb might have been far larger.

The Necropolis (Fig. 8) Roman graves from different phases have been found almost everywhere around the legionary fortress. The earliest piece of evidence for a burial in Vindobona is the stele fragment to Caius Atius, a soldier of the legio XV Apollinaris (Fig. 2). It can be dated on epigraphic criteria to the time of the emperor Tiberius.27 The stone was found, reused, in the area Stephansplatz/Brandstätte (Fig. 8A). Little can be said about the location or the exact expanse of this earliest cemetery. There are references to settlement activity in the area of Vindobona for the early C1st, but its spatial delimitation is currently not comprehensible (Figs. 3 and 4). There are more concrete references to a necropolis when it came to the stationing of an auxiliary troop in the second half or the end of the C1st (ala I Flavia Britannica milliaria c.R.), and probably under the emperor Domitian of a legion (legio XIII Gemina). Three gravestones of auxiliary soldiers, dug up in 1559 and 1901 in the area of the Stallburg (Fig. 8B) and the most recent results from the excavation at the Albertina (Fig. 8C) make clear that this Flavian-Trajanic cemetery was laid out along the limes road (Sakl-Oberthaler 1999: 110–127) which led eastwards, in the direction of Carnuntum and westwards, in the direction of Klosterneuburg. On this eastern radial road a necropolis was established from approximately the C1st down to the C3rd in the area of today's Karlsplatz (Fig. 8D: Kenner 1897: 118, 148; Kenner 1900: 74–82), like the western one in the area Sigmund-Freud-Park (Fig. 8E) and in the Währingerstraße.28

In order to get guiding values for the expanse and development of the settlement in the course of the time, canabae areas which showed different distances to the legionary fortress and of which sufficient dated finds have remained, were selected. From the evaluation of the mould decorated terra sigillata of these areas, certain key differences stand out in different districts.24 Those areas which are closest to the legionary fortress were, as expected, occupied in the first settlement phase (see Fig. 7.1). They show the highest percentage of products of the Flavian period, which correlates well with the chronology of the legionary fortress built under Domitian. The curve of the mould decorated terra sigillata finds rises steadily in the Trajanic-Hadrianic period. The pattern is seen in other adjacent areas, as in todays Neuer Markt (see Fig. 7.3) and the Singerstraße/Liliengasse near the Stephansdom (see Fig. 7.2). The finds from the excavations in the Johannesgasse25 which date from the late C2nd onwards, prove therefore a further opening up of the settlement area at this time (see Fig. 7.4). A break in settlement at the turn of the C2nd into the C3rd in the area surrounding the fortress seems to stand out. It was possibly given up for defensive reasons in the immediate vicinity of the legionary fortress. In connection with this, the results of the evaluation of the excavations in the area of the Freyung and the Renngasse are eagerly anticipated.26 From the finds a further settlement is certain in the other districts already mentioned down to the middle of the C3rd. The vicinity of the Neuer Markt especially should be noted here where this sequence of development has been also confirmed epigraphically (Kronberger & Mosser 2001). The find of an altar dedicated to the goddess Fortuna, which is dated by the legion’s epithet SEVERIANA to the time of the emperor Severus Alexander (222–235), indicates municipal life in this area is proved (Kenner 1900: 61–62; Neumann 1961/62: 13 Nr. 19; Pohanka 1997: 102). Although it has to be assumed that a large part of the population had already established itself within the fortress walls, certain sites prove the on-going existence of canabae life in the late C4th, especially workshop areas, as the latest results of the analysis of the excavation at Michaelerplatz have shown (Fig. 7.5).

In the late C1st the canabae legionis started to extend to areas next to the legionary fortress. The settlement should have extended to at least the height of the Neuer Markt to the limes road until the C3rd (Fig. 8F). Settlement pits and wall structures were also found in the Maysederstraße (Fig. 8G) which is only approximately 100m off the limes road in the Augustinerstraße (Fig. 8H).29 From the results of last year’s excavation at the Albertina, it can be proved that the area south of the limes road was used exclusively as a necropolis in the period from the late C1st to the late Roman period (Fig. 8C: cf. Huber 2001). In the second half of the C3rd large parts of the canabae legionis were left and levelled. In their place inhumation cemeteries developed at the latest by the end of the C3rd, with stone or tile chests and plain inhumations (cf. Fig. 8).

23 Work on the Vindobona tiles by B. Lörincz (Die Ziegel von Vindobona - in print); digital mapping by M. Mosser. 24 The publication of Weber-Hiden (1996) was used as a basis for the evaluation of the terra sigillata. Definite statements can be made only after the evaluation of all find genres. That evaluation is being carried out now. 25 Excavation 1907: FA-Rz, I., H–K, Johannesgaße 1907; excavation 1912: FA-Rz, I., H–K, Johannesgaße 1912; excavation 1913: FA-Rz, I., H–K, Johannesgaße 1913. 26 Evaluation of the excavations Freyung 1987 and Palais Harrach 1992– 94: M. Kronberger, of the excavation of Renngasse : W. Börner.

27

See above n. 12. Sigmund Freud Park: Kenner 1897: 103–104; Kenner 1905: 187–196; A. Neumann FÖ 5, 1946–1950, 134, approximately from the C3rd to the C4th. Währinger Straße: Kenner 1911: 120a–123a; Neumann 1967a: 71– 72, perhaps from the C1st to late Roman times. 29 Sakl-Oberthaler 1999: 113 Abb. 1; FA-Rz, A–B, Augustinerstraße, 1935; R. Wadler FÖ 2, 1934/37, 104. 28

576

Michaela Kronberger & Martin Mosser: Vindobona – legionary fortress, canabae legionis and necropolis

Kronberger M. & Mosser M. 2001 Das spātrōmische Grāberfeld von Vindobona Neuen Markt und Umgebung. Fundort Wien 4: 158-221. Kühlborn J.-S. 2000 Schlagkraft. Die Feldzüge unter Augustus und Tiberius in Nordwestdeutschland. In Die Römer zwischen Alpen und Nordmeer. Ausstellungskatalog Rosenheim (Mainz): 27-33. Ladenbauer-Orel H. 1984 Mittelalterliche Quellen zur römischen Lagermauer von Vindobona. Wiener Geschichtsblätter 39/2: 75-79. Ladenbauer-Orel H. 1996 Erfolge und Probleme auf drei Baustellen in Vindobona. In Römerzeit–Mittelalter. (Festschrift Herma Stiglitz, Wien): 61–65. Lörincz B. 1981 Some remarks on the history of the Pannonian legions in the late first and early second centuries A.D. Alba Regia 19: 285 - 288. Mosser M. 1998 Das Legionslager Vindobona – EDV gestützte Erfassung alter und neuer Grabungen. Fundort Wien 1: 74-88. Mosser M. 1999a Befunde im Legionslager Vindobona. Teil I: Altgrabungen am Judenplatz und Umgebung. Fundort Wien 2: 48-85. Mosser M. 1999b Innere Stadt – Rabensteig 3. Fundort Wien 2: 4. Mosser M. 2001 Die Porta Principalis Dextra im Legionslager von Vindobona. In Akten zum 8. Österreichischen Archäologentag 1999 (Wien): in print. Müller M. 1997 Herrengasse. Fundberichte aus Österreich 36: 876–881. Müller M. 2000 Römische und neuzeitliche Funde aus Wien 3, Eslarngasse 20. Zur Befestigung der Zivilstadt von Vindobona. Fundort Wien 3: 76-102. Neumann A. 1961/62 Inschriften aus Vindobona. Jahrbuch des Vereines für Geschichte der Stadt Wien 17/18: 7-29. Neumann A. 1967a Forschungen in Vindobona 1948–1967. I. Teil: Lager und Lagerterritorium. (Der römische Limes in Österreich 23: Wien). Neumann A. 1967b Die Skulpturen des Stadtgebietes von Vindobona. (CSIR Österreich I, 1: Wien). Neumann A. 1973 Ziegel aus Vindobona. (Der römische Limes in Österreich 27: Wien). Pitts L.F & St. Joseph J. K. 1985 Inchtuthil - The Roman legionary fortress. Britannia Monograph Series 6. (London). Pohanka R. 1997 Das römische Wien. (Wien). Polaschek E. 1944 Die Kunst der Römerzeit in Wien. In R. K. Donin (ed.) Geschichte der bildenden Kunst in Wien I: Von der Urzeit bis zur Romantik. (Wien): 91-121. Sakl-Oberthaler S. 1999 Untersuchungen zur Limesstraße in Wien–Simmering. Fundort Wien 2: 110–127. Šašel-Kos M. 1995 The 15th Legion at Emona – some thoughts. Zeitschrift für Papyrologie und Epigraphik 109: 227–244. von Schnurbein S. 2000 Die augusteischen Stützpunkte in Mainfranken und Hessen. In Die Römer zwischen Alpen und Nordmeer. Ausstellungskatalog Rosenheim (Mainz): 34–37. Soproni S. 1985 Die letzten Jahrzehnte des pannonischen Limes. (Münchner Beiträge zur Vor- und Frühgeschichte 38: München). Speidel M. 1978 Die römischen Reitertruppen und die Wiener Ala Britannica. In Vindobona – die Römer im Wiener Raum. (Wien): 20-23. Strobel K. 1988 Zur Dislozierung der römischen Legionen in Pannonien zwischen 89 und 118 n. Chr. Tyche 3: 193222.

Bibliography CIL - Corpus Inscriptionum Latinarum CSIR - Corpus Signorum Imperii Romani RIU - Die römischen Inschriften Ungarns Boppert W. 1992 Militärische Grabdenkmäler aus Mainz und Umgebung. (CSIR Deutschland II, 5: Mainz). Chmelar W. 1998 Judenplatz. Fundberichte aus Österreich 37: 829-830. Chmelar W. & Helgert H. 1998 Die römischen Kasernen unter dem Judenplatz. Fundort Wien 1: 20-26. Donat P. 1999a Feinkeramik aus Vindobona – Hinweise auf eine lokale Produktion? Fundort Wien 2: 32-46. Donat P. 1999b Zur Herkunft der Terra Sigillata von der Ausgrabung Michaelerplatz. Fundort Wien 2: 210-215. Fitz J. 1986 Geschichtliche Probleme des Forums von Gorsium. Oikumene 5: 329-369. Fitz J. 1993 Die Verwaltung Pannoniens in der Römerzeit II. (Budapest). Gabler D. & Lörincz B. 1977 Some remarks on the history of the Danubian limes of the first and second century. Archaeologiai Értesitö 104: 145-175. Gaisbauer I. & Mosser M. 2001 Befunde im Legionslager Vindobona. Teil II: Altgrabungen im Bereich der principia. Fundort Wien 4: 114-157. Genser K. 1986 Der österreichische Donaulimes in der Römerzeit. (Der römische Limes in Österreich 33: Wien). Gómez-Pantoja J. 2000 Legio X Gemina. In Y. Le Bohec (ed.) Les Légions de Rome sous le Haut-Empire. Actes du Congrés de Lyon 17–19 septembre 1998. (Collection du Centre d´Études Romaines et Gallo-Romaines, Nouvelle série, Nr. 20, Lyon): 169–190. Harl O. 1978a Erforschung, Name, Lage, Befestigungsanlagen. In Vindobona – die Römer im Wiener Raum. (Wien): 8489. Harl O. 1978b Die römischen Ziegel. In Vindobona – die Römer im Wiener Raum. (Wien): 141-143. Harl O. & Süss K. 1995 Tuchlauben. Fundberichte aus Österreich 34: 726-727. Helgert H. 1997 Judenplatz. Fundberichte aus Österreich 36: 880-884. Huber E.H. 2001 Albertina. Fundort Wien 4: 258-259. Kandler M. 1974 Die Ausgrabungen im Legionslager von Carnuntum 1968–1973. Anzeiger der Österreichischen Akademie der Wissenschaften 111: 27-40. Kandler M. 1978a Vorläufiger Bericht über die Grabungen im Legionslager Carnuntum in den Jahren 1974–1977. Anzeiger der Österreichischen Akademie der Wissenschaften 115: 335–351. Kandler M. 1978b Zu den Innenbauten des Legionslagers. In Vindobona – die Römer im Wiener Raum. (Wien): 90100. Kenner F. 1897 Die archäologischen Funde aus römischer Zeit. In Geschichte der Stadt Wien I. (Wien): 42–159. Kenner F. 1900 Bericht über römische Funde in Wien in den Jahren 1896–1900. (Wien). Kenner F. 1904 Römische Funde in Wien aus den Jahren 1901 bis 1903. Jahrbuch der k. k. Zentral-Kommission N. F. 2: 103-170. Kenner F. 1905 Römische Funde in Wien aus den Jahren 1904 und 1905. Jahrbuch der k. k. Zentral-Kommission N. F. 3: 137-230. Kenner F. 1911 Römische Funde in Wien 1908–1910. Jahrbuch für Altertumskunde 5: 107–162. Keppie L. 1984 The making of the Roman army. (London).

577

Limes XVIII

Süss K. & Bauer W. 1997 Freyung. Fundberichte aus Österreich 36: 870-876. Vindobona–Katalog 1978 Vindobona – die Römer im Wiener Raum. (Wien). Weber E. 1968–71 Die römischen Meilensteine aus dem österreichischen Pannonien. Jahreshefte des Österreichischen Archäologischen Instituts 49: 121-145. Weber-Hiden I. 1996 Die reliefverzierte Terrasigillata aus Vindobona (Wiener Archäologische Studien 1: Wien).

578

Michaela Kronberger & Martin Mosser: Vindobona – legionary fortress, canabae legionis and necropolis

Fig.1. Vindobona - legionary fortress, approximately mid-2nd century AD (dig.: M. Kronberger & M. Mosser)

Fig. 2. Tombstone of C. Atius, miles legionis XV Apollinaris (Photo: Kunsthistorisches Museum Wien, Archivphoto – Nr. III 28.669, Inv. Nr. ASIII 101).

579

Limes XVIII

Fig. 3. Pre-Flavian pottery from Vindobona. 1 and 7 – Spiegelgasse/Plankengasse; 2 – Michaelerplatz; 3 – Sterngasse; 4-6 – Freyung; 8 – Wollzeile. Scale 1: 2 (Dig.: M. Kronberger).

580

Michaela Kronberger & Martin Mosser: Vindobona – legionary fortress, canabae legionis and necropolis

Fig. 4. Vindobona – pre-Flavian artefacts (dig.: M. Kronberger & M. Mosser).

Fig. 5. Profile: courtyard within the principia of the legionary fortress in Vinodobona (dig: I. Gaisbauer & M. Kronberger)

581

Fig. 6. Period table of the legionary fortress and canabae of Vindobona (M. Kronberger & M. Mosser).

Limes XVIII

582

Fig. 7. The extent and chronology of the canabae legionis of Vindobona. 1 – Freyung/Renngasse; 2 – Singerstraße/Liliengasse; 3 – area of Neuer Markt; 4 – Johannesgasse; 5 – Michaelerplatz; 6 – Fleischmarkt/Laurenzerberg. (M. Kronberger & M. Mosser)

Michaela Kronberger & Martin Mosser: Vindobona – legionary fortress, canabae legionis and necropolis

583

Fig. 8. Roman cemeteries of Vindobona within the 1st district of Vienna (M. Kronberger & M. Mosser).

Limes XVIII

584

Vindobona – the Roman civil settlement at Vienna (poster-presentation) Ingrid Mader north-west of the Roman settlement.3 Of course this does not provide proof for the existence of a previous Celtic settlement, but single farms certainly must have existed. Recently a La Tène period house, filled with pottery (Grubenhaus/La Tène D) has been found.4 But all these finds still lack the structural connection indicating a settlement.

The topography of Vindobona Within the limits of the modern city of Vienna people have settled since Neolithic times. The sites geological terraces have always been regarded as suitable for settlements. The Romans erected a legionary fortress on one of the terraces (Stadtterrasse) in what is today’s 1st district. One still can recognise its elevated position in today’s cityscape.

The Roman civil-settlement Vindobona The municipium Vindobonense Finds made in the C18th and C19th in the 3rd district of Vienna (Fig. 2) such as the fragment of a marble statue, a finger and a foot fragment cast in bronze (Neumann 1967: Nr. 5-7), led researchers to assume that the Roman town was situated around the ‘Rennweg’ (a street later identified as the former limes street).5 A milestone found in the C16th formerly marked the eastern border of the settlement.6 This stone dated back to the mid-C3rd AD, therefore informing us of the probable distance to the legionary fortress at this time (Harl 1979: 152). Further proof for the eastern boundaries being situated here in that period are traces of the fossae fastigatae marking the south-east of the settlement that have been excavated there recently.7 From the mapping of the house structures one can clearly see that the town developed aside the east–west leading limes street (Fig. 2).8

Together with the military fortress and the canabae legionis, the construction of a civil settlement began. The name Vindobona is regarded as Celtic in origin.1 The earliest reference to it is in the Greek writer Ptolemaeus (II.14.3.). Later, the Tabula Peutingeriana shows the name Vindobona marked in the vicinity of two houses. Other instances of the name are given by many tile-stamps made by the civilian manufacturer M. Antonius Tiberianus. Sometimes he stamped in short: VINDOB. (Neumann 1973: Taf. LXIX). An inscription found in the C16th mentions Vindobona as a municipium (Harl 1979: 140). The original stone itself is not preserved, but a copy of the text made by the Viennese humanist Wolfgang Lazius reporting this, still exists. According to Lazius, the stone was dedicated by C. Marcius Marcianus who was a decurio of the municipium of Vindobona and a praefectus of the cohors fabianae (CIL III/I 4557). The reading coh(ors) fabi(anae) was later on corrected by T. Mommsen into co[ll(egii)] fabr(um).

The Roman settlement-first phase (Fig. 2) The discovery of two fossae fastigatae in the beginning of the C19th gives the impression that the settlement was once surrounded by a system of fortifications (Kenner 1911: 150a). A. Neumann (1980: 15) interpreted these ditches as military structures. He explained his theory with the earlier discovery of gravestones mentioning the ala I milliaria Britannica.9 Although these stones were found a great distance from the remains of the civil town, he thought that there must be a connection with the ditches in question, and concluded that an auxiliary fortress in the 3rd district must have existed. Only recently, another part of the same ditches has been found.10 That makes us believe rather in a Roman fortified town than a military fortress (Kenner 1911: 150aq; Piso 1991: 174).

In the C17th century, another inscription was found this time outside of Vienna that also mentioned the cohors fabianae and the municipium Vindobonae (Hormayr 1823: Bd I. 93). But both inscriptions are very questionable, because of the uncertain identity of the cohors fabianae. This inscription is omitted from most scientific literature.2 Up until today, these two inscriptions were the only evidence that Vindobona had been a municipium. No other inscriptions are preserved recording the municipal status of Vindobona. Is there a pre-Roman settlement to the Roman town ? (Fig. 1)

3

Der Ofen wurde 1926 gefunden, konnte aber nicht vollständig freigelegt werden. Fundprotokolle 1926; Fundberichte aus Österreich 1930–34: 77. Auswertung der Grabungsbefunde und –Funde der Grabung Klimschgasse/Rudolfstiftung obliegt E. Pichler. 5 Kubitschek 1893: 34. N. J. Freiherr von Jacquin, Hortus Botanicus (1770–1776); Kenner 1897: 121ff. 6 Gefunden 1588, in einem Weingarten unweit vom Krankenhaus St. Marx: Kenner 1898; Weber 1968: 132; CIL III 4647. 7 Bericht über die Grabungen am Aspangbahnhof: Huber 1999. 8 Zum Verlauf der römischen Straßen in Vindobona siehe: SaklOberthaler 1999. 9 Zur Geschichte und den Standorten der ala I Britannica: Kennedy 1977; Lörincz 1979. 10 Müller 1997; 1998; 2000.

Mapping of the finds from the La Tène period suggest the possibility of a previous settlement lying in the northern region of the later Roman settlement (Fig. 1). Apart from a few isolated finds, a pottery kiln was found in the 1930s,

4

1

Überlegungen zu seiner Etymologie zusammenfassend bei Genser 1986. Zuletzt hat sich J. Haberl bemüht, den Lazius-Stein und die Lanzendorfer Inschrift mit der Erwähnung der cohors Fabiana als Beweis für die Theorie, Vindobona sei Favianis, heranzuziehen; Haberl 1976. Dagegen Harl 1979.

2

585

Limes XVIII

In the years 1989-1990 traces of a handicraft-district (Fig. 2) were discovered in which at least six periods could be distinguished. Together with the first wooden remains, were found some terra sigillata Padana (Gabler 1996: 389f.), while the metal objects have been dated to the 2nd half of the C1st.11 This evidence implies that people had already settled here between the end of the C1st AD and the first half of the C2nd.

Bibliography FÖ 1930–34 = Fundberichte aus Österreich: 77. FP 1926 = Fundprotokolle des Historischen Museums der Stadt Wien 2b. Fwien = Fundort Wien Gabler D. 1996 Die frühe Sigillata aus Wien 3, Rennweg 44: 389– 401. In I. Weber-Hiden Die reliefverzierte Terrasigillata aus Vindobona. Teil 1. Legionslager und Canabae. WAS 1 (Wien). Genser K. 1986 Der Österreichische Donaulimes in der Römerzeit. Der römische Limes in Österreich 33: 468– 471. Haberl J. 1976 Favianis, Vindobona und Wien. Eine archäologisch-historische Illustration zur Vita Sancti Severini des Eugippius. Harl O. 1979 Vindobona. Das römische Wien. Wiener Geschichtsbücher Band 21/22. Harl O. 1979 (Review of Haberl 1976). Gnomon 51: 578-581. Hormayr, Freiherr von J. (ed.) 1823 Wiens Geschichte und seine Denkwürdigkeiten. Bd I. Huber E.H. 1999 Fundchronik Römerzeit: Wien 3, Aspangbahnhof. FWien 2: 165-166. Kennedy D.L. 1977 The ala I and cohors I Britannica. Britannia VIII: 249-255. Kenner F.1897 Die archäologischen Funde aus römischer Zeit. In Geschichte der Stadt Wien I (Wien): 42-159. Kenner F. 1898 Neueste Fundergebnisse in Wien. Mittheilungen der K.K. Centralkommission Neue Folge 24: 190. Kenner F. 1903 Römische Funde aus Wien (1902). Mittheilungen der K.K. Centralkommission. Folge III, Band 2: 45-48. Kenner F. 1905 Römische Funde aus den Jahren 1904 und 1905. Jahrbuch der K.K. Centralkommission 3. Neue Folge, Band 3: 138-227. Kenner F.1911 Römische Funde in Wien 1908–1910. Jahrbuch für Altertumskunde V: 107-162. Kubitschek W. 1893 Vindobona. Xenia Austriaca. Festschrift der österreichischen Mittelschulen zur 42. Versammlung deutscher Philologen und Schulmänner in Wien 34. Lörincz B. 1979 Beitrage zur Geschichte der Ala I Flavia Britannica Milliaria C.R. Alba Regia 17: 357–359. Müller M. 1997 Herrengasse. Fundberichte des Österreich 36: 876–881. Müller M. 1998 Fundchronik Römerzeit: Wien 3, Eslarngasse 20. FWien 1: 169–171. Müller M. 2000 Römische und neuzeitliche Funde aus Wien 3, Eslarngasse 20. Zur Befestigung der Zivilstadt von Vindobona. FWien 3: 76-104. Neumann A. 1967 CSIR = Corpus Signorum Imperii Romani. Corpus der Skulpturen der Römischen Welt. Österreich (I). Die Skulpturen des Stadtgebietes von Vindobona. (Wien). Neumann A. 1973 Ziegel aus Vindobona. Der römische Limes in Österreich 27. Neumann A. 1980 Vindobona – Die römische Vergangenheit Wiens. (Wien). Piso I. 1991 Municipium Vindobonense. Tyche 6: 171–177. Sakl-Oberthaler S. 1999 Untersucheungen zur Limesstraße in Wien-Simmering. FWien 2: 110-127. Weber E. 1968 Die römischen Meilensteine aus dem östereichischen Pannonien. Jahrbuch des Österreichischen Archäologischen Instituts 49: 71.

Further indications for the first phase we might have found are some graves and fossae filled with grave contents. Ocassionally the fossae cut some of the graves under them. I think the graves may mark this presumed earlier phase (Kenner 1905: 214ff). Only a few finds though can be dated to the C1st AD.12 Another cemetery found near the southern boundary was in use from the beginning until the end of the settlement (Kenner 1905: 220; 1903: 45ff. Von 0.7m bis 2.70m). The Roman settlement Vindobona - second phase (Fig. 3) The later town extended westwards and eastwards (see the “fossae fastigatae”). The most important finds marking this development are a Roman bath, the milestone mentioned above and the recently found ditches in the south-east sector. In general, pottery found all over the town dates back into the C2nd and C3rd. Completely lacking are all traces of temples and of the forum of this town. Only very sparse finds have given us the clue though, that we must certainly expect to find them some day. Traces of late Roman period housing have not been found either, although coins and pottery from this period have occasionally been recovered. Summary We are not yet sure if Vindobona was a municipium or not. Besides the inscriptions mentioned above, which were lost in the meantime, we do not have any other proof of its existence. The name Vindobona has a Celtic origin, but this does not automatically mean that there has been a Celtic settlement here before the Roman one. Only some scattered Celtic objects and remains have been found within (and north of) the later Roman settlement. The Roman town developed from the C1st to the C3rd. Especially since the C2nd, the settlement extended east and westwards. As mentioned above, some statue fragments speak of a forum existing here too. Future finds should help to clarify our picture of the history of this town.

11

Datierung von H. Sedlmayer Stadtarchäologie Wien. Kenner 1905: 151b: “Die Brandmulde wurde später, aber noch in römischer Zeit, mit einer Erdschicht überdeckt, welche Bau- und Gräberschutt enthielt.“ 12

586

Ingrid Mader: Vindobona – the Roman civil settlement at Vienna (poster-presentation)

Fig. 1. Third District of Vienna – Latène – time period

587

Fig. 2. Third District of Vienna (S) – Possible Extension of the Early Phase of the Roman Town.

Limes XVIII

588

Fig. 3. Third District of Vienna (S) – Possible Extension of the Later Phase of the Roman Town.

Ingrid Mader: Vindobona – the Roman civil settlement at Vienna (poster-presentation)

589

Limes XVIII

590

Die Ausgrabungen am Michaelerplatz – ein Bereich der canabae legionis von Vindobona Sylvia Sakl-Oberthaler & Patrizia Donat et al (Mai–Juli 1990) wurden großflächige römische Gebäudestrukturen im Süden des Areals (Abb. 1/8) freigelegt, deren Chronologie und Funktion derzeit noch unklar sind. Einen Hinweis auf die wirtschaftlichen Grundlagen dieses Abschnitts der Vindobonenser canabae geben spezifische Belege der Buntmetallverarbeitung. In diesem südlichen Grabungsareal läßt sich ein dichtes Vorkommen von Halbfabrikaten, Gußresten und Fertigprodukten (Altmetall) aus Buntmetall beobachten.5 Westlich davon zeigten sich außerdem Teile eines weiteren Gebäudes (Abb. 1/3), das mit Estrichböden und Wandmalerei ausgestattet war. Zusätzlich kam im südlichen Areal der Rest eines Wasserbeckens zum Vorschein (Abb. 1/3). Über dieser kaiserzeitlichen Bebauungsphase befanden sich ein umfangreicher Komplex von Münzen des 4. Jh. n. Chr. sowie die Reste einer spätantiken Holzkonstruktion (Abb. 1/9). Alle diese Befunde waren durch die renaissance- und barockzeitliche Verbauung des Michaelerplatzes (Paradeisgartlmauer Abb. 1/c und Hofburgtheater Abb. 1/d) massiv gestört. Es existieren Hinweise auf mehrfache Planierungen sowohl in römischer als auch in nachantiker Zeit.6 Dieser Umstand erschwert die exakte Deutung der Befunde. Die zweite Grabungsphase (Oktober 1990-Oktober 1991) konzentrierte sich vor allem auf den Teil nördlich der Paradeisgartlmauer. Erwähnenswert ist hier vor allem ein Werkstättenareal. Im zentralen Bereich der Grabung gelang der direkte Nachweis der vor Ort betriebenen Verarbeitung von Eisen durch eine überkuppelte Esse (Abb. 1/7). Die Mehrphasigkeit der Werkstatt indizieren mehrere, unter der jüngsten Essenkonstruktion befindliche Straten mit abgelagerten Eisenschlacken und Holzkohle.

1. Der archäologische Befund (P. Donat, S. SaklOberthaler & H. Sedlmayer) 1.1: Anlaß, Durchführung Anlaß für die Ausgrabungen am Michaelerplatz war seine geplante Umgestaltung als Fußgängerzone. Es fanden in der Folge zwei Grabungskampagnen der Stadtarchäologie Wien statt, die erste zwischen Mai und Juli 1990, eine weitere von September 1990 bis zum Jahresende 1991. Die erste Grabungskampagne mußte unter großem Zeitdruck durchgeführt werden. Nach dem Abschluß der Arbeiten übernahm der Architekt Hans Hollein die Platzgestaltung, wobei ein Teil der Ausgrabungen konserviert werden konnte (Abb. 2). 1997 begann die Gesamtbearbeitung der römischen Funde und Befunde der Ausgrabungskampagnen 1990/1991 durch ein neu zusammengestelltes Team. Nach einer ersten Sichtung der vorhandenen Dokumentation wurden zunächst die Grabungspläne digitalisiert.1 Zugleich begann die Bearbeitung des Fundmaterials.2 Zur Zeit erfolgt auch die Interpretation des archäologischen Befundes. An dieser Stelle soll der gegenwärtige Stand der Bearbeitung vorgestellt werden. Da eine durchgehende Stratigraphie nicht zu erwarten ist, wird die Materialbearbeitung nach typologischen Gesichtspunkten durchgeführt. Geplant ist auch eine Aufarbeitung der mittelalterlichen und neuzeitlichen Funde und Befunde. 1.2: Stellung des Michaelerplatzes in römischer Zeit

Dazu kam ein aufwendig ausgestatteter römischer Gebäuderest (Abb. 1/3). Dieser wurde von den Mauern eines in der Neuzeit angelegten Kellersystems unter einer schon im Mittelalter entstandenen Häuserzeile (Abb. 1/a) durchschnitten. Die beiden erhaltenen gedeckten Räume (Abb. 1/3) waren mit einem mehrphasigen Heizsystem (Fußboden- und Wandheizung) und Wandmalerei ausgestattet. Daran grenzte im Westen ein mit Steinplatten gepflasterter Bereich (Abb. 1/5). Das gesamte Gebäude wurde mehrfach verändert. Über das gesamte Graben ß Gelände verteilt befanden sich außerdem verschiedene Gruben mit mehreren Verfüllungen, einige davon entstanden mit Sicherheit vor der Errichtung des Gebäudes 3.

Der 1990/1991 ergrabene Gebäudekomplex befand sich am Kreuzungspunkt zweier wichtiger Straßen, nämlich der Limesstraße (Abb. 1/1) und der zur porta decumana des Legionslagers führenden Abzweigung (Abb. 1/2). Eine massive Basis aus behauenen Steinquadern markierte diese Kreuzung (Abb. 1/10). Beide Straßen waren geschottert. Die Limesstraße verlief südlich am Legionslager von Vindobona vorbei, querte die canabae legionis3 am Michaelerplatz, durchlief anschließend die Zivilstadt (cf. I. Mader, this volume) um dann weiter über die Kastelle Ala Nova (Schwechat) und Aequinoctium (Fischamend) Carnuntum zu erreichen.4 Die Limesstraße kreuzte sich an dieser Stelle mit der sog. „pannonischen Transversale“ in Richtung Scarbantia (Sopron), wo sie in die Limesstraße einmündete. Während der ersten Grabungskampagne

Bislang unklar bleibt die Funktion und die Zeitstellung der parallel verlaufenden Mauern (Abb. 1/11) unmittelbar südlich der neuzeitlichen Kelleranlagen. Nur die südlichste

1

W. Börner, Stadtarchäologie Wien. Siehe Einzelkapitel. 3 Börner 1997: 247-249; Harl 2001: 25-47. Zu den canabae allgemein: siehe Beitrag Mosser & Kronberger 3-7, fig. 7. 4 Sakl-Oberthaler 1999: 110-127. 2

5

Sedlmayer 1998: 22. Als Hinweis darauf kann die Streuung anpassender Scherben aus den obersten Schichten über das gesamte Grabungsareal gelten.

6

591

Limes XVIII

von ihnen gehört zweifelsfrei zum römischen Befund.

unter der glatten und barbotineverzierten Ware bilden.12

Bei den Ausgrabungen kamen als mittelalterliche und neuzeitliche Befunde die massiven Fundamente der Umfassungsmauern des Paradiesgartls7 (Abb. 1/c) sowie jene des Hofburgtheaters8 (Abb. 1/d) zum Vorschein. Während der zweiten Kampagne legte man ein mehrstöckiges System von Kelleranlagen (Abb. 1/a) frei, die zu einer Häuserzeile gehörten, die bereits im Mittelalter entstand und den Platz beherrschte, bis sie im 19. Jh. geschliffen wurde.9

Die Hauptmenge der Terra sigillata stammt aus trajanischhadrianischer Zeit bis ca.180 n. Chr. Im zweiten und dritten Drittel des 2. Jh. n. Chr. herrscht die mittelgallische reliefverzierte Terra sigillata des Cinnamus vor. Weiters sind Mapillo, Criciro und die spätesten mittelgallischen Töpfer13 zu dieser Zeit belegt. Außerdem laufen etwa um die Mitte des 2. Jh. n. Chr. die südgallischen und norditalischen Waren aus und werden durch die frühen Produkte aus Rheinzabern ersetzt. Bei der reliefverzierten Terra sigillata ist die Bernhard–Gruppe Ia mit den Töpfern Reginus I, Januarius I, Firmus I und Cobnertus I, II, III anzutreffen.14 Die glatte und barbotineverzierte Terra sigillata ist mit der Form Drag. 18/31 am häufigsten vertreten.

2: Das Fundmaterial Die folgenden Kapitel geben einen kurzen Überblick über den derzeitigen Bearbeitungsstand. Bevorzugt wurde hier auf jene Fundgruppen eingegangen, die die präzisesten Datierungsansätze bieten. Die Klassifikation von Scherbentypen ist beispielgebend für die Terra sigillata präsentiert. Diese Methode wird jedoch für das gesamte keramische Fundmaterial angewandt. Für die Wandmalerei werden Farbpigmente und Mörtelzusammensetzung chemisch analysiert. Die angesprochenen Analysen werden von R.Sauer (Institut f. Silikatchemie) durchgeführt. Chemische Analysen sind auch für die Glasfunde geplant. 2.1.

Auch die reliefverzierten Produkte der Bernhard-Gruppen Ib und IIa, die am Ende des 2. und in den ersten beiden Jahrzehnten des 3. Jh. n. Chr. auftreten, sind am Michaelerplatz gut belegt.15 Daneben treten einige frühe Erzeugnisse von Comitialis aus Westerndorf auf.16 Bei glatten und barbotineverzierten Stücken trifft man am häufigsten die Formen Drag. 31 und Drag. 32 an, wobei der Anteil der Form Drag. 31 größer ist17. In der Zeit ab 220 n. Chr. bis zum Ende des 3. Jh. n. Chr. sind nur mehr wenige Terra sigillata-Lieferungen aus Rheinzabern18, Westerndorf und Pfaffenhofen bezeugt. Aus Rheinzabern liegen Dekorationsserien der Bernhard-Gruppe IIc bis IIIa und aus Westerndorf19 sowie Pfaffenhofen Produkte von Helenius vor. Aber auch ein Schüsselbruchstück des in Pfaffenhofen am spätesten belegten Formschüsselherstellers Dicanus konnte geborgen werden.20

Terra sigillata (P. Donat)

2.1.1: Zur Herkunft und Datierung der Terra Sigillata (Abb. 3) Die frühesten Terra sigillata Funde gehören noch in die spätflavische Zeit, aber der Beginn der Masseneinfuhr ist von der trajanischen Zeit bis in die Mitte des 2. Jh. n. Chr. zu setzen. Ein Argument, das nahelegt, die Datierung nicht allzusehr in die flavische Zeit zurückzuverschieben, ist der Umstand, daß am Michaelerplatz die Form Drag. 29, die während der flavischen Zeit ausläuft, gänzlich fehlt. Zu diesen früheren Funden gehören die Produktionen der spätesten Töpfer aus La Graufesenque, die verschiedenen Banassac-Importe10 und die frühesten mittelgallischen Waren.11 Bei der glatten und barbotineverzierten Terra sigillata sind in diesem Zeitraum das übliche Formenspektrum der süd- und mittelgallischen Produkte vorhanden. Außerdem ist bemerkenswert, daß die Formen Consp. 39 und 43 der Tardo Padana eine große Gruppe

Außerdem sind einige Stücke afrikanischer Waren der Produktionsgruppen Africana A und C vorhanden.21 Unter den Einzelstücken finden sich Fragmente von Saturninus/Satto bzw. aus Ittenweiler und Heiligenberg sowie ein Stück ostmediterraner sigillata ESB 2.

7

Czeike 1995: 490. s.v Paradiesgarten. Es handelt sich dabei um eine renaissancezeitliche Gartenanlage der Hofburg. Czeike 1992: 522. s.v. Hofburgtheater. Dieses ursprünglich im 16. Jh. als Veranstaltungsort für Festlichkeiten errichtete Gebäude wurde im 18. Jh. zum Hofoperntheater umgestaltet. 9 Harrer 1958: 363-404. 10 La Graufesenque - Töpfer: Mercator; Biragil; C. Cingius Senovirus; Banassac -Töpfer: Germanus; Biragil; Natalis; Tiere und „taillierte Ornamente“ unter der Dekoration. Faber 1994: 162-181; Mees 1995. 11 Frühe Mittelgallische Töpfer: Attianus; Bassus; Drusus I; Ioenalis; Libertus; Priscinus; Quintillianus; Sacer; Sissus II; Töpfer –X6; Töpfer – X8/Medeto – Ranto; Töpfer –X10/Silvio; Töpfer –X13/Geminus; Töpfer –X14; Töpfer –P5. Rogers1974; Stanfield & Simpson 1990; Faber 1994: 181-185.

12

Ettlinger 1990. Spätere mittelgallische Töpfer: Albucius; Censorinus; Fgientinus; Iullinus; Laxtucissa; Paternus; Pugnus. 14 Ricken & Fischer 1963; Bernhard 1981: 87; Zanier 1992: 126; 1994: 61. 15 Töpfer Bernhard Gruppe Ib/IIa: Cerialis II-IV; Belsus II Bernhard 1981: 87; Zanier 1992: 126; 1994: 61. 16 Kellner 1981: 123-157. 17 Zanier 1992: 132-135. 18 Bernhard 1981: 87-93. 19 Westerndorfer - Töpfer: Helenius. Gabler & Kellner 1994:189-248. 20 Christlein & Kellner 1969: 4, 88, 130-141, 150. 21 Hayes 1972; Carandini & Baldassarre 1981.

8

13

592

Sylvia Sakl-Oberthaler & Patrizia Donat et al: Die Ausgrabungen am Michaelerplatz

Formtyp

Produktionsgruppe Anzahl

Datierung

Hayes 2

Africana A

3

60–200 n. Chr.

Hayes 3

Africana A

2

60–200 n. Chr.

Hayes 14c

Africana A

3

175–225 n. Chr.

Hayes 20

Africana A

3

70–150 n. Chr.

Hayes 26

Africana A

1

150–250 n. Chr.

Hayes 45a

Africana C

2

230/240–320 n. Chr.

Hayes 52b

Africana A?

1

275–400 n. Chr.

Hayes 138, Nr. 3

Africana A

1

175–225 n. Chr.

Hayes 147, Nr. 3

Africana A

1

150–215 n. Chr.

(Wels) ergänzt wurden25, erfolgreich angewandt worden. Hierzu wurden bei den mit archäologischen Mitteln klassifizierten Scherben, unter Zuhilfenahme eines Binokulars, die Matrix und die Magerungsbestandteile am frischen Bruch nach standardisierten Vorlagen erfaßt und die Differenzierung der sog. Referenzscherben durch naturwissenschaftliche Untersuchungen abgesichert. Diese aufgestellten Referenzstücke bilden die weitere Basis zur Bestimmung nach Produktionsstätten, da die einzelnen Scherben durch Betrachtung unter dem Binokular nur mehr diesen Referenzstücken zugeordnet werden müssen. Mit dieser Methode konnte eine nachvollziehbare optische Zuweisung der Terra sigillata-Scherben zu Produktionsstätten erreicht werden. 2.2: Feinware (Abb. 4) (P. Donat & I. Pavic) Die Feinware kann generell von c.70 n. Chr. bis in das 3. Jh. n. Chr. datiert werden. Eine Konzentration ist für das ganze 2. Jh. n. Chr. feststellbar. Diese Tendenz ließ sich bereits bei der Terra sigillata beobachten.

Tabelle 1. Afrikanische Terra Sigillata

Die am Michaelerplatz vorhandenen Feinwarengruppen repräsentieren das für diese Zeitspanne und für die pannonische Region übliche Spektrum sowohl lokal produzierter als auch importierter Stücke.26 Zwei weitere Gruppen, nämlich bauchige Becher mit Überzug und Barbotinetropfendekor oder mit Barbotineschuppendekor, und tongrundige Ware mit Barbotinedekor, sind als lokale Produktionen anzusehen.27

2.1.2: Terra sigillata – Scherbenklassifizierung (S. Radbauer) 2.1.2. Die Provenienzzuweisung bei reliefverzierter und vor allem glatter Terra sigillata mit den traditionellen archäologischen Bearbeitungsmethoden, wie der typologischen Formenentwicklung, der Dekoranalyse und der makroskopischen sowie taktilen, meist aus der Erfahrung beurteilten Scherbenbeschreibung, kann nicht in allen Fällen eindeutig erreicht werden. Dies zeigt sich bei reliefverzierter Terra sigillata, wo nur mehr Punzenreste vorhanden sind, aber besonders bei glatter, barbotine- und kerbschnittverzierter Terra sigillata ohne Namenstempel bzw. wegen großer Formengleichheit. Dasselbe Problem tritt auch bei Namenstempel und Punzen auf, die in mehreren Produktionsstätten vorkommen (z. B., am Michaelerplatz, Biragil oder Helenius). Darüber hinaus liegt das sigillata-Material meist in kleinzerscherbten Fragmenten vor. Daher wurde an der hier vorgestellten Terra sigillata statt der herkömmlichen Scherbenbeschreibung das von D.P.S. Peacock entwickelte und in der Folge von V.Gassner und R.Sauer leicht modifizierte standardisierte Beschreibungsund Klassifikationsverfahren durchgeführt.22 Dieses Verfahren ist bereits an Terra sigillata in dem von R.Tomber und J.Dore herausgegebenen Handbuch The National Roman und parallel dazu etwas fabric collection“23 weiterentwickelt bei der Bearbeitung der Terra sigillata aus den östlichen canabae von Carnuntum24, wo die gebildeten Referenzgruppen mit der von P.Karnitsch publizierten reliefverzierten Terra sigillata aus Ovilava

Auffallend ist auch das Vorkommen von sechs Fragmenten von "Soldatenbechern" der Typen Magdalensberg 27-70.28 Sie sind in die Zeit von c.10 v. Chr.– c.20. n. Chr. zu datieren und bilden damit die älteste Feinwaregruppe. 2.3: Pannonische Glanztonware (P. Donat & I. Pavic) Die Pannonische Glanztonware läßt sich allgemein in die Zeit vom Ende des 1. bis ins 3. Jh. n. Chr. datieren.29 Der Großteil der Stücke ist reduzierend, etwa ein Drittel oxidierend gebrannt. Die Ware weist ein großes Spektrum an Formen auf. Es finden sich sowohl Imitationen von Terra sigillata-Gefäßen (vor allem Imitationen von Tellern Drag. 36), Ringschüsseln als auch eigenständige Formen. Zu den frühesten Funden zählen reduzierend gebrannte bauchige Töpfe mit eingeschnürter Wandung und hohem, verdicktem Rand, bei denen eine Vielzahl von Formvarianten feststellbar ist. Töpfe dieser Form sind in Nordwestpannonien in verschiedenen Keramikgattungen von der 2. Hälfte des 1. bis Anfang des 3. Jh. n. Chr.

25

Karnitsch 1959. Drexel 1911; Gassner 1992: 445; Gassner, Jilek & Sauer 1997: 179; Petznek 2000: 193–316. 27 Donat 1999a: 32-46. 28 Siehe Beitrag Kronberger & Mosser Fig. 3.2; 4; Schindler-Kaudelka 1975; 1998. 29 Gassner 1993: 359f.; Wölfl 1996.

22

26

Peacock 1977: 21-34; Orton, Tyers & Vince 1993: 67-75. 132-151. 231-242; Gassner 1997: 189-193; Gassner 2000: 185-190. 23 Tomber & Dore 1998: 25-41, 223-226. 24 Radbauer Diplomarbeit am Institut für Klassische Archäologie der Universität Wien, vor der Fertigstellung; Donat & Radbauer 1999: 208209; Gassner & Radbauer in Druck.

593

Limes XVIII belegt.30

nachgewiesen werden.36 Manche Nachläufer dieser Form37 dürften nach mineralogischen Untersuchungen nach Vindobona importiert worden sein, wobei Pannonien und Noricum als Herkunftsgebiet in Frage kommen.38 Für die ei- und birnenförmigen Gefäße gibt es sowohl Anzeichen für lokale Herstellung als auch für Import, wobei Pannonien und Oberitalien als Herkunftsort wahrscheinlich scheinen. Das Vorhandensein mehrerer Gefäßfragmente (Deckel, Töpfe, Reibschale, Krug, eiförmiges Gefäß), die sowohl verbrannt, verglast und verformt sind, sollte besonders analysiert werden. Leider sind diese letztgenannten Fragmente durch die hier angewendeten archäometrischen Methoden schwer zu untersuchen, da sie sehr stark verglast sind. Bevor man diese als Töpferofenabfälle identifiziert, muss die stratigraphische Auswertung der Grabung fertig gestellt werden.39

Einige reduzierend gebrannte Gefäße sind auf der Innenseite gestempelt (verschiedene Blattmotive, Rosetten, rechteckige Namenstempel und solche in planta pedisForm). Stempeldekor auf der Außenseite findet sich lediglich bei wenigen oxidierend gebrannten Schüsseln. 2.4: Gebrauchskeramik 2.4.1: Die oxidierend gebrannte Gebrauchskeramik mit und ohne Oberflächenveredelungen (R. Chinelli) Das Formenspektrum bei der oxidierend gebrannten Gebrauchskeramik mit und ohne Oberflächenveredelungen umfaßt zahlreiche Krüge, viele Teller, Reibschalen und Räucherschalen, weiters Deckel, Töpfe, Becken, Schüssel, Gesichtsgefäße (9 Exemplare, darunter einige reduzierend gebrannte), eiförmige und birnenförmige Gefäße (sog. Amphorenstöpsel) und sehr wenige Fragmente anderer Formen (Abb. 5).

2.4.2: Reduzierend gebrannte Gebrauchskeramik (U. Eisenmenger) Die reduzierend gebrannte Gebrauchskeramik - Töpfe, Schüsseln, Teller, Deckel, große Vorratsgefäße und Krüge - steht in engem typologischen Zusammenhang mit dem Material von Linz bis Carnuntum und datiert überwiegend 1. bis 3. Jh. n. Chr. Einige Stücke von einglättverzierter Keramik belegen den Zeitraum ab c.380 n. Chr. und wenige Fragmente der sogenannten Horreumware können der Spätantike zugeordnet werden.

Die Oberflächenveredelungen der hier untersuchten Keramik bestehen aus Engobe, Bemalung (sog. pannonische streifenverzierte Keramik), rotem Überzug sowie in seltenen Fällen Glasur und marmoriert-roter Bemalung. Graffiti und Stempel sind kaum vertreten.

Frühe Dreifuß-Schüsseln sind selten, der zeitliche Schwerpunkt liegt bei diesen im 2.Jh. n. Chr. Bei den Tellern läßt sich eine Formgruppe gut in das 2., eine weitere in das 3. Jh. n. Chr. einbinden.

Die Datierung dieser oxidierend gebrannten Gebrauchskeramik ist noch in Bearbeitung, aber es scheint, daß sie sich auf den Zeitraum zwischen dem Ende des 1. und der Mitte des 3. Jh. n. Chr. konzentriert. Die größte Menge des Materials läßt sich dem 2. Jh. n. Chr. zuweisen. Gewisse Keramiktypen können vor dem Ende des 1. Jh. n. Chr. datiert werden, wobei berücksichtigt werden muß, daß diese normalerweise eine lange Lebensdauer haben.31

2.5: Amphoren (T. Bezeczky) Die Anzahl der am Michaelerplatz gefundenen Amphoren ist nicht besonders hoch. Chronologisch sind sie in einem Zeitrahmen vom Ende des 1. bis 2. Jh. n. Chr. einzubinden. Der Wein wurde aus Italien, genauer aus Kampanien (Dressel 2-4), Südfrankreich (Gauloise 4), von der Insel Kreta (AC4) und aus Knidos importiert. Das Olivenöl stammt aus Istrien, die Oliven aus Norditalien (Schörgendorfer 558). Die verschieden gewürzten Fischsaucen wurden aus der iberischen Halbinsel (Dressel 8; Pelichet 46) eingeführt. In der späten Kaiserzeit wurde Wein aus Kleinasien, wahrscheinlich aus dem Mäandertal (Kapitän II) transportiert.

Die spätantike glasierte Keramik ist sehr spärlich vertreten (9 glasierte Fragmente, darunter 4 von Reibschalen). Die bis jetzt genauer untersuchten Keramikformen (Reibschalen, Räucherschalen, ei- und birnenförmige Gefäße) bestätigen diesen Datierungansatz. (Abb. 6). Die durchgeführten archäometrischen Untersuchungen (Schwermineral- und Dünnschliffanalysen- von R.Sauer) für bestimmte Keramikformen (Reibschalen und ei- und birnenförmige Gefäße32) weisen auch auf lokale Produktionen hin. Besonderes wichtig sind die Ergebnisse, die auf eine lokale Produktion der Reibschalen mit rotem Überzug (früher als „rätische“ Reibschalen bezeichnet33) hindeuten.34 Importe von schweren Reibschüsseln aus Italien, die nach archäometrischen Untersuchungen in Carnuntum festgestellt wurden35, konnten hier nicht

Hauptimportweg war die Bernsteinstraße. Als Beweis dafür dienen die Funde aus Carnuntum, wo die Verhältnisse ähnlich liegen.

30

Gabler & Rauchenwald 1992: 447–487, Nr. 478; Sedlmayer 1996: 22f. Gassner 1989: 134; Kronberger 1997: 80. 32 Chinelli 1997; 1998; Chinelli & Sauer 1 im Druck – a; im Druck – b. 33 Ladstätter 2000: 125 Anm. 698. 34 Chinelli 1998: 155. 35 Gassner 1989: 143, 148; Gassner & Sauer 1991: 15, Abb. 6, 3.1. 31

36

Chinelli 1998: 151. Kronberger 1997: Taf. 10.89. Chinelli 1997: 122; 1998: 156. 39 Schindler-Kaudelka & Zabehlicky-Scheffenegger 2000: 571-574. 37 38

594

Sylvia Sakl-Oberthaler & Patrizia Donat et al: Die Ausgrabungen am Michaelerplatz

Fibeltyp Bogenfibel Nauheim/Feugère 540 Ringfibel Fowler B1 Scharnierfibel Feugère 26b1 Kräftig profilierte Fibel Almgren 70/73 Kräftig profilierte Fibel Jobst 4F Kräftig profilierte Fibel Almgren 184 Kniefibeln Jobst 13D und 13F Tierfibeln Scharnierarmfibel Scheibenfibel Böhme 45 Zwiebelknopffibel Pröttel 3/4B

Material Ae Ae Ae, Email Ae Ae Ae Ae Ae Ae Ae, Glaseinlage fehlt Ae

Anzahl 1 1 1 1 5 3 2 4 1 1 1

Datierung 70/60-30/20 v. Chr. 1. Jh. n. Chr. 50/80 n. Chr. 80/100-150/180 n. Chr. 80/100-150/180 n. Chr. 150/130-200/220 n. Chr. 130/150-200/220 n. Chr. 3. Jh. n. Chr.(?) 3. Jh. n. Chr. 3. Jh. n. Chr. 330/400 n. Chr.

Tabelle 2: Fibelspektrum Anzahl

Ansprache

Typ

Datierung

1 4

Teller/Schale aus Millefioriglas Schalen m. n. außen gefaltetem vertikalem Rand Flache Schalen m. n. außen gefaltetem horizontalem Rand Rippenschalen Teller m. Schliff u. horizontalen Bearbeitungsspuren Schalen m. Kragenrand, Schliff u. horizontalen Bearbeitungsspuren. Sackförmige Fläschchen m. zylindrischem Hals (Balsamaria) Konische Fläschchen m. langem zylindrischem Hals (Balsamaria) Kugelförmige Fläschchen m. Ösenhenkeln (Aryballoi) Tintenfaß Becher m. abgesprengtem Rand u. Wanddellen Hoher konischer Becher m. Facettenschliff Dünnwandige glockenförmige Becher m. Schlifflinien Flaschen m. einem Bandhenkel u. quadratischem Querschnitt Zylindrische Schalen m. Standfuß u. horizontalen Bearbeitungsspuren Geschwungener Becher m. rundgeschmolzenem Rand u. Netzfadendekor Halbkugelige Schalen m. rundem Facettenschliff Konische Becher m. nach innen verdicktem Rand u. horizontalen Bearbeitungsspuren

unbestimmt Is. 44

0-150n. Chr. 50-100n. Chr.

Is. 45/46a/48

50-100n. Chr.

Is.3a AV V 34.2

50-250n. Chr. 60150n. Chr.

AR 22 oder AR 16.2

50-250n. Chr.

Is. 28 Varianten

0-125n. Chr.

Is. 82B1

80-110n.Chr.

Is. 61 Varianten

50-250n. Chr.

Is. 77 Bar. 50a Is. 21 Bar.34

50-250n. Chr. ca. 90-110n. Chr. 70-140n. Chr. 70-275n.

Is. 50

0-300n. Chr.

Is.85b u. Varianten

100-300n. Chr.

Bar. 57a

100-300n. Chr.

Bar. 25b

200-300n. Chr.

Bar. 39

250-400n. Chr.

Hoher Fußbecher m. Fadenauflage Schalen m. ausgestelltem Rand u. Wanddellen

Bar. 80 Bar.17, Nr. 29

275-300n. Chr. 350-400n. Chr.

8 3 1 1 4 2 3 1 1 2 1 30 6 1 2 2 1 1

Tabelle 3: Formenüberblick Hohlgläser

40

Siehe auch Beitrag Kronberger & Mosser fig. 4.

595

Limes XVIII

Jh. n. Chr. charakteristischen Typen am häufigsten vertreten. Das Repertoire wird von kräftig profilierten Fibeln dominiert, spezifische, ausschließlich der indigenen Frauentracht zugehörige Formen fehlen (Tab. 2). Interaktionen zwischen den Bewohnern der canabae und dem Militär verdeutlichen die Waffenfunde aus dem Siedlungsareal. Eindeutig dem militärischen Kontext zuzuweisen sind die Fragmente von Schutz- wie auch Trutzwaffen, u.a. Lanzen, Teile von Schienenpanzern des Typs Corbridge und des Typs Newstead/Jütting 2, Schildnagel Oldenstein 565/566, Randverstärkungen von Schilden oder Dolch- bzw. Schwertscheiden und Schwertscheidenklammer.

2.6: Lampen (S. Sakl-Oberthaler) Ca. 120 bestimmbare Fragmente ließen sich in folgende Typen einteilen: 1) Modelgepreßte Bildlampen mit Volutenschnauze (Loeschke Typen 1B und 1C41). 2) Modelgepreßte Firmalampen (Loeschke Typen IXB, XA, XC) mit den Stempeln Fortis, Atimentus, Cresces. 3) Scheibengedrehte Tiegellampen (Loeschke Typ XIII). 4) Achtförmige Lampen (Loeschke Typ XI). 5) Scheibengedrehte Lampen mit eckiger Schnauze (ähnlich Loeschke Typ XIX in Bronze). 6) Handgeformte mehrschnäuzige Lampen (ähnlich Loeschke Typ III).

2.9: Glasfunde (S. Sakl-Oberthaler) Unter den c.230 bestimmbaren Glasobjekten bilden die Hohlgläser die größte Gruppe. Darüber hinaus wurden Fragmente gläserner Fensterscheiben, ein für kosmetische Zwecke benutztes Glasstäbchen, eine gläserne Schmuckperle und - als verwandte Materialgruppe mehrere Melonenperlen aus Kieselkeramik gefunden.

Davon bilden Firmalampen der Typvarianten Loeschke XA und XC mengenmäßig die größte Gruppe, gefolgt von den Bildlampen mit Volutenschnauze (Loeschke 1C). Bei den Gruppen 3-6 handelt es sich um seltene Exemplare bzw. Einzelstücke. Damit entspricht das vorhandene Typenspektrum durchaus dem in Pannonien üblichen.

Unter den Hohlgläsern ließen sich 20 Formgruppen einwandfrei definieren (Tab. 3). Die überwiegende Anzahl dieser Formen wurde in der Zeitspanne zwischen der 2. Hälfte des 1. und dem 3. Jh. n. Chr. produziert. Ein Schwerpunkt ist dabei zwischen 50 n. Chr. und 250 n. Chr. zu erkennen. Die Dellenschale Bar.17 ist dabei die einzige für die Spätantike typische Form.

Chronologisch zeichnet sich ein deutlicher Schwerpunkt zwischen der 2. Hälfte des 1. und dem Ende des 3. Jh. n. Chr. ab. Spätantike Formen fehlen dagegen völlig. 2.7: Münzen (C. Litschauer-Vrba) Die insgesamt 217 Fundmünzen gliedern sich in mindestens 115 römisch-kaiserzeitliche und 2 vorkaiserzeitliche Münzen. Übrig bleiben die nicht näher bestimmbaren Münzen sowie c.10 nicht eindeutig als solche zu identifizierenden Objekte.

2.10: Beinartefakte (S. Sakl-Oberthaler) Die c.30 bestimmbaren Beinartefakte wurden entweder aus Tierknochen, Tiergeweih oder aus Tierhorn hergestellt.

Beim Münzmaterial überwiegt eindeutig Buntmetall, Silber tritt mit Ausnahme der zum kleingeld zählen den Antoninana selten auf, Gold gar nicht. Das zeitliche Spektrum (Abb. 7) der kaiserzeitlichen Münzen reicht von Augustus bis Valens II (d.h. vom Anfang des 1. bis in die 2. Hälfte des 4. Jh. n. Chr.).42 Aus diesem zeitlichen Rahmen fallen eine hellenistische (2. Jh. v. Chr.) sowie eine republikanische Prägung. Am häufigsten vertreten sind die Münzen der 1. Hälfte des 2. Jh. n. Chr., wobei trajanisches Material mit 19 Stück und hadrianisches mit 23 Stück überwiegt. Im Gegensatz dazu ist die relativ hohe Anzahl spätantiker Münzen aus dem 4. Jh. n. Chr. chronologisch weniger aussagekräftig, da der größere Teil verstreut in einem kleineren spätantiken Bereich aufgefunden wurde, der sich bei der Holzkonstruktion über der kaiserzeitlichen Bebauung befindet (Abb. 1/9).

Bei den aus diesen Materialien hergestellten Gegenständen handelt es sich um die allgemein üblichen Nähnadeln, Haarnadeln, aber auch um medizinisches Gerät (Ohrlöffelchen). Dazu kommen tesserae, sowie mehrere Beschläge und Griffe. Erwähnenswert ist dabei die quergerippte Griffhülse eines gladius-Griffes (1. Jh. n. Chr.)43, die die schon bei den Metallfunden beobachtete Verbindung zum Militärlager bestätigt (siehe H. Sedlmayer). Außerdem fallen unfertige Werkstücke und Schnitzabfälle einer Beinschnitzerei auf. Die dazugehörige Werkstatt konnte bisher nicht lokalisiert werden. 2.11: Wandmalerei (N. Willburger) Während der Grabung am Michaelerplatz wurden über 1000 Wandmalereifragmente gefunden. Trotz des stark fragmentarischen Zustandes war es möglich, Hinweise zur Ausstattung der archäologisch untersuchten Gebäude zu erhalten. Im folgenden sollen die ausgewählten Funde kurz vergestellt werden.

2.8: Metall – Kleinfunde (H. Sedlmayer) Die kaiserzeitlichen Fibeltypen datieren vom 1. bis ins 4. Jh. n. Chr., mit 16 Funden sind die für das Spektrum des 2.

Aus der ersten Kampagne (Abb. 1/4) liegen außer einfach 41 42

Loeschke 1919. Siehe auch Beitrag Kronberger & Mosser fig. 4.

43

596

Deschler-Erb 1998: Taf. 40-3998.

Sylvia Sakl-Oberthaler & Patrizia Donat et al: Die Ausgrabungen am Michaelerplatz profilierten Stuckleisten und Fragmenten einer Nischengliederung auch Reste einer gewölbten Decke mit Tapetenmalerei vor. Es handelt sich hierbei um Girlanden, die in alternierender Farbgebung (grün und ockerbraun) ein quadratisches Rastersystem bilden. Diese in unendlichem Rapport ausgeführten Muster sind in den nördlichen Provinzen weitläufig belegt und dort vom 2. Jh. n. Chr. bis in die Spätantike nachgewiesen.44

Einige Fragmente erlauben die Teilrekonstruktion der Mittelzone eines einfachen, weißgrundigen, linearen Felderschemas, wie es häufig vom ersten nachchristlichen Jahrhundert bis in die Spätantike zu finden ist.48 Teile eines Sockelbereiches einer zweiten weißgrundigen Dekoration wurde noch in situ im Hypokaustraum (Abb. 1/3) gefunden. Er war mit grünen Girlanden bemalt.

Über 100 Fragmente stammen von einer Felderdekoration in Rot-Schwarz-Kontrast, einem in den nördlichen Provinzen sehr beliebten Dekorationssystem.45 Während für die roten Felder ein gelber Innenrahmen anzunehmen ist, waren die schwarzen Trennfelder mit nicht mehr exakt zu bestimmender, vegetabiler Malerei in Rot, Ocker und Grün versehen.

Eine Datierung der hier kurz vorgestellten Fragmente muß gegenwärtig offenbleiben, da die Befundauswertung hierzu noch aussteht. 2.12: Die gestempelten Ziegel vom Michaelerplatz (Tab. 4) (M. Mosser) Vor einer endgültigen Auswertung der gestempelten Ziegel vom Michaelerplatz können folgende Aussagen zum Material und seiner Chronologie getroffen werden:

Während der zweiten Kampagne fanden sich eine Reihe von Fragmenten (Abb. 1/3), die die für die Zeichnung eines Leopardenfells charakteristischen Ringe zeigen und ihrer Größe nach von einer Megalographie stammen.46

Zunächst ist auffallend, daß kein explizit spätantikes Ziegelmaterial am Michaelerplatz anzutreffen ist, also zum Beispiel die typischen Stempel des 4. Jh. n. Chr. des Ursicinus (magister bzw. dux) oder Bonus magister.49 Die frühesten Stempel sind jene der legio XV Apollinaris und der legio XIII gemina. Ob der einzige Ziegel der 15. Legion mit einer frühen Stationierung in Vindobona in tiberischer Zeit oder mit ihrem Aufenthalt in Carnuntum ab claudischer bis in spättrajanische Zeit zusammenhängt, ist noch zu klären.50 Vom Ende des 1. Jh. n. Chr. stammen die beiden gestempelten Ziegel der legio XIII gemina, die in domitianischer Zeit ein neues Legionslager errichtete und bis zu den Dakerkriegen in Vindobona stationiert und auch an der Errichtung von Gebäuden der canabae beteiligt war. Überproportional viele Ziegel, im Vergleich zum Gesamtspektrum von Vindobona, stempelte die 14. Legion,51 die von 101 bis 114 n. Chr. die 13. Legion in Wien ersetzte. Für die Datierung stellt sich das selbe Problem wie bei den Stempeln der 15. Legion, da diese Legion ebenfalls nach ihrer Stationierung in Vindobona nach Carnuntum verlegt wurde und dort bis zur Spätantike verblieb. Es war bisher nicht zu entscheiden, ob die zahlreich in Vindobona gefundenen Stempel der 14. Legion aus der Stationierungsphase in Vindobona oder von Bautrupps aus Carnuntum stammen, daher sind sie vorerst zu Datierungszwecken nur bedingt einsetzbar.52 Überraschend wenige Ziegel sind von der ab 114 n. Chr. bis zur Spätantike in Vindobona stationierten legio X gemina vom Michaelerplatz bekannt, auch diese sind auf

Weitere Fragmente belegen die Darstellung eines zweiten Tieres; hierbei könnte es sich um einen Hund handeln. Sowohl diese Stücke als auch die Reste des Leoparden weisen Zweiphasigkeit auf. Soweit mit bloßem Auge zu erkennen, haben sie auch den gleichen Putzaufbau und wären somit möglicherweise derselben Wand zuzuordnen. Chemische Analysen hierzu stehen jedoch noch aus. Der Dekorationszusammenhang beider Darstellungen bleibt unklar, denkbar wäre eine Jagdszene. Zweiphasigkeit läßt sich bei einer weiteren Fragmentgruppe feststellen. Es handelt sich hierbei um grüne, vegetabile Malerei auf einer schwarzen Fläche, die mittels einer weißen Linie von einer grünen Fläche oder einem grünen Band getrennt wird; diese Fragmente könnten sowohl der Sockelzone als auch dem Hauptbildfeld zugewiesen werden. Wiederum ist eine Dekoration in rot-schwarzem Kontrast nachgewiesen. Während die gelbe Binnenrahmung der roten Felder hier gesichert ist, muß die Dekoration der schwarzen Trennfelder offenbleiben. Die Fragmente zeichnen sich durch eine sehr glatt polierte Oberfläche aus, wie sie bei dieser Dekorationsart hauptsächlich aus dem 1. und der 1. Hälfte des 2. Jh. n. Chr. bekannt ist. Zwei kleinere Stücke stammen ebenfalls von einer rotschwarzen Felderdekoration, deren rote Felder jedoch blau eingerahmt waren.47

48

Strocka 1975: 101-106; Thomas 1993: 19f.; Gogräfe 1999: 31f. Zum Gesamtbestand der Ziegel von Vindobona siehe Neumann (1973). 50 Zur Stationierung der 15. Legion und der weiteren Truppenabfolge in Vindobona siehe Beitrag Kronberger & Mosser in diesem Band. 51 Zu den in der Tabelle angeführten 9 Stempel der 14. Legion kommen noch mindestens 6 weitere, die allerdings nicht mehr auffindbar sind (Lörincz a. a. O. Nr. 3009 bis 3013, 3019). 52 Ein Projekt der Stadtarchäologie Wien in Zusammenarbeit mit der Österreichischen Akademie der Wissenschaften hat zum Ziel mit Hilfe archäometrischer Analysen die Herkunfts- und Produktionsfrage dieser Ziegel bzw. auch jener in Carnuntum von der 14. Legion gefundenen zu klären. 49

44

Peters 1965-66: 140f; Davey & Ling 1981: 199f.; Gogräfe 1999: 54ff.; Barbet, Douaud & Lanièpce 1997: 18ff.; Thomas 1993: 55f.: Thomas 1995: 306. 45 Praschniker & Kenner 1947: 218; Linfert 1975: 19ff.; Davey & Ling 1981: 31. 46 Thomas 1993: 108f., Abb. 20 198f., Abb. 74; Gogräfe 1997: 101, Abb. 100; 1999: 53f., Abb. 102. 47 Reusch 1966: 187-199; Gogräfe 1999: 25, 29, 239.

597

Limes XVIII

Fundnr.

Produzent

Ziegelart

KF 271/1 KF 533/1 Ohne FNr. KF 619/1 KF 160 KF 675/1 KF 685/1 KF 426/1 KF 436/1 KF 630/1 KF 160 KF 563 KF 366 3000/38 KF 713/1 KF 516/1 KF 525/1

Legio XV Apollinaris Legio XIII G Legio XIII Gemina Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio XIIII GMV Legio X Gemina Legio X Gemina Legio X Gemina Legio X Gemina Legio X Gemina

tegula Later

tegula

KF 526/1 KF 686/1 KF 375 KF 450/1 KF 623/1 KF 132/1 KF 690/1 KF 75 KF 79 KF 98 KF 566/1

Legio X Gemina Ala I Thracum victrix M Antonius Tiberianus M Antonius Tiberianus M Antonius Tiberianus M Antonius Tiberianus M Antonius Tiberianus C Valerius Constans C Valerius Constans C Valerius Constans P L( ) Se( )

tegula tegula tegula Later tegula Later Later tegula tegula tegula tegula

Typ (Lörincz)56 60

Later Later Later tegula tegula tegula tubulus ? Later tegula tegula tegula tegula

Vgl. 121 275 219 107 398 381

331, 361

4 3 oder 4 4 1 1 1 1

Taf. (Lörincz) 66/12 34/8

Datierung

Bis 114 89-101 89-101 Ab 101 Ab 101 Vgl. 44/13 Ab 101 Ab 101 53/13 Ab 101 65/16 Ab 101 50/10 Ab 101 Ab 101 43/13 Ab 101 23/14 Ab 114 22/16 Ab 114 222-235? Ab 114 Vgl. 20/4, Ab 114 21/16 Ab 114 2.Jh. 2. Jh. 62/1 2. Jh. 61/18 o.62/1 2. Jh. Vgl. 61/9 2. Jh. 62/1 2. Jh. 63/13 1. H. 2. Jh. 63/13 1. H. 2. Jh. 63/13 1. H. 2. Jh. 63/3 2./3.Jh.

Tabelle 4: Spektrum der Ziegel mit Stempel Gebäudestrukturen, bestehend aus Wohneinheiten und Werkstätten befinden sich im halbmilitärischen Bereich die zu Lager und der canabae legionis.57 Obwohl canabae gehörigen Gräberfelder sich sichtlich entlang der auch am Michaelerplatz vorbeilaufenden Limesstraße befanden, wurden am Platz selbst keinerlei Hinweise dafür gefunden. In unmittelbarer Nähe sind in diesem Zusammenhang die Auffindung eines Soldatengrabsteines in der Habsburgergasse sowie eines Sarkophagbruchstückes am Josefsplatz58 zu erwähnen.

Grund des langen Aufenthaltes der Legion nur schwer chronologisch eingrenzbar. Erstmals konnte in Wien ein Stempel der ala I Thracum victrix nachgewiesen werden, die im 2. Jh. n. Chr. im Auxiliarkastell Carnuntum belegt ist.53 Der Anteil an privaten Produzenten (M. Antonius Tiberianus, C. Valerius Constans) mit etwa einem Drittel der Gesamtzahl der Ziegel ist verhältnismäßig hoch, allerdings für den Bereich der canabae legionis durchaus zu erwarten. C Valerius Constans, der im 2. Jh. in Carnuntum produzierte,54 ist in der bisherigen Forschung in Vindobona nur aus dem Bereich der Zivilstadt bekannt.55

Nach dem hier präsentierten Überblick über das Fundmaterial lassen sich folgende chronologische Schwerpunkte definieren. Zu einem ersten deutlichen mengenmäßigen Anstieg an Funden kommt es in der

3: Resumée Die am Michaelerplatz während der Kampagnen 1990/1991 freigelegten römischen Gebäudereste befanden sich am Kreuzungspunkt zwischen der Limesstraße und der das Lager südwestlich-nordöstlich durchquerenden sogenannten pannonischen Transversale. Die

56 Eine Gesamtpublikation zu den Ziegelstempeln von Vindobona durch B. Lörincz ist bereits abgeschlossen. Der Druck des Manuskripts, auf das sich die Typen- und Tafelangaben beziehen, ist in der Reihe der Wiener Archäologischen Studien (WAS) in Vorbereitung. 57 Weitere Hinweise auf die Situierung der canabae rund um das Legionslager im ersten Bezirk fanden sich im Verlauf der Grabungstätigkeit der letzten Jahre auf der Freyung sowie in der Renngasse. Börner 1997: 247-249. 58 Siehe Beitrag Kronberger & Mosser fig. 7; 8; Harl 1977: 102; Kenner 1897: 139-143.

53

Stiglitz 1997: 75. Lörincz 1989: 99. 55 Neumann 1973: 102, Nr. 1795. 54

598

Sylvia Sakl-Oberthaler & Patrizia Donat et al: Die Ausgrabungen am Michaelerplatz spätflavischen Epoche. Aus der Zeitspanne davor gibt es nur wenige Einzelfunde, darunter eine Fibel aus der 2. Hälfte des 1. Jh. v. Chr., einige Feinwarefragmente aus augusteisch-tiberischer Zeit und eine sowie eine Bronzemünze aus dem 2. Jh. v. Chr.59 Über ihre Bedeutung hinsichtlich einer vorflavischen Besiedlungsphase am Michaelerplatz kann zu diesem Zeitpunkt noch nicht entschieden werden. Die größte Anzahl von Funden stammt schließlich aus der Zeit zwischen 100/110 und den Markomannenkriegen (bis 180 n. Chr.). Aus dem darauffolgenden Zeitraum zwischen 180 und 230 n. Chr. stammt ebenfalls ein beachtlicher, jedoch kleinerer Prozentsatz des Fundmaterials. Die Entwicklung während der 2. Hälfte des 3. Jh. ist derzeit auf Grund der geringen Anzahl von Funden, die zweifelsfrei in diese Zeit datiert werden können schwer nachzuvollziehen. Die Importe an Terra sigillata enden etwa um diese Zeit. Auffallend ist auch, daß der Anteil des typisch spätantiken Fundmaterials äußerst gering bleibt.

Enciclopedia italiana fondata da Giovanni Treccani. (Roma). Chinelli R. 1997 Analisi archeometriche sui mortai di Michaelerplatz. A proposito di alcuni mortai di probabile provenienza africana. In S. Santoro Bianchi & B. Fabbri (edd.) Il contributo delle analisi archeometriche allo studio delle ceramiche grezze e comuni: il rapporto forma/funzione/impasto, Atti della 1a Giornata di archeometria della ceramica, Studi e scavi del Dipartimento di Bologna, 28 febbraio 1997 (Bologna): 121–127. Chinelli R. 1998 DeiReibschalen der Grabung Michaelerplatz 1990-91: Archäometrische Analysen. Fundort Wien. Berichte zur Archäologie 1: 153-159. Chinelli R. 2001 Drei Gefüßfragmente mit bacchischer Darstellung vom Michaelerplatz. Fundort Wien. Berichte zur Archäologie 4: 30-63. Chinelli R. & Sauer R. im Druck – a Die Reibschalen der Grabung Wien 1, Michaelerplatz. Archeologie et Histoire Romaine. Chinelli R. & Sauer R. im Druck – b Eiförmige und birnenförmige Gefäße aus Wien. Wiener Archäologischen Studien. Christlein R. & Kellner H.J. 1969 Die Ausgrabungen 1967 in Pons Aeni. Bayerische Vorgeschichtsblätter 34: 76159. Czeike F. 1992 Historisches Lexikon Wien Bd. 1. (Wien). Czeike F. 1995 Historisches Lexikon Wien Bd. 4 . (Wien). Davey N. & Ling R. 1981 Wall painting in Roman Britain. Britannia Monograph Series No.3. (London). Deschler-Erb S. 1998 Römische Beinartefakte aus Augusta Raurica. (Forschungen in Augst 27/1 u. 27/2: Römermuseum, Augst). Donat P. 1999a Feinkeramik aus Vindobona – Hinweise auf eine lokale Produktion? Fundort Wien, Berichte zur Archäologie 2 : 32-46. Donat P. 1999b Zur Herkunft der Terra Sigillata von der Ausgrabung Michaelerplatz. Fundort Wien, Berichte zur Archäologie 2: 210–215. Donat P. & Radbauer S. 1999 Klassifikation von Scherbentypen an Terra Sigillata. Fundort Wien, Berichte zur Archäologie 2: 208-209. Drexel F. 1911 Das Kastell Faimigen (Der ObergermanischRätisches Limes B 66 c. 30). Ettlinger E. 1990 Conspectus formarum terrae sigillatae italico modo confectae. (Materialien zur römischgermanischen Keramik 10. (Bonn). Faber A. 1994 Das Römische Auxiliarkastell und der Vicus von Regensburg - Kumpfmühl. Münchner Beiträge zur Vorund Frühgeschichte 49. (München). Feugère M. 1985 Les fibules en Gaule méridionale de la conquête à la fin du Ve s. ap. J.-C. Revue Archéologique de Narbonnaise Suppl. 12. (Paris). Fowler E. 1960 The origins and development of the penannular brooch in Europe. Proceedings of the Prehistoric Society 26: 149-177. Gabler D. & Kellner H.J 1994 Die Bilderstempel von Westerndorf. II. Helenius und Onniorix. Bayerische Vorgeschichtsblätter 58: 189-248. Gabler D. & Rauchenwald A. 1992 Töpferkunst und Keramikindustrie. In W. Jobst (ed.) Carnuntum, das Erbe Roms an der Donau. (Katalog der Ausstellung des Archäologischen Museums Carnuntinum in Bad Deutsch Altenburg Carnuntum). Gassner V. 1989 Gelbtonige Keramik aus datierten Fundkomplexen in Carnuntum. Ein Überblick über die Gefäßformen. Carnuntum Jahrbuch: 133-149.

Hinsichtlich der Zusammensetzung des Fundmaterials ist erwähnenswert, daß unter den Kleinfunden auch Waffenfunde sind, die auf Beziehungen zwischen den Bewohnern der canabae und dem Militär hindeuten. Die dort aufgefundenen Ziegel wurden teilweise in Legionsziegeleien hergestellt. Hingewiesen werden muß darüber hinaus auf einige Einzelfunde. Dabei handelt es sich um ein steinernes Altarfragment, eine tönerne Maske und ein reliefverziertes Gefäß mit bacchischen Motiven.60 Dazu kommen Produktionsreste von Glas- und Metallverarbeitung sowie möglicherweise von Töpfern.

Bibliographie Almgren O. 1923 Studien über nordeuropäische Fibelformen. (Leipzig). Barbet A., Douaud R. & Lanièpce V. 1997 Imitations d’opus sectile e decors a reseau. Essai de terminologie. (Paris). Barkóczi L. 1988 Pannonische Glasfunde in Ungarn. Studia Archaeologica IX: Verlag der ungarischen Akademie der Wissenschaften. (Budapest). Bernhard H. 1981 Zur Diskussion um die Chronologie Rheinzaberner Relieftöpfer. Germania 59/1: 79–93. Böhme A. 1972 Die Fibeln der Kastelle Saalburg und Zugmantel. Saalburg-Jahrbuch 29: 5-112. Börner W. 1997 Wien - Vindobona. Legionslager – Canabae Autonome Stadt. In H. Friesinger & F. Krinzinger (Hsrg.) Der Römische Limes in Österreich. Verlag der Österreichischen Akademie der Wissenschaften. (Wien): 241-253. Brein F. & Sauer E. 2001 Eine tönerre Maske – O Jefetl, der Mon-Mon ! Fundort Wien, Berichte zur Archäologie 4: 4-17. Carandini A. & Baldassarre I.A. 1981 EAA. Atlante delle forme ceramiche. I. Ceramica fine romana nel bacino del Mediterraneo. Medio e tardo impero. (Enciclopedia dell’arte antica classica e orientale. Istituto della 59 60

Siehe Beitrag Kronberger & Mosser fig. 4. Brein & Sauer 2001; Pavic & Sauer 2001; Chinelli 2001.

599

Limes XVIII

Gassner V. 1992 Feinware in Carnuntum. Import und lokale Produktion. Rei Cretaria Fautorum Acta 31-32: 445463. Gassner V. 1993 Pannonische Glanztonware mit Stempelverzierung aus Carnuntum. In Ptujski arheoloski zbornik ob 100-letnici muzeja in Muzejskega drustva. Gassner V. 1997 Der Töpferofen von Carnuntum. In H.Stiglitz & S.Jilek (edd.) Das Auxiliarkastell Carnuntum I. Sonderschriften des Österreichischen Archäologischen Instituts 29. (Wien). Gassner V. 2000 Die Keramik mit Ausnahme der Terra sigillata. 2.2. Scherbentypen. In V. Gassner et al. Das Kastell Mautern – Favianis. Der Römische Limes in Österreich 39, Verlag der Österreichischen Akademie der Wissenschaften. (Wien). Gassner V., Jilek S. & Sauer R. 1997 Der Töpferofen von Carnuntum In H. Stiglitz (ed.) Das Auxiliarkastell Carnuntum 1. Forschungen 1977-1988. Sonderschriften des Österreichischen Archäologischen lnstitutes 29. (Wien): 179-229. Gassner V. & Radbauer S. im druck Produktionszuweisung bei Terra Sigillata durch Scherbenklassifizierung. (Kolloquium „Römische Keramik - Herstellung und Handel“ im Regionalmuseum Xanten Juni 2000). Gassner V. & Sauer R. 1991 Archäometrische Untersuchungen zur Keramikproduktion in Carnuntum. In Römischer Limes in Österreich. Verlag der Österreichischen Akademie der Wissenschaft 37. (Wien). Gogräfe R. 1997 Die Wand- und Deckenmalereien der Villen von Bad Kreuznach und Bingen-Kempten. Mainzer Archäologische Zeitschrift 4: 1-110. Gogräfe R. 1999 Die Römischen Wand- und Deckengemälde im nördlichen Obergermanien. Archäologische Forschungen der Pfalz 2. (Neustadt an der Weinstraße). Harl O. 1977 Friedhöfe und Canabae des Legionslagers. In: Vindobona-Die Römer im Wiener Raum. (Katalog der 52. Sonderausstellung des Historischen Museums der Stadt Wien, Karlsplatz, Eigenverlag des Museums der Stadt Wien): 101-103. Harl O. 2001 Die Römerzeit. In P. Csendes & F. Opll (edd.) Wien, Geschichte einer Stadt von du Anfangen bis zur ersten Türkenteilangesung. Bd. 1. (Wien). Harrer P. 1958 Wien, seine Häuser, Menschen und Kultur Bd. 6 (maschinenschriftl. Mskr. Wien). Hayes J.W. 1972 Late Roman pottery. (London). Isings C. 1957 Roman Glass from dated finds. (Groningen). Jobst W. 1975 Die römischen Fibeln aus Lauriacum. Forschungen in Lauriacum 10. (Linz). Jütting I. 1995 Die Kleinfunde aus dem römischen Lager Eining-Unterfeld. Bayerische Vorgeschichtsblätter 60: 143-230. Karnitsch P. 1959 Die Reliefsigillata von Ovilava. Institut für Landeskunde von Oberösterreich 12. (Linz). Kellner H. 1981 Die Bildstempel von Westerndorf. Comitialis und Iassus. Bayerische Vorgeschichtsblätter 46: 123157. Kenner F. 1897 Die Archäologischen Funde aus römischer Zeit. In F. Kenner Geschichte der Stadt Wien (Wien): 139-143. Kronberger M. 1997 Ausgewähltes keramisches Fundmaterial aus stratifizierten Fundkomplexen. In H. Stiglitz & S. Jilek (edd.) Das Auxiliarkastell Carnuntum 1. Sonderschriften des Österreichischen Archäologischen Institutes 29: 77-135. Ladstätter S. 2000 Die materielle Kultur der Spätantike in den

Ostalpen. (Wien). Linfert A. 1975 Römische Wandmalerei der nordwestlichen Provinzen. (Köln). Loeschke S. 1919 Lampen aus Vindonissa: Ein Beitrag zur Geschichte von Vindonissa und des antiken Beleuchtungswesens. (Zürich). Lörincz B. 1989 Zu den Verbindungen zwischen Pannonien und Barbaricum. Die Verbreitung und Datierung der Ziegelstempel. Klio 71: 96-106. Mees A.W. 1995 Modelsignierte Dekorationen auf südgallischer Terra Sigillata. Forschungen und Berichte zur Vor- und Frühgeschichte in BadenWürttemberg 54. (Stuttgart). Neumann A. 1973 Ziegel aus Vindobona. Der römische Limes in Österreich 27. (Wien). Oldenstein J. 1976 Zur Ausrüstung römischer Auxiliareinheiten. Berichte der Römisch-Germanischen Kommission 57: 49-284. Orton C., Tyers P. & Vince A. 1993 Pottery in archaeology. (Cambridge). Pavic I. & Sauer R. 2001 Das römische Altärchen vom Michaelerplatz. Fundwort Wien, Berichte zur Archäologie 4: 20-29. Peacock D.P.S. 1977 Ceramics in Roman and medieval archaeology. In D.P.S. Peacock (ed.) Pottery in early commerce. Characterization and trade in Roman and later ceramics. (London). Peters W.J.T. 1965-66 Mural painting found in the Roman castra at Nijmegen. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoeck 15/16: 113-144. Petznek B. 2000 Römerzeitliche Gebrauchskeramik von Carnuntum. Carnuntum Jahrbuch 1999: 193–316. Praschniker C. & Kenner H. 1947 Der Bäderbezirk von Virunum. (Wien). Pröttel Ph. M. 1988 Zur Chronologie der Zwiebelknopffibeln. Jahrbuch des Römisch-Germanischen Zentralmuseums Mainz 35/2: 347-371. Radbauer S. Diplomarbeit am Institut für Klassische Archäologie der Universität Wien vor der Fertigstellung. Reusch W. 1966 Wandmalereien und Mosaikboden eines Peristylhauses im Bereich der Trierer Kaiserthermen. Trierer Zeitschrift 29: 187-199. Ricken H. & Fischer C. 1963 Die Bildschüsseln der römischen Töpfer von Rheinzabern. Materialien zur römischgermanischen Keramik 7. (Bonn). Rogers G. 1974 Potteries sigillées de la Gaule centrale. Les motifs non figurés. Gallia supplément 28. (Paris). Rütti B. 1991 Die römischen Gläser aus Augst und Kaiseraugst. Forschungen in Augst 13/1 u. 13/2. (Augst). Sakl-Oberthaler S. 1999 Untersuchungen zur Limesstraße in Wien – Simmering. Fundort Wien, Berichte zur Archäologie 2: 110-127. Schindler-Kaudelka E. 1975 Die dünnwandige Gebrauchskeramik vom Magdalensberg. Kärntner Museumsschr. 58. (Klagenfurt). Schindler-Kaudelka E. 1998 Die dünnwandige Gebrauchskeramik vom Magdalensberg. 2. Die pareti sottili vom Südhang des Magdalensberg. Die Ausgrabungen auf dem Magdalensberg 1980 bis 1986. Magdalensberg Grabungsberichte 16. (Klagenfurt): 389-427. Schindler-Kaudelka E. & Zabehlicky-Scheffenegger S. 2000 Es war der Schadensbrand und nicht der Fehlbrand. Rei Cretariae Fautorum Acta 36 : 571-574. Sedlmayer H. 1996 Römische Grabfunde aus Müllendorf und Oggau. Wissenschaftliche Arbeiten aus dem Burgenland 96: 3-57.

600

Sylvia Sakl-Oberthaler & Patrizia Donat et al: Die Ausgrabungen am Michaelerplatz Sedlmayer H. 1998 Fabri aerarii und plumbarii in Vindobona. Instrumentum 8: 22. Stanfield J.A. & Simpson G. 1990 Les potiers de la Gaule centrale. (Marseille). Stiglitz H. (Hrsg.) 1997 Das Auxiliarkastell Carnuntum 1. Sonderschriften des Österreichischen Archäologischen Institutes 29. (Wien). Strocka V.M. 1975 Pompejanische Nebenzimmer. In A. Andreae & H. Kyrieleis (edd.) Neue Forschungen in Pompeji. (Recklinghausen): 101-106. Symonds R. P. 1992 Rhenish Wares. Fine dark coloured pottery from Gaul and Germany. Oxford University Committee for Archaeology Monograph 23. (Oxford). Thomas R. 1993 Die römische Wandmalerei in Köln. Kölner Forschungen Bd. 6. (Mainz). Thomas R. 1995 Die Dekorationssysteme der römischen Wandmalerei von augusteischer bis in die trajanische Zeit. (Mainz). Tomber R. & Dore J. 1998 The national Roman fabric reference collection. A handbook. (London). Wölfl K. 1996 Pannonische Glanztonware aus dem Auxiliarkastell von Carnuntum. (Diplomarbeit: Universität Wien). Zanier W. 1992 Das römische Kastell Ellingen. Limesforschungen 23. (Mainz). Zanier W. 1994 Handelsfragen der Rheinzaberner Sigillata. Beiträge zur antiken Handelsgeschichte 13: 60-69.

601

Limes XVIII

Abb. 1

Abb. 2 602

Sylvia Sakl-Oberthaler & Patrizia Donat et al: Die Ausgrabungen am Michaelerplatz

Abb. 3 Herkunft, Menge und Datierung der verzierten Terra Sigillata.

Abb. 4. Herkunft, Menge und Datierung der Feinware.

Abb. 5. Verteilung der Formen der oxidierend gebrannten Gebrauchskeramik.

603

Limes XVIII

Abb. 6. Datierung Reibschalen. H e lle n is t . S y r ie n ? 2 . J h . v . C h r . 1 C a e sa r 4 9 - 4 8 v . C h r. 1 A u g u stu s 2 7 v . - 1 4 n . C h r. 1 C a iu s 3 7 - 4 1 n . C h r. 2 N e r o 5 4 - 6 8 n . C h r. 1 V e s p a s ia n u s 6 9 - 7 9 n . C h r . 1 D o m i ti a n u s 8 1 - 9 6 V n .a le C nhsr .I I . 3 6 4 - 3 7 8 n . 3C h r . N e r v a 9 6 - 9 8Vna.leCn h t inr .ia n u s I . 3 6 4 - 3 7 5 n . 1C h r . T ra ia n u s 9 8 - 1 1 7 n . C h r. 19 C o n s t a n t iu s I I . 3 3 7 - 3 6 1 n . C h r . H a d r ia n u s 1 1 7 - 1 3 8 n . C h r . 23

M P b e s tim m b a r e r ö m is c h e M ü n z e n 6 4 9

C o n s ta n s 3 3 7 - 3 5 0 n . C h r.

2

A n t o n i n u s P i u s C1o3n8s ta - n 1 tin 6 1u ns .I.C3h0r.6 - 3 3 7 n .7C h r . L . V e ru s 1 6 1 - 1 6 9 n . C h r. 1 M . A u re liu s 1 6 1 C- o1n8s0tann.tiu Csh rI.. 3 0 5 - 3 0 6 n .2C h r .

13 1

L ic in iu s I. 3 0 8 - 3 2 4 n . C h r .

2

S e p tim iu s S e v e r u s 1 9 3 - 2 1 1 n . C h r . 3 A u r e lia n u s 2 7 0 - 2 7 5 n . C h r . C a r a c a lla 2 1 1 - 2 1 7 n . C h r . 1

2

G a llie n u s 2 5 9 - 2 6 8 n . C h r .

2

Herrscher

S e v e ru s A leSxeavnedr uesr A 2 le 2 2x a-n 2d 3e r5 2n2.2 C- h2r.3 5 n .4C h r . G a l lie n u s 2 5 9 - 2 6 8 n . C h r . 2 C a r a c a lla 2 1 1 - 2 1 7 n . C h r . A u r e lia n u s 2 7 0 - 2 7 5 n . C h r . 2 L ic in iu s I. 3 0S8e p- tim 3 2iu 4 snS. eCvhe r. r u s 1 9 3 - 2 1 1 n .2C h r . C o n s t a n tiu s I. 3 0 5 - 3 0 6 n . C h r . 1

4 1 3

M . A u r e liu s 1 6 1 - 1 8 0 n . C h r .

2

r u. sC1h6r. 1 - 1 6 9 n .1C C o n s t a n ti n u s I . 3 0 6 - L3.3V7e n 3h r. C o n s t a n s 3 3 7 - A3n5to0n nin.uC s hPr.iu s 1 3 8 - 1 6 1 n .2C h r .

1 7

H a d r ia n u s 1 1 7 - 1 3 8 n . C h r .

C o n s t a n ti u s II . 3 3 7 - 3 6 1 n . C h r.

23

9

T r a ia n u s 9 8 - 1 1 7 n . C h r .

19

V a le n t in ia n u s I . 3 6 4 - 3 7 5N ne r. v aC 9h6r. - 9 8 n .4C h r . V a le n s II. 3 6 4 - 3 7 8 n . C h r. 6

1

V e s p a s ia n u s 6 9 - 7 9 n . C h r .

1

N ero 5 4 - 6 8 n . C h r.

1

D o m itia n u s 8 1 - 9 6 n . C h r .

3

C a iu s 3 7 - 4 1 n . C h r .

2

A u g u s tu s 2 7 v . - 1 4 n . C h r .

1

C a es a r 4 9 - 4 8 v . C h r.

1

H e lle n is t . S y r ie n ? 2 . J h . v . C h r .

1

0

5

10

15 Anzahl

Abb. 7. Zeitliches Spektrum der Münzen.

604

20

25

Legionslager an der Wende zur Spätantike – Ein Überblick zu Carnuntum und vergleichbaren kaiserzeitlichen Standlagern des Rhein-Donau-Raumes in einer Periode des Umbruchs Raimund Kastler Published in a series of short preliminary reports, the architecture and the stratification uncovered during the 1968–1977 excavations in the eastern praetentura of the legionary fortress at Carnuntum are now being studied in more detail. Besides questions of stratification and chronology, this project jointly conducted by the excavator Dr. M. Kandler and the author is designed to draw a more precise picture of the fortress’ architectural layout in different phases of its development. An issue of special interest is the transitional stages of the legionary fortress’ buildings in to late antiquity. Already described by Dr. Kandler, there is a remarkable change in the architectural structures between the buildings of Period 4 dating from the later C3rd or early C4th AD and the reconstruction in Period 5 of the later C4th and usually connected with rebuilding under the emperor Valentinian II. The earliest phase is a traditional rebuilding of the barracks following the dimensions and design of the previous ones. The latter period abandoned the barrack system for a new design. Even from this small glimpse of the eastern praetentura a new kind of standardised architectural scheme emerges, distinct from the loosely grouped habitations which followed in the next period. In the area of the 1968– 1977 excavations heated rooms in the north contrast with open spaces in the south, supplied with centrally placed water basins in the centre. Several questions arise from this evidence. How were the old legionary fortresses used in the course of the C4th and how do they reflect the changes in the military organisation in their architecture, compared to newly built fortresses ? Does the use of the internal space in the fortresses change due to reduced numbers of troops ? How long is the standard barrack buildings scheme used ? Are there new late antique building types along the Danube region like those identified in the fortlets of Britain ? Answers to these questions are sought through comparison with other legionary fortresses and fortlets in the northern provinces built and used in late antiquity.

Als 1968 unter der Leitung von H. Vetters und M. Kandler die Ausgrabungen im Legionslager von Carnuntum nach mehr als fünfzigjähriger Unterbrechung wiederaufgenommen wurden, galt das hauptsächliche Interesse neben der Schließung der topographischen Lücke im Ostbereich des Lager-N-Teiles (praetentura), der Erforschung der einzelnen Bauperioden (Abb. 1). Hatten die Grabungen M. Grollers v. Mildensee im Lagerzentrum (latera praetorii) und im S-Bereich (retentura) des Lagers zwar zahlreiche Hinweise auf zumindest zwei größere Umbauphasen in der Lagerbebauung erbracht, so blieb es den Untersuchungen E. Novotnys vorbehalten, in der Wpraetentura sieben aufeinanderfolgende Bauhorizonte zu beobachten (Genser 1986: 576ff. bes. 588ff.).

unterging (Kandler 1989: 313-15). Das mit dieser Periode bzw. deren Erbauung zu verbindende keramische Spektrum ist durch rot überzogene Platten, Faltenbecher und glasierte Reibschalen charakterisiert, weshalb M. Grünewald Periode 4 in den Zeitraum vor 350 n. Chr. einordnete. Nach Befunden in der Lagersiedlung (canabae) von Carnuntum und Vergleichskontexten aus Ungarn ist jedoch bereits mit einem Vorkommen glasierter Reibschalen am Ende des 3. Jh.s n. Chr. bzw. am Anfang des folgenden Jahrhunderts zu rechnen (Gassner & Jilek 1999: 52f, 5762, Abb. 7-9). Dementsprechend ist ein Ansatz dieser Bauperiode bereits in tetrarchischer Zeit wie ihn M. Kandler vertritt, durchaus im Bereich des Möglichen (Kandler 1974: 35). Diese Datierung harmoniert mit dem Baubefund, wie im folgenden gezeigt werden soll, auch besser als eine konstantinische oder jüngere Entstehung.

Bis zu der nicht vorhersehbaren Einstellung der neuen Ausgrabungen im Jahr 1977 gelang es im Legionslager, die Existenz eines Holz-Erde-Lagers aus der Mitte des 1. Jh.s n. Chr. zu erhärten und darüber hinaus drei weitere Ausbauphasen in Stein sowie zwei weitere Perioden mit geändertem Bebauungsschema in Ausschnitten zu erfassen (Kandler 1974: 28ff.).

Mit Periode 4 lässt sich für die NO-praetentura eine grundsätzliche Neubebauung feststellen. Die Mauern von Periode 3 werden bis in die Fundamente hinab abgetragen, die Bodenniveaus zum größten Teil beseitigt. Möglicherweise waren die Kasernen der Periode 3 während der Zeit, als die Legion unter Gordian oder Philippus Arabs der mobilen Eingreiftruppe angehörte, baufällig geworden (Wilkes 2000: 112).

Im Rahmen der 1998 gemeinsam mit M. Kandler begonnenen Aufarbeitung der Bau- und Schichtbefunde aus den neueren Grabungen in der O-praetentura erwiesen sich die spätantiken Bauperioden des Lagers von besonderem Interesse.

Die neu errichteten Bruchsteinmauern der Kasernen folgen in ihrer Planführung zumeist fast identisch den Baufluchten der vorangegangenen Periode, wobei sich jedoch erstmals in Periode 4 ein klares Gliederungsschema der Mannschaftsbaracken aus dem Befund herausarbeiten

Anhand des Fundmaterials ist bereits die dritte Steinbauphase oder Periode 4 (Abb. 2) diesem Zeitraum zugehörig, die dem Befund nach in einer Erdbebenkatastrophe wohl um die Mitte des 4. Jh.s n. Chr. 605

Limes XVIII

lässt.

Germaneneinfällen 275/76 n. Chr. aufgegeben wurde, erfolgte in Bonn möglicherweise bereits in den 80er Jahren des 3. Jh.s n. Chr. ein grundlegender Neubau des ebenso wie die Zivilsiedlung von Zerstörungen betroffenen Standlagers. Bislang sind von diesen Baumaßnahmen die Umfassungsmauer und eine im Bereich des ehemaligen Wehrganges angelegte fabrica erforscht. Auskünfte über den weiteren spätantiken Innenausbau des Lagers sind nicht vorhanden (Gechter 1980: 531ff.).

Die teilweise als Doppelkasernen mit gemeinsamer Mittelmauer errichten Gebäude, denen eine Veranda (porticus) mit Holzpfeilern (teilweise auf Steinbasen) vorgeschaltet war, wurden durch Zwischenmauern gegliedert. Kennzeichnend für die Grundrissführung der Mannschaftsquartiere ist dabei ein schmaler, parallel zum Vorraum (arma) angeordneter Korridor. Er führt ausgehend von der porticus in den Schlafraum (papilio) und erschließt gleichzeitig auch den Vorraum.

Hinsichtlich der Nutzung des Innenareals ist jedoch zu bemerken, dass die ehemaligen canabae nicht wieder aufgebaut wurden, sondern als Nekropole dienten. Dementsprechend wird davon ausgegangen, dass das bis in das 9./10. Jh. n. Chr. genutzte Legionslager zunehmend auch als ziviles Siedlungsareal diente (Gechter 1987: 370372).

Die in der O-praetentura von Carnuntum festgestellte Erneuerung der Kasernen stellt für die tetrarchische Periode 4 nicht die einzige Bautätigkeit dar. Stratigraphisch und anhand des charakteristischen weißen kiesigen Mörtels lassen sich die komplette Neuanlage der Umfassungsmauer mit Zwischenturm, verbunden mit einem langrechteckigen Einbau in den Wall, diesem Zeitabschnitt zuweisen (Abb. 2; Kandler 1979: 338-343).

Mainz, D – Mogontiacum: legio XXII Primigenia (Abb. 5)

Die Existenz von zwei Bauperioden an Umwehrung, Kasernenanlagen und Sondergebäuden in der Wpraetentura und in der retentura erlaubt nach M. Kandler, die jüngeren Mauern und die entsprechenden jüngeren Phasen der innenangebauten Zwischentürme bzw. Torgestaltungen mit dem Ausbau in Periode 4 zu verbinden (Abb. 1). Ebenfalls in diesen Zeitraum dürfte die O- Erweiterung der retentura für eine große Werkstattanlage (fabrica) angelegt worden sein (Kandler 1977a: 627ff. bzw. 645f.).

Das nur aus spärlichen Resten unter der rezenten Stadtverbauung bekannte Legionslager war bereits im 3. Jh. n. Chr. mit dem Mauerring der Zivilsiedlung verbunden worden. Vor oder zu Anfang des 4. Jh.s n. Chr. erhielt das Lager eine neue Umwehrung in Form einer einfachen Steinmauer. Der einzige bisher bekannte Turm an der SWSeite weist zwei Bauperioden auf und gehörte vermutlich zu einer Toranlage (Baatz 1962: 27-30, 73). Die Lagertherme belegt eine vermutlich konstantinische Umbauphase. Eine militärische Nutzung des Lagers in dieser Zeit gilt als gesichert, zumal die durch eine Mauer (3.Jh. n. Chr.) gesicherten canabae teilweise bis in das 4. Jh. n. Chr. bestanden.

Das Legionslager von Carnuntum lässt demnach für den Bauzustand am Ende des 3. Jh.s n. Chr. eine umfassende Neugestaltung des gesamten Lagerareals erkennen, die gemäß dem traditionellen kaiserzeitlichen Lagerbauschema durchgeführt wurde.

Mit dem Untergang seiner Stammlegion in den Kämpfen bei Mursa wurde das Lager Mitte des 4. Jh.s n. Chr. aufgegeben und abgerissen und die Umwehrung der Zivilstadt über das Lagergelände gezogen (Baatz 1962: 78, 87; Petrikovits 1971: 190).

Entspricht dieses Vorgehen der üblichen Praxis? Welche Rückschlüsse bzw. Fragestellungen ergeben sich damit bezüglich der Organisationsstruktur der Legion am Beginn der Spätantike? Wie sind die bestehenden Legionsstandlager im Umfeld der Neuanlagen von Fortifikationen einzuordnen? Lassen sich dabei Veränderungen in ihrem Status und ihrer strategischen Rolle erkennen, bzw. finden diese Ausdruck im baulichen Befund?

Straßburg, F – Argentorate: legio VIII Augusta (Abb. 4) Mit der Zerstörung der Zivilsiedlung im späteren 3. Jh. n. Chr. scheint diese in das Gebiet des Legionslagers verlegt worden zu sein. Unter Julian oder Valentinian wurde die Festung mit einem neuen Mauerring, verstärkt von UTürmen, umgeben. Welchen Veränderungen die Innenbauten des Legionslagers unterzogen wurden, lässt sich aufgrund massiver Umgestaltungen aus der Zeit Konstantins nicht mehr feststellen (Johnson 1983: 142).

Zahlreiche in der Früh- und Mittelkaiserzeit gegründete Legionslager an Rhein und Donau bestanden bis in die Spätantike und könnten somit prinzipiell zu Carnuntum als Vergleich herangezogen werden (Wilkes 2000: 114-116). Bei genauerer Betrachtung des archäologischen Befundes im Einzelnen sind allerdings über diese Ausbauphasen mangels Differenzierung kaum präzisere Aussagen bekannt.

Regensburg, D – Reginum (castra Regina): legio III Italica (Abb. 6) Großflächige, in fast allen untersuchten Bereichen des Lagers festgestellte, Verwüstungen bzw. Brandschichten in der Mitte oder 2. Hälfte der siebziger Jahre des 3. Jh.s n. Chr. werden geläufig mit den Alamanneneinfällen verbunden (Fischer 1990: 32; Mackensen 1999: 218).

Bonn, BRD - Bonna: legio I Minerva (Abb. 3) Während die Nachbargarnison Xanten - Vetera II nach den 606

Raimund Kastler: Legionslager an der Wende zur Spätantike Die Reparaturen an der Umfassungsmauer, die Schleifung des Erdwalles und die Umwandlung dieser Flächen in Siedlungsareal lassen sich derzeit innerhalb der Spätantike nicht fixieren (Aumüller 2000: 12; Wintergerst 2000: 1920).

stand, muss ebenso fraglich bleiben, wie die von H. Vetters geäußerte Vermutung eines in der NO-Ecke zu ergänzenden Rückzugskastells des 5. Jh.s n. Chr. (Genser 1986: 153). Wien, A– Vindobona: legio X Gemina (Abb. 8)

Gesichert der ersten von zwei spätantiken Bauperioden zuzuweisen ist hingegen der ohne größere Veränderungen und teilweise auf den alten Fundamenten erfolgte Wiederaufbau der Mannschaftskasernen unter dem Niedermünster d. h. in der Nordostecke des Lagers. Erst in einer jüngeren, vor dem 5. Jh. n. Chr. anzusetzenden, Phase erfolgte die kleinräumliche Untergliederung der Baracken unter teilweiser Überbauung der Lagergassen mit Trockenmauern und Schlauchheizungen (Konrad 2000: 21).

Auch das Erscheinungsbild der Carnuntiner Schwestergarnison Vindobona ist bislang nur in wenigen Ausschnitten erkennbar (Genser 1986: 435-515; Mosser 1998; 1999). Jüngste Grabungen im Bereich der Kasernen des westlichen latus praetorii lassen für die 2. Hälfte des 3. Jh.s n. Chr. einen großzügigen Wiederaufbau der Kasernen erkennen, wobei die Mannschaftswohnblöcke nach dem gleichen Grundrisstyp mit Korridor und seitlich betretbarem Vorraum (arma) gestaltet wurden wie ihre zeitgleichen Entsprechungen in Periode 4 in Carnuntum (Chmelar & Helgert 1998: 24-26; Eisenmenger & Elefteriadou 2000: 34, Abb. 1).

Der Befund der in der südlichen Lagerecke gelegenen Grasgasse bestätigt einen Wiederaufbau der Lagerbaulichkeiten im späten 3. oder frühen 4. Jh. n. Chr. Grundrisse und Fundmaterial, vor allem die hier im Gegensatz zum Niedermünster fehlenden spätantiken Militaria, lassen nach Ansicht der Ausgräber auf primär zivile Bewohner schließen (Fischer & Riekhoff-Pauli 1982: 52f.).

Die geographische Ausdehnung und der bauliche Umfang dieser spätantiken Lagererneuerung ist beim derzeitigen Stand der Forschung noch nicht absehbar. Einfache Holz- bzw. Trockenmauerbauten kennzeichnen die späteste antike Besiedlung im Lagerareal gegen Ende des 4. Jh.s n. Chr. (Chmelar & Helgert 1998: 24-26).

Wenn auch im Zuge der laufenden Forschungen zur Innenbebauung des Legionslagers von Regensburg Vorsicht bei der Formulierung weitergehender Schlussfolgerungen zu walten hat, kann bereits im späten 3. oder frühen 4. Jh. n. Chr. nicht mehr von einer rein militärischen Nutzung des Lagerareals ausgegangen werden. Weiterhin unbewiesen hingegen bleibt die in der Literatur häufig erwähnte Vermutung eines Reduktionskastells in der NO-Ecke des Legionslagers, das dem am Ende des 4. Jh.s n. Chr. erfolgenden Zuzug elbgermanischer Gruppen im Lagergebiet vorausgeht (Dietz & Fischer 1996: 200-201; Mackensen 1999: 218).1

Szőny, H – Brigetio: legio I Adiutrix (Abb. 9) Die bislang auf dem Gebiet des antiken Brigetio erfassten Baureste von drei verschiedenen Militärlagern wurden in der Forschung unterschiedlich interpretiert. Zuletzt wurden die Anlagen von B. Lőrincz chronologisch und funktionell geordnet, wobei nach S. Soproni dem kaiserzeitlichen Standlager (bezeichnet als Lager II) möglicherweise in valentinianischer Zeit eine neue Befestigungsanlage vorgeschaltet wurde (Soproni 1985: 55-56).

Enns / Lorch, A – Lauriacum: legio II Italica (Abb. 7)

Da die Fläche des Legionslagers erst an der Wende vom 5. zum 6. Jh. n. Chr. als Begräbnisplatz gedient hat, scheint demnach die ältere Anlage bis zur Errichtung des neuen Lagers noch in Verwendung gestanden zu haben. Seitens der Ausgräber wird dabei für das Legionslager eine als konstantinisch bezeichnete spätantike Ausbauperiode angenommen, mit der auch Veränderungen der Umfassungsmauer einhergehen (Barkóczi 1976: 33-35). Während für das mittelkaiserliche Lager die Existenz von Außentürmen ungesichert bleibt, scheinen aufgrund von Vermessungen des 19. Jh.s hufeisenförmige Seitentürme und vielleicht auch Ecktürme vorhanden gewesen zu sein (Visy 1988: 55).

Für das Legionslager Lauriacum stehen primär Befunde von Altgrabungen des 19. Jh.s und der Zwischenkriegszeit zur Verfügung, die drei anhand von Münzbeobachtungen und Nachgrabungen konstruierten Hauptbauperioden nicht detailliert zugewiesen werden können (Petrikovits 1975:157 Taf. 10). Auf die 270/71 n. Chr. durch die verursachten Zerstörungen folgten Juthungen2 Baumaßnahmen unter Aurelian und Probus. Weiter wird mit Baumaßnahmen unter der Regierung des Diokletian und schließlich unter Valentinian gerechnet, wobei dieser Türme und Tore einer umfassenden Erneuerung unterzog. Diese Periode kennzeichnet nach K. Genser wohl die letzte militärisch genutzte Phase (Genser 1986: 151-152). Ob die gesamte Innenfläche dabei noch in militärischer Nutzung

Bedingt durch die industrielle Überbauung des Areals und Kriegsverluste an der Dokumentation sind zum Aussehen der Innenbebauung und damit auch zu Umfang und Gestaltung des Legionslagers am Beginn der Spätantike keinerlei Informationen vorhanden.

1

M.Gschwind (München) verdanke ich den Hinweis auf diese jüngste Zusammenstellung. Skeptisch zu einer Verbindung der Zerstörungen mit bestimmten historischen Ereignissen äußert sich Ubl 1997: 193.

2

Inwieweit sich der für das letzte Drittel des 3. Jh.s n. Chr. 607

Limes XVIII

festgestellte Schrumpfungsprozess der Zivilsiedlung auch auf die militärischen Anlagen auswirkte, muss daher offen bleiben (Barkóczi 1961: 113-114). Noch in der Notitia Dignitatum (Oc. XXXIII.51) wird der Ort als Standort der legio I Adiutrix genannt.

Belgrad, YU (Serbien) - Singidunum: legio IIII Flavia Das Areal des unter Trajan angelegten 18ha großen Legionslagers von Singidunum wird von der neuzeitlichen Festung und teilweise von Parkanlagen in der Oberstadt von Belgrad bedeckt. Die bislang freigelegten Teile der mit Ecktürmen und nach außen vorspringenden Zwischentürmen versehenen Umfassungsmauer werden der Gründungsphase des Lagers zugeschrieben. Nach außen vorspringende rechteckige Tortürme an der porta decumana lassen jedoch auf eine weitere Bauperiode schließen. Informationen zur Innenbebauung des Lagers sind bislang nur aus einzelnen Sondagen bekannt. Eine in den bisherigen Befunden deutlich erkennbare Zäsur der Bebauung nach dem 3. Jh. n. Chr. wird mit einer unter Konstantin I, wohl im Zuge der Sarmatenkriege um 328 n. Chr. erfolgten Aufgabe des Lagers in Verbindung gebracht. Inwieweit diese vollständig vollzogen wurde, ist angesichts von Ziegelstempeln mit der Nennung der pars superior der Legion aus Kontexten der Mitte des 4. Jh. n. Chr. zu diskutieren.

Budapest, H – Aquincum: legio II Adiutrix3 (Abb. 10) Von den Barbareneinfällen der Jahre 259/60 n. Chr. im Gegensatz zu den canabae verschont, weist das 24ha große Legionslager dennoch zahlreiche Reparaturen und Umbauten aus der 2. Hälfte des 3. Jh.s n. Chr. auf (Madarassy 1999: 645). Die Innenbebauung, das Kommandogebäude (principia), die Häuser der Tribunen (scammnum tribunorum), die Lagerthermen und Straßen des im 2. Jh. n. Chr. errichteten Steinlagers wurden nach dem traditionellen Schema erneuert bzw. Erweitert (Szirmai 1986: 427; Kocsis 1990: 712; Póczy 1990: 690693). Nach Befunden der praetentura wurden auch die Mannschaftsbaracken erneuert (Szirmai 1991: 260-261). An der W-Front wurde der Erdwall des Wehrganges beseitigt, und in diesem Bereich zwei neue Magazinbauten oder fabricae angelegt.4 Die westlichen Lagerumwehrung wurde durch einen neuen U-förmigen Graben gesichert und das münzdatiere probuszeitliche oder spätere Lagerhaupttor (porta praetoria) errichtet (Póczy 1990: 690-693).

Eine militärische Wiederbesetzung des Lagers durch seine alte Stammeinheit gegen Ende des 4. Jh.s n. Chr. wird mit im Schema und Orientierung abweichenden Bauten über den alten Baracken in Verbindung gebracht (Boiović 1996: 66-68). Aussehen und Ausdehnung dieser Wiederbesetzung lassen sich bislang nicht näher erschließen.

Die bisherigen Belege sprechen demnach für eine intakte herkömmliche Lagerbebauung aus der 2. Hälfte des 3. Jh.s n. Chr. Dieses Lager wurde in konstantinischer Zeit durch eine nur wenig kleinere Festung von 700m x 350m Ausmaß ersetzt, wobei die ursprüngliche Prätorialfront des kaiserzeitlichen Standlagers wiederverwendet wurde. Die übrigen Seiten des bis in das frühe 5. Jh.n. Chr. genutzten neuen Lagers wurden nach moderner Fortifikationstechnik mit bis zu 3m dicken Mauer und hufeisenförmigen Türmen befestigt (Soproni 1985: 71).

Požarevac / Kostolac YU (Serbien) – Viminacium: legio VII Claudia Das östlich direkt an die unter Hadrian zum municipium erhobene Zivilstadt anschließende Legionslager ist bislang nur in groben Zügen bekannt. Den Ausgrabungen an der Wende des 19./20. Jh.s zufolge lassen sich auf dem Gebiet des Legionslagers auf die beiden Bauperioden bis zum frühen 3. Jh. n. Chr. folgende deutlich spätantike Verbauungen in Trockenmauertechnik und mit provisorischem Charakter orten (Popović 1967: 51-52). Die Notitia Dignitatum (Or. XLI.16.31) und spätantike Ziegelstempel mit dem Signet der legio Viminaciensis weisen jedoch auf eine andauernde militärische Präsenz hin (Popović 1967: 52; Dietz 1993: 293-294).

Wie die Befunde im Haus des tribunus laticlavius belegen, waren bereits Anfang des 4. Jh.s n. Chr. Teile des ehemaligen Legionslagers zerstört oder nur mehr in Teilen bewohnt. Während die retentura zum größten Teil aufgelassen und um die Mitte des 4. Jh.s n. Chr. als Gräberfeld diente, wurden andere Bauten wie die Thermen weitergenutzt, oder neue zivile Bauten als Ersatz für die im späten 3. bzw. im frühen 4. Jh. n. Chr. sukzessive aufgegebene Lagerstadt errichtet (Soproni 1985: 71; Póczy 1986: 401-402; 1990: 697; Kocsis 1990: 712; Madarassy 1999: 645).

Über das Aussehen und den Umfang des spätantiken Legionslagers lassen sich anhand der derzeitigen Befunde keine Aussagen gewinnen. Archar, BG - Ratiaria: legio XIII Gemina Diese bedeutende Stadt an der unteren Donau erhielt unter Aurelian die legio XIII Gemina als Garnison zugewiesen. Die Lage des Legionslagers ist bislang nicht gesichert. Die festgestellten Befestigungsmauern werden mit der Zivilstadt in Verbindung gebracht (Atanassova & Georgieva 1986: 438; R. Ivanov 1996: 162-163).

3

Die neuesten Ergebnisse zum Legionslager (Németh 1994: 139-152) war waren mir leider nicht zugänglich. 4 Falls diese Anlagen nicht bereits früher erfolgten. Nach Pócsy 1986: 400 wird diese bauliche Veränderung im Gegensatz dazu bereits für die Zeit vor der Mitte des 3. Jh.s n. Chr. angesetzt.

608

Raimund Kastler: Legionslager an der Wende zur Spätantike Gigen, BG - Oescus: legio V Macedonica (Abb. 11)

Festung kleineren Umfangs errichtet. Das ehemalige Lager, welches noch Bautätigkeiten im 5./6. Jh. n. Chr. an Außenmauern und Innenbebauung aufweist, wurde hingegen für zivile Siedlungszwecke genutzt (Donevski 1990a: 932-934; 1990b: 243-244; R. Ivanov 1997: 589590, Abb. 46).

Mit der Aufgabe der drei dakischen Provinzen 271 n. Chr. unter Kaiser Aurelian wurde die legio V Macedonia von ihrem Lager in Potaissa wieder in ihre angestammte Garnison in Oescus zurückgeführt. Während unterhalb der Zivilstadt Oescus I nun Reste des frühkaiserzeitlichen Legionslagers lokalisiert wurden, ist die Lage der aurelianische Garnison noch nicht exakt bekannt. Zwischen dem letzten Viertel des 3. und dem Anfang des 4. Jh.s n. Chr. lässt sich jedoch die 10ha große Ost Erweiterung der Stadt Oescus II datieren. Errichtet über den Resten der von den Goten in der Mitte des 3. Jh.s n. Chr. zerstörten Wohnanlagen schließt die Befestigung als eigener Bereich an die früheren Stadtmauern an. Die neuen Befestigungswerke sind von runden Ecktürmen und hufeisenförmigen Zwischentürmen gesichert, wobei letztere auch die drei Tore flankierten (T. Ivanov 1990: 917-921; R. Ivanov 1996: 167; 1997: 548-554).

Von den Befestigungsanlagen des dem Gelände angepassten spätantiken castrums sind bislang nur Teile der augenscheinlich polygonal geführten Außenmauer mit einem Eckturm und eine von Rechtecktürmen flankierte Eingangspforte bekannt (Donevski 1990b: 218, Abb. 2,3). Auswertung Trotz der, wie die Zusammenstellung erkennen lässt, eher dünnen Materialbasis zum Schicksal der kaiserzeitlichen Legionslager in der Spätantike seien einige Thesen und Beobachtungen formuliert.

Da die Innenfläche von Oescus II von modernen Wohnbauten bedeckt ist, sind keine näheren Angaben zur Struktur des spätantiken Lagers der legio V Macedonia zugänglich.

Hinsichtlich ihrer Befestigungsanlagen lassen die in der Spätantike genutzten mittelkaiserzeitlichen Standlager kaum Adaptionen an den zeitgemäßen Stand der Fortifikationstechnik5, wie er sich unter anderem an tetrarchischen Lagerneubauten, wie etwa Portchester (Abb. 15; Lander 1984: 193-194, Abb. 187), Caelius Mons/Kellmünz (Mackensen 1999: 207-213, Abb. 7.5-9) Pilismarot (Soproni 1985: 62, Abb. 35), Troesmis (Abb. 18; Lander 1984: 218-221, Abb. 229) oder El Lejjun (Abb. 19) und Udruh, Parker (2000: 128-133, Abb. 10.7; 10.12), offenbart, erkennen.

Steklen, BG – Novae: legio I Italica (Abb. 12) Das über dem Holz-Erde-Lager der legio VII Augustana gegen Ende des 1. Jh.s n. Chr. errichtet Steinlager der legio I Italica wurde im Zuge der Goteneinfälle der 2. Hälfte des 3. Jh.s n. Chr. gründlich zerstört. Die Wiederherstellung des Lagers begann Ende des 3. bzw. Anfang des 4. Jh.s n. Chr. Das Areal des Lagers wurde mit einer mehr als 3m dicken Wehrmauer neu befestigt, die Tore wurden moderner fortifiziert, und das gesamte Areal wurde nach O erweitert (Donevski 1996: 203). Die neue Befestigungsanlage erhielt einen gemischten militärischen und zivilen Charakter, wobei die Funktionen der noch erhaltenen Militärgebäude verändert wurden (Miltscheva & Gentscheva 1996: 191-192). Ein eigener Militärkomplex im Sinne einer Garnison lässt sich bislang nicht erkennen. Unter Konstantin II. wurden U-förmig nach außen vorspringende Türme in die Befestigungsmauern eingesetzt (R. Ivanov 1996: 168, Abb. 10; 1997: 560ff., Abb. 27-28).

Während in Noricum und Pannonien die weiter existierenden mittelkaiserzeitlichen Hilfstruppenlager mit U- oder fächerförmigen Turmanlagen modifiziert wurden, scheinen derartige Modernisierungen bei Legionslagern tetrarchischer Zeit kaum bzw. nur in Ausnahmen durchgeführt worden zu sein (Pietsch 2000: 365ff.).6 Die anhand des Geländereliefs erschlossenen und nicht durch Grabung verifizierten hufeisenförmigen Eck- und Zwischentürme für das Legionslager Brigetio (Abb. 9) lassen sich innerhalb der Spätantike nicht genauer datieren, wären jedoch noch vor Anlage des späten castrums anzusetzen (Visy 1988: 55). Nicht bestätigt werden konnte die Annahme zeitgemäßer Eckturmformen für Aquincum. Pietsch (2000) Hier bringt lediglich die ebenfalls in die Zeit nach 260 n.Chr. datierte Umgestaltung der porta praetoria zeitgemäßere Bezüge in die Umwehrungsmauern des Lagers ein.7

Silistra, BG - Durostorum legio XI Claudia (Abb. 13, 14) Von dem heute von der modernen Stadt Silistra bedeckten Legionslager wurden 1972 Teile der SW-retentura, der Umfassungsmauern und zweier Kasernen freigelegt (Donevski 1990b: 237, Abb. 1; R. Ivanov 1997: 587-589, Abb. 44-45). Das ursprüngliche Standlager aus der 1. Hälfte des 2. Jh.s n. Chr. wurde um die Mitte des 3. Jh.s, vielleicht im Zuge der Goteneinfälle, stark beschädigt. Der nachfolgenden Reparaturphase, kann die Errichtung eines Außenturmes an der SW-Lagerecke zugewiesen werden (R. Ivanov 1996: 170). Ende des 3. bzw. zu Beginn des 4. Jh.s wurde am Ufer der Donau eine neue spätantike

5

Zur tetrarchischen und späteren Befestigungsarchitektur allgemein vgl. vor allem v. Petrikovits 1971: 182ff.; Johnson 1983: 139ff.; Lander 1984: 151ff.; für Rätien: Mackensen 1999: 199ff.: zu Noricum und Pannonien Pietsch 2000: 365ff. 6 Ähnliches scheint mit Rundtürmen gelegentlich auch bei niederrheinischen Kastellen erfolgt zu sein vgl. Thoen & Vermeulen 1998: 1-12 bes. 3f. Abb. 3 b.c 7 Siehe Póczy 1990: 690f. – Dagegen tritt Wellner 1980: 349ff. für eine Datierung der Anlage in das 4. Jh. n. Chr. ein.

609

Limes XVIII

Das konservative Festhalten an tradierten Befestigungsformen gilt auch für Neubauten von Umfassungsmauern der seit der mittleren Kaiserzeit bestehenden Lager von Mainz, Bonn und Carnuntum.

Legionslager im Schwergewicht eines aktuellen Bedrohungsbildes, wie an der unteren Donau, werden dem aktuellen Standard entsprechend aufgerüstet (Brigetio ??) bzw. neu gebaut (Aquincum und Oescus). Gleiches scheint für einige Legionslager zu gelten, die, wie Strassburg, Novae oder auch Oescus, verstärkt als castrum bzw. befestigte Zitadelle für die Zivilsbevölkerung und vor allem auch als Verwaltungssitz dienten.

Als Zugeständnis an die modernere Verteidigungsarchitektur bei der spätantiken Umfassungsmauer des Legionslagers Bonn (Abb. 3) ist lediglich der Verzicht auf einen aus Erde aufgeschütteten Wehrgang zu bezeichnen. Statt dessen trug die Mauer einen auf Strebepfeilern gestützten Umgang (Gechter 1980: 531f., Abb. 33.1-2). Die Form der Mauer mit innen angesetzten Zwischentürmen und Torbauten in Form einfacher Rechtecktürme stimmt mit den Bauformen der Lagermauer von Periode 4 in Carnuntum überein und entspricht traditionelleren Lagerbefestigungsweisen, wie sie u. a. auch das spätestens um 200 n. Chr. errichtete Legionslager von Albano erkennen lässt (Benario 1972: 258).

Demgegenüber scheinen die übrigen in der Kaiserzeit gegründeten Legionslager auf den Status militärischer Anlagen zweiten Ranges herabgesunken zu sein. Hinsichtlich der Innenbebauung von Legionslagern an der Wende zur Spätantike sind hier Veränderungen der militärischen Organisationseinheit Legion als solche anzusprechen, da sie wohl auch Auswirkungen auf die entsprechenden Baulichkeiten besaßen. So betont Vegetius (III.8), dass die Größe eines Lagers stets proportional zur Zahl der darin garnisonierten Truppen sein sollte.

Eher vergleichbar mit der von nach außen vorspringenden U-förmigen Zwischentürmen und fächerförmigen Ecktürmen verstärkten Befestigungsmauer der tetrarchischen Legionslager El Lejjun (Abb. 16) oder Troesmis (Abb. 17), ist die neu errichtete Anlage von Oescus II (Abb. 11). Ihre Entstehung lässt sich nicht näher zwischen dem späten 3. und frühen 4. Jh. n.Chr. festlegen (R. Ivanov 1996: 167; 1997: 553).

Unter Diokletian lässt sich eine drastische Erhöhung bzw. Verdoppelung der Legionenzahl (bzw. der allgemeinen Truppenanzahl) feststellen (Coello 1996: 14f.; Southern & Dixon 1996: 31). Die Standlager dieser Legionen weisen auf wesentlich geringere Truppenstärken gegenüber den kaiserzeitlichen hin. So umfasst das Lager der unter Diokletian ausgehobenen legio II Herculia in Troesmis (Iglitsa, Rumänien) (vgl. Abb. 17) nur mehr eine Fläche von 2.8ha, das ist weniger als ein Siebtel eines kaiserzeitlichen Lagers. Mit 4.6ha ist das um 300 n. Chr. errichtete Standquartier der legio IV Martia in El Lejjun (Jordanien, vgl. Abb. 16) etwas größer (Coello 1996: 32, 59-61; Southern & Dixon 1996: 31-32; Hodgson 1999: 549; Tomlin 2000: 169-172).9

Wenn auch nach außen vorspringende Türme häufiger bei Um- (Strassburg) oder Neubauten (Aquincum) von Legionslagern späterer Zeit (konstantinisch und danach) anzutreffen sind, scheint es sich dabei weder um ein zeitlich später einzusetzendes Phänomen, noch, wie entsprechende tetrarchiche Neu- und Umbauten von Kastellen im Umfeld der Legionslager belegen, um regionale Vorlieben zu handeln. 8 Viel eher scheint sich in der Altertümlichkeit zahlreicher Legionslagerbefestigungen gegenüber den in den gleichen Provinzen befindlichen Kastell- oder Burgusbauten deutlich eine geänderte bzw. verringerte strategische Bedeutung der traditionellen Legionsstandorte und möglicherweise auch eine geänderte Aufgabenstellung widerzuspiegeln. Dass Veränderungen der strategischen Lagen nicht unbedingt immer zur Verlagerung eines Legionsstandlagers führten, belegt das bis in die Spätantike bestehende Legionslager Regensburg, dessen Positionierung sich für die Krisen des 3. Jh.s als deutliche Fehlplanung erwies. Fischer (1990: 28) Auflösungen und Neugründungen von Legionsstandorten stellen, abgesehen von Truppenneugründungen bzw. im Zuge von Provinzaufgaben (Dakien), eher die Ausnahmen im 3. Jh. n. Chr. dar (Wilkes 2000: 108; 111).

Reduzierte Truppenstärken lassen sich auch in zahlreichen Hilfstruppenlagern anhand des archäologischen Befundes erschließen. So wird bereits in nachaurelianischer Zeit für Gerulata eine Verkleinerung des Lagerareals erkennbar, um 300 n. Chr. in Eining das Binnenkastell errichtet. Vergleichbare Verkleinerungen lassen sich in konstantinischer Zeit für Arrabona erkennen (Mackensen 1994: 189-190; Jilek 1999: 126; Varsik 1996: 28). Diese reduzierten Mannschaftsstärken betrafen nicht nur Neugründungen von Einheiten, sondern, wie u. a. die vom 2.-5. Jh. n. Chr. in Eining stationierte cohors III Britannorum belegt, auch Traditionsverbände. Dementsprechend wird die Erhöhung der Legionenzahl nicht ausschließlich durch Neuaufstellungen sondern auch

8

Vgl. zu den Bauformen die Verweise oben Zur tetrarchischen und späteren Befestigungsarchitektur allgemein vgl. vor allem v. Petrikovits 1971: 182ff.; Johnson 1983: 139ff.; Lander 1984: 151ff.; für Rätien: Mackensen 1999: 199ff.: zu Noricum und Pannonien Pietsch 2000: 365ff. - Eher regional gebunden sind die sogenannten Fächertürme: Pietsch 2000: 365-371.

9

Ob das mit 10ha im Vergleich zum annähernd zeitgleichen Legionslager Lauriacum nur halb so große Standlager der legio II Parthica schon auf reduzierte Mannschaftsstärken bei Legionsneugründungen unter Septimius Severus hinweist, muss offen bleiben. vgl. Benario 1972: 258259.

610

Raimund Kastler: Legionslager an der Wende zur Spätantike durch Aufgliederungen älterer Verbände erklärt (Coello 1996: 59-60). Diese Teilungen älterer taktischer Verbände stellen dabei weniger das Ergebnis einer gezielten Reform dar. Sie sind vielmehr als Produkt einer bereits früher, besonders jedoch ab der Severerzeit, einsetzenden Entwicklung zum verstärkt selbständigen Einsatz und zur Stationierung von Legionsdetachments zu verstehen, die schließlich zu neuen, als Legion bezeichneten Einheiten mit geringerer Truppenstärke führen (Dietz 1993: 302-303; Tomlin 2000: 162-165).

Existierten hier so stark divergierende Truppengrößen oder besaßen die Lager nun zusätzliche u. U. auch nichtmilitärische Nutzungen? Ein bereits im ausgehenden 3. Jh. n. Chr. üblicher und im 4. Jh. n. Chr. weiter verbreiteter Zustand scheint das Verschmelzen von Zivilsiedlungen und tradierten Legionslagern zu Festungsstädten gewesen zu sein. Während in Argentorate vielleicht die Anlage von der zum comitatus12 versetzten Truppe übernommen wurde, werden in Novae oder Oescus Zivil- und Militärbereich in Ausbauten integriert. Der militärische Bereich scheint dabei nicht immer deutlich abgetrennt worden zu sein, wie in Oescus (Abb. 11) durch die Beibehaltung der älteren Mauer und des Grabens.

Grundsätzlich scheint man zunächst bei der Auswahl und Zusammenstellung von Detachments (vexilationes) versucht zu haben, das organisatorische Gefüge der Stammtruppenkörper möglichst wenig anzutasten, mit der Zeit jedoch sank der Stammverband mehr und mehr auf den Mannschaftsbestand der vexillationes herab (Dietz 1993: 300-301, 302-303; Southern & Dixon 1996: 32).

Trotz der unsicheren Befundlage scheinen auch im Falle von Reginum (Abb. 6) bereits Ende des 3. bzw. Anfang des 4. Jh.s n. Chr. zivile Bewohner für Teile des Lagers nachgewiesen zu sein. Damit würde sich eine reduzierte militärische Nutzung der Anlage abzeichnen, die schließlich zu dem in der Notitia Dignitatum dokumentierten Abzug des Restsegmentes der ursprünglichen Stammlegion aus der nun Castra Regina benannten Zitadelle führt (Dietz & Fischer 1996: 206; Mackensen 1999: 218).

Die Zersplitterung besonders der Grenzlegionen nimmt, wie die gegen Ende des 4. Jh. n. Chr. zusammengestellte Notitia Dignitatum erkennen lässt, noch weiter zu, so dass z. B. die legio V Macedonica an sechs Standorten auftritt und noch zusätzliche Einheiten im comitatus besitzt.10 Die für Bonna, Mogontiacum, Reginum, Carnuntum, Brigetio, Aquincum und Durostorum gesicherten Restaurierung bzw. Neubauten der Lagerumwehrungen im späteren 3. Jh. n. Chr. erfolgen in vollem Umfang der kaiserzeitlichen Lagerfläche. Dies würde die Feststellung von K. Dietz unterstreichen, dass auch um 300 n.Chr. noch mit einer rechnerischen Sollstärke der Legionen von 6000 Mann kalkuliert worden ist (Dietz 1993: 301 Anm. 87). Die Realität dürfte hingegen häufig eine andere gewesen sein.

Der fehlende Wiederaufbau der canabae von Bonna (Gechter 1987: 370; Andrikopoulou-Strack & Klaus 1988: 60f.) im späten 3. Jh. n. Chr. bzw. der Schrumpfungsprozess der Zivilsiedlung um das Lager Brigetio bis zur Mitte des 4. Jh.s n. Chr. könnte auf einen ähnlichen Prozess schließen lassen (Barkóczi 1961: 113114).

Wenn auch die in der Notitia Dignitatum vorhandene Aufgliederung der Grenzlegionen sich nur bedingt auf das späte 3. Jh. n. Chr. anwenden lässt, so zeigen literarische Quellen oder auch der am Beginn des 4. Jh.s n. Chr. erfolgte Neubau des Lagers in Durostorum (Abb. 14: Donevski 1990b: 244), dass auch mit deutlich kleineren Mannschaftsstärken für die Legionen gerechnet wurde (Coello 1996: 32; Tomlin 2000: 169).11 Demgegenüber erweist sich das konstantinische Lager von Aquincum (Abb. 10) jedoch nur als unwesentlich kleiner als sein kaiserzeitlicher Vorläufer.

Ob die Aufnahme der Zivilbevölkerung bei der Neuanlage dieser Lager bereits einkalkuliert wurde, erscheint jedoch angesichts der für Britannien erzielten Ergebnisse eher unwahrscheinlich (Bidwell 1991: 11ff.). Ein Ausbau der Lager für eine vollständige militärische Nutzung, ähnlich wie sie für Aquincum festzustellen ist, dürfte den Planungen und Neubauten auch in Bonna und Reginum (wohl auch Brigetio ?) zugrunde liegen. Eine Vermischung zwischen Militär- und Zivilbereich lässt sich in Aquincum erst parallel zur Errichtung der neuen konstantinischen Befestigung feststellen. Die zuerst im Bereich des ehemaligen Legionslagers untergebrachten Zivilsiedlungen der canabae wurden dabei im Laufe des 4. Jh.s n. Chr. mehr und mehr in das neue castrum verlagert. Daneben existierte in reduziertem Umfang auch das Leben in der Zivilstadt weiter (Nagy 1976: 371, 377-378; Soproni 1985: 71-72).

10

Ein ausführlicher Vergleich zwischen der Truppenstationierung einzelner Provinzen in severischer Zeit und in der Notitia Digniatum findet sich bei Jones 1990: 1438-1445. Für die Notitia Dignitatum sind hier auch die Grenzlegionen mit ihren Standorten und die zum comitatus oder sonst als vexillatio abkommandierten Einheiten aufgelistet. – Ergänzend ist hier noch die auf den Ziegelstempeln beruhende Zusammenstellung von Dietz 1993: 291-297 zu nennen. 11 Auch der Erweiterungsbau Oescus II umfasst 10ha, vgl. T. Ivanov 1996: 184. Dies entspricht in der Größe dem severischen Legionslager von Albano s.o. Benario 1972; Tomlin 2000: 165 nennt die legio V Macedonica zusammen mit Ihren Nachbareinheiten, IIII Flavia, VII Claudia und XIII Gemina unter jenen Truppen, deren Detachments 295 n. Chr. in Ägypten eingesetzt wurden. Die vexillationes der legio V und XIII wurden fest dort stationiert. Hatte dies bereits Auswirkungen auf die Lagerdimensionierung ?

Die Stammtruppe des Carnuntiner Lagers, die legio XIV, ist nach Ausweis der Notitia Dignitatum im späteren 4. Jh., 12 Vgl. Jones 1990: 1443; Tomlin 2000: 160 erwähnt von dieser Einheit erbaute burgi in der Schweiz.

611

Limes XVIII

mit Ausnahme einer zum comitatus abkommandierten Abteilung, auf die Standorte Carnuntum und Arrabona aufgegliedert, jeweils ergänzt durch Flotteneinheiten (Jones 1990: 1442).

militärische Nutzungen sind hier ebenfalls denkbar. Dass die römische Armee Kasernenanlagen flexibel je nach Bedarf gestaltete, belegt u. a. die Baugeschichte des Kastells South Shields (Hodgson 1999: 547). In dem im späteren 3. Jh. n. Chr. aufgegebenen Auxiliarkastell von Carnuntum selbst wurden in der Periode 3 (d. h. zwischen den Steinkastellen I und II) Werkstätten, bzw. Versorgungsbauten im Bereich von Kasernenanlagen errichtet (Kandler 1998: 785).

Der das gesamte Lagerareal umfassende Neubau in Periode 4 am Ende des 3. bzw. Anfang des 4. Jhs n. Chr. lässt in seiner Planung hingegen für diese Zeit noch kaum an eine reduzierte Lagerbesatzung denken (vgl. Abb. 1). Eine Zivilsiedlung innerhalb der Lagermauern ist für Periode 4 in Carnuntum gesichert auszuschließen. V. Gassner und S. Jilek haben in ihrer Zusammenstellung zu Carnuntum in der Periode der Soldatenkaiser verdeutlicht, dass die canabae sicher im ganzen 3. Jh. n. Chr., wohl auch bis zum Erdbeben um die Mitte des 4. Jh.s n. Chr., besiedelt waren. Zwar lassen sich bereits ab der 2. Hälfte des 3. Jh.s n. Chr. Bestattungen in den canabae feststellen, Gräber und Siedlungen dürften jedoch parallel bestanden haben (Gassner & Jilek 1999: 50ff.; 63f.).

Entsprechende Nutzungen von nicht sinngemäß verwendeten Kasernenblöcken im Legionslager von Carnuntum, wie sie nun auch für die spätantiken Kasernen von Vindobona beobachtet wurden, wären durchaus denkbar (Eisenmenger & Elefteriadou 2000: 37). Eine Überprüfung dieser These steht für Carnuntum jedoch noch aus und erweist sich anhand der Dokumentation der Altgrabungen schwierig. Eine Erhöhung der Versorgungstätigkeit im Rahmen des Legionslagers von Carnuntum deutet sich jedoch auch für die in der tetrarchischen Periode 4 vorgenommene Erweiterung in der Ost-retentura an, die für eine große fabrica angelegt wurde. Die Errichtung zusätzlicher Versorgungseinrichtungen, unmittelbar anschließend an die Umwehrungsmauer, lassen auch die Lager von Bonn und Aquincum an Ende des 3. Jh. erkennen (Gechter 1980: 537 Abb. 33.6; Póczy 1990: 690). Ob in diesen Tendenzen bereits die in der späteren Notita Dignitatum Oc. IX.19-21 für Lauriacum, Carnuntum und Aquincum belegten Schildmanufakturen zu erkennen sind, oder ob noch für das späte 3. Jh. n. Chr. einige der traditionellen Legionslager auch als Versorgungsbasen für die Verbände der mobilen Eingreiftruppe (comitatus) fungierten, kann erst weitere Forschung erhellen.14

In der vermutlich im 3. Jh. n. Chr. massiv ummauerten Zivilstadt lassen zahlreiche Ausbauphasen ein ungebrochenes Florieren bis in das spätere 4. Jh. n. Chr. feststellen, Siedlungsspuren reichen sogar noch bis in das nachfolgende Jahrhundert (Gassner & Jilek 1999: 63; Ertel, Gassner, Jilek & Stiglitz 1999: 128).13 Kinderbestattungen, die nach P. Bidwell als bester Hinweis für zivile Ansiedlungen in Militäranlagen gewertet werden können, lassen sich im Lagerbereich erst in den von M. Kandler ausführlich beschriebenen Perioden 5 und 6, d. h. gegen Ende des 4. Jh. n. Chr. und später, feststellen (Bidwell 1991: 12ff.; Kandler 1981: 87, Abb.7). Zu dieser Zeit sind die canabae bereits aufgegeben, ein Zuzug von Zivilbevölkerung in das Lager ist daher erklärbar. Ähnlich wie für Regensburg oder Lorch angenommen, scheint sich der militärische Bereich, vermutlich auch mit einem burgus versehen sich in der nordöstlichen Lagerecke zu konzentrieren, bevor schließlich das Lager zur Föderatensiedlung wird (Kandler 1977a: 658; 1981: 87-92; Genser 1986: 152; Mackensen 1999: 218).

Betrachtet man kaiserzeitliche Legionslagerstandorte in der Spätantike, so bleiben zusammenfassend mehr offene Fragen als Antworten. Durch Veränderung der Heeresstruktur, Aufbau von mobilen Krisenreaktionskräften und verstärkter Anlage kleinerer Stützpunktnetze, sowie mit der Veränderung geostrategischer Lagen (Truppenkonzentrationen am falschen Ort) verlieren zahlreiche der traditionellen Legionsstandorte ihre herausragende militärische Bedeutung. Im Gegensatz zu früheren Zeiten werden die Legionsstandlager jedoch nicht aufgelassen. Die nach Zerstörungen des 3. Jh. n. Chr. wiedererrichten Anlagen kennzeichnet jedoch ein deutlicher Konservatismus besonders in der fortifikatorischen Gestaltung.

Die jedoch auch in der Neubebauung von Periode 4 Ende des 3./Anfang des 4. Jh. n. Chr. beibehaltene Gliederung des Lagerareals in Kasernenblöcke zeigt an, dass man zumindest in der Planung noch von einer vollständigen Besatzung des Lagers ausgegangen ist. Ob tatsächlich alle Kasernenblöcke in dieser Phase auch dann wirklich als Truppenunterkünfte genutzt wurden, muss hingegen besonders bei den Anlagen an der Decumanfront offen bleiben. Die in den Planunterlagen erkennbaren zahlreichen Umbauten und Untergliederungen lassen sich kaum mit dem mit Periode 4 zu verbindenden Contubernienschema, noch mit anderen Kasernenformen, vereinbaren (Kandler 1977a: 650-653). Andere

Die bei Neuaufstellungen im späteren 3. Jh. n. Chr. deutliche Verringerung der Truppensollstärken von Legionen findet bei den Lagerbauten von „Traditionseinheiten” zunächst keinen Widerhall. In der Planung der Innenbebauung wird unter Beibehaltung der herkömmlichen Lagergliederung von der traditionellen

13 Die Gleichzeitigkeit von Siedlung und Bestattung wird unter Hinweis auf das fehlende Stadtrecht der canabae erklärt – Humer & Rauchenwald 1993: 163 münzdatierter valentininianischer Bodenhorizont.

14

612

Zu diesem Problemkreis für das 4. Jh. vgl. Borhy 1996: 207-224.

Raimund Kastler: Legionslager an der Wende zur Spätantike Sollstärke der Legionen ausgegangen.

Archaeological Reports Int. Ser. 245. (Oxford). Dietz K. H. 1993 Cohortaes, ripae, pedaturae. Zur Entwicklung der Grenzlegionen in der Spätantike. In K.H. Dietz, D. Hennig & H. Kaletsch (Hrsg.) Klassisches Altertum, Spätantike und frühes Christentum. Adolf Lippold zum 65. Geburtstag gewidmet (Würzburg): 279-329. Dietz K. & Fischer T. 1996 Die Römer in Regensburg. (Regensburg). Donevski P. 1990a Durostorum. Lager und Canabae der Legio XI Claudia. In H. Vetters & M. Kandler (Hrsg.) Akten des 14. internationalen Limeskongresses 1986 in Carnuntum. Der Römische Limes in Österreich Heft 36/2, Österreichische Akademie der Wissenschaften. (Wien): 931-939. Donevski P. 1990b Zur Topographie von Durostorum. Germania 68: 236-245. Donevski P. 1996 Some aspects of defensive system of the Roman camp Novae (Moesia inferior) in Ist-IIIrd century. In P. Petrović (Hrsg.) Roman limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 201-203. Eisenmenger U. & Elefteriadou E. 2000 Ein neues Schlangengefäß aus dem Legionslager Vindobona. Fundort Wien. Berichte zur Archäologie 3/2000: 34-39. Ertel C., Gassner V., Jilek S. & Stiglitz H. 1999 Untersuchungen zu den Gräberfeldern von Carnuntum, Bd1 Der archäologische Befund. Der Römische Limes in Österreich Heft 40. Ősterreichische Akademie der Wissenschaften. (Wien). Fischer T. 1990 Das Umland des römischen Regensburg Veröffentlichung der Kommission zur archäologischen Erforschung des spätrömischen Rätien. Münchner Beiträge zur Vor und Frühgeschichte 42 (München): 28-34. Fischer Th. & Riekhoff-Pauli S. 1982 Von den Römern zu den Bajuwaren. Stadtarchäologie in Regensburg. Bavaria Antiqua: 44-65. Gassner V. & Jilek S. 1999 Carnuntum zur Zeit der Soldatenkaiser - eine Bestandsaufnahme. In J. Tejral (Hrsg.) Das mitteleuropäische Barbaricum und die Krise des römischen Weltreiches im 3. Jahrhundert. Materialien des IX. Internationalen Symposiums „Grundprobleme der frühgeschichtlichen Entwicklung im nördlichen Mitteldonaugebiet” Kravsko, Dezember 1996. (Brno): 51-70. Gechter M. 1980 Das spätantike Bonner Legionslager. In W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies 1979. Papers presented to the XIIth International Congress of Roman Frontier Studies. British Archaeological Reports Int. Ser. 71 (ii). (Oxford): 531540. Gechter M. 1987 Bonn. In H.G. Horn (Hrsg.) Die Römer in Nordrhein-Westfalen. (Stuttgart): 364-383. Genser K. 1986 Der Österreichische Donaulimes zur Römerzeit. Ein Forschungsbericht. Der Römische Limes in Österreich Heft 33. (Wien). Grünewald M. 1979 Die Gefässkeramik des Legionslagers von Carnuntum (Grabungen 1968-1974). Der Römische Limes in Österreich Heft 29. (Wien). Hodgson N. 1999 The late Roman plan at South Shields and the size and status of units in the late Roman army. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 547-554. Humer F. & Rauchenwald A. 1995 Zivilstadt Carnuntum 1992. Untersuchung eines Wohnhauses im Bereich der „Schlossstrasse”. Carnuntum Jahrbuch (1996): 153ff.

Offen bleibt somit die Frage, ob es hier zumindest Anfangs des 4. Jh. n. Chr. Abstufungen in den Stärken zwischen Traditionseinheiten und Neugründungen gab. Wozu die überzähligen Kasernen planungsgemäß dienen sollten, ob für die temporäre Aufnahme von Einheiten des comitatus vorgesehen oder zur Beherbergung von zusätzliche Versorgungs- und Instandsetzungselementen, ist dabei noch ungeklärt. Die Ansiedlung von Zivilisten in den Militärlagern vollzieht sich von Ort zu Ort chronologisch unterschiedlich und ist meist aus den lokalen Bedingungen zu erklären. Räumliche Trennungen zwischen militärischem Sektor und Zivilsiedlung unterstreichen, dass es sich bei diesen Truppen um reguläres Heer gehandelt hat und nicht um Bauernmilizen oder Föderatenverbände. Literaturverzeichnis Andrikopoulou-Strack N. & Klaus J. 1988 Untersuchungen im Bereich des nördlichen Brückenkopfes der KennedyBrücke in Bonn. Archäologie im Rheinland 1987: 60f. Atanassova-Georgieva J. 1986 Résultat des fouilles de la ville antique de Ratiaria au cours des années 1976-1982. In C. Unz (Hrsg.) Studien zu den Militärgrenzen Roms III. 13. Internationaler Limeskongress Aalen 1983. Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 20. (Stuttgart): 437-440. Aumüller T. 2000 Bauforschung an der Lagerbefestigung und Porta Praetoria. In S. Codreanu-Windauer u. a. (Hrsg.) Römerforschung in Regensburg an der Jahrtausendwende 18.-19.02.2000. (Regensburger Kolloquien zur Archäologie. Zusammenfassung der Vorträge): 11-13. Baatz D. 1962 Mogontiacum. Neue Untersuchungen am Römischen Legionslager in Mainz. Limesforschungen 4. (Berlin). Barkóczi L. 1961 Adatok Brigetio későrómai történetéhez. Folia Archaeologia XIII: 95-115. Barkóczi L. 1976 Brigetio In J. Fitz (Hrsg.) Der römische Limes in Ungarn. Taschenbuch für die Teilnehmer des XI. Internationalen Limeskongresses. Az Istvan Király Múzeum Közleményei Serie A, No. 22. (Székesfehérvár): 32-35. Benario H.W. 1972 Albano and the second Parthian legion. Archaeology 25: 256-263 Bidwell P. 1991 Later Roman barracks in Britain. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 9-15. Boiović D. 1996 Le Camp de la Legion IV Flavia à Singidunum. In P. Petrović (Hrsg.) Roman Limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 53-68. Borhy L. 1996 Non castra sed horrea. Zur Bestimmung einer Funktion spätrömischer Binnenfestungen. Bayerische Vorgeschichtsblätter 61: 207-224. Chmelar W. & Helgert H. 1998 Die römischen Kasernen unter dem Judenplatz. Fundort Wien. Berichte zur Archäologie 1/1998: 20-26. Coello T. 1996 Unit sizes in the late Roman army. British

613

Limes XVIII

Ivanov R. 1996 Der Limes von Dorticum bis Durostorum (1-6Jh.). In P. Petrović (Hrsg.) Roman limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade):161-171. Ivanov R. 1997 Das römische Verteidigungssytem an der unteren Donau. Berichte der Römisch-Germanischen Kommission Mainz: 467-590. Ivanov T. 1990 Das Befestigungssystem der Colonia Ulpia Oescensium. In H. Vetters & M. Kandler (Hrsg.) Akten des 14. internationalen Limeskongresses 1986 in Carnuntum. Der Römische Limes in Österreich Heft 36/2. (Wien): 913-924. Ivanov T. 1996 Das Forum der Colonia Ulpia Oescensium (1980-1987) - Die Architektur. In P. Petrović (Hrsg.) Roman limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 183-186. Jilek S. 1999 The limes in Pannonia Superior 1986-1997. New results of research, excavations and publications. In N. Gudea (Hrsg.) Roman Frontier Studies. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 121-129. Johnson S. 1983 Late Roman fortifications. (London). Jones A.H.M. 1990 The later Roman empire 284-602 (Reprint, Oxford). Kandler M. 1974 Die Ausgrabungen im Legionslager Carnuntum 1968-1973. Eine vorläufige Zusammenfassung. Anzeiger der phil.-hist. Klasse der Österreichischen Akademie der Wissenschaften 111: 27-40. Kandler M. 1977a II. Legionslager und canabae von Carnuntum. In H. Temporini & W. Haase (Hrsg.) Aufstieg und Niedergang der römischen Welt II.6. (Berlin): 626-660. Kandler M. 1977b Zum territorium legionis in Carnuntum. In J.Fitz (Hrsg.) Limes. Akten des XI Internationalen Limeskongresses (Székesfehérvár, 30.8.-6.9.1976). (Budapest): 145-154. Kandler M. 1979 Vorläufiger Bericht über die Grabungen im Legionslager Carnuntum in den Jahren 1974 bis 1977. Anzeiger der phil.-hist. Klasse der Österreichischen Akademie der Wissenschaften 115: 335-351. Kandler M. 1981 Archäologische Beobachtungen zur Baugeschichte des Legionslagers in Carnuntum am Ausgang der Antike. In H. Wolfram & F. Daim (Hrsg.) Die Völker der mittleren und unteren Donau im fünften und sechsten Jahrhundert. Berichte des Symposiums der Kommission für Frühmittelalterforschung. Oktober 1978, Stift Zwettl, Niederösterreich. Österreichische Akademie der Wissenschaften philosophisch – historische Klasse Denkschriften 145 Veröffentlichung der Kommission für Frühmittelalterforschung 4. (Wien): 83-91. Kandler M. 1989 Eine Erdbebenkatastrophe in Carnuntum? Acta Archaeologica Academiae Scientiarum Hungaricae 41: 313-336. Kandler M. 1998 Fundberichte aus Österreich 37: 785. Kerdő K.1990 Forschungen im nördlichen Teil der retentura des Legionslagers des 2.-3.Jh.s von Aquincum in den Jahren 1983-1984. In H. Vetters & M. Kandler M. (Hrsg.) Akten des 14. internationalen Limeskongresses 1986 in Carnuntum. Der Römische Limes in Österreich Heft 36/2. Österreichische Akademie der Wissenschaften. (Wien): 703-707. Kocsis L. 1990 Zur Periodisierung des Hauses des tribunus laticlavius im Legionslager von Aquincum. In H. Vetters & M. Kandler M. (Hrsg.) Akten des 14.

internationalen Limeskongresses 1986 in Carnuntum. Der Römische Limes in Österreich Heft 36/2, Österreichische Akademie der Wissenschaften. (Wien): 709-714. Konrad M. 2000 Die Mannschaftsbaracken unter dem Niedermünster. In S. Codreanu-Windauer u. a. (Hrsg.) Römerforschung in Regensburg an der Jahrtausendwende 18.-19.02.2000. (Regensburger Kolloquien zur Archäologie. Zusammenfassung der Vorträge): 21. Lander J.1984 Roman stone fortifications. Variation and change from the first century AD to the fourth. British Archaeological Reports Int. Ser. 206. (Oxford). Mackensen M. 1994 Die Innenbebauung und der Nordvorbau des spätrömischen Kastells Abusina /Eining. Germania 72: 479-513. Mackensen M. 1999 Late Roman fortifications and building programmes in the province of Raetia: the evidence of recent excavations and some new reflections. In J.D. Creighton & R.J.A. Wilson (edd.) Roman Germany. Studies in cultural interaction. Journal Roman Archaeology Supplementary Series 32: 199-244. Madarassy O. 1999 Canabae legionis II Adiutricis. In N. Gudea (Hrsg.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalǎu): 643-645 Miltscheva A. & Gentscheva E. 1996 Die Architektur des Militärlagers und der frühbyzantinischen Stadt Novae (Erkundungen 1980-1994). In P. Petrović (Hrsg.) Roman Limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 187-193. Mosser M. 1998 Das Legionslager von Vindobona – EDV– gestützte Erfassung alter und neuer Grabungen. Fundort Wien. Berichte zur Archäologie 1: 74-88. Mosser M. 1999 Befunde im Legionslager von Vindobona Teil I: Altgrabungen am Judenplatz und Umgebung. Fundort Wien. Berichte zur Archäologie 2: 48-85. Nagy T. 1976 Aquincum. Stadt und Lager im 4. Jh. Acta Archaeologica Academiae Scientiarum Hungaricae 24: 369-381. Németh M. 1994 Roman Military Camps in Aquincum. In G. Hajnóczi (Hrsg.) La Pannonia e l’Impero Romano. Atti del Convegno Internazionale. Accademia d’Ungheria e l’instituto Austriaco di Cultura (Roma 13-16 gennaio 1994). Annuario dell Accademia d’Ungheria. (Milano): 139-152. Parker S. T. 2000 Roman legionary fortresses in the East. In R.J. Brewer (ed.) Roman fortresses and their legions. Papers in honour of George C. Boon. Occasional papers of the Society of Antiquaries of London No. 20: 121-138. Petrikovits H. v. 1971 Fortifications in the north-western Roman empire from the third to the fifth centuries A.D. Journal of Roman Studies 61: 178-218. Petrikovits H. v. 1975 Die Innenbauten römischer Legionslager während der Prinzipatszeit. Abhandlungen der Rheinisch-Westfälischen Akademie der Wissenschaften 56. (Opladen). Pietsch W. 2000 Spätantike Festungstürme in Mautern. In V. Gassner u. a. Das Kastell Mautern – Favianis. Der Römische Limes in Österreich Heft 33. Österreichische Akademie der Wissenschaften. (Wien): 361-380. Póczy K. 1986 Das Legionslager von Aquincum. Ergebnisse der Ausgrabungen 1973-1983. In H. Unz (Hrsg.) Studien zu den Militärgrenzen Roms III. 13. Internationaler

614

Raimund Kastler: Legionslager an der Wende zur Spätantike Limeskongress Aalen 1983. Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 20. (Stuttgart): 398-403. Póczy K u.a.. 1990 Zur Baugeschichte des Legionslagers von Aquincum zwischen 260 und 320. In H. Vetters & M. Kandler (Hrsg.) Akten des 14. internationalen Limeskongresses 1986 in Carnuntum. Der Römische Limes in Österreich Heft 36/2. Österreichische Akademie der Wissenschaften. (Wien): 689-702. Popović V. 1967 Introduction à la topographie de Viminacium. Starinar 18: 29-53. Schnitzler B. 1996 Cinq siecles de civilisation romaine en Alsace. Les collections du Musee Archeologique, tome 4. (Editions les Musees de la Ville de Strasbourg): 3540. Soproni S. 1985 Die letzten Jahrzehnte des pannonischen Limes. Veröffentlichung der Kommission zur archäologischen Erforschung des spätrömischen Rätien. Münchner Beiträge zur Vor und Frühgeschichte 38: (München). Southern P. & Dixon K.R. 1996 The late Roman army. (London). Stiglitz H. 1997 Anhang 1: Gedanken zur Kastellvermessung. In H. Stiglitz (Hrsg.) Das Auxiliarkastell von Carnuntum 1. Österreichisches Archäologisches Institut Sonderschriften Band 29. (Wien): 136-139. Szirmai K. 1986 Das Straßennetz des Legionslagers von Aquincum im 2. und 3. Jahrhundert. In C. Unz (Hrsg.) Studien zu den Militärgrenzen Roms III. 13. Internationaler Limeskongress Aalen 1983. Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 20. (Stuttgart): 426-428. Szirmai K. 1991 Barrack-blocks in the praetentura of the legionary fortress of Aquincum 1987-1988. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 259262. Thoen H. & Vermeulen F. 1998 Phasen der Germanisierung Flanderns in der mittel- und spätrömischen Zeit. In C. Bridger & K.J. Gilles (Hrsg.) Spätrömische Befestigungsanlagen in den Rheinund Donauprovinzen. Beiträge der Arbeitsgemeinschaft ‘Römische Archäologie’ bei der Tagung des West- und Süddeutschen Verbandes der Altertumsforschung in Kempten 08.06-09.06.1995. British Archaeological Reports Int. Ser. 704. (Oxford). Tomlin R. S. O. 2000 The legions in the late empire. In R. J. Brewer (ed.) Roman fortresses and their legions. Papers in honour of George C. Boon. Occasional papers of the Society of Antiquaries of London No. 20: 159-181. Ubl H. 1997 Enns – Lauriacum. Legionslager – Canabae – Autonome Stadt. In H. Friesinger & F. Krinzinger (Hrsg.) Der Römische Limes in Österreich. Führer zu den archäologischen Denkmälern. (Wien): 187-194. Varsik V. 1996 Archäologische Topographie des antiken Gerulata. In. K. Kuzmanová & J. Rajtár (Hrsg.) Gerulata I. (Archäologisches Institut der slowakischen Akademie der Wissenschaften, Nitra): 7-46. Visy Z. 1988 Der pannonische Limes in Ungarn. (Stuttgart). Wellner I. 1980 Das Legionslager von Aquincum und die vermuteten Principia. Alba Regia 18: 349ff. Wilkes J. J. 2000 Roman legions and their fortresses in the Danube lands. In R.J. Brewer (ed.) Roman fortresses and their legions. Papers in honour of George C. Boon. Occasional papers of the Society of Antiquaries of

London No.20: 101-119. Wintergerst E. 2000 Die Ausgrabungen im Niedermünsterkreuzgang. Ergebnisse zur Römerzeit. In S. CodreanuWindauer u. a. (Hrsg.) Römerforschung in Regensburg an der Jahrtausendwende 18.-19.02.2000. (Regensburger Kolloquien zur Archäologie. Zusammenfassung der Vorträge): 18-20.

615

Limes XVIII

Abb. 1 Legionslager Gesamtplan

Carnuntum,

Abb. 2 Legionslager Carnuntum, Grabungen 1968-77 Befundplan Periode 4

616

Raimund Kastler: Legionslager an der Wende zur Spätantike

Abb. 3 Legionslager Bonna nach Gechter (1980: Abb. 33,5)

Abb. 4 Legionslager Argentorate nach Schnitzler (1996: 36)

617

Limes XVIII

Abb. 5 Legionslager Mogontiacum nach Baatz (1962: Beil. 2).

618

Raimund Kastler: Legionslager an der Wende zur Spätantike

Abb. 6 Legionslager Reginum Mackensen (1999: Abb. 7,13).

nach

Abb. 7 Legionslager Lauriacum nach Petrikovits (1975: Abb. 36).

619

Limes XVIII

Abb. 8 Legionslager Vindobona nach Mosser (1999: Abb. 3).

Abb. 9 Legionslager Brigetio nach Visy (1988: Abb. 41).

620

Raimund Kastler: Legionslager an der Wende zur Spätantike

Abb. 10 Legionslager Aquincum nach Pocsy (1986: Abb. 2).

Abb. 11 Befestigung von Oescus II nach R. Ivanov (1996: Abb. 4).

621

Limes XVIII

Abb. 12 Befestigung von Novae nach R. Ivanov (1966: Abb. 10).

Abb. 13 Legionslager Durostorum nach Donevski (1990b: Abb. 1).

622

Raimund Kastler: Legionslager an der Wende zur Spätantike

Abb. 14 Legionslager Durostorum nach Donevski (1990b: Abb. 2).

623

Limes XVIII

Abb. 15 Kastell Portchester nach Lander (1984: Abb. 187).

Abb. 16 Legionslager El Lejjun nach Parker (2000: Abb. 10,7).

Abb. 17 Legionslager Troesmis nach Lander (1984: Abb. 224).

624

The history and perspectives of the research of the Csörsz Ditch (‘Limes Sarmatiae’)1 Eszter Istvánovits and Valéria Kulcsár

The next step was taken by the team of Pál Patay, Sándor Soproni, and Éva Garam who made a field survey of the entire Hungarian part of the Csörsz, carried out several cross-section excavations at different places and published the results in a monograph (Garam, Patay & Soproni 1983).

The Csörsz Ditch (also known as Ördög - or “The Devil’s Ditch”) is an earthwork spanning a total of 1260kms that enfolds the Great Hungarian Plain. The width of the ditch varies from 3.4 to 8m. Its depth is 1.5 - 3m from the current surface. The surviving parts of the vallum are about 2m high. The system is composed of two, three, or even, at several points, four lines. Each line consists of a fossa and a vallum; the fossa turned consistently to north or east (with some rare exceptions). The starting point of the ditch is the bank of the Danube, almost opposite Aquincum, the capital of Pannonia Inferior. From here it runs almost exactly from east to west and turns to the south below Nyíregyháza. From here it passes the eastern part of the Hungarian Plain, crosses the Banat and ends at the lower Danube opposite Viminacium in Moesia Superior. Today we can see the vallum and fossa only at a few points in eastern Hungary, mostly at places where forests have preserved the earthwork from ploughing. However, a hundred years ago and earlier, the whole system could be traced quite well and thus attracted the attention of both scholars and lay people. A number of folk tales try to explain the origin of the great earthwork.

The main issue regarding the earthwork remains its date. A lot of different opinions on the question have been published; suggestions have included both prehistoric and mediaeval origins. The test excavations conducted by Patay, Soproni and Garam provided some evidence for dating. At the northern part of the Csörsz Ditch (at Mezőkövesd, Jászfényszarú, Kötegyán and Tarnazsadány) several features of Sarmatian settlements of the C2nd and C3rd were found under the vallum, and not far from these sites (at Tarnazsadány) a Sarmatian grave dated between 220-300 was unearthed under the vallum as well. The only datum serving as a terminus ante quem is the period of the Hungarian Conquest, because several graves from the C10th and C11th were found dug into the soil of the vallum (at the village called German). Lacking more concrete data, this wide interval of time can be narrowed only through logical speculation. Most researchers agree the Csörsz Ditch must have been constructed in Roman times, because the territory surrounded by the earthwork can be identified with the Sarmatian barbaricum of the Carpathian Basin. According to Sándor Soproni, the period before 270, that is to say before the surrender of Dacia, can be eliminated from consideration: it seems highly unlikely that the Romans would allow the construction of a defensive system at the borders of a province. From the other point of view, if the Sarmatians had wanted to defend themselves against the Romans, they would not have done so on their eastern border, because Roman attacks generally came from the west (Pannonia) and south (Moesia). The situation changed fundamentally after the surrender of Dacia, when Sarmatia and Rome both found themselves under threat from eastern peoples moving westward. On the basis of historical considerations related to events in the wars between Rome and the Sarmatians and Goths, Soproni placed the terminus ante quem of the Csörsz in 322, in the age of Constantine, who was famous for strengthening the defence of the limes at several points (Soproni 1969).

The first literary mention of the legend of the Csörsz is from the C16th in the so-called Cracovian Chronicle. According to this account, the earthwork was built for the defence of the country before the time of Attila. However, the earliest record on the vallum and fossa near the village of Szihalom comes from 1067 (Balás 1961: 104). In the first quarter of the C18th, the Austrian military engineer Marsigli mapped the line of the Csörsz Ditch. In historical works of the C18th and C19th, the vallum and fossa system was mentioned in many places and it can be well traced on the military maps made by the engineers of Austrian emperor Joseph II at the end of the C18th. In the second half of the C19th, at the dawn of Hungarian archaeology, data on different parts of the Csörsz (and on earthworks in Hungary more generally) were systematically collected by Flóris Rómer, who was among the first Hungarian archaeologists (Balás 1961: 5-11). These surveys were made in different parts of the country (for example, in the territory between the Danube and Tisza, the Banat etc.) before the regulation of the rivers and the beginning of intensive agricultural activity that destroyed most of the earthwork. Thanks to these early surveys, today we have a more or less precise picture of the whole system. In the 1960s Vilmos Balás published his thorough study of the Csörsz Ditch with a detailed history of the research (Balás 1961; 1963).

While agreeing that Csörsz was built under Roman control or support in order to keep northern and eastern Germanic tribes away from the limes, András Mócsy reasoned that these construction works must have followed the wars between the Sarmatians and Diocletian, after the year 294. This is the period when fortified ports were built on the left bank of the Danube, on the barbarian side, to assist Roman military movements in Sarmatia. Mócsy considered these

___________________________________ 1

The article was written with the support of the OTKA (Hungarian Scientific Research Foundation), project N T 032778.

625

Limes XVIII

to be elements of the same concept (Mócsy 1972: 84-86). The weakness of this theory is that such a monumental construction project in barbaricum could be managed only with the agreement and support of the Sarmatians, and this would have been highly improbable given that this period was characterised by constant wars with them (Istvánovits 1998: n. 63).

At another point on the salvage operation for the M3 motorway, several small 30-40cms wide ditches were excavated running south of and parallel to the main ditch. Pottery found in the upper part of the fill of the main ditch has shown that the vallum and fossa must have been renewed and reused sometimes in the C9th-C10th, perhaps by the Slavic population (Fischl 1995).

In recent years construction works on various new motorways in Hungary have brought to light new data on the archaeology of the Csörsz earthwork. During salvage excavations conducted before the construction of the motorway 2/A at Fót, we had a chance to investigate a part of the Csörsz earthwork never before visible on the surface. This is a system consisting of three parallel ditches. The largest one was almost 5m wide; its maximum depth (from the current surface) was 1.8m. A smaller ditch, which disappears entirely at points, runs directly alongside the larger one, cutting into it at some places. Its maximum width was 3.2m and maximum depth 1.6m. The third ditch runs 6m south of the latter one. Its maximum width was 3m and its maximum depth was 1m. In all of them, potsherds from the imperial age were found. These pieces of pottery were very few in number and not easily classifiable; it is difficult to say whether they were Sarmatian or Quadian (Germanic) in origin, and it is impossible to date them precisely.

Summarising the traditional hypotheses on the origin and chronology of the Csörsz Ditch and the results of recent excavations, we have to assume the following: 1.

2.

In an aerial photograph, about 250m of the ditch can be seen east of the excavation. West of the excavated part, which touched only the track of the future road, we tried to trace the ditch by making boreholes in the soil. We also made a test trench 170m from the main excavation. The three ditches could be traced over this entire distance. In the part of the ditch found in the test trench, some scattered human bones were unearthed. This may indicate that the ditch disturbed an earlier grave.2 During the salvage work connected with the M3 motorway at Mezőszemere (-Kismari-fenék), about 150kms east of Fót, the aforementioned site, the Csörsz and its surroundings were revealed using traditional archaeological methods along with magnetometric examination. Here the excavated part of the ditch was 55.5m wide and could be traced for 250m. About 10m south of it, a smaller one 60-70cms in width ran parallel to the main ditch. According to the excavator, it could be the foundation ditch of an inner palisade which had several interruptions assumed to be “entrances” to the palisade. (It is to be noted that traces of palisade had been found earlier between Kál and Kápolna, some not far from the site in question in the course of the test cuts of the 1960s: Garam, Patay & Soproni 1983: 30.) In the direct vicinity of the ditch, a small barbarian military cemetery dating back to the end of the C4th or beginning of the C5th was found. It has been suggested that this could be the cemetery of the guards of the defence system (Domboróczky 1997: 102).

3.

At certain parts of the Csörsz, for example between the rivers Danube and Tisza, it seems obvious that whenever it was built, it consolidated an old border line between the Sarmatians and the Germans. According to recent excavations in the area of the Danube knee, Sarmatians and Quadians had begun jointly inhabiting this region no later than the beginning of the C2nd AD. The construction of such an extensive earthwork of more than 1000kms would have taken many years, perhaps decades. It must have been a long term project, too long perhaps to have been completed under the reign of one emperor (20-25 years in case of Diocletian and Constantine the Great). According to the Soproni, Patay & Garam team, the construction of the 1260kms long vallum and fossa system must have taken 10,000,000 workdays (at 1 cubic metre per day). This means that if we assume 120 workdays (about 5 months), that is to say one year, 83,300 people were needed to erect the system (Garam, Patay & Soproni 1983: 15). Unfortunately we do not have any data for comparison. The existence of the several parallel lines of the earthwork may refer to the moves of the borderline and its corrections, which would also indicate a prolonged period for the full completion of the vallum and fossa system. The several lines of the system, which sometimes cross each other, may be construed to mean that they are not contemporary. For example, it has been suggested that one or more of the earthwork lines in the Banat could have already been constructed in the early empire as the western frontier of Dacia.

Bibliography Balás V. 1961 Az alföldi hosszanti földsáncok. (Longitudinal earthwalls of the Great Hungarian Plain). Régészeti Füzetek Ser. II. No. 9. (Budapest). Balás V. 1963 Die Erdwälle der Ungarischen Tiefebene. Acta Archaeologica Academiae Scientiarum Hungaricae 15: 309–336. Domboróczky L. 1997 Mezőszemere-Kismari-Fenék. Sarmatian settlement and cemetery from the A.D. 4th century. Paths into the past. Rescue excavation on the M3 Motorway. (Budapest): 99-102. Fischl K. 1995 Előzetes jelentés a Csörsz-árok kutatásáról

2

The excavation was conducted by Valéria Kulcsár and the Ukrainian archaeologist Alexandr Simonenko.

626

Eszter Istvánovits and Valéria Kulcsár: The history and perspectives of the research of the Csörsz Ditch Csincsén [Vorbericht über die Forschung des Csörsz Grabens bei Csincse]. Somogy Megyei Múzeumok Közleményei XI: 33-38. Garam E., Patay P. & Soproni S. 1983 Sarmatisches Wallsystem im Karpatenbecken. Régészeti Füzetek Ser. II. No. 23. (Budapest). Istvánovits E. 1998 Szarmaták a Kárpát-medencében. (Sarmatians in the Carpathian Basin) In: Jazigok, roxolánok, alánok. Szarmaták az Alföldön. (Jazygi, Roxolani, Alani. Sarmatians at the Great Hungarian Plain) (Szerk. Havassy Péter). Gyulai Katalógusok 6. (Gyula): 35-48. Mócsy A. 1972 Das letzte Jahrhundert der römischbarbarischen Nachbarschaft im Gebiete des heutigen Ungarn. Cumania I: 83-93. Soproni S. 1969 Limes Sarmatiae. Archaeologiai Értesítő 96: 43-53.

627

Limes XVIII

Fig. 1. The Csörsz (Devil’s) Ditch of the Great Hungarian Plain (after Soproni 1971).

628

Decorated lead sarcophagi in Moesia Superior Snežana Golubović Recent archaeological excavations conducted in the province of Moesia Superior are yielding large amounts of information for a preciser picture of burial practices there in Roman times. Most of the information is coming from the necropolis of Viminacium. In the course of 20 years of systematic research, more than 13,000 graves have been excavated. Compared with other burial modes in Moesia Superior, the lead funerary sarcophagi are proportionally rare. A few more lead sarcophagi, but still chance finds, are known from the rest of Moesia Superior. However, at this stage of research it is still impossible to find out if this form of funeral was a privilege limited to certain social classes. Besides extremely richly decorated examples found in memorial buildings, there are a lot of sarcophagi without any decoration that were placed in simple pits. In the majority of cases these contain the remains of children. The central Balkans were abundant with supplies of lead, so the raw material was easy obtainable. It is interesting that decorated lead sarcophagi are found in the west principally in Britain where supplies of lead were plentiful, in Gaul and in Roman Germany, and, above all, in the east, in Syria and Palestine, where the lead must have been imported. However, this form of art flourished prolifically in Syria and Palestine than elsewhere in the Roman world. The presence of orientals at Viminacium is confirmed by the discovery of three inscriptions there. A lead ossuary found at Slatina and samples from Smederevo and Nis have evident Christian features (as christograms) and are dated from the late C3rd to the late C4th. Although the greatest number of lead sarcophagi, especially ornamented ones, derive from Viminacium, obvious Christian characteristics are not present at those found during long-term systematic excavations there. However, it is well-known from other sources that Viminacium was a bishop’s residence in the C4th. The identity of two of Viminacium’s bishops, Amantinus and Cyriacus, has been confirmed. The reason for such a situation could be that at Viminacium the solar theology, expressed through Mithraism, had a priority unlike in other centres like Singidunum and Nais where the supreme god Jupiter was a dominant one. As in the eastern areas, the lead sarcophagi are almost exclusively rectangular with the ends of equal length with the ornamentation, on a small scale and very repetitive which includes geometrical and linear ornaments. All the motifs employed for the decoration of the lead sarcophagi have in fact a place in the standard iconographic repertory of Roman funerary art, as for instance, Medusa-heads. There are also numerous rosettes, garlands, leaf-sprays and symbols of the fruitful afterlife. Most of the applied ornaments on lead sarcophagi are analogous to those from Syria. The territory of Viminacium, inhabited by numerous veterans and with extraordinary traffic communications for trade between the east and west, represented an ideal ground for the merchants and craftsmen - Syrians, Greeks and Italians who represented the main initiators and innovators of various crafts.

Archaeological excavations and protection work in the territory of Moesia Superior has revealed a number of cemeteries of the Roman period as the base for establishing a typology of burial practices (Jovanović 1984: 10). Compared to other burial rites, those in lead sarcophagi are proportionally rare. The most numerous are the coffins found in the necropolis of Viminacium which has been systematically explored for more than 20 years. From other parts of Moesia Superior there are a few lead coffins found mostly during rescue excavations. Besides very rich decorated ones found in tombs, there are many lead coffins without any decoration, put in a simple grave pit. In most the cases, children were buried children in lead sarcophagi.

sarcophagi of local stone or imported marble. It is interesting that decorated lead sarcophagi are found in the west principally in Britain where supplies of lead were plentiful, in Gaul and in Roman Germany, but above all, in the east, in Syria and Palestine, where the lead must have been imported. However, this form of art became the most valued in Syria and Palestine than anywhere else in the Roman world. The use of this particular funeral mode could be explained by the considerable deposits of lead in Asia Minor or, perhaps, by the special tendency for using the metal by the Near Eastern population (Bertin 1974: 44). The territory of central Balkan had abundant lead supplies, so the raw material was easy obtainable. Continuous archaeological research has obtained much data about early mining in the region of Moesia Superior. During excavations at Municipium DD near the village of Sočanica (Fig. 1) a few metal objects have been found on the periphery of antique settlement. Up until 1920, two or three rectangular structures considered as buildings for melting ore, cut in to caves, were discernable. O. Davies, having found traces of water channels, pointed to structures being involved in the smelting of lead ore. Simultaneously, he recognized the remains of a melting kiln that was built, according to him, of stone and plastered with clay (Čerškov 1970: 55). The abundance of and easily obtained lead supplies is obvious from the detailed metalgenetic map made for the so-called Ridensko-Krepoljinska

Historical sources refer to lead mines and the trade in antiquity. In the Roman period lead was transported all over the ancient world (Boulakia 1972: 143). The mines in Britain had been worked before the Roman conquest. Spanish mines too had been worked for a very long time. Its exceptionally low melting point (3270 C) made possible the extraction of lead from its ore by a rather simple metallurgical operation which enabled the wide-spread use of the metal (Boulakia 1972: 139). Lead served for manufacturing coffins in various parts of the Roman empire mainly from the mid-C2nd to the early C4th AD. Workshops or, perhaps itinerant artisans, served a well-to-do middle class that could afford more than simple wooden coffins, and yet did not aspire to 629

Limes XVIII

district in which was founded the camp and town of Viminacium.

1986: 403, Pl. xxixA). Smederevo (Vincea) (Fig. 1/3)

There is general opinion that lead sarcophagi fabricated in the west came under a Syrian influence. They are not found only in Europe but in North Africa too, especially regions rich with lead ores. Recent archaeological excavations combined with historical sources (Bertin 1974: 44) emphasise the extent internments in lead coffins all over the Roman empire. However to claim that all decoration applied to them was as a result of Near Eastern influences is too general.

Of considerable importance is the accidental discovery of a grave with an encased lead sarcophagus at the site of Ćirilovac (Fig. 3; Cunjak & Marković-Nikolić 1997: 39). The lid is very richly decorated. At the central part is a deltoid space framing a cross. The deltoid is composed into the hasta of the Christ monogram in relief made by using the motif of fir tree branches and covering the entire surface. The iconographic scheme is enriched by three Dioynius masks - two at the upper part of the lid and one at the central part. In addition, on the lid was incised an inscription relating to the person buried in the coffin. The inscription was incised along the edge and is considerably damaged, but it is possible to reconstruct it: Aureli(ae) S(alleliae) F(iliae), C(arissimae). After epigraphic and anthropologic analysis we can say that the burial was that of a 14 year old girl from high-birth who suddenly died and was buried in a sarcophagus originally intended for an adult (Cunjak & Marković-Nikolić 1997: 40).

Beograd (Singidunum) (Fig. 1/1) In Singidunum exploration has found five lead coffins at different locations throughout the city. Any attempt at undertaking systematic archaeological excavations is impossible as we are dealing with a Roman cemetery in the territory of modern Belgrade-city. However, it can be expected that the discovery of new examples of lead coffins from different sites in town will be unavoidable as a consequence of future building work. (Kondić 1960: 29; Todorović, Kondić, & Birtašević 1956: 80; Saria 19241925: 160)

Considering all iconographical analysis of the decoration and the inventory of the grave (fragmented golden necklace, pins with head in the shape of cantharos and silver ring) it could be concluded that the coffin was made at the end of the C3rd or in the first decades of the C4th. Also, the obvious Christian symbol proves the significance of Christianity at this territory (Cunjak & MarkovićNikolić 1997: 43).

Slatina (Fig. 1/2) Though it is not precisely a lead sarcophagus, the very interesting find of a lead urn (Fig. 2) in the village Slatina on the slopes of the Kosmaj mountain, has to be emphasized for further analyses (Pop-Lazić, Jovanović, & Mrkobrad 1992: 135). Half of an already completed lead sarcophagus was used to manufacture the urn. Its dimensions are remarkable (diameter 0.6m and height 0.45m). This implies the possibility of it having been used as a ossuary with two or three urns. The urn is ornamented with linear, geometrical decorations. On its narrower side, framed by the vertical zones of crosshatched cast decoration, are two intersecting cable lines at the intersection of which is a star with arrows at the ends. On the other side is a radial ornament combined with floral details and cable lines (Pop-Lazić, Jovanović & Mrkobrad 1992: 136).

Niš (Naissus) (Fig. 1/4) During archaeological research in antique Naissus (Petrović 1976: 86, 87), a Christian sarcophagus was found during rescue excavations in the late antique cemetery associated with a basilica and martyrium which are dated to the C5th. This early Christian sarcophagus, now in the Museum of Niš, was found in the southern part of the basilica’s nartex. On the lid the cross, with three busts, is modelled at the ends of a longer shank, while only one figure in a long garment is presented on the shorter shanks. The shorter sides of the coffin bears three busts (Fig. 4). By the comparison with the ingot found at Feldioara this coffin is dated to the later half of the C4th, with the three busts representing Gratian I, Valentinian II and Theodosius I (Nikolajević 1989: 2447; Buschhausen & Buschhausen 1991: 52).

The influence of the eastern cultural region is indirectly suggested by the radial-solar motifs and rosettes connected with the cult of Jupiter Heliopolitanus. However, in this case the part of the coffin used as an ossuary contained the remains of a cremated person; this is contrary to oriental burial rites which tended to practice inhumation. Therefore, it is considered that persons buried in this sacral object originated, ethnically, from Dalmatia. This is particularly so as the inhabitants of this province are epigraphically confirmed in the territory of mountain Kosmaj (Pop-Lazić, Jovanović & Mrkobrad: 1992: 138). The decoration, as applied on the example from Slatina, is quite similar to those on a lead tank from Oxborough which could be indicative of the widespread use of ornamental patterns at this time (Frere, Hassall & Tomlin

The Roman empire created exploitation centres in the southern regions of Moesia Superior which were particularly rich in lead ore. One such centre was established in the Ibar region, at the confluence of the Sočanica River into Ibar – Municipium DD (Čerškov 1970, 80). At the end of the C19th the river exposed a group of 50 sarcophagi (Fig. 1/5). The original records suggest that all of them were sealed with lead, suggesting that there were no traces of robbery. Therefore they must have contained well preserved skeletons and inventories of 630

Snežana Golubović: Decorated lead sarcophagi in Moesia Superior grave goods. Unfortunately, not one of those sarcophagi survives today. There is one lead sarcophagus of obscure origin in the National Museum of Belgrade. It has been said that it could be from the site of Ulpiana. (dimensions: length 1.9m, width 0.57m, height 0.5m). Only the longer lateral sides are decorated with ornamentation, with so called “des baguettes de perles et de pirouettes”. This motif is very common, not only on Near Eastern sarcophagi but also in the western parts of the Roman empire. In fact, the variations of the motif are so numerous that researchers from France consider it is possible to distinguish different workshops in Gallia (Santrot & Frugier 1982: 285). The wide circulation of the motif is confirmed also by the fragments of lead coffin from the Roman cemetery at Brigantium (Konrad 1997: 31).

following the usual custom (Zotović 2000: 11). The inventory of grave goods consists of two golden earrings, fragments of a paste pearl and a bronze coin which dated this inhumation to the end of the C2nd or first half of the C3rd (Marcus Aurelius or Septimius Severus). Grave 349 (Fig. 10, 11): The coffin was found at a depth of 1.25m. The length of the lid was 1.5m, its width 0.4m and its height 0.07m. The length of the coffin was 1.45m, 0.35m wide and 0.25m high. The grave was orientated east-west. The lid, was originally fastened to the coffin by the usual lead tongues at the end of the coffin, passing through a slit in the lid and folded over. The long sides and the base were fashioned from a single sheet of lead. The pattern consists of intersecting cables dividing the entire surface of the lid, into triangles, rectangles and trapeziums. The well preserved skeleton of a child about 5 years old was found in this coffin. The usual bronze coin put into child’s mouth could be Constans II (337-361) or Valentinian I (364-371) and so dates this inhumation to the mid-C4th. The other goods consisted of three bone pins, bone sawing pins, two golden earrings, an unidentified wooden object and a bone gambling cube.

Kostolac (Viminacium) (Fig. 1/6) The largest collection of data has been obtained from the cemetery at Viminacium following the systematic archaeological excavation of 13,000 graves. In the course of 500 years, the cemetery served the military camp and canabae, encompassing a chronological span between C1st and C4th (Jovanović 1985: 13-18). All the lead coffins were found in the locality Pećine (Fig. 5). It has been during the last 20 years of excavation that most of the lead coffin burials have come to light. From 13 excavated, 7 are ornamented.

Grave 1221 (Fig. 12): In the 1982 season, an exceptionally damaged lid of a lead sarcophagus was found. It occurred during construction work of the TE Drmno thermo-energy plant. The length of the coffin was 1.5m, by 0.38m by 0.29m. The grave orientation was north-west – south-east. Horizontally laid ceramic tiles covered the coffin. The cover was destroyed during the construction of the plant. Under the coffin, the floor consisted of six horizontally laid bricks. The remains of the deceased were not found. The grave was plundered when the lid was broken. Still the decoration is quite remarkable. The upper compartment contains a representation of Mars Ultor standing to the right. The figure of Mars is in the frame in the four corners of the upper compartments. His right arm is raised. The spear usually held by this arm is missing. In the space between the emblems is a rosette while the sides and lid are ornamented with angels in relief. A representation of Mars Ultor is well-known from a lead sarcophagus from Jerusalem (Rahmani 1988a: 49). Representations of Mars like these are frequently encountered, for instance, on gems and on C2nd and C3rd coins, from both the Roman mint and the city mints in Palestine, and from early C3rd altars like Altar of Faustinianus found at Carnuntum dated to 219 (Obermayr 1967: 197-198), helmets (Garbsch 1978: 73) and other objects connected with the Roman army. In the case of the Viminacium sarcophagus, the figures of Mars in the corners are made with wooden stamps pressed to form the decoration.

Grave 228 (Fig. 6, 7): The coffin was found at a depth of 1m. Unfortunately, the coffin is lost and the measurements are from the documentation: length 1.15m, width 0.33m, height 0.24m. The orientation of the grave pit in which the coffin was laid was north-south. The lid was originally fastened to the coffin by the usual lead tongues at the end of the coffin, passing through a slit in the lid and folded over. The long sides and the base were fashioned from a single sheet of lead. The lid was deformed in its central part under the soil pressure. The ornamentation consisted of intersecting cable lines, forming two joined triangles in the middle of the lid. The well preserved skeleton of a child, together with fragments of linen material and leather shoes were found in this coffin. Grave 339 (Fig. 8, 9): The coffin was found at a depth of 1.3m. The length of its lid was 0.8m, width 0.28m. The length of the coffin was 0.8m, 0.24m wide and 0.2m high. The grave orientation was west-east. The lid was originally fastened to the coffin by the usual lead tongues at the end of the coffin, passing through a slit in the lid and folded over. The long sides and the base were fashioned from a single sheet of lead. The lid is a little deformed in its central part. The bottom of the coffin is in quite bad repair. Also, holes made with a metal stick - the work of a treasure hunter - are visible on the lid. The pattern consists of intersecting cables dividing the entire surface, including the lid, into triangles and lozenges. The poorly preserved skeleton of a child about 12 months old was found in the coffin. Discernible were traces of bronze on the facial bones which point to a bronze coin put in the mouth

Grave 3337: During the 1983 campaign, another lead sarcophagus was found during construction work and by that work had become quite damaged. The length of the lid was 1.6m by 0.7m by 0.08m. The coffin is 1.85m x 0.6m x 0.35m. The grave orientation was north-west - south-east. The northern, narrower, side of the lid and coffin was destroyed. The sides and base were fashioned from a single 631

Limes XVIII

sheet of lead. The pattern consists of intersecting cables dividing the entire surface of the lid into rhomboids. The lid was broken and deformed in its central part under the soil pressure. The southern end was flanked by one brick and at the northern side by two bricks. The remains of three deceased people were found. One was a mature person with a fragmented skull, and an outstretched right arm. The left arm was twisted at the elbow and the legs crossed at the ankles (the length of the skeleton was 1.55m). Another person was younger, again with a fragmented skull and the rest of the body badly preserved (length – 0.85m). The bones of a third deceased person was laid on its back, (length – 0.85m), and was that of a young person.

the sarcophagi from Viminacium had considered apotropaique purposes (Cumont 1966: 339). The multishanked star with arrows at the end of shanks is also a rather common ornament found on lead sarcophagi at various parts of Roman empire – in the Gauls, Dalmatia and Pannonia (Cochet 1982: 140; Duval, Jeremić & Marin 2000: 498). The precise significance of this symbol is still obscure, so it is not possible to connect it with the particular religious appertainance of the deceased (Cochet 1978: 228). In the western part of the coffin fragmented and dislocated bones were found. The deceased had been dressed in a linen dress made of purple brocade shot with golden threads of which remains were still visible after the unearthing of the coffin. This confirms that it was a burial of a person from high-class society. Related to the position of the graves and tomb, a certain order of burials and the building of the tomb could be established. In the first instance there was a lead sarcophagus, then two stone sarcophagi, after which followed the building and flooring of the tomb, then the burial of the fourth stone sarcophagus and finally, a burial placed in a simple grave pit.

Grave 3640: During the 1984 campaign, a lead sarcophagus was found during building work that damaged it. The length of the coffin is 0.77m, by 0.3m wide and 0.2m high. The thickness of the lead was 0.4cms. The grave orientation was north-west - south-east. Inside the coffin were found the dislocated bones of a child. The lateral side of the coffin was ornamented with two diagonal cable lines intersected by a vertical one.

Chance find: Particularly interesting is a piece of lead coffin given to the National Museum at Požarevac as a gift (Fig. 16, 17). The piece was found at the site of Selište – in the territory of Viminacium. Geometrical ornaments and figural images decorate the fragment - motifs of Europa and the bull and the Three Graces. Europe are common on gravestones (Klemenc 1961: 41, Pl. 35). As a specific sepulchral motif, the famous group of the Three Graces, may symbolize, just like the Muses in funerary contexts, culture and the arts as the road to immortality. This motif has been also found on a few marble pieces (Toynbee 1982: 277). The same motif is presented on the side of a lead coffin in the Museum at Beirut (Chebab 1935: 52, Pl. 27). However, it is also interesting that the Three Graces were applied as a decorative element on a bronze mirror found in one of the Viminacium cemeteries Više Grobalja (Spasić 2000).

Tomb (Grave 3971 - D): During archaeological excavation in 1985, a tomb with a stone sarcophagus were found (Fig. 5/10). Five burials were recorded. The tomb has been conserved and is now on display (Fig. 13). The entrance into the tomb was on the west side. The tomb was rectangular, vaulted, built of bricks and with walls damage during grave robbing. The walls were painted with red, black, blue and green ochre applied in geometrical and floral motifs. The tomb is west-east oriented. Of the five burials, three were in stone sarcophagi, one in the lead one and the last one placed into a simple pit dug into the soil. G-3971 Grave D: Length of coffin 1.55m, width 0.4m and height 0.35m. This lead sarcophagus was found placed into a masonry burial chamber and containing an exceptional relief on its lid (Fig. 14). The relief is divided into three spaces by the ornamentation (geometrical pattern - lemniskata), applied with a stamp (Fig. 15). Beyond these decorative elements are animal heads - most probably bears. On the edges is a well-known motif consisting of angles, often seen as decoration on other sarcophagi from Viminacium. The central zone is decorated with eightshanked star with arrows at the ends of the shanks. Two terminated zones are decorated with Medusa heads framed with angels in relief. As in a previous example, the lid was badly damaged during the robbery. The sides of the coffin are also decorated with the same ornamentation, comprising a complicated geometrical motif (lemniskata), Medusa heads and bear heads. Medusa heads are a very common motif not only on lead sarcophagi but for stonegraves too, as on a cavalry stela at Norique for legio VIII Augusta (Cumont 1966: 228). Similar masks representing satyrs or gorgons are common on lead coffins. Medusa masks especially are considered to be a motif coming from the Hellenistic period (Mouterde 1929: 248). With respect to the iconographic scheme, the Medusa head applied to

The sarcophagus itself symbolizes, just as the urn, the house of the deceased - the temple. From both scenes, the Graces and Europe, applied on the piece from Viminacium, the iconography is apparent. The architectural background confirms such a comprehension (Froning 1990: 530). Among the motifs which include the symbols of eternity from the iconography of the Bacchus cult or the Psyche myth, are also those known as apotropaique as it is Gorgona, sphynx or lion rather than mythological motifs as Minerva or Victoria and, at the end, geometrical or floral ornamentation that have a certain significance in the subordination of the context (Bertin 1974: 48). Most of the Viminacium coffins bear designs characteristic of Jerusalem workshops - diagonally crossed cable lines forming lozenges and triangles (Rahmani 1988a: 47). However there is one rare but nearer exemplar, found at the necropolis of ancient Scupi. The sides and upper were protected with stone slabs. The lid is richly decorated with 632

Snežana Golubović: Decorated lead sarcophagi in Moesia Superior a geometrical motif that is often employed on sarcophagi at Viminacium. What is very interesting is that inside the coffin were found the remains of a cremation together with a luxurious inventory: confirmation of a deceased person of high social status (Korakevik 1977: 165). The division of the lid surface into five rectangular fields is typical of coffins in southern France (Goudineau 1979: 554).

those found in the territory of Moesia Superior, have in fact, a place in the standard iconographic repertoire of Roman period funerary art. Although there are many analogies, particularly with sarcophagi from Syria and since the orientals are epigraphically confirmed, the patterns and decorative motifs used for lead coffins at Viminacium are more likely the product of local artisans. They could comprise some imported patterns as the expression of a widely accepted fashion. Chronologically, lead coffins cover the span from the late C2nd to the late C4th without the usual symbols to point at Christianity, except the one allegedly found in the territory of Viminacium. It was recorded by N. Vulić as a chance find, but without recording the precise place of discovery. Each of the narrower sides was decorated with a vertical branch, while the lateral sides figure three crosses each. Unfortunately, the coffin does not survive today and the description is known only from literature. The grave could be dated from C4th to the C6th (Vulić 1909: 133; Zotović 1995: 343).

As is well known, decorated marble and stone sarcophagi were meant to be admired and meditated on by the passerby. Many of them bear inscriptions that were intended to be read. Conversely, decorated lead sarcophagi very rare carry texts. One exception is an example from Smederevo with an additional incised text. Moreover, many of them were hidden away from sight - as the examples from Viminacium demonstrate - encased in an outer coffin made of bricks. In other parts of the Roman empire they were lowered into rock-cut cavities into the floors of tombs and covered over with slabs of stone as in the case of burial at Syria and Palestine (Rahmani 1976: 77-78) or concealed beneath the earthen mound of a tumulus (Toynbee 1982: 275).

The percentage of burials in lead coffins is proportionally low at Viminacium necropolis which confirms the special status of the persons interred in them. At the necropolis of Tournai, sarcophagi decorated with scenes from the Dionysius’ cult and one more found earlier, made up only 1% of all excavated graves. (Brulet 1996: 304).

A considerable number of Viminacium’s inhabitants were veterans of legio VII Claudia with their families. Colonists from Italy and other Roman provinces made the majority, at least, among those rich members of society who could afford gravestones. Colonists from Syria too must have made a considerable community at Viminacium. Until now, epigraphically there are confirmed just three persons from Komagena, from the district of Doliha (the villages Hairomut, Sige, Jadanata) (Mirković 1968: 69; Mirković 1986: 59, 175, 176). Therefore, parallel cremation and inhumations within the large civil and military centres such was Viminacium may be explained by the presence of orientals, who, on their part, and according to their protoChristian beliefs, had never accepted the Roman custom of cremation. Most of the applied ornaments on the lead sarcophagi are analogues to those of Syrian examples. The territory of Viminacium, inhabited by numerous veterans and with extraordinary traffic communications for trade transit between the east and west, presented an ideal ground for merchants and craftsmen - Syrians, Greeks and Italians who represented the main initiators and beginners of various crafts.

There are different explanations of the reasons for decorating coffins at all, for the choice of the elements of decoration and their possible significance of eventual beneficial influence on the deceased. Even when observing most modest lead coffins, it could be maintained that quite common ornaments should be understood as symbols. The application of deity representations, their symbols or partial Dionysiac scenes had the main assignment to protect deceased from evil. Quite simple motif, very often at the sarcophagi in Moesia Superior, such as cables applied all around the coffin, might have been intended to protect the deceased. It had to be designated that the period when the lead coffins were most in use belongs to one of decline in the stature of the pagan gods and belief in their power. Also, the artisan often applied ready-made patterns. It is likely that artisans collected motifs in their own pattern-books (application of patterns culled from metal work, sigillata ware, lamps, gems, sarcophagi and architectural decoration), choosing those which they found suitable to a funerary context. Accordingly, they would often change, add or alter the motifs while the clients, choosing ready-made coffins or desirable motifs from the pattern-book, would be inspired by common religious and vague superstitious notions frequent in their society. The same ornamental panel containing a vine scroll with bunches of grapes alternating vine leaves from the Christian lead coffin from Caesarea, was applied to nonChristian lead coffins. It is approval of the different clients order to the same artisan (Rahmani 1988b: 248) as well as not to disregard the wish to impress family and friends at the burial ceremony when the sarcophagus was laid open for a while. Thus enhanced their social standing in the

Until now archaeologically it was thought that lead was used for manufacturing coffins in various parts of Roman empire, from a period from the mid-C2nd to the early C4th. The great majority came from the east where the main centres for their production were Tyre, Sidon and Beirut on the Lebanese coast, and Ascalon, Caesarea and Jerusalem in modern Israel (Rahmani 1987: 145-147). However, in the past few years the number of excavated lead coffins in the western parts has increased. This has made possible a more comprehensive understanding of the use of lead sarcophagi. Essentially all the motifs employed for the decoration of lead sarcophagi, both in the east and west and including 633

Limes XVIII

community. The other assumption is that rich decoration, ornamental motifs on sides or lids of lead sarcophagi, have been designed only and strictly to please the dead, whose souls, while located in paradise, were yet believed to inhabit, in some sense, or at least to visit from time to time, the places where their bodies reposed (Toynbee 1982: 275). There has to compliment very strong and universal psychological impulses expended for the benefit of the dead, originating from a complex of feelings of guilt, love and fear concerning the deceased, just as from vague ideas that such prodigal might in certain way endow joy, consolation and safety to the deceased. Beyond that, much of the decoration might have been used or chosen without too much reflection, predominantly because it pleased the eye (Rahmani 1987: 146). Only with the christogram we have the symbol of values expressed in contemporary Christian writing, intended to protect from all evil, to assure help and to express faith.

srednjovekovne nekropole Smedereva. (Smederevo). Duval N., Jeremić M. & Marin M. 2000 SALONA III, Manastirine. In N. Duval & E. Marin (edd.) Salona III. (Split-Rim). Frere S.S., Hassall M.W.C. & Tomlin R.S.O. 1986 Roman Britain in 1985. Britannia XVII: 354-454. Froning H. 1990 Nachrichten aus dem Martin-von-WagnerMuseum Würzburg zu Syrischen Bleisarkophagen der Tyrus-Gruppe. Archäologischer Anzeiger: 523-535. Garbsch J. 1978 Römische Paraderüstungen. (Münich). Goudineau C. 1979 Circonscription de Côte-d’Azur. Gallia 37/2: 554-555. Jovanović A. 1984 Rimske nekropole na teritoriji Jugoslavije. (Beograd). Jovanović B. 1985 Nekropola na Pećinama i starije gvozdeno doba Podunavlja. Starinar xxxvi: 13-18. Klemenc J. 1961 Rimske izkopanine v Šempetru. (Ljubljana). Kondić V. 1960 Rimski nalazi iz Višnjiceve ulice. Godišnjak grada Beograda VI: 29-36. Konrad M. 1997 Das Römische Gräberfeld von BregenzBrigantium I. (München). Korakevik D. 1977 Sastojbata na arheološkite iskopuvanja na antički Skupi. Macedoniae Acta Archaeologica 3: 143180. Mirković M. 1968 Rimski gradovi na Dunavu u Gornjoj Meziji. (Beograd). Mirković M. 1986 Inscriptions de la Mésie Supérieure, Vol. II. (Beograd). Mouterde R.P.R. 1929 Sarcophages de plomb trouvé an Syrie. Syria x: 238-251. Nikolajević I. 1989 Recherches nouvelles sur les monuments chrétiens de Serbie et du Monténégro. Actes du XIe Congrès International d’Archéologie chrétienne III. (Rome): 2441-2462. Obermayr A. 1967 Römerstadt Carnuntum. (Vienna): 197-198. Petrović P. 1976 Niš u antičko doba. (Niš). Pop-Lazić P., Jovanović A. & Mrkobrad D. 1992 Novi arheološki nalazi na južnim obroncima Kosmaja. Glasnik 8: 135-143. Rahmani L.Y. 1976 Roman tombs in Nahal Raqafot, Jerusalem. Atiqot E.S. xl: 77-88. Rahmani L.Y. 1987 More lead coffins from Israel. Israel Exploration Journal 37/2-3: 145-147. Rahmani L.Y. 1988a Roman lead coffins in the Israel Museum collection. The Israel Museum Journal 25: 47–60. Rahmani L.Y. 1988b A Christian lead coffin from Caesarea. Israel Exploration Journal 38/4: 246-248. Santrot J. & Frugier D. 1982 Sarcophage en plomb ouvragé découvert a Cenon (Gironde). Gallia 40: 271-286. Saria B. 1924-1925 Arheološke beleške, Starinar III n.s.: 159164. Spasić D. 2000 Reljefna ogledala iz Viminacijuma, Viminacium 12, Požarevac (in press). Todorović J., Kondić V. & Birtašević M. 1956 Arheološka nalazišta u Beogradu i okolini.Godišnjak muzeja grada Beograda III: 75-98. Toynbee J.M.C. 1982 Death and burial in the Roman world. (London). Vulić N. 1909 Antički spomenici u Srbiji, Spomenik Srpske kraljevske akademije xlvii: 109-158. Zotović Lj. 1995 Early christianity in Viminacium. In D. Srejović (ed.) The age of Tetrachs. (Belgrade): 336348. Zotović Lj. 2000 Promene pogrebnih rituala tokom pet vekova istorije Viminacija. Saopštenja xxx-xxxi/1998-1999. (Beograd): 7-18.

Lead osuaria found at Slatina and coffins from Smederevo and Niš have obvious Christian symbols (eg. monograms) and are dated to the end of the C3rd to the end of the C4th. Although those from Viminacium derived the most of lead coffins, especially decorated coffins, there are no Christians symbols on them. All the specimens cover the chronological span from the mid-C2nd to the mid-C4th. However, it is well-known from historical sources that Viminacium was a bishop’s residence in the C4th. Two of Viminacium’s bishops, Amantinus and Cyriacus, have been confirmed (Mirković 1968: 72). The reasons for such a situation could be explained by the fact that in Viminacium, the solar theology, expressed through Mithraism, had a priority unlike in other centres such as Singidunum and Nais where the supreme god Jupiter was the dominant deity. Mithraism, supported by the state, was a strong bulwark in the struggle against Christianity for a long time (Zotović 1995: 346). Bibliography Bertin A.-M. 1974 Les sarcophages en plomb Syriens au Musée du Louvre. Revue Archéologique 1: 43-82. Boulakia J.D.C. 1972 Lead in the Roman world. American Journal of Archaeology 76: 139-144. Brulet R. 1996 Dionysisch in der Unterwelt-Ein gallo-römischer Bleisarkophag von Tournai. Antike Welt: 297-304. Buschhausen H. & H. 1991 Der imperiale Bleisarkophag aus dem Martyrium zu Niš. Römisches Österreich 17-18: 45-59. Čerškov E. 1970 Municipium DD kod Sočanice. (Priština – Beograd). Chébab M. 1935 Sarcophages en plomb du Musée National Libanais. Syria XVI: 51-72. Cochet A. 1978 Les Sarcophag de plomb du Musée de Rouen. Actes du Colloque International d’Archéologie Rouen 3-4-5 Juillet 1975. (Rouen): 217-233. Cochet A. 1982 Note sur le sarchophage d’enfant, en plomb, decouvert a la Boisse en 1980. Revue archéologique de l’Est et du Centre-Est 33: 139-141. Cumont F. 1966 Recherches sur le Symbolisme Funéraires des Romains. (Paris). Cunjak M. & Marković-Nikolić Lj. 1997 Antičke i

634

Snežana Golubović: Decorated lead sarcophagi in Moesia Superior

Fig. 1. 1-6 Sites with lead sarcophagi in the territory of Moesia Superior (after M. Mirković, Istorija srpskog naroda I (1981):73.

635

Limes XVIII

Fig. 2. Lead ossuary from Slatina (after Pop-Lazić, Jovanović & Mrkobrad (1992): 137, fig. 2.

Fig. 3. Lead sarcophagus from Smederevo.

Fig. 4 The busts at the shorter side of a lead coffin from Naissus.

636

Snežana Golubović: Decorated lead sarcophagi in Moesia Superior

Fig. 5. Plan of Viminacium (after G. Milošević, New data on the topography of Viminacium. In Acta of a conference Roman and Late Roman City (Veliko Trnovo, July 2000): fig. 4 (in press).

Fig. 6. Lead coffin from Viminacium (Grave 228).

Fig. 7. Lead coffin from Viminacium (Grave 228).

637

Limes XVIII

Fig. 8. Lead coffin from Viminacium (Grave 339).

Fig. 9. Lead coffin from Viminacium (Grave 339).

Fig. 10. Lead coffin from Viminacium (Grave 349).

638

Snežana Golubović: Decorated lead sarcophagi in Moesia Superior

Fig. 11. Lead coffin from Viminacium (Grave 349).

Fig. 12. Lead coffin from Viminacium (Grave 1221).

Fig. 13. A tomb with stone sarcophagus from Viminacium.

Fig. 14. Lid with Medusa’s heads.

639

Limes XVIII

Fig. 15. Lid with Medusa’s heads – detail.

Fig. 16. Fragment with the Three Graces and Europe on the bull.

Fig. 17. Fragment with the Three Graces and Europe on the bull.

640

Artillery troops detached north of the lower Danube in the late Roman period Dorel Bondoc This study is meant to outline a rarely discussed problem in Romanian history: artillery troops detached to the northern Danube fortification bridgeheads, by the military administration of the Moesia Prima, Dacia Ripensis, Moesia Secunda and Scythia provinces in the late Roman period. In Romania, after 1989, there has been less and less archaeological research because of the extremely reduced funds. This is why this study is based on published archaeological information rather than fresh research.

Among these stamps, only those of the equites sagittarii give certain information about its nature. The vexillation of legio VII Claudia must have been an infantry detachment, but about the third, attested only by the tegulae stamps, there is not much to say. Three pieces of artillery discovered inside the fortification offer further information about the character of the troops that were stationed at Gornea (Gudea & Baatz 1974: 5072; Gudea 1977a: 47-60; 1977b: 82-83; 1978: 69-75). I refer to three cylindrical pieces of iron (kambestrion) (Fig. 1), each one made up of two rings bars each. The description is according to Gudea 1977b:

Artillery units detached to the late Roman fortifications north of the Danube are relatively unknown. The literary sources, with few exceptions (Heron Cheirobalista, according to Marsden 1971: 212-217; Vegetius II.22-25; III.24; Anonymus de Rebus Bellicis VII, XVIII: Exposito ballistae fulminalis, according to Thompson 1952; Ammianus Marcellinus XVII.1.12; XVIII.9.1 and especially XXIII.4.1-3) are silent about the types of troops involved and the archaeological discoveries are sporadic. For these reasons the character of the present study is relative, but future research should increase our knowledge of this problem.

1: A cylindrical piece of kambestrion made up of two rings with bar of 1.7-2cms width, 0.4cms high. The external diameter of the rings is 8cms and 9cms respectively; the internal diameter is 5.4cms. The two rings are bound together with two vertical bars 12.5cms long, fastened to square holes on the rings and then rivetted. On the rings there are two holes on each side of the rivets to 3-3.5cms. The holes are 0.4cms in diameter and are placed at 4cms one to another. The vertical bars have a 1.3-2cms width and 0.6cms height. One of the bars, the widest one, has in the middle a semi-circular vault 3.5cms wide. The distance between the two vertical bars is 6.7cms. At each end, up or down, of the vertical bars there are two ‘ears’ made up of a bar bent in the shape of the letter ‘U’, fixed to holes on the vertical bars and then rivetted. The dimensions of the bar the ‘ears’ are made of are 1.02 x 0.4cms. On the superior side (at the ring 8cms in diameter) the size of the ‘ears’ cannot be specified because both were broken and restored. For the inferior ring (9cms in diameter) the dimensions of the ‘ears’ are 2.4 x 3.5cms. The distance between the rivets was 2cms at the upper ‘ears’ and 2.9cms at the lower ones. The ‘ears’ on the superior side were smaller than the ones on the inferior side. The total height of the piece is 13.3cms. The present weight is 330.5g. The piece was found inside the tower in the south-east corner, at 1.70m depth.

For artillery, this includes all the military machines (catapult, onager, cheirobalista-manuballista, tormenta muralia, ballista, ballista fulminalis). Seemingly, the detachment of artillery from the legions happened during the reign of Constantine the Great (Marsden 1969: 195196). For the historical context, the following considerations are integrated: it is marked by the administrative and especially military reforms which aimed first of all at the efficiency of the efforts of Roman emperors starting, with Gallienus, made in order to face successfully the barbarian attacks which happened especially in the eastern Roman empire. It is not my intention to consider the typology of war machines. The problem has been discussed extensively elsewhere (Marsden 1969; 1971). However I will point out those archaeological sites where there are attested traces of Roman artillery equipment. There are a few late Roman fortifications that have been investigated which offer information about the existence of artillery troops inside their walls; they too will be presented below.

2: A cylindrical piece of kambestrion made up of two rings, with bar of 1.2- 1.5 x 0.4cms. The external diameter of the rings is 8.5cms; the internal diameter is 5.9cms. The distance between the rings is 13.6cms. The rings are bound together with two vertical bars of 13.6cms long each, fixed to holes on rings and rivetted. On each side of the rivets to 3-3.5cms distance there are two holes 0.4cms one to another. The vertical bars are wide of 1.5cms and 2cms and 0.7cms high. The wider bar has in the middle a part in the shape of a vault, wide of 3.5cms and high of 0.7cms. The distance between the bars is 6.9cms. Each of the two bars has two ‘ears’ in the shape of the letter ‘U’ at each end. The ‘ears’ are made up of a bar of 0.8 x 0.4cms. Like in the case of the above piece the ‘ears’ on the superior side are smaller than the ones on the inferior side. The

1. Gornea, (Sicheviţa commune, Caraş-Severin district, Romania); a fortification of the quadriburgium type (41.5 x 41.5m; cf. Fig. 6; Gudea 1977b: 42-58) which lodged many military units during its existence. Thus, there are attested the following: a vexillation of legio VII Claudia (IDR III/1.31); an anonymous military unit which is designated by the tegulae stamps as DA(ciae) R(i)P(ensis) (CIL III.8075b): these stamps might be from the import of tegular materials; a unit of equites sagittarii (IDR III/1.32).

641

Limes XVIII

‘ears’ are fastened to holes and then rivetted. At the upper ring the sizes of the ‘ears’ are 1.6 x 20cms and at the lower ring the dimensions of the ‘ears’ are 2 x 3.1cms. The total length of the piece is 14.4cms and the present weight is 425.7g. The piece was found inside the tower in the southeast corner at 1.5m depth.

11.5cms the lower ring; the internal diameter is 7.5cms, respectively 8cms. The two rings are connected between them by the two bars of 32cms length. One of the bars has a semi-circular bent in the middle. The bars are fastened to holes on the rings and then rivetted. On both sides of the rivets at 2cms distance there are two holes 0.8cms in diameter placed at 1.5cms one to another. The aspect of the rings is not circular; at the upper side they are a little prolonged to the ends where the ‘ears’ are fixed to the bars. Each bar has two ‘ears’ at each end. At the superior side of the piece there are fastened two smaller ‘ears’; they are 2.5 x 5cms in the exterior and 2 x 3cms in the interior side. The ‘ears’ are made of a bar measuring 4.5 x 3-3.5cms in the interior side. The ‘ears’ are made up of a bar of 4.5 x 0.8cms in the superior side of the piece and 5.5 x 1cms in the inferior side. The part to the ring of the ‘ears’ was widened, pierced and fastened to the holes on the vertical bar and rivetted. The total length of the piece is 36cms; the present weight is 8kg.The piece was discovered inside the tower in the south-west corner at a depth of 2.5m.

3: A cylindrical piece of kambestrion made up of two rings with the external diameter of 7.9cms on the superior side and 8.2cms on the inferior side; the internal diameter is 5.4cms. The rings are made up of a 1.3 x 0.5cms bar. The bars are bound together with two vertical bars of 13.7cms length, 1.5 and 2cms width, 0.5cms high. The vertical bars are fixed to holes and rivetted to their ends. On each side of the rivets there are two holes 0.4cms in diameter; the distance between bars is 6.5cms. In the middle of the widest bar there is a vault of 3.8cms high. On the vertical bars there were rivetted two ‘ears’ at the extremities. Their dimensions differ at the same end. Unlike the two pieces above (see also piece no. 4) the ‘ears’ are distributed up and down, one small and one big. At the end where the ring is 7.9cms in diameter, the dimensions of the ‘ears’ are 2.2 x 2.5cms in the exterior and 1.7 x 1.5cms in the interior; the other ‘ear’ has 2.8 x 3.8cms in the exterior, respectively 2 x 2.9cms. The ‘ears’ have been destroyed partially and then restored. At the lower end (8.2cms in diameter) the smaller ‘ear’ has 1.7 x 2.5cms (in the exterior) and 1.3 x 2cms (in the interior); the bigger ‘ear’ is 2.8 x 4cms (in the exterior) and 2.6 x 2.9cms in the interior. The ‘ears’ are made up of a bar of 1.1 x 0.4cms. The total length of the piece is 14.7cms; the present weight is 447.3g. The piece was discovered inside the tower of the south-tower at a depth of 1.6m.

2. A piece of iron 1.45m long (kamarion). In the middle of the piece there is a semi-circular curve 30cms in diameter; the bend is 16cms high; on both sides on the external extremities of the 10cms curve there are two round holes, 0.9-10cms in diameter. Near the holes the bar is 3 x 1.2cms wide; these dimensions change at the ends of the bar where they are 2.5-2.7 x 1.5cms. At the two ends the bar has two forked arms in the shape of the letter U. The ends of the arms have been destroyed, so it is difficult to specify the dimensions. The distance between the arms is 22cms and the depth is 10cms. On the better-preserved arm, there is a hole 0.9cms in diameter and to 2cms there is a part of another hole. The bar the two forked arms are made of is 0.4cms thick. The piece was discovered inside the tower in the south-west corner at 2.5m depth. It piece is dated towards the end of the C4th (Gudea 1977a: 59). Altogether these pieces of iron certify the presence of ballistarii in the garrison at Dierna and the realities of its Roman military history.

The date of the three pieces is towards the C4th on the evidence of coin finds of Valentinian I - Theodosius. We are dealing with a unit which had artillerists specialized in throwing arrows - manuballistae. These machines were easy to handle by one man, so they could be used on walls as well in the open-field. 2. Dierna (Orşova, Mehedinţi district, Romania); a late Roman fortification (Fig. 7) of quadriburgium type (36 x 35.5m), has been identified here due to systematic archaeological research (Gudea 1972: 177). The garrison had initially soldiers belonging to legio XIII Gemina (IDR III/1. 47, 51). Later, there were quartered anonymous military detachments designated on stamped bricks as D(acia) R(i)P(ensis) DIERNA, DIERTRA and DA(cia) R(ipensis) DIANA (IDR III/1. 44-46). The final examples might be imported tegular materials.

The failure of legionary infantry when confronted by barbarian horsemen called for changes within the framework of old style Roman military units. It seems that confrontations in the open field were avoided and engagement at a distance was preferred. Thus the war machines derived a significant importance in battle. In such contexts, the ballista from Dierna should be taken into consideration. It was a heavy weapon, so it was placed on a tower platform and was handled by many men. Its function was most probably defensive, because its size and weight meant it could hardly have been used for attack (for reconstruction of this type of war machine, see Anstee 1998; cf. Fig. 5).

The discovery of two pieces (Fig. 2) belonging to a ballista, dated to the end of the C4th, inside the fortification (Gudea & Baatz 1974; Gudea 1977a; 1978), gives rise to the problem of the character of the troops in Dierna. Descriptions according to Gudea 1977a:

3. Praetorium (Mehadia, Caraş-Severin district, Romania); the old stone Roman camp (142 x 116m) was rebuilt with the same dimensions (Fig. 8, acc. Marsigli), in the time of Constantine the Great (Macrea 1949: 139-141). At that time protection of the fortification was secured by a

1: A cylindrical piece of kambestrion formed of two rings with the external diameter of 14.5cms (the upper ring) and 642

Dorel Bondoc: Artillery troops detached north of the lower Danube in the late Roman period detachment made up of soldiers belonging to legio XIII Gemina and legio V Macedonica (IDR III.120, nr. 102; Macrea, Gudea & Moţu 1993: 51-53).

(there are no photos or drawings) which were considered to be parts of a war machine, and also some spherical missiles (Fig. 3) of burnt clay and of stone (Davidescu 1980: 134). It seems that these were used by artillery to defend Drobeta. It cannot be specified if the pieces belong to the C2nd-C3rd or to the C4th.

Archaeological research has allowed, broadly speaking, the identification of the armament used as long as this fort lasted. A spherical projectile, 10cms in diameter, of sandy grit stone is important in this respect (Macrea, Gudea & Moţu, 1993: 44, 114, presented without photos or drawings). This evidence shows without any doubt the existence of a war machine, very probably an onager, inside Mehadia. It cannot be specified if the projectile belongs to the C2nd-C3rd or to the C4th.

5. Hinova, (Mehedinţi district, Romania); a late Roman quadriburgium (45.85 x 39.8m: cf. Fig. 10). It was investigated systematically between 1976-1981 (Davidescu 1989). The units based in the fortification are recorded on the stamped bricks that have been discovered here: L(egio) V M(acedonica), DIANA, Da(ciae) R(ipensis) DIANA and DRVBETA (Davidescu 1989: 33-41). Only the first type of stamp permits identification of the unit that the detachment belonged to; for the rest we have no clear archaeological clues (= the bricks represent imports?).

4. Drobeta-Theodora (Turnu-Severin, Mehedinţi district, Romania); the old Roman camp built at the beginning of the C2nd, suffered many restorations and architectural changes (Fig. 9) within the same basic dimensions (137 x 123m; cf. Zahariade 1997: 167). As in the previous period, this point was given a special attention.

The existence of artilllerists at Hinova is confirmd by the discovery of some spherical projectiles of stone inside the fortification (Davidescu 1989: 67, without photos or drawings). They might have been thrown from war machines by the defenders of the fortification. The war machines were very possibly placed on the towers of the fortification, especially on the one situated in the middle of the north side.

After the Aurelian withdrawal there a unit was stationed in the Roman camp as demonstrated by an altar dedicated to Jupiter Cohortalis in the name of all centuriones (Bárcácilá 1934: 82-84). At Drobeta there was also stationed a vexillation of legio V Macedonica, which left many tegulae stamps, the majority dated to Aurelian-Diocletian (Benea 1977: 176). The presence of a detachment of legio XIII Gemina is also possible because its centre of command was near by, at Ratiaria. Surprisingly, from the latter, evidence comes to us only in a late stamped brick- L(egio) XIII G(emina) P(ars) S(uperior) (Bárcácilá 1938: 22).

6. Ostrovul Mare, (Gogoşu commune, Mehedinţi district, Romania); a late Roman fortification (Fig. 11) which has been investigated in very difficult conditions because it is almost entirely destroyed (Davidescu 1989: 113-117). There is no information about the military unit which protected this fortification. Among the pieces in the extremely poor inventory of finds, the investigator mentions three round pieces of limestone (Davidescu 1989: 116) considered to be weights for fishing nets. Alternatively they could be projectiles for war machines. However this hypothesis is not supported by sure archaeological evidence (there being no published photos and drawings).

The structure of the military disposition at Drobeta changed radically beginning with the reign of Constantine the Great, at the time of the transformation of the inner arrangement of the old camp, in the cross shape (Zahariade 1997: 167-182). The Notitia Dignitatum (Occ. XLII.16; XLII.24) mentions two border units here, a cuneus equitum Dalmatarum Divitensium and an auxilium primorum Daciscorum, both dated c.395 or later.

7. Bistreţ, (Dolj district, Romania); the military unit in this late Roman fortification (Fig. 12), built by Constantine the Great, north of the Danube, is unknown (Tudor 1978: 265). The archaeological discoveries give an idea about its character (Vládescu & Zahariade 1986: 29-40); as they include some spherical stone artillery balls (Fig. 4), which certify the existence of a troop, made up partly of artillerists, inside the fortification.

There were other units at Drobeta, if we take into consideration the reading of the stamps that have been discovered, unless they are imported: DA(ciae) R(ipensis) DIANA (IGLR 404 with bibliography), D(aciae) R(i)P(ensis) AQUIS (IGLR 405, with bibliography), [D(aciae) R(i)P(ensis)] DIERN(a) (Pärvan 1913: 50-51), DRVBETA (IGLR 406, with bibliography).

8. Sucidava-Sykibida (Celei, Olt district, Romania); a late Roman fortification (Fig. 13) built during the second half of the C3rd, perhaps under the emperor Gallienus (Barbu 1973: 27). The permanent presence of military detachments at Sucidava is proved by the many tile stamps in the area of the fortification (the types were published in IGLR 277-299, with bibliography; Toropu & Tátulea 1987: 103-104). A vexillation belonged to legio V Macedonica (cohs III and IIII), detached from Oescus under Diocletian-Constantine the Great. The Notitia

In connection with the preceding discussion, attention should be paid to the interval towers, between the gates and the corners of the fortress. Their function was probably to support the emplacement of machines for throwing arrows and missiles over a large distance (Tudor 1978: 449). This hypothesis is supported by archaeological evidence (Davidescu 1980: 134): two pieces of iron bent at the centre, with the extremities in the shape of a letter T 643

Limes XVIII

Dignitatum (Or. XLII.39) records a praefectus legionis quintae Macedonicae here. Other stamps record the presence of infantry and cavalry sent from the Varinia, Almus and Utus fortifications (IGLR 289, 294, 293). Bricks with the stamps of legio XIII Gemina and legio VII Claudia (IGLR 287, 288), prove the importation of material from the military centres at Ratiaria and Viminacium. There are also the ordinary stamps Da(cia) R(i)p(ensis) (IGLR 295). The external towers of the fort functioned until the end of the C4th and possessed platforms for war machines which provide interesting clues about the character of the troops at Sucidava (Tudor 1978: 430). The war machines threw arrows and projectiles at a distance. This assertion is supported by the spherical projectiles of stone discovered in the area of the civilian settlement and fortification (Toropu & Tátulea 1989: 73, 112, but without photos and drawings).

were light, they could be used in any context. However the two pieces discovered at Orşova merit special discussion. They belong to a heavy type of machine, a ballista fulminalis. Because of this, they needed a special spot for emplacement which could only be one of the four towers at this fortification (Gudea 1977a: 59). The towers are square (8 x 8m) and have thick walls (1.5m). These characteristics assured the solidity necessary for supporting the ballista. The fortifications at Gornea and Orşova are tetrarchical in date, and the pieces discovered inside them are dated towards the end of the C4th. At Drobeta, the intermediary towers between the gates and the corners of the fortification, were considered by Tudor (1978: 449) as necessary to support ballistic machines. Having a square shape, the towers were placed two on the south side (3 x 3m), two on the north one (4.35 x 4.35m). They have thick walls (2m) and were built in the time of Constantine the Great.

9. Dafne; a late Roman fortification built by the Constantine the Great north of the Danube, across Transmarisca, cannot be identified in the field because it was probably destroyed by erosion by the Danube and Argeş. It is mentioned by some important literary sources (Ammianus Marcellinus XXVII.5; Notitia Dignitatum Or. VIII. 45-46; Procopius De Aed. IV.7).

The probability that the tower in the middle of the north side of the fortification at Hinova would have been used for a similar purpose seems plausible. The tower in question is square (with interior dimensions: 3.9 x 3.8m), and thick walls (1.35-1.5m) and is dated to the time of Diocletian (Davidescu 1989: 13-14).

The troops stationed in the fortification at the start of the C4th are known from the Notitia Dignitatum (Or. VIII.4546), which registers Constantini Dafnenses and Ballistarii Dafnenses regiments. Two units are mentioned without specifying the particular name of the fortification which implies that c.395 the fortification had not been a bridgehead for the empire at the north of the Danube for a long time. The units had been withdrawn before this date to south of the Danube.

At Ostrovul Mare, the northern tower of the partially uncovered enclosure is round in shape, with an interior diameter of 3.5m. It seems that such shaped towers were used to accommodate artillery (Zahariade 1999: 9) and it is possible that the one from Ostrovul Mare would have had the same function. Fortifications with round corner towers lead to many possibilities of dating; one of them is of the Valentinian-Valens period (Gudea 1982: 102).

For the subject under discussion, the Ballistarii Dafnenses are of special importance. Zahariade (1977: 398) discussed the possibility that the weapon used by this regiment was the ballista fulminalis, recorded by the Anonymus de Rebus Bellicis (XVIII: Exposito ballistae fulminalis) which could be handled by three or at least two men. In contrast, Vegetius (II.25) suggests that a ballista should be handled by 11 men, the figure is clearly an exaggeration for a unit which had no more than 1000 men. The conclusion that a regiment of ballistarii had ballistae and a considerable number of manuballistae or other weapons seems acceptable from this point of view (Zahariade 1977: 401).

Although it belongs to the C6th, the southern pentagonal tower at Ostrovul Mare probably had the same use. Likewise Towers I and J at Sucidava, with a similar shape, seem to have been used for the same purpose (Tudor 1978: 430), but again in the C6th. Sucidava’s interior towers F and G were considered by Tudor (1965: 82) to be high platforms meant to support ballistic machines for throwing projectiles. Tower F is of rectangular shape (7.35 x 3.5m) with walls of 1.55m thick. Tower G has not yet been explored.

Looking at the placement of the war machines within northern Danubian fortifications, there are clues for analysing the characteristic elements of the fortifications. For instance, Tudor noted the fact that some exterior towers at Drobeta and Sucidava were designed especially to be emplacements for such ballistic machines (Tudor 1978: 449, 430). I shall resume the discussion here.

Both constructions are dated to the time of Constantine the Great (Toropu & Tátulea 1987: 84-85). I have mentioned above that the exterior towers I and J, although they belong to the C6th, were considered to have the same function (Tudor 1978: 430). Tower I is of irregular pentagonal shape (4.2 x 1.3 x 2.3 x 2.75 x 1.55m). The exterior tower J had four external sides (3.4 x 5 x 6 x 3m).

At Gornea, where there have been discovered three pieces of manuballistae, it is not the case of special constructions being needed for such a purpose. Because the weapons

For the fortifications Mehadia and Bistreţ we do not have clear archaeological evidence and for Dafne it is totally missing. 644

Dorel Bondoc: Artillery troops detached north of the lower Danube in the late Roman period For the moment, it is difficult to construct a typology of the elements of fortification destined to house ballistic machines. It was necessary to ensure a certain security and solidity for them, and they be placed in dominant places with a good visibility.

Gudea N. & Baatz D. 1974 Teile spätrömischer Ballisten aus Garnea und Orsova (Rumonien). Saalburg Jahrbuch XXXI: 50-72. Macrea M. 1949 Sápáturile arheologice de la Mehadia. In Studii 2/1: 139-141. Macrea M., Gudea N. & Moţu I. 1993 Praetorium. Castrul şi aşezarea romaná de la Mehadia. (Bucureşti): 51-53. Marsden E.W. 1969 Greek and Roman artillery, I. Historical development. (Oxford). Marsden E.W. 1971 Greek and Roman artillery, II. Technical treatises. (Oxford). Pärvan V. 1913 Stiri nouá din Dacia Malvensis. Analele Academiei Romäne. Memoriile secţiunii istorice 36. (Bucureşti): 39-68. Thompson E.A. 1952 A Roman reformer and inventor. (Oxford). Toropu O. & Tátulea C. 1989 Sucidava- Celei. (Bucureşti). Tudor D. 1965 Sucidava. Une citê daco-romaine et byzantine en Dacie. (Bruxelles). Tudor D. 1978 Oltenia romaná. (Bucureşti). Vládescu C.M. & Zahariade M. 1986 In Oltenia. Studii şi comunicári, V-VI. (Craiova): 29-40. Zahariade M. 1977 Constantini Dafnenses şi Ballistarii Dafnenses. Studii şi cercetári de istorie veche şi arheologie 28/3. (Bucureşti): 391-402. Zahariade M. 1997 The late Roman Drobeta I. The cruciform building and the fort garrison in the 4th century a.d. Acta Musei Napocensis 34/1. (Cluj): 167-182. Zahariade M. 1999 The tetrarchic building activity at the lower Danube: I. Quadriburgia. In G. v. Bulow &A. Milčeva (Hrsg) Der limes an der unteren Donau von Diokletian bis Heraklios. Vorträge der Internationalen Konferenz Svištov (1-5 September 1998). (Sofia): 3-13.

The presence of a unit of ballistarii at Dafne and the archaeological evidence, direct or indirect, suggests that in the late Roman period, fortifications north of the Danube often used artillery. Procopius (De Aed. IV.5.1-8) appreciated that in the late Roman fortifications north of the Danube there existed a small number of people. The utilization of the artillery could compensate for the small number of defenders. This is the present stage of the investigation. Late Roman fortifications on the north ripa of the lower Danube limes give us clues about the existence of special artillery detachments as parts of the garrison on the watch. As I mentioned at the beginning of this paper, future archaeological investigation should increase the number of known examples and modify these conclusions.

Bibliography CIL IDR IGLR

Corpus Inscriptionum Latinarum. (Berlin, 1863 and following). Inscripţiile Daciei romane. (Bucureşti, I (1975), II (1977), III/1 (1977) and following). E. Popescu Inscripţiile greceşti şi latine din secolele IV-XIII descoperite în Romania. (Bucureşti, 1976).

Anstee J. 1998 In Studia Danubiana, Pars Romaniae, Series Symposia, I. The Roman frontier at the lower Danube 4th-6th centuries. The second International Symposium (Murighiol/Halmiris, 18-24 August 1996). (Bucureşti): 131-139. Barbu V. 1973 Fortáreaţa romano-bizantiná de la Sucidava ín lumina cercetárilor din sectorul de sud-est. Studii şi cercetári de istorie veche şi arheologie 24/1. (Bucureşti): 27-53. Bárcácilá A. 1934 Arhivele Olteniei 13. (Craiova): 82-84. Bárcácilá A. 1938 Une ville daco-romaine: Drubeta. (Bucarest). Benea D.1977 Vexilaţii ale legiunii a V-a Macedonica la Drobeta ín secolele III-IV. Acta Musei Porolissensis 1: 176. Davidescu M. 1980 Drobeta ín secolele I-VII e.n. (Craiova). Davidescu M. 1989 Cetatea romaná de la Hinova. (Bucureşti). Gudea N. 1972 Befestigungen am Banater Donau-Limes aus der zeit der Tetrarchie. Actes Mamaia – Actes du IXe Congrès International d’Ètudes sur les frontiers romaine, Mamaia 1972. (Bucureşti): 171-180. Gudea N. 1977a Studii şi materiale de muzeografie şi istorie militará 10. (Bucureşti): 47-60. Gudea N. 1977b Gornea. Aşezári de epocá romaná şi romaná tärzie. (Reşiţa). Gudea N. 1978 Contribuţii la istoria artileriei antice. Drobeta 1978: 69-75. Gudea N. 1982 Despre graniţa de nord a provinciei Moesia I şi sectorul vestic al frontierei de nord a provinciei Dacia Ripensis de la 275 la 378 e.n. Drobeta 5: 93-111.

645

Limes XVIII

Fig. 1. Three cylindrical pieces of iron (kambestrion) found inside the Gornea fortification.

Fig. 2. Two pieces of iron belonging to a ballista dated to the end of the C4th found inside the Diernea fortification.

Fig. 3. Spherical missiles of burnt clay and stone from Drobeta.

646

Dorel Bondoc: Artillery troops detached north of the lower Danube in the late Roman period

Fig. 4. Spherical stone artillery balls from Bistreţ.

Fig. 5. Reconstructed ballista.

Fig. 6. Gornea; fortification of the quadriburgium type.

Fig. 7. Dierna. Late Roman fortification quadriburgium.

Fig. 8. Praetorium Mehadia. Roman camp.

647

Limes XVIII

Fig. 10. Hinova. Late Roman quadriburgium.

Fig. 9. Drobeta-Thedora. Roman camp.

Fig. 12. Bistreţ. Late Roman fortification. Fig. 11. Ostrovul Mare. Late Roman fortification.

Fig. 13. Sucidava-Sykibida. Late Roman fortification.

648

The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior) Andrzej B. Biernacki Since 1970, the interdisciplinary team of the Archeological Expedition of the Adam Mickiewicz University in Poznań has been researching the legionary camp and the later Roman and early Byzantine city of Novae on the lower Danube, in the former Roman province of Moesia Inferior (now Svishtov, Bulgaria).1

3.3m deep. The pool in the apse of the tepidarium is of the diameter of 6.4m and the depth of 2.41m. The northeastern part of the parallel row complex of the bath was a rectangular room of the dimensions of 27.1 x 10.2m (90 x 34 Roman feet), whose function was apparently that of an apodyterium.6 In the southern wall of this room, we found an entrance, which most probably had a double door. It is remarkable that the entire building of the legionary bath was hypocausted. The entrance to the complex of the bath was a columned portico with a large vestibule in the east. The portico led to the street which separated the insula where the bath was, from the complex of the principia. There also might have existed another entrance from the north, ie. from the via principalis, which however is not very likely. South of the room identified as the apodyterium, there was another large room: the frigidarium with a rectangular pool (natatio) of the dimensions of 9.8 x 6.5m.7 At the east wall of the pool, there were stone seats and two sets of stairs leading into the water. Exactly on the middle west-east axis of the room, there was a semicircular niche in the wall of the pool. Inside, a stone cascade was installed, with 7 steps and two statues of dolphins on the sides.8 Directly south of the natatio, which probably was not roofed, we have started to expose another large room which was a part of the complex of the legionary bath. Like the rooms that we explored earlier, this one was also hypocausted. The recorded values of the thickness of the walls in the north-western corner of the room, which amount to 0.93 and 0.96m, indicate that this interior was of substantial dimensions.

In the years 1986–2000, the exploration focused on the central part of the legionary camp and the early Byzantine city, directly west of the principia of the legio I Italica.2 We have now sufficient evidence to conclude that this was an area of bustling building and civilizational activity in Novae (Moesia Inferior) from the early C2nd to the late C6th AD. Under the complex of the enormous early Christian episcopal basilica, the minor basilica and the Bishop’s residence, a dozen rooms were unearthed, belonging to two complexes of baths dated to the first half of the C2nd and the turn of the C4th AD. Following this discovery, we were able to determine in more detail the layout and cubic capacity of the buildings of the legionary bath, and consequently to establish the chronological and architectural layers of the bath facilities. The north-western part of the facilities of the legionary bath in Novae, which are arranged in a row, consists of looking from the north - a praefurnium, two alvei, a caldarium, a sudatorium (?) or caldarium II (?), and a tepidarium.3 These rooms, which had semicircular pools in the west, were of very similar cubic capacity. The caldarium of the legionary bath was a rectangular room of the dimensions of 14.55 x 5.85–5.95m, with a semicircular pool in the west, of the diameter at the base amounting to 5.95m, and the depth at the base of 3.7m. In the northern part of the caldarium, two small pools (alvei) were discovered.4 On the north, these pools adjoined a large praefurnium. In the next two, identical rooms, construed as the sudatorium and the tepidarium of the legionary bath, which are in the same row as the caldarium and adjacent to it,5 it was possible to establish the location and dimensions of their apses, which housed bathing pools. The diameter of the apse in the sudatorium (?) is c.6m, and

The large interior located south of the tepidarium of the legionary bath, has the dimensions 32.8 x 18.1m, or approximately 110 x 60 Roman feet, and characteristic rectangular buttresses at the walls. So far, 10 of these internal buttresses have been studied. Their spacing is between 1.8 and 2.2m. Based on the collected data, due to its unusual layout and architectural design, the large interior has been designated as the palaestra or basilika diskoperta, used for recreation and rest.9 In the southeastern part of this room, directly north of the open-air natatio, a section of a sewer running along the west-east axis and covered with monolithic stone slabs was encountered several years ago, and its contents studied.

1

The results of the exploration of Novae by the Archeological Expedition of the Adam Mickiewicz University in Poznań are regularly reported in the annual Archeologia and in the book series Novae. Studies and Materials published in Poznań. This paper was translated by P. Znaniecki. 2 cf. Novae - Sektor Zachodni. Sprawozdania tymczasowe z kampanii wykopaliskowych Ekspedycji Archeologicznej UAM [Novae: the Western Sector. Interim Reports of the Excavation Seasons of the Archeological Expedition of the Adam Mickiewicz University in Poznań] in the annuals Archeologia, beginning in 1986, and Archeologia 48. 1997 (1998): 35– 42. 3 Archeologia 47. 1996 (1997): 65-84; Archeologia 48. 1997 (1998): 3742. 4 A similar design of baths is described, eg. in Stockstadt & Zugmantel City Bath at Carnuntum. Great City Bath II at Cambodunum, Small Eastern Bath at Timgad, Northeastern Bath at Epidauros, Double Bath at Ptolemais, cf. Nielsen 1990. 5 Archeologia 47. 1996 (1997): 70, 74, 75, fig. 3.

6

My earlier account of this room as the frigidarium, cf. Archeologia 47. 1996 (1997): 70, fig. 3, was mistaken. I was able to revise my opinion because of new archaeological data gathered during subsequent seasons. 7 Archeologia 39. 1988 (1990): 151–155. 8 Although A. Sadurska and M. Biernacka-Lubańska described and discussed this relic on several occasions, they had never seen it, and their accounts are based on drawings and photographs only. For a proper interpretation of this relic, based on personal on-site study, cf. P. Pawlak 1997: 95-97. 9 Basilikai diskopertai of similar layouts may be found, eg. in the Julia Memmia Baths in Bulla Regia (Hammam-Barradji) in Africa Proconsularis, in the large ‘Palais du Légat’ baths in Lambaesis, or in the bath-gymnasium of Odessos (Varna).

649

Limes XVIII

The studied length of the sewer is 11.7m. In the southwestern part of the palaestra or basilika diskoperta, an entrance with a monolithic stone threshold has been discovered. A legionary bath of a similar layout, featuring a palaestra in the south, and dated to the third-quarter of the C2nd, is known from Potaissa (Turda: cf. Barbulescu 1997: 37–39, abb. 5, 6, 19). Having divided the floor projection of the bath into layers, we classify it as type III, or I. Nielsen’s parallel row type.10 This type of baths developed fully in the late C2nd. We know of six facilities of this type: in Bumbeşti (Dacia), Slaveni (Dacia), Boudobriga (Germania Superior), Dinogetia (Moesia Inferior), Lauriacum (Noricum) and Iciniacum (Rhaetia Nielsen 1990: II 137–139, 148). Baths of this type are particularly often encountered in the legionary camps in the provinces of Dacia and Moesia Inferior (Nielsen 1990: I 76). The layout of the legionary bath in Novae is similar to that of the legionary bath in Lauriacum, which is dated to the period of the Severi (? - Vetters 1953: 49–53).

3.22m. During the exploration, fragments of large rectangular ceramic slabs were found, with cylindrical appendages (pins) 0.013m high and 0.033-0.038m diameter. A slab of the dimensions of 56.0 x 0.43 x 0.0250.036m was furnished with three such pins: two in the corners of one shorter side, and a third one in the middle of the other shorter side. The slabs were the tiles which originally lined the heated walls in the room of the legionary bath. They were secured by means of the pins and spacing bobbins. The early room in the south-western part of the complex of the legionary bath extends farther to the east. Its archeological study will continue in 2001; this will provide more data on its layout, appearance and purpose. We may now conclude that the complex of the rooms of the legionary bath was of the length of c.90m, or 300 Roman feet, along the north-south axis. The width of the studied area along the west-east axis is more than 55m, or 180 Roman feet. Thus, the total area of the complex of the rooms of the legionary bath exceeds 4950sq m. The data collected so far substantiate the postulate that in the early C2nd, the complex of the legionary bath took the area of an entire residential quarter inside the legionary camp.

Fragments of floors and hypocaust spaces of other rooms, which were of similar designs, stratification and chronology as those described earlier, were discovered during other probing excavations south of the remains of the palaestra or basilika diskoperta which are discussed above.

Incidentally, the issue of the spatial layout of the western edge of the quarter occupied by the legionary bath in Novae, still remains to be investigated. In the years 1999 and 2000, further archaeological data on the spatial arrangement of this part of the complex of public buildings were acquired. During the exploration, we discovered the foundations and lower above ground parts of three huge rectangular stone pillars. The first, easternmost one, is 2.08m long and 1.46m wide; it has been preserved up to the height of 0.64-1.56m. It is made of classically rusticated stone blocks, laid in irregular stripes. The dimensions of the blocks are 0.62 x 0.50-0.95 x 0.46 x 0.75m, and they have been carefully fitted to one another.

In the south-western part of the complex of the bath, a room was unearthed which at a later stage of its existence had been divided into two smaller rooms by means of a wall. The western is smaller, of the dimensions of 2.7m (northern wall), 2.67m (southern wall), 2.05m (western wall) and 1.9m (eastern wall). [The eastern wall was added at a later time]. On the entire exposed floor of the interior, 12 brick pilae are preserved; the southern wall features a vaulted opening, which suggests that the praefurnium should be located in this direction. The pilae, extant up to a maximum height of 0.6m, consist of a base, made of a ceramic slab of the dimensions of 0.4 x 0.4 x 0.07m, on which another square slab is placed, 0.28m x 0.04m, and on top of this, several bricks of 0.19m x 0.085m, laid one on another. A vaulted brick sewer was found in the southern wall, 0.64m wide, 0.9m high, and 1.13m deep. In the year 2000, a further eastern section of this room was explored. The northern wall was exposed along a length of 3m, the southern, along a length of 4.4m, and the western, along a length of 1.9m. In the southern wall, a vaulted sewer was revealed, 0.65m wide, 0.88m high, 0.96m deep, an opening, apparently used for ventilation, 0.24m wide, 0.21m high, and 1.05m deep, and an entrance 1.98m wide with a preserved monolithic threshold 0.16m thick. A further 18 supporting posts were also unearthed, extant up to a maximum height of 0.6m. Each post consists of a base, made of a square ceramic slab of 0.4m x 0.06m, on which further square slabs are placed, between 0.27–0.29m square and 0.04 or 0.07m. The original southern wall of the room was exposed in certain locations up to a height of

Another pillar, 5.35m long (along the west-east axis) and 1.46m wide, was discovered in 1997, and the third one, 5.3m long and 1.45m wide, in 1998. All three pillars have profiled foundation offsets on their longer sides. Their ends at the shorter sides had also the function of fender beams, guarding the pillars from the traffic of vehicles near and among them. The distances between the pillars are similar, amounting to between 5.89 and 5.91m. The discovered foundations and lower parts of three rectangular pillars apparently are remnants of an enormous aqueduct, supported by arches; we believe that this structure supplied water to the legionary bath and the nymphaeum, although our knowledge of the ancient Novae is still too slight to make this postulate more than an assumption. Similar facilities are the remnants of the nymphaeum in the camp of the legio III Augusta in Lambaesis and the so-called Septizonium in the civilian city of Lambaesis.11 The Septizonium, dating to the second

10 Nielsen based her typology on Krenckner’s classification: cf. Krencker, Kruger, Lehmann & Wachtler 1929; Nielsen 1990: I 76–80.

11

650

The author expresses his gratitude to Prof. Tadeusz Sarnowski for

Andrzej B. Biernacki: The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior) quarter of the C3rd AD, had a dedicated aquaeduct which supplied water to it (Rakob 1979: 383, 394-396, abb. 5-7). Similarities between the architectural designs and floor layouts of the two facilities are striking. Nielsen (1990: I 166), discussing the spatial arrangements of numerous legionary baths, emphasizes that “nymphaea could be associated with thermae, in a variety of forms. They are seen only in connection with the palaestra, and could form an imposing frontage facing the natatio.” In Novae, the enormous structures discovered directly west of the palaestra warrant - as we believe - a tentative supposition that an immense nymphaeum was located here, supplied with water by an aqueduct. Both this arrangement and the aqueduct itself would be the first finds of this type encountered in the Roman provinces on the lower Danube. Among the essential objectives of our archaeological exploration in Novae in the years 2001–2005 will be a complete spatial reconstruction of the western edge of the quarter occupied by the legionary bath, including the palaestra and other auxiliary rooms and structures (the nymphaeum (?) and the aqueduct (?)).

Istrum was rapidly growing. A. Poulter postulates that it was at that time that the essential edifices, columned porticoes and the monumental entrances to the agora in Nicopolis-ad-Istrum were erected (Poulter 1995: 10–11; 1994: 11–13). It is now virtually incontestable that settlers from Asia Minor played the key rôle in the history of the city, the emergence of its spatial layout and the applied orders of architecture and stone material. By no means is it a coincidence that the agora in Nicopolis-ad-Istrum is most similar to those in the cities in western Turkey which are of a Greek origin. Direct evidence of the impact of the culture of Asia Minor may be found in the inscriptions discovered in Nicopolis which attest to the long activity of the stonemasons from Nicomedia and house builders (domotektoi) from Nicaea in that area (Mihailov 19561971: II 674; II 690). After the period of devastation in the aftermath of the invasion of the Kostoboks in 170, a second period of demand for stone from Hotnitsa followed, both at Nicopolis-ad-Istrum and in Novae. The reign of Commodus, and then of the Severi, saw in both cities a period of prosperity and, consequently, of extensive consumption of organogenous detrital limestone from Hotnitsa.12 Thus, business and trade relations naturally developed between Nicopolis-ad-Istrum and the Roman legionary camp, located at the fairly short distance of approximately 45 Roman leagues. It is fully plausible that the citizens of Nicopolis, who themselves were settlers from Asia Minor, transferred the idea of an establishment which combined the functions of a bath and a gymnasium, to the legionary camp in Novae between the years 130 and 160. We hope that the coming excavation seasons in Novae will provide complete archaeological and architectural evidence, and thereby make it possible to reconstruct the spatial layout of the complex of the legionary bath in Novae, and gain a deeper insight into its typological classification and chronology.

Based on the studies of such parts of the complex of the legionary bath in Novae as have been exposed so far, yet another postulate may be put forward. Namely, the buildings in the quarter of the legionary camp in Novae might have constituted a combined architectural-andfunctional complex of a bath and a gymnasium. Obviously, this would not be such a typologically pure design as those encountered in Asia Minor (Nielsen 1990: I 105–113), but an attempt at introducing such facilities into the province of Moesia Inferior. To sum up, let us once again emphasize such features as the total area of the complex of buildings exceeding 4950sq m, the central location of the palaestra, the parallel row layout of the double caldarium with the alvei, the room which provided communication between the apodyterium and the frigidarium bypassing the palaestra, the outdoor natatio pool and the large interiors in the southern part of the complex. All of these are also seen in the Asia Minor type of the bath-cum-gymnasium. The assumption that in Novae that a similar complex of edifices was erected, may be substantiated by the following circumstances. We know of an eminent instance of the architectural complex of a bath-cum-gymnasium in Odessos (now Varna), whose first stage is dated to AD150 and the rebuilding, c.300 (Nielsen 1990: II 45, 208, fig. 250). We also have evidence of close business and trade relations between Novae and Nicopolis-ad-Istrum (Nikyup). Let us juxtapose certain historical facts and data collected during the many years of the archeological exploration of both cities by discussing the contemporaneous developments in Nicopolis-ad-Istrum. Between 130 and 160 (the reigns of Hadrian and Antoninus Pius), the recently founded city of Nicopolis-ad-

Bibliography Barbulescu M. 1997 Das Legionslager von Potaissa (Turda). (Zalǎu). Biernacki A.B. & Skoczylas J. in press The classification of rock material in juxtaposition with the typology of the inscribed pedestals in Novae. In the proceedings of a conference, The Roman and late Roman city, held at Veliko Tărnovo (Bulgaria) by the University of Nottingham (Great Britain). Krencker D., Kruger E., Lehmann H. & Wachtler H. 1929 Die Trierer Kaiserthermen I. (Augsburg). Mihailov G. (ed.) 1956-1971 Inscriptiones Graecae in Bulgaria Repertae, I-IV. (Sofia). Nielsen I. 1990 Thermae et Balnea. The architecture and cultural history of Roman public baths. Vols. I-II. (Aarhus). Pawlak P. 1997 Das platische Programm der Wasserkaskade von Novae. In A.B. Biernacki & P. Pawlak (edd.) Late Roman and early Byzantine cities on the lower Danube. (Poznań): 95-97.

turning his attention to this possible purpose of the north-western part of the complex of buildings west of the principia in Novae: Rakob 1979: 375–397, taf. 103–117 (with a bibliography of previous literature).

12 For a discussion of the stone architectural details and elements from Hotnitsa found in Novae, see Biernacki & Skoczylas in press.

651

Limes XVIII

Poulter A. 1994 Nikopolis ad Istrum - anatomija na grăkorimskija grad Bulletin. Museum of History Veliko Tărnovo, IX: 11–13. Poulter A. 1995 Nicopolis ad Istrum: A Roman, late Roman and early Byzantine city. Excavations 1985-1992. (London). Rakob F. 1979 Das Groma - Nymphaeum im Legionslager von Lambaesis. Mitteilungen des Deutschen Archäologischen Instituts, Römische Abteilung, Band 86: 375–397, taf. 103–117. Vetters H. 1953 Das Legionsbad von Lauriacum. Forschungen in Lauriacum, I.

652

Andrzej B. Biernacki: The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

Fig. 1. A fragment of the Tabula Peutingeriana, with a section of the Lower Danubian limes and the location AD NOVAS.

Fig. 2. A map of Bulgaria and the region of Novae. Drawing by L. Fijał.

653

Limes XVIII

Fig. 3. A schematic street map of Novae. Drawing by A.B. Biernacki.

654

Andrzej B. Biernacki: The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

Fig. 4. Novae. General plan of the legionary baths. Al: alveus; C: caldarium; S/C: sudatorium/caldarium II; S: sudatorium; T: tepidarium; PR: praefurnium A: apodytherium; P/B: palaestra-basilica discoperta; F: frigidarium; N: natatio; AQ: the aqueduct. Drawing by A.B. Biernacki.

655

Limes XVIII

Fig. 5. Novae. The caldarium of the legionary baths, later rebuilt as a part of the late 3rd century public bath. Photograph by P. Namiota.

Fig. 6. Novae. The praefurnium of the legionary bath, rebuilt later as a part of the late 3rd century public bath. View from the north-west. Photograph by P. Namiota.

Fig. 7. Novae. A part of the legionary bath with the caldarium and the sudatorium/caldarium II. View from the south-east. Photograph by P. Namiota.

656

Andrzej B. Biernacki: The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

Fig. 8. Novae. The natatio of the legionary bath. Stone seats and stairs leading from the frigidarium into the pool. View from the north. Photograph by P. Namiota.

Fig. 9. Novae. A sewer which, among other purposes, was used for removing water from the natatio. View from the west. Photograph by A.B. Biernacki.

Fig. 10. Novae. A stone pillar of the ?aqueduct, with a wall closing in the space between pillars and added later. View from the north-east. Photograph by P. Pawlak.

Fig. 11. Novae. The aqueduct (?). The above surface part of a pillar with a foundation offset (a fender beam), made of rusticated stone blocks. View from the north. Photograph by P. Pawlak.

657

Limes XVIII

Fig. 13. Bumbesti. Castellum baths. After D. Tudor Castra Daciae Inferioris I. Castrele romane de la Bumbesti-Gorj. (1940): Fig. 12.

Fig. 12. Potaissa (Turda). A plan of the bath with the palaestra. After M. Barbulescu Das Legionslager von Potaissa (Turda). (1997): Abb.19.

Fig. 14. Sarmizegetusa (Ulpia Traiana) City Baths, After D. Tudor Arheologia Romana. (Bucuresti 1976): fig. 49.

658

Andrzej B. Biernacki: The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

Fig. 15. Lauriacum (Lorch). Legionary Baths. After A. Vetters Das Legionsbad von Lauriacum. Forschungen in Lauriacum I. (1953): Abb. 24.

Fig. 16. Colonia Ulpia Traiana (Xanten). City Baths. After H. Hinz Die Colonia Ulpia Traiana bei Xanten in Germania Romana I. (Heidelber): Abb. 11.

Fig. 17. Bulla Regia (Hammam-Barradji). Julia Memmia Baths. After A. Beschaouch, R. Hanoune & Y. Thebert Les Ruines de Bulla Regia. (Rome 1977): fig. 15.

659

Limes XVIII

Fig. 18. Odessos (Warna). Bath – gymnasium. After B. Böttger Die römischen Thermen in Varna. Altertum 23 (1977): abb.1.

Fig. 19. Lambesis (Lambese). Large Baths. After D. Kencker, E. Krüger, H. Lehmman & H. Wachtler Die Trierer Kaiserthermen I. (Augsburg 1929): Abb. 295

660

Andrzej B. Biernacki: The Roman legionary bath from the C2nd AD in Novae (Moesia Inferior)

Fig. 20. Lambesis City. General plan of Septizonium. 1-4 aqueduct. After F. Rokob Das Groma – Nymphaeum im Legionslager von Lambaesis. Mitteilungen des DAI, Römisches Abteilung 86 (1979): abb.5.

Fig. 21 Lambesis City. Septizonium. Isometric reconstruction. After F. Rokob Das Groma – Nymphaeum im Legionslager von Lambesis, Mitteilungen des DAI, Römisches Abteilung 86 (1979): Abb. 7.

661

Limes XVIII

Fig. 22. Lambesis. Groma-Nymphaeum. Isometric reconstruction. After F. Rokob Das Groma – Nymphaeum im Legionslager von Lambaesis. Mitteilungen des DAI, Römisches Abteilung 86 (1979): abb.19.

662

Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes? Gerda v. Bülow Die immer wieder Verwirrung stiftende scheinbare Inkonsequenz vieler antiker, ganz besonders spätantiker Autoren bei der Benennung von Siedlungen als „Stadt“ (colonia, municipium, polis, civitas...) im Sinne einer Zivilsiedlung unterschiedlicher Rechtsstellung oder als „Militärlager“ (castra, castellum, phrourion, burgus...) ist ein Faktum, mit dem der Archäologe leben muß. Es wird kaum jemals möglich sein, die vielen verschiedenen Begriffe eindeutig genug zu definieren, um sie in ihrer Bedeutung klar gegeneinander abgrenzen zu können.1 Allem Anschein nach sind sie auch nicht ohne weiteres miteinander vergleichbar, weil sie entweder eine bestimmte Siedlungsform oder einen festgelegten Rechtsstatus oder auch einfach nur die tatsächliche Funktion eines Ortes ausdrücken (Tarpin 1999: 1–10). Außerdem haben sich die Dinge, die sie bezeichnen sollen, d. h. die Orte selbst im Laufe ihrer Existenzzeit in ihrem iuristischen Status, in der Zusammensetzung ihrer Bewohnerschaft und in ihrem äußeren Erscheinungsbild ebenso verändert wie die Begriffsinhalte, so daß in den Beschreibungen die Unsicherheit bei der Wahl der Termini ständig wuchs.

Quellen erwähnt, doch diese stammen einerseits aus der Frühzeit und andererseits aus der Endphase seiner Existenz, so daß sie für die hier angestrebte Betrachtung aussagefähig sind. Abgesehen von der Nennung auf der Tabula Peutingeriana wird Iatrus mit dem Zusatz castellum in der Notitia Dignitatum (Or. XL.8) aufgeführt als Standort eines cuneus equitum scutariorum im Zuständigkeitsbereich des dux Moesiae secundae (Ivanov 1966a: 5-6). Über die zeitliche Einordnung des in der Notitia dignitatum dargestellten Zustandes einzelner Grenzabschnitte ist wiederholt nachgedacht worden. Für den hier relevanten niedermösischen Limes ist es nicht auszuschließen, daß die Angaben auf die konstantinische Zeit zu beziehen sind.3 Die Gründung von Iatrus erfolgte nach Aussage des Münzspiegels in der gemeinsamen Regierungszeit von Konstantin I. und Licinius, im zweiten Jahrzehnt des 4. Jh (Schönert-Geiß 1979: 167-209; 1991: 213-240; Döhle 1995: 9; cf. Abb. 1). Bereits T. Ivanov hat auf die große Ähnlichkeit der Festungsmauer von Iatrus mit der von Tropaeum Traiani (Provinz Scythia), welche durch eine Inschrift in die Jahre 315/317 datiert werden kann (Papuc 1974: 325-337), hingewiesen und nimmt die gleiche Entstehungszeit auch für Iatrus an (Ivanov 1966b: 38-40). Eine Gegenüberstellung mit einem „klassischen“, d. h. kaiserzeitlichen Kastellplan läßt grundlegende Unterschiede deutlich hervortreten: Der Verlauf der Festungsmauer von Iatrus orientiert sich in erster Linie an den Geländegegebenheiten, wobei man aber zugleich versucht hat, die einzelnen Mauerabschnitte geradlinig zu führen und aufeinander zu beziehen, so daß ein unregelmäßiges Vieleck mit annähernd parallel verlaufenden Mauerstrecken entstanden ist (Ivanov 1966b: 23-24). Charakteristische Unterschiede zu kaiserzeitlichen Anlagen bestehen in der Massivität der Festungsmauer (Dicke der Mauer von Iatrus etwa 3m), in der Form der Festungstürme und ihrer Verteilung entlang der Kurtine4 sowie darin, daß nur ein einziges Tor den Zugang ermöglichte und daß dieses Tor überdies durch einen Mauereinzug und zusätzliche Türme außerordentlich stark gesichert war.

Dazu kommt aber noch ein anderer Gesichtspunkt: Betrachtet man beispielsweise Prokops Abhandlung de aedificiis, so gewinnt man den Eindruck, daß der Autor manchmal den Begriff allein aus ästhetischen Erwägungen gewechselt oder an anderer Stelle mehrere Orte unter einem Begriff zusammengefaßt hat, der, streng genommen, gar nicht für alle gleichermaßen zutraf. Zuweilen hat der Autor auch dieselben Ortsnamen in unterschiedlichen Zusammenhängen mit verschiedenen Bezeichnungen charakterisiert. Abgesehen von den künstlerischen Beweggründen beruht diese Beliebigkeit in der Wortwahl auf einer Angleichung des ihnen zugrundeliegenden Inhalts, so daß sie in gewissem Grade austauschbar wurden. Diese Frage kann und soll hier nicht als philologisches oder semantisches Problem behandelt werden, sie soll vielmehr den Hintergrund dafür bilden, um an einem konkreten Beispiel die in den antiken Schriftquellen überlieferten Termini genauer zu definieren, indem diese mit bekannten archäologischen Befunden in Beziehung gesetzt werden.2 Ausgangspunkt dieser Überlegungen ist Iatrus an der unteren Donau, im heutigen Bulgarien, das allgemein als spätantikes Limeskastell bezeichnet wird (Iatrus-Krivina 1966; Iatrus–Krivina I–V).

Die durch die einzelnen Mauerabschnitte vorgegebene Richtung wird bis zu einem gewissen Grade in der Ausrichtung der Innenbebauung wiederaufgenommen, 3

Sarnowski 1985; van Berchem 1952: 90, nimmt an, daß der Truppenbestand konstantinischer Zeit bis zum Ende des 4. Jh. gültig blieb; Hoffmann 1969: 154; Zahariade 1988: 187–192 (englisches Resumee), datiert dagegen die Angaben für die beiden Provinzen um 395; Dietz 1993: 279–329. 4 Der für Iatrus charakteristische, besonders große viereckige Turm in der Mitte der Südmauer, der von U- und fächerförmigen Türmen in dichter Abfolge flankiert wird, findet sich nicht nur in Tropaeum Traiani an der Westseite wieder, sondern auch in anderen spätantiken Festungsanlagen an der unteren Donau, s. dazu Scorpan 1980: 190, Taf. 46; Ivanov 1980: 30, Abb. 10 (bulgarisch); Ivanov & Stojanov 1985: 14, Abb. 9.

Iatrus wird namentlich zwar nur in wenigen antiken 1

Von den zahlreichen Arbeiten zu diesem Thema sei hier nur angeführt: Ravegnani 1982, wo viel ältere Literatur zitiert ist. Speziell für die Provinzen an der unteren Donau beschäftigt sich Torbatov 2000a, mit dieser Frage. 2 Damit wird eine Anregung von P. Schreiner aufgegriffen, cf. Schreiner 1986, 26 mit Anm. 8.

663

Limes XVIII

wobei zu berücksichtigen ist, daß der Verlauf der Mauer an der Nordseite nicht mehr feststellbar ist.5 Die Innenbebauung von Iatrus läßt für die Gründungsperiode insgesamt eine einheitliche Planung erkennen, die noch wesentliche Elemente der kaiserzeitlichen Kastellarchitektur aufgreift, gleichzeitig aber auch signifikant davon abweicht: Im Zentrum erheben sich die principia, welche durch die von Kolonnaden flankierte Hauptstraße mit dem einzigen Kastelltor verbunden sind (Döhle 1995: 25-26). Weitere Bauformen der kaiserzeitlichen Militärarchitektur sind das praetorium (Stančev 1999: 151-154), Kasernen hinter den principia (Döhle 1999: 141-150) und tabernae entlang der Hauptstraße (v. Bülow 1999: 155-163).

über den Kasernenruinen entstanden gehöftartige Baukomplexe, die sowohl Wohnräume für Einzelfamilien wie auch Wirtschaftsräume und kleine Werkstätten umfaßten. Auch die Principia wurden als administratives, wiewohl auch geistiges Zentrum des konstantinischen Kastells aufgegeben (Döhle 1991: 42-46). Ihre Integrationsfunktion ging zumindest teilweise auf eine christliche Kirche über, welche am höchsten Geländepunkt innerhalb des Mauerringes errichtet wurde. Es ist also anzunehmen, daß es seit dem Ende der Siedlungsperiode A in Iatrus keine Verwendung mehr für derartige militärische Funktionsbauten gegeben hat. Und das steht in einem gewissen Widerspruch zu den Angaben in der Notitia Dignitatum, wenn man den dort erwähnten cuneus equitum scutariorum als eine reguläre Truppeneinheit versteht.

Insgesamt gesehen, tritt der militärische Charakter von Iatrus in der Struktur der Innenbebauung und in einzelnen Gebäudetypen deutlich hervor. Dieser Befund läßt sich versuchsweise mit der nach der Notitia Dignitatum hier stationierten Heeresabteilung von ungefähr 500 Bogenschützen berittenen in Beziehung setzen, und die Bezeichnung als Kastell scheint durch den Baubefund gerechtfertigt.

Andererseits erhielt Iatrus in der Siedlungsphase B außer der christlichen Basilika zwei weitere Neubauten, je ein langgestrecktes, zweischiffiges Bauwerk an der südwestlichen Festungsmauer und im Westteil nahe dem Jantraufer. Sie werden als Lagerhäuser zur Aufnahme von Lebensmittelvorräten wie auch von Ausrüstungsgegenständen und Kriegsgerät (horrea) gedeutet.7 Beide Gebäude wurden im frühen 5. Jh., am Ende der Siedlungsphase B zerstört und ihre Ruinen in eine grundsätzlich anders strukturierte Bebauung eingebunden (Döhle 1966a: 89–93; Mitova-Dzhonova 1968: 13–23). Die Deutung als horrea ist daher zwar als sehr wahrscheinlich anzusehen, aber mangels aussagekräftiger Funde und Befunde nicht mehr zu beweisen.8

Es sei noch einmal darauf hingewiesen, daß es für die These, in der Notitia Dignitatum sei für diesen Limesabschnitt der Zustand in konstantinischer Zeit wiedergegeben, keine schlüssigen Beweise gibt. Aber das offensichtlich von militärischen Funktionsbauten geprägte Siedlungsbild von Iatrus in ebendieser Zeit kann durchaus als Indiz dafür herangezogen werden. Die archäologischen Befunde sind mit der Charakterisierung des Ortes als castellum vereinbar.

Diese drei Großbauten stellen im Siedlungsbild von Iatrus am Ende des 4. Jh. einen architektonischen Kontrast zu der ansonsten kleinteiligen und wohl überwiegend einstöckigen Bebauung des ummauerten Areals dar. Wegen ihrer Größe und wegen ihrer sorgfältigen Bauweise müssen die beiden Horrea ebenso wie die Basilika als öffentliche Gebäude gelten, deren Errichtung eine übergeordnete, d. h. wohl staatliche Planung zugrunde gelegen haben dürfte. Diese Befunde lassen die Interpretation zu, daß Iatrus auch nach dem Abriß der militärischen Funktionsbauten und der Errichtung von eher zivil geprägten Wohn-Wirtschaftsgebäuden eine ziemlich genau bestimmbare Aufgabe im Verteidigungssystem am mösischen Grenzabschnitt zu erfüllen hatte, und zwar als befestigtes Nachschublager, das unter militärischer Aufsicht gestanden hat. Darüber hinaus erfüllte der Ort aber noch eine weitere wichtige Funktion im Grenzsicherungssystem, die sich aus seiner geographischen Lage an der Mündung eines schiffbaren Nebenflusses (Iatrus, heute Jantra) in die Donau ergab. Von hier aus konnte einerseits ein gewisser Bereich im Grenzvorfeld am nördlichen Donauufer überblickt werden.

Geht man dagegen davon aus, daß in der Notitia Dignitatum der Zustand des untermösischen Limes im späten 4. bzw. frühen 5. Jh. überliefert ist, so müssen diese Angaben auf die Siedlungsperiode B/C, Phase B in Iatrus bezogen werden (zur Periodisierung cf. v. Bülow 1995a: 29–31: Abb.2). Nach dem Münzspiegel ist der Beginn der Siedlungsperiode B/C in das dritte Drittel des 4. Jh. zu datieren (Döhle 1995: 25; v. Bülow 1994: 10-14) und das Ende der Phase B um 410.6 Der Übergang von der Periode A zur Periode B/C ist nach den archäologischen Befunden nicht durch eine Zerstörung der älteren Anlage hervorgerufen worden. Jedoch war zu beobachten, daß fast sämtliche Bauten aus der Gründungszeit aufgegeben und die Mauern abgetragen worden sind, oft bis zu den untersten Fundamentlagen. Die so gewonnenen Steine wurden sogleich für die Errichtung neuer Gebäude eingesetzt. An Stelle des Prätoriums, der Tabernen und 5

Döhle 1995: 12-13; Inzwischen ist im Nordostbereich des ummauerten Areals eine Gebäudeecke freigelegt worden (Objekt XXX, Nordostecke), deren Mauerverlauf auf einen Straßenknick Bezug nimmt, welcher durch die Richtungsänderung in der Kurtine bedingt ist, s. dazu v. Bülow 1999: 156, Abb. 1. 6 Vgl. dazu den Hortfund von 260 Bronzemünzen, die im Zerstörungsschutt der Phase B im Obj. XLI zutage gekommen sind: Schönert-Geiß 1995: 131–133, wo noch eine inzwischen korrigierte Objektbezeichnung (Obj. XVIII) verwendet wurde.

7

Döhle 1966a: 57–136; 1966b: 137-151; 1979: 53-62; v. Bülow 1988: 65-72. Das Fehlen der für diesen Bautyp charakteristischen Kontraforce wird kompensiert durch die dichte Reihe massiver Stützpfeiler in der Längsachse beider Gebäude. 8 Das gilt gleichermaßen für die von V. Dinčev vorgeschlagene Deutung des Objektes VII als Kaserne, cf. Dinčev 1999: 165-174.

664

Gerda v. Bülow: Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes? Andererseits konnte und sollte der Zugang ins Provinzinnere durch das Jantratal gegebenenfalls abgeriegelt werden. Das heißt, daß die strategische Bedeutung von Iatrus für das Grenzsicherungssystem nach wie vor groß war. Daß sein architektonisches Erscheinungsbild dieser Funktion scheinbar nicht mehr entsprach, ist vielleicht damit zu erklären, daß die regulären Truppen, der cuneus equitum scutariorum der Notitia Dignitatum, aus Iatrus abgezogen und durch eine grundsätzlich anders strukturierte und organisierte Besatzung ersetzt worden war. Es liegt nahe, in dieser neuen Bewohnerschaft Föderaten zu vermuten. Dafür sprechen archäologische Indizien aus Iatrus selbst, sowie auch die historische Situation in den Provinzen an der unteren Donau seit den Gotenkriegen der 60er Jahre und nach der schweren Niederlage des römischen Heeres gegen die Goten in der Schlacht bei Adrianopel (9 August 378: Zosimos XV.34.6; Ammianus Marcellinus XXXI 13.12; vgl. auch Vetters 1950: 30-32; Wolfram 1980: 153-163).

Werkstatt- und Vorratsräumen mit der Wirtschaftsweise von Familien verbinden. Auf der anderen Seite weisen die zwei großen Horrea auf die Unterhaltspflicht des spätantiken Staates für die Soldaten und Föderaten hin (v. Bülow 1987: 215–224; 1998: 145-157) und können somit als Illustration zu dem Bericht von Ammianus Marcellinus über die Bemühungen der Kaiser Valentinian I. und Valens dienen, durch die Errichtung befestigter Nachschubbasen die Versorung sowohl der Besatzungen in den Grenzstationen wie auch des mobilen Feldheeres zu gewährleisten (Ammianus Marcellinus XXVIII. 2.1; 3.7). Unter den Funden aus der Siedlungsperiode B/C finden sich zahlreiche Militaria, von denen viele schon beschädigt und unbrauchbar waren, bevor sie in die Erde gelangten (Gomolka-Fuchs 1999a: 514). Eine weitere Gruppe von Kleinfunden aus dieser Siedlungsperiode ist als völkerwanderungszeitlich und nicht-römischer Provenienz anzusprechen und kann mit einiger Vorsicht für Angehörige fremder Ethnien in Iatrus sprechen (Gomolka– Fuchs 1999a: 515; 1995: 91–105; 1999b: 189–194). Diesen beiden Faktoren steht gegenüber, daß der Anteil von Amphoren am gesamten Keramikbestand aus dieser Siedlungsschicht mit über 43% sehr hoch ist und daß im Formenbestand dieser Amphoren kein wesentlicher Unterschied zwischen den Perioden A und B/C zu erkennen ist (v. Bülow 2000: 211–216). Amphoren gelangten als Transportgefäße mit allen möglichen Nahrungsmitteln aus verschiedenen Reichsteilen nach Iatrus und können so einen provinzübergreifenden Gütertransport belegen. Dieser ist angesichts des zu erschließenden großen Umfang ohne eine zentrale Organisation kaum vorstellbar. Es liegt also nahe, die zahlreichen Amphorenfunde als Hinweise auf das Funktionieren einer staatlich gelenkten Versorgung der Bewohner mit bestimmtem Produkten zu deuten.

Der Föderatenstatus beinhaltete für die Betroffenen - oder sollte man sagen: Begünstigten - sowohl einen militärischen wie auch einen zivilen Aspekt.9 Durch den Vertrag wurden sie verpflichtet, die militärische Sicherung ihres Ansiedlungsortes und seiner Umgebung zu übernehmen. Dafür wurden sie in das Versorgungssystem des römischen Heeres eingebunden, d. h. sie erhielten außer Kriegsgerät und Ausrüstungsgegenständen auch Lebensmittellieferungen, sofern die allgemeine militärische Situation in den entsprechenden Provinzen solche Transporte zuließ (Velkov 1980: 171-198). Außerdem hatten sie Anspruch auf regelmäßige Geldzuwendungen.10 In den Föderatenverträgen, die Kaiser Theodosius I. nach der schweren Niederlage der Römer bei Adrianopel (378) mit den Goten schloß, mußte er ihnen darüber hinaus zusichern, daß sie Führern aus ihren eigenen Reihen unterstellt wurden, so daß sie eine noch größere Selbständigkeit innerhalb des spätantiken Heeres und Staates erhielten (Wolfram 1980: 153-163; Stallknecht 1967: 74-87; Schmidt 1934: 419-422). Zu den Besonderheiten der Föderatenansiedlung gehörte, daß nicht nur wehrfähige Männer, sondern auch deren Familien Wohnrecht am Ort erhielten. Die Familie wurde so zum bestimmenden Element für die Organisation des gesellschaftlichen Lebens in einer solchen Siedlung.

Eine weitere Quelle, in der Iatrus namentlich erwähnt wird, findet sich im Werk des Prokop, wo der Name mit dem Terminus phrourion verbunden ist (Prokop IV.7.6; Ivanov 1966a: 6). Dieser Begriff wird allgemein als gleichbedeutend mit dem lateinischen castellum angesehen (Ravegnani 1982: 273; Dölger 1959: 70-71). Prokop berichtet, daß Iustinian in den phrouria Iatrus und Tigras am mösischen Donauabschnitt die „baufälligen Teile durch einen Neubau ersetzt“ habe. Mit dieser Nachricht läßt sich die letzte der vier Siedlungsperioden von Iatrus auf Grund von sicher stratifizierten Münzfunden verbinden (v. Bülow 1991: 6162; cf. Abb. 3). Betrachtet man jedoch den Plan der erhaltenen Baureste aus dieser Periode D2, so vermißt man zweifellos typische Militärbauten. Allein die mächtige und offenbar unter Iustinian ausgebesserte Verteidigungsmauer verlieh dem Ort das Aussehen einer Festung (v. Bülow 1995b: 61-66, Beilage 6). Die Innenbebauung dagegen bestand aus einigen wenigen Gebäuden in ortsüblicher Bauweise, d. h. aus Stein-Erde-Sockeln mit fachwerkartigen Aufbauten aus luftgetrockneten Lehmziegeln und Holzstützen. Zum größeren Teil war das

Die beiden Aspekte des Föderatenwesens, der militärische und der zivile, scheinen sich im Siedlungsbild von Iatrus seit der zweiten Siedlungsperiode, d. h. seit dem letzten Drittel des 4. Jh. widerzuspiegeln: Auf der einen Seite lassen sich die gehöftartigen Einzelkomplexe aus Wohn-, 9

Wolfram 1983: 5-35; Cesa 1984: 307-316; Claude 1988: 13–16; Mirković 1993: 425–434, vermeidet weitgehend den Begriff „Föderaten“. Darauf weisen die Funde von römischen/byzantinischen Goldmünzen hin, die nach allgemeiner Ansicht vorrangig zur Deckung der Zahlungsverpflichtungen an verbündete Barbaren geprägt worden waren, vgl. dazu Pohl 1980: 282. In diesem Sinne können auch die Goldmünzen aus Iatrus gedeutet werden: v. Bülow 1998: 145-157. 10

665

Limes XVIII

ummauerte Areal von hüttenartigen Bauten aus Holz und Lehm bedeckt. Die gleichmäßige Verteilung der von der Zerstörung übriggebliebenen Schuttschicht zeigt aber eindeutig, daß die gesamte Fläche innerhalb des Mauerringes bebaut gewesen ist, daß man also beim Wiederaufbau die Siedlung nicht, wie in vergleichbaren Fällen, verkleinert hat.11 Dominiert wurde das Siedlungsbild von Iatrus in iustinianischer Zeit von einer christlichen Basilika, der größten, die je in Iatrus gestanden hat (Ivanov 1966c: 168-170; Stančev 1995; 123-124).

Diese Zusammenstellung zeigt, daß sich eigentlich alle Siedlungen, an denen Iustinian Festungswerke wieder instandsetzen oder neu errichten ließ, an Plätzen von strategischer Bedeutung befanden. Das gilt vermutlich auch für diejenigen Orte, die von Prokop erwähnt, die aber bisher noch nicht lokalisiert worden sind. Andererseits sind durch archäologische Untersuchungen zahlreiche spätantike Festungsanlagen im Innern der Provinzen festgestellt worden, denen bislang kein antiker Name zugeordnet werden konnte (die Karte, Abb. 4, kann die auch in diesem Punkt keinen Anspruch auf Vollständigkeit erheben). Ergänzt man die Karte der von Prokop überlieferten Orte durch jene, so zeigt sich, daß in der Spätantike die gesamte Balkanhalbinsel südlich der Donau eine auf Konstantinopel bezogene Verteidigungszone gebildet hat. Nicht nur die eigentliche Grenze entlang der Donau und der Meeresküsten war befestigt, sondern auch die das Land durchziehenden Straßen, wobei man anscheinend den Gebirgspässen besondere Aufmerksamkeit widmete.

Vergleicht man die drei Periodenpläne von Iatrus miteinander, so zeigen sich strukturelle Gemeinsamkeiten in den Anlagen der Phase B und der Periode D2, während zwischen diesen beiden und dem der Periode A grundsätzliche Unterschiede in der Organisation der Innenbebauung hervortreten. Aber auf alle drei Siedlungsbilder wird in zeitgenössischen Quellen der Begriff „Kastell“ angewendet. Was die von Prokop gebrauchte Bezeichnung phrourion betrifft, so ist bedeutsam, daß er durchaus auch andere Begriffe einsetzt, um den Status oder die Funktion eines Ortes zu charakterisieren (Beševliev 1970, passim). Allerdings geht er dabei nicht so konsequent vor, daß für den heutigen Leser ein klarer Sachbestand entstünde. Es kommt wiederholt vor, daß ein und derselbe Name an verschiedenen Textstellen unterschiedlichen Kategorien zugeordnet wird. Es scheint also zu Prokops Zeiten keine allzu strenge Abgrenzung zwischen der inhaltlichen Bedeutung der verschiedenen Bezeichnungen gegeben zu haben.

Wenn also Prokop von phrourion spricht, hat er wohl weniger das architektonische Erscheinungsbild einer Festung im Auge, sondern vielmehr deren strategischen Stellenwert innerhalb des flächendeckenden Verteidigungssystems. In diesem Sinne ist Iatrus, gelegen am Zusammenfluß von Donau und einem schiffbaren Nebenfluß, der tief in diese Pufferzone um Konstantinopel hineinführt, zweifellos als eine Militärstation, als Kastell zu bezeichnen, da die vorrangige Aufgabe seiner Bewohner, unabhängig von ihrer ethnischen Zugehörigkeit und ihrer rechtlichen Stellung, die Überwachung und gegebenenfalls Verteidigung dieser Einfallschneise war.

Von den aufgelisteten Orten in dem vorliegenden Gebiet konnte bis heute nur etwa die Hälfte lokalisiert werden (Abb. 4).12 Aus der Kartierung allein dieser Orte ergibt sich jedoch ein bemerkenswert einheitliches Bild: Genannt sind hier die meisten Limesstationen am südlichen Donauufer, ergänzt durch einige Gegenposten auf der Nordseite des Flusses. Dazu kommen etliche Städte an der westlichen Schwarzmeerküste, die Prokop meistens als phrouria auf führt. Das gilt auch für eine größere Zahl von Orten an der nördlichen Propontisküste, also im unmittelbaren Vorfeld von Konstantinopel, sowie für die Orte an der Südküste Thrakiens. Die von ihm verzeichneten Orte im Provinzinnern liegen, soweit sie überhaupt lokalisierbar sind, fast durchweg an Straßen, in erster Linie an der von Westen kommenden Hauptstraße über Naissus - Serdica - Philippopolis - Adrianopel nach Konstantinopel. Eine weitere wichtige Straße scheint danach auch von Odessus an der Schwarzmeerküste über Marcianopolis zur Donau beim Legionslager Durostorum geführt zu haben (Beševliev 1962: 57-80; Torbatov 2000b: 59-72).

Aufgrund der politischen Gesamtsituation der Balkanprovinzen hat es in iustinianischer Zeit keine klare Trennung mehr zwischen militärischen und zivilen Siedlungen gegeben. Vielmehr waren alle funktional in die großflächige Verteidigungszone eingebunden, unabhängig von ihrem iuristischen Status. Daher ist auch die Frage, ob Iatrus im 6. Jh. noch als „Kastell“ bezeichnet werden kann, nicht mehr von der Siedlungsstruktur, nicht von bestimmten Gebäudetypen oder davon abhängig, ob hier reguläre Soldaten untergebracht waren. Nachdem es am Beginn der Spätantike als Kastell an einem strategisch wichtigen Platz eingerichtet worden war, erfüllte es dieselbe Funktion in der Grenzsicherung, solange der militärpolitische Kontext, d. h. der Limes an der unteren Donau existierte, unbeschadet der nachgewiesenen Veränderungen im Baubestand sowie in Status und Herkunft der Bewohner. Diese Schlußfolgerungen aus der Gegenüberstellung von schriftlicher Überlieferung und archäologischen Befunden machen deutlich, daß ein spätantikes Kastell nicht nach den für die Prinzipatszeit gültigen Kriterien definiert werden kann. Ein spätantikes Kastell ist nicht an einer bestimmten Anordnung bestimmter Bautypen innerhalb einer Ummauerung zu erkennen oder daran, daß eine fest strukturierte, benennbare Heereseinheit hier untergebracht

11 Zu den Diskrepanzen zwischen nomineller und tatsächlicher Größe von Truppenverbänden in der Spätantike s. Castritius 1996: 215–234. 12 Auf dieser Karte sind die von Beshevliev 1970 publizierten Lokalisierungen eingetragen, ergänzt durch die Angaben der Tabula Imperii Romani 1999; v. Kampen; Scorpan 1980: 144-147, Taf. 1 und 2. Diese Karte kann keine Vollständigkeit beanspruchen, aber zumindest einige wesentliche Auswahlkriterien der Listen bei Prokop aufzeigen.

666

Gerda v. Bülow: Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes? war. Ein Ort ist vielmehr als Kastell anzusprechen, wenn ihm aufgrund seiner Lage eine besondere strategische Bedeutung zukam und damit die Bewohner dieses Ortes in erster Linie militärische Aufgaben im Verteidigungssystem des jeweiligen Verwaltungsbezirkes zu erfüllen hatten. So definiert, war Iatrus ein spätantikes Kastell am niedermösischen Limes.

strategische Bedeutung des Platzes nicht nur die wesentliche Voraussetzung für die Errichtung des spätantiken Kastells am Anfang des 4. Jh. war, sondern sie blieb auch existenzbestimmend bis zur Aufgabe von Iatrus in den ersten Jahren des 7. Jh.

Es gibt noch eine dritte Quelle, in der Iatrus gleich zweimal namentlich erwähnt ist, das Geschichtswerk des Theophylaktos Simokates (VII.2.13; Ivanov 1966a: 6-7). Dabei handelt es sich um einen Bericht über die byzantinisch-awarischen Kriege zwischen 582 und 602. Diese wurden im wesentlichen auf dem Territorium des heutigen Bulgarien ausgetragen und endeten mit der Aufgabe des Limes an der unteren Donau. Theophylaktos war kein unmittelbarer Augenzeuge dieser Ereignisse, sondern verfaßte seine Historien zwischen 630 und 640.13 Bei der Schilderung des Kriegsverlaufes kommen mehr als 30 Ortsnamen vor, von denen 24 am Südufer der Donau liegen, und die übrigen verteilen sich im Provinzinnern entlang größerer Heerstraßen (Schreiner 1986: 30-33). Die meisten von ihnen werden durch den Begriff polis charakterisiert, darunter auch Iatrus. Aus der zeitlichen Nähe zu den Listen des Prokop ergibt sich, daß dasselbe Siedlungsbild von Iatrus einmal mit phrourion und einmal mit polis verbunden erscheint. Dieser Widerspruch darf nicht überbewertet werden. Er weist wohl kaum auf einen kürzfristig vollzogenen Statuswechsel des Ortes hin, sondern signalisiert vielmehr, daß beide Begriffe synonym verwendet wurden. Darin spiegelt sich offenbar die Realität des politischen Lebens auf der Balkanhalbinsel in der 2. Hälfte des 6. Jh.: Der traditionelle Gegensatz von Militärlager und Stadt oder Zivilsiedlung war weitgehend aufgehoben. Alle oder doch die meisten Siedlungen hatten Festungscharakter angenommen. Besondere Bedeutung kam darunter denjenigen Orten/Festungen/Städten zu, die nicht allein der Bevölkerung Schutz in Gefahrensituationen boten, sondern denen aufgrund ihrer geographischen Lage eine aktive Rolle bei der Sicherung der Grenzen und großen Verbindungsstraßen zukam. In diesem Sinne besaßen die meisten Orte, die Theophylaktos Simokates als polis bezeichnet, eine herausragende strategische Bedeutung für beide kriegführenden Seiten. Das gilt auch für Iatrus an der Jantramündung, von wo aus eine direkte und relativ kurze Verbindung von der Donau über das Balkangebirge bis zur Hauptstraße von Italien nach Konstantinopel bestand (Schreiner 1986: 34, vgl. auch 27, Karte). Der Begriff polis bezeichnet bei Theophylaktos Simokates demnach recht allgemein einen wichtigen Ort, und das ist in dem gegebenen historischen Kontext wohl gleichzusetzen mit militär-strategischer Bedeutung. Insofern kann man eine weitgehende inhaltliche Übereinstimmung mit dem von Prokop gebrauchten Begriff phrourion annehmen.

Ammianus Marcellinus Römische Geschichte. Lateinisch und deutsch und mit einem Kommentar versehen von W. Seyfahrt. Bd. I – IV. (Berlin 1968 – 1971). Prokop de aedificiis. Werke. Griechisch und deutsch ed. Otto Veh. Bd. V: Bauten. (München 1977). Theophylaktos Simokates Geschichte. Übersetzt und erläutert von P. Schreiner. (Stuttgart 1985). Zosimos Historia nova. Edidit L. Mendelssohn. (Leipzig 1887).

Literatur

Beševliev V. 1962 Zur Geographie Nordost-Bulgariens in der Spätantike und im Mittelalter. Linguistique Balkanique 4: 57-80. Beševliev V. 1970 Zur Deutung der Kastellnamen in Prokops Werk „de aedificiis“. (Amsterdam). v. Bülow G. 1987 Wandlung der Siedlungsstruktur im Kastell Iatrus vom 4. zum 5. Jh., dargestellt am Beispiel eines Architekturkomplexes. Ratiariensia. (Bologna) 3/4: 215-224. v. Bülow G. 1988 Funktionsbestimmung archäologisch nachgewiesener Bauwerke durch Vergleich und Analogieschluß. Wissenschaftliche Zeitschrift der W. Pieck - Universität Rostock, G-Reihe 37: H. 2, 65-72 v. Bülow G. 1991 Die Bebauung des 6. Jh. über dem ehemaligen Zentrum des Kastells. In Iatrus – Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. IV: Ergebnisse der Ausgrabungen 1975–1981. (Berlin): 59-67. v. Bülow G. 1994 Das spätantike Kastell Iatrus am UnterdonauLimes in Bulgarien. Stand und Probleme der Erforschung. Bericht der Römisch-Germanischen Kommission 75: 5-22. v. Bülow G. 1995a Die Entwicklung des Siedlungsbildes von Iatrus in der Periode B/C. In Iatrus - Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. V: Studien zur Geschichte des Kastells Iatrus (Forschungsstand 1989). (Berlin): 29-53. v. Bülow G. 1995b Die Siedlungsperiode D 2 in Iatrus. In Iatrus Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. V: Studien zur Geschichte des Kastells Iatrus (Forschungsstand 1989). (Berlin): 61-66. v. Bülow G. 1998 Föderaten im spätantiken Limeskastell Iatrus (Moesia secunda)? In U. Peter (Hrg.) stephanos nomismatikos. Edith Schönert-Geiss zum 65. Geburtstag (Berlin): 145-157. v. Bülow G. 1999 Die Bebauung an der Nordseite der via praetoria in Iatrus. In G. v. Bülow & A. Milčeva (Hrsg.) Der Limes an der unteren Donau von Diokletian bis Heraklios. Vorträge der Internationalen Konferenz Svištov, Bulgarien, September 1998. (Sofia): 155-163. v. Bülow G. 2000 Anmerkungen zur Keramik aus dem apätantiken Limeskastell Iatrus (Moesia secunda). In RCRF Acta 36: 211-216. Castritius H. 1996 Die Wehrverfassung des spätrömischen Reiches als hinreichende Bedingung zur Erklärung des

Auf Iatrus bezogen, läßt sich also schlußfolgern, daß die 13 Schreiner Einführung. In Theophylaktos Simokates, Geschichte (übersetzt und erläutert von P. Schreiner; Stuutgart): 1-5.

667

Limes XVIII

Untergangs? Zur Interdependenz von wirtschaftlichfinanzieller Stärke und militärischer Macht. In: R. Bratož (Hrsg.) Westillyricum und Nordostitalien in der spätrömischen Zeit. (Ljubljana): 215-234. Cesa M. 1984 Überlegungen zur Föderatenfrage. Mitteilungen des Instituts für österreichische Geschichtsforschung 92: 307-316. Claude D. 1988 Zur Ansiedlung barbarischer Föderaten in der ersten Hälfte des 5. Jahrhunderts. In H. Wolfram & A. Schwarcz (Hrsg.) Anerkennung und Integration. Zu den wirtschaftlichen Grundlagen der Völkerwanderungszeit 400 - 600. Berichte des Symposions der Kommission für Frühmittelalterforschung. Mai 1986, Stift Zwettl, Niederösterreich. (Wien): 13-16. Dietz K. 1993 Cohortes, ripae, pedaturae. Zur Entwicklung der Grenzlegionen in der Spätantike. In K. Dietz, D. Hennig & H. Kaletsch (Hrsg.) Klassische Altertumskunde, Spätantike und frühes Christentum. A. Lippold zum 65. Geburtstag gewidmet. (Würzburg): 279-329. Dinčev V. 1999 Sur la caracteristique d’ Iatrus (deuxième moitié du Ive - début du Ve s.). In G. v. Bülow & A. Milčeva (Hrsg.) Der Limes an der unteren Donau von Diokletian bis Heraklios. Vorträge der Internationalen Konferenz Svištov, Bulgarien, September 1998. (Sofia): 165-174. Döhle B. 1966a Der Sektor VII. Klio 47: 57-136. Döhle B. 1966b Das römische Gebäude in Sektor I. Klio 47: 137-151. Döhle B. 1979 Untersuchungen im Bereich des Gebäudes VII (früher Sektor VII) in der Kampagne 1966. In Iatrus – Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. I: Ergebnisse der Ausgrabungen 1966-1973. (Berlin): 53-62. Döhle B. 1991 Die Principia (Objekt XX/XXII). In Iatrus Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. IV: Ergebnisse der Ausgrabungen 1975-1981. (Berlin): 35-58. Döhle B. 1995 Die Siedlungsperiode A in Iatrus. In Iatrus – Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. V: Studien zur Geschichte des Kastells Iatrus (Forschungsstand 1989). (Berlin): 9-28. Döhle B. 1999 Beobachtungen zur Bebauung und Bauabfolge im Westabschnitt von Iatrus. In G. v. Bülow & A. Milčeva (Hrsg.) Der Limes an der unteren Donau von Diokletian bis Heraklios. Vorträge der Internationalen Konferenz Svištov, Bulgarien, September 1998. (Sofia): 141-150. Dölger F. 1959 Die frühbyzantinische und byzantinisch beeinflußte Stadt (V.-VIII. Jahrhundert). In Atti del 3o Congresso internazionale di Studi sull’ alto Medioevo. (Spoleto): 65-100. Gomolka-Fuchs G. 1995 Die Kleinfunde und ihre Aussagen zur Bevölkerung von Iatrus. In Iatrus - Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. V: Studien zur Geschichte des Kastells Iatrus (Forschungsstand 1989). (Berlin): 81105. Gomolka-Fuchs G. 1999a Zur Militärbesatzung im spätrömischen Limeskastell Iatrus vom 4. bis zum 2. Viertel des 5. Jahrhunderts. Eurasia Antiqua 5: 509522. Gomolka-Fuchs G. 1999b Eine frühvölkerwanderungszeitliche

Nephritschnalle aus dem spätrömischen Limeskastell Iatrus in Nordbulgarien. In G. v. Bülow & A. Milčeva (Hrsg.) Der Limes an der unteren Donau von Diokletian bis Heraklios. Vorträge der Internationalen Konferenz Svištov, Bulgarien, September 1998. (Sofia): 189-194. Hoffmann D. 1969 Das spätrömische Bewegungsheer und die Notitia dignitatum, Bd. 1. (Düsseldorf). Iatrus-Krivina 1966 Das Limeskastell Iatrus in Moesia inferior (Ergebnisse der Grabungskampagnen 1958, 1960 und 1962). Klio 47. Iatrus-Krivina I – V Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau, Bd. I - V. Berlin 1979 – 1995. (Schriften zur Geschichte und Kultur der Antike Bd. 17). Ivanov T. 1966a Schriftquellen und geographische Karten zur Geschichte von Iatrus. Klio 47: 5-10. Ivanov T. 1966b Die Festungsmauer des Kastells Iatrus. Klio 47: 23-56. Ivanov T. 1966c Zwei altchristliche Basiliken des 4. - 6. Jh. in Sektor III. Klio 47: 153-191. Ivanov T. 1980 Abritus. Rimski kastel i rannovizantijski grad na Dolna Mizija. (Sofia) - in bulgarisch. Ivanov T. & Stojanov S. 1985 Abritus. (Rasgrad). v. Kampen A. n.d. Die Welt der Antike (Atlas antiquus). Taschenatlas, bearb. von A. v. Kampen, 13. Aufl. (Gotha o.J.). Mirković M. 1993 ‛γπήκοοι und σύµµαχοι. Ansiedlung und Rekrutierung von Barbaren bis zum Jahr 382. In K. Dietz, D. Hennig & H. Kaletsch (Hrsg.) Klassische Altertumskunde, Spätantike und frühes Christentum. A. Lippold zum 65. Geburtstag gewidmet. (Würzburg). 425-434. Mitova-Džonova D. 1968 Rannovizantijsko žilištno stroitelstvo v kastela Jatrus. Arheologija (Sofija) 10. 3: 13-23. Papuc G. 1974 Cîteva consideraţii privind construirea zidului de incită al cetăţii Tropaeum Traiani. Pontica 7: 325-337. Pohl W. 1980 Die Gepiden und die Gentes an der mittleren Donau nach dem Zerfall des Attilareiches. In H. Wolfram & F. Daim (Hrsg.) Die Völker an der mittleren und unteren Donau im 5. und 6. Jh. (Wien): 239-305. Ravegnani G. 1982 Kastron e Polis: ricerche sull’ organizzazione territoriale nel V secolo. Rivista di Studi Bizantini e Slavi 2: 271-282. Sarnowski T. 1985 Die legio I Italica und der untere Donauabschnitt der Notitia dignitatum. Germania 63/1: 107-127. Schmidt L. 1934 Geschichte der deutschen Stämme bis zum Ausgang der Völkerwanderung. Die Ostgermanen. (2. Auflage; München). Schönert-Geiß E. 1979 Die Fundmünzen von Krivina. In Iatrus Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. I: Ergebnisse der Ausgrabungen 1966–1973. (Berlin): 167-209. Schönert-Geiß E. 1989 Spätantike Befestigung und Frühmittelalterliche Siedlung ander unteren Donau. Band V. Studien zur Geschichte des kastells Iatrus Forschungsstand. (Berlin). Schönert-Geiß E. 1991 Die Fundmünzen. In Iatrus - Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. IV: Ergebnisse der Ausgrabungen 1975–1981. (Berlin): 213-240. Schönert-Geiß E. 1995 Der Münzfund von Krivina 1989. In

668

Gerda v. Bülow: Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes? Iatrus - Krivina V (s. Anm. 4): 131-133. Schreiner P. 1986 Städte und Wegenetz in Moesien, Dakien und Thrakien nach dem Zeugnis des Theophylaktos Simokates. In R. Pillinger (Hrg.) Spätantike und frühbyzantinische Kultur Bulgariens zwischen Orient und Okzident (Schriften der Balkankommission Antiquarische Abteilung Bd. 16). (Wien): 25-35. Scorpan C 1980 Limes Scythiae. British Archaeological Reports Int. Ser. 88. (Oxford). Stallknecht B. 1967 Untersuchungen zur römischen Außenpolitik in der Spätantike (306–395). (Bonn). Stančev D. 1995 Grabungen im Kastell Iatrus von 1984 - 1988. In Iatrus - Krivina. Spätantike Befestigung und frühmittelalterliche Siedlung an der unteren Donau. Bd. V: Studien zur Geschichte des Kastells Iatrus (Forschungsstand 1989). (Berlin): 123–130. Stančev D. 1999 Das Prätorium des Kastells Iatrus (Vorläufiger Bericht). In G. v. Bülow & A. Milčeva (Hrsg.) Der Limes an der unteren Donau von Diokletian bis Heraklios. Vorträge der Internationalen Konferenz Svištov, Bulgarien, September 1998. (Sofia): 151–154. Tabula Imperii Romani 1999 Tabula Imperii Romani K-35, Philippopolis, bearb. von T. Spiridonov und P. Vălev. (Sofia). Tarpin M. 1999 Colonia, Municipium, Vicus: Institutionen und Stadtformen. In N. Hanel & C. Schucany (Hrsg.) Colonia - municipium - vicus. Struktur und Entwicklung städtischer Siedlungen in Noricum, Rätien und Obergermanien. Beiträge der Arbeitsgemeinschaft „Römische Archäologie“ bei der Tagung des West- und Süddeutschen Verbandes der Altertumsforschung in Wien 21-23. 5. 1997. British Archaeological Reports Int. Ser. 783. (Oxford): 1-10. Torbatov S. 2000a Funkcionalno i tipologičesko sădăržanie na kăsnoantičnata fortifikacionna terminologija. Arheologičeski Vesti (Sofia) 1, Beilage. Torbatov S. 2000b The Roman road Durostorum Marcianopolis. Archaeologia Bulgarica 4.1: 59-72. van Berchem D. 1952 L’armée de Dioclétien et la réforme Constantinienne. (Paris). Velkov V. 1980 Themistius as a source of information about Thrace. In V. Velkov Roman cities in Bulgaria. Collected studies. (Amsterdam): 171-198. Vetters H. 1950 Dacia ripensis. (Wien). Wolfram H. 1980 Geschichte der Goten. (2. Auglage; München). Wolfram H. 1983 Zur Ansiedlung reichsangehöriger Föderaten. Erklärungsversuche und Forschungsziele. Mitteilungen des Instituts für österreichische Geschichtsforschung 91: 5-35. Zahariade M. 1988 Moesia secunda, Scythia şi Notitia Dignitatum. (Bukarest).

669

Limes XVIII

Fig. 1. Iatrus-Krivina, Plan der Siedlungsperiode A (1.H.4.Jh.)

Fig. 2. Iatrus-Krivina, Plan der Siedlungsphase B (Ende 4./ Anfang 5.Jh.)

670

Gerda v. Bülow: Iatrus - spätantikes Kastell oder befestigte Zivilsiedlung am Unterdonaulimes?

Fig. 3. Iatrus-Krivina, Plan der Siedlungsperiode D 2 (2.H.6.Jh.)

671

Limes XVIII

Fig. 4. Die Balkanprovinzen in spätantiker Zeit.

672

Archaeological survey on the Roman frontier on the lower Danube between Novae and Sexaginta Prista. Preliminary report (1997-2000) Sven Conrad & Dimităr Stančev Mehr als vierzig Jahre archäologischer Forschungstätigkeit im Legionslager Novae (bei Svistov, Bulgarien) und in dem c.15kms flußabwärts an der Jantra-Mündung gelegenen spätantiken Kastell Iatrus (beim Dorf Krivina) erbrachten grundlegende Beiträge zur Geschichte des römischen Limes an der unteren Donau. Seit einigen Jahren richtet sich der Blick verstärkt auf die Beziehungen der Befestigungsanlagen zu ihrem Umland. Im Mittelpunkt eines seit 1997 von der Römisch-Germanischen Kommission des Deutschen Archäologischen Instituts in Zusammenarbeit mit bulgarischen Archäologen durchgeführten und von der Deutschen Forschungsgemeinschaft unterstützten Projektes stehen neben der Erforschung der Siedlungstätigkeit im betreffenden Gebiet und ihres Einflusses auf die Umwelt auch die Erkundung von Limesanlagen einschließlich der Suche nach der zu erwartenden kaiserzeitlichen Limesstation Iatrus. Das im Jahre 45 gegründete Legionslager Novae diente zunächst der legio VIII Augusta, ab dem Jahre 69 der legio I Italica als Standquartier. Wie die Feldbegehungen erbrachten, existierte im 2. und 3. Jh. um das Legionslager herum mit den canabae (60-70ha.) und dem 2.5kms donauabwärts gelegenen vicus (c.25-30ha.) eine bedeutende Siedlungsagglomeration mit städtischem Charakter. Außerdem entwickelten sich in den weiten, für die landwirtschaftliche Nutzung geeigneten Tälern im Umland von Novae größere Siedlungen und Villen, die zumindest teilweise mit Veteranenansiedlungen in Verbindung gebracht werden müssen. Aus verschiedenen Hinweisen ist zu erschließen, daß die c.60kms lange Donaustrecke bis zum Auxiliarkastell Sexaginta Prista lediglich durch drei kleinere Militärposten (Iatrus, Scaidava, Trimammium) abgesichert war. Sie dürften mit Détachements der legio I Italica besetzt gewesen sein. Obwohl die Barbareneinfälle des 3. Jh. gewiß erhebliche Auswirkungen auf die Siedlungstätigkeit hatten, läßt sich - mit gewissen Einschränkungen - an Hand des Fundmaterials eine weitgehende Siedlungskontinuität bis zum Ende des 4. oder das beginnende 5. Jh. im Untersuchungsgebiet beobachten. Voraussetzung dafür waren die Wiederherstellung und der Ausbau der Befestigungsanlagen am Donau-Limes unter Diokletian und Konstantin I. Neben der Osterweiterung des Legionslagers Novae (Novae II) war vor allem die Gründung des Kastells Iatrus mit der Stationierung eines cuneus equitum scutariorum an der Jantra-Mündung von außerordentlicher Bedeutung für die militärische Absicherung dieses Limesabschnitts. Der Bau einer Straße quer durch das Niederungsgebiet der JantraMündung verkürzte die Distanz zwischen Novae und Iatrus erheblich. Für die valentinianische Zeit läßt sich ähnlich wie an der mittleren Donau und am Rhein nochmals eine Verstärkung der Limesstrecke mit Türmen und Kleinfestungen nachweisen. Außerdem wurden nunmehr verstärkt Föderaten in die Grenzsicherung eingebunden. Die Hunnenkriege in der 1. Hälfte des 5. Jh. führten offenbar nicht nur zum zeitweiligen Zusammenbruch des Limes, sondern auch zu fast völliger Aufgabe jeglicher Siedlungstätigkeit im Binnenland. Erst im Laufe der 2. Hälfte des 5. Jh. wurden die Limesbefestigungen als Verteidigungsanlagen und Rückzugsorte wiederhergestellt, die gemeinsam mit einem Befestigungssystem im Landesinnern die Zugehörigkeit der Gebiete nördlich des Balkan zum Oströmischen Reich bis zum Ende des 6. Jh. sicherten.

1.

Historical Museum, Ruse. The programme has been supported by the Deutsche Forschungsgemeinschaft Bonn since autumn 1998.2 The central aims of the project are to investigate settlements and the reconstruction of ancient landscapes in this region in all periods from the palaeolithic onwards as well as of the limes components including the search for the presumed limes-fort at Iatrus of the C1st-C3rd AD. The results are to be compared with the envelopment of the well-known building periods of Iatrus and Novae and with the results of a similar programme in the region of Nicopolis-ad-Istrum (cf. Poulter 1999a; 1999b). Our on-going interdisciplinary investigations include field survey, the use of geophysical methods, aerial photography as well as palynological and soil analyses in a study territory of approximately 50kms long, and up to

Introduction

More than 40 years of archaeological research in the legionary camp of Novae (near Svištov, Bulgaria) and the late Roman fortress of Iatrus (at the mouth of the River Jantra, about 15kms as the crow flies downstream from Novae) has produced important contributions to the history of the Roman frontier on the lower Danube.1 Because of the poor level of knowledge, for several years attention has been focused on the links of both fortresses to their surrounding territories (cf. v. Schnurbein & Wachtel 1995: 6). The investigations started with a short survey in autumn 1997 (Conrad & Stančev 1998). The RömischGermanische Kommission Frankfurt/M. collaborates with Bulgarian colleagues from the Archaeological Institute of the Bulgarian Academy of Sciences, Sofia and the

2

The authors are grateful to G. v. Bülow and S. v. Schnurbein (Frankfurt/M.), R. Gančev (Ruse), P. Donevski (Svištov), R. Krauß (Berlin) and L. Vagalinski (Sofia) for their help and support. The support of the programme by the Deutsche Forschungsgemeinschaft Bonn is here acknowledged. Many thanks to A. Poulter (Nottingham) who read the translation and to K. Ruppel & A. Posluschny (Frankfurt/M.) for technical help.

1

Iatrus: Iatrus-Krivina I-V; v. Bülow 1994. Preliminary reports on the results of the excavations since 1992 in v. Bülow & Milčeva 1995: 140 sqq. – Novae: Summary of the results by Press & Sarnowski 1990; Ivanov 1997: 556-574, 599-601; id. 1999: 184-189, 267-269; annual reports of the excavations in Archeologia (Warszawa).

673

Limes XVIII 20kms wide – in all about 1000kms2 (Figs. 1, 2-4). With respect to the limes we are collecting new data which goes beyond the research level recently summarised by R. Ivanov (1997; 1999).

Dermen dere.5 Parts of the necropolis were found to the south. There are also some indications that the area covered by the later extension Novae II and at the mouth of the Dermen dere in the eastern part belonged to the civilian settlement of the fort in the C1st-3rd (Press & Sarnowski 1990: 353-355; Ivanov 1997: 560, 600 - cf. Dimitrov et al. 1970: 55). All-in one could calculate an area of about 70– 80ha for the civilian settlement for the C1st-3rd.6 Compared with the c.130ha of the canabae of the legionary camp of Carnuntum7 this size seems to be realistic. Deductions derived from the distribution of the surface pottery, indicate the area of the late Roman canabae of Novae occupied only one-third of the former territory.

The history of the Roman limes in this section of the Danube begins with the creation of the province Moesia in AD44 and the transfer of the legio VIII Augusta to Novae simultaneously or within a short time after that.3 The territory of the Moesian province and with it the border along the Danube extended to the River Jantra (ancient Iatrus). From Dimum to the mouth of the Danube, the Portorium Ripae Thraciae was established. The organisation of the military protection along the Danubian bank east of the Jantra is unknown for the early Roman period. In general a redistribution of troops is accepted in this section only in the Flavian times as supported by some archaeological evidence (Suceveanu & Barnea 1991: 2830; Ivanov 1997: 506sq. and n. 140, 509sq.; Vagalinski 1999: 232). We suppose that the legio VIII Augusta (replaced by the legio I Italica in 69/70) at Novae was responsible for the military protection of this part of the frontier at this time. 2.

Remains of walls uncovered by erosion processes on the Danube prove the urban character of the settlement. Until now little archaeological investigation has taken place in this area. Reliable data exists only from the excavations of the so-called “Villa of the legatus” (Čičikova 1987; Čičikova & Božilova 1990; cf. Ivanov 1997: 569 sq., fig. 32, 574) near the western gate of the camp and of the mithraeum in the south-western part of the canabae (Naydenova 1994; 1999). To date we know nothing about the original structure of the canabae Novenses.

The fortifications Many years ago a large Roman settlement was identified c.2.5kms east of the legionary camp on the Danubian bank (Figs. 2; 3: cf. Ivanov 1997: 600sq.). A considerable number of architectural fragments (stones, building ceramics) visible on the ground surface proved the monumental character of some of its buildings. We recovered a lot of ceramics and small-finds (metal objects, quernstones etc.), as well as brick stamps of the legio I Italica and of private producers.8 The identification with the vicus belonging to the legionary camp of Novae and connected with it by the limes road seems to be beyond dispute.9 It is tempting to speculate about the existence of a small garrison at this site.

The following preliminary results of our project are based on surface pottery and small-finds, local observations, aerial photographs and in some cases on additional geophysical investigation and archaeological excavation. The Kaleto mound situated in the centre of Svištov near the Danubian bank was inhabited from prehistory onwards (Stefanov 1958: 337-365). Pottery of the C2nd-C3rd has been found too. At this strategic position with an excellent overview along it and across to the opposite side of the river a watch-tower probably existed during the principate. Finds of brick stamps of the legio I Italica were recorded too (cf. Press & Sarnowski 1990: 226, fig. 2). Excavations carried out by P. Donevski proved the extension of the fortification in the late Roman and early Byzantine times (Donevski 1997: 35-38 - cf. also Stefanov 1958: 345-353). The circuit wall probably included the whole mound - an area of about 5-6ha. V. Beševliev made the identification of the site with the fortress Theodoroupolis mentioned by Procopius (De Aed. IV.7) and Theophylactus Simokates (VII.1) in the C6th (Figs. 2-4).4

The relatively small area of the vicus (c.20ha) compared to the canabae seems to argue against the identification with the municipium Novense of that settlement as suggested by R. Ivanov (1997: 600). However to come to a definite opinion on this matter further data is necessary. So one has to take into account the loss of the northern part of the vicus caused by the erosion of the Danubian bank. During the C4th the extent and the intensity of the settlement declined and for the C5th-6th there are no signs of continued occupation.

The canabae of the legionary camp at Novae extended c.1.5kms along the road to Svištov; the route is roughly identical with the former limes road (Figs. 2, 3 & 5). The settled area was interrupted only by several hills; in the south-west it reached the ridge that dropped steeply to the

5

The archaeological fieldwork in this area is very restricted because it is completely covered by private estates. The authors thank P. Donevski (Svištov) for his support during the survey carried out in the spring of 1999. 6 This extension is much more than supposed by Press & Sarnowski 1990: 226, fig. 2; Ivanov 1997: 600, fig. 49. 7 Kandler 1981: 3; Humer & Jobst 1995: 12 sq. For the canabae of Carnuntum cf. Stiglitz, Kandler & Jobst 1977. 8 There are also earlier finds of brickstamps of the legio I Italica, cf. Press & Sarnowski 1990: 220, fig. 2. 9 Cf. Ivanov 1997: 600f.; 1999: 268. For the distance between a legionary camp and its vicus see Piso 1991.

3 Cf. Ivanov 1997: 506. In Augustan times the limes along the Danube had reached only the modern town of Nikopol. The Ripa Thraciae downstream of the river belonged to theThracian kingdom which was dependent on Rome. 4 Beševliev 1970: 123. Other proposals for the identification can be excluded, cf. Ivanov 1997: 485.

674

Sven Conrad & Dimităr Stančev: Archaeological survey on the Roman frontier on the lower Danube Before starting our survey we assumed the simultaneous erection of a fort at the mouth of the Iatrus river with the camp of Novae or at latest during the reign of Trajan. The most suitable place seemed to be about 500m south of the late Roman fortress where we picked up the C2nd-3rd pottery (Figs. 2-4, 6). Geophysical investigations carried out at this site did not confirm our assumption.10 We located only small undated anomalies and could not get any evidence for a fortification or other big building complex.

uncovered by erosion prove the existence of a vicus belonging to a late Roman fort. In contrast, traces of a settlement in the early Byzantine period in this area are rare. About 2kms north-west of the village of Batin and 2kms west of the fort of Scaidava a burgus was excavated in the 1980s (Figs. 3, 7; cf. Stančev 1999). It is only partly preserved and lies in the mouth of a valley running out of the highlands to the Danube and was erected in the time of Valentinian I (Stančev 1999: 203 sq.). The plan of the preserved parts (Fig. 7) suggests a second tower with an entrance between. The closest parallel in the eastern Balkans can be found in the burgus at Topraichioi in the Dobroudja. This fortification possessed only one gate tower and is dated also to the Valentinian period (Zahariade & Opaiţ 1986). Two other sites producing late Roman pottery and in a topographical situation similar to the burgus near Scaidava were found at the edge of the Danube lowlands between Iatrus and the excavated fortification, with a distance between them of 2-3kms (Figs. 3, 12). The topographical situation of these sites is very similar to the middle Danube, where the burgi are situated at the mouth of valleys running to the Danube (Ubl 1980: 597).

Another large Roman settlement is situated on the left bank of the Jantra on a south-facing location (about 4kms southwest of Iatrus) with an extensive spread of surface debris covering c.25ha.11 We can exclude this place for the location of the hypothesised limes fort because it is too distant from the River Jantra and its mouth and we have not got any evidence from geophysics. Based on the summary of the data to hand we suppose that there did not exist a large fort at the mouth of the Jantra in the C2nd-3rd. The reasons for this opinion are: 1. Iatrus is not mentioned in the itineraries of the C2nd3rd. 2. There are no references regarding a relocation of an auxiliary unit or a vexillatio at that time.12 3. For the construction of the well-known late Roman fortress only a small amount of spolia was used, which indicates no earlier buildings of large size in the vicinity. 4. In many cases, late Roman forts were built over their predecessor. There is no evidence for this at Iatrus.

If we are correct in our interpretation, the burgi/turres north-east of Iatrus belong to the echelon defence system at the mouth of the River Jantra. A fort on the left bank of the Danube near the village of Pietroşani was included in this system (Figs. 2, 3).13 Until now we have no information about its date, but we can suppose a late Roman origin.14 [See Postscript below].

Instead there probably existed only a modest military post or a secured mansio occupied by soldiers of the legio I Italica. This supposition remains to be proven by archaeological investigations.

To date we know of five forts in the frontier section between Iatrus and Sexaginta Prista: 1. Scaidava15 (Figs. 24, 8); 2. Trimammium (to the west of Stălpište/Mečka)16 (Figs. 2-4, 9): and three sites in the neighbourhood of Pirgovo (Figs. 2-4): 3. at the 512kms point on the Danube (fields ‘Drakata’; ‘Gornoto skele’; Dremsizova-Nelčinova 1983: 55 Nr. 138); 4. at the 510kms point (fields ‘Dolnoto skele’; identified as Mediolana from the literary sources: Dremsizova-Nelčinova & Ivanov 1983: 54 sq. no. 137; Ivanov 1997: 486. For the identification as Mediolana cf. Torbatov 1991: 231): at the 508kms point on the Danube (fields ‘Kaleto’: Dremsizova-Nelčinova & Ivanov 1983: 55, no. 139). They have been recorded in several publications but with unproved, incomplete or insecure data.

Near the late Roman fort at Iatrus, which was erected in the reign of the emperor Constantine I (cf. Döhle 1995: 912 and n. 13), we confirmed the existence of a civilian vicus, today overlaid in its major part by the village of Krivina. The vicus covered a relatively small area in its first period (the first half of C4th) and consisted probably only of a few houses situated along the main road running out of the fort. The vicus reached its maximum extent, at about 4-5ha, at the end of the C4th and in the first half of the C5th when it included part of the area of the settlement of the C2nd-3rd south of the fort. Pottery, coins, some small finds and wall segments c.100m north-east of the fort 10 The geophysical investigations were carried out by N. Tonkov & I. Katevski (Archaeological Institute of the Bulgarian Academy of Science Sofia, (in 1999) and members of the staff of the Archaeological Institute of the Slovakian Academy of Science Nitra under the direction of J. Rajtár (in 2000). 11 The site is already mentioned by Stefanov 1956: 46 (and map). 12 The small number of inscriptions reused in the construction of the Iatrus fort and mentioning members of the legio I Italica may have come from Novae or from a Roman site in the vicinity. Döhle 1995: 27 supposed the origin of the inscriptions and spolia was Novae. Against this, see Wachtel 1991: 207.

13 The site is briefly mentioned by Tudor 1978: 286, fig. 74,3; Zahariade & Gudea 1997: 74 no. 17. 14 A brick stamp of the legio I Italica was published, cf. CIL III 12522. Similar bridgeheads existed on several places along the lower Danube, cf. for instance the fortifications of Gornea (cf. Gudea 1977a); Drobeta (Tudor 1978: 56-68, 274-277; TIR L34) and Sucidava (Tudor 1965; TIR K35). 15 Itinerarium Antonini 222,1. – Cf. Škorpil 1905: 455; 1914: 558; Dremsizova-Nelčinova & Ivanov 1983: 27 Nr. 13; Ivanov 1997: 48. 16 Itinerarium Antonini 222,2. Cf. Škorpil 1905: 454f.; DremsizovaNelčinova & Ivanov 1983: 46 sq. no. 99; Ivanov 1997: 486.

675

Limes XVIII

Excluding No. 3, all these sites are characterised by a similar topographical situation which we were able to study in the locality and on aerial photographs. The forts are situated at a spur/ridge watching over a valley mouth to the Danube. The late Roman and early Byzantine walls seem to be overlaid by ramparts belonging to the early middle ages. This is evident at Trimammium (No. 2) and Mediolana (No. 4). The sites are protected by deep ditches separating the spur/ridge from the plateau. At Trimammium the trackway to the eastern gate appears on the aerial photograph (Škorpil 1905: 454, plan b records two gates at Trimammium, at the east and the northwestern sides).

north of the fort of Iatrus. We could identify it on aerial photographs taken in spring 2000 (Fig. 10). An excavation trench through the road, about 3kms west of Iatrus, showed the method of its construction (Fig. 11). First an embankment was deposited with earth taken from the side ditches. Secondly a substructure in the form of three bonds was carried out with limestone slabs. Finally the pavement of small quarry stones and gravel was filled up. Unfortunately the pavement has been badly damaged by agricultural machines but we assume that there existed only one pavement layer.20 There is some admittedly limited evidence for its construction date. Near the trench we picked up a forged nail, besides two sherds of a pot and a late Roman amphora 1 of the C4th. Because there seems to exist no evidence for a renovation of the pavement we suppose that this road could not have been used for long. We can probably connect it with the defence activities of the mid-C4th. In any case it was impossible to use that road section during flooding when one then had to make a much longer circumvention around the lowlands.21 This had to be the usual way in the C1st-3rd and in earlier times.

With the exception of No. 4 we found at the sites small amounts of pottery fragments of the C2nd-C3rd. These finds confirm the Itinerarium Antonini which mentions Scaidava and Trimammium on the road from Novae to Sexaginta Prista. We assume that at that time only small garrisons existed for the observation and the defence of the Danubian bank in this sector. We do not have any evidence of the troops located in this part of the limes from the C1st to 3rd. The plan of the fortifications, measuring between c.0.75ha (No. 5) and c.1.5ha (Trimammium), suggest detachments of the legio I Italica or parts of auxiliary units.17 For late antiquity the Notitia Dignitatum mentions milites Constantini for Trimammium (Orr. XL.20) and milites Dacisci for Mediolana (Orr. XL.21). Brick stamps of Rumorid(us) found in Trimammium show fortification activities during the reign of Valentinian I.18 According to the surface finds all these fortifications continued to be used into the early Byzantine period.

We also discovered some parts of the route down the Danube from Iatrus. On the way to Scaidava it was partly cut into the precipitous slope of the Danube bank (Fig. 12). Here no pavement is preserved, possibly because the route had repeatedly to be renewed on account of the erosion in ancient times. In the vicinity of Mediolana we also documented sections of the destroyed pavement of the limes way in the fields near the Danube.22 4. The system of settlements in the vicinities of Iatrus and Novae

3. The limes road One of the most important aims of the project is to investigate the settlement pattern in the region in all historic periods. According to the finds picked-up by the survey a relatively dense settlement system existed in the hinterland of the limes in the C2nd and 3rd (Fig. 2). We mainly find sites of larger extent (vici) in the great river valleys and in the wide-open valleys south of Novae. Smaller places that we can interpret as villae are situated in small side-valleys. In the area east of the Jantra characterised by deep narrow valleys the sites are situated at greater intervals. In any case the water supply or the possibility of access to water is a prerequisite for settlement. The average interval is about 5-6kms and in the vicinity of Novae 2-3kms. These differences show a peculiarity of our region: the closed military system of the

Up until now only small parts of the limes road connecting the fortifications along the Danube have been established to the west of Svištov.19 To the east of Novae the ancient route is probably partly identical with the modern road to the village of Vardim. To the east of Vardim we discovered the ancient route at the first very low bank terrace which was free from floods (Figs. 2-4). It was a surprise to find about 3kms east of Vardim a turning that runs 4kms straight through the lowlands at the mouth of the Jantra and therefore predating the 1950s regulation of this marshy area which was subject to floods. The route reaches the right bank of the Jantra about 100m 17 For size of the forts see Johnson 1983: 292 sq.; Bennett 1986. There is no standard and a great variety of the fort areas; without further data it is not possible to establish the unit from it. A gravestone of a mensor apprentice of the legio I Italica was found in the vicinity of Trimammium, cf. Velkov 1968: 4-7, no. 2, fig. 2: D(is) M(anibus) /Aur(elio) Muciano / discente me(n)/sore leg(ionis) I / Ital(icae) vix(it) / ann(is) XX mil(itavit) / ann(is) I... (1st half of C3rd). 18 Škorpil 1905: 455 = CIL III Nr. 2655. For the stamps of Rumorid(us) cf. Sarnowski 1985: 124 sq.; Wachtel 1991: 209 with n. 10; 11. 19 Škorpil 1905: 460 plan h can obviously be right with his information (questioned by Stefanov 1956: 72), that one route of the ancient way runs east of Dimum through the – in the past marshy – lowlands near the Danube.

20 This is northern of the route supposed by Škorpil 1905: 456 plan g. We also could not find any traces of the supposed bridge (ibid.) south-west of Iatrus. After interviewing local people, we learnt of a ford in the Turkish period of the Bulgarian history. 21 From the information from local people before the regulation of the Danube in the 1950s the flood could reach about 4kms to the south, in the vicinity of the village of Karamanovo. 22 After Škorpil 1905: 455 the route between Scaidava and Trimammium circumvented the territory characterised by a strong undulated landscape and the swamps at the Danube by a southern by-pass through the highlands. There is no evidence for a Roman military station or settlement in this section.

676

Sven Conrad & Dimităr Stančev: Archaeological survey on the Roman frontier on the lower Danube of the lower Danube.24

frontier fortifications had a great influence on the settlement system, especially as an economic factor, influencing how the population was settled.

In the hinterland of the Roman frontier an active network of settlements evolved consisting of vici and villae. The Gothic wars of the second half of the C3rd obviously badly affected the region but did not end this occupation.

The consolidation of the Roman frontier after the wars of the C3rd, during the reigns of Diocletian and Constantine I, was a supposition for the continuation or the renewal of most of the settlements at the beginning of the late Roman period. In some of the settlements identified we established a reduction of the settled area but altogether we can summarise a continuity from the C2nd-3rd through to the beginning of the late Roman period, when some places were even established for the first time (Fig. 3).

According to the available evidence the end and redefinition of the limes in this part took place - as on the middle Danube (cf. Mócsy 1974: 266-296; Soproni 1978: 9) - in late antiquity when small garrisons and observation posts were converted into fortresses. The Iatrus fort was erected and the limes road improved. With the help of the brick stamps T. Sarnowski has shown that at the beginning of the late Roman period the legio I Italica was partly divided into garrisons of the diocese Thracia - among them surely forts between Novae and Sexaginta Prista - probably to establish a more regular distribution of troops (Sarnowski 1985. cf. Ivanov 1997: 520-529). But the defence system on the lower Danube obviously did not reach the developed level and the density of forts and burgi as found in Pannonia, especially, if one takes into account the echelon limes system (cf. Soproni 1978: 192-202). Could the width of the river be the reason?

Profound crises were caused by the Gothic wars in the second half of the C4th and the invasions of the Huns at the beginning of the C5th when all settlements in the hinterland of the frontier were apparently abandoned (Fig. 4). Similar results were found by the English-Bulgarian expedition in the hinterland of Nicopolis-ad-Istrum.23 Undoubtedly these events were followed by a considerable reduction in the population in the territories north of the Balkan range. The surviving population settled in or near the fortifications, while new fortresses were erected (cf. in the area of our research at the fortresses of Theodoroupolis, Bjala, Polsko Kosovo and Koprivec). These results correspond with the situation at the Iatrus fort in the first half of C5th when all free areas between existing buildings and in the demolition of horrea I and VII were filled with dwelling houses (cf. v. Bülow 1995: 32, 38sq., 41). In these late times such fortifications aimed not only at the defence of the region, but also served an important function as centres for retreat as well as agriculture.

By the time of Valentinian I the intervals between the forts was partly secured by burgi and watch towers. The defensive efforts were concentrated in the threatened region around the mouth of the Jantra. Here the echelon system is very similar for instance to the defence system at the mouth of the River Neckar in Germania Prima (Wieczorek 1995). After the destruction of the disastrous Gothic wars in the second half of the C4th and the Hunnic assaults in the first half of the C5th, the function of defence relied more upon the rebuilt or newly constructed great fortifications along the Danube and later on a second line in the hinterland of the limes. These fortifications had a double function: as military centres as well as civilian settlements.

5. Conclusions On present evidence, in the C1st-C3rd there did not exist a fort at the mouth of the Jantra. We assume an observation post or a military defended mansio at that time. Naturally this hypothesis has still to be proven by archaeological excavation. We conclude the existence of a loose linear control system in this frontier section between Novae and Sexaginta Prista in the same period. It consisted of small posts spaced c.7-10kms from each other. The distance between the large forts of Novae and Sexaginta Prista is about 50kms as the crow flies; this is more than the ‘norm’ on a river frontier and on the Danube itself (cf. Breeze 1997: 73 sq.). The reason for it could be that there existed partial Roman control over the ‘barbarian’ territories north

After an eventful C6th the history of the limes on the lower Danube ceased with the catastrophic invasions of the Slavs and Avars at the end of that century and in the beginning of the C7th. Postscript In the spring of 2001 we continued our survey on the northern bank of the Danube with the specific aim of searching in the area of the Roman fortress Pietroşani. We can now say that a castrum of the C2nd/C3rd as well as of the late Roman period in the extensions described by Tudor (160 x 80m in the preserved part) does not exist. There are no traces and no finds on the ground surface which could justify such a hypothesis.

23 Abandonment of the settlements in the territory of Nicopolis-ad-Istrum by c.400; cf. Poulter 1995 (the fortified annexe at Nicopolis II after the destruction of Nicopolis-ad-Istrum in the mid-C5th); 1999a: 145-147 (supposing a reduction of the settlement density); 1999b (building up the fortress at ‘Gradishte’ near Dichin in the territory of Nicopolis-ad-Istrum in the early C5th); Cočev 1998 (end of the coin circulation in the late Roman settlements of Pavlikeni and Butovo in the middle or the second half of the C4th). Cf. similar observances at several sites along the Danube (Henning 1987: 35-40).

24 Cf. Gudea 1977b: 851-876. South-east Dacia libera seems to have been at the very least partly under control of the Romans after the reforms of Hadrian.

677

Limes XVIII

There are very few late Roman pottery sherds and some brick fragments of the late type (including late - ie. end of the C3rd/beginning of C4th - brick stamps of the legio I Italica) from the area south-west of the village of Pietroşani. They could be connected with a small burgus or a watch tower in that area which existed in any case only for a very short time. We will continue the discussion of this problem at a later stage of research.

Landwirtschaft des 1. Jahrtausends u.Z. Schriften zur Ur- und Frühgeschichte 42. (Berlin). Humer F. & Jobst W. 1995 Rundgang durch die Ausgrabungen und das archäologische Museum Carnuntum. Archäologischer Park Carnuntum. Die Ausgrabungen 2. (Bad Deutsch-Altenburg). Iatrus-Krivina I-V 1979-1995 Iatrus-Krivina I-V. Schriften zur Geschichte und Kultur der Antike 17. (Berlin). Ivanov R. 1997 Das römische Verteidigungssystem an der unteren Donau zwischen Dorticum und Durostorum (Bulgarien) von Augustus bis Maurikios. Berichte der Römisch-Germanischen Kommission 78: 468-640. Ivanov R. 1999 Dolnodunavskata otbranitelna sistema meždu Dortikum i Durostorum ot Avgust do Mavrikij. (1999). Johnson A. 1983 Roman forts of the 1st and 2nd centuries AD in Britain and the German provinces (London). Kandler M. 1981 Ein Tempelbezirk in der Lagerstadt von Carnuntum. (Wien). Mócsy A. 1974 Pannonia and Upper Moesia. (London). Naydenova V. 1994 Un sanctuaire syncrétiste de Mithra et de Sol Augustus découvert à Novae (Mésie Inférieure). In J.R. Hinnels (ed.) Studies in Mithraism. Storia delle religioni 9. (Roma): 225-228. Naydenova V. 1999 Le culte de Mithra à Novae (Mésie Inférieure). In G. v. Bülow & A. Milčeva (edd.) Der Limes an der unteren Donau von Diokletian bis Heraklios (Sofia): 117-120. Piso I. 1991 Die Inschriften vom Pfaffenberg und der Bereich der Canabae legionis. Tyche 6: 131-169. Poulter A. 1995 Nicopolis ad Istrum. A Roman, late Roman and early Byzantine city. Journal of Roman Studies Monograph 8. (London). Poulter A. 1999a The transition to late antiquity on the lower Danube: An interim report (1996-8). The Antiquaries Journal 79: 145-185. Poulter A. 1999b ‘Gradishte’ near Dichin: A new late Roman fortress on the lower Danube. In G. v. Bülow & A. Milčeva (edd.) Der Limes an der unteren Donau von Diokletian bis Heraklios (Sofia): 207-227. Press L. & Sarnowski T. 1990 Novae. Römisches Legionslager und frühbyzantinische Stadt an der unteren Donau. Antike Welt 21/4: 225-243. Sarnowski T. 1985 Die legio I Italica und der untere Donauabschnitt der Notitia Dignitatum. Germania 63: 107-127. v. Schnurbein S. & Wachtel K. 1995 Vorwort. In IatrusKrivina V (Berlin): 5-6. Škorpil K. 1905 Nekotorija iz dorog vostočnoj Bolgarii. Izv. Russkogo Arh. Inst. Konstantinopel 10: 443-502. Škorpil K. 1914 Opis na starinite po tečenieto na reka Rusenski Lom. Materiali za Arheologičeskata karta na Bălgarija 1 (Sofija). Soproni S. 1978 Der spätrömische Limes zwischen Esztergom und Szentendre. (Budapest). Stančev D. 1999 Eine Befestigung beim Dorf Batin (Gebiet Ruse). In G. v. Bülow & A. Milčeva (edd.) Der Limes an der unteren Donau von Diokletian bis Heraklios (Sofia): 201-205. Stefanov S. 1956 Starinite po dolnija basejn na Jantra. Materiali za Arheologičeskata karta na Bălgarija 8 (Sofija). Stefanov S. 1958 Prinos kăm starata istorija na Svištov do sredata na XVII vek. In Sto godini narodno čitalište Svištov 1856 g. – 1956 g. Jubileen sbornik (Svištov): 337-365. Stiglitz H., Kandler M. & Jobst W. 1977 Carnuntum. In H.

Bibliography Bennett J. 1986 Fort sizes as a guide to garrison type: A preliminary study of selected forts in the European provinces. In C. Unz (ed.) Studien zu den Militärgrenzen Roms III. 13. Internationaler Limeskongreß Aalen 1983. (Stuttgart): 707-716. Beševliev V. 1970 Die Deutung der Kastellnamen in Prokops Werk „de aedificiis“ (Amsterdam). Breeze D.J. 1997 Regiments and frontiers: patterns of distribution on rivers and artificial frontiers. In W. Groenman-van Wateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 73-74. v. Bülow G. 1994 Das spätantike Kastell Iatrus am UnterdonauLimes in Bulgarien. Stand und Probleme der Erforschung. Vortrag zur Jahressitzung der RömischGermanischen Kommission. Berichte der RömischGermanischen Kommission 75: 5-22. v. Bülow G. 1995 Die Entwicklung des Siedlungsbildes von Iatrus in der Periode B/C. In Iatrus-Krivina V (Berlin): 29-53. v. Bülow G. & Milčeva A. (edd.) 1995 Der Limes an der unteren Douau von Diokletian bis Heraklios. (Sofia). Čičikova M. 1987 Su un edificio con peristilio extra muros di Novae. Ratiariensia 3-4: 185-192. Čičikova M. & Božilova V. 1990 Nouvelle inscription d’un sénateur anonyme découverte à Novae (Mésie Inférieure). Mél. École Française Rome 102/1: 611619. Cočev M. 1998 Moneti i monetna cirkulacija v antičnite keramični centrove Pavlikeni i Butovo (Veliko Tărnovo). Conrad S. & Stančev D. 1998 Ein Limeskastell und sein Umfeld. Archäologie in Deutschland 2: 64. Dimitrov D., Čičikova M., Sultov B. & Dimitrova A. 1970 Arheologičeskie raskopki v vostočnom sektore Nove v 1966 godu. Izv. Arh. Inst. 32: 55-71. Döhle B. 1995 Die Siedlungsperiode A in Iatrus. In IatrusKrivina V (Berlin): 9-28. Donevski P. 1997 New late Roman fortress in Svištov to the west of Novae. In A. Biernacki & P. Pawlak (edd.) Late Roman and early Byzantine cities on the lower Danube from the 4th to the 6th centuries A.D. (Poznań): 35-38. Dremsizova-Nelčinova C. & Ivanov D. 1983 Arheologičeski pametnici v Rusenski okrăg. (Sofija). Gudea N. 1977a Gornea. Aşezări din epoca romană şi romană tîrzie. (Reşiţa). Gudea N. 1977b Der Limes Dakiens und die Verteidigung der obermoesischen Donaulinie von Trajan bis Aurelian. In H. Temporini & W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II.6 (Berlin): 849887. Henning J. 1987 Südosteuropa zwischen Antike und Mittelalter. Archäologische Beiträge zur

678

Sven Conrad & Dimităr Stančev: Archaeological survey on the Roman frontier on the lower Danube Temporini & W. Haase (edd.) Aufstieg und Niedergang der Römischen Welt II.6. (Berlin): 583-730. Suceveanu A. & Barnea A. 1991 La Dobroudja romaine. (Bucarest). Torbatov S. 1991 Recensija: M. Zahariade, Moesia Secunda, Scythia şi Notitia Dignitatum (Bucureşti 1988). Dobrudža 8: 229-233. Tudor D. 1965 Sacidava. Une cité daco-romaine et byzantine la Dacie. Collection Latomus 80 (Bruxelles). Tudor D. 1978 Oltenia romană (Bucureşti). Ubl H. 1980 Der Österreichische Abschnitt des Donau-Limes. Ein Forschungsbericht (1970-1979). In W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies XII. British Archaeological Reports Int. Ser. 71. (Oxford): 587-611. Vagalinski L. 1999 Die spätrömische nördliche Festungsmauer von Transmarisca. In G. v. Bülow & A. Milčeva (edd.) Der Limes an der unteren Donau von Diokletian bis Heraklios. (Sofia): 229-236. Velkov V. 1968 Epigrafski prinosi kăm starata istorija na Ruse i Rusensko prez Rimskata epoha. Izv. Narod. Muz. Ruse 3: 3-10. Wachtel K. 1991 Die Inschriften. In Iatrus-Krivina IV. (Berlin): 207-211. Wieczorek A. 1995 Zu den spätrömischen Befestigungsanlagen des Neckarmündungsgebietes. Mannheimer Geschbl. N.F. 2: 9-90. Zahariade M. & Gudea N. 1997 The fortifications of lower Moesia. (Amsterdam). Zahariade M. & Opaiţ A. 1986 A new late Roman fortification in the territory of Romania: The Burgus at Topraichioi, Tulcea County. In C. Unz (ed.) Studien zu den Militärgrenzen Roms III. 13. Internationaler Limeskongreß Aalen 1983. (Stuttgart): 565-572.

679

Limes XVIII

Fig. 1. The limes on the lower Danube in the late Roman period and the position of the study area (scanning) (drawing by K. Ruppel, Frankfurt a. M.).

Fig. 2. Frontier and settlements of the C1st – C3rd in the study area (autumn 2000).

680

Sven Conrad & Dimităr Stančev: Archaeological survey on the Roman frontier on the lower Danube

Fig. 3. Frontier and settlements in the late Roman period (C4th – beginning of the C5th ) (autumn 2000).

Fig. 4. The study area in the early Byzantine period (end of the C5th – 6th ) (autumn 2000).

681

Limes XVIII

Fig. 5. The vicinity of Novae from early to late Roman times (after Press & Sarnowski (1990): fig. 2; Ivanov (1997): fig. 49).

Fig. 6. 3-D perspective of the area around the mouth of the River Jantra (from south-west; double vertical exaggeration). Extensions of the scene: c.8kms (west-east) to 20kms (south-north). The late Roman defence system: 1 – Iatrus-fort with vicus; 2 – bridgehead-fort near Pietro.

682

Sven Conrad & Dimităr Stančev: Archaeological survey on the Roman frontier on the lower Danube

Fig. 7. The burgus north-west to Batin (Rusedistrict) (drawing K. Ruppel, Frankfurt/M.).

Fig. 8. Aerial photograph of the Scaidava fort (from north-east). The arrow marks the ditch.

Fig. 9. Aerial photograph of the Trimammium fort (from south-east) showing the trapezoidal rampart over the wall, the ditch south of the site and the trackway to the eastern gate. South of the ditch there are traces of a civilian settlement belonging to the fort.

683

Limes XVIII

Fig. 10. The limes road near the mouth of the Jantra (arrow 1; from west); in the background the Iatrus fort (arrow 2).

Fig. 11. Western section of the limes road in a trench, c.3kms west of Iatrus (cf. Fig. 10).

Fig. 12. The supposed track of the limes road on the northern hill slope near the Danubian bank (arrow 1; from west). Arrow 2 marks a small site, probably a burgus, producing late Roman pottery.

684

Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae (Moesia Inferior) Piotr Dyczek The question of the supply of the Roman army has yet to be treated exhaustively. There is little epigraphic data on the subject and the results of archaeological excavations are even more debatable. Our understanding of the Roman army supply system is still very general and full of gaps. Epigraphic and archaeological data suggests that only some chosen products and army provisions were centrally imported. Occasionally, purchases were made on behalf of a unit as a whole - on the market, at contract prices, or as taxes. The vici and travelling salesmen were also a source of provisions, which, apparently, covered more than just food supplies, ie. grain and animals, but also such products as clothes, pottery, metal objects, household commodities.

(CIL XIII 6935). This is again confirmed by finds from Wiesbaden and Mainz, where millstones have been found with inscriptions referring to individual centuriae (Breeze 1977: 138). Big groups of soldiers could have travelled even to very distant places for provisions. For example, 105 soldiers from the cohors I Hispanorum stationed in Macedonia, but operating in lower Moesia, provisioned themselves with clothes from Gaul, grain from the conquered lands, and cattle from the Balkans (Fink 1958: 13). From the list of absentees - ex eis apsentes – we learn that they were: in gallia vestitum item fumintat[u]m trans err[e]m equatum in is eq ( Col.ii, 18-20)

Under the early empire, there was in Rome a special office charged with army supply; the secretaria a rationibus. In the provinces, governors were responsible together with the procurators and quaestors. Officials working in the officium rationorum constituted the lower administration (CIL III 1099; VIII 3229). Some scholars believe that the central provisioning office was still operating in the C3rd AD (van Berchem 1937: 142). In the opinion of this scholar, it was the officium a copiis or cura copiarum exercitus. It was represented in each legion, where the dispensatores were responsible for supply. This organization of army supplies is confirmed only for Africa and there is no certainty that it was common to the rest of the empire. It is known that during military operations supply was in the hands of (CIL III 333) dispensatores rationum copiarum expeditionis (CIL V 2155; VI 8541). Grain provisions were charged to the frumentarii (CIL VI 3340). These persons may have also served as a secret police (Rickman 1971: 274-275).

and even in dardania ad metalla (Col. ii, 21) On the command of the legate Fabius Iustus the detachments from intra provinciam went to: pirob[o]ridavae in praesidio buridavae in vexillatione trans danuvium in expeditionem in is sesq[ .................................................................. item tras ad annona[m] defendendam .................................................................. in.a.ario ad naves frumentarias in is dec i[ .................................................................. ad haemum ad armenta addu[cenda] (Col. ii 2729; 31, 33, 34, 36)

The rôle of the negotiatores and mercatores in provisioning the army should also be kept in mind. The question is well documented and has been widely discussed; I am noting it here only for the sake of completeness.

Occasionally, soldiers themselves arranged for their needs - some minor objects and food articles. Small shops with pottery and glass vases discovered at Inchtuthil along the via principalis suggest such a method of provisioning. At Cardean pits were found where the garrison had dumped the pottery just before abandoning the fort (Breeze 1977: 138). The discoverers were of the opinion that this pottery had been purchased/produced with the intention of centrally supplying the camp. Also, based on the excavations at the valetudinarium in Novae, we may assume that a considerable part of the pottery and lamps were acquired by the legion from a central source. The offices and workshop were probably also provisioned from a central source. Surely the stores in the horrea were maintained from a central source. Some of the provisions reached the military units thanks to the frumentum emptum system. We know the rules of this system thanks to papyri from Koptos, dated 185-186. The ala commander sent the duplicarius to the villages to buy the required amounts of barley for the ala Heraclina. The transaction was carried

Epigraphic documents confirm that provisions came from different areas of the empire; for instance, a papyrus from Oxyrhynchos records the purchase of 200 and 775 blankets for the army. Another document records the purchase in Egypt of garments for the army in Judea. Sometimes acquisitions were commissioned by individual units, eg. two camels bought by the ala veterana Gallica (Breeze 1977: 137). At Usk a mortarium bore the following inscription made on it before firing: pelveis contubernio Messoris. The inscription: 7 Iusti Superi was found on a mortar from Colchester (Breeze 1977: 138). A quern from Saalburg featured the mark contubernium Brittonis (CIL XIII 11954a), and a bread stamp from Mainz suggests that at least occasionally the centuria baked their own bread

685

Limes XVIII

out between the representative of the unit and the strategos or his deputy, basiclico grammateus. The operation was financed by a ‘bank’, which held the money deposited for this purpose (Rickman 1971: 272). Occasionally, larger quantities of pottery were acquired in the same way, as can be concluded from the above-mentioned papyrus from Oxyrhynchos.

vicinity of Madara, Pliska and Kovačevce there are: Aug(ustorum) (duorum) pra(edium) or pra(ta), Marcia, AVGGAVX, AVGGSARM, AVXAN, SARM (Gerov 1988: 74-78; Sarnowski 1988: 65). We also know that near the fortress of Abritus established in the Flavian period there was an estate or estates belonging to Fisc(i) (Sarnowski 1988: 65), and at the same time the boundaries of the legionary area were marked within them (CIL III 13.726). In the opinion of some of the scholars, these epigraphic sources indicate that the workshops producing stamped bricks were located inside the legion’s territory, set apart from the imperial domain. I think that another solution is also possible. The workshops on the emperor’s land produced bricks on commission from the legion. In any case, there is no doubt that there were very close ties between the administration of the imperial domains and the military administration. It is also to be assumed that the provisioning bases of the army may have been located within the imperial domains.

The army could also be provisioned, primarily in food and natural resources, by exploiting the conquered regions. From the formal point of view, these territories belonged to the emperor and all the legion had to do was to administer them (Gerov 1988: 19). These areas associated with legionary fortresses were called prata legionis in inscriptions. The name suggests that these had been pastures at first (Gerov 1988: 27; Mason 1988: 164-165). This is indeed possible, as each legion owned a large number of horses and asses (connected with each contubernia), not counting other beasts of burden, cattle and sacrificial animals. The land was cultivated and the natural resources – water, clay, wood – exploited. Finally, military training grounds were established (Mason 1988: 165).

The costs of erecting a valetudinarium were covered undoubtedly from a central source. The architectural plan, while following the principles outlined by Hyginus and confirmed by archaeological excavations at legionary fortresses and castella on the Rhine and Danube (Press 1988: 69-89), provides suggestions concerning the provisioning system. The ends of the corridors in the hospital at Novae were occupied by open rooms. Large quantities of lamps and tableware were found in them (Szubert 1987: 231-252). Free access to the stores suggests that both staff and patients could use the objects deposited there as needed. Thus, it is highly probable that the commodities had been purchased by the legion command. Another room apparently served as a store for amphorae (Dyczek 1996: 58-59). The vessels, discovered here in fragments once contained wine, oil and salsamenta; hence, it cannot be excluded that these products had also been bought centrally. An interesting piece of a barrel with the inscription immune in r(ationem) val(etudinarii) leg(ionis) II Adi(utricis) comes from Aquincum (Whittaker 1989: 64); it probably contained wine used for healing purposes in the legionary hospital there. And Amineum wine (CIL IV 4535) was perceived by recipients not so much in terms of the place of manufacture, but in terms of having an abundance of fruit (Pliny NH XIV.4.21). The ancients believed it had healing properties. An amphora once containing this wine was found in the military hospital at Carleon (Davies 1970: 92). Wine mixed with herbs was also consumed for medicinal reasons (Davies 1970: 92). Fish sauces and their mixtures were also used in treatment: oleogarum, οξυγαρον, hydrogarum, oenogarum (Curtis 1991: 8, 35, 138, 165, 189), to disinfect wounds, in all kinds of gastric illnesses, and even for migraine and depression. It seems fairly unlikely that patients could afford whole amphora of expensive wine or fish sauce strictly as a cure. One may assume that the valetudinarium received these kinds of products from a central source.

New light on the problem of supplying the Roman army is shed by the excavations carried out by the Center for Archaeological Research of Warsaw University in Novae, the fortress of the legio I Italica. One object category, lead seals, given to the Archaeological Museum in Svištov, clearly has connections with the provisioning of the legion (Mrozewicz 1981: 79-84, figs. 1-20). On one of these, No. 13, there is the inscription: [...]ONA[...], on another, No. 17, AANM (Mrozewicz 1981: 81-82, figs. 13,17), (Fig. 1), which T. Sarnowski (1988) reads as: (ad) [ann]onam and (ad) ann(o)nam. Hence, the seals may have served to certify and protect a transport of the annona. More data was obtained from the exploration of the army hospital (Fig. 2). The valetudinarium at Novae is the best preserved structure of the type in all of empire (Fig. 3). It was erected in the time of Trajan and abandoned in the time of Caracalla, presumably because of the emperor’s decree of 214. Despite this most of the original endow have been preserved, leading to some provisional conclusions. The hospital itself was undoubtedly built by the legionaries themselves. Numerous roof tiles bear the stamp of the legio I Italica. However, there is a small number of tiles that are stamped by the legio I Minervia Pia Fidelis, possibly suggesting that units from other legions may have also participated in the building of the hospital, units that could have been transferred to the Danube for the Dacian wars. Of course, the different types of building ceramics must have been produced in local production centres. There is no data to prove that these were legionary workshops. At relatively numerous sites in Moesia – taking into account the state of research - bricks are found that bear the stamps of the imperial domains. For instance, stamped on the ceramic elements from the area of Ratiaria are the following: Imper(atoris) Had(riani) and Im(pe)r(atoris) T. A(eli)ari(a)e (CIL III 14.589); from the

Religious issues constitute a separate and highly interesting problem, which while not directly connected with the issue 686

Piotr Dyczek: Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae of supply for the hospital, sheds new light on the inner structure of the cult and its practice inside the legionary fortress. In keeping with the general building principles, the valetudinarium at Novae has a big portico-lined courtyard in the middle. Archaeological research has revealed a small temple, sacellum, of the healing deities Asclepios and Hygieia in the centre of this courtyard (Dyczek 1995: 125-128). Numerous altars and bases for silver statues, to judge by the inscriptions on them, were found in front of the facade and in the immediate vicinity (Fig. 4). Altars devoted to the other deities were also found around the courtyard, thus letting us assume that the courtyard as a whole had served as a cult temenos and that the hospital served as a specific kind of ‘Asclepeion’. An analysis of the inscriptions leads to some interesting conclusions (Kolendo 1999: 55-71). The main altar was dedicated to Asclepios/Aesculapus on behalf of the entire legion. Individual patients expressed their gratitude with smaller altars and bases. One might conclude that the cult area and the installations there were financed from a central source too. Cured patients and the physicians filled it with their private vota.

caskets may have also once held particularly valuable personal belongings (Dyczek in print). A small number of bronze vessels were also discovered in the hospital. These vessels are simple and plain, mostly with handles terminating in lion’s heads. There is one exception – a richly decorated handle with a representation of Harpocrates (Dyczek 1993: 197-204; 1994: 135-139). There is no way to exclude the possibility that the vessel belonged to a sick legionary, but the symbolism Harpocrates was, after all, the god of silence and patience – may suggest that the vessel was part of the hospital furnishings. The tableware assemblage – bowls, beakers, cups, plates, jars and lamps, as well as mortaria of different size (unstamped) - were surely bought from central resources, to judge by the quantities that they occur in and the variety of types that is represented. Neither do they bear any graffiti, possibly identifying a private owner. Unbroken vessels were also found, clearly abandoned, presumably because they had been filled with food leftovers or medicine when the hospital was being cleared. Had they belonged to individual soldiers, they would not have been left in the valetudinarium. Some of these vessels were quite obviously private ceramic copies of bronze vessels used by the legionaries, for example, a clay griffschale.

Most hospital rooms were arranged in series of two sick rooms and a vestibule in between, an arrangement of threes. The material from the vestibules was the richest, as these spaces presumably operated as handy stores. The finds may be separated into two groups: the assumed personal belongings; and various goods, which probably originated from the central stores.

The issue of pottery supplies is debatable. For D.J. Breeze it is the variety of ceramics found on Roman army sites in Britain that is proof against a system of central provisioning. Proof is also provided by the ceramic production centres that worked on military commissions, discovered at York, Holt (Breeze 1977: 136; Greene 1977: 113-127), as well as Viminacium in upper Moesia (Jordović 1995: 95-106). There also exist epigraphic documents from different parts of the empire providing grounds for conclusions about legionaries in charge of ceramic workshops. At Mainz, where the legio XXII Primigenia was stationed, an inscription of 220 states: Cus(tos) castel(li) Figlina(rum): at Noviomagus another inscription, mag(ister) fig(linarum), (CIL XIII 8729): at Bonna, immunes figlinae (Breeze 1977: 136). At Drobeta a brick was found, the inscription on it referring to a soldier who was charged with the operation of a military brickyard: Aurelis Mercurius milis C(o)h(or)tis I Sagitt(ariorum) in figlinis magister super milites LX (Barcacila 1938: 38; Breeze 1977: 136; Popilian 1997: 17). Presumably units belonging to the legio XI Claudia produced building ceramics at Transmarisca as indicated by the stamps: F(iglina) Tra(ns) M(arisca) (Zahariade 1988: 112).

The first group surely included medical instruments. Few of these were discovered at the hospital in Novae. They included spatulae, probes, hooks, needles, and tweezers. All were damaged in some way and unsuitable for recasting which is why they were abandoned in the ruins. They must have once been the private property of physicians. We know from inscriptions that the medical staff included civilian physicians (Kolendo 1999: 62-64). The legionaries-patients were most probably the owners of armaments, fibulae, pendants, cosmetic instruments strigilla - and perhaps terra sigillata, (cf. Dimitrova 1987: 109-151) of which very few fragments were identified at the hospital. Armour, which is quite frequent at Novae, is of particular interest. So far we have discovered fragments of four separate pieces of the lorica segmentata type, a fragment of a helmet, pieces of swords (Fig. 5). This proves that patients took their private equipment to the hospital. It is not to be excluded that the small bronze figurines and small plaques with representations of Asclepios also constituted the private possessions of the soldier-patients.

Analyzing the tableware from the valetudinarium at Novae, we can state with certainty that these vessels were produced at the ceramic manufacturing centres at Butovo, Pavlikeni and Hotnitza. Archaeological excavations on these sites seem to prove that they were not army property and only produced vessels on commission from the army (Soultov 1985).

Remains of medical caskets and glass unguentaria were found in practically all of the vestibules (Dankova 1993: 77-126). They were of a variety of types. Most caskets had mountings of bronze with an impressed motif of concentric circles. These caskets and flasks may have also been part of the hospital equipment, intended to store medicines. The 687

Limes XVIII

The absence of cooking pots from the assemblage uncovered in the hospital at Novae, indeed the absence of cooking facilities of any kind, suggests that even while at the hospital the sick were fed by comrades from the contubernia. The small number of uncovered and fully investigated army infirmaries precludes more positive conclusions. It should be noted however, that in pits in the central court at Caerleon, for example, a certain number of cooking pots was found (Threipland 1969: 115-123). And at Novae a tegulae–lined pit in the courtyard portico was found to contain charcoal. It is difficult to say whether the charcoal was used in braziers for heating or for warming and cooking meals.

opportunity to compare different kinds of amphorae dated to the C1st through C3rd, discovered in the scamnum tribunorum, principia, thermae legionis and valetudinarium. Despite a relative abundance of archaeological material, we are only able to identify general trends. Detailed conclusions will be possible after the excavations have been completed. The first phase of the scamnum tribunorum architecture is provisionally dated from c.AD50 until the end of the C1st AD. Amphorae, whole and fragmentary, are found mainly in pits. The predominant vessels are those from the Greek speaking territories - Dr. 43, Kn 19. Also present, but in smaller numbers, are western amphorae of type Dr. 2-4. This set of types indicates that consumption of the more luxurious Greek wines was relatively frequent here, presumably supplementing the regular everyday diet with Campanian wine. This regularity is easily explained by the status of the residents of the scamnum. On the other hand, the Flavian legionary bath yielded primarily amphorae for wine of Zeest 64 and related types. There were fewer sherds of amphorae once containing Istrian olive oil. While the sherds from these areas could well reflect the trends observed in the fortress at large, the amphorae from the valetudinarium are an important link in discussing the kinds of products consumed in army infirmaries. Here, the definite leader is olive oil from Asia Minor, stored in Zeest 90 amphorae. It was used for consumption foremost, but cosmetic use cannot be excluded; the oil could also have been used in lamps and for dressing wounds.

In virtually every vestibule of the hospital at Novae and in many of the sick rooms amphorae for olive oil were found, along with others for wine and salsamenta. Of special interest are amphorae of type Zeest 90, which I think could have contained olive oil from Asia Minor. On one of the amphorae from Romula there is a fragment of a dipinto: OLEVM (Tudor 1968: 122, No. 170). On another from Novae the inscription is also legible: ΠΟΝΤΙΚΟ [...]; obviously some Pontic product had once been transported in it. The amphorae or rather their contents were purchased most probably for the legion as a whole. Testifying to this are the dipinti on some of the vessels from Novae. At Tirgu Secuiesc (now Romania) an amphora was found with the legend L V M, which I believe is not a record of the volume, as suggested by Z. Székely (Székely 1975: 349), but rather an abbreviation of l(egio)V M(acedonica). From Buridava there is a more developed dipinti referring to the same legion: LE V M - le(gio) V M(acedonica) (Bichir 1985, 100, fig. 11, 2) and a fragment of yet another: Leg (Bichir 1985: 102, fig. 11, 1). In both cases I should think it is a form of address, not a designation of the owner. So in the valetudinarium at Novae amphorae were found with the inscription LEG I ITAL, ]G I IT . Also in upper Moesia, in Boljetin, an amphora rim fragment was found to bear an inscription: LEG (Bjelajac 1996: 55, No. 86). This proves that some of the amphorae of the discussed type were sent directly to specified military units.

Taking into account the finds from other sites, we can discuss some general trends in the economic development of the territories on the lower Danube. In the C1st AD, when the Danubian limes was established and new military units were stationed here, the main line of provisioning was from the western provinces of the empire and to a lesser extent the Greek territories, from where goods were imported quite certainly via the Greek colonies/towns on the Black Sea. Age-long urban traditions and economic power made these colonies the key element in the system of trade exchange for the newly formed provinces. The presence of products from the western parts of the empire may, in turn, reflect a situation that was common elsewhere, eg. in Cyrenaica. When detachments from Spain were stationed there, the number of amphorae from this province grew (Riley 1979: 412). The Hunt papyrus documents a similar situation developing on the lower Danube. In the wake of the military units came commodities from their former area of deployment. Needless to say, they maintained their ties with the old provisioning centres. The incipient network of local economic ties was probably frequently destroyed by military operations. Supplementary provisioning reached this area via the Greek towns in the Aegean. Once the political situation had stabilized in the middle of the C2nd a trend to reorient imports appeared. Ties with the western provisioning centres were slowly broken off. The only commodities that were purchased were those that were unavailable on the nearer Greek or Pontic markets. Local markets grew in importance. Their development was

An analysis of amphorae types (kinds of products transported inside them; Dyczek 1999) gives a general idea of the directions followed by the presumably centrally funded import of products (Fig. 6). We are faced here with the methodological problem of having to estimate the amounts of products transported in the amphorae. Occasionally, estimates are made of the numbers of the amphorae themselves. The results reflect ancient reality only in the case of amphorae of similar volume. On the lower Danube calculations start to become difficult when we see that the amphorae had a small and varied volume. The recipient hardly cared about the number of amphorae he received, as long as he had whatever amount he had ordered. Hence the need to apply a suitable conversion rate. The excavations at Novae have provided a unique 688

Piotr Dyczek: Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae favoured by new settlement in the area, the building of new Roman villas with their domestic facilities, establishment of imperial domains and organization of legionary territories. The military organization and administration of the province presumably exerted no small influence. It should be remembered that the regions that the Romans conquered did not have a strong local agricultural tradition. The local populace was engaged mostly in pastoral activities and fishing. Consequently, the costs of maintaining the legions grew. Provisioning costs turned out to be one of the major drivers in stimulating trade. At the same time, the army was indirectly involved in developing trade contacts and favoured the influx of new merchants.

cuppae (barrels) in Moesia Superior and wells in Rome and Aquincum that had the walls made of old barrels. The representations on steles are dated to the early C2nd AD (Kolendo 1997: 129, 133) and testify to the importance of the local wine industry.

A considerable rôle in the provisioning system was played by Roman villas, of which most were of an agricultural character – the so-called villae rusticae. In Bulgaria today 33 such villas are known; 7 of these were in lower Moesia (Dinčev 1997: 115-119), mainly in the area between the Ciabrus and Iatrus rivers (Mrozewicz 1984: 120). In Dobrudja 93 villas have been identified (Baumann 1983: 148; cf. also Mrozewicz 1984: 120). Most of the villa owners were connected with the military units stationed in lower Moesia. Taking advantage of old ties, these owners surely sold their produce to the Roman army, but not only. The fact that most of them lie on Roman roads in the provinces is proof of this. The villas are usually of medium size (Dinčev 1997: 132). Exceptions include the villas in Vârbovki Livadi, the residential villa in Madara, (Dinčev 1997: 45-47; 74-79) and the villas in the vicinity of Troesmis (Baumann 1983: 107-122). Some of these concentrated on a special kind of production, like the villas in the vicinity of Pavlikeni, which produced pottery and building ceramics, or the villa at Telita in Izvorul Maicilor, where an iron-smelting centre based on local ores was discovered (Baumann 1983: 90-93). The majority, however, was engaged in a variety of agricultural production, including grain cultivation – especially at Dobrudja (Baumann 1983: 27-29), and shepherding. And so, in the Noviodunum area four of the 10 villas were engaged in pastoral activities and the remaining six in agriculture (Baumann 1983: 71-75). From Novae there is a very interesting inscribed stele (Fig. 7):

Neither should we forget that the local populace must have met some of the demand for provisions. Skins, wool and some agricultural products were surely sold on the local markets. What with the progressing romanization, which is reflected in the epigraphic material (Mrozewicz 1982: 117118; 1984: 119-123), a gradual but incremental process was taking place, incorporating the Thracian population in the Roman trade exchange system.

In the middle of the C3rd many of the villas experienced a brief recession (Baumann 1983: 37; Dinčev 1997: 132). Some of them were destroyed, quite probably via barbarian raids. Prosperity returned soon enough. Stable political conditions helped the small villas to rebuild their economic potential and to supply successfully the province with commodities that the general crisis in the empire made impossible or unaffordable to import from other regions.

Data from the excavations at Novae provides the grounds for different conclusions. In this paper I have presented my view of the provisioning of the Roman army on the lower Danube in the C1st-C3rd, based on material from a single characteristic building. Bibliography Baumann V.H. 1983 Ferma romana dîn Dobrogea. (Tulcea). Bărcăcilă A. 1938 Une ville daco-romaine. Drubeta (Bucureşti). Berchem D. van 1937 L’annone militaire dans l’empire romain au 3e siècle. (Paris). Bichir G. 1985 Centrul militar roman de la Buridava. ThracoDacia VI: 93-104. Bjelajac L. 1996 Amfore Gornjo Mezijskog Podunavlja. (Beograd). Breeze D.J. 1977 The fort at Bearsden and the supply of pottery to the Roman army. In J. Dore & K. Greene (edd.) Roman pottery studies in Britain and beyond. British Archaeological Reports Brit. Ser. 30 (Oxford): 133145. Curtis R.I. 1991 Garum and Salsamenta. Production and commerce in materia medica. Studies in Ancient Medicine 3. Dankova G. 1993 Kieliszki i unguentaria z Novae. Materiały do katalogu. Novensia 6: 77-126. Davies R.W. 1970 The Roman military medical service. Saalburg Jahrbuch 27: 84-104. Dimitrova-Milčeva A. 1987 Importierte Terra Sigillata aus Novae. Izvestija na Archeologičeskija Institut 37: 108151. Dinčev V. 1997 Rimskite vili v dnešnata bâlgarska teritorija. (Sofia). Dyczek P. 1993 Harpocrates z Novae. Novensia 5: 197-204. Dyczek P. 1994 Harpocrates on Bronze Handle from the Valetudinarium at Novae. Akten der 10. Internationalen Tagung über Antike Bronzen, Freiburg, 18. - 22. Juli 1988. Forschungen und Berichte zur vor- und frühgeschichte in Baden - Württemberg, Band 45.

Iulius Iero neg [o]tiator [vi]x(it) an(nis) LX h(ic) s(itus) e(st) Iul[ia---] There are two barrels represented on the stele, suggesting that Iulius Iero had been engaged in the wine trade. The barrel here is a symbol, referring to the most popular, the most common kind of situation. For the sake of comparison, let us look at steles known from lower Moesia, including three examples now in the National Museum in Sofia, originating from Kunio, Kamennopole and Teteven (Kolendo 1965: 134, figs. 5, 6, 7), as well as other kinds of evidence from other areas, like the locality 689

Limes XVIII

(Stuttgart): 135-139. Dyczek P. 1995 Was it “Sacellum Aesculapi” in the valetudinarium at Novae, Nunc de suebis dicendum est. Studia archaeologica et historica Georgii Kolendo ab amicis et discipuli dicta. (Warszawa): 125-128. Dyczek P. 1996 Novae - Western Sector, 1992-1995. Archeologia 47: 51-68. Dyczek P. 1999 Amfory rzymskie z obszaru dolnego Dunaju. Dystrybucja amfor i transportowanych w nich produktów w I-III w. po Chr. (Warszawa). Dyczek P. in print Bronze appliqués of wood caskets from the site of the valetudinarium at Novae. Journal of Roman Archaeology. Fink R.O. 1958 Hunt’s Pridianum, British Museum Papyrus 2851. Journal of Roman Studies 48: 102-116. Gerov B. 1988 Landownership in Roman Thracia and Moesia (1st-3rd century). (Amsterdam). Greene K. 1977 Legionary pottery and the significance of Holt. In J. Dore & K. Greene (edd.) Roman pottery studies in Britain and beyond. British Archaeological Reports Brit. Ser. 30. (Oxford): 113-127. Jordović C. 1995 Grnčarski i ciglarski centar u Viminacijumu. Saobštenja Republičkog zavoda za zaštitu spomenika kulture XXVI: 95-106. Kolendo J. 1965 Études sur les inscriptions de Novae. Archeologia 16: 124-148. Kolendo J. 1999 Inscriptions en l’honneur d’Esculape et d’Hygiae du valetudinarium de Novae. Archeologia 49 (1998): 55-71. Kolendo J. & Božilova V. 1997 Inscriptions grecques et latines de Novae. (Mésie Inférieur). (Bordeaux). Mason D.J.P. 1988 Prata Legionis in Britain. Britannia XIX: 163-189. Mrozewicz L. 1981 Les plombes de Novae. Archeologia 32: 7084. Mrozewicz L. 1982 Rozwój ustroju municypalnego a postępy romanizacji w Mezji Dolnej. (Poznań). Mrozewicz L. 1984 Romanizacja Mezji Dolnej - zarys problemu. Balcanica Posnaniensia I. (Poznań): 109124. Popilian G. 1997 Les centres de production céramique d’Olténie. Études sur la céramique romaine et dacoromaine de la Dacie et de la Mésie Inférieure I. (Timişoara): 7-20. Press L. 1988 Valetudinarium at Novae and other Roman Danubian hospitals. Archeologia 39: 69-89. Rickman G. 1971 Roman granaries and store buildings. (Cambridge). Sarnowski T. 1988 Wojsko rzymskie w Mezji Dolnej i na północnym wybrzeżu Morza Czarnego. Novaensia 3. Soultov B. 1985 Ceramic Production on the Territory od Nicopolis ad Istrum (IInd-IVth Century). Godišnik na Sofijskija Universitet „Kliment Ohridski” LXXVI. 2: 1983. Szubert W. 1987 Lampy wolutowe z budowli legionowej w północno-zachodniej części Novae. Novensia 1: 231252. Threipland M. 1969 The Hall, Carleon 1964: Excavations on the site of the legionary hospital. Archaeologia Cambrensis 118: 87-123. Tudor D. 1968 Oltenia Romana. (Bucureşti). Whittaker C.R. 1989 Les frontiéres de l’empire romain. Centre de Recherches d’Histoires Ancienne 85. (Besançon). Zahariade M. 1988 Moesia Secunda, Scythia şi Notitia Dignitatum. (Bucureşti).

690

Piotr Dyczek: Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae

Fig. 1. Lead seal from Novae. Drawing P. Dyczek, J. Janowski.

Fig. 2. Plan of the legionary fortress at Novae. Drawing P. Dyczek, J. Janowski.

691

Limes XVIII

Fig. 3. Plan of the valetudinarium at Novae. Drawing P. Dyczek, T. Słowik.

692

Piotr Dyczek: Remarks on supply of the Roman army from the point of view of the valetudinarium at Novae

Fig. 4. Computer reconstruction of the sacellum at Novae. M. Chruśliński, A. Sawicki.

Fig. 5. Fragments of a lorica segmentata from Novae. Phot. M. Dąbski.

693

Limes XVIII

Fig. 6. Amphorae types found at Novae. Drawing P. Dyczek.

Fig. 7. Stele of Iulius Nero from Novae. Drawing P. Dyczek, J. Janowski

694

Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior) Elena Klenina Ceramic studies are aimed towards establishing the economic boundaries of items based on archaeological and written sources and towards explication of the processes operating in the ceramics industry itself. The pottery represented in this paper was collected during the last field survey at the legionary baths and the Episcopal residence at Novae (Moesia Inferior), situated on the southern bank of the Danube in north Bulgaria. Bulgarian and Polish archaeologists have successfully excavated here for more than 40 years. However there are some groups of pottery which have not been thoroughly characterised. An analysis of the assemblage was made in order to obtain an impression of the general groups and the technical peculiarities of their production. The origins of some groups were also taken into account. According to the chronology of the site the pottery was divided into assemblages to facilitate analysis. We would like to present the preliminary results of our investigations. The early Roman levels (the first half of the C2nd AD) present the entire spectrum of local coarse wares and terra sigillata of North Italian and South Gaulish manufacture. Stability created the conditions for the creation of a local fine ware industry. The second half of the C2nd AD levels consist of pottery produced at centres near modern Pavlikeni for fine wares, and from Hotnitsa (north Bulgaria) and Troesmis for more coarse utensils. Utilitarian pottery from Troesmis is distinguished by its yellowish-brown fabric consisting of sand. Wares from Hotnista are characterised by light-brown fabrics. The pottery produced at the ceramic centre near Butovo spread in northern Bulgaria in the C3rd and C4th. Kitchenwares of the C3rd and C4th, recorded in large numbers, are present in various forms made of a yellowish-brown and grey-brown fabrics. The question of the origins of this pottery is still open. There is a problem in distinguishing ceramics of the C5th and C6th because their study at Novae has only just begun. The amphorae are mostly from Syria, Palestine and western Asia Minor. The red slip fine wares were transported from the Canakkale region of the Dardanelles or the Pergamum region (late Roman C) with some examples from North Africa (present day Tunisia). Local manufacture satisfied the needs of the consumers for utilitarian kitchen grey clay pottery. Ceramic material from Novae will help to solve specific problems concerning pottery production, its distribution and provide new evidence for the site's chronology.

At the site of Novae the study of ceramic assemblages forms an important part of its archaeological investigation. The study is directed towards the problems of naming, dating and the classification of the ceramics as well as analysing their function. It is hoped that the pottery can help in reconstructing aspects of the economic and social character of the Roman to early Byzantine occupation of the site.

and the development of regional manufacture. Contemporary archaeological research is helping academics to understand and define the technical development level of the ceramic production and its resources. The current investigations at Novae attempt to systematize the ceramic material from the site and make comparison with other excavations, both within the region as well as from other parts of the Roman empire. The well-recorded archaeological levels provide the main evidence to make some provisional conclusions. To obtain significant results a whole array of finds from the deposit should be used. The written and epigraphic sources need to be correlated during pottery research. However despite the fact that there is not much by way of surviving written and epigraphic texts the data obtained can still be usefully deployed. The use of a variety of pottery fabric distribution maps and the search of analogies is an important part in the investigation and understanding of pottery production and distribution.

Ceramics are rarely datable in their own right, although there are exceptions such as vessels bearing painted inscriptions or stamps which include the date of manufacture in terms of a consular year (Peacock 1982: 161). Artefacts which may help in giving a date range to ceramic deposits are all taken into consideration, including coins, datable fine wares and other correlating material. However coins may be hoarded and remain out of circulation for a long period or they have a long life before being lost. Therefore they can be difficult to use as a dating material, but if there are enough of them in a given deposit, the latest is likely to be the date when the layer etc. was formed, providing a terminus ante quem for any associated pottery.

This paper is a contribution to the typology and chronology of the table and kitchen pottery dated from the C2nd to the C6th AD from Novae in northern Bulgaria on the southern bank of the Danube. The pottery represented in this paper was collected during the last field season in the area of the legionary bath house and the episcopal residence there

The chronologies of the principal fine wares are now well established in ceramic studies. The pottery provides the most important evidence of the commercial relationship 695

Limes XVIII

(Figs. 1 & 2). The ceramics from the site have been studied by Bulgarian and Polish archaeologists who have excavated there for more than 40 years.

Italian and south Gaulish workshops (Dimitrova-Milčeva 1987: 111; Źmudźinski 1999: 108) and Tralles in Asia Minor (Domżalski 1998: 127). It is known the legio I Italica was recruited in Italy (Иванов 1999: 90) and auxiliaries were formed of Gallic and German recruits (Герасимова 1970: 30). The ethnic structure of the army could explain the fine ware composition. The roads used to supply the military with goods are well-known. The fine wares, with other commodities from Gaul and Italy, were transported by ship to Aquileia and then overland to Emona and finally transported via the Sava and Danube rivers to Novae (Bezeczky 1987: 191). Commodities from Asia Minor were supplied probably through sites along the western Black Sea coast (Sartre 1997: 275). The kitchen ware forms were naturally connected with native traditions of ceramic production (Peacock 1982: 148). Obviously the potters were civilian. The opportunity to sell and trade their products attracted their attention. The Thracian type hand-made kitchen wares represented in the assemblages of the C1st to early C2nd are more than 30% of the entire kitchen ware collection (Милчева & Генчева 1992: 31; Karasiewicz-Szczypiorski 1998: 191, 193). The reason why wheel-made pottery changed its colour from grey to brownish-grey and on rare occasions, to red and brownishorange was a result of firing the pottery at different temperatures (Милчева & Генчева 1992: 31; Karasiewicz-Szczypiorski 1998: 192).

The first attempt at the systematization of the ceramics from Novae was made in the 1960s when the BulgarianPolish archaeological team was first formed. B. Rutkowski made an important contribution to pottery study in Novae and attempted to systematize the material when understanding of the ceramic chronology and typology was still not well-founded (Kołkowna & Rutkowski 1962; Rutkowski 1965). Unfortunately the ceramic material has been studied irregularly in the subsequent years, although there are published catalogues of the vessels (Biernacki 1973; Kuraszkiewicz 1975). Thanks to these studies there is enough evidence to support statistical analysis. Efforts to publish assemblages and use the archaeological material widely were made at the end of the 1980s and in spite of some inaccuracy in interpretation it is necessary to take into account the significance of these observations for future research (Dyczek 1987; 1991; 1992). Terra sigillata of the C1st-C2nd AD, from north Italian and south Gaulish workshops, were researched by A. Dimitrova-Milčeva (1987; 1996; 2000; Милчева & Генчева 1992). In the last few years the ceramic studies have been complimented by a new observations and research (Domżalski 1998). The table and kitchen wares of the Roman and early Byzantine periods have not attracted the attention of researchers until now. Some papers have appeared in recent years dedicated to these types of pottery (Karasiewicz-Szczypiorski 1998; Klenina 1999; in print).

At the beginning of the C1st, 150,000 people emigrated from the Danube and this had an effect on pottery production in Moesia (Sartre 1997: 275). After the conquest of Dacia the stabilization and subsequent economic growth began in lower Moesia. New cities, production centres and trade relationships appeared in this period.

In this paper I will attempt to systematize the pottery found during the excavations carried out by Adam Mickiewicz University from Poznań (Poland). In the 1993, 1994, 1997 and 1998 seasons 754 whole vessels and typologically sensitive fragments were recorded. Analysis of the material was made in order to obtain a general grouping of the pottery according to the chronological periods. The origin of the groups was also taken into account. Grouping by archaeological assemblages enables a comparison of the overall date to be obtained. The pottery was sub-divided into four principal cultural horizons.

The Roman period (second half of C2nd - mid-C3rd) (Fig. 3) The evolved economy in the region attracted new immigrants with mainly veterans occupying the villa estates (Герасимова-Томова 1976: 53; Торбатов 1991: 23). The elite were replaced by the descendants of the Italian veterans and settlers originating from the Greek east who were traders and artisans (Тачева-Хитова 1972: 2932, 38; Sartre 1997: 276). Large estates (villa rustica) owned by veterans appear in the lower Moesia in the C2nd, as indicated by epigraphic and archaeological data (Велков 1962: 31-33; Султов 1964: 63; Генев 1986: 34; Динчев 1997: 28). Pottery production centres within the territory of Nicopolis-ad-Istrum had been established by the early C2nd. There was a quarry and kiln site at Hotnista (Sultov 1985: 21; Biernacki 1997: 79). At first grey wares were produced here for the stonemasons while the red slip wares appeared somewhat later. Hotnitsa supplied the sites nearby with some of the types of kitchen vessels and censers (Dyczek 1992: 68-69; 1995: 237-238; Poulter 1999: 108).

The early Roman period (C1st AD - first half of the C2nd AD) Our knowledge of the early Roman pottery from Novae is based on the published material. There are no early wares from the area of the legionary baths. The first period of occupation was conditioned by the unstable situation in the region. The abandoned area of the lower Danube and the Dacian wars did not help with trade development and establishing local production (Sartre 1997: 275). There are no workshops in this period producing good quality pottery. That found is from north

The ceramic centre near Pavlikeni was probably the main 696

Elena Klenina: Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior). supplier of the fine wares during the reign of Hadrian (117138). This centre comprised a villa with living and household buildings. In its first phase, brick kilns were erected. The production of fine ware and toys started in the second period. The main production period appears to be in the time of Marcus Aurelius (138-170; Sultov 1985: 2223). The fine wares from Pavlikeni seem to have reached Novae in significant quantities (Милчева & Генчева 1992: 33; Dyczek 1992: 68-69). The kiln site was destroyed during the Costoboci invasions with fine ware ceramic products being replaced by coarse greyish-black earthenware (Sultov 1985: 24).

The late Roman period (second half of the C3rd - first half of the C5th) (Fig. 4) The continuous military activities of the second half of the C3rd were one of the reasons which led to the crisis of the Roman empire. Dacia was abandoned c.272 and the borders of the empire established on the right bank of the Danube. The changes and division of the old provinces into smaller units provoked some changes in military organisation. Border and mobile forces were different to the permanently stationed units which had different needs and possibilities for commodity supply (Sarnowski 1988: 123-124). It was the same situation in other provinces (Маджаров 1986: 45; Bjelajac 1996: 126). Land routes began to be used more often for supplying the army. The decreased supply of imported commodities led to an increase in local supply. Workshops erected at the centre of Novae produced glass, bronze and ceramic wares (Dyczek 1999b: 99; Genčeva 1999: 95).

The ceramic centre at Butovo commenced production in the second half of the C2nd and by the early C3rd had become the main production centre, producing fine wares for the lower Danube region. It was situated on a provincial boundary, of lower Moesia and Thracia, and on a major cross-road. This location gave it the opportunity to be a leader in pottery production in the region. The advantageous conditions for an emporia were established under Marcus Aurelius (Sartre 1997: 268). Emporia were markets for local and imported commodities and were controlled by the authority of Nicopolis-ad-Istrum (Poulter 1999: 34).

The kiln site near Pavlikieni stopped production in the second half of the C3rd. The manufacture of pottery for everyday life at Hotnista was reduced to a minimum. However the Gothic invasions did not interrupt life in Butovo. Production continued there until the end of the C4th, though on a more limited scale. Butovo’s pottery was in a great demand in Novae and at sites along the lower Danube (Sultov 1985: 93-94).

These changes, which had an effect on the supply of pottery to the military together with the removal of an export tax, are seen at Novae where its C3rd levels are characterised by the pottery from Botovo (Dyczek 1987: 255; 1991: 101-102; Милчева & Генчева 1991: 33; Klenina in print).

Valens agreement to settle the Visigoths in Moesia Secunda led to full-scale warfare in c.376. Peace was concluded in 382 and the Visigoths settled as foederati in Moesia Secunda (Hoddinott 1975: 340). The archaeological data from the excavations at Novae confirms the written evidence (Парницки-Пуделко 1981: 15; Dyczek 1997: 93; Genčeva 1999. 95). The cessation of ceramic production in Butovo should be connected with the events of the last quarter of the C4th.

The kitchen wares of the second half of the C2nd came from Hotnista and probably Troesmis (Opait 1980: 364; Dyczek 1991: 101; 1992 74). According to evidence from other provinces, some of the coarse vessels may have been produced at legionary workshops (Peacock 1982: 148, 150). The kilns of the second half of the C3rd to the midC4th founded in the territory of Novae and behind the eastern defensive wall near the River Dermendere have been published (Митова-Джонова: 1966: 38; Вълов 1966: 48; Genceva 1999: 98). It was a suitable place for establishing a legionary ceramic centre. There was a good source of both clay and water and a slightly slanting terrain for erecting two chamber kilns. It seems that there was an earlier ceramic centre here. Most of the kilns behind the wall have not been investigated because of the setting out of a beach on the Danube shore.

The destruction of the cities, fortresses and settlements and the military success by the Huns resulted in the need to fortify the defensive system. This also changed the economic activities in the military zone on the lower Danube. The entire defensive system was rebuilt or repaired at the end of the C4th – the first quarter of the C5th (Парницки-Пуделко 1981: 15, 17). Civilian buildings appeared within the legionary camp in this period (Genčeva 1999: 98). It is likely that ceramic manufacture behind the eastern wall ceased to function during the conflict with the Visigoths at the end of the C4th (Митова-Джонова 1966: 40). The principia at Novae served both a military purpose and for the good supply of foodstuffs (Sarnowski 1999: 63). There is no obvious break in pottery production. Many of the old forms and the same fabrics as earlier were still used. The assemblages consist of the local red slip and grey coarse wares. Imported fine wares, notably late Roman C ware from western Asia Minor, reached Novae (Hayes 1972: 323). However amphorae imports coming from the Aegean and western Asia Minor (Peacock & Williams 1986: 189,

The repeated Gothic invasions, starting from c.227, devastated most of the extramural area and disrupted agricultural and industrial production. Novae itself does not seem to have been destroyed. Trebonianus Gallus at the head of an army won a battle near Nicopolis-ad-Istrum (Sarnowski 1988: 94). However there was a decline in the quantity of fine wares at Novae although it has been suggested that the major ceramic manufacturing centres continued to function as late at the start of the C4th.

697

Limes XVIII

193) are rare.

територия. (София). Иванов Р. 1999 Отбранителна система между Дортикум и Дурасторум от Август до Меврикий. (София). Маджаров М. 1986 Казармата на Първа Аврелианова кохорта в Суб радице. Археология. (София). Книга 1: 45-55. Милчева А. & Генчева Е. 1992 Scamnum tribunorum на военния лагер Нове. Археология. (София). 2: 24-35. Митова-Джонова Д. 1966 Пещи за керамиди от Нове, Археология. (София). 1: 38-45. Парницки-Пуделко С. 1981 Крепостните порти на Нове. Археопогия. (София). 4: 9-21. Султов Б. 1964 Една villa rustica край с. Присово, В. Търновский окръг. Известия на окръжния музей В. Търново (Варна). II. 49-63. Тачева-Хитова М. 1972 Ролята на преселниците от Мала Азия, Сирия и Египет в икономическия и культурния живот на Западнопонтийские градове през I-IIIв. В Известия на Народния музей-Варна. (Варна). VIII (XXII): 17-44 Торбатов С. 1991 Римска военна диплома от 145 г. от Нигринианис, Долная Мизия. Археология (София). 1: 23-27. Bezeczky T. 1987 Roman amphorae from the Amber Route in western Pannonia. British Archaeological Reports Int. Ser. 386. (Oxford). Biernacki A. 1973 Ceramika naczyniowa. In S. ParnickiPudełko (ed.) Novae – Sektor Zachodni 1970: 125-129. Biernacki A. B. 1990 Remarks on the basilica and episcopal residence at Novae. Acta Associationis Internationalis Terra Antiqua Balcanica V (Sofia): 187-208. Biernacki A.B. 1997 Remarks on early Christian architectural details made of Proconnesian marble and found in Novae (Moesia Inferior). In A.B. Biernacki & P. Pawlak (edd.) Late Roman and early Byzantine cities on the lower Danube from the 4th to the 6th century A.D. (Poznań): 71-80. Bjelajac L. 1996 Amfore Gornjo Mezijskog Podunavlja. (Beograd). Conrad S. 1999 Zur Typologie und Funktionsbestimmung der Amphoren aus dem Kastell Iatrus. In G. von Bülow & A. Milčeva (Hsrg.) Der Limes an der underen Donau von Diokletian bis Heraklios. (Sofia): 175-188. Dimitrova-Milčeva A.1987 Importierte terra sigillata aus Novae. In Известия на Археологческия институт XXXVII: 108-152. Dimitrova-Milčeva A. 1996 Römische kunstlerische importgegenstande aus Novae. In Зборник Народног музеја XVI-1 (Београд): 167-173. Dimitrova-Milčeva A. 2000 Terra Sigillata und Dünnwandige Keramik aus Moesia Inferior (Nordbulgarien): (София). Domżalski K. 1998 Terra sigillata z komendantury w Novae. Ceramika stołowa z principia w Novae. Wypełnisko jamy nr 4. Novensia 11: 127-137, 141-155. Dyczek P. 1987 Uwagi na temat ceramiki odkrytej w Novae 1979 roku. Odcinek IV, Budowla z portykami. Novaensia 1: 253-287. Dyczek P. 1991 Ceramika z odcinka IV (1981, 1983). Novaensia 2: 101-127. Dyczek P. 1992 Z wstępnych badań nad ceramiką z wypełniska latryny szpitalnej w Novae. Novaensia 4: 67-79. Dyczek P. 1995 Kadzielnica z Novae. Balcanica Posnaniensia VII: 235-245. Dyczek P. 1997 New late Roman horreum from sector IV at

Many towns situated in Moesia Secunda were destroyed in the Hunnic invasion of the 440s (Hoddinott 1975: 239). From the archaeological excavations there is evidence of destruction in the first half of the C5th in Novae as well as in Ratiaria and Nicopolis-ad-Istrum (Велков 1984: 93; Poulter 1999: 20). The early Byzantine period (mid-C5th - C6th) (Fig. 5) Moesia Secunda was included in a new administrative unit of the Quaestura Exercitus. The purpose of this union was to ensure that the army on the Danube received its annonae (Poulter 1999: 47). The episcopate which was established in Novae c.430 or 449 (Ilski 1984: 305-309; Димитров 1992: 84) could have had an influence on the supply of different commodities. The episcopal residence and basilica were erected in the second half of the C5th (Biernacki 1990: 189; Parnicki-Pudełko 1995: 54). There was an increase in imported amphorae from the Aegean and the northern Syria, Palestinian and Gaza regions (Peacock & Williams 1986: 182, 186, 191, 198). One of the amphorae types originated from the southern Black Sea coastal region (Conrad 1999: 184). Some fine ware imports which appeared in this period are not common. African red slip and late Roman C wares had been supplied with other basic commodities in Novae. Local workshops produced the necessities for consumers, including utilitarian kitchen grey clay pottery, amphorae and storage vessels. The abundant ceramic material from Novae will help to resolve a number of problems concerning pottery production and distribution as well as providing further evidence for developing the site’s chronology. Bibliography Велков В. 1962 Към въпроса за аґрарните отнощения в Мизия през ІІ в. На н.е. Археолоґия (София). Книґа 1: 31-34. Велков В. 1984 Възпоменателен надпис за импертор Анастастасий (491-518) от Рациария Археология (София). 2-3: 92-93. Вълов В. 1966 Пещ за строителна керамика от Нове. Ареология. (София). 1: 46-51. Герасимова В. 1970 Дислокация на римските помощни войски в провинция Мизия от 44 до 86г. на н.е. Археология. (София). 2: 22-30. Герасимова-Томова В. 1976 Военна диплома от Nicopolis ad Istrum от 162 г. Археология. (София). 1: 50-54. Генев Г. 1986 Вила рустика при Търговище. Археология. (София). 1: 29-36. Димитров К. 1992Нове на Долния Дунав като раннохристиянски център (V-VI) в. сл. Хр. Balkan Antiquities 2 (София); 83-87. Динчев В. 1997 Римските вили в днешната Българска

698

Elena Klenina: Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior). Novae. In A.B. Biernacki & P. Pawlak (edd.) Late Roman and early Byzantine cities on the lower Danube from the 4th to the 6th century A.D. (Poznan): 87-94. Dyczek P. 1999a Amfory rzymskie z obszaru dolnego Dunaju. Dystrybucja amfor i transportowanych w nich Productow w I-III w, po Chr. (Warszawa). Dyczek P. 1999b A glass atelier from sector IV in Novae. In G. von Bülow & A. Milčeva (Hsrg.) Der Limes an der underen Donau von Diokletian bis Heraklios. (Sofia): 99-104. Genčeva E. 1999 La production dans le camp militaire romain Novae pendant la basse antiquité (a l instar du scamnum tribunorum). In G. von Bülow & A. Milčeva (Hsrg.) Der Limes an der underen Donau von Diokletian bis Heraklios. (Sofia): 95-98. Hayes J.W. 1972 Late Roman pottery. (London). Hoddinott R.F. 1975 Bulgaria in antiquity. (London). Ilski K. 1984 Biskupstwo w Novae a zagadnienie chrystianizacji Mezji Dolnej. Balcanica Poznaniensia 1: 305-309. Karasiewiecz-Szczypiorski R. 1998 Ceramika kuchenna z komendantury w Novae. Wypełnisko jamy nr 4. Novaensia 11: 191-198. Klenina E. 1999 Table and cooking pottery of the IV-VI AD from the excavation of the Episcopal residence in Novae. In G. von Bülow & A. Milčeva (Hsrg.) Der Limes an der underen Donau von Diokletian bis Heraklios. (Sofia): 87-93. Klenina E. in print Earthenware of the II-III centuries AD from the excavation of the legionary bath canal in Novae. Kołkówna S. & Rutkowski B. 1962 Ceramika. In K. Majewski (ed.) Novae 1961. Tymczasowe sprawozdanie z wykopalisk Ekspedycji archeologicznej Uniwersytetu Warszawskego. Archeologia XIII: 114-119. Kuraszkiewicz G. 1975 Ceramika naczyniowa. In S. ParnickiPudełko (ed.) Novae – Sector Zachodni 1972: 115-129. Opaiţ A. 1980 Consideraţii preliminare asupra ceramicii romane timpurii de la Troesmis. In Peuce VIII. Muzeul ‘Deltel Duněrii’. Raporte asupre cercetărilor de arheologie, numistmatică şi istorie 1977-1978: 328366. Parnicki-Pudełko S. 1995 The episcopal basilica in Novae. (Poznan). Peacock D.P.S. 1982 Pottery in the Roman world; an ethnoarchaeological approach. (London). Peacock D.P.S. & Williams D.F. 1986 Amphorae and the Roman economy. (London). Poulter A.G. 1999 Nicopolis-ad-Istrum: a Roman and early Byzantine city. The pottery and glass. (London). Rutkowski B. 1965 Ceramika. In S. Parnicki-Pudełko (ed.) Novae – Sektor Zachodni 1964. Archeologia XVI: 169184. Sarnowski T. 1988 Wojsko rzymskie w Mezji Dolnej i na Północnym wybrzeżu Morza Czarnego. Novaensia 3. Sarnowski T. 1999 Die Principia von Novae im späten 4. Und 5. Jh. In G. von Bülow & A. Milčeva (Hsrg.) Der Limes an der underen Donau von Diokletian bis Heraklios. (Sofia): 57-63. Sartre M. 1997 Wschód rzymski. (Wrocław). Sultov B. 1985 Ceramic production on the territory of Nicopolis ad Istrum (IInd – IVth century). (Sofia). Żmudziński M. 1999 Z badań nad gospodarką i relacjami ekonomicznymi Novae (I-III w.n.e.). Antiquitas XXIV: 101-132.

699

Limes XVIII

Fig. 1. Map of Northern Bulgaria.

Fig. 2. Plan of Novae (C4th – C6th AD).

700

Elena Klenina: Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior).

Fig. 3. Pottery of the Roman period.

701

Limes XVIII

Fig. 4. Pottery of the late Roman period.

702

Elena Klenina: Some remarks about the Roman and early Byzantine pottery from Novae (Moesia Inferior).

. Fig. 5. Pottery of the early Byzantine period.

703

Limes XVIII

704

Was there a militarisation of the southern Balkans during late antiquity?1 Archibald Dunn A valid topic for research is a comparison of the impact of the military in the late Roman era upon the limes and upon the interior. The ability to make such comparisons assumes that many strands of provincial history have already been clarified and modelled, not least that a military factor is easily isolated, which is certainly not yet the case in the southern Balkans. But since a military factor could always be assumed in the northern Balkans for late antiquity, and is gradually becoming clearer with the progress, above all, of archaeology, there are at least patterns to which the southern Balkan evidence, also mainly archaeological, can be usefully compared and contrasted. The archaeology of the late Roman southern Balkans is in dire need of analysis, and even in a semi-analysed state it calls into question the tacit distinction between frontier and interior which has been inherited from the study of the principate. Fieldwork has documented the existence of vast numbers of fortifications of apparently differing types, but relating these to military, administrative, political or demographic history is fraught with problems. There is also relatively much less excavation of such sites in the south than around the limes. Nevertheless an assessment of the unevenly distributed archaeological data from one or two southern Balkan provinces, the Macedoniae, suggests that a military factor was affecting the status and almost certainly the economies of many settlements during late antiquity.

A valid topic for research would be a comparison of the impact of the Roman military upon the limes and upon the interior. The ability to make such comparisons though presupposes that many strands of provincial history have already been clarified and modelled, which is certainly not yet the case in the southern Balkans. But since a military factor could always be assumed in the northern Balkans for late antiquity, and is gradually becoming clearer with the progress above all of archaeology, there are at least patterns to which the southern Balkan evidence, also mainly archaeological, can be usefully compared and contrasted. The archaeology of the late Roman southern Balkans is in dire need of analysis, but even in a semianalysed state it calls into question the tacit distinction between frontier and interior which has been inherited from the study of the principate. Fieldwork has in descriptive terms documented the existence of vast numbers of fortifications of apparently differing types, but relating these to military, administrative, political, or demographic history is fraught with problems. There is also relatively much less excavation of such sites in the south than around the limes. Nevertheless an assessment of the unevenly recorded data from one or two southern Balkan provinces, the Macedoniae, suggests that a military factor was affecting the status and almost certainly the economies of many settlements during late antiquity. Here we must confine ourselves to establishing whether there was a military factor, its periodicity, and some of the concerns which the new fortifications seem to reflect. The question “Was there a militarisation of the province or provinces of Macedonia in the late Roman period?” is a question begged by the archaeology of the region. It calls for a working definition of militarisation, an evocation of the archaeology and its problems, and reference to the wider problems of historical sources.

To introduce the problem of historical sources before that of archaeology, the only literary reference to the stationing of soldiers in the hinterland of Macedonia (as opposed to the presence of soldiers with emperors staying in Thessalonica), after the mid-C3rd crises and before the reign of Justinian, are Malchus’ references from the reign of Zeno (474-491) to a garrison (stratiôtai) at Stobi2 and, crucially, to “those units (tagmata) scattered among the cities (poleis)” (Historia Malchi p.251. I.13-16). Can we evaluate these crucial references of the late C5th archaeologically? The broad distinctions, chronological and typological, adopted here will be familiar to people studying the Balkan limes. Firstly, clear distinctions can be made between the designs and building materials of Hellenistic to midRoman fortifications, fortifications built hastily during the crises of the mid-C3rd, fortifications built between then and the early C7th, and later Byzantine fortifications.3 Once these broad phases are recognised, it becomes clear that the greatest investment in fortifications took place after the mid-C3rd crises and before the middle Byzantine period. This comparative statement is partly based upon a study of central and eastern Macedonia in the middle Byzantine period (cf. Dunn in progress). Secondly, it can be recognised in Greek Macedonia, as it has been by specialists working elsewhere in the southern Balkans (in Albania and ex-Jugoslav Macedonia (= FYROM) - which includes the northern parts of late Roman Macedonia), that late C3rd to early C7th defenses fall into different, if overlapping, categories according to the size of their defended areas, their architectural scale, and their relationship to the topography of ancient settlements and communications.4 Scholars in Albania and

If one defines militarisation here as the long-term distribution of regular soldiers across parts or all of a province, with whatever that implies for diversion of resources to defenses and personnel, for settlement patterns, and for fiscal arrangements, then I suggest that we have a set of headings under which what is thought to define frontier provinces can be sought for comparative purposes in provinces of the interior during late antiquity.

1

I would like to acknowledge the support and advice of the Greek Archaeological Service (the Ephoreias of Kavala) which I received when carrying out the site inspections which underlie some of this work. 2 Historia Malchi p.245, 1.2-4. I do not propose to discuss the precise date (?479) of the events concerned, the attacks by one part of the Ostrogoths upon the southern Balkan provinces. 3 For Hellenistic and Roman, see Velenis 1988: Chs. 1 & 2; Lander 1984; for Roman and late Roman fortifications in the Balkans, see for convenience Biernacka-Lubanska 1982; for later Byzantine plans and structures Foss & Winfield 1986. 4 For instance, Baçe 1976: 62-74; Mikulčić 1974: 191–212; 1986: 221-

705

Limes XVIII

FYROM have been tempted to pin Latin labels to these walled sites, namely, oppidum, castrum (or castellum), vicus, pagus, refugium, and of course civitas (polis) in some documented cases. Now these interpreters, Mikulčić, Baçe, and others5, while noting garrison-forts, have tacitly implied the civilian status and functions of all other walled settlements. They indeed have some unmistakable late Roman garrison-forts, with masonry-built living quarters lining the interior faces of enceintes, which they naturally call castra.6 But these form a tiny fraction of their totals of fortifications, and this feature has not yet been identified at a single reported site in Greek Macedonia (though very few relevant sites have been properly explored there, it has to be said).7 Furthermore excavations in the northern Balkans also show that by no means all late Roman garrisoned fortifications had soldiers’ quarters built against their enceintes.8 And Malchus’ reference to the presence of soldiers at Macedonian ‘cities’ is partly to physical cities at which there is no evidence of garrisons’ quarters as such (eg. in the excavations of Stobi). Barracks are elusive then, and their absence or presence not the most useful archaeological yardstick of militarisation. But even without Malchus, and without obvious soldiers’ quarters, we would still need to ask, among other questions: can one discern among the fortifications, many of them massive, but many of them not, which ones were probably the responsibility of the imperial army? And although one can confidently ascribe many walls, or phases of walls, to the late Roman era, can any phase or phases of militarisation yet be identified within that era?

been the same concerted effort in all parts of Greek Macedonia, where some late Roman material has not been recognised as such, so that sites (or phases of sites) to which I refer have often been reported as ‘Roman’, ‘ancient’, ‘Byzantine’, ‘medieval’ or even ‘Turkish’. Naturally the archaeological case for a degree of militarisation has architectural, geographical, topographical, and material-cultural aspects. But, owing to the state of research and publication, not all of these are yet clear at many sites. The military-architectural arguments would be based upon an interpretation of the differences among late Roman fortifications. There are a number of architectural or constructional features, which are familiar to all limesinvestigators, on the basis of which I would divide nearly all late Roman fortifications in the province into either imperial or communal - of which the majority would be categorised as communal. There are probably also a few private ecclesiastical fortifications, as in other Balkan provinces, which cannot be discussed here.9 What I identify as ‘communal’ are those fortifications in which a simple wall encircles a hilltop, forming the ‘Ringwall’ familiar to archaeologists of northern and central Europe and of the Balkans.10 They lack any ‘architectural’ features; the defensive wall is sometimes weak, around 1m thick, sometimes lime-mortared, sometimes of drystone construction. Those that have been explored, though in all but one case only by non-intensive surveys11, contain various combinations of a simple early Christian church, cisterns, stone house-footings, undecorated late Roman pottery types, and late Roman coins (Dunn 1997: 138-140). They do not tend to overlook lines of communication. They vary in size between 0.5ha and c.20ha.12 This is the same range of sites which archaeologists in FYROM have called vici, pagi, refugia, or, if larger, oppida.13 Many sites of this kind situated further north and east have been assumed to be military foundations (Biernacka-Lubanska 1982), though the assumption has been sensibly questioned on the basis of settlement-patterns and topography (Poulter 1983: 97-100). There is therefore a lack of consensus about the origins and functions of these simple fortifications in the Balkans.

Regarding the first question, it is already clear that there are several grounds for doubting that the presence of barrack-rooms is the only useful structural indicator of the presence of soldiers in a southern Balkan settlement, and for doubting that the hypothetical interpretation of nearly all walled sites as oppida, vici, pagi, refugia, or civitates, puts them in a fully ‘civilian’ category which would be distinct from diagnostic castra or castella in Macedonia and the southern Balkans in general. This does not simplify the task, but a large area of potential misunderstanding is thereby removed. To explore further the implications of archaeology for the question of a late Roman militarisation of Macedonia (Fig. 1), I must focus upon parts of the late Roman province within Greece, so as to use first-hand knowledge, though not to the exclusion of parts of the late Roman province within FYROM, where concerted efforts have been made to record late Roman material of all kinds. There has not

Arguably the contrast between these and other types of late Roman fortification points to their radically different status and functions however. The other types that one can identify in modern Greek Macedonia are: powerful 9

See Moutsopoulos 1987: 129-194; Kissas 1988: 207-217. See in FYROM for instance pagi murati and refugia in Lilčić 19881989: 115-136; for Thrace Gočeva 1985: 97-108. 11 For the clear exception see Peristeri 1990: 469-476. Further reports may place the site of ‘Kouri’ at Ossa, excavated by Polyxeni Adam-Veleni, in this category though. 12 Respectively ‘Kastello’ (Toroni) and ‘Kalelik’ (Xiropotamos) (work in progress). For ‘Kastello’, see Kampitoglou & Papadopoulos 1989: Fig. 1 (p.440). 13 The Tabula Imperii Roman (TIR) for Macedonia, which is based on the same research, is more circumspect however: Tabula Imperii Romani. Naissus - Dyrrachion - Scupi - Serdica - Thessalonike. 10

251, 253-277; Dunn 1997: 137-151. 5 See citations (other than Dunn) at n.4. They are followed by others, notably Lilčić. 6 For instance, Popović 1984: 181-243; Baçe 1995: I: 502-508; Karaiskaj 1995: II: 892-897. 7 Some late Roman garrison forts are identifiable on the bases of design and of finds (v. i.). 8 But late C5th to late C6th Nicopolis-ad-Istrum contained a long barracks or storage-facility towards the centre of the enceinte: Poulter 1995: 4042.

706

Archibald Dunn: Was there a militarisation of the southern Balkans during late antiquity? systems built in a careful opus mixtum with towers or a version of Philo of Byzantium’s ‘saw-toothed’ façade, and with proteikhisma, as found at the great civitates of Thessalonica and Philippi (Fig.2, Sites 1 & 2).14 Without obvious proteikhismata, and with or without opus mixtum, the same powerful towers and massive walls up to 2.5m thick occur at many other civitas-sites, particularly on the line of the via Egnatia (eg. at Pella’s late Roman location (Fig.2, Site 3), Amphipolis (Fig.2, Site 4), Neapolis, modern Kavala (Fig.2, Site 5) - only visible in rescueexcavations, and at Topeiros, technically in Rhodope (Fig.2, Site 6).15 One can also identify the rectangular, or at least rectilinear, castrum, with the same powerful military architecture and with opus mixtum, but in non-civic, essentially non-urban contexts (Fig. 3, Sites 1 & 2)16, and a less powerful military architecture with opus mixtum, which was used to fortify selected stations of the via Egnatia - for instance the station of “Pennana” between Apollonia and Amphipolis (Fig.3, Site 3) (Samsaris 1986). Finally one can identify a type of very small roughly rectangular fort with no obvious original towers, of limemortared undressed stone, on the line of the via Egnatia for instance within the medieval site of Rentina by the site of an abandoned Greco-Roman town (Fig.4, Site 1)17, or at the two forts of Khortokopi which overlook the junction of two side-roads or by-ways of the Egnatia to the west of Philippi (Fig.3, Sites 4 & 5).18

defend what had become under the principate a provincia inermis.19 A second set of administrative decisions is indicated by the well-known early C6th list of civitates, the Synekdémos of Hierokles20, which offers considerable evidence of Macedonian civitates being in areas where the only significant late Roman sites are small, if in a few cases powerful, fortifications (see Fig. 2). Some of these civitates bear the names of Greco-Roman towns which had no urban successors, eg. around ancient Akanthos and the Heraclea of eastern Macedonia21, while several others bear the names of upland districts, names never demonstrably attached to cities or towns, Greek, Roman, or Byzantine, eg. EORDAIA and ALMOPIA in western Macedonia22, PELAGONIA in northern Macedonia23, and the KLIMA MESTIKON (“district of/by the River Nestos”) (cf. Sophocles 1887; Lampe 1981), which precedes in the Synekdémos AKONTISMA (literally “Set of Javelins”) (Synekdémos 640a and b), an identifiable fort on the via Egnatia (Koukouli-Khrysanthaki 1972). PRIMANA, another unlocated late Roman civitas, thought from its position in Hierokles’ sequence to be in southern Macedonia, seems to bear a name derived from that of a military unit under the control (in the C4th) of the eastern Magister Militum Praesentalis, the Primani (Notititia Dignitatum Or. vi.45). These 8 late Roman Macedonian civitates consisted of small fortifications and rural settlements, and moreover almost certainly do not constitute the complete checklist of this category.24 Such units of administration would almost certainly have been dominated by the praetorian prefecture of Illyricum (based in Thessalonica for most of the late Roman period: v.i.) and by military commanders.

Turning to the geographical and topographical aspects, the distribution of these types of late Roman fortifications in relation to the geography of earlier and contemporary urban settlements, is suggestive, bearing in mind the scale of the investment in new fortifications, of several kinds of administrative decision. Firstly, the relationship of late Roman fortifications to earlier and contemporary urban settlements shows a concentration of resources upon only about half of the settlements that had enjoyed urban, municipal, or colonial status under the principate (Albanian and Jugoslav scholars have found the same patterns and proportions in the neighbouring provinces of Epirus Nova and Praevalis) (Dunn 1997: 138). There is a relationship to be explored between the increased strength of the new defenses relative to their predecessors’, the scale of the investment at both urban and non-urban sites, the reduction in the number of traditional large settlements privileged with defenses, and an allocation of soldiers to

A third set of administrative decisions is indicated by (Fig. 4) the growing density of late Roman fortifications along the via Egnatia, both of towns and of “stations” which were selected for fortification. When contrasted with the hinterlands on either side of the road scattered with abandoned Greco-Roman towns25, and when compared with similar fortification-programmes in Epirus Nova 19 For Roman Macedonia’s demilitarization, see for convenience Kos 1977: 277-296. 20 Honigmann 1939. The idea that it is not a list of civitates but in the first instance a list of bishoprics has never quite ceased to circulate, but there is no opportunity to deal with it here. 21 There is no dispute about the location of Greco-Roman Akanthos on the south-east coast of the Khalkidiki. I follow Papazoglou 1988: 368-371, for the area in which this Heraclea should be sought. 22 Papazoglou 1988: 159-167 (Eordaia), 169-173 (Almopia), in extensive reviews of the Greco-Roman and late Roman evidence, can find no “evidence” of towns other than that of the Synekdemos itself, which then becomes “proof’ of the existence of late Roman “cities” of these names. No late Roman towns or cities have been identified in these areas since the mid-1980s either. 23 Papazoglou 1988: 266-267, 283-289, for exactly the same problems and ‘solutions’ as offered regarding Eordaia and Almopia. 24 Other civitates dominated by fortresses probably include Kellai/Cellae (Fig. 2, no. 11) and Nicaea, the “Nikedê” of the Synekdémos (Fig. 2, no.12), which cannot for reasons of space be discussed here. 25 Papazoglou 1988: 436-442 and Tableau Synoptique, particularly for the ancient districts of Bottiaia and Mygdonia.

14 For Thessaloniki, see now Velenis 1988: Chs. 3 & 4; for Philippi, still Ducoux & Lemerle 1938: 5-19; Roger 1938: 20-41. 15 For late Roman Pella, see now Khrysostomos 1995: 117-136; 1997: 215-232; for late Roman Amphipolis Bakirtzis 1996: 229-241; for some other civitates, Dunn 1997: 140-141 with further references. 16 For instance the first phase of the walls of Drama (Fig. 3, Site 1. Not its late Roman name): see Dunn 1997: 139 for illustrated reports, which do not however recognize the primary late Roman phase; for the demonstrably late Roman castrum at Mandraki, Samsaris 1976: 125 (where it is called ‘ancient’); AD 27/B3, 1972: 577 (where it is called ‘medieval’). 17 Phase 1 of this late Roman and Byzantine fortification: Moutsopoulos 1989: 199-201; 1990: 199-208, Figs. 1 & 2 for delineations of the late Roman phase. 18 See AD 34/B2, 1979, p.332; AD 37/B, 1982, p. 327; Schütz & Unger 1981.

707

Limes XVIII “Klima Mestikon”.31 There are rectilinear castra or smaller forts very obviously over-looking parts of the Egnatia and its tributaries briefly introduced above under the military architectural heading. There is the probability that some refortified civitates in south-eastern Macedonia, such as Aralos, also had their own strongly fortified refugia in the hills west of the Strymon (Fig. 3, Sites 9 - 12)32, a development attested in northern Macedonia c.479 around Heraclea Lyncestis (Historia Malchi p.249, 1.16-21) and, just beyond historic Macedonia, around Scupi (Skopje) in Dardania in the C6th (Dunn 1998: 800-1, fig. 4 (p.806)). The problem then becomes one of distinguishing between an apparent minority of upland fortresses which were designed by and for the military (eg. one of the C6th successors of the civitas-site of Scupi), and the majority which resemble other lower-status communal refugia. Sites in the hinterland of Aralos pose this problem. There is, I would suggest, an example of the former category, a posturban imperial and military successor to a civitas-site, at the fortress of “Neposi”, near Palaiokhorion, in the Khalkidiki, which could have become the centre of the bythen non-urban civitas of Akanthos (Papangelos 1998: 87, pl. 12; in press re the primary C6th phase). Furthermore there is other good evidence, literary and archaeological, that the late Roman administration built or re-built fortifications at or near several other traditional urban centres of varying status in this region south of the via Egnatia (at Apollonia, Chalcidician Beroia, and Kassandreia for instance).33

(Baçe 1976) and the provinces of Aegean Thrace along the route of the via Egnatia (Moutsopoulos 1979), this suggests that the principal east-west axis of the southern Balkans was becoming militarised in our general sense of the word. It was connected of course outside Thessalonica with a great north-south route leading to the limes26, which in northern Macedonia has been shown to have been heavily fortified in late antiquity (Mikulčić 1984; 198889), a route which also continued southwards. In southern Macedonia Evterpi Marki and Andrew Poulter have identified a cluster of late Roman roadside fortifications (Fig. 2, Sites 9 & 10) between Pydna and Dium on this route which are unquestionably of imperial (public) origins (Marki 1996; Poulter 1998). The epigraphic record of repairs to the Egnatia seems to stop in the 360s or 370s27, but the epigraphic habit is a cultural phenomenon. The munus romangareia of road-building did not stop28, nor did the protection of these two routes. The Notitia Dignitatum supplies evidence in the toponymics of military units for the western end of the Egnatia at a slightly later date (Hoffman 1969: 20, 52-54, 226). The excavations and survey conducted on the north - south route reveal highstatus occupation in the C6th (Poulter 1998). Procopius records new or renovated fortifications on the Egnatia, though without calling it by its Roman name, for instance (in central and eastern Macedonia and on the boundary between Macedonia and Rhodope) at Kyrros, Artemision, Neapolis and Topeiros (Fig.4, Sites 2, 1, 3, and 4 respectively) (Procopius De Aedificiis iv.4; iv.3.30; iv.4, iv.11), fortifications which can be identified.29 And 7 of the 8 late Roman tombstones of soldiers found throughout the province of Macedonia before 1983 were found at four civitates on or beside the Egnatia in Greek Macedonia. All seem to be of the C5th or C5th to C6th.30

The significance of a late Roman military presence in Macedonia, its rationale, periodicity, and impact, have never been posed.34 Discussion must here be limited to further problems of rationale and periodicity. As already observed, the late Roman archaeology of this province strongly indicates a wide distribution of soldiers. Besides the protection of Illyricum, the rationale of their distribution around civitates, and especially among civitates and stations of the via Egnatia, is that, even if the Egnatia were not militarily important, this distribution would have been a logical response to the logistical and fiscal problems analysed by Ramsey McMullen, whose

The archaeological, geographical and topographical aspects of the case for a degree of militarisation of this Aegean province could be explored much further. There is for instance the possibility that several fortifications or ‘walled settlements’ near to the River Nestos in unurbanised upland areas (Fig.2, Site 7, and Fig.3, Sites 6 8) were military sites of the non-urban civitas of the

31 For Akontisma, see Koukouli-Khrysanthaki 1972. For Fig. 3, Site no.6 see AD 25/B2, 1970, p.402. Fig. 3, Site no.7 is at Ano Pyxari, which Professor Nicholas Moutsopoulos (Univ. of Thessaloniki) confirms is late Roman. For Site no.8, see Peristeri 1990. 32 For the location of Aralos, see now Papazoglou 1988: 362-363. For its late Roman walls (though not dated by the author), see Welch 1918-1919: 64-66. The other diagnostic late Roman features are not recorded by Welch or Papazoglou though. For one possible refugium (Fig. 3, Site no. 10) see Samsaris 1976: 172, and for its standard late Roman masonry pl. 11 & 12 (p.173) (inexplicably dated here to “the Macedonian era”). For another of the possible refugia of Aralos, ibid. 118-119 (including pl. 3, whose typically late Roman masonry is called “ancient”). 33 For late Roman Apollonia, see Moutsopoulos 1993: 999-1110; for the late Roman phase of Chalcidician Beroia, see Pazaras 1996: 313-332; for late Roman Kassandreia Pazaras 1987: 157-192. 34 See the general histories of late Roman and Byzantine Macedonia: Theokharidis 1980: Lemerle 1945. But the late Roman military archaeology of northern Macedonia (areas now within the FYROM) is the focus of much primary analysis. See Mikulčić 1974; 1986; 1998-1989 for instance.

26

See TIR 1976 the map “K34 Sofia” in fine. 27 Collart 1976: 196. Subsequent publications, notably by the Centre for the Study of Roman Antiquity of the Greek National Research Foundation, have not altered this finding. 28 See for instance The Theodosian Code and Novels and the Sirmondian Constitutions 1952: 11-16-10, 11-16-15, 11-16-18, 15-1-36, 15-3-3, 15-34, 15-3-5 and 15-3-6, of the emperors Julian to Theodosius II. 29 For Kyrros: Bakalakis 1970: 172-183; Khrysostomos 1990: 205-238, site no. 12B. For Artemision the identification with the late Roman phases of Rentina proposed by Moutsopoulos, the excavator, seems reasonable. For a Roman and/or late Roman outer enceinte at Neapolis (late medieval and modern Kavala) and associated C6th finds see AD 41/B, 1986: 175. For Topeiros see Koukouli-Khrysanthaki 1987-1990: 81-102, for a convenient synthesis of the literature and a plan, and PolykhronidouLoukopoulou 1989: 577-599; AD 47/B2, 1992: 503, 515. 30 Feissel 1983: nos. 26-27 (Edessa), 63 (Beroia), 150 and 152 (Thessalonica?), 153 and 205 (Thessalonica), and 246 (Philippi). Beroia was and is situated between the via Egnatia and the great north-south route passing via Pydna and Dion (Fig. 2; site nos. 9 & 10).

708

Archibald Dunn: Was there a militarisation of the southern Balkans during late antiquity? study of Valentinian I’s redistribution of soldiers to new small forts on parts of the limes deserves to be quoted: “there would be tremendous savings through distributing men in small stationary groups at natural market-centres: in riverine forts, in towns, and at crossroads” (MacMullen 1984: 577). This deserves detailed examination, civitas by civitas. This is arguably the context in which to set also the creation of numerous non-urban civitates in southern Macedonia. The rationale of those 8 (but almost certainly more) late Roman civitates of the Synekdémos situated in southern Macedonia which lacked towns or cities is most probably that their creation was never about fostering cities (sometimes the stated aim of an emperor), but that they were destined to support a long-term redistribution of soldiers, and the up-keep of highways and fortifications. But defensive, logistical, and fiscal factors affecting the distribution of soldiers so far from the limes are arguably not the whole story. In Macedonia there would also have been underlying politico-administrative factors stemming from the functions of Thessalonica within the late Roman state.

Narrative sources, imperial legislation, and other genres can be unhelpful when it is a question of developments in a particular region or province, but the aspect of late Roman Macedonia which concerns us, some of the circumstances in which southern Macedonia’s numerous late Roman noncommunal fortifications were built, is arguably traceable without mere extrapolation from references to exhortatory imperial decrees about the protection of “Illyricum”. The key is the relationship between epigraphy and the Notitia Dignitatum (that tabulation of the senior civil and military positions in the late Roman state, and of their officia, and of the distribution of military units; cf. Seeck 1876). The multiple arguments of Hoffmann that the Notitia Dignitatum’s disposition of military units in eastern Illyricum reflects the situation in the reign of Theodosius I seem to stand (Hoffman 1969: 52-54). The Notitia provides enormous lists of military units stationed in the northern provinces of eastern Illyricum (Moesia I and Dacia Ripensis) which were under the direct command of duces (ND Or. xli; xll). One could infer then that the 26 units under the direct command of the Magister Militum per IlIyricum were essentially distributed further south (ND Or. ix). No provincial breakdown is given, but the Greek epitaphs of late Roman soldiers found at cities on or by the via Egnatia (dateable to the C5th or C5th to C6th) record accurately four of these very units, the Ascarii Juniores, Atecotti, Germaniciani, and Secundani.38 One could argue that others, not yet recorded epigraphically, among the magister’s 26 units, not forgetting also the Primani (v.s.), were distributed in other parts of Macedonia, to the north and south of the Egnatia (v.s.). This is not to suggest that all these arrangements were made so late, but that the distribution of the magisters, and probably other, troops around Macedonia by the late C4th (following Hoffmann’s dating of the relevant part of the Notitia), when combined with the frequent presence there of the praetorian prefect from 375 onwards, would have provided a long-term stimulus to fortification-building of all kinds.

These latter putative factors immediately bring into play the much discussed origins of Thessalonica’s own great fortifications, the problem of the periodicity of such building programmes, already posed, and of course the problems of the purely historical evidence (or lack thereof). In this respect the frequently discussed temporary presences of emperors in Thessalonica, principally Galerius, Constantine I, and Theodosius I, were not necessarily decisive. Detailed analyses and arguments have not established that any of these emperors were responsible for the two total reconstructions of Thessalonica’s fortifications which certainly occurred during late antiquity.35 The new arguments that either Galerius or Constantine (or both) were responsible for additions to the first reconstruction of the fortifications (not forgetting Constantine’s harbour-works), and that Theodosius I at least initiated the second and definitive rebuilding of the fortifications, are persuasive (Velenis 1988: 89-90, 94 and ch.IV). But there is almost a consensus that that task was only completed in the reign of Theodosius II, and the association of this with a praetorian prefect of Illyricum in the early 440s now seems certain.36 If, as has been well argued, the praetorian prefect of Illyricum was closely linked with the city of Thessalonica from the very foundation of the prefecture c.37537, then prefects would have been primarily responsible for the programme (and even for its protractedness). The prefecture, given its powers and responsibilities, would also from that date have been the catalyst for the re-fortification of provincial sites. The problems of the historical sources have to be faced though.

The lack of conventional narrative references to such a process should not in itself be deemed problematic. The mundane business of the praetorian prefect of Illyricum, including defense, its infrastructure, and their fiscal support was not the stuff of narrative history. But the process, once recognised, is of immense significance for understanding the archaeology, hence history, of late Roman Macedonia. On one level it is a process which rather marginalises emperors. The tendency to assume that every significant late Roman monument in Macedonia, secular or religious, must have been an imperial initiative was undermined in a series of articles by Michael Vickers concerning the monuments of Thessalonica39, but the significance of this methodological gain for interpreting the archaeology of the wider region may not have been

35 I follow the arguments of Spieser 1984: 63ff. for one major mid-C3rd rebuilding, and 1999: 570-571, for a late C4th to mid-C5th rebuilding. 36 Spieser 1999: 570-571; Croke 1978: 251-258 for its completion in the early 440s. Croke’s research seems to have been ignored by the archaeologists. 37 Theokharidis 1980: 49-50, 56, 67, 75, 97, 99-100, developing arguments to be found in Lemerle 1945.

38

Feissel 1983: nos. 153, 205, 26 and 27 respectively (with the editor’s observations that these units were under the command of the Magister Militum per Illyricum). 39 Some of Vickers’ chronologies have been successfully challenged by Spieser and others, but the methodological gain remains.

709

Limes XVIII

properly recognised. Of course an individual emperor might endow an individual city with powerful walls, as Diocletian seems to have done at Diocletianopolis in south-western Macedonia (Papazôtos 1988). But there are now good reasons for suspecting that great programmes featuring all kinds of fortification, from the vast enceinte of Thessalonica to the tiniest lookout-post by the Egnatia, were the works of successive praetorian prefects from about 375 onwards, rather than of an individual emperor. Certainly not all late Roman non-communal fortifications, nor even all phases of such fortifications, can belong to this era of delegation to the praetorian prefects. The tetrarchic emperors were quite active. And the era of Justinian arguably experienced some kind of reversion of the initiative to the emperor (and one does not have to take Procopius’ De Aedificiis at face-value to accept this). A reasonable working hypothesis might be that, after the response to the invasions and civil wars of the mid-C3rd, and the activities of the tetrarchy and of Constantine the Great, the next, and greater, phase of public fortificationbuilding was instituted in the southern Balkans by the new praetorian prefecture of Illyricum.

Procopii Caesariensis opera omnia III/2. VI Libri ... De Aedificiis (J. Haury (ed.) Leipzig 1913) Tabula Imperii Romani (K34) Slovenian Academy of Science 1976 AD Arkhaiologiken Deltian Baçe A. 1976 Fortifications de la Basse Antiquité en Albanie. Monumentet 11: 62-74. Baçe A. 1995 Einige Erscheinungen der Spätantike in der Festung von Paleokastro. In Akten des XII. Internationalen Kongress für christliche Archäologie I. (Munster): 502-508. Bakalakis G. 1970 To latomeio tês arkhaias Kyrrou? Arkhaia Makedonia (I), (Thessaloniki): 172-183. Bakirtzis Kh. 1996 Anaskaphê khristianikês Amphipoleôs. Praktika tês Arkhaiologikês Etaireias: 229-241. Biernacka-Lubanska M. 1982 The Roman and early Byzantine fortifications of Lower Moesia and Northern Thrace. (Wroclaw). Collart P. 1976 Les milliaires de la Via Egnatia. Bulletin de Correspondance Hellénique 100: 177-200. Croke B. 1978 Hormisdas and the late Roman walls of Thessalonika [sic]. Greek, Roman & Byzantine Studies 19: 251-258. Ducoux H. & Lemerle P. 1938 L’acropole et l’enceinte haute de Philippes. Bulletin de Correspondance Hellénique 62: 5-19. Dunn A. 1997 Stages in the transition from the late antique to the middle Byzantine urban centre in S.Macedonia and S.Thrace. Aphierôma ston N.G.L. Hammond. (Thessaloniki): 137-151. Dunn A. 1998 Heraclius’ ‘reconstruction of cities’ and their sixth-century Balkan antecedents. Acta XIII Congressus Internationalis Archaeologiae Christianae II. (Vatican): 795-806. Feissel D. 1983 Recueil des inscriptions chrétiennes de Macédoine du IIIe au VIe siècle. (Ecole française d’Athènes). Foss C. & Winfield D. 1986 Byzantine fortifications: an introduction. (Pretoria). Gočeva Z. 1985 Der thrakische Festungsbau und sein Fortleben im spätantiken Fortifikationssystem in Thrakien. Griechenland – Byzanz – Europa. (Berlin): 97-108. Hoffman D. 1969 Das spätrömische Bewegungsheer und die Notitia Dignitatum. Vol. 1. (Dusseldorf). Kampitoglou A. & Papadopoulos G. 1989 Oi anaskaphes stên Torônê. Arkhaiologiko Ergo stê Makedonia kai Thrakê 3: 439-449. Karaiskaj G. 1995 Städte der Spätantike in Albanien. In Akten des XII. Internationalen Kongress für christliche Archäologie II. (Munster): 892-897. Khrysostomos P. 1990 Ê topographia tês voreias Vottiaias. Ê Pella, ê apoikia tês Pellas, kai oi khôres tous. Mnêmê D. Lazaridê. (Thessaloniki): 205-238. Khrysostomos P. 1995 Anaskaphê stê rômaikê kai vyzantinê Pella kata to 1995. Arkhaiologiko Ergo stê Makedonia kai Thrakê 9: 117-136. Khrysostomos P. 1997 Anaskaphikes ereunes stên pellaia khôra kata to 1997. Arkhaiologiko Ergo stê Makedonia kai Thrakê 11: 215-232. Kissas S. 1988 Ê anaskaphê stên Kolkhida Kilkis. Arkhaiologiko Ergo stê Makedonia kai Thrakê 2: 207-217. Kos M. 1977 The military role of Macedonia from the civil war to the establishment of the Moesian Limes. In J. Fitz (ed.) Limes. Akten des XI. Internationalen

This has been a generalising approach to the origins of certain types of late Roman fortification found in Macedonia. It does not, and cannot, deal for instance with the problem of the refinement of the chronology of late Roman communal fortifications. Only stratigraphic excavations of the kinds practised on the fortifications of the limes will enable this (or an alternative approach) to be tested. Military gravestones point the way, and the right kind of exploration of foundation-trenches has certainly begun: around Pydna (v.s.), at Amphipolis (Bakirtzis 1996), at Thessalonica itself (Velenis 1988), and at the civitas-site of Dium (Dion) in south-western Macedonia, where the excavators have established a terminus post quem for the powerful late Roman walls at c.375. This they link with Theodosius I’s response to the aftermath of the Battle of Adrianople (378).40 But equally we are at the foundation of the new praetorian prefecture of Illyricum. Progress is likely to be slow, but arguably the problem is worth posing for an orientation of future archaeological enquiries (one among many naturally) into the largely unwritten history of late Roman Macedonia, enquiries which it would be logical to conduct with frequent reference to the archaeology of the limes. Bibliography The Theodosian Code and Novels and the Sirmondian Constitutions. C. Pharr tr. (Princeton, 1952). Historia Malchi. B.Niebuhr ed. (Bonn, 1829). Honigmann E. 1939 Le Synekdémos de Hiéroklès et l’Opuscule géographique de Georges de Chypre. (Brussels). Notitia Dignitatum. Accedunt Notitia Urbis Constantinopolitanae et Latercula Prouinciarum. (O. Seeck (ed.) Frankfurt, 1962 rep.). 40 Stephanidou-Tiveriou 1990: 197-199. Unfortunately StephanidouTiveriou’s final report on the fortifications of Dion is unavailable to me at the time of writing.

710

Archibald Dunn: Was there a militarisation of the southern Balkans during late antiquity? Limeskongress. (Budapest): 277-296. Koukouli-Khrysanthaki Kh. 1972 Via Egnatia – Akontisma. Arkhaiologika Analekta Athênôn 5/3: 474-485. Koukouli-Khrysanthaki Kh. 1987-1990 Portraito tês Agrippinês tês Neôterês. Thrakikê Epetêrida 7: 81-102. Lampe G. 1981 A Patristic Greek Lexikon. (Oxford). Lander J. 1984 Roman stone fortifications: variations and change from the first century A.D. to the fourth. British Archaeological Reports Int.Ser. 206. (Oxford). Lemerle P. 1945 Philippes et la Macédoine orientale à l’époque chrétienne et byzantine. (Paris). Lilčić V. 1988-89 Docnežnoantičite tvrdini vo Tikveš i Vitačevo. Filozofski Fakultet na Universitetot. Skopje. Godišen Zbornik 41-42: 115-136. MacMullen R. 1984 The Roman emperor’s army costs. Latomus 43: 571-580. Marki E. 1996 Symperasmata apo tis anaskaphes tês 9ês Ephoreias Vyzantinôn Arkhaiotêtôn stê Voreia Pieria. Arkhaiologiko Ergo stê Makedonia kai Thrakê 10: 239258. Mikulčić I. 1974 Über die Grösse der spätantiken Städte in Makedonien. Živa Antika 24: 191–212. Mikulčić N. 1984 Problemot na Antigoneja/Antigonea am Axios. Filozofski Fakultet na Universitetot. Skopje. Godišen Zbornik 37: 111-138. Mikulčić I. 1986 Frühchristlicher Kirchenbau in der S.R.Makedonien. XXXIII. Corso di cultura sull’ arte ravennate e bizantina: 221-251. Mikulčić I. 1986 Spätantike Fortifikationen in der S.R.Makedonien. XXXIII. Corso di cultura sull’ arte ravennate e bizantina: 253-277. Mikulčić N. 1988-89 Topografija na Stenae – Prosek – Demir Kapija”. Filozofski Fakultet na Universitetot. Skopje. Godišen Zbornik 41-42: 65-87. Moutsopoulos N. 1979 ‘De via militari Romanorum’. Mutatio, mansio e castra nella parte tracese della via egnatia. Studi castellani in onore di Piero Gazzola. (Rome): 193-222. Moutsopoulos N. 1987 Monasteries outside the walls of Thessaloniki during the period of Slav raids. Cyrillomethodianum 11: 129-194. Moutsopoulos N. 1989 Anaskaphê Rentinas. Praktika tês Arkhaiologikês Etaireias: 199-204. Moutsopoulos N. 1990 Anaskaphê Rentinas. Praktika tês Arkhaiologikês Etaireias: 199-208. Moutsopoulos N. 1993 Ê thesê tês Mygdonikês Apollônias kai ê paralimnia (?) kharaxê tês Egnatias odou. Arkhaia Makedonia V/2: 999-1110. Papangelos I. 1998 Ê Khalkidikê kata tous mesous khronous. In Ê istoria tês Khalkidikês. (Thessaloniki). Papangelos I. in press Arkhaiologiko Ergo stê Makedonia kai Thrakê (February, 1999). Papazoglou F. 1988 Les villes de Macédoine à l’époque romaine. (Ecole française d’Athènes): 368-371. Papazôtos Th. 1988 Anaskaphê Dioklêtianoupoleôs. Oi prôtes ektimêseis. AD 43/A: 195-218. Pazaras Th. 1987 To ‘diateikhisma’ tês Kassandras. Praktika tou Prôtou Panellêniou Symposiou Istorias kai Arkhaiologias tês Khalkidikês. (Thessaloniki). Pazaras Th. 1996 Apo tên klasikê Vrea sto mesaiôniko kastro tês Vryas. Arkhaiologiko Ergo stê Makedonia kai Thrakê 10: 313-332. Peristeri K. 1990 Prôtê anaskaphikê ereuna stên “Akropolê” Platanias Dramas. Arkhaiologiko Ergo stê Makedonia kai Thrakê 4: 469-476. Polykhronidou-Loukopoulou L. 1989 Topeiros, polis tês Thrakês: provlêmata istorikês geôgraphias kai

topographias. In Kh. Bakirtzizs(ed.) Byzantine Thrace. Image and character. (Byzantine Forschungen XIV): 577-599. Popović V. 1984 Slaves et autochthones dans les provinces de Prévalitane et Nouvelle Épire. Villes et peuplement dans l’Illyricum protobyzantin. (Rome): 181-243. Poulter A. 1983 Town and country in Moesia Inferior. In A. Poulter (ed.) Ancient Bulgaria. (Nottingham): 74-118. Poulter A. 1995 Nicopolis ad Istrum: a Roman, late Roman, and early Byzantine city. Excavations 1985-1992. (London). Poulter A. 1998 Field survey at Louloudies: a new late Roman fortification in Pieria. Annual of the British School at Athens 93: 463-511. Roger J. 1938 L’enceinte basse de Philippes. Bulletin de Correspondance Hellénique 62: 20-41. Samsaris D. 1976 Istorikê geôgraphia tês anatolikês Makedonias kata tên arkhaiotêta. (Thessaloniki). Samsaris D. 1986 PENNANA. Enas rômaikos stathmos (mutatio) tês Egnatias odou. Dôdônê 15: 69-84. Schütz E. & Unger H. 1981 Wanderungen im Pangaion: ein Gebirge und sein Bergbau. (Ländshut). Sophocles E. 1887 Greek Lexikon of the Roman and Byzantine periods. (Harvard). Spieser J.-M. 1984 Thessalonique et ses monuments du IVe au VIe siècle. Contribution à l’étude d’une ville paléochrétienne. (Paris). Spieser J.-M. 1999 Les remparts de Thessalonique. À propos d’un livre récent. Byzantinoslavica 60: 557-574. Stephanidou-Tiveriou Th. 1990 Dion: ê okhyrôsê tês opsimês arkhaiotêtas. Arkhaiologiko Ergo stê Makedonia kai Thrakê 4: 195-204. Theokharidis G. 1980 Istoria tês Makedonias kata tous mesous khronous (285-1354). (Thessaloniki). Velenis G. 1988 Ta teikhê tês Thessalonikês. (Thessaloniki). Welch F. 1918-1919 Ancient sites in the Strymon Valley. Annual of the British School at Athens 23: 64-66.

711

Limes XVIII

Fig. 1. Late Roman Macedonia, with area of Fig. 3 at inset.

Fig. 2. Late Roman civitates of Macedonia.

Fig. 3. South-eastern Macedonia: Reported late Roman fortifications in pre-modern landscapes.

Fig. 4. The protection of the Via Egnatia in late Roman Macedonia.

712

Tilurium, Burnum and Bigeste. A new contribution to the dating of the Delmataen limes Mirjana Sanader The Delmataen limes is a system of Roman legionary camps and forts found among the mountain massive of the Dalmatian coastal hinterland. Both the name of this defensive line and its dating have been the subject of much debate. Based on new information gathered by recent archaeological research I shall, in this presentation, forward my thoughts on the limes and its date.

victorious in battle against the Delmatae. When the Roman province of Illyricum was established cannot be said for certain. According to Mommsen this occurred in the time of Sulla's rule. It is, however, evident that Caesar administered the provinces of Illyricum and Gaul. Illyricum was first ruled as a separate province when government of the area was given by Caesar to the Roman legate Publius Vatinius (45 to 44BC) whose seat was in Narona.

Rome began to show an interest in the region of the future province of Dalmatia once they became aware that the pirating activities of the Illyrians, inhabitants of what is today the Adriatic coast and islands of Croatia, were causing great damage to their shipping, and with that to their commerce. Following the successful wars waged on Carthage the Romans were interested in the areas of Greece and Macedonia, reachable by ship or by a land route along the eastern Adriatic shore. As Polybius reports (II.9.10), the Romans intervened in the Adriatic in 229BC. Roman naval advances against the Illyrians and their queen Teuta were exceptionally successful so that in this period they managed to establish some sort of control over the islands and possibly in the narrow coastal belt. This control, however, did not endure. Within 10 years the Romans engaged their ships against the Illyrians for a second time, in 219, during which Pharos was devastated (Livy XLV.26).

Octavian decided in 35BC to secure the northern border of Italy. In this period he turned his attention also to Illyricum aiming to secure the passages to the east in which he was engaged in difficult and exhausting wars in which he was himself wounded. In these advances he reached the River Sava up to, it appears, its mouth, as Cassius Dio reports that Roman soldiers reached the Danube. Siscia at this time became a Roman stronghold. In these battles Octavian captured the forts of Delmataen in the hinterland of Salona. In the years 12 to 11BC. Tiberius waged the 1st Pannonian War and crossed the River Sava making the new border the River Drava to the north and the Danube to the east. In this time, Illyricum, until then a senatorial province, was put under the rule of the emperor. It was most likely then that the province was divided into superior provincia Illyricum and inferior provincia Illyricum.

The Illyrian leaders, in an attempt to prevent Roman incursions in their territories entered into an alliance with Macedon. So it was that the fall of Macedon following the Battle of Pydna in 168 was also the demise of the Illyrian state at whose head stood King Gentius. By this event the area that Gentius controlled, the islands and a narrow coastal belt, came under Roman rule. As the Romans had emerged victorious in an earlier war with the Istrians from 179 to 178, they also occupied the Istrian peninsula. This means that following the victory against Gentius the land route to Macedonia was free.

All these military actions did not achieve the pacification of Illyricum. This occurred only after the 2nd Pannonian revolt that occurred between AD6 and 9. The opinions of experts are divided in this question; did upper Illyricum then or later become the province of Dalmatia, and lower Illyricum the province of Pannonia? What is certain is that, with the suppression of this bloody rising, peace was established in the area bringing economic prosperity. As is evident from this short historical overview of the Roman conquest of Dalmatia the Romans had considerable difficulty in achieving their military goals. It is certain that in these conquests military strongholds were of vital importance. It is, however, not known to us where the armies camped when not at war in the first period of the conquest of Dalmatia. If they built fortifications in this early period they are not preserved. There are no material records of the early period, and written records mention only Narona as a stronghold of the Roman commander G. Marcius Figulus in 156BC (Appian III: XI). This commander led his attacks on the Delmatae and their capital Delminium from Narona, from the flank therefore, using the easy passable valley of the River Trebižat and the easily crossed area in Ljubuški and Imotsko polje (Pašalić 1960: 55, 58; Zaninović 1966: 224).

The Romans however did not simultaneously advance into the interior with the result that the following several decades were marked by wars against the rebellious Illyrian tribes amongst whom the Delmatae are conspicuous. Numerous Roman advances into the Illyrian interior are recorded setting off the 1st Delmataen War in 156 under the leadership of Gaius Marcius Figulus to whom the city of Narona already served as a base. Cornelius Scipio Nasica destroyed in 155 the city Delminium, the chief stronghold of the Delmatae. Servius Fulvius Flaccus waged war successfully in 135 against the Ardiaei and the Pleraei. Wars of Lucius Caecilius Metellus against the Iapodes and his siege of Segestica are recorded in 119. Metellus wintered that year in the later renowned city of Salona, the first mention of the city. In 117 he was

713

Limes XVIII

The same source mentions L. Caecilius Metellus’s wintering in Salona in 119BC, Narona and Salona remained the chief Roman strongholds in the later Roman conquests.1 To the north was, as has been noted earlier in the text, Siscia.

To illustrate the range of the debate, and partly the conflicting positions in this context, I shall cite as an example Wilkes according to whom we are not dealing here with an independent limes, nor with the issue of whether this supposed limes was built before AD9 (Wilkes 1969: 91, 95). The author based this assertion on a few, in his opinion, undeniable facts. The first is that there are no epigraphic records that would document construction in this period. The second is that to Octavian, who was just coming into power, the Delmatae were unimportant; and the third that the limes is too short to impede the advance of Delmatae and other peoples across the Dinaric massif.

In this time the Roman army had only begun to build systematically permanent camps of which the oldest is, as it appears today, the one at Renieblas, constructed in the mid-C2nd BC (Schultern 1933; Johnson 1983: 246). In later decades when the Roman army becomes a professional one and when it resided far from Italy, the construction of permanent camps was an integral part of Roman military strategy. Such a strategy had undoubtedly the advantage that professional Roman soldiers residing in the camps were always prepared for movement and action. Besides this these camps were built at strategically important points that enabled constant control of conquered areas and over eventual unforeseeable movements of larger groups of people. It does not require emphasising that military encampments in certain provinces were systematically and thoughtfully distributed and planned. This enabled rapid communication between forces, including logistical and intelligence communication.

Following the presentation of these opinions in 1969, when they were found convincing, much archaeological research has been undertaken on the territory of the former Roman province of Dalmatia. I shall consider here only excavations at Burnum, Bigeste and Tilurium. Burnum The legionary camp at Burnum is located on the left bank of the River Krka (Titius), in an almost identical strategic position to Tilurium. Burnum too was originally an Illyrian settlement mentioned by Pliny (NH III.142). Zaninović (1996: 284) states that this was once an important Illyrian stronghold conquered by Octavian (35-33BC). Remnants of the camp, in the area of Šupljaja, are found today about 3kms east of the village of Ivoševci and about 14kms west of the city of Knin. The camp itself was located on a plateau rising sharply above the river. This position enabled excellent visibility and a complete overview of events in the surrounding area.

Between the legionary camps at Burnum and Tilurium the Romans raised several forts: Promona, Kadina glavica, Magnum and Andetrium. This line of defence ends with the fort at Bigeste on the River Trebižat, north-west of Narona. The legionary camps together with the forts form a line of defence and Patsch (1922: 57) was the first to note that they must have constituted a sort of Roman limes against the Delmatae. Experts on the subject generally accept his hypothesis on the limes (Zaninović 1996: 24 ; Bojanovski 1988; Šašel-Kos 1997: 284). Contrary to them, Wilkes considers them to be only a part of a greater defence system (Wilkes 1969: 92). As the existence of these military fortifications is undeniable and unquestionable, I am of the opinion that, if a name is to be given, there are no obstacles to calling this defensive line, consisting of legionary camps and regularly distributed forts the Delmataen limes as does Šašel-Kos (1997: 284).

From Burnum, as from other localities, numerous artefacts have been deposited in local museums, among which the epigraphic stand out. It is precisely with the aid of these epigraphic sources that an account of troop movements through the camp is possible. Experts are, presently, of the opinion that a short stay of legio XX in Burnum was followed by the arrival of the legio XI. Further, after legio XI left Burnum it was replaced by the legio IV Flavia which resided there until, at the latest, 86. After that the military significance of Burnum faded and it was used by smaller military units. A civil settlement developed in its proximity, and even attaining the rank of municipium.

Even more debate among experts has been provoked by the issue of dating the construction of the Delmataen limes. As the experts who have studied this issue have invested a great deal of energy in the study of ancient sources and epigraphic artefacts, so there has crystallized two hypothesises concerning the time of its construction. The first places the construction of the fortifications after Octavian's Illyrian War (35-33BC), while the second speaks of construction in the time following the suppression of Bato's revolt in AD9.

In contrast to the great interest in mobile artefacts from Burnum, the area of the former legionary camp has seen little and then unsystematic research. The first archaeological research was carried out in two summer campaigns in 1912 and 1913 (Reisch 1913: 122). Research was renewed in 1973 and 1974 again in two summer campaigns (Zabehlicky, Scheffenegger & Kandler 1979). The published analysis of the portable material, particularly coins and ceramics, demonstrate a time period spanning from the Republican era to the mid-C4th. As the research was, as the authors note, limited in area and time, only probing excavations were undertaken. Even though such research can hardly be considered representative for

1

Witness to Narona’s importance as a Roman military stronghold is, among other things, the well-known case in which Cicero begged the consul P. Sulpicius Rufus to find the escaped slave Dionysus, his librarian, who had made away with some important books. Flaccus answers Cicero who notes that Flaccus wrote him ex castris Narona (Cicero Ad fam. V.9).

714

Mirjana Sanader: Tilurium, Burnum and Bigeste. A new contribution to the dating of the Delmataen limes architecture the authors nevertheless concluded that Burnum had been built in two phases of which the earlier they place in the time of the emperor Claudius. Finds of Republican coins and numerous coins from the Augustan age, as well as finds of tiles with the stamps of the Pansiana workshop (which in the time of Augustus passed into the emperors’ possession so that bricks, from the rule of Tiberius onwards, also carried the name of the reigning emperor), have been interpreted by some authors as evidence of an older camp in the close proximity (Zabehlicky, Scheffenegger & Kandler 1979: 14).

The Cetina is a natural obstacle to routes to the east and north-east, that is, to the very heart of Dalmatia. The importance of this route was, of course, evident in the earliest times and it was clear also to the Romans that the Cetina was vital to the transport of goods and the passage of people. The Romans defended this strategically important communication by building the military camp at Tilurium. A linguistic analysis of the origin of the camps name points to an Illyrian origin, suggesting that the Illyrians, that is Delmatae, were aware of the importance of this exceptional position.2

Bigeste

The camp at Tilurium was raised by the Romans on the north-eastern part of the plateau that rises above the right bank of the River Cetina (Hyppus). The shape of the camp is an irregular rectangle within which is the larger part of the modern village of Gardun which is under the jurisdiction of the city of Trilj. For the moment we can assume that the area of the camp was some 20ha. At the locality of St. Peter’s church, to the east, the absolute height peaks at 430m. From this dominant and strategic place the River Cetina and its crossing can be viewed as well as all surrounding fields and plateaux. Of course from this position routes could be overseen, which was of importance because the road from Salona headed toward Tilurium, and here, on the River Cetina, mentioned in ancient itineraries as Pons Tiluri, it forks into two routes. One route went to the former Delminium, to the north-east, and the other towards Narona, to the south-east.

In the locality of the former fort at Bigeste archaeological research has been undertaken in three campaigns between 1977 and 1979 (Bojanovski 1985: 65–94). Although only a small portion of the fort’s area was explored at that time Bojanovski, who led the excavations, holds that the fort was raised as early as the period of intensive Roman conflict with the Delmatae, that is from 53 to 39BC. This was followed by the construction of the first walled fort during Octavian’s campaign against the Delmatae (35 to 33BC). Proof positive for this claim the author, along with numerous coins from the early period of Augustus’ rule, is to be found in several tiles with very early stamps of the Pansiana workshop. Who built these early forts is unknown; it has only been proven that the cohors III Alpinorum resided at Bigeste during the C1st. In the C2nd it was replaced by the cohors I Belgarum. In the later C2nd there are no further traces of military activity in the fort although civil life continued for some time. The author notes that in the River Trebižat valley there existed several smaller fortifications; at Vitina, Tihaljina and Posuški Gradac in whose locality there have been no excavations whereby the conclusion of the author is derived from preserved architectural and numismatic finds.

The area of the camp rises in steps to the north to a height of 430m, to the locality of Međine, while to the south towards the locality Podvornice the ground drops to 377m. To the west the area of the village of Gardun borders with the territory of the village of Vojnić. At the entrance to the Vojnić valley are preserved to this day the remnants of an Illyrian rampart called Prizida. These ramparts probably played a vital rôle in the defence of the passages to the Illyrian hillfort once located in the area of Tilurium.

Tilurium The legionary camp at Tilurium, located in the area of the modern village of Gardun, was placed in the hinterland of Salona – the chief and most powerful city of the province of Dalmatia, at a distance of some 30kms, the same distance as from the contemporary Croatian city of Split (Salona). It is worth noting that the entire, very elongated, coastal belt of the province is characteristic in its geomorphic structure. The narrow coastal belt is protected by a mountain chain that runs parallel to the coast. Access to the sea is possible via a large number of passes and rivers. Salona is located in a bay protected to the north by the hills of Kozjak and Mosor. The only pass to the north is through the narrow Kliš pass. Behind these hills is the fertile Sinj plain again defended by the heights of Svilaj and Moseć to the west and the almost impassable mountain massive of the Dinara and Kamešnica. The Sinj plain is rendered fertile by the River Cetina that pours into the sea some 30kms east of Salona, at the city of Omiš (Oneum).

Several papers have been written on the historical and archaeological question of the Roman military camp Tilurium, its artefacts, communications and topography (cf. Sanader 1998 for a review of published work). At a symposium held in Sinj in 1980 with the theme of The Cetina krajina from prehistory to the arrival of the Turks all existing knowledge about the Cetina krajina was compiled. At that meeting Marin Zaninović expounded the history of the establishment of Tilurium, from an Illyrian hill fort to a beneficium, summarised the distribution of military units in the camp through the centuries and established its certain military significance in ancient times. In his study on Tilurium, based on written sources 2

Pliny (NH III.142) mentions Tilurium, Burnum i Andetrium as nobilitata proeliis castella. Zaninović 1996: 272 equates Pliny use of the term castellum to a Illyrian town and thereby differentiating it from the military camp castrum.

715

Limes XVIII

and survey, Professor Zaninović gathered and analysed all available information on the camp based on ancient sources and epigraphic sources (Zaninović 1984: 65–77). He then expounded the opinion that the first legion to reside in Tilurium was the legio IX Hispana. It was followed by the legio VII. The time of its arrival is still a subject of debate, but it seems to have arrived in Tilurium at the latest by AD6. No later than 61, the legion left the camp after which it was occupied by smaller units until the mid-C3rd.

resided during the military operations of Caesar and Octavian, in temporary or wooden camps of which no remains exist. We hope that further research and discoveries in Tilurium may bring to light some clue to the resolution of this and other questions of which there may be discussion here. Bibliography Bojanovski I. 1985 Epigrafski i topografski nalazi s područja antičke Bigeste. In: 100 godina Muzeja na Humcu. (Ljubuški). Johnson A. 1983 Roman forts of the first and second centuries AD in Britain and the German provinces. (London). Pašalić E. 1960 Antička naselja i komunikacije u Bosni i Hercegovini. (Sarajevo). Patsch C. 1922 Historische Wanderung im Karst und an der AdriaI. In Die Herzegowina einst und jetzt. (Wien). Reisch E. 1913 Das Standlager in Burnum. Östrreichische Jahreshefte 16. (Wien). Sanader M. 1998 Tilurij – rimski vojni logor. Predhodno izviješće s arheoloških istraživanja u 1997 i 1998. Opuscula archaeologica 22. (Zagreb): 243-255. Sanader M. 2000 Tilurij – rimski vojni logor. Predhodno izviješće s arheoloških istraživanja u 1999. Obavijesti 1. XXXII/10. (Zagreb): 51-61. Schultern A. 1933 Geschichte von Numantia. (München). Šašel-Kos M. 1997 Dalmatiae. Dalmatia. Der Neue Pauly 3. (Stuttgart). Zabehlicky H., Scheffenegger S. & Kandler M. 1979 Das Standlager in Burnum I. Schriften der Balkankomission XIV. (Wien). Zaninović M. 1984 Vojni značaj Tilurija u antici. In Cetinska krajina od predhistorije do dolaska Turaka. (Split): 6577. Zaninović M. 1996 Burnum, Od Helena do Hrvata. (Zagreb). Wilkes J.J. 1969 Dalmatia. (London).

It is interesting, however, that until 1997 there had been no systematic archaeological research on the camp, so that archaeologists compiled their data exclusively on the basis of numerous artefacts sold to museums by the inhabitants of Gardun. Among such objects the most numerous are epigraphic artefacts, tombstones, votive tables and numerous medical instruments, weapons and coins. Systematic archaeological research on the camp following two years of preparations, began in 1997 (Sanader 1998; 2000). In these excavations architectural structures of the former camp were unearthed for the first time. Parts of the north-western walls, 3.2m wide, have been discovered, for example, with rare and interesting masonry which includes constructions of massive horizontally and vertically placed wooden beams. These were beams joined with an exceptionally strong mixture of mortar and crushed rock, which formed the in-fill of that part of the wall. The outer face of the rampart was formed of smoothly dressed stone blocks of smaller sizes. A well-preserved cistern and the remains of other constructions we uncovered are still undergoing research. Tilurium was for almost a century the target of collectors and antiques merchants during which time numerous smaller artefacts had been removed. In spite of this we have unearthed numerous ceramic, metal and glass fragments and Roman coins. As this portable material is still undergoing analysis the final results of the excavations have not been published. The larger number of the excavated Roman coins, however, were identifiable in the field. The coins chronologically span the later Republican period to the end of the C3rd. This short presentation of the results of research of Burnum, Bigeste and Tilurium, can say something about the time of their construction, or at least suggest answers to the dilemma. In this respect it is noteworthy that there are still no epigraphic or architectural proof that any of the localities was built prior to the beginning of the C1st. However, at all the sites there are finds, numismatic and ceramic, which pre-date that time. These are predominantly from the time of the late Republic but also from the very early principate. I would therefore, in ending, dare to assume that this defensive line of camps and forts was indeed built at the beginning of the C1st, that is with the arrival of permanent military garrisons. However, it appears that the hypothesis that these camps were raised in the same locations at which Roman troops 716

Mirjana Sanader: Tilurium, Burnum and Bigeste. A new contribution to the dating of the Delmataen limes

Fig. 1. The area of the Delmataen limes

717

Limes XVIII

Fig. 2. The Roman province of Dalmatia

718

Géographie antique (Ptolémée, Tabula Peutingeriana, Ravennatus) et stratégie impériale en Dacie Ioana Bogdan Cătăniciu aucun repère pour les rivières de la Valachie, l’Aluta3 étant le cours d’eau mentionné le plus à l’est, jusqu’à l’Hierasus. Pour Ptolémée il n’était pas clair que la province de Dacie, pour laquelle il disposait de renseignements concernant des routes récentes, ne recouvrait pas tout l’espace géographique où vivaient les Daces4; ainsi il a accordé des valeurs exagérées aux degrés de longitude, produisant une déformation accentuée, et unanimement reconnue, sur l’axe ouest-est. Suivant la théorie préconçue, toutes les localisations des villes situées dans l’est de la Dacie ont été cherchées en Moldavie centrale et du Nord, et les occidentales à l’ouest des Carpates Occidentales.

Pour commencer, il convient de souligner que le titre ne reflète que le résultat d’une certitude, à savoir que pour les Romains le territoire n’est que l’attribut d’une personne ou d’une communauté, définie juridiquement1 de sorte qu’il nous faut tenir compte en permanence du fait que toutes nos conclusions concernant les recherches sur le terrain, les traces archéologiques et leur disposition dans l’espace sont à expliquer par l’existence de cette conception romaine. La question est dans quelle mesure, étudiant les restes matériels, nous pouvons nous hasarder à comprendre les décisions (et leur contenu idéologique, juridique, etc…) qui ont produit le résultat que nous étudions, en tentant d’inclure aussi son évolution historique. Nous croyons qu’en évaluant les résultats des efforts réalisés par des générations d’archéologues, la réponse doit être positive. Ces dernières années, mes préoccupations se sont orientées vers l’étude systématique des références géographiques antiques concernant la Dacie, en vue de réaliser une corrélation entre les recherches modernes et l’interprétation traditionnelle2 de ces sources.

Ressentant le besoin d’équilibrer dans un espace inconnu les données sur les routes dont il disposait et en l’absence de coordonnées mesurées directement, Ptolémée s’est permis une distribution des localités que l’on a considéré chaotique. Les Romains définissaient l’espace par rapport aux routes (Purcell 1990: 12) qu’ils construisaient dès qu’ils occupaient une formation ethnique ou politique. Dans la disposition relative des localités inscrites sur la carte de la Dacie, on peut voir des chemins transmis par la Tabula Peutingeriana (= TP); cela étant, on peut mettre en évidence le réseau routier de la Dacie des deux dernières décennies après la conquête, en remarquant que Ptolémée a indiqué une série d’itinéraires5 que l’on retrouve en partie

Le principal acquis des investigations est la mise à l’écart de l’assertion que Ptolémée (= P) se rapporte à la géographie du Royaume Dace, et non à celle de la province de Dacie (Fig. 1: Bogdan Cătăniciu 1990). La compréhension des données de la Géographie de Ptolémée relative à la Dacie royale repose principalement sur le fait que la description géographique du chapitre III.8.1 englobe tout l’espace sur lequel s’est développé la civilisation dace, c’est à dire le nord de la Transylvanie, la Moldavie et aussi la Valachie; mais dans ce très large espace sont représentées les populations et les villes de la Dacie (P. III.8.3 et 4) et celles situées au nord du Danube, en Moldavie du sud, au nord de la Mésie Inférieure (P. III.10.7 et 8), ou en Moldavie centrale (P. III.7 - et III.10 et III.11 Tiragetii, Peucini, Carpiani, Bastarnii) et plus au nord (P. III. 9 - Costoboci transmontanoi).

3

Le Géographe de Ravenne IV.14.53-54 nous apprend même que le fluvius Flutassis finit patriam, transcription d’une lecture erronée pour Fl. Alutis, erreur qu’avait déjà faite Iordanes Getica 33, mais dans le texte, il est de toute évidence qu’il s’agit de l’Olt. 4 Schütte 1917: 77-90; l’espace géographique est défini d’après une source plus ancienne - le prototype Ac; les autres données appartiennent à deux autres étapes d’information – les prototypes Ad et Ae; nous sommes de l’avis qu’il n’existe qu’un seul prototype pour les itinéraires et les populations, car il n’y a aucune fracture chronologique entre les informations concernant le matériel de caractère dacique accentué et le matériel romain, précisément parce que l’on peut déduire de la Tabula Peutingeriana que les toponymes conservés par les Romains sont pour la plupart des cas autochtones. 5 Il s’agit du tracé Zeugma-Arcidava-Aizizis où manque la localité de Berzobis, connue des commentaires de Trajan, et du chemin Drobeta défilé Vulcan - Sarmizegetusa Regia, principale route d’accès et tracé le plus court entre le Danube et la capitale dacique, mais aussi vers le Pays de Hateg où a été fondé Ulpia Traiana Sarmizegetusa. Le tracé Dierna Tibiscum n’est que mentionné; il n’y avait qu’un seul chemin rudimentaire, si l’on tient compte du fait que par la Porte Domasnea il était, et il est encore, difficile de traverser, de sorte qu’il ne deviendra une route couramment utilisée que quand des troupes romaines se seront établies sur ce tracé, vers le milieu du IIe siècle. Une route qui apparaît nettement sur la carte de Ptolémée est celle de Novae, où durant les guerres stationnait la légion I Italica vers Cumidava (Râsnov) et plus loin les parallèlement à la frontière Est, au pied des Carpates Orientaux. Tel que nous avons essayé de le démontrer, le chemin Flămânda - le défilé de Bran a été utilisé par Trajan lors des combats contre les Daces, qui ont attaqué la Mésie Inférieure. Ces remarques soutiennent l’hypothèse que Trajan s’est efforcé d’assurer une circulation rapide et sans difficultés majeures le long des Carpates Orientales celes jusqu’à Porolissum.

Il n’en demeure pas moins que l’on peut formuler la conclusion que, pour les données comprises dans P. III.8.34 pour la province de Dacie, il n’avait en vue qu’une partie de la Transylvanie, et de l’Olténie, comme il ne signale 1

Mommsen 1887: 824-832; c’est la raison pour laquelle tous les actes que relatent les sources se réfèrent à civitas, gens, rex et leurs territoires. Les données géographiques concernant la Dacie n’ont plus été analysées systématiquement dans le passé récent. Mommsen, Kiepert et Goos et pour la zone sud-carpatique, Tocilescu et Pârvan se trouvent à la base de toutes les considérations modernes. En ce temps-là, les limites de la province de la Dacie n’étaient pas définies avec certitude et les étapes d’organisation du territoire dacique n’avaient pas été tirées au clair non plus. Leurs points de vue sont fondés sur l’état des connaissances du XIXe siècle et du début de celui qui vient de s’achever, et, sans aucune exception, ils partent de l’impression que les renseignements de Ptolémée, de la Tabula Peutingeriana et du Géographe de Ravenne sont incorrects, soit à cause de l’ignorance de leurs auteurs, soit en raison d’un manque d’attention et d’une encore plus grande ignorance des copistes.

2

719

Limes XVIII

seulement parmi ceux transmis par le Ravennatus (= GR: cf. Annexe 1).

recherche n’a pas joui d’un financement suffisant, de sorte que la multitude des données recueillies à partir de l’étude des cartes réalisées à grande échelle à la fin du XIXe siècle et les observations sur les photographies aériennes prises par les spécialistes de l’Institut de Photogrammétrie et de la Direction Topographique militaire n’ont pas pu être contrôlées et validées par des recherches sur le terrain. Pourtant les résultats de l’étude méritent d’être présentés justement parce que j’ai revu à travers l’oeil de celui qui a assimilé la somme de découvertes effectuées dans le courant de ce siècle, les données des trois sources géographiques et parce que celles-ci constituent une provocation pour les recherches à venir.

On constate dès le début que Ptolémée a retenu les données les plus importantes concernant les routes de la Dacie et cette circonstance, croyons-nous, est une confirmation du fait qu’il avait obtenu les données les plus récentes et les plus complètes se rapportant à la province trajane. Le fait que ces routes ne se retrouvent pas en totalité dans les deux autres sources analysées peut être dû aussi aux changements qui se sont produit dans le développement de certains tracés, en fonction des besoins de la vie de la société6, mais sans doute aussi du fait que la Tabula peinte tout comme la Cosmographie de Ravenne reflètent à une époque tardive les routes tracées dans un espace qui n’était plus une province romaine et présentent un choix de tracés en fonction des critères qui se trouvent à la base de la sélection réalisée par ceux qui ont rédigé les deux sources géographiques tardives.

Les données géographiques conservées - étant partie intégrante de la conception romaine du monde (Strabon Geog. I.1.16; Purcell 1990: 8-9,12; Wolff 1993: 179) reflètent des attitudes et matérialisent les décisions du pouvoir de Rome dans des cas concrets. La multitude de routes que l’on distingue sur la carte de Ptolémée pour l’époque trajane et le début du règne d’Hadrien indique non seulement l’existence de liens avec le sud, par Viminacium, en franchissant un pont en amont des Portes de Fer, à Zeugma = Pons, un autre par Drobeta - le défilé Vulcan - Sarmizegetusa et, par la suite, vers le Nord jusqu’à Porolissum, mais aussi un réseau de routes orientées ouest-est.

Parmi toutes les localités mentionnées par Ptolémée et par les routes, seules sont localisées avec certitude en vertu des matériaux épigraphiques Porolissum, Napoca, Potaissa, Apulum, Germisara, Aquae, Tibiscum, Sarmizegetusa, Dierna, Drobeta, Romula et Cumidava. Parmi les stations localisées avec certitude, en dehors des établissements civils7, la coexistence d’établissements civils avec un fort et des canabae ou des vici militari a été constatée archéologiquement à Drobeta, Tibiscum, Apulum, Porolissum, probablement aussi à Potaissa (cf. Fig. 2). Il a été proposé des situations pour d’autres localités aussi, mais sans autre raisons que leur établissement relatif dans un espace habité par les Daces.

Le tracé marqué par Ziridava - Singidava par Apulum à Praetoria Augusta et Sangiava est facilement identifiable avec la route qui remonte le cours du Mures et qui a certainement été utilisée au cours des guerres de Trajan (Fig. 3). Moins facile à reconnaître est le tracé qui, venant de l’ouest, par Porolissum, débouche vers le nord-est de la province; la distance trop grande entre les établissements de Rucconium et Dacidava, par rapport à Porolissum et entre cette localité et Arcobadara-Triphulum, doit être corrigée tout comme celle entre les localités situées sur la ligne du Mures et sur la route Drobeta-Sornum, en se guidant sur leurs positions relatives par rapport à la longitude de Gigen (colonie Oescus) et Svistov (Novae). En ce qui concerne les deux villes de Rucconium et de Dacidava, il faut aussi en considérer l’orientation fausse dans l’espace, parce qu’elles sont à une latitude plus au nord que Porolissum, ce qui les situerait en dehors de l’espace de la province.

Les données de Ptolémée acquièrent un sens si’en partant de localisations certaines, on met à profit les indications milliaires signalées dans la Tabula Peutingeriana et si l’on étudie les traces sur le terrain. Toutes les études concernant les identifications des stations figurant sur l’itinéraire peint sont parties de l’assomption que celles-ci sont de caractère militairecastra, castella ou mansiones sur les artères publiques. Un premier doute concernant ce point de vue m’a été suggéré par la reconsidération des identifications proposées pour le tronçon de route entre Romula et Caput Stenarum lors des investigations que j’ai entreprises sur la ligne fortifiée le long de l’Olt. A partir de ce moment, j’ai proposé une étude aérophotogrammétrique des chemins romains, en vue de contrôler l’exactitude des données de la TP.8 La

Afin que ces propositions soient acceptables, il faut considérer toutes les distorsions qui sautent initialement aux yeux sur la carte de Ptolémée, par rapport aux localisations certaines: - Ptolémée ne sait pas que les Monts Carpates font un arc et rejoignent le Danube, même si les villes décrivent cet arc.

6

Ptolémée ne connaissait pas le chemin de la vallée de l’Olt, bien qu’il nous indique l’existence de la population qui avait pour centre Buridava. 7 Les établissements civils sont: Napoca, Germisara, Aquae, Dierna et Romula 8 Surtout que Baradez 1959: 19-24, qui est l’auteur de cette étude pour l’Afrique du Nord, a conclu de façon explicite à l’exactitude des données concernées par cet itinéraire; plus récemment, des recherches dues à Morizot 1991: 341-346 ont confirmé à égale mesure la correction de ces données. Les objectifs de notre étude ont été rendus publics dans le

Programme pour une recherche systématique du réseau routier de la Dacie (Tibiscum, 1986: 355-358); le programme a été accepté comme thème de recherche, mais il n’a joui que de moyens extrêmement modestes.

720

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie - par conséquence, il ignore l’existence des montagnes méridionales. Sarmizegetusa est placée trop au sud et devrait se trouver en Olténie; néanmoins, puisqu’il manque des localités entre les latitudes de 450 20` et de 460, il apparaît que les itinéraires ont imposé un espace qui suggère la présence de ces montagnes, mais par erreur, Ptolémée l’a situé au nord de Sarmizegetusa et d’Aquae.

pontons qui permet la liaison avec Viminacium; la route s’oriente vers le nord jusqu’à la ‘ville’ de Zurobara (Fig. 4). La TP indique cette route avec des stations espacées regulièrement de XII mp (=17,760kms), jusqu’à Aizizis, d’où, avec une station intermédiaire (Caput Bubali), il n’y a plus que XIII mp jusqu’à Tibiscum; c’est pourquoi nous pouvons affirmer que Ptolémée indique le premier des chemins construits sous la direction de Balbus en Dacie, dans l’ouest du Banat10; toutes les stations même si elles ont des noms à consonnance autochtone, sont des étapes de chemin, édifiées à intervalles réguliers. Le même rhytme de XII mp est conservé entre les stations se trouvant entre Apulum et Potaissa11, ce qui nous porte à croire que sur ce tronçon aussi, la Tabula signale toutes les stations construites initialement sur la route impériale.

- les distances entre les villes figurant sur la carte accompagnant la géographie de Ptolémée ne respectent aucun rapport proportionnel avec les indications milliaires connues sur la Tabula Peutingeriana (Fig. 1). - Ulpianum qui ne peut être que metallum Ulpianum9 (il n’y a aucun municipe de Trajan en Dacie) dans les Carpates occidentales devait être situé sur un itinéraire en corrélation avec Potaissa et Napoca, sur la latitude desquelles celui-ci s’inscrit .

Toutes les stations ne sont pas habituellement inscrites sur les routes, mais seulement une sélection selon certains critères. On peut observer qu’en dehors des grandes villes les points de croisement12 des voies sont marqués sur la TP: Tibiscum en est un, même si celui qui a peint l’itinéraire n’a pas réussi à ramener au même point les tracés des routes qui, à travers le Banat, rejoignaient Tibiscum, comme cela a été réussi dans le cas du grand centre d’Apulum, où se réunissent les chemins traversant la Dacie et celui qui longe l’Olt. La source de Ptolémée, et sans doute aussi l’auteur de la première rédaction de la Tabula, ignoraient le fait que les deux routes traversant le Banat convergeaient vers Tibiscum, car celui-ci n’est signalé qu’une seule fois au bout du chemin (sans stations) qui se dirige vers Dierna, et une seconde fois comme étape au début du chemin vers Apulum, par Aquae et Germisara, Ulpia Traiana Sarmizegetusa étant exclue.

- la première mention de Tibiscum se trouve beaucoup trop au sud, mais en rapport relativement correct à la direction de Dierna; Sarmizegetusa Regia est cependant située de façon erronée par rapport à Tibiscum (seconde mention, transcription incorrecte - Tiriscon), se trouvant à l’ouest de celle-ci. Les coordonnées sont celles d’Ulpia Traiana Sarmizegetusa et il semble qu’elles ont été établies en fonction de Drobeta, par rapport à laquelle elles se trouvent en relation correcte; nous supposons que l’absence de Sarmizegetusa Ulpia Traiana à son emplacement naturel entre Tibiscum et Aquae se trouve être la conséquence de l’élimination faite par Ptolémée, qui a cru à un doublet avec celle notée sur la route de Drobeta, avec l’épithète de regia, précisément parce que pour lui ou pour sa source il s’agissait d’une même localité – l’ancienne capitale royale, refaite.

Le Géographe de Ravenne indique à Arcidava un chemin qui relie le Banat à Potula13 et à Bacaucis14, et à Tibiscum il mentionne une bifurcation vers Acmonia15; en outre il indentifie Cedonie en tant que carrefour avec un lien direct du chemin longeant l’Olt avec Sarmizegetusa, sans passer par Apulum au nord (Fig. 5).

L’extension vers l’est de la carte de la Dacie entière est – comme on l’a déjà dit - la conséquence du besoin de ne pas laisser vide l’espace hors de l`Empire, là où il n’y avait pas de routes.

Les données embrouillées du Ravennat concernant les chemins du Banat et de l’Olténie peuvent être éclaircies par leur interprétation comme représentation des villes de la carte de Ptolémée: il s’agit de la route Drobeta-

La déformation de la carte de Dacie vers l’est est la même que celle pour les localités d’Oescus et de Novae. Par rapport à ces deux repères on peut ajuster les coordonnées des stations distribuées vers l’est: le chemin DrobetaSornum doit être restreint au seul espace de l’Olténie; les coordonnées de la rivière Olt = Alutus se déversant dans le Danube indiquent la bonne distance d’Oescus, mais pas la bonne position, le confluent avec le Danube étant indiqué de façon erronée à l’ouest de la ville.

10 Les identifications proposées ne nous semblent pas entièrement satisfaisantes. Nous croyons que Apo flumen est un point où la route coupe cette rivière et ne peut être le nom d’une rivière que suit la route 11 Le milliaire d’Aiton (CIL III.1627) placé en l’an 108 sur la route Napoca-Potaissa, construite par la cohors I Fl. Ulp. Hispanorum se trouve à X mp de Potaissa 12 Remarque que nous devons à Baradez 1959, qui l’a obtenue par l’étude des photo aériennes du le nord de l’Afrique. 13 Le Ravennat confirme de cette manière les données dans la quatrième série de trois populations, Potulatenses qui doivent être localisées en Olténie du nord-ouest 14 Par Cannonia, nom certainement corrompu, mais dont il est malaisé de dire de Ad Pannonios, qui apparaît encore une fois comme Panonion (GR IV.53-54) ou est la Με̃τατηεςέ pour Acmonium que l’on connaît dans la description de Ptolémée. 15 Avec la graphie AGMONIA, facile à confondre avec ACMONIA.

La première route attestée comme route d’accès de l’armée à la tête de laquelle se trouvait Trajan (Priscian VI.13 p.205) est indiquée par Ptolémée sans la mention de Berzobis; Zeugma = Pons indique l’existence d’un pont à 9

Gostar 1969: 2, 173 a lancé l’hypothèse que Ulpianum est ou bien un metallum ou un aurarium Ulpium, mais considérant la position sur la carte ptolémaïque, il préfère croire qu’il s’agit d`un castellum situé dans les monts Meses.

721

Limes XVIII Acmonia-Sarmizegetusa.16 L‘Acmonia de Ptolémée apparaît de façon explicite comme également reliée à Tibiscum (GR IV.14.5354).

Everghita-Corzu21 - Grădistea-Cotofenii din Dos Cerboaia (où existe le toponyme Cetatea, sur le Jiu) Craiova-Popânzălesti - le long de la rivière TesluiDobroslăveni-Resca22 (Fig. 6). Notre proposition repose sur le fait que le chemin a été repéré sur le terrain jusqu`à Everghita, mais de là les recherches se sont dirigées vers le Motru, pour découvrir Amutria.23 Si nous ne considérons pas nécessaire que l’antique Amutrium soit sur la rive du Motru, mais seulement en un point d’où l’on se dirigeait vers cette rivière, alors la voie que nous avons décrite, jonchée de toponymes révélant l’existence de traces antiques et débouchant dans la vallé du Teslui à RescaRomula, est tout à fait plausible; mais les localisations pour les trois stations doivent être effectuées au terme d’une étude jointe des photographies aériennes et du terrain.

Un chemin orienté ouest-est (Fig. 4) est répété aussi plus bas, avec comme stations Arcidaba-Potula-Bacaucis. Il s’agit de trouver quel toponyme a été transmis de façon erronée par le GR: Pannonios sur le tracé Dierna, Tibiscum ou de nouveau Acmonia; de toute façon l’existence de la route est certainement validée par les données de la géographie de Ptolémée, où apparaît Acmonia, tandis que le vicus Potula est le centre d’une civitas Potulatenses, dans le quatrième groupe de populations, celui qui correspond aux villes ayant la latitude 45°. La route est celle qui, le long de la vallée de la Cerna, à travers les monts Mehediti, passait dans les sous-Carpates de Gorj17, là où il faut chercher Acmonia, tandis qu’à l’est on devrait repérer un établissement à caractère civil, Potula et, plus à l’est, Bacaucis. Il s’agit de la zone de l’Olténie la moins connue du point de vue archéologique, où sont signalés très peu d’établissements romains.18

Sur l’itinéraire peint les localités mentionnées par Ptolémée ne sont pas répétées, peut-être parce que les sites n’ont pas évolué au-delà de l’époque trajane, ou bien parce qu’ il s’agit de points intermédiaires entre ceux marqués sur la Tabula à des distances si grandes que l’on peut admettre l’existence d’autre stations24 qui auraient été éliminées pour diverses raisons lors de la rédaction réalisée aux IV-Ve siècles. La TP mentionne Castra Nova, toponyme qui renvoie à l’existence de fortifications à un moment de l’histoire de la province.

Pour en revenir aux routes visibles sur la carte de Ptolémée, un autre tracé de direction ouest-est part du Danube, traverse l’Olténie et est marqué comme d’autres stations sur la TP, n’ayant en commun que Amutrium, à XXXVI mp de Drobeta. Cette localité a été recherchée sur la rivière Motru19, considérée comme la variante moderne de Mutrium.20 Les recherches effectuées sur les cartes modernes à grande échelle nous ont amené à proposer pour son tracé une variante descendant de Turnu Severin par

La Tabula Peutingeriana continue néanmoins son chemin à travers Romula, au-delà des montagnes à ApulumPorolissum. La route qui longe l’Olt, de Romula à Caput Stenarum, a fait l’objet de recherches systématiques, dirigées par Gr. Tocilescu.

16 Chemin que l’on connaît grâce aux découvertes archéologiques, marqué de fortifications à partir de Turnul Severin (Drobeta), Cătunele, Vârtu, Bumbesti-Vârtop à travers le défilé Vâlcan vers Sarmizegetusa regia, avec la mention que la route devait traverser l’endroit où avait été crée Sarmizegetusa Ulpia Traiana, car il est peu probable qu’il s’agisse du chemin remontant la montagne, suivi, tel qu’on l’a établi, par la cavalerie de Lusius Quietus lors des guerres contre les Daces (en témoignent les camps de marche de Jigoru, Comărnicel, Vârful lui Petru- présenté par Stefan 1997: 517-525). On doit tenir compte du fait que cet itinéraire a été suivi par les Romains pendant la première guerre de Trajan uniquement grâce aux efforts des exploratores et bien sûr par une trahison des Daces voisins, probablement des habitant de Drobeta, qui avait un statut privilégié déjà au commencement du IIe siècle ap. J.C.: cf. comme dit explicitement Strabon, la géographie avait un but politique - Wheeler 1999. 17 Du point de vue du relief, la zone de montagne ne permet la circulation que sur le tracé Dalboset- Lăpusnitel- Mehadia- valea Cernei - Obârsia Closani- Baia de Aramă. Ce chemin part, tel que le dit le Ravennat, de la route reliant le Danube, par Arcidava à Tibiscum, voie peu définie par recherches sur le terrain (ce qui s’explique par le fait que ces dernières 50 années il a été zone frontalière, difficilement accessible). Compte tenu de l’aspect géographique de la zone et des données de la Tabula peinte, on peut proposer pour Apo Flumen l’identification avec la rivière Nera; la route qui traversait le Danube venant de Viminacium par Ram monterait vers le nord, traverserait la Nera dans la zone Zlatita, où se trouve un établissement romain qui n’a pas encore été étudié. Les identifications acceptées de nos jours proposent la direction nord-ouest longeant le Caras. Arcidava ne peut pas être identique au castellum (récemment on y a découvert un deuxième castellum) de Vărădia, beaucoup plus au nord que ne l’indique les XXIV mp, soit 35.5kms du Danube. 18 TIR L-34 (1968) la carte. 19 On a accepté la localisation proposée par Tocilescu à Butoiesti voyez Tudor - TIR L-34: s.v. Admutrium 20 Inconnu dans les sources antiques.

Les recherches ont été reprises ces dernières décennies, mais toujours en vertu de l’hypothèse que la TP représente exclusivement les fortifications militaires le long de cette voie; là où les distances ne sont pas conformes aux coordonnées des fortifications connues, les données de la TP ont été décrétées fausses.25 Compte tenu de cette 21

Marele Dictionar Geografic al României 1900. II s.v. Corzu - à 51kms de Turnu Severin il y a un fossé, que les paysans appellent Valul lui Traian et qui ressemble à un chemin romain; l’endroit nommé Grădistea et la colline Jidovilor (des Juifs) toponymes caractéristiques pour des ruines antiques, se trouvent à quelques centaines de mètres vers l’est: Je n’ai pas fait de recherches sur le terrain. 22 La distance entre Turnu Severin et Resca (sur la route actuelle il n’y a que 170kms, sur la variante proposée par nous la distance est plus grande de 20kms environ) est plus courte que ne l’indique la Tabula – CLX mp; Goos a supposé que le chemin descend jusqu’à Celei, pour remonter vers Romula; l’hypothèse ne saurait être soutenue, parce que, bien que la route Sucidava-Romula ait un role majeur dans le traffic en Dacie (Constantin construisant ici un nouveau pont traversant le Danube), le vicus Sucidava n’apparaît dans aucune de nos sources géographiques. Il est probable que la distance entre Castra Nova et Romula de LXX mp est une transcription erronée. Tudor 1969: 57-60 considère qu’il ne s’agit que de XX mp. Nous croyons plutôt qu`un des bras uni du X s’est effacé et que le copiste tardif l’a interprété commé un L 23 Recherches réalisées par l’ingénieur P.Polonic en 1894, ses observations ont été utilisées par Tudor, loc. cit. 24 Entre Drobeta et Amutrium s’accommoderaient deux stations à XII mp, éventuellement même Arcina et Frateria évoquées par Ptolémée. 25 Par exemple, la distance de XLIIII mp entre Pons Vetus et Caput

722

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie perspective, les choses n’ont pas pu évoluer. Il nous semble édifiant dans la série des abus dans l’interprétation des sources d’avoir accepté la localisation de Praetorium à Racovita-Copăceni dans le défilé de l’Olt, sans tenir compte du fait que les deux castella ont été édifiées à la fin de l’époque d’Hadrien, et que le nom de la localité fait référence à la présence du gouverneur Laberius Maximus, sinon de l’empereur Trajan.

les Romains aux populations daciques habitant la province a tenu compte de ce stade de développement étatique, motif pour lequel l’organisation administrative de la province est fondée sur des communautés territoriales autonomes - civitates - dont le nom dérive généralement de celui de la localité centrale (Galsterer 1982: 76; Bogdan Cătăniciu 1991). Les 15 “populations” définies par Ptolémée sont plus nombreuses que celles mentionnées pour les provinces voisines, circonstance que nous pouvons comprendre comme preuve d’une relative richesse démographique, d’autant plus que, à l’identique des autres provinces, il faut prendre en compte les territoria des villes de droit romain de province (pour le début du IIe siècle J.C., Ulpia Traiana Sarmizegetusa, Drobeta et Napoca).

Il va de soi que les distances indiquées sur la Tabula ne correspondent pas à cette identification; elles ont alors été ajustées. Nous avons pris des mesures à partir d’ApulumAlba Iulia et il nous faut signaler que la distance inscrite sur la TP est beaucoup plus importante que la distance réelle26 jusqu’ à Romula-Resca. En prenant des mesures à partir d’Alba Iulia, Caput Stenarum est localisé à Tălmaciu. La distance de là à travers le défilé de l’Olt correspond à l’indication de XLIV mp signalée par l’auteur de la Tabula peinte jusqu’ à Pons Vetus, que l’on doit alors localiser à Jiblea, où l’a fixé Gr. Tocilescu.27 A cet endroit, il est possible qu’il y ait eu un pont ancien qui ait donné son nom à la localité, pour passer sur la voie praticable à l’époque des guerres daciques. Déduites des mesures de Romula, les localisations d’ Arcidava et de Rusidava doivent être envisagées avec une certaine réserve: la première n’est pas le castellum d’Enosesti, mais l’un des établissements situés au sud de celui-ci; et Rusidava n’est pas un castellum parce que celui-ci n’a pas été identifié dans la zone de Drăgăsani, où sont signalées des traces romaines.28

On doit mentionner le fait que Ptolémée a exclu de sa carte (en tant que doublets pour les populations?) le nom des établissements qui ont constitué le centre des civitates29 peut-être parce que celles-ci étaient des localités sises sur des chemins secondaires ou pas encore construits à la manière des Romains; le fait que Potula doit être localisée d’après les informations de Ravennatus sur une dérivation des chemins du Banat vers l’est plaide pour cette dernière explication. Quant à Buridava, la localité se trouvait sur un chemin dont le tracé n’était pas connu, n’étant pas encore inclus parmi les routes praticables, et qui n’a été finalisé qu’à l’époque d’Hadrien. La route qu’on peut reconnaître dans l’est de la Dacie, avec des localités connues aussi grâce aux sources épigraphiques (Cumidava et Ramidava), est en rapport avec la longitude de Novae et Cumidava (localisable dans la zone de Râsnov), et doit être identifiée avec la route préhistorique, utilisée par les Romains comme tracé de la ligne des fortifications transalutanes. Faute de correspondance entre les routes sélectionnées par l’auteur de la TP, il est plutôt hasardeux de proposer des localisations que nous serions tentés de choisir parmi les castella30 sur cette ligne fortifiée. Aucune des localités transcrites par le Ravennat pour la voie sur laquelle on circulait de Tyras, à travers l’est de la Dacie, à Porolissum, n’apparaît introuvable chez Ptolémée, ce qui porte à croire qu’il s’agissait d’un autre chemin ou bien qu’avaient été choisies comme étapes du tracé des localités qui ont évolué par la suite sans lien avec celles créés immédiatement après la conquête.

De Stolniceni-Buridava jusqu’à Sâmbotin, la distance est celle rapportée par la TP et il est fort probable que la Castra Traiana se trouvait à cet endroit et que le chemin pouvait continuer sur le tracé à l’est des monts Cozia, où, comme l’a aussi considéré Goos, doivent être cherchées Arutela et Praetorium, la distance de XXXIII mp = 48kms jusqu’à Greblesti-Câineni étant approximative. Nous considérons que le doublage de la route n’a pas été très clair pour l’auteur de la Tabula et que c’est pourquoi le dessin ne suggère pas la situation réelle (Fig. 7). Les Romains considéraient le monde barbare comme étant composé de peuples et de localités. La Dacie conquise est la descendante d’un État de facture originale, un hybride entre la forme de l’État hellénistique et celle des unions tribales celtes; nous estimons que le traitement accordé par

Il est intéressant de constater que le niveau des connaissances de Ptolémée est beaucoup plus important en ce qui concerne le nombre des localités portant des noms

Stenarum a été corrigée et est de XIIII mp seulement; Tudor 1969: 57, 59. 26 La distance qui sépare Alba Iulia de Resca est, de nos jours, de 275kms, tandis que sur la Tabula la distance est de CCIV mp, c’est-à-dire 301.5kms. La voie romaine telle que l’a également répéré C. Goos, entre Sibiu et Alba Iulia, avait un tracé plus direct et plus court à l’est , à partir de Alba Iulia au-delà de Mures-Hăpria-Berghin-Miercurea-Apoldul de Jos- Ruscior- Sibiu-Vestem. Cette voie a été identifiée et est marquée par des établissements civils romains - voir TIR L-34, où sont proposées aussi des localisations pour Sacidava (Dostat) et Cedonia (proche de Sibiu), pour lesquelles nous proposons Ruscior. 27 Fouilles et recherches archéologiques en Roumanie (1900) carte. 28 Il est probable qu’Arcidava est plutôt l’établissement civil de Brâncovenesti, et à Drăgăsani il n’y a aucune trace d’une fortification romaine.

29 Il semble que Rucconium serait vicus Anartorum et elle est transcrite parmi les localités du nord-ouest de la Dacie. Notre hypothèse (dans l’article cité) pour compléter la lacune figurant sur le milliaire d’Almasul mare - CIL III 8060 avec Rucconium. 30 On sait que sur cet alignement de fortifications marquant le chemin le plus aisément accessible en direction de la Transylvanie il n’est accompagné ni d’établissements civils ni de canabae; en outre, entre l’Olt et cette ligne les établissements civils sont rares et n’apparaissent qu’après le milieu du IIe siècle - Bogdan Cătăniciu 1997: 129ff.

723

Limes XVIII

daciques que les autres; la situation nous permet de supposer que, peu après la conquête31, sous Balbus, ont été construites de nombreuses routes32 qui reliaient les établissements les plus importants de la province, dont on a gardé les noms daciques. Les établissements identifiés et étudiés (Apulum, Dierna, Drobeta, Napoca, etc) ont été transposés de leur foyer dacique d’origine dans le voisinage. On a aussi l’exemple de la Cumidava dacique, qui est probablement la forteresse de Râsnov, dont le nom a été transmis non seulement comme celui d’un point de passage sur la route utilisée par Ptolémée pour l’est de la Dacie, mais aussi comme un épithète de la COH(ors) VI NO(va) CUMIDAVENSIS, qui a stationné au début du IIIe siècle dans le castellum33 édifié par Trajan à Râsnov dans Tara Bârsei. Il est sûr que Ramidava a été un vicus appartenant à la zone voisine de Cumidava, où a été recruté C. Valerius Herculanus stator alae Aravacorum.34 Les deux ‘villes’ de la carte de Ptolémée (Cumidava et Ramidava) n’apparaissent plus dans l’est de la Dacie chez le Ravennat, se trouvant peut-être au sud de la route venant de Tyras vers la Transylvanie.35

Ptolémée, nous estimons avoir suffisamment de raisons pour souligner leur caractère civil, et non militaire, la majorité étant sans doute des établissements et non pas des castella. C’est la raison pour laquelle, croyons-nous, il ne signale aucune légion dans la province, bien que tout au moins la legio XIII Gemina se trouvât dans le camp couronnant la colline Cetatea (Cité) d’Alba Iulia-Apulum. Pour ce qui est de la route traversant l’ouest du Banat qu’avait suivie Trajan en l’an 101, où se trouve le camp de la legio IIII Flavia Felix à Jidovin, de nos jours généralement asssimilée à Berzovia, Bersobis36 n’y est pas inclus; cette absence est peut-être explicable par le fait que la fortification aussi avait été abandonnée lors du transfert de la légion en Mésie Supérieure. La Tabula Peutingeriana a été rédigée après un long intervalle de temps après l’abandon de la province de Dacie. Même si l’on a constaté sous Constantin que des fortifications ont été refaites et même si l’on a accepté la présence permanente du pouvoir impérial au nord du Danube, du moins dans le secteur sud-carpatique, il est difficile d’accepter pour le monde scientifique que lorsque la Tabula évoque une série de chemins au nord du Danube, ceux-ci sont autre chose qu’une préférence savante pour l’antiquité; le problème est le même pour ce qui touche la Cosmographie ravennate. Les deux sources sont estimées correctes seulement pour la période d’existence de la province (jusqu’à 271 ap.Chr.)

La carte de Ptolémée représente le premier niveau de développement des routes de la province et on peut en déduire quelques principes fondamentaux: 1. l’accomplissement de la liaison avec le sud du Danube se fait par trois points, à savoir: Viminacium-Zeugma, EgetaDrobeta et Novae-Tiasum, auxquels on peut ajouter éventuellement Taliatis-Dierna: 2. le développement à l’intérieur du territoire provincial d’un axe de circulation rapide entre les établissements les plus importants et, en plus, un chemin suivant le tracé ouest-est le long du Mures et un autre à travers l’Olténie. 3. le chemin périmétral qui cerne le territoire inclu dans les provinces de Dacie, après la réforme administrative d’Hadrien, du Danube vers le Nord, puis vers l’ouest, nous semble être caractéristique.

La Tabula n’indique que des chemins tracés à travers l’ouest et le centre de la Dacie intracarpatique, vers Porolissum. Ptolémée, ainsi que le Ravennat, enregistrent des points qu’on peut situer dans l’est de la Dacie. Il est remarquable que le Ravennat ne transmette aucun des établissements situés à l’extérieur de la Dacie, présentés chez Ptolémée au chapitre référant au voisinage de la province, ou bien à la description géographique (III.6.15 prope Daciam; III.10.14), et que la Tabula ne présente aucune des routes traversant les zones qui avaient été abandonnées par Hadrien. En ce qui concerne l’absence dans la Tabula de tracés des routes pour la partie est de la Dacie, celle-ci a été considérée comme preuve de l’abandon de l’est de la Dacie par Gallien et l’on a essayé de dater la source de la Tabula précisément entre Gallien et Aurélien.37

Pour ce qui est du caractère des routes utilisées par 31

Bien que dans Ptolémée abondent les noms daciques, nous estimons que l’on peut soutenir que les toponymes sont des stations sur des routes romaines identifiées par leur rapport entre elles, en précisant que celui-ci n’est qu’une orientation, puisqu’on ne peut pas en établir avec certitude la distance ou la position exacte. 32 Balbus ad Celsum p. 92, Les routes sont celles que Trajan a employé pour pénétrer en Dacie. 33 Jusqu’à présent, on n’a pas découvert à proximité du castellum un établissement civil pour l’époque romaine, mais nous estimons pouvoir supposer qu’il existait justement cette référence du IIIe siècle au nom que l’on connaît sur la carte de Ptolémée. 34 Florescu 1951: 125ff. inscription découverte à Gârliciu en Mésie Inférieure: DM [Her]CULANUS VETeranus EXTATORE praefe]CTI ALAE II ARAVACOR[um]---X NATUS VICO RAMID----ANNisXXVI Caia VALERIA C[on]iux ---- ANNis LX NATA LOCO E[odem] ----AB EO IN CONUBIO ---S ET C VALerius VALENTIN[us; Il va de soi que l’auteur fixe Ramidava en Valachie suivant l’identification proposée par Macrea 1941-42: 25 avec le castellum de Drajna de Jos et par conséquence le recrutement s’est produit à l’epoque où ce territoire a été incorporé dans la Mésie Inférieure à l’époque trajane. 35 Angustia, laquelle est localisée unanimement à Bretcu, étant identifiée au défilé d’Oituz n’a pas été citée non plus par le Ravennat; les coordonnées de Ptolémée semblent la fixer plus au nord, et s’il s’agit de Bretcu, nous croyons qu’elle n’aurait pu manquer sur la route de Tyras à Porolissum.

Comme la Tabula Peutingeriana a été transcrite vers la fin du IVe siècle (Weber 1976: 20-23), elle ne nous offre pas une image complète des voies de la Dacie. L’apparition des routes au nord du Danube est une exception38; l’intérêt sélectif pour les chemins de cette région à une date tardive doit être considéré comme une démonstration relative des liens de l’Empire avec cette partie de l’Europe. La route 36 C’est une preuve réelle que Ptolémée a utilisé des itinéraires et non pas les commentaires de Trajan à propos des guerres contre les Daces. 37 Horedt 1974; 1977 use d’arguments archéologiques qui contredisent “la perte militaire à l’est de la Dacie”, parce que là ont été mises en évidence des fortifications tardives. 38 Seule la route allant de Vindonissa à Sumelocenis est hors des frontières, mais c’est la zone d’intérêt et d’action face aux Alamans.

724

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie Flămânda - le défilé de Bran - Transylvanie transformée par Hadrien en chemin de rocade pour la frontière, n’a plus été incluse parmi les routes utilisées couramment; pourtant l’utilisation du chemin est prouvée par des découvertes monétaires à Jidava-Câmpulung et par le trésor monétaire du IVesiècle de Săpata de Jos.39

défaite de l’an 332, et une installation substantielle dans la partie est de la Dacie n’est enregistrée qu’à la fin du IVe siècle.43 Du point de vue archéologique, on constate le fonctionnement de la vie urbaine dans l’ancienne province avec les mêmes paramètres durant quelques décennies, et ce n’est qu’au IVe siècle que se fait ressentir une dégradation de la vie et qu’apparaissent des changements pertinents dans la culture matérielle.44

Le Géographe de Ravenne ne transmet plus les toponymes jalonnant la route entre le Danube et le défilé de Bran, bien qu’à l’époque tardive où ont été accumulées ses informations on a considéré nécessaire de noter une route reliant Tyras à Porolissum, sans doute la route entre l’un des plus importants ports du nord de la Mer Noire et le nord de la Transylvanie. L’inclusion de la localité de Certie, qui figure sur la TP, suggère le lien entre ce chemin et celui constituant la voie principale vers le sud du Danube, même à une époque tardive.

Conformément à la conception romaine, le territoire pouvait se trouver sous l’hégémonie du Pouvoir de Rome uniquement si des relations juridiques avec la population se maintenaient.45 Dans le cas de l’abandon de la Dacie, il faut considérer que du point de vue archéologique (comme l’affirment aussi les sources46), on ne saurait soutenir l’existence d’un désastre dans la province à la suite d’attaques barbares. Il faut également prendre en considération le fait que la province était constituée, tout comme l’Empire entier, par la somme de communautés autonomes. Comme partout, l’administration romaine était formée des honestiores locaux et d’un nombre restreint de personnages qui représentaient le pouvoir central (Wolff 2000). Il faut considérer que les structures locales de l’administration romaine n’ont pas disparues avec la décision d’Aurélien de retirer les légions et l’administration. Il ne faut pas oublier que c’était un principe fondamental de la pratique romaine que de recourir à des réglementations juridiques avec chacune des communautés autonomes, établies par la lex provinciae et modifiées aussi par des lois impériales, et celles-ci semblent perdurer, sauf si elles étaient annulées par l’Empire.

Les différences qui peuvent être signalées entre les trois principales sources pour la géographie de la Dacie sont dues, à notre avis, aux périodes de temps où elles ont été réalisées. Le principe fondamental de la stratégie romaine de contrôle des mouvements des peuples se trouvant au-delà des fines imperii était déjà en usage à l’époque d’Auguste, qui a perfectionné le système républicain des relations diplomatiques avec ces populations, système qui a été maintenu aussi à l’époque romaine tardive (Schulz 1993). Or, il est une idée absolument aberrante, même si l’on interprète généralement de cette façon la situation après Aurélien, qui est que la retraite ait été totale, que le sens pratique des Romains ait disparu brusquement et que, ayant besoin des troupes existant au nord du Danube, ils aient abandonné un vaste territoire, le sacrifiant au bon plaisir des barbares, qui pouvaient s’organiser et y créer une base d’attaque.40 Essayant d’éliminer cette aberration, E. Chrysos41 propose l’hypothèse qu’ Aurélien aurait cédé la Dacie aux Goths, en leur offrant le titre de foederati, sans tenir compte du fait qu’à la date de l’abandon de la province, les Goths se trouvaient dans la zone est de la Dacie, que ceux-ci avaient déjà eu des foedera avec l’Empire, qu’ils pillaient et, en coalition avec les Carpes, attaquaient les provinces au sud du Danube.42 Archéologiquement parlant, sur le territoire de la Dacie, on ne peut pas soutenir une pénétration même sporadique des Goths avant les premières décennies du IVe siècle, après la

Gallien aussi bien qu’Aurélien – les deux empereurs à qui l’abandon de la Dacie a été attribué – ont beaucoup apporté à la réforme et au redressement de l’Empire; en conséquence, on doit regarder le retrait des légions et de l’administration centrale de la Dacie comme une des mesures prises pour renforcer l’Empire. On peut considérer qu’ils ont eu recours au principe stratégique de renforcer les frontières par économie de forces propres, insuffisantes dans des conditions où les frontières sont menacées d’un bout à l’autre. Il est dans la mentalité juridique et politique

43

Horedt 1982: 190-194; Harhoiu 1990: 169-208; 1999, aussi les contributions de Kokolowski et Bierbrauer 1999 concernant la chronologie et l’interprétation ethnique, voir les cartes p.191, fig. 7 et p.213, fig. 1, et plus précisément p.229 – “Die Gründe, die zur Erweiterung der gothischen Siedelgebietes bzw zur gotischen Landnahme westlich der Prut führten, sind unbekannt. Die Räumung der Prowinz Dacia war es sicher nicht” 44 Nous avons constaté les mêmes aspects de la culture matérielle à Tropaeum Traiani-Adamclisi, où l’aspect spécifique de la céramique change seulement à partir de la seconde partie du IVe siècle; uniquement du point de vue stratigraphique et par rapport aux autres matériaux datables, on peut considérer que la céramique date du IVe siècle; Barnea, Barnea & Bogdan Cătăniciu 1979: 226, fig. 156-157. 45 Schulz 1993: 18 les relations diplomatiques sont établies avec populus, civitas ou rex. 46 SHA Aurel 39.7 - cum vastatum Illyricum ac Moesiam deperditam videret, provinciam Transdanuvianam Daciam a Traiano constitutam sublato exercitu et provincialibus reliquit disperans eam posse retineri.

39 Les découvertes monétaires de Jidava s’échelonnent jusqu’ au VIe siècle et le trésor de Săpata de Jos contient des monnaies de TrajanValentinien; Mitrea 1960: 591. 40 Whittaker 1989: 87-88 “à aucun moment, dans aucune source, on ne reconnut officiellement que les frontières du Haut-Empire avaient diminué, alors que même la Dacie au nord du Danube et les Champs Décumates à l’est du Rhin avaient été perdus dans les troubles du III-e siècle.” 41 Chrysos 1992 propose un acte d’abandon (E. Menzel) d’Aurélien à l’égard de la Dacie, car son retrait a été “auf der Grundlage des Kaiserlichen Verzichts auf Ausübung der Gebietshoheit vollzogen”. Et en effet, le renoncement doit avoir été définitif, car Orosius affirme qu’il a été in perpetuum. 42 Petrus Patricius 8 (Niebur 124-126); Iordanes Getica 89.

725

Limes XVIII romaine que l’autorité47 sur une communauté ou sur un État ne soit pas conditionnée par la présence d’une force militaire, étant une réalité définie par des rapports juridiques; l’Empire n’a jamais interrompu les rapports juridiques avec une communauté avec le retrait des légions du territoire de cette communauté, au contraire il a eu la sagesse de la transformer en filtre contre des actions menaçantes.

de l’Olténie52 par Constantin et la construction du pont entre Oescus et Sucidava, ainsi que la réfection de la route vers Romula, constatées archéologiquement, mais sans que l’on sache à la suite de quelles actions, nous semblent pouvoir être expliqué uniquement par la capacité réelle de l’Empire, rétabli à l’intérieur, d’agir sur le territoire au nord du Danube. L’insuffisance des sources pour les régions septentrionales de l’Empire et le fait qu’elles sont dans la plupart des cas sans caractère officiel nous oblige à trouver des arguments d’ordre archéologique pour définir les aires d’habitation et de domination des diverses gentes barbares (Elton 1996: 574).

Il est à supposer que les Romains ont maintenu certaines clauses pour les communautés provinciales et ont défini les rapports en leur accordant le droit de se défendre et de se constituer en une force fidèle à Rome; il semble que cet objectif a eu du succès, se concrétisant en une force que nous ne pouvons connaître par les sources qui nous sont parvenues, toutes étant tardives et non officielles; du point de vue archéologique, on peut pourtant constater qu’il n’y a pas eu de destructions d’établissements fouillés, ni de castella de Dacie.48 On constate la présence de céramique typique des Daces libres particulièrement dans les derniers niveaux des castella où ceux-ci peuvent être considérés comme individus ou groupes.49 Les recherches que nous avons effectuées à Urluieni suggèrent que ce castellum fonctionnait même après Aurélien.50

Le fait que les éléments gothiques sont inconnus à l’ouest de la ligne des fortifications transalutanes est la preuve que l’entrée sur le territoire de l’ancienne province leur a été interdite. En Transylvanie aussi la vie de la population romanisée dans la plupart des centres urbains a été mise en évidence, même si les investigations ne se sont pas déroulées à un rythme et avec les moyens nécessaires pour comprendre à fond le phénomène.53 Ce n’est que dans la deuxième moitié du IVe siècle qu’apparaissent des éléments de la culture Sântana de Mures et c’est de là qu’on peut supposer que les Goths sont recepti, en vertu d’une alliance en tant que foederati. Il est difficile d’imaginer dans quelle mesure l’administration des Goths s’était substituée à celle de la Daco-Romania, mais on peut faire l’analogie avec l’établissement des Goths au sud du Danube.54

C´est uniquement si l’on considère que l’autorité romaine s’étendant sur une communauté (quelle que soit son organisation politique ou administrative) n’était pas liée à la présence ou à l’absence de la force militaire, étant une autorité de iure, que l’on peut comprendre que même la retraite militaire d’un territoire n’était pas considérée par le Pouvoir de Rome comme une interruption des rapports, le renoncement à ses droits sur la population ou bien au statut concerné. On doit s’attendre à ce que, dans une ancienne province incluant des communautés urbaines, une autorité centrale ne soit pas instituée à la place du pouvoir romain, qui ne disparaît pas de iure.

Dans les territoires nord-danubiens, la diplomatie romaine face aux barbares immigrés reposait sur l’affirmation permanente des droits de l’Empire. Les actions de Valens décrites par Ammien Marcellin (XXVII.5-6) sont typiques de ces relations. Les Goths sont, explicitement, des foederati en vertu du principe général de contrôle par des relations diplomatiques des populations externae et par l’admission de forces barbares ‘conquérantes’, dans les territoires nord-danubiens. Rome n’a jamais cessé d’assurer un certain contrôle, qui, en fonction de sa situation et de la force des barbares, était plus ou moins étendu (Whittaker 1992).

Ainsi les rapports de chaque communauté avec Rome se perpétuent, aucune raison sérieuse de les annuler n’existant; la formule permettait la continuité de l’autorité romaine in perpetuum; l’absence d’une puissance centrale a rendu ce système en fait peu solide. Nous adopterons la dénomination employée par Demougeot (1985) de ‘Daco-Romania’ pour définir l’espace où s’est conservé l’hégémonie romaine après la retraite des légions.51 La reconquête longuement discutée

Le développement du système routier en Dacie jusqu’à la fin de la domination romaine est saisisable précisément dans les références enchevêtrées de l’Anonyme de

47 On devrait se rapporter uniquement à Ammien Marcellin pour souligner les actions romaines dans les territoires abandonnés, au-delà du Rhin, ou à ceux de Constantin au nord du Danube, afin de trouver des arguments contre leur abandon à perpétuité. 48 Protase 1987 avec la bibliographie du problème. 49 Opreanu 1993: 235-260 a fait une étude traitant de la céramique dacique trouvée dans les castella - voir aussi sa thèse Dacia romană si Barbaricum (Timişoara, 1998: 116sqq.) comme toujours les arguments des hypothèses qu’il nous offre doivent être attentivement jugés. 50 Bogdan Cătăniciu 1994: 327-353. 51 Ruscu 1998: 235-254 reprend la discussion à propos de toutes les contributions plus ou moins inspirées relatives aux sources concernant l’abandon de la Dacie.

52 Les recherches archéologiques ont constaté une présence romaine effective à Drobeta, Sucidava et une population romaine christianisée apparaît aussi à Slăveni, mais probablement aussi dans le nord de l’ancienne Dacia Malvensis, où n’ont pas été effectuées d’explorations sur le terrain, mais où il existe une continuité de la circulation monétaire en bronze jusqu’au VIesiècle, étant donné que la monnaie en argent et en or n’entrait dans le monde barbare qu’en tant que subsides, dons ou butin de guerre; Codex Iust. IV.63.2 interdit vers la fin du IVe siècle l’utilisation de l’or dans le commerce avec les populations vivant en dehors de l’Empire. 53 Horedt 1977; 1982: 11, 59-70; à propos de l’état des investigations dans les villes romaines, voir notre étude 1993: 125-135. 54 Themistios discours XVI.210.

726

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie Ravenne. Il est évident que la localisation sur le terrain des anciennes provinces ne lui est pas claire, même si l’on peut supposer qu’Acmonia faisait partie, durant la période trajane, avec l’est de l’Olténie et la Valachie, de la Mésie Inférieure. Bien qu’à la date où a été rédigée la Cartographie ravennate il est probable que seuls les toponymes aient été conservés et que nombre de localités étaient déjà en ruines, la préoccupation de l’auteur de reproduire la majeure partie des routes de l’ancienne province avec toutes les principales voies secondaires peut indiquer que l’on désirait s’informer sur cette région. L’apparition de la route entre Tyras et Porolissum est - pour la date de la rédaction du document - un indice qu’après la chute du limes scythique on pouvait maintenir des rapports avec les zones nord de la Dacie par la mer. Compte tenu des données géographiques dont nous disposons, nous croyons pouvoir soutenir que la stratégie romaine en Dacie a connu plusieurs étapes: 1. A l’époque du Principat, les Romains s’appuyaient sur une excellente modalité pour définir l’espace à l’aide des routes et ils l’applicaient dès qu’ils conquéraient un territoire en vue de le transformer en une province viable militairement et économiquement, sous l’effet d’un phénomène d’‘acculturation’. 2. Lors de la période de crise du IIIe siècle, un programme d’ ‘abandon’ de la Dacie fut établi, dès Gallien, en parallèle avec les autres réformes destinées à sauver l’Empire, avant tout parce que sa défense n’était plus rentable avec un sud déstabilisé, mais tout en assurant le maintien de l’hégémonie sur les communautés de l’ancienne province. 3. A l’époque postérieure à la tétrarchie, le contrôle du territoire de l’ancienne province55, du moins intentionnellement, est une issue majeure de la diplomatie romaine tardive, jusqu’à la fin du VIe siècle - “Justinien envoya une mission auprès de ces barbares, leur demandant de s’installer dans une ancienne fortification du nom de Turris, qui se trouve au-delà de l’Istros et qui avait été édifiée par l’empereur Trajan, mais elle avait été abandonnée depuis longtemps parce que les barbares des lieux l’avait ravagée. Justinien avait promis qu’ il leur offrirait la forteresse avec le territoire qui l’entourait, parce qu’ elle appartenait aux Romains dès le début”.

55 Procope A propos des guerres VII.7.14.32; Fontes Historiae DacoRomaniae II. 1970: 441-442. Le géographe de Ravenne rassemble des données sur ces territoires, mais en présentant aussi les chemins venant de l’est qui faisaient la jonction à la mer avec Constantinople.

727

Limes XVIII

Annexe 1 Tableau des villes de la Dacie enregistrées dans les trois sources géographiques analysées Première route Tabula Peuting

MP

Ptolemeu

Viminacium

10

Viminacium

lederata

12

Ravennatus

Route collatérale

Zeugma (pons) apo flumen

12

Arcidava

12

Centum putea

12

Berzovia

12

Azizis

3

caput bubali

10

Arcidava

Arcidaba*

-canonia-potulabacaucis

Berzovia Aezisis

Zizis gubali

Zurobara Tibiscum 1

Tibiscum

Tibiscum Tema

Seconde route Tabula Peuting

MP

Ptolemeu

Ravennatus

Taliatis

20

Taliastis

DIERNA

11

DIERNA

ad Mediam

14

Medilas

Praetorium

9

Pretorich

ad Pannonios

9

Panonin

Gaganis

11

Gazanam

Masclianis

14

Masclunis

TIBISCUM

14

Agnaviae

8

pons Augusti

15

SARMIZEGEUSA

14

[Sarmizegetusa]

SARMIZEGETUSA

AD AQUAS

13

AQUAE

Aquas

GERMISARA

Germisara

Petris

9

GERMISARA

9

Blandiana

8

TIBISCUM

Route collatérale

DROBETA [DIERNA]

Tibis

- agmonia

Augmonia Augusti

Petris Blandiana

-cedonia

Burticum APULUM*

12

APULUM

APULUM

-ROMULA

Marcodava Brucla

12

Salinis

12

SALINAE

Brutia Salinis

POTAISSA

24

POTAISSA

POTAISSA

=macedonica

NAPOCA

16

NAPOCA

NAPOCA*

-ulpianum

Optatiana

15

largiana

17

Certiae

4

Certie

728

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie

Troisième route Tabula Peuting

MP

***

Ptolemeu

Egeta

21

Egeta

DROBETA

36

DROBETA

Ravennatus

Route collatérale

Arcina Frateria Amutrium

35

Amutrium Pirum Sorum

Pelendava

20

Castra Nova

70

ROMULA

13

30 ROMULA

Acidava

24

Rusidava

14

pons Aluti

13

Burridava

12

Castra Traiana

9

Arutela

15

Praetorium

9

pons Vetus

44

Vetera

Stenarum

12

caput Stenarum

16

Aluti

Cedonie

24

Cedonia

Acidava

15

Sacidava

APULUM

APULUM

APULUM

*** correction proposée Quatrième route Tabula Peuting

MP

Ptolemeu

Ravennatus

Route collaterale

DROBETA Acmonia

Potula-Bacaucisa

[Sarmizegetusa] SARMIZEGETUSA regia

Cinquième route Tabula Peuting

MP

Ptolemeu

Ravennatus

Partiscum Ziridava Singidava APULUM* Praetoria Augusta Sangidava

729

Route collatéral

Limes XVIII

Sixième route Tabula Peuting

MP

Ptolemeu

Ravennatus

Route collatérale

NOVAE Tiasum Netindava Zusidava Ramidava Comidava Polonda Angustia Utidava Petrodava Patridava Carsidava Triphulum Arcobadara POROLISSUM* Docidava Rucconium

Septième route Tabula Peuting

MP

Ptolemeu

Ravennatus

Route collatérale

TYRAS Tirepsum Iscina Capora Alincum Ermenium Urgum Sturium Congri Porolissum* Certie

Huitième route Tabula Peuting

MP

Ptolemeu

Ravennatus Acmonia (Canonia)

Route collatérale Arcidava

Potula Bacaucis

Mures-Cernjachov-Kultur. In G. Gomolka-Fuchs (ed.) Die Santana de Cernjachov-Kultur. (Bonn): 211-233. Bogdan Cătăniciu I. 1990 Ptolémée et la province de Dacie. Dacia NS. XXXIV: 223ff. Bogdan Cătăniciu I. 1991 À propos de civitates en Dacie. Ephemeris Napocensis I: 59-67. Bogdan Cătăniciu I. 1993 Bemerkungen zur Römischen Stadt in der Rumänischen Forschung. Historia Urbana 1.2: 125-135. Bogdan Cătăniciu I. 1994 Castella de la Urluieni. Studii si

Bibliographie Baradez J. 1959 Réseau routier de commandement, d’administration et d’exploitation de la zone arrière des limes de Numidie. In R. Laur-Belart (ed.) LimesStudien, Vorträge d. 3 Internationalen Limeskongress in Rheinfelden-Basel. (Basel): 19-30. Barnea A., Barnea I. & Bogdan Cătăniciu I. 1979 Tropaeum Triani, Cetatea I. (Bucureşti). Bierbauer V. 1999 Die ethnische Interpretation der Santana de

730

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie Cercetari de Istorie Veche si Arheologie 45/5: 327-353. Bogdan Cătăniciu I. 1997 Muntenia in sistemul defensiv al imperiului Roman sec. 1-III p. Chr. (= Wallachia in the defensive system of the Roman Empire 1st – 3rd centuries A.D.). (Alexandria). Bossio L. 1983 Tabula Peutingeriana, una descrizione pittorica del mondo antici. (Rimini). Braund D. 1996 River frontiers in the environmental psychology of the Roman world. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 43-47. Brun P. 1993 Genèse d’une frontière d’ Empire, La frontière nord de l’ Empire romain. In P.Brun, S. van der Leeuwen & C. Whittaker (edd.) Frontières d’empire, Nature et significations des frontières romaines. (Nemours): 21-31. Chirilătilde E. & Gudea N. 1982 Economie, Populatie si societate în Dacia intracarpatică în primul secol după abandonarea provinciei (275-380 e.n). Acta Musei Porolissensis 6: 123-140. Chrysos E. 1992 Von der Räumung der Dacia Traiana zur Entstehung der Gothia. Bonner Jahrbücher 192: 175193. Demougeot E. 1985 Constantin et la Dacie. In E. Frézouls (ed.) Crise et redressement dans les provinces européeenes de l’Empire Romain. Actes du Colloque de Strasbourg (décembre 1981): 94ff. Dilke A.O.W. 1985 Greek and Roman maps. (London). Dillemann L. 1975 La carte routière de la Cosmographie de Ravenne. Bonner Jahrbücher 75: 165-170. Dyson S. 1985 The creation of the Roman frontier. (Princeton). Elton H. 1996 Romans and Goths: recent approaches. Journal of Roman Archaeology 9: 566-574. Florescu G. 1951 Un nou document epigrafic referitor la teritoriul de la nordul Dunarii Moesice. Studii si Cercetari de Istoire Veche si Arheologie 2/2: 125-135. Galsterer H. 1982 Stadt und Territorium. In Stadt und Herrschaft. Historische Zeitschrift 7. (München). Gostar N. 1969 Ulpianum. Analele Stiintifice ale Universitătii A.I.Cuza, Iasi XV.2: 173ff. Harhoiu R 1990 Chronologische Fragen der Völkerwanderungszeit in Rumenien. Dacia NS XXIV: 169-208. Harhoiu R. 1997 Die Frühe Völkerwanderungszeit in Rumänien. Archaeologia Romanica I. Harhoiu R. 1999 Das Ende der Santana de Mures-Cernjachov Kultur und die Phase der Hunnischen Expansion. In G. Gomolka-Fuchs (ed.) Die Santana de MuresCernjachov-Kultur. (Bonn): 59-68. Horedt K. 1974 Dacia amissa. Studii si Cercetari de Istoire Veche si Arheologie 25/4: 551-558. Horedt K. 1977 Die Siebenbürgische Limesstrecke Dakiens. In D. Haupt & H.G. Horn (edd.) Studien zu den Militärgrenzen Roms II. Vorträge des 10 Internationalen Limeskongress in der Germania Inferior. (Bonn): 331-338. Horedt K. 1982 Die Städtischen Siedlungen Siebenbürgens in spätrömischer Zeit. Sargetia XIV (Deva, 1979). Isaac B. 1996 Eusebius and the geography of Roman provinces. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 156-166. Janni P. 1984 La mapa e il periplo. Cartografia antica e spazio odologico. (Roma). Kokolowski A. & Bierbrauer 1999 Vorschlag zur relativen Chronologie der südöstlichen Kulturen des ‘Gotenkreises’, Die Forsuchngsergebnisse der

Maăomăcz-Gruppe. In G. Gomolka-Fuchs (ed.) Die Santana de Mures-Cernjachov-Kultur. (Bonn): 197ff. Lee A.D. 1993 Information and frontiers: Roman foreign relations in late antiquity. (Cambridge). Litterature gréco-romaine et géographie historique, Caesarodnum, IX bis. Mélanges offerts a Roger Dion, R. Chevalier ed, (Paris, 1974). Macrea M. 1941-42 Anuarul Institutului de Studii Clasice IV: 25ff. Mann J.C. & Breeze D. 1987 Ptolemy, Tacitus and the tribes of north Britain. Proceedings of the Society of Antiquaries of Scotland 117: 85-91. Manning W.H. 1995 Ptolemy and the pre-Flavian military sites of Britain. In W. Groenman-van Waateringe, B.L. Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies. Proceedings of the XVth International Congress of Roman Frontier Studies. (Oxford): 33-41. Marele Dictionar Geografic al României 1898-1902 ed. G.I Lahovary. Mitrea B. 1960 Découvertes monnétaires en Roumanie. Dacia NS IV: 591ff. Mommsen T. 1887 Römisches Staatsrecht III. (Leipzig). Morizot P. 1991 Les stations de la Tabula Peutinger entre Lambases et Ad Calceum Herculis. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. (Exeter): 337-345. Nicolet C. 1991 Space, geography and politics in the early Roman empire. (Ann Arbor). Opreanu C. 1993 Elemente ale culturii materiale dacice tarzii. Ephemeris Napocensis III: 235-258. Opreanu C. 1998 Dacia romană si Barbaricum. (Timisoara). Parker S.T. 1997 Geography and strategy. In W. Groenmanvan Waateringe, B.L. Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies. Proceedings of the XVth International Congress of Roman Frontier Studies. (Oxford): 115-122. Potter D. 1996 Emperors, their borders and their neighbours: the scope of imperial mandata. In D.L. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 49-66. Protase D. 1987 Die Dakisch-römische Bevölkerung nördlich der Donau in der Periode von Aurelian bis zu den Slawen (7Jhd) im Lichte der Aktuellen Dokumente, Die Völker Süd-osteuropas in 6 bis 8 Jhd. Süd-osteuropa Jahrbuch 17. (Berlin): 231-249. Purcell N. 1980 Maps, lists, money and power. Journal of Roman Studies 70: 178-182. Purcell N. 1990 The creation of provincial landscape: the Roman impact on Cisalpine Gaul. In T. Blagg & M. Millett (edd.) The early Roman Empire in the west. (Oxford): 7-29. Ruscu D. 1998 L’abandon de la Dacie romaine dans les sources littéraires. Acta Musei Napocensis 35,1: 235254. Schulz R. 1993 Die Entwicklung des römischen Völkerrechts im Vierten und Fünften Jahrhundert n. Chr. Einzelschriften Hermes 61. (Stuttgart). Schütte G. 1917 Ptolemy’s maps of northern Europe: a reconstruction of the prototypes.. (Copenhagen): 77-90. Sherk R.K. 1974 Roman geographical exploration and military maps. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt I.1. (Berlin): 534-560. Stefan A.S. 1997 Les guerres daciques de Trajan: les opérations du front alpin. In W. Groenman-van Waateringe, B.L. Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies. Proceedings of the XVIth International Congress of Roman Frontier

731

Limes XVIII

Studies. (Oxford): 517-525. Stevenson E. 1931 Geography of Claudius Ptolemy, based upon Greek and Latin manuscripts and important late fifteenth and early sixteenth century printed editions. (New York) - Carte de la Dacie, ms Ebner cca 1460. Tudor D. 1968 Oltenia Romana. (Bucureşti). Tudor D. 1969 Tabula Imperii Romani, L-34. (Bucareşti). Weber E. 1976 Tabula Peutingeriana Codex Vindobonensis 324. (Wien). Wheeler E. 1999 sv. Strategie. In Mensch und Landschaft in der Antike, Lexikon der Historischen Geographie. (Stuttgart). Whittaker C.R. 1989 Les frontières de l’Empire Romain. (Paris). Whittaker C.R. 1992 What happens when frontiers come to an end? In P. Brun, S. van der Leeuwen & C. Whittaker (edd.) Frontières d’empire, Nature et significations des frontières romaines. (Nemours): 133-141. Whittaker C.R. 1996 Where are the frontiers now? In D. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 25-41. Wirth G. 1967 Zur Frage der foederierten Staaten in der späteren röm Kaiserzeit. Historia 16: 234ff. Wolff G. 1993 Roman peace. In J.Rich & G. Shipley (edd.) War and society in the Roman world. (NewYork): 171-193. Wolff H. 2000 Administrative Einheiten in den Nordprovinzen und ihre Beziehungen zu römischen Funktionsträger. In W. Eck (ed). Locale Autonomie und römische Ordnungsmacht in der Kaiserzeitlichen Provinzen von 1 bis 3 Jahrhundert. (Oldenburg): 47-60.

732

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie

Fig. 1. Carte de la Dacie après les coordonnées de Ptolémée; nous avons introduit la distance en mp entre les localités qu’ on connaît de la Tabula Peutingeriana

733

Fig. 2. Carte de la Dacie romaine.

Limes XVIII

734

Ioana Bogdan Cătăniciu: Géographie antique et stratégie impériale en Dacie

Fig. 3. Territoires daciques au temps de Trajan.

Fig. 4. Carte du sud-ouest de la Dacie, sont tracées les voies qui peuvent être les voies enregistrées par les sources géographiques

Fig. 5. Le probable carrefour de Cedonie, les routes romaines probables

735

Limes XVIII

Fig. 6. La route débouchant à Romula.

Fig. 8. La carte avec les découvertes de la Culture Sântana-Černjakov, après G.Fuchs-Gomolka 1999: 213.

Fig. 7. La carte de la zone de Buridava avec le carrefour de Jiblea.

Fig. 9. Schéma avec les voies qu’on peut déchiffrer dans les itineraires romains.

736

Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia Cristian Gazdac Ancient authors have different opinions about the moment when the province of Dacia was abandoned by the Roman army and administration. Eutropius (IX.8), Rufus Festus (VIII) and Iordanes (Romana 217) speak about a double withdrawal of Roman administration from Dacia during the reigns of Gallienus and Aurelian; Aurelius Victor (2223) and Orosius (VII.22.7) consider that Dacia was lost in the reign of Gallienus; finally, the Historia Augusta affirms that this province was lost by the Romans only under the reign of Aurelian (SHA Vita Aureliani 39: see also Iliescu 1971: 425).

The first aspect revealed by the graphics of the finds/period made for the auxiliary forts is the cessation of the coinfinds for the issues later than the reign of Philip I (see Figs. 2-7). For some of the auxiliary forts the coin-finds stop even earlier, eg. Războieni - Cumidava, Arutela, but the quantity of coins recovered is too small to allow any certain conclusions (see Tab. 3-5). Also for the coins issued in the later periods than reign of Philip I, except the fort of Drobeta, there is no certain information that they were found on the site of the auxiliary forts and not in the civilian settlement, beside the fort.

The purpose of this paper is to present the numismatic evidence for some of the auxiliary forts in the province of Dacia in the second half of the C3rd, at about the moment when the abandonment of this province is supposed to have occurred.

The general trend of coin-finds/period for the auxiliary forts in Dacia shows a strong decrease of coin-finds after the reign of Philip I, which reaches the lowest value in the period of 253-268 (see Fig. 1). This co-efficient records a slight increase for the next period, 268-275, followed by another gradual decrease in 275-284 and 275-284. The decrease of coin-finds after 275 can be explained by the fact that Dacia in this period was already a territory outside of the Roman empire while the increase of coin-finds starting with the period of Constantine I may be explained by the issue of abundant quantity of the bronze coinage as well as the re-conquest of the southern part of the former province of Dacia during his reign.

The material used for this paper is the isolated coin finds from the sites of auxiliary sites of Dacia and, for comparison, Pannonia. For the method, I have divided the history of Roman province of Dacia into chronological periods and calculated the finds/period and the percentage of this co-efficient to allow comparisons with other areas. Up to the early C3rd, the chronological sequences are most easily formulated in terms of the length of imperial reigns (Reece 1987: 73). The Severan dynasty is divided into two periods: the first group contains the reigns of Septimius Severus to Macrinus (193-218); the second includes the period from Elagabal to Maximinus I Thrax (218-238). The reigns of Gordian III (238-244) and Philip I (244-249) are studied separately based on the monetary changes which took place in these periods. The next chronological period, 249-253, is based on the type of commonly found coins, ie. antoniniani and bronze issues of Viminacium and PROVINCIA DACIA. The period after Trebonianus Gallus/Aemilian was divided in two groups, 253-268 and 268-275, following the ancient authors and about the moment when Dacia was abandoned by the Roman administration. The following period, 275-284, is characterised by radiates of slightly higher silver content. Due to the state of publication and also because this paper is not focused on this period, the last chronological sequences have been carried out by the periods: 284-305 and 306-337. The last three chronological sequences represent the period when Dacia was lost to the Roman empire followed by the partial re-conquest by Constantine I of the southern part of Dacia.

The marked decrease in coin-finds in the period 253-268 may be explained either as: a decrease of coin production; or a cessation of the regular supply to the auxiliary troops. To have a clearer explanation for this situation, we compared the numismatic evidence from the auxiliary forts in Dacia with other similar items. The comparative percentage of coin-finds/period from Dacia (Gazdac PhD thesis) and the auxiliary forts shows a similar pattern for the fluctuations of coin-finds between these two categories (see Fig. 10, Tab. 8). A slight difference can be observed for the period after the reign of Philip I, especially for 253268, when the co-efficient of auxiliary forts records a stronger decrease than that of the province. According to this graph it may be suggested that whole province could have suffered a strong diminution of coin supply. But the reason for the diminution of coin-finds in this period does not allow us to see if it is a particular phenomenon for this province or is a general pattern for the empire, following possibly low coin production. Going further, I have made comparative graphs of some settlements and their auxiliary forts, Porolissum, Gherla and Drobeta (see Figs. 8, 9, Tab. 8). In two of the cases, Porolissum and Gherla, the coin-finds recovered from the territory of the forts stopped with the issues of Philip I while later issues continue to be found in a very small quantity on the civilian part of the same site (see Fig. 9, Tab. 9). For our purposes, the comparative graph of the site of Drobeta is the most important. Until the period 253-268,

A major problem for this subject was the state of research and the publication of the numismatic material. Only for some auxiliary forts has this category of material been published but without stratigraphical context. Another difficulty for this approach is that in many cases, the forts were only partially excavated.

737

Limes XVIII

the percentage of finds/period of auxiliary fort is higher or slightly higher than the one of the finds recovered from the territory of the town of Drobeta, except for the reign of Philip I. Then, for the following periods of 253-268 and 268-275, the co-efficient of the auxiliary fort records a dramatic decrease in comparison with the coefficient of the town for which the period of 253-268 has a ‘normal’ decrease, like in the case of the province of Dacia. At the same time for the next period, 268-275, both co-efficients raise again, but the co-efficient for the town records a huge value - the highest value of the entire period of study while that of the auxiliary fort of Drobeta has just a medium value. And, again, for the period after the abandonment of Dacia the territory of the auxiliary fort produces more coin-finds than the town of Drobeta proving that the auxiliary fort was reused by the inhabitants probably as a fortified place. The reuse of this fort under Constantine as military place is marked by the huge increase of the coin-finds on its territory comparative with the town’s coin-finds in the same period.

Another possibility to establish if the auxiliary forts of Dacia were facing a bad supply of coinage or it was a general phenomenon in the Roman empire in the period of 253-268 is by comparing with the situation of the auxiliary forts from neighbouring provinces. For the moment, due to the state of publication, only numismatic material from auxiliary forts of Pannonia was available for study (FMRU 1999). The comparative graphic of the percentage of coinfinds/period indicates that the intensity of coin-finds in Dacia has a much higher value in the auxiliary forts of Dacia than those of Pannonia until the period of 249-253 (see Fig. 11, Tab. 11). Starting with this period the comparative trends of the coin-finds for the auxiliary forts in these two provinces follows two opposite directions. The co-efficient for Pannonia indicates a huge increase and an ascending tendency for the period of 249-275 while the same co-efficient records a dramatic decrease, especially for the period of 253-268, for the auxiliary forts of Dacia. In conclusion, the pattern of coin-finds in the auxiliary forts of Dacia compared with different categories of sites indicates that the strong decrease of the coins issued in the period of 253-268 is a common pattern of the auxiliary forts in Dacia. At this stage of research, it may be affirmed that if the auxiliary forts of Dacia were not abandoned under the reign of Gallienus, they were at least facing a serious problem of coin supply in this period.

For the auxiliary forts of Porolissum and Gherla the lack of any coins issued after the reign of Philip I may be a matter of chance. The case of Drobeta suggests that the auxiliary garrison, if this was still garrisoned there, had a bad coin supply in the period following the reign of Philip I.

Tab. 1 - Table of coin-finds from Porolissum, Micia, Hoghiz POROLISSVM - auxiliary fort Pomăt Period D No

42

MICIA - VEŢEL Period

98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 249-253 TOTAL

% 50 60 62.5 75 50 94.7 100

3 3 5 3 1 18 9

98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 TOTAL

No 2 2

2(p) 2

8 (2p)

Ant

No

S

%

No 2

% 33.3

2

75

D

1

5.2

1 2 4

100 100 7.1

No

No 1 1 1 1

25

4

S % 50 100

Dp

7.1

%

4 1

50 12.5

66.6 66.6

1

33.3

24.2

1 6

100 18.1

No

1 2 1

4

% 16.6 20 12.5 25

4

Dp

As

7.1

As

%

No 2

12.5 25 33.3

12.1

738

3 4 2

11

No

finds/ period

%

1

20

1

50

2

3.5

prov. issues

% 50 37.5 50 66.6

33.3

No

Total No

0.3 0.2 0.3 0.2 0.1 0.7 0.4 0.1 0.4

finds/ period

%

6 5 8 4 2 19 9 1 2 56

Total No

1

12.5

1

33.3

1

100

2

6

0.2 0.09 0.3 0.4 0.2 0.1 0.1 0.1 0.2

4 2 8 8 3 3 3 1 1 33

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia HOGHIZ Period

98-117 117-138 138-161 161-180 193-218 218-238 238-244 244-249 275-284 284-305 306-337 TOTAL

D

Ant

No 2 1 1

% 100 100 33.3

1 2

50 66.6

7

29.1

No

S

%

No

117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 TOTAL

No 14 (4p) 3 (2p) 12 (4p) 6 (1p) 1 27 21 (8p)

84 (19p)

ILIŞUA Period

98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244

66.6 100 50

66.6

1 1

33.3 100

1

100

3

12.5

6

%

%

4 3 7

4.1

Dp

As

prov. issues

% 21.1

No 32

9.6

6

19.3

2

6.4

19

61.2

34.2

9

25.7

5

14.2

9

40

6

40

1

6.6

16.6 96.4 100

3

50

1

16.6

2 2 38

28.5 40 17.3

71.4 60 3.6

Ant % 10.3

4 (2p) 2 (1p) 22 (15p) 13 (7p) 3(p)

No

24

S

%

10.9

Dp

% 45

1.4

31

25.7

1.5

35

2

13.3

0.7

15

1 1

16.6 3.5

0.5 1.1 1

6 28 21

1.1 1

7 5 219

finds/ period

Total

64

29.2

As Qd No % 29 50

No 11

% 18.9

3.7 35.8

12 2

22.6 5.1

33 15 1

62.2 38.4 2.5

57.1

2 14 1 cast 1

14.2

1

14.2

1

50

1

25

1

50

3 (2p) 6 (1p) 9 (3p)

1

1

18

No

1

52 1 cast

26

739

80 1

% 3

1

1.8

No 58 53 39

14.2

0.3

7

25

0.3

4

22.2

72

0.4

prov. issues

1

4.3

0.9

23

1

5.5

0.9

18

1

6

5

25

50 24

3.2

2.5 1.6

95.6 4

%

Total No 71

% 20.6

72.2

No

finds/ period 3.7

No 12

9.4 17.9

No 2 1 3 1 2 3 3 1 1 4 3 24

0.1 0.04 0.1 0.05 0.08 0.1 0.5 0.2 0.1 0.1 0.09

100 100 29.1

No 15

5 3 8

Total

%

33.3

1

S

No

finds/ period

% 14

38.3

No

25

N

No 10

No 6 (4p) 5 7 (3p)

62 (35p)

No

1 2

D

244-249 TOTAL

% 19.7

%

2 1 1

Tab. 2 - Table of coin-finds from Buciumi, Ilişua BUCIUMI Period D Ant

98-117

As

4 3

233

Limes XVIII

Tab. 3 - Table of coin-finds from Gherla, Certiae, Războieni GHERLA Period D Ant S

98-117 117-138 138-161 161-180 180-192 193-218 218-238

No 22 (1p) 4 (1p) 7 (1p) 7 (1p)

% 70.9

20 (5p) 14 (1p)

% 12.9

No 2

% 6.4

1.6

No 31

19

3

14.2

2

9.5

12

57.1

1

21

53.8

1

7.6

1

7.6

4

30.7

0.5

13

58.3

3

25

2

16.6

0.6

12

95.2

1 1

100 4.7

0.08 0.8

1 21

0.7

14

0.1 1.6

1 8

60.6

1 2 (1p) 3 (1p)

218-238

RĂZBOIENI Period

2 1 3

75

2.4

18

14.7

No 1 1 1 2 2 1

90

41.2

3 (1p)

66.6 100 37.5

5.7

20

Dp % 20 9 14.2 66.6 40 12.5

No 1 1 1 1

16.3

As

% 9 14.2 33.3 20

No 3 4 2

122

N

% 60 36.3 28.5

No

prov. issues %

No 1

finds/ perio d

% 9

Total

0.2 0.5 0.3 0.1 0.4 0.3

No 5 11 7 3 5 8

1

10

0.5

10

66.6

1

33.3

0.5

3

10

7

70

2 0.04

10 1 63

4.7

S %

7

S

%

D No

98-117 138-161 161-180 193-218 218-238 TOTAL

No

2 (1p) 1 26 (3p)

6

40 87.5

238-244 244-249 284-305 TOTAL

100 25

Ant % 20 36.3 42.8

2 7 (2p) 9 (1p)

Total

No 4

100

74 (10p)

No 1 4 3

finds/ period

% 9.6

CERTIAE - ROMITA Period D

98-117 117-138 138-161 161-180 180-192 193-218

%

As

No 3

238-244 244-249 TOTAL

No

Dp

16

25.3

finds/ period

No 1 1 2 1

% 100 100 100 33.3

5

62.5

0.05 0.04 0.1 0.1 0.05

2 7

Total No 1 1 2 3 1 8

740

20 11.1

9

14.2

1 1

100 1.5

1

1.5

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia Tab. 4 - Table of coin-finds from Cumidava, Praetorium, Tibiscum CVMIDAVA - RÎŞNOV Period D S As No

%

No

98-117 117-138 138-161 193-218

1 1

33.3 100

218-238 TOTAL

1 3

50 33.3

%

1 2

100 66.6

1 4

50 44.4

No 2

% 100

2

22.2

finds/ period

Total No 2 1 3 1

0.1 0.04 0.1 0.04 0.1

2 9

PRAETORIVM - MEHADIA Period

98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 Total

D

Ant

No 5 2 (1p) 4

% 35.7 16.6

1 9 (3p) 20 (1p)

50 56.2

117-138 138-161 193-218 218-238 TOTAL

%

50

3 7 1 1 4 6

28

41 (5 p)

No 3 (1p) 3 5 1 1 12 (1p)

No

Dp %

1

8.3

2 1

25 50

No 1

As Sem. % 7.1

No 7 1 1

%

N No

prov. issues %

No

22

D

37.5 58.3 16.6 33.3 100 100

15

finds/ period % 100 100 100 100 100 100

0.1 0.1 0.2 0.04 0.05

%

58.3 8.3 12.5 1 1

64.5

TIBISCVM - JUPA Period

98-117

No

S

50 6.2

unid. denom.

finds/ period

Tota l

0.7 0.5

No 14 12 8 2 2 16

No 8 1

% 57.1 8.3

1 1

12.5 50

6

37.5

0.3 0.1 0.1 0.6

1

3.2

11

35.4

1.5

31

2 1 3 2

25 8.3 50 66.6

3 4 2

37.5 33.3 33.3

1.3 2.4 1.5 0.2 0.5 0.6 0.04 0.6

37

25.3

8 12 6 3 4 6 1 20 146

13

8.9

1

0.6

Total No 3 3 5 1 1 12

741

8 1

5.4 0.6

1 20 21

100 100 14.3

2

1.3

Limes XVIII

Tab. 5 - Table of coin-finds from Cristeşti, Bologa, Arutela, Inlăceni CRISTEŞTI Period

98-117 117-138 138-161 161-180 218-238 238-244 244-249 268-275 275-284 284-305 306-337 TOTAL

Au No 1

D % 50

No

%

5

1

3.2

5

16.1

218-238 TOTAL

%

1 2 1 1

20 40 100 100

5

29.4

No 2 (1p) 1

3 (1p) 1 7 (2p)

INLĂCENI Period

117-138 161-180 193-218 218-238 238-244 244-249 TOTAL

%

No

%

1

9

1

100

6

100

2 2

100 100

1

100

5

16.1

No

No 2

% 50

1

20

6

35.2

3

17.6

S

No

As %

No 1 10

N % 50 90.9

No

finds/ period %

100

20

2 1

40 25

75

1

25

100 41.1

4

23.5

S

No

40

1 3

17.6

1

1

finds/ period

No 1

% 50

1 1 3

100 100 33.3

0.09 0.05 0.1 0.05 0.1 0.2

No

2 2

25

5.8

4

Total No 2 1 3 1 1 1 9

742

11

No 4 5 5 1 1 1 17

0.2 0.2 0.2 0.04 0.05 0.2

No

finds/ period

%

40 50

23.5

100 100 16.1

Total

prov. issues

%

2 3 5

35.4

%

As

%

3.2

finds/ period

2

Dp %

1

As

% 50 40 40

No

66.6

9.6

Dp

% 100

% 50 100 100 100

3

No 2 2 2

D No 1 1 3 1

1

S

ARVTELA - CĂLIMĂNEŞTI Period D

117-138 138-161 180-192 193-218

No

Dp

100

D No

98-117

S

1

BOLOGA Period

98-117 117-138 138-161 193-218 218-238 244-249 TOTAL

Ant

1

100

1

5.8

Total

0.1

No 2

0.2 0.1 0.08 0.1

5 4 1 4

0.05

1 17

0.1 0.5 0.04 0.05 0.2 0.1 0.2 0.2 0.2 0.09 0.09

Total No 2 11 1 1 5 1 1 2 2 2 3 31

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia Tab. 6 - Table of coin-finds from Pojejena, Răcari, Drobeta POJEJENA Period D Ant S No 117-138 138-161 193-218 218-238 238-244 244-249 249-253 253-268 TOTAL

%

No

1 1

33.3 100

1

100

3

1 1

100 33.3

3

20

D

98-117 117-138 138-161 161-180 193-217 218-238 238-244 244-249 249-253 TOTAL

%

% 66.6

6

54.5

5 1

71.4 33.3

%

1

100

1

3.3

% 100

2 1 1 8

66.6 100 100 53.3

No

Total No 1 3 1

0.04 0.1 0.04 0.05 16.6 0.6 0.7 0.06

1

S

No

46.6

No 1

finds/ period

66.6

Ant

No 2

14

No 2

20

RĂCARI Emperor

%

prov. issues

%

1

33.3

1

3.3

1 1 3 3 1 15

6.6

unid. denom No % 1 33.3 2 100 5 45.4 1 100 2 28.5 1 33.3 1 100 1 14

finds/ period 0.1 0.09 0.4 0.05 0.1 0.1 0.1 0.2 0.2

100 46.6

Total No 3 2 11 1 7 3 1 1 1 30

DROBETA - fort Period

D No

98 - 117 117 - 138 138 - 161 161 - 180 180 - 192 193 - 218 218 - 238 238 - 244 244 - 249 249 - 253 253 - 268 268 - 275 275 - 284 284 - 305 306 - 337 TOTAL

Ant %

1

8.3

1 4 (2p) 5 (p)

50 50 17.8

11 (7p)

4.2

No

S %

4 1 4 8 7 2

40 9 100 100 100 18.1

26

10

No 2 2 6 3 1

4 6 10

34

Dp % 33.3 25 50 75 50

44.4 60 90.9

13

As

No

%

1

12.5

1

25

1

3

No 2 1 2

N

% 33.3 12.5 16.6

4 23 4

11.1

1.1

No

5

743

1.9

31

%

prov. issues No %

unid. denom. No % 2 33.3 4 50 3 25

50 82.1 44.4

11.9

9 132 141

81.8 100 54.2

9

3.4

finds/ period

Total

0.3 0.3 0.5 0.2 0.1 0.3 1.4 1.5 2 2.7 0.2 1.6 0.7 0.5 4.2

6 8 12 4 2 8 28 9 10 11 4 8 7 11 132 260

Limes XVIII

Tab. 7 - The site-finds from the auxiliary forts of Dacia AD 98 - 117 Au POROLISSVM MICIA Hoghiz Buciumi Ilişua Gherla CERTIAE Războieni CVMIDAVA PRAETORIVM TIBISCVM Cătunelei Cristeşti Bologa ARVTELA Inlăceni Pojejena Sighişoara Sânpaul Olteni Răcari BVRIDAVA Teregova Orhei Enoşeşti Gilău Tihău Odorhei Urluieni Breţcu Bumbeşti Comalău Călugăreni Săpata de Jos DROBETA CASTRA TRAIANA Brâncoveneşti Gornea TOTAL

D

S

Dp

3 2 2 10 4 21 1

2

1

As

plated

AD 117 -138 prov. mint

unid. denom

f/ year

TOTAL

D 3 2 1 1 5 2 4

1

6 4 2 71 58 31 5 1 2 14 3

1

2 4 2

2 10 12 3 1 1

15 11 4

32 29 2 3

4 2 1

2 5 2

1

1

8 1

2

2

1

1 3 1

S

6 2 3 1

Dp

As

1

1

2 12 2 1

19 33 12 4

1 1

7

1 2 4

10 2 2 1

1

1

1

plated

2

prov. mint

unid. denom

f/ year

1 1

2

1

1

1 3

1

2

1 12 3

1

4 3

1 1

1 1 1 1 1 3

1 1 1 1 1

2

4

2

1

1 1 3

2

11 5 6 2 1 4 2

2

2

3 3

4 1 6

1

1

4

8

1

2

2

2

2

1

26

26

20

96

S

Dp

As

plated

1 2

4

6

1

1 1

58

39

D

S

Dp

5

2 4 2 9 14 1 1 1 2 2

1 1

3

5 2 1 1

9 15 4 2

1 8 4 6 3 1 4 5 2

2 1

38

73

9

1

12

12.1

plated

unid. denom

f/ year

TOTAL

D

8 8 3 35 39 13 7 1 3 8 5

3

231

As

Qd

1

4 3 1

1 1 1 1

1

2

1 5 4

1

3 4

5 2 6

1 1 6 1 1 2 2

1 1 1 1

2 1 4

1 1

AD 161 -180 prov. mint 1 1 2 1

1 1

5

3

2 1 1

6

1 2 1

1

4

1

2 1

6

2

53

14

1 44

1

8

1

1 2

1

1

1 11

1 1 1 1 1

1 2 1 3 7

2

2

1 1

1 1 1

3 4

3 4

3

12

1 54

1 1 1

5

1

1

16

unid. f/ year TOTAL denom 4 8 1 15 7 13 3 2

1 1 2

TOTAL 5 2 1 31 53 21 11

1 3

AD 138 - 161 POROLISSVM MICIA Hoghiz Buciumi Ilişua Gherla CERTIAE Războieni CVMIDAVA PRAETORIVM TIBISCVM Cătunelei Cristeşti Bologa ARVTELA Inlăceni Pojejena Sighişoara Sânpaul Olteni Răcari BVRIDAVA Teregova Orhei Enoşeşti Gilău Tihău Odorhei Urluieni Breţcu Bumbeşti Comalău Călugăreni Săpata de Jos DROBETA CASTRA TRAIANA Brâncoveneşti Gornea TOTAL

Sem

16

8.2

3

1

21

8

4

1

1 1 190

18

744

1 11

4

1

12

3.9

75

9.2

194

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia

AD 180 - 192 D POROLISSVM MICIA Hoghiz Buciumi Ilişua Gherla CERTIAE Războieni CVMIDAVA PRAETORIVM TIBISCVM Cătunelei Cristeşti Bologa ARVTELA Inlăceni Pojejena Sighişoara Sânpaul Olteni Răcari BVRIDAVA Teregova Orhei Enoşeşti Gilău Tihău Jidava Odorhei Urluieni Breţcu Bumbeşti Comalău Călugăreni Săpata de Jos DROBETA CASTRA TRAIANA Brâncoveneşti Gornea TOTAL

POROLISSVM MICIA Hoghiz Buciumi Ilişua Gherla CERTIAE Războieni CVMIDAVA PRAETORIVM TIBISCVM Cătunelei Cristeşti Bologa ARVTELA Inlăceni Pojejena Sighişoara Sânpaul Olteni Răcari BVRIDAVA Teregova Orhei Enoşeşti Gilău Tihău Jidava Odorhei Urluieni Breţcu Bumbeşti Comalău Călugăreni Săpata de Jos DROBETA CASTRA TRAIANA Brâncoveneşti Gornea TOTAL

S

Dp

As

1

1 2

1 1 1 2

3 1 1 2

1

1 1

plated

2

unid. denom

f/ year

TOTAL

Ant

D

2 3

1

18 1 27 7 15 5 2 1 6 1

6 4 1 5

1

1

1

1

prov. mint

1

2

1

1

1

1 2 3 1 3 1

4

S

As

AD 193 -218 prov. plated mint 2

1

15 5 2

1

3

1

unid. denom

f/ year

19 3 2 28 23 20 8 3 1 16 1

1 1 1 1

1

6

1 4 3 1 4 1

1 1

5

2

7

2

2

1

1

1 7 1 1 1 1

1

1 10

D

S

9 2 2 13 6 12 8 1 1 19 1 1 5 1 1 1 1 1 1 1 1 1

4

5

1

AD 218 - 238 prov. mint

As plated

1 1 4 1 1

8 7 1 1

1 1

1

1

1 1

1

1

2

2

4

2

1

34

unid. f/ year TOTAL denom 9 3 3 21 18 14 10 1 2 11 32 1 1 5 1 1 1 1 2 1 3 1 3 1

119

4

1

Ant

S

Dp

plated

1 1 2

3

2

8

1

1

2 5 1 1 1

4

1

33

8

2 10

7

AD 238 -244 prov. unid. f/ year TOTAL mint denom 1 1 1 3 7 5 6 1 1 3 3

8

1

1 1

1 1

1

1

1

4 2

2 1

1

6

1

2

1

9

2 2

1

2 2

5

105

2,8

3 5

2

1

10

1 9

2

3 5

1 1 1 2

2

10

1

5

23

25

1 31

18

7 28

9,5

1 190

17

745

TOTAL

4

1

1 12

1

4

6

5

9

4

7,5

1 45

176

Limes XVIII

AD 244 - 249 Ant POROLISSVM MICIA Hoghiz Buciumi Ilişua Gherla CERTIAE Războieni CVMIDAVA PRAETORIVM TIBISCVM Cătunelei Cristeşti Bologa ARVTELA Inlăceni Pojejena Sighişoara Sânpaul Olteni Răcari BVRIDAVA Teregova Orhei Enoşeşti Gilău Tihău Jidava Odorhei Urluieni Breţcu Bumbeşti Comalău Călugăreni Săpata de Jos DROBETA CASTRA TRAIANA Brâncoveneşti Gornea TOTAL

S

As

plated

2

unid. f/ year TOTAL denom 2

Ant

AD 249 -253 unid. f/ TOTAL denom year

S plated 1

3 5 1 1

1 2 18 6 7

7

1

1

1 1

2 4

1

12

1 2 1

1 3 1

1

1

1

1 1

1 1 1 1

6

25

50

3

2

1

6

17,4

AD 275-284 Ant

1

1

10

2

18

f/year

AD 284-305 Ant

N

Total

4

4

1

1

1

3

AD 306 - 337 f/year

N

f/year

3

1

6

1

1

12

2

2

2

2

3

1

3

1 1

1 2

1

7

18

1

11

4

2,4

1

4

87

1

17

3

1

1

1 2

1

10

6

1 1

1 2

2

TOTAL

4

10

2

1

f/ year

1

1

4

S

1 1

1 1

AD 253 - 268 Ant

1 5 25 8 10

1

AD 268275 Ant f/year POROLISSVM MICIA Hoghiz Buciumi Ilişua Gherla CERTIAE Războieni CVMIDAVA PRAETORIVM TIBISCVM Cătunelei Cristeşti Bologa ARVTELA Inlăceni Pojejena Sighişoara Sânpaul Olteni Răcari BVRIDAVA Teregova Orhei Enoşeşti Gilău Tihău Jidava Odorhei Urluieni Breţcu Bumbeşti Comalău Călugăreni Săpata de Jos DROBETA CASTRA TRAIANA Brâncoveneşti Gornea TOTAL

Dp

2

2

9

11

4

16

20

132

0,9

746

2 157

5

6

24

7

4 2

0,6

9

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia Tab. 8 -Comparative tables auxiliary fort - settlement in Dacia POROLISSVM PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 306-337 TOTAL

Settlement No. finds/ coins period 80 4.2 78 3.7 118 3.8 70 3.6 27 2.2 211 8.4 116 5.8 19 3.1 24 4.8 5 1.2 1 0.06 1 0.1 4 0.1 757 41.06

PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 253-268 268-275 275-284 284-305 306-337 TOTAL

Settlement No. finds/ coins period 37 1.9 39 1.8 34 1.4 21 1.1 5 0.4 39 1.5 23 1.1 5 0.8 13 2.6 1 0.06 2 0.2 1 0.1 1 0.04 6 0.1 13.1 227

PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

Town No. finds/ coins period 7 0.3 8 0.3 12 0.5 3 0.1 2 0.1 1 0.04 16 0.8 10 1.6 14 2.8 11 2.7 24 1.6 50 7.1 5 0.5 3 0.1 35 1.1 19.64 201

%finds/ period 10.2 9 9.2 8.7 5.3 20.4 14.1 7.5 11.6 2.9 0.1 0.2 0.2

No. coins 6 5 8 4 2 19 9 1 2

56

Auxiliary fort finds/ %finds/ period period 0.3 11.1 0.2 7.4 0.3 11.1 0.2 7.4 0.1 3.7 0.7 25.9 0.4 14.8 0.1 3.7 0.4 14.8

2.7

GHERLA %finds/ period 14.5 13.7 10.6 8.3 3 11.4 8.3 6.1 19.8 0.4 1.5 0.7 0.3 0.7

No. coins 31 21 13 12 1 21 14 1 8

122

Auxiliary fort finds/ %finds/ period period 1.6 22.9 1 14.3 0.5 7.1 0.6 8.5 0.08 1.1 0.8 11.4 0.7 10 0.1 1.4 1.6 22.9

6.98

DROBETA %finds/ period 1.5 1.5 2.5 0.5 0.5 0.2 4 8.1 14.2 13.7 8.1 36.1 2.5 0.5 5.6

No. coins 6 8 12 4 2 8 28 9 10 12 4 8 7 11 132 261

747

Auxiliary fort finds/ %finds/ period period 0.3 1.7 0.3 1.7 0.5 2.9 0.2 1.1 0.1 0.5 0.3 1.7 1.4 8.3 1.5 8.9 2 11.9 3 17.8 0.2 1.1 1.6 9.5 0.7 4.1 0.5 2.9 4.2 25 16.8

Limes XVIII

DACIA PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

General statistics No. coins 725 722 811 432 142 864 836 282 402 169 149 92 40 57 392 6115

finds/ period 38.1 34.3 35.2 22.7 11.8 34.5 41.8 47 80.4 42.2 9.9 13.1 4.4 2.7 12.6 430.7

%finds/ period 8.8 7.9 8.1 5.2 2.7 8 9.7 10.9 18.6 9.7 2.2 3 1 0.6 2.9

No. coins 231 194 190 75 34 176 190 45 87 25 9 17 18 20 157 1468

Auxiliary forts finds/ %finds/ period period 12.1 12.7 9.2 9.7 8.2 8.6 3.9 4.1 2.8 2.9 7 7.3 9.5 10 7.5 7.9 17.4 18.3 6.2 6.5 0.6 0.6 2.4 2.5 2 2.1 0.9 0.9 5 5.2 94.7

Comparative table of the auxiliary forts coin-finds from Dacia and Pannonia DACIA No. coins 231 194 190 75 34 176 190 45 87 24 9 17 18 20 157 1467

PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

finds/ period 12.1 9.2 8.2 3.9 2.8 7 9.5 7.5 17.4 6 0.6 2.4 2 0.9 5 94.5

%finds/ period 12.8 9.7 8.6 4.1 2.9 7.4 10 7.9 18.4 6.3 0.6 2.5 2.1 0.9 5.2

No. coins 77 83 106 80 23 75 91 47 52 52 234 234 132 144 764 2194

PANNONIA finds/ %finds/ period period 4 2.6 3.9 2.5 4.6 3 4.2 2.7 1.9 1.2 3 1.9 4.5 2.9 7.8 5.1 10.4 6.8 13 8.5 15.6 10.2 33.4 21.9 14.6 9.5 6.8 4.4 24.6 16.1 152.3

Tab. 9 - Tables of coin-finds from the auxiliary forts in Pannonia ARRABONA –Györ PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

Au No 1

2

% 14.2

22.2

D No % 1 14.2 4 6 2 5 7

Ant No

28.5 46.1 66.6 55.5 70

1 2 8 11 1 3

2.2

25

18.3

23

S % No 1 4 5 3 1 2 2 1 3 33.3 2 100 100 100 14.2 16.9

24

% 14.2 57.1 35.7 23 33.3 22.2 20 100 100 66.6

Dp No

2 1

As % No % 4 57.1 3 42.8 14.2 3 21.4 7.6 3 23

1

Arg No

3

2.2

748

14

finds/ period

10.2

2

28.5 1.4

1 1

4 2.6 37 0.7 41

TOTAL

%

10

2 17.6

%

Sil N No % No

57.1 97.3 30.1

0.3 0.3 0.6 0.6 0.2 0.3 0.5 0.1 0.6 0.7 0.1 1.1 1.2 0.3 1.2

7 7 14 13 3 9 10 1 3 3 2 8 11 7 38 136

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia

VETVS SALINA - Adony PERIOD 98-117 117-138 138-161 161-180 193-218 218-238 238-244 TOTAL

D No

%

1

100

1

6.6

S No

Dp % No 1

4

66.6

1

100

2 7

100 46.6

% 50

As No 1 2

1

6.6

3

% 50

unid. denom. No

finds/ period %

2

100

1

100

3

20

33.3

20

0.1 0.09 0.2 0.05 0.04 0.05 0.3

TOTAL 2 2 6 1 1 1 2 15

INTERCISA - Dunaujvaros PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

D No 2 4 7 3 2 18 22 1

59

% 25 33.3 33.3 30 22.2 85.7 81.4 7.1

10.2

Ant No

%

S No 2 2 6 3 4 2 2 6 9 8

1

4.7

4 2 5 100 72 50 10

28.5 18.1 38.4 100 100 100 27

244

42.4 44

Dp % No % 25 1 12.5 16.6 1 8.3 28.5 1 4.7 30 44.4 1 11.1 9.5 7.4 42.8 3 21.4 81.8 61.5

7.6

7

S No 1

As No 3 3 4 4 1

% 37.5 25 19 40 11.1

prov. issues No

3

1.2

15

%

As No

%

Qd No 1

50

1

50

2.6

3

%

N No

%

unid. denom. No

finds/ period %

2 3

16.6 14.2

1

11.1

6

1

11.1

0.5

27 72.9 170 100 197 34.2

%

N No

0.4 0.5 0.9 0.5 0.7 0.8 1.3 2.3 2.2 3.2 6.6 10.2 5.5 1.7 5.4

TOTAL 8 12 21 10 9 21 27 14 11 13 100 72 50 37 170 575

ANNAMATIA - Baracs PERIOD 98-117 117-138 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 275-284 284-305 306-337 TOTAL

D No

1

%

Ant No

%

% 33.3

Arg No

50

1

1

3.4

1

3.4

1

3.4

1

3.4

1

3.4

749

2

66.6

2

6.8

unid. finds/ denom. period % No % 2 66.6 0.1 0.09 1 50 0.1 1 100 0.08 1 100 0.04 3 100 0.1 1 100 0.1 1 100 0.2 1 100 0.2 1 100 0.06 1 100 0.1 1 33.3 0.1 9 100 0.2 10 34.4 13 44.8

TOTAL 3 2 2 1 1 3 1 1 1 1 1 3 9 29

Limes XVIII

AD MVRES - Acs PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

D No

%

1

14.2

9 8

90 80

Ant No

1 3 3

1 8 13 5 3 18

%

S No

20 100 100 100 33.3

30

1 1 3 4 4

Dp % No 2 14.2 2 42.8 2 42.8 2 2 10 10 1 75 1 80 1 80

% 66.6 28.5 28.5 28.5 100

As No 1 4 6 3

% 33.3 57.1 85.7 42.8

finds/ period

N No

66.6 100

0.1 0.3 0.4 0.4 0.1 0.4 0.5 0.6 1 1.2 0.5 1.8 0.5 0.4 1.7

prov. issues No

%

N No

1 1

1.8 1.7

10 25 20

20

13

14

S % No 6 10 16 15 1 5 15 11 44 10 9 28.1 22 9 30 18 123 100 141 100 65 100 15 17

Dp No 14 5 6 9 1

% 25.4 9.4 10.3 20 12.5

2

8

1

3.3

As No 26 26 12 6 3 1 1 1 1 2

6 53 59

TOTAL

% 3 7 11 9 2 10 10 4 5 5 8 13 5 9 53 154

SOLVA - Esztergom PERIOD 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337 TOTAL

D No 8 6 12 7 3 27 24 1

88

% 14.5 11.3 20.6 15.5 37.5 81.8 60 4

6.8

Ant No

373 28.8 118

% 10.9 18.8 27.5 33.3 12.5 15.1 37.5 40 68.7 60

9.1

38

2.9

79

% 47.2 49 20.6 13.3 37.5 3 2.5 4 3.1 6.6

6.1

2

%

73 495 568

0.1

unid. denom. No 1 5 11 8

% 1.8 9.4 18.9 17.7

25

1.9

82.9 100 43.9

finds/ period

TOTAL

2.8 2.5 2.5 2.3 0.5 1.32 2 4.1 6.4 7.5 8.2 20.1 7.2 4.1 15.9

55 53 58 45 8 33 40 25 32 30 123 141 65 88 495 1291

Tab. 10 - The site-finds from the auxiliary forts of Pannonia AD 98 - 117

ARRABONA

Au

D

S

1

1

1

VETVS SALINA 2

INTERCISA

2

Dp 1

1

2

1

3

8

ARRABONA

4 4

2

As 3

1

7

3

2

2

12

1

2

4

54

6

10

5

26

1

5

1

9

3.9

f/ period

TOTAL

1

11

9

18

36

D

S

Dp

4

5

2

AD 138 - 161 prov. As mint

4

77

10

18

8

37

unid. denom

f/ period

TOTAL

D

S

Dp

As

14

6

3

1

3

3

6

1

4

3

2

6

1

2

2

2

3

21

3

12

16

6

12

1

11

TOTAL

23

30

11

25

1

16

4.6

7

AD 161 - 180 unid. denom

1 10

1

2

6

8

45

16

10

11

1

3

2

3

58

7

15

9

106

18

24

12

750

53

13

4

1

ANNAMATIA SOLVA

2

1 3

TOTAL

2 1

26

AD MVRES

f/ period

3

14

7

Dp

3

VETVS SALINA INTERCISA

S

1

6

2

D

AD 117 - 138 prov. unid. mint denom

2 8

SOLVA

TOTAL 7

1

AD MVRES

f/ period

4

ANNAMATIA

TOTAL

As

unid. denom

9 4.2

80

83

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia

AD 180 - 192

ARRABONA

D

S

2

1

Dp

As

unid. denom

f/ period

TOTAL

Au

D

3

2

5

Ant

2

4

1

1

ANNAMATIA

1

9

1

1

2

AD MVRES

18

SOLVA

3

1

1

3

7

6

4

4

D

S

Dp

ARRABONA

7

2

VETVS SALINA

1

INTERCISA

22

2

AD MVRES

8

1

SOLVA

24

15

TOTAL

62

20

AD 218 - 238 prov. As mint

1.9

23

unid. denom

f/ period

1 3 3

2

3

Ant

S

3

4.5

Dp

As

f/ period

TOTAL

5

1 1

Ant

D

S

Dp

10

1

1

2 4

1 10

11

1

33 1 AD 238 -244 unid. As denom

3

75

f/ period

TOTAL 1 2

6

3

3

1

14 1

11

1

10

2

1

91

15

2

22

6

1

Ant

S

3

1

2

11

5

8

Dp

1 4

40

TOTAL

3

ARRABONA

21

1

AD 244 - 249 unid. denom

1

2

59

10 1

1

9

1

3

1

TOTAL

1

27 2

27

ANNAMATIA

f/ period

1

8 2

1

9

2

TOTAL

As

2

VETVS SALINA INTERCISA

S

AD 193 -218 prov. unid. mint denom

25 1

AD 249 -253 unid. f/ As denom period

7.8

47

TOTAL 3

VETVS SALINA INTERCISA

2

9 1

ANNAMATIA 4

AD MVRES SOLVA

9

22

TOTAL

11

38

AD 253 - 268 Ant ARRABONA

f/ period

1 1 1

1

1

13 1

1

10.4

AD 268-275 f/ period

1

5

1

4

32

9

18

1

2

52

16

32

1

2

Ant

Arg

N

Total

2

4

7

27

37

1

3

9

6

9

53

5

AD 275-284

30 1

13

52

AD 284-305 f/ period

Ant

Ant

2

8

11

1

72

50

10

AD 306 - 337 f/ period

Sil

N

1

37

f/ period

VETVS SALINA INTERCISA

100

ANNAMATIA

1

AD MVRES

8

SOLVA

123

TOTAL

234

1 13

5

141 15.6

234

2 3

65 33.4

132

15 29

14.6

4

73

88

111

144

170

495 1

6.8

764

Tab. 11 - Comparative table of the auxiliary forts coin-finds from Dacia and Pannonia DACIA PERIOD

PANNONIA

No. coins

finds/ period

%finds/ period

No. coins

finds/ period

%finds/ period

98-117

231

12.1

12.8

77

4

2.6

117-138

194

9.2

9.7

83

3.9

2.5

138-161

190

8.2

8.6

106

4.6

3

161-180

75

3.9

4.1

80

4.2

2.7

180-192

34

2.8

2.9

23

1.9

1.2 1.9

193-218

176

7

7.4

75

3

218-238

190

9.5

10

91

4.5

2.9

238-244

45

7.5

7.9

47

7.8

5.1

244-249

87

17.4

18.4

52

10.4

6.8

249-253

24

6

6.3

52

13

8.5

253-268

9

0.6

0.6

234

15.6

10.2

268-275

17

2.4

2.5

234

33.4

21.9

275-284

18

2

2.1

132

14.6

9.5

284-305

20

0.9

0.9

144

6.8

4.4

306-337

157

5

5.2

764

24.6

16.1

TOTAL

1467

94.5

2194

152.3

751

24.6

Limes XVIII

Denomination abbreviations: Ant D Dp N S prov. issues unid. denom.

= Antoninianus = Denarius = Dupondius = Nummus = Sestertius = provincial issues = unidentified denomination

Bibliography Gazdac

PhD thesis Monetary circulation in the province of Dacia in the period from Trajan to Constatine I (AD 106-337). (Oxford University). FMRU 1999 passim Die Fundmünzen der römischen Zeit in Ungarn 1990-1999, Band. I-III. (Berlin). Iliescu Vl. 1971 Părăsirea Daciei în lumina izvoarelor literare, Studii şi Cercetări de Istorie Veche şi. Arheologie 22. 3: 425-442. Reece R. 1987 Coinage in Roman Britain. (London).

752

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia

20

17.4

15 12.1 9.2

10

9.5

8.2

7 3.9

5

finds/ period 7.5

6.2

5

2.8

2.4

2

268275

275284

0.6

Fig. 1 - Coin-finds/ period in the auxiliary forts from Dacia.

0.9

0 98117

117138

138161

161180

180192

193218

218238

238244

244249

249253

253268

POROLISSVM - auxiliary fort Pomăt finds/ period

0.8 0.3

0.2

0.2

0.3

finds/ period

0.4 0.2

306337

MICIA - VEŢEL

0.7

0.6 0.4

284305

0.1

0.5 0.4 0.3 0.2 0.1 0

0.4 0.1

0 98- 117- 138- 161- 180- 193- 218- 238- 244117 138 161 180 192 218 238 244 249 PERIOD

0.4 0.3 0.2

0.2

0.2 0.1

0.09

0.1

0.1

98- 117- 138- 161- 180- 193- 218- 238- 249117 138 161 180 192 218 238 244 253

PERIOD

HOGHIZ 0.6 0.5 0.4 0.3 0.2 0.1 0

finds/ period

0.5 0.2

0.1 0.04 0.1 0.05 0.08 0.1

Fig. 2 - Graphics of finds/ period in the auxiliary forts of: Porolissum, Micia, Hoghiz.

0.1 0.09

98- 117- 138- 161- 193- 218- 238- 244- 275- 306117 138 161 180 218 238 244 249 284 337 PERIOD

ILIŞUA

BUCIUMI 3.7

4 3

1.4

2

1.5

1

0.7

0.5

161180

180192

1.1

6 5 4 3 2 1 0

finds/ period 1.1 1 1

0 98117

117138

138161

193218

218238

238244

98117

244249

PERIOD

5

finds/ period 3

2.5

117138

1.6

138161

0.3

0.3

161180

180192

0.9

0.9

1

193218

218238

238244

244249

PERIOD

GHERLA 2

1.6

1.5

1

1

0.5

0.5

1.6

finds/ period 0.8

0.6 0.08

0.7 0.1

Fig. 3 - Graphics of finds/ period in the auxiliary forts of: Buciumi, Ilişua, Gherla

0 98117

117- 138- 161- 180- 193- 218- 238- 244138 161 180 192 218 238 244 249 PERIOD

753

Limes XVIII

CERTIAE - ROMITA

RĂZBOIENI

finds/ period

3 2 1 0

0.5

0.2

0.3

0.4

0.1

0.5

0.3

finds/ period

0.15

2 0.5

0.1

0.04

0.05

98- 117- 138- 161- 180- 193- 218- 238- 244- 284117 138 161 180 192 218 238 244 249 305

0.1

0.1 0.05

0.05

0.04

98-117

138-161

0

PERIOD

161-180

193-218

218-238 PERIOD

CVMIDAVA - RÎŞNOV 0.15 0.1

0.1

0.1 0.04

0.05

0.1

finds/ period 0.04

Fig. 4 - Graphics of finds/ period in the auxiliary forts of: Certiae, Războieni, Cumidava

0 98-117

117-138

138-161

193-218

218-238 PERIOD

PRAETORIVM - MEHADIA 3 2.4

2 .5 2

f i nd s/ p er i o d 1.5 1.3

1. 5 1

1.5

0.77

0.6

0.5

0.5

0.3

0 .5

0.1

0.6

0.2

0.1

0.6 0.04

0 98-117 117-138 138-161 161-180

180192

193218

218238

238244

244249

249253

253268

268275

275284

284305

306337 PER I OD

CRISTEŞTI 0.5

finds/ period

0.4

0.4 0.3 0.2

0.2 0.1

0.2

0.2

0.1

0.04 0.05

0.1

0.2

0.09 0.09

0 98- 117- 138- 161- 218- 238- 244- 268- 275- 284- 306117 138 161 180 238 244 249 275 284 305 337 PERIOD

BOLOGA 0.25 0.2 0.15 0.1 0.05 0

0.2

98-117

0.2

117-138

0.2

138-161

finds/ period

0.04

0.05

193-218

218-238

0.2

244-249 PERIOD

754

Fig. 5 - Graphics of finds/ period in the auxiliary forts of: Praetorium, Cristeşti, Bologa.

Cristian Gazdac: Monetary circulation and the abandonment of the auxiliary forts in Roman Dacia ARVTELA – CĂLIMĂNEŞTI

INLĂCENI 0.3

0.3

0.2

0.2

0.1

0.1

0.08

0.2

0.2

finds/ period

0.1

finds/ period

0.05

0.09

0.1

0

0.1

0.08

0.05

0.05

0

98-117

117-138

138-161

193-218

218-238

98-117

117-138 193-218 218-238 238-244 244-249 PERIOD

PERIOD Fig. 6 - Graphics of finds/ period in the auxiliary forts of: Arutela, Inlăceni, Bologa.

POJEJENA

RĂCARI finds/ period

0.8 0.6 0.4 0.2 0

0.6

0.04

0.1

0.04

0.05

0.1

117138

138161

193218

218238

238244

0.5 0.4 0.3 0.2 0.1 0

0.6

0.06 244249

249253

253268 PERIOD

finds/ period

0.4 0.2 0.1

0.09

98117

117138

138161

193218

0.2 0.1

0.1

0.1

218238

218238

238244

0.2

244- 251249 253 PERIOD

DROBETA - auxiliary fort 4.5

4.2

finds/ period 2.7

3 1.5

1.4 1.5

2

1.6

0.3 0.3 0.5 0.2 0.1 0.3

Fig. 7 - Graphics of finds/ period in the auxiliary forts of: Pojejena, Răcari, Drobeta.

0.7 0.5

0.2

0 306-337

284-305

275-284

268-275

253-268

249-253

244-249

238-244

218-238

193-218

180-192

161-180

138-161

117-138

98-117

PERIOD

POROLISSVM 30

GHERLA

%f i nd s/ p er i o d

25. 9

site

25

auxiliary fort

20. 4

20 14. 8 14. 1

15

11. 6

11. 1

11. 1 10. 2

10

9. 2

9 7. 4

%finds/ 25 period 20

14. 8

15

8. 7 7. 5

7. 4

3. 7

3. 7

10 2. 9 0. 1

0. 2

5

0. 2

14.5

13.714.3

7.1

11.4 11.4 10 8.3 8.5

8.3 6.1 3 1.1

0

site auxiliary fort

22.9 19.8

10.6

5. 3

5

22.9

1.4

0.4

1.5

0.7

0.3

0.7

0

98- 117- 138- 161- 180- 193- 218- 238- 244- 253- 268- 275- 284- 306117 138 161 180 192 218 238 244 249 268 275 284 305 337

PER I OD

PERIOD

Fig. 8 - Comparative graphic auxiliary fort-settlement (% of finds/ period): Porolissum, Gherla .

755

Limes XVIII

DROBETA 40 35

% finds/ period

36.1

town auxiliary fort

30

25

25 20

17.8 14.2

15 8.3 8.1 8.9

10 5

13.7

11.9

1.5 1.7 1.5 1.7

9.5

8.1

4

2.5 2.9

0.5 1.1 0.5

0

1.7

2.5

1.1

0.5 0.2

5.6

4.1

2.9 0.5

Fig. 9 - Comparative graphic auxiliary fort-settlement (% of finds/ period): Drobeta

98- 117- 138- 161- 180- 193- 218- 238- 244- 249- 253- 268- 275- 284- 306117 138 161 180 192 218 238 244 249 253 268 275 284 305 337 PERIOD

25

PANNONIA

% finds/ period

21.9

20

DACIA

18.3 16.1

15 12.7 10.2

10

9.7

10

8.6

7.9

7.3

9.5

8.5 6.5

6.8 5.1

5 2.6

3

2.5

2.9

2.7 1.2

5.2

4.4

4.1 2.9

2.5

1.9

2.1 0.9

0.6

0 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337

Fig. 10 - Comparative graphic of the coin-finds from Dacia and its auxiliary forts 20 18

% finds/ period

18.6 18.3

16 14

12.7

12 10 8 6 4

DACIA

10.9 9.7 10

9.7 8.8 7.9

8.1 8.6

8

9.7

auxiliary forts

7.9

7.3

6.5 5.2

5.2 4.1 3

2.7 2.9

2.2

2

0.6

2.9

2.5

2.1 1

0.6 0.9

0 98-117 117-138 138-161 161-180 180-192 193-218 218-238 238-244 244-249 249-253 253-268 268-275 275-284 284-305 306-337

Fig. 11 - Comparative graphic of the coin-finds from auxiliary forts in Dacia and Pannonia

756

Deserted forts – the Moesian limes after the conquest of Dacia Miroslava Mirković The whole system of defence on the Danube was altered by the Roman conquest of Dacia following her victory in the wars against Decebalus. The annexation of the territory north of the Danube caused changes to the provincial boundaries and brought about a new distribution of troops along with changes in the military command on the lower Danube, from Pannonia to the Black Sea. Altogether the new situation on the Danube, with a new Roman province in the north poses many questions and a long lasting discussion about the new military command, the defence of Dacia, the provincial boundaries of Moesia Superior, Dacia and Moesia Inferior, and about the legions and legates of Dacia, Pannonia and of Moesia Superior, not only immediately after the wars, but later still, until the end of the C2nd and the Severan dynasty.

Boljetin and Taliata and in some of the forts in Donja Klisura) dates back to the C1st, but is no later than the civil wars of 68-69; the second phase, dating back to the Flavians, is found at Singidunum, Aureus Mons, Margum and Viminacium, as well as in many forts in the Iron Gates; the most important building phase, with roads, forts and the bridge across the Danube, dated broadly to a time from Trajan to Hadrian and Marcus Aurelius, is in fact primarily connected with Trajan’s intensive preparation for the war against the Dacians. The remains of legionary and auxiliary forts are numerous from Singidunum to Pontes. Repairs under Marcus Aurelius are attested on the limes section in Moesia Superior only in the fort at Pontes (Petrović & Vasić 1996: 21). The first general reconstruction of the forts on the middle Danube after Trajan is to be dated to the Severan dynasty. This building phase is followed by that which took place under Aurelian (cf. Petrović & Vasić 1996: 15ff.).

1. It is natural to suppose that the first step after organizing the new province of Dacia was to move the boundary of both Moesian provinces and the frontier of the empire from the Danube to the north (cf. v. Domaszewski 1890: 129ff.; Patsch 1937: 1ff.; Swoboda 1939: 1ff.; Syme 1938: 84110; 1959: 122-135; Piso 1993: 1-49). The large swathe of territory north of the lower Danube, Oltenia and the part of Moldau, came under the control of the legate of Moesia Inferior; less clear are the changes on the Danube in the province of Upper Moesia.

The expected consequence of moving the frontier to the north was the abandonment of some forts on the right bank of the Danube. Recent archaeological research in the Djerdap has revealed that there is a gap between the building phase belonging to the time of Trajan’s preparation for the war with Decebalus and the next one, dated many decades later, to the time of Severi. This means that the forts on the southern Danube bank, before all those in the Djerdap canyon, must have been deserted after the Romans expanded their control to the former Dacian kingdom. A gap between the building phase of Trajan’s time (III) and the stratum dated in the time of Severi (IV) is clear in all the excavated forts; at Nova, Boljetin, Taliatae, Pontes and Diana. In Boljetin, in the middle of the upper Klisura of Djerdap these two strata are clearly divided by alluvium (Zotović 1984: 219.). The lack of any building activity on the right Danube bank in Moesia Superior from the time of Hadrian to the Severan dynasty indicates that there was a break in occupation in these forts for many decades. It means that they were abandoned for a longer period of time after the conquest of Dacia.

New light on the position of the forts on the right Danube bank in Upper Moesia from the time from Hadrian to Septimius Severus was thrown by the excavations of 19661970 (at the Iron Gates) and again between 1979-1982 (the Danube section downstream, between the exit of the Iron Gates and the Danube' great curve near Trajan’s bridge). The most important result of these excavations and research so far published is the discovery of the chronological concordance of strata in the majority of forts between Novae (Čezava) at the entrance of the Djerdap pass and that of Pontes, a fort which was linked by the bridge with Drobeta on the northern bank of the Danube. The chronology of constructions and reconstruction which was established in the auxiliary fort at Novae (Vasić 1984: 91ff.; 1987: 131-134; 1990: 897-911), is confirmed at Pontes (Garašanin & Vasić 1980: Vasić 1984: 25ff.: 1987: 711ff.) and Diana, (Karataš: Kondić 1980: 51ff.; 1984: 7ff.; 1987: 5ff.). There are a good reasons and reasonable grounds to believe that the same situation existed in the castrum of Boljetin.1 The stratification at Novae corresponds to that of Pontes and in Diana, in the fort not far from the the bridge built by Trajan. In the chronological framework from the time of Tiberius to the later Roman empire the following strata can be distinguished: the first horizon in the most important military bases (Novae,

The military situation west of the Djerdap canyon, in the legionary forts at Viminacium and Singidunum must have been different after the establishment of the province of Dacia. Archaeological research was expanded in the 1970s to Viminacium, but not to the legionary camp. Instead it was limited primarily to the cemeteries. The preserved inscriptions, however small in number, and finds from numerous graves of the huge necropolis allow us to conclude that the chronological framework established for the forts in Djerdap does not correspond to that in the legionary camps to the west. Even in the inscriptions there is clear evidence that the legio VII Claudia was stationed in Viminacium continuously under Trajan and Hadrian as well as later. This site, not far from the embouchure of the

1 Zotović 1984: 211ff., especially 219, with the explicit statement that the castrum was abandoned in the first decades of the C2nd and rebuilt in the second half of the C3rd.

757

Limes XVIII

small River Mlava into the Danube, was also important for communications from the Danube to the south after the creation of the province Dacia to the north. A new road leading from Viminacium to Dardania was built at the time of Hadrian, probably in 119, as reported in an inscription (IMS II 50). There are other monuments in Viminacium dating from the time of Hadrian’s reign. Among others, an inscription relating to recruits enlisted for the VII Claudia in the years 133-134 and discharged 159 has been found here (IMS II 51). This means that the legionary fort was occupied even after Dacia was organized as a Roman province.

other hand, forts on the southern Danube bank at the entrance of the Iron Gates as are Novae, and in the Gorge, as Boljetin (Smornae?), Taliatae4 and others built at first under Tiberius, at the time of Flavians, or during the Trajan’s preparation for the war against Decebalus, lost their military importance after the final Roman victory in 107 and the establishment of the new province of Dacia north of the river. If the defensive line on the middle Danube no longer existed from Hadrian’s reign until Marcus Aurelius or even the Severan dynasty, that could mean that the limes in Moesia Superior east of the legionary fort in Viminacium was neglected and abandoned during the most part of the C2nd and did not function as a defensive system at that time at all. Therefore it would be reasonable to pose the question: where was the new limes established and the frontier line of the Empire in this part of the Roman world? The actual state of archaeological research allows more than one answer.

The position of the legionary fort in Singidunum is less clear in the C2nd. Archaeological research on the legionary camp is not finished nor is the material properly published.2 New discoveries made in the modern town show that two legionary forts in Singidunum followed each other in the C1st and C2nd.3 The earliest one must have belonged to the second half of the C1st and could be dated to the time of the Flavian dynasty; the chronology of the second, stone built legionary camp is still not established. It is not even possible to say whether it existed in the second half of the C1st. However, the conclusion that the latter camp was in use continuously during the C2nd is important. It was suggested as one building phase without a break beginning after Nerva and lasting until the Severan dynasty. There is no Latin inscription which could be dated exactly to the first half of the C2nd, but this proposed chronology with the continuous stationing of the legio IV Flavia might be supported by the fact that bricks marked with IIII FF, as it was usual before Hadrian and the later ones stamped LEG IIII FL have been found together. The information in Ptolemy (III.9.3) about the legio IV Flavia in Singidunum cannot be dated exactly, but even so it attests the presence of a legion there before the end of the C2nd. It could reflect the situation immediately after Trajan, or later, but not after Marcus Aurelius.

II. New defensive lines must be sought in the large plain of Banat north of the Danube which was often threatened by Sarmatian tribes. Scattered finds in the vast plain, Roman monuments and ruins of the forts there, as well as inscriptions and bricks stamped by military units, do not help in determining the Roman policies, if considered in isolation. But if connected together, following rivers valleys and the possible directions of roads and passages across the plain of Banat and arranged chronologicaly, they allow us to recognize more than one line of defence, and not just along the Roman roads known from the itineraries, but also in other directions. The fact that most of them do not belong to the same time, enables us to try to use them in reconstructing possible Roman policy north of the middle Danube at different times, beginning with Hadrian. The following directions of communication are to be supposed: 1. The line marked by the forts to the south of the plain, along the Danube’s left bank, from the river passage near Banatska Palanka to the bridge constructed by Trajan between Drobeta and Pontes is nearest to the former limes of Moesia Superior. But it is not the earliest one following the subjugation of the Dacians. Sites are known in front of the forts on the south side of the river, as at Banatska Palanka, opposite to Lederata, Pojejena de sus vis-a-vis Pincus, Gornea and Sviniţa fronting Taliatae on the right Danube bank and Dubova. To this group of forts could be added Moldava nova, some distance from the Danube, known as the findspot of Roman monuments. Banatska Palanka is known because of bricks produced by the auxiliary units ala II Pannoniorum, cohors I Cretum, cohors II Hispanorum, and by two Upper Moesian legions, IIII Flavia (IIII FF) and VII Claudia, the latter with many types, such as VII CP, VIICL PS, LEG VII SC ADVENTINI, VII CL SC EUF PF BESSIO.5 In Pojejena

The archaeological material and inscriptions, although not abundant, make it possible to suggest that there was a remarkable difference between the situation in the northwest corner of the Moesia Superior, on the Danube between the embouchure of the Sava and that of the Mlava with its two legionary camps, Singidunum and Viminacium and that on the Danube bank east of Viminacium, in the Djerdap Gorge and further downstream at the time after the subjugation of the Dacian kingdom. It seems resonable to suggest that the legionary camp at Viminacium and probably that of Singidunum existed continously from the middle of the C1st and after the conquest of Dacia until the Severan era and later. On the 2

See Petrović & Vasić 1996: 20. Bojović’s 1996 dating of building phase II of the castrum in the long period from Nerva to the C4th is of no help in discussing the question of the early military occupation of this site. At least two different building phases supposed in this period of time are marked by the legionary bricks stamped, IIII FF and IIII Fl. 3 Until an unknown fort south-west of the legionary camp on the hill Calemegdan was discovered in 1989; M. Popović 1998: 7ff.; Mirković 1999: 108ff.

4

For dating the construction periods in Taliatae, see V.Popović 1984: 265ff. IDR III. (1977): 34ff, no. 5 ala II Pannoniorum, no. 6 cohors I Cretum, no. 7; cohors II Hispanorum, legio IIII Flavia, legio VII Claudia, nos. 8 – 5

758

Miroslava Mirković: Deserted forts – the Moesian limes after the conquest of Dacia have been found altars devoted by a praefectus cohortis and a praefectus cohortis V Gallica, many funeral monuments and bricks stamped by the legions IV Flavia and VII Claudia and cohors V Gallorum. (IDR III 39 nos. 10, 11, 21, 22, 22a, 23). Gornea has produced bricks with stamps LEG VII CL, SC BUBALI PP LEG VII CL MVIT and those of equites sagitarii (IDR III 55ff, nos. 30-32). At Sviniţa have been found bricks stamped by Hermogenes, SC HERMOGENI PP RI LEG VII CL PART CIT (IDR III 61, nos. 33-34), all from the C4th and later. In Dubova have been discovered many bricks with stamp DRPDIERNA - D(acia) R(ipensis) Dierna (IDR III 63). Most of the auxiliary units attested by brick stamps are those which took part in the Trajan’s campaigns against Decebalus6; the simple legion stamps of the legions IV Flavia, LEG IIII FF or VII Claudia LEG VII CL suggest an early date: not later than the beginning of Hadrian’s reign. The stamps bearing the name of a praefectus as well as those with the stamp DRPDIERNA - D(acia) R(ipensis) Dierna - belong to the later empire, mostly to the C4th. The only monuments to be dated to the C2nd are two altars dedicated by the praefecti coh. V Gallorum at Pojejena. There is reason to believe that the V Gallorum cohors was under the command of the legio VII Claudia in Viminacium after the mid-C2nd. The standards of weights in the cohort was regulated after that in the legio VII Claudia, as testified in the inscription on hemina found in the probable camp of the cohors V Gallorum at Trandrierna (Mirković 1980: 173ff.). If so, the troop on the opposite side of the Danube might be part of the Moesian army. There is no reason to believe that a defensive line was established close to the former limes of Upper Moesia, but that it was independent of it at the time immediately after Trajan’s Dacian campaigns. If some of forts on the left Danube bank were garrisoned in the C2nd, they must have been part of the defensive system in Moesia Superior, not of Dacia.

Aurelius were not discovered along this road, but on the one running to the east of it, from Dierna to the north, to Tibiscum.9 3. The findspots along the road linking Dierna with Tibiscum bear material dating from Antoninus Pius to Marcus Aurelius and the Severan dynasty. The Peutinger Table notes the following stations on this road: Dierna Ad Mediam - Praetorio - Ad Pannonios - Gaga Masclianis. Most of the inscriptions known until now have been found in Baile Herculane (Ad Mediam) and in Mehadia (Praetorium), among them an inscription dated in 157, and inscriptions vowed to Marcus Aurelius and the Severi (Baile Herculane: IDR III 57 ( = CIL III 1564), 58 (= CIL III 1565), IDR III 77). 4. Communications from Partiscum (Szeged) alongside Marisus (Mureas), to Apulum is confirmed by the discovery of stamped bricks. This route is marked by finds of legionary bricks at Cenad, Sinnicolau Mare, Arad and Bulci10, all with the stamp of the legio XIII Gemina, excepting one of the IV Flavia found in Arad (IDR III 242). The connection to Apulum as the legionary camp of the legio XIII Gemina must have been important. The road directed further to the north, from Napoca to Potaissa was constructed in 108, many decades before the coming of legio V Macedonica to Napoca (CIL III 1627). It is possible to suppose two further defensive lines in the south of Banat: 5. The ruins of an auxiliary fort at Dalboset and the archaeological finds from Bozovici, Petnic and Prigor suggest the existence of a communication line connecting Lederata, Pojejena and Praetorium (Mehadia) on the road from Dierna to the north. Building activity by the coh. III Delmatarum in the Nera valley is suggested a brick stamped with C III D from Petnic (IDR III 121f., n.103). It is dated to the C3rd (Wagner 1938: 133).

2. The forts at Arcidava, Centum Putei, Berzobis, Aizis, Caput Bubali, Tibiscum on the line running from Banatska Palanka to the north, known as stations on the road from Lederata to the Dacian capital Sarmizegetusa, must have been built during Trajan’s preparations for the expedition against Decebalus.7 The most important discovery was the legionary fort at Berzobis constructed by the legio IIII Flavia (IDR III 134ff.; Syme 1971: 10f.; Piso 1993: 6f.). The only inscription dedicated to Trajan, in 108, after the conquest of Dacia, was found at Vršac, not far from the road (IDR III n.106 (= CIL III 6273)). There is no proof that this road was in use after Trajan.8 Those epigraphic monuments dating from the time of Hadrian and Marcus

6. Starting from Acumincum (Slankamen) near the embouchure of the Tisa into the Danube two communications routes appear to lead to Dacia, that in the valley of Tibiscus, the other one following the River Berzovia to the forts at Berzobis and Tibiscus. The latter route is marked by the discovery of a brick stamped with IIII FF at Denta (IDR III 131f.). Roman inscriptions from Acumincum which date from the time of Caracalla’s reign suggest a later use of this route (CIL III 3252, 3256, 3258, 10197. cf. Dušanić 1967: 202ff.). III. The chronology of different defensive lines in Banat could be based on the dated inscriptions or bricks. The

9; leg. VII Claudia. 6 ala II Pannoniorum: Wagner 1938: 61; cohors I Cretum took part in the construction of the bridge between Pontes and Drobeta together with the cohors II Hispanorum - Wagner 1938: 127. 7 Tab.Peut. Seg.VI. On road construction at the time of Trajan’s preparation for the war, see Patsch 1937; Syme 1938: 207ff.; Mirković 1976b: 220ff. 8 cf. Protase 1967: 341, n.6, basing his conclusion on his own archaeological excavations. For further bibliography see IDR III.134.

9

Tab.Peut. Seg. VII. For the Roman inscriptions and other monuments on this line, see IDR III .63ff. IDR III 242ff. Bulci : LEG XIII G; Arad: LEG IIII F; Sinicolau Mare: LEG XIII G AVR ENT, LEG XIII G AVR GIDE; Cenad: LEG XIII G, LEG XIII G AVR ENTIM, LEG XIII GEM FLAVI MARTINVS, LEG XIII G VLP FRON. Most of these bricks must be dated at the time of later empire.

10

759

Limes XVIII

history of the legions IV Flavia, VII Claudia, XIII Gemina and some of the auxiliary units could be of help here.

appears in the text of a funerary monument from Cuppae (Spomenik Srpske Akademije Nauka 98, 36 n.85). The names of the legio XIII Gemina and V Macedonica are attested bricks from Baile Herculane (Ad Mediam) on the road connecting Dierna and Tibiscum. Bricks of the XIII Gemina have not only been found at Dierna on the northern Danube bank (IDR III 73), but also in the Banat, on the River Mureş, in Cenad, Sinnicolau Mare, Arad Nou and Bulci (IDR III 242-246). The legio V Macedonica belongs to the time after 167, when it was transfered from Troesmis in Moesia Inferior to Potaissa in Dacia Superior (Ritterling 1924: 1578ff.). Stamped bricks bearing its name have been found at Mehadia, and elsewhere in Dacia (IDR III 119, nos. 100-102. cf. Ritterling 1924: 1580).

Bricks marked by the IV Flavia with the stamp IIII FF belong to the very early stage, and should not be dated after Trajan. The established list of the findspots11 marked Trajan’s route to the Dacian capital and were found between Lederata and Berzobis, not only on the road line (Palanka, Cetum Putei, Berzobis) but also west from it, in Denta (IDR III 134.) and east of it, in Bocşa Romonâ (IDR III 134) and in Tibiscum, as well in Sarmizegetusa and the Roman fort in the Dacian capital at Grădiştea Muncelului (I. Glodariu 1966: 327-329; Benea 1983: 154 and n. 71). Bricks marked by legio IV Flavia were discovered in Pojejena de Sus, not far from Palanka (IDR III 49f.: LEG IIIIF and LEG VII CL.). The gap between Pojejena and Dierna is remarkable. Drobeta, opposite Pontes, on the other site of the Danube, is again known as a findspot of bricks marked IIII FF (cf. IDR III 242 PROV). On the line north of Dierna, leading to Tibiscum, the presence of the legion or its detachments is indicated by bricks found at Baile Herculane (Ad Mediam) and in Mehadia (Praetorium: IDR III 98: LEG IIII FF, LEG V, LEG VII CL, (LEG) XIII (G?)), at the junction of two roads: one known from the itineraries, connecting Dierna and Tibiscum and the other supposed in the Nera valley. Significant is the lack of any trace of this legion north of this point. The discovery of the brick stamped IIII FF in Arad Nou on the River Mureş, one of the possible boundaries of Dacia could indicate the legion’s building activity there.

The findspots of dated inscriptions on the north of the Danube are significant. The only inscription dedicated to Trajan was found in Vršac (CIL III 6273), not far from the road to Berzobis used by the emperor and the troops as a military base in the 1st Dacian War. In Baile Herculane, on the road leading from Dierna to the north altars dedicated in 157, 160 and to emperors of Severan dynasty have been discovered.13 It has been supposed that the road and defensive line from Lederata via Berzobis to Tibiscum and Sarmizegetusa was abandoned soon after the conquest of Dacia, in 117 or 118 (Protase 1967: 341). There is no doubt that the second one, from Dierna, via Ad Mediam and Praetorium to Tibiscum, constructed for the Trajan’s second campaign grew in importance in the time of Antoninus Pius and his war on the western Dacian border (Ritterling 1924: 1294-1295). It is from this time that the defensive line of Dacia to the south-west should be sought on this line.

It is known that the IV Flavia took part in the Trajan’s campaigns and was left in Dacia until 117. Thereafter it was despatched to Singidunum in Moesia Superior.12 To the time after its transfer to Singidunum probably belongs brick production stamped IIII F F, found at Singidunum and between it and Lederata, at Pančevo and Kovin (CIL III 6326, 8276, 1-2) on the left Danube bank.

To sum up: it seems possible to establish the following chronology of the defensive lines or communications in Banat, based on the findspots of the stamped bricks and inscriptions: To the earliest time belongs the road from Lederata to Berzobis; the defensive line on the Mures must have been created soon after the annexation of the Dacian kingdom, as is shown by the discovery of the bricks of the IV Flavia and XIII Gemina; the defensive system connecting Dierna and Tibiscum could be dated to a later period, to the time of Pius and Marcus Aurelius; not earlier than the Severi is the connecting of Pannonia and Dacia by the line which leads from Acumincum to Tibiscum; at the same time the old passage across the Danube near Drobeta grew in importance.14 Acumincum might have been of military importance at the time of Domitian’s wars on the Pannonian frontiers or later, as could be inferred from the passage in Ptolemy (II.15.3) mentioning the legion at this site. The defensive system following the left Danube bank, and probably a part of the upper Moesian limes, cannot be

Bricks stamped by the legiones VII Claudia and the XIII Gemina are not numerous in the region. Bricks produced by the VII Claudia are less numerous, except those bearing the stamp s.c. p.p. legionis VII Claudiae. They are concentrated on the Danube bank and have to be dated to the later empire. To an earlier period belongs the simple stamp LEG VII CL found in Pojejena and north of Dierna (IDR III 49 (Pojejena) and 71f. (Dierna)). In the C2nd, under the command of this legion, were some of the auxiliary units, attested on bricks and inscriptions from the south Banat plain, cohors V Gallica, which was located probably in Transdierna and cohors III Campestris which 11

See Glodariu 1967: 330: Aradul Nou, Baile Herculane, Berzobis, Covin, Denta, Tibiscum, Mehadia, Palanka, Pancevo, Pojejena, Sarmizegetusa, Centum Putei, Drobeta; cf. Protase 1967: 337ff. Benea 1983: 156ff. adds some new findspots, as Bosca Romana, Patas, Surduk, Micia, Apulum, Eftimie Murgu. 12 On the legio IV Flavia in Dacia, see the still valuable Ritterling 1924: 1279; Syme 1971: 84ff.; Glodariu 1967: 34; Protase 1967: 337-342; Benea 1983: 159ff., 178ff., 198ff.

13

CIL III.1568 (IDR III.84, n.60), 1573 (IDR III.89, n.66, 1575 (IDR III.94, n.70, 1565 (IDR III.81, n.57, 1565 (IDR III.83, n.58), 1569 (IDR III.84f, n. 61). 14 An inscription dedicated to Caracalla by Marius Perpetuus, governor of the Dacian provinces was found in Transdrobeta; cf. Mirković 1979: 417418. For his career, cf. PIR2 M 308, 205ff. Thomasson 1984: 157, n. 53. It corresponds to the time of a renewal of the forts in Djerdap.

760

Miroslava Mirković: Deserted forts – the Moesian limes after the conquest of Dacia dated before the second half of the C2nd. It grew in importance in the C4th and later.

one province on the middle Danube, in tres Daciae and Moesia Superior (PIR2 C 874; Thomasson 1984: 127, 151, 154, n.37; Piso 1993: 82-94), as is seen from an inscription from Rome: M.Claudio [Ti.] f. Q[uir] Frontoni cos., leg. Aug. pr.pr. provinciarum Daciarum et super. simul leg. Aug. pr. provincia Daciar., leg. Augg. pr.pr. Moesiae Super. Daciae Apulesis simul leg. Augg. pr.pr. provinciae Moesiae super. ..etc. (ILS 1098). The sequence of his appointments, as J. Piso arranged it, is as follows: in 168 leg. pr.pr. Moesiae Sup. in 168-febr. 169 leg. pr. pr. Moes. Sup. et Daciae Apulensis simul, in 169-170, leg. pr. pr. provinciar. Daciar. 170, leg. Aug. pr. pr. provinciarum Daciarum et Moes. Sup. simul (Piso 1993: 97). Significant in this case is the link Dacia - Moesia Superior, neglected until Marcus Aurelius. At the north-west corner of the province, Singidunum was left without a garrison and Fronto's military command in Moesia Superior must have been limited to Viminacium and the fortresses on the east of it. It could mean that control over the Jazygi was left to the governor of Pannonia, to whose command was added the legio IV Flavia from Singidunum16, and that Fronto's governorship in Moesia Superior, left with only one legion, (VII Claudia at Viminacium), was extended to two legions in the province of the united tres Daciae. It is to be supposed that by this time had begun the renewal of the forts to the east, from Viminacium, before all those in Lederata, in Transdierna or in Pontes, as well as at points of passage across the Danube and the starting point of roads leading into Dacia. The cohors V Gallica in Transdierna and III Campestris at Cuppae might be dated to this period.

Significant for Roman politics in Dacia and on the middle Danube under Hadrian and later, under Marcus Aurelius, were the extraordinary appointments and the careers of two prominent personalities, Marcius Turbo, the friend of Hadrian, and Claudius Fronto, general under Marcus Aurelius. To these could be added Iulius Pompilius, a commander of different legions in 70s of the C2nd. As the emperor’s friend Marcius Turbo was appointed governor both of Dacia and Pannonia in 118 with the duty to survey the northern frontier of the empire. His career and the chronology of his numerous honors have been the subject of much discussion.15 As a man of equestrian rank, he was left in Dacia by Hadrian with the command of two legions, with a rank commensurate with that of the prafectus Aegypti, according to the SHA (7.3): Dacia Turboni credita titulo Aegyptiacae praefecturae, quo plus auctoritatis haberet, ornato; and 6.7: Marcium Turbonem praefecturae infulis ornatum Pannoniae Daciaeque ad tempus praefecit. For how long he remained in this post is uncertain. The auxiliary troops from both provinces under his command were granted citizenship and conubium in 123 when he advanced to be praefectus praetorio in Rome (RMD I.21). His appointment as commander in two provinces reveals in what area the Roman frontier on the middle Danube was vulnerable in 117-118. Marcius Turbo's command concerned Dacia and Pannonia, divided by the plain of Banat and not Dacia and neighbouring Moesia Superior, as might be expected. This would mean that the Pannonian border and the north-west Dacia were exposed to Sarmatian attacks after Trajan’s death. As Marcius Turbo in Dacia had to co-operate with Pannonian troops, it is to be expected of that communications were improved between the two provinces. It is suggested that the Roman army used the existing route along Marusius (Maroš). Bricks stamped IIII F and others marked by the legio XIII Gemina indicate building activity by these legions on this route or defensive line at the start of Hadrian’s reign or even before. The other possible line of communication could run more to the south, connecting Acumincum on the Tisa’s embouchure with Tibiscum alongside the River Tibiscus or alongside Birzava to Berzobis. However, it must be admitted that the known Roman inscriptions from Acumincum belong to the Severan period.

The former link of the military command in Dacia with that of Pannonia at the start of Hadrian’s reign was replaced by connecting the Dacian governorship with that of Moesia Superior under Marcus Aurelius. This could indicate the beginning of a retreat to the Danubian border to the empire. The presence of the emperor at Viminacium in 17517 might be linked with the reorganisation of the limes in Upper Moesia. The legio IV Flavia, which has been moved from Singidunum to Pannonia Inferior in 164, probably returned to its Upper Moesian camp. The career of Iulius Pompilius Piso T.Vibius (Varus?) Laevillus Berenicianus, senatorial commander of many legions is significant for dating the return of that legion. Attested in an inscription from Lambaesis, (ILS 1111), dated 177, he joined the command of the legio IV Flavia with that of the Dacian XIII Gemina at first, then with I Italica stationed at Novae in Lower Moesia. He was legatus leg. XIII [Geminae item IIII Flaviae] praepositus legionibus I Italicae et III[I Flaviae cum omnibus copiis] auxiliorum dato iure gladii. His command of the legiones XIII Gemina and IIII Flavia might be connected with a war in Dacia soon after 170.18 It would be difficult to have the single command of these two legions if the former was still in

The extraordinary career of Claudius Fronto at the time of Marcus Aurelius reveals the changes in the Roman politics on the western border of Dacia and in the defence from attacks of Sarmatian tribes, as well as changes in the military occupation along the upper Moesian Danube. This prominent politician had joint command over more than

16

leg. IV Flavia in Pannonia Inferior 164: Mirković 1994: 355ff. See Plassart 1970: 127, no. 326 and p.123, no. 323, the emperor’s letter to Delphi signed in Viminacium, cf. M. Mirković IMS II. 23. 18 PIR2 J 477, p. 252: quae fortasse in Dacia bellum gesserunt, paulo post annum 170. 17

15 On Marcius Turbo, see PIR2 M 249, 189ff.; Pflaum 1960: n. 94, 199216; Alföldy 1979: 233ff.; Eck 1993: 247/255; Thomasson 1984: 112; Piso 1993: 30.

761

Limes XVIII

Pannonia, the latter in Moesia Inferior. Rather he could have been appointed commander of two legions belonging to two provinces neighbouring to each other and to Dacia.

returned to the policy of defending the empire as it was before the annexation of the Dacian kingdom, renewing the forts in Novae, Taliatae, Diana and Pontes. Caracalla also rebuilt forts on the Pannonian part of the Danube, among others probably that at Acumincum.

Some changes might have even occured between Hadrian and Marcus Aurelius. The reign of Antoninus Pius was also marked by a war in Dacia, but evidence about it is very scanty. It was suggested that the war took place in Dacia Superior.19 An altar from Baile Herculane (Ad Mediam?) on the road from Drobeta to Tibiscum dated to 157 could indicate the return to the fortified line in western Dacia (CIL III 1568 = AE 1960: 343 = IDR III 60).

Other points to be considered are the real Roman occupation of the plain of Banat and the establishment of the western boundary of Dacia. It can be pointed out at first that all Roman findspots in Banat are situated on the southern and eastern fringes of the plain; no Roman town, forts or important settlement have been discovered between the line marked by the Lederata – Berzobis road and the River Tisa, as well as north of Tibiscus, between that river and Marisius. The explanation should be sought in the fact that this region was swampy and not inhabited. Amelioration works date from the time of Maria Theresa’s reign (in the C18th). It seems possible that this zone was outside the Roman frontier, but not necessarily beyond Roman control. It was overseen by Roman forces from Dacia, Pannonia and from Moesia Superior. The only two Roman roads leading from the south to this region are those presented in the Peutinger Table (Seg. VII), following the eastern edge of the Banat plain, on a line leading from Lederata or Dierna to Tibiscum and further to Apulum and Sarmizegetusa. Other communication routes along river valleys might not have been constructed as roads but served probably as tracks.

Roman policy on the middle Danube from the time of Domitian’s Danubian wars to Marcus Aurelius can be summarised as follows. Until the time of Domitian’s Danubian wars the main forts on the Danubian frontier in Upper Moesia, from the mouth of Mlava, to the exit of Iron Gates, were located in Viminacium, at Novae (Čezava) at the entrance in the Djerdap Canyon, in Boljetin in the middle of the Gornja Klisura and probably in Taliatae and Transdierna. Domitian, as it was supposed, brought to the Danube five new legions at the time of his Sarmatian campaign and invoked important changes to the Pannonian and Moesian garrisons. Trajan concentrated a large number of forces on the right Danube bank at the eve of the war against Decebalus. Forts he used as starting points in the first campaign were Viminacium and Lederata. In the second war that position was taken by the new fort Pontes, in front of Drobeta on the northern Danube bank.20 The immediate consequence of the Roman victory over Decebalus was the creation of the new province of Dacia. The organization of the military command on the middle and lower Danube thereafter is not known in detail. Trajan’s solution must have not been definitive, because he needed Danubian forces for his Parthian war. Hadrian had to solve the problem. He abandoned the forts in Moesia Superior in Djerdap, from Novae to Pontes soon after Trajan’s death and entrusted the defence of Dacia from Sarmatian raids 117-118 to the army of two provinces, Dacia and Pannonia Inferior, joined together, by establishing a temporary joint command. The forts in Upper Moesia east of Viminacium were abandoned for most of the C2nd. The first step in returning to the Danube bank east of the legionary fort at Viminacium as a line of defence from the north might have been made during a war of Antoninus Pius on the north-western frontier of Dacia. The first emperor after Trajan to return to the system of defending Dacia with the army of Moesia was Marcus Aurelius. He could have entrusted temporarily the defence of Dacia to the governor of Moesia Superior and he probably made further efforts to establish the old system again, with a fortified line on the Danube in Moesia Superior. Septimius Severus or more probably Caracalla,

Bibliography Alföldy G. 1979 Marcius Turbo, Septimius Clarus und die Historia Augusta. Zeitschrift der Papyrologie und Epigraphik 36: 233ff. Benea D. 1983 Din istoria militara a Moesiei Superior si a Daciei. (Cluj/Napoca). Bjelajac Lj. & Ivanišević V. 1991 Les temoignages archéologiques des grandes invasions a Singidunum. Starinar 42: 123-141. Bojović D. 1996 Le camp de la légion IV Flavia à Singidunum. In P. Petrovic (ed.) Roman Limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 53-68. Domaszewski A.v. 1890 Studien zur Geschichte der Donaupprovinzen, I. Die Grenzen von Moesia Superior und der illyrische Grenzzoll. Arch.-Epigr.Mitt. 13: 129ff. Dušanić S. 1967 Novi i revidirani natpisi iz istočnog Srema. Živa Antika 17: 202-215. Eck W. 1993 Q. Marcius Turbo in Niedermösien. In K.H. Dietz, D. Henning & H. Kaletsch (Hsrg.) Klasssische Altertum, Spätantike und frühes Christentum. Adolf Lippold zum 65. Geburtstag gewidmet. (Würzburg): 247-255. Garašanin M. & Vasić M. 1980 Le Pont de Trajan et le castellum Pontes, rapport préliminaire pour l’année 1979. (Trajanov most - kastel Pontes). In V. Kondić (ed.) Djerdapske sveske. Cahiers des Portes de Fer II. (Belgrade): 25-41. Garašanin M. & Vasić M. & Marjanović-Vujović G. 1984 Pontes – camp et pont de Trajan. Fouilles de 1980 (Trajanov most – Castrum Pontes). In V. Kondić (ed.)

19

Ritterling 1924: 1294ff. An altar to Victoria Augusta pro salute Imp.Antonini was vowed by the governor of Dacia Superior, Statius Priscus in 157 (CIL III.1416). 20 On the changes in Roman policy in this part of the Danubian frontier see Mirković 1996: 27ff.

762

Miroslava Mirković: Deserted forts – the Moesian limes after the conquest of Dacia Djerdapske sveske. Cahiers des Portes de Fer II. (Belgrade): 41-85. Glodariu I. 1966 Acta musei Napocensis III: 327-329. Glodariu I. 1967 Legio IV Flavia felix et la Dacie. In Acta of the Fifth International Congress of Greek and Latin Epigraphy. (Oxford 1971): 327-336. Kondić J. 1980 Les fouilles de Karataš (Iskopavanja na Karatašu). In V. Kondić (ed.) Djerdapske sveske. Cahiers des Portes de Fer I. (Belgrade): 51-69. Kondić J. 1984 Statio Cataractarum Diana. In V. Kondić (ed.) Djeradpske sveske. Cahiers des Portes de Fer II. (Belgrade): 7ff. Kondić J. 1987 Statio Cataractarum Diana. In V. Kondić (ed.) Djerdapske sveske. Cahiers des Portes de Fer IV. (Belgrade): 16-38. Mirković M. 1976a Inscriptions de la Mésie Supérieure. (Beograd). Mirković M. 1976b Cohors I Montanorum u Gornjoj Meziji. Arheoloćki vestnik (Ljubljana) 26: 220-224. Mirković M. 1977 L. Marius Perpetuus, consularis. Živa Antika xxvii: 443-448. Mirković M. 1979 L.Marius Perpetuus, consularis. In M. Pippidi (ed.) Actes du VIIe Congres intern. d’épigraphie grecque et latine, Mamaia 1977. (Bucuresti): 217-219. Mirković M. 1980 Cohors V Gallorum u Transdierni. Arheološki vestnik (Ljubljana) 31: 173ff. Mirković M. 1994 Beneficiarii consularis in Sirmium. Chiron 24: 345-404. Mirković M. 1996 The Iron Gates (Djerdap) and the Roman policy on the Moesian limes AD 33-117. In P. Petrović (ed.) Roman Limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographie 2 (Belgrade): 27-40. Mirković M. 1998 The Legio VIII Augusta in the Balkans. Novae and the Romans on Rhine, Danube, Black Sea and beyond the frontiers of the empire. Acta of the International Conference Wykno, Poland, 1995. (Warszawa). Novensia 10: 89-97. Mirković M. 1999 Porta praetoria semper hostem spectare debet. In M. Vasić (ed.) Le Djerdap /Portes de Fer a la deuxieme moitié du premier milenaire av. J.Chr. jusqu’aux guerres daciques, Kolloquium in Kladovo Drobeta - Turnu Severin 1998. (Beograd): 108-114. Patsch K. 1937 Der Kampf um den Donauraum unter Domitian und Trajan, Beiträge zur Völkerkunde von Südosteuropa V/2, Akademie der Wissenschaft in Wien, Philos.-hist. Klasse, Sitzb. 217 Bd.1. (Wien). Petrović P. & Vasić M. 1996 The Roman frontier in Upper Moesia: archaeological investigations in the Iron Gates area - main results. In P. Petrović (ed.) Roman limes on the Middle and Lower Danube. Cahiers des Portes de Fer, Monographies 2. (Belgrade): 15-26. Pflaum H.-G. 1960 Les carrieres procuratoriennes équestres sous le Haut-Empire romain. (Paris). Piso I. 1993 Fasti Provinciae Daciae I. Die Senatorischen Armtstzäger. Antiquitas Reihe 1 Bd. 43. (Bonn). Plassart A. 1970 Fouilles de Delphes III, Fasc. 4. (Paris). Popović M. 1998 Antički Singidunum, dosadašnja otkrića i mogućnost daljih istraživanja. Singidunum I (1997): 7 ff. - with summary. Popović V. 1984 Donji Milanovac - Veliki Gradac (Taliata), rimsko i ranovizantijsko utvrdjenje (res. fr.: Donji Milanovac - Veliki Gradac (Taliata), forteresse romano-byzantine). Starinar 33/34: 265-282. Protase D. 1967 La legion IV Flavia au nord du Danube et la premiere organisation de la Dacie romaine. Acta of the Fifth International Congress of Greek and Latin

Epigraphy. (Oxford 1971): 337-342. Ritterling E. 1924 Legio. Paulys Real-Encyclopädie der classischen Alterwißenschaft 24 (Stuttgart): Cols. 1086-2828. Strobel K. 1984 Untersuchungen zu den Donaukriegen Trajans. (Bonn). Swoboda E.v. 1939 Forschungen am obermöesischen Limes. Schriften der Balkankommission, Antiquarische Abteilung X. (Wien). Syme R. 1938 The first garrison of Trajan´s Dacia. Laureae Aquincenses I: 267ff. (= Danubian Papers. (Bucharest, 1971): 84-110). Syme R. 1959 The lower Danube under Trajan. Journal of Roman Studies 49: 26ff. (= Danubian Papers. (Bucharest, 1971: 122-135). Thomasson B. 1984 Laterculi praesidum I. (Götteborg). Vasić M. 1984 Čezava - Castrum Novae. Starinar 33/34, 19821983; 25ff; 91ff. Vasić M. 1987 Čezava - Castrum Novae. Ratiarensia 3-4: 131134; 711. Vasić M. 1990 Čezava - Castrum Novae. La stratigraphie, la chronologie et les phases architectoniques. In H. Vetters & M. Kandler (edd.) Akten des 14. Internationalen Limeskongresses 1986 in Carnuntum II (Wien): 897-911 Wagner W. 1938 Die Dislokation der römischen Auxiliarformationen in den Provinzen Noricum, Pannonien, Moesien und Dakien von Augustus bis Gallienus. (Berlin). Zotović Lj. 1984 Boljetin (Smorna), rimski i ranovizantijski logor. Starinar 33/34: 211ff.

763

Limes XVIII

Fig. 1 Roman Roads in Moesia Superior.

Fig. 2 Roman Roads from Moesia Superior to Dacia and possible communications in the plain of Bauaf.. 764

The military equipment of oriental archers in Roman Dacia Liviu Petculescu Roman Dacia had a large garrison consisting mainly of auxiliary units among which many were recruited from the Orient. As many of the forts in Dacia occupied by oriental troops have been partially excavated – eg. Drobeta, Tibiscum, Micia, Porolissum, Jidava – one can make some observations about the equipment of this class of auxiliaries.

The presence at Micia in only one context of very differently shaped and sized ear laths suggests that in the Roman army and even in the same military unit simultaneously bows of different shape and width of the stave were used. If the variations in shape of the ends of the ear laths had an ornamental meaning the curving and especially the width of the laths, of the limbs respectively, determined the stiffness of the bows, that is their propulsion power. This means that the bows with the narrowest staves were less rigid and shot arrows with greater speed but a smaller striking power than ‘normal’ bows. When used with light arrows they were fit for the hunting of small animals, especially birds.

The forts garrisoned by orientals offers plenty of evidence of archery equipment: bow laths, bone nocks of arrows, arrowheads. In the fort of Micia (Veţel, Hunedoara County) was excavated a store dated c.106-170 containing bow laths, bone arrow nocks and antler debris or waster (unpubl. excavations of the author). The laths are made from antler with the exception of a bone grip lath. From the debris two pieces are of bone and the rest of antler. None of the finished ear laths are entirely preserved. All are broken at the lower end and some at the upper end. Many of them were burnt together with the buildings of the Phase I of the fort and are bent. Among the 35 ear laths which are illustrated (Figs. 1-3, nos. 1-35) only 21 have the inner edge entirely preserved allowing their typological identification (nos. 1-21). Most have the upper end rounded but three of them have angular ends and two have square ends. The majority of the laths with rounded tips have U-shaped nocks; two have semi-circular nocks (nos. 12, 14). The laths with angular and square ends have, but for one (no. 20), shallow semi-circular nocks. Some of the laths are heavily curved (nos. 7, 20, 27) and others are almost straight (nos. 3, 21-23).

In the store from Micia were also found two grip laths. One is long and narrow and was applied on the transversal face of the grip (no. 36). The second lath with one straight side and the other arched was placed laterally on the grip of the bow (no. 37). The transversal grip laths even if much rarer than the ear laths are still usual among Roman bow finds. On the contrary the lateral grip lath has close parallels in the steppes of central Asia but are met only sporadically in Roman contexts and so it appears as a direct eastern influx (Simonenko & Lobai 1991: 13, fig. 6. 1-4, esp. nos. 2 and 4 from Porogi and western Siberia). In the same store were found unfinished items and antler debris or wasters from which I illustrate a selection of representative pieces (nos. 38-52). The association of the finished and unfinished laths and wasters indicates that the store belonged to a workshop specializing in the production of composite bows.

I think that the laths with angular and square upper ends were intended to be applied to the lower limb of the stave which would rest on the ground when the string was put on the bow. In support of this assumption, the early Hunnic bow from Qum Darya (Sinkiang) has the short ear laths, the only ones preserved, with square upper ends (Coulston 1985: 248, fig. 4). However the Parthian bow from Yrzi on the Euphrates, dating between the C1st BC and C3rd AD has both sets of ear laths with rounded upper tips (Coulston 1985: 248, fig. 2). At Micia the number of rounded laths is nearly four times more than the pieces differently shaped and a similar situation is met with the other finds of bow fragments in Dacia and the whole Roman empire. Consequently it seems to me that the end of the upper limbs of the Roman bows were always, or at least in most cases, rounded, and the tip of the lower limbs also rounded or less frequently angular or square.

Also at Micia, in the first ditch of the stone fort was discovered an almost straight ear lath (no. 53). In the 7.2ha fort at Micia were simultaneously quartered three units: cohors II Flavia Commagenorum from the time of Trajan but attested only during Hadrian’s reign; ala I Hispanorum Campagonum, from the first half of the C2nd AD and Mauri Micienses = numerus Maurorum Miciensium at the latest from 204 (Petolescu 1977. 369 with refs.) As cohors II Flavia Commagenorum was equitata sagittariorum it is almost certain that its soldiers produced and used the composite bows whose pieces were found in the workshop store and elsewhere in the fort. In the fort and the civil settlement at Tibiscum (Jupa, Caraş-Severin County) were unearthed five laths, among which four ear laths (nos. 54-57: Bona et al 1983: 417-8 nos. 1, 4-5, pl.11/1, 2, 11; Petrescu & Rogozea 1990: 117, no. 2, pl. 11/5) and one transversal grip lath (no. 58, Benea 1983, fig. 2/2). Tibiscum was garrisoned by at least two archer units: cohors I sagittariorum milliaria equitata and numerus Palmyrenorum Tibiscensium (Petolescu 1997: 109-110, 124-125 with refs.). As the former unit was

The ear laths from Micia are differently sized. If their fragmentation does not allow us to estimate their original lengths it is obvious that their widths are considerably dissimilar. From this point of view the most suggestive example is made by comparing nos. 18 and 19 which are virtually identically shaped but the first is nearly double in width than the latter. 765

Limes XVIII

transferred to Drobeta towards the end of the C2nd AD the three pieces dated in the C3rd AD (nos. 55-6; 58) were probably parts of the equipment of the Palmyrene archers.

Roman arrowheads of the C2nd-C3rd have been grouped in three main types: trilobate tanged (nos. 69-90); pyramidal tanged (nos. 91-105); pyramidal socketed (nos. 103-105: for the typology of the arrowheads see, Erdmann 1976: 5-10, 1982: 5-11; Zanier 1988: 5-27). Trilobate tanged arrowheads were spread over the whole empire by oriental archers and are the most numerous type, both in Dacia and in the other provinces. The arrowheads belonging to all the three types are remarkable different in shape, size and standard of work. The pyramidal tanged form seems to date mostly from the C3rd and the pyramidal socketed type to the second half of the C2nd AD and until the end of Dacia province.

Three ear laths from Apulum (Alba Iulia, Alba County: nos. 59-61) were found on the spot of the civil settlement (canabae?) and so their assignment to the equipment of the legio XIII Gemina is uncertain (Ciugudean 1997: 77 nos. 459-461, pl. 30/2-4). The ear lath from Romita (Sălaj County, no. 62) was used more probably by an oriental from cohors I IS (Ituraeorum Sagittariorum?) which was quartered in this fort at the beginning of the C2nd AD or perhaps by a soldier from cohors VI Thracum rather than one from cohors II Britannica (Matei & Bajusz 1997: 129, pl. 82/1; 67-95 for the military units quartered in the fort,) as the Thracians were also attested as bowmen but not the Britanni.

Examples of trilobate tanged and pyramidal tanged heads are often associated in the same forts: Jidava (nos. 69-71; 91-93: Popescu & Popescu 1970: 261, fig. 12/2), Tibiscum (nos. 75-77; 97-8: Bona et al 1983: 412, nos. 122, 14, pl. 11/8, 122-14), Micia (nos. 78-90; 102: unpubl. excavations by the author), Urluieni B (nos. 72-3; 94-96: BogdanCătăniciu 1994: 348, figs. 4/2; 13). In the first three forts were quartered oriental archers; the garrison of Urluieni B is unknown. Socketed heads are less frequent but they appear in the forts, as at Micia (nos. 103-4), which produced heads of other types. If one considers that in the same forts, for example once again at Micia, are found all the types of arrowheads and within these types the entire range of shapes, sizes and standards of work it is obvious that oriental archers used simultaneously arrows of different shapes, sizes and weight.

The last lath accessible to me was discovered in the fort at Urluieni (Argeş County, no. 63) whose garrison is unknown but should have consisted of auxiliaries (BogdanCătăniciu 1994: 348, fig. 14a). From the large amount of arrows which were used in Roman Dacia have survived a few bone nocks and an important number of metal heads. The bone nocks have a cylindrical body with a notch at the bottom for receiving the bow string and a narrow extension at the tip which was put in the shaft of the reed arrow. Thus the arrows were provided with a strong nock which should not split on the pressing of the bow string. The nocks are usually decorated with grooves or ribs and mouldings.

The variety of ‘ammunition’ used by the bowmen resulted from adapting to the targets to be hit. So two pieces from Micia (nos. 78, 81) weighing at present only 0.8 and 1.5gms respectively seem to me too small and too light to be the heads of fighting arrows, even counting with a certain loss of their material to corrosion. Consequently it is a reasonable assumption that the smallest and lightest heads together with bone nocks were parts of the arrows shot from light bows for hunting small animals.

The five Dacian bone nocks (nos. 64-68) originated in the forts of Micia (2 pieces: unpubl. excavations of the author) and Tibiscum (3 pieces: Bona et al 1983: 418, nos. 6-8, pl. 11/4-6) where there are attested Oriental archers and where bow laths were also unearthed. The Micia examples (nos. 64-5) came from a layer dated c.106-170 and the Tibiscum ones (nos. 66-68) from the hypocaustum Building A dated to the C3rd. Consequently light arrows with reed shaft and bone nock were used in Dacia during the entire period that this province existed. I know only one bone nock in the other Roman provinces, at Sebaste in Judaea (Reisner et al 1924: 374, no. 17b, fig. 241/J17b), but two more examples are known from the C1st AD ‘free’ Dacian hill-fort at Poiana (Petculescu & Nicu 2000: 211, nos. 11-12)..

N. Gudea discovered at Porolissum c.200 bone arrowheads which are still unpublished (Zanier 1988: 17, no. 88). If these heads date to the C2nd-C3rd and not a later period of time, they represent a rarity in the Roman empire (nb. G. Stiebel communicated at the Amman Congress a bone head from Masada). At the end of a discussion about bows and arrows one can make some general remarks. First of all one should stress that there are no typological differences between the laths from Dacia and those from the rest of the Roman empire. The contemporary Dacian examples are different in shape and size which is evidence for the use in the same period of time of differently shaped and sized bows. As expected, most of the laths were found in the forts where were quartered units of oriental archers, who, at least in one case (Micia) manufactured them on the spot. However the presence of the laths at Apulum and perhaps also at Potaissa, the headquarters of the legio V Macedonica (M.

Arrowheads are very frequent in Roman contexts but it is certain that many of them were meant to be shot from European bows and besides, even from the heads probably associated with composite bows, not all were part of the fighting equipment of the oriental archers. In the Vetera I legionary fortress were found no less than 2370 pieces; in Dacia the largest hoard of arrowheads consists of c.400 examples which was unearthed in the auxiliary fort at Jidava.

766

Liviu Petculescu: The military equipment of oriental archers in Roman Dacia Barbulescu pers. comm. apud Ciugudean 1997: 37, n.2) proves that, like in the other frontier provinces, the use of the composite bows was not restricted to sagittarii units.

1979). Acta Musei Napocensis 20: 405-432. Ciugudean D. 1997 Obiecte din os, corn şi fildeş de la Apulum. Biblioteca Musei Apulensis 5 Alba Iulia. Coulston J.C.N. 1985 Roman archery equipment. In M.C. Bishop (ed.) The production and distribution of Roman military equipment. Proceedings of the second Roman military equipment research seminar. British Archaeological Reports. Int. Ser. 275. (Oxford): 220366. Erdmann E. 1976 Dreiflügelige Pfeilspitzen aus Eisen von der Saalburg. Saalburg Jahrbuch 33: 5-10. Erdmann E. 1982 Vierkantige Pfeilspitzen aus Eisen von der Saalburg. Saalburg Jahrbuch 38: 5-11. Hica-Cîmpeanu I. 1982 Din colectiile Muzeului de Istorie al Transilvaniei – Donaţia Torma Karoly. Acta Musei Napocensis 19: 593-605. Matei A. & Bajusz I. 1997 Castrul roman de la RomitaCertiae. Das Römergrenzkastell von Romita-Certiae. Führer zu archäologischen Denkmälern in Dacia Porolissensis 4 (Zalău). Petculescu L. 1982 Obrăzare de coifuri romane din Dacia. Acta Musei Napocensis 19: 291-300. Petculescu L. 1983 Prinzătoare de teacă romane din Dacia. Acta Musei Napocensis 20: 451-466. Petculescu L. 1999 Fragments of helmets from the Roman fort at Jidava. Acta Musei Napocensis 36/1: 189-196. Petculescu L. & Gherghe P. 1979 Coiful roman de la Bumbeşti. Studii şi Cercetări de Istorie Veche şi Arheologie 30: 603-606. Petculescu L. & Nicu M. 2000 Echipamentul militar roman din cetatea Daciă de la Poiana (Jud. Galaţi). In M. Iacob, E. Oberländer-Târnoveanu & F. Topoleanu (edd.) IstroPontica. (Tulcea): 203-220. Petolescu C.C. 1977 L’organisation de la frontière romaine dans la Vallée du Mureş (Dacia superior). In Studien zu den Militärgrenzen Roms II. (Köln):367-372. Petolescu C.C. 1997 Die Auxiliareinheiten im römischen Dakien. Acta Musei Napocensis 34: 66-141. Petrescu M.S. & Rogozea P. 1990 Tibiscum – principia castrului mare de piatră (I). Banatica 10: 107-137. Popescu Em. & Popescu Eug. 1970 Raport preliminar asupra săpăturilor efectuate în anii 1962-1967 în castrul roman de lîngă Cîmpulung Muscel (Jidova). Materiale şi Cercetări Arheologice 9: 251-263. Reisner G.A. et al 1924 Excavations at Samaria. 1908-1010. I. (Cambridge, Mass.). Simonenko A.V. & Lobai B.I. 1991 Sarmaty SeveroZapadnogo Pričernomorja v I v.n.e. (Kiev). Zanier W. 1988 Römische dreiflügelige Pfeilspitzen. Saalburg Jahrbuch 44: 5-27.

The presence of bone nocks in Dacia attests a direct eastern influence probably because some of the archer units which settled in Dacia were recruited from the Orient. The light bows and light arrows seem to have been used for hunting small animals, especially birds but heavier bows could certainly have been used for hunting larger quarry. As I believe that at Micia the workshop manufactured both military and hunting bows it looks that part of the laths found in military contexts all over the empire should also belong to hunting bows. Therefore the widespread diffusion of the laths in the forts should not be considered as definitive evidence for the extensive use in the C2nd-C3rd of the composite bows in warfare by all classes of Roman military units, except of course in emergencies. Besides composite bows manufactured at least partly on the spot, the oriental archers in Dacia were probably provided with short swords like the one unearthed in the fort at Tibiscum (Bona et al 1983: 413, no. 21, pl. 12/1 ae). The ivory scabbard slides of eastern types, but perhaps also some bone pieces, probably arrived in Dacia directly from the Orient together with their owners enlisted in sagittarii units (Drobeta: Petculescu 1983: 460, no. 22; Romula (Resca, Olt County, unpublished). If in the C2nd oriental auxiliaries seem to have been equipped with conical helmets with narrow cheek-pieces as in the examples found at Bumbeşti (Petculescu & Gherghe 1979: 603-606) and Micia (Petculescu 1982: 293 nr. 2, fig. 3/1), the hoard of arms from the Jidava fort suggests that in the C3rd they adopted the Niederbieber type helmets which were generally fashioned in the Roman army (Petculescu 1999: 189-196). The Commageni from Micia had armour made of differently shaped and sized iron and bronze scales (unpubl. excavations of the author) but for the other bowmen units there is not enough evidence to specify what kind of armour they wore. As for the fittings adorning the military dress and the horse harness of the archers they seem to have been quite common Roman military equipment accessories. In conclusion, during the C2nd-C3rd AD the oriental soldiers in Dacia maintained the specialized weapons linked to their traditional fighting techniques but their armour, dress and harness came closer and closer to the standard equipment of the Roman army. Bibliography Benea D. 1983 Cercetări arheologice în aşezarea civilă de la Tibiscum – Edificiul VII. Banatica 7: 203-223. Bogdan-Cătăniciu I. 1994 Castella de la Urluieni. Studii şi Cercetări de Istorie Veche şi Arheologie 45: 327-355. Bona P. et al 1983 Tibiscum – Cercetări arheologice (II) (1976-

767

Limes XVIII

Fig. 1. Micia.

Fig. 2. Micia.

768

Liviu Petculescu: The military equipment of oriental archers in Roman Dacia

Fig. 3. Micia.

Fig. 4. Micia.

769

Limes XVIII

Fig. 5. 53, 64-65 Micia; 54-58, 66-68 Tibiscum, (after Bona; Petrescu & Rogozea; Benea); 59-61 Apulum (after Ciugudean); 62 Romita (after Matei & Bajusz); 63 Urluieni (after BogdanCătăniciu).

Fig. 6. 69-71, 91-93 Jidava (after Popescu & Popescu); 72-73, 94-96 Urluieni (after Bogdan-Cătăniciu); 75-77, 97-98 Tibiscum (after Bona); 78-90, 102-105 Micia; 100 Feldioara; 101 Ilişua-Cristeşti (after Hica-Cîmpeanu).

770

Die römischen Lager bei Renieblas, Prov. Soria (Spanien). Ergebnisse der Vermessungskampagnen 1997-2000 Martin Luik Das Ziel der Vermessungsarbeiten ist die Erstellung eines archäologisch-topographischen Planes in Maßstab 1:500. Die Arbeiten werden im Auftrag der Abteilung Madrid des Deutschen Archäologischen Instituts von den Fachhochschulen für Technik Karlsruhe und Stuttgart durchgeführt. Bisher liegen für das Gebiet römischen Lager bei Renieblas Pläne vor, die während der Ausgrabungen unter der Leitung von Adolf Schulten durch den bayerischen Topographen General Lammerer in den Jahren 1909 und 1912 angefertigt wurden und deren kartographische Schönheit auch noch heute beeindruckt. Allerdings genügen diese Pläne in vielen Punkten nicht mehr modernen Anforderungen. Im verlauf der vier Vermessungskampagenen wurde das Gebiet des Lagers V vollständig neu aufgenommen, das rund 60ha umfasst. Die Auswertungsarbeiten dauern noch an Während im Süden große Teile des Aufnahmegebietes durch lang anhaltende Beackerung verebnete landwirtschaftliche Flächen darstellten, waren die Befunde vor allem im Norden gut erhalten. Besonders die Süd-, die Ostund die Nordumwehrung des Lagers waren stellenweise noch gut zu erkennen. Im Norden waren noch zahlreiche Spuren der Ausgrabung zu beobachten. Die Grabungsschnitte blieben dort offen liegen, im lockeren Erdreich konnten sich Bäume und Sträucher entwickeln. Hier konnten zahlreiche Gebäudestrukturen dokumentiert werden, deren Mauerfluchten mitunter erhebliche Abweichungen von den Angaben Lammerers aufweisen. Auch konnten weitere, eindeutig römische Mauerzüge aufgenommen werden, die in den alten Planunterlagen nicht verzeichnet sind. Vor allem zeigte sich, dass der das Lagerareal in seiner gesamten Länge (rund 1km) durchziehende Steilhang wohl vollständig terrassiert war. Datierende Funde fehlen bisher. Jedoch spricht die Beobachtung, dass diese Terrassen jeweils unmittelbar an der Lagerumwehrung enden, eindeutig für eine römische Erbauungszeit.

In den Jahren von 1997 bis 2000 wurden umfangreiche Vermessungsarbeiten in den römischen Lagern bei Renieblas, Provinz Soria, durchgeführt (Breuer, Luik & Müller 1999). Auftraggeber war die Abteilung Madrid des Deutschen Archäologischen Instituts. Wichtigstes Ziel war die Erstellung eines archäologisch-topographischen Planes des Lagers V (über 60ha). An den vier Vermessungskampagnen waren insgesamt 20 Diplomanden beteiligt, die von Prof. Dr. Peter Breuer und Dipl.-Ing. Dieter Müller, Hochschule für Technik Stuttgart, und Prof. Dr. Heinz Saler, Hochschule für Technik Karlsruhe, betreut wurden. Die wissenschaftliche Leitung des Unternehmens lag in den Händen des Verfassers. Der vorliegende Beitrag fasst wesentliche, gemeinsam erarbeitete Ergebnisse dieser vierjährigen Tätigkeit zusammen.

vom Ebrotal kommend an der Sierra del Moncayo vorbei über Numantia nach Westen führt. Renieblas liegt wie ein Sperrriegel vor dieser Verkehrsachse. Die höchste Stelle der Anhöhe liegt bei 1150m, von wo sich das Gelände nach Norden allmählich senkt, nach Süden fällt es steil ab. Unter diesem Steilhang erstreckt sich eine breite, wellige Terrasse, die mehrere leichte Erhebungen aufweist und bis zum Moñigón reicht. Durch diese exponierte Lage erklärt sich die rauhe, extrem windige Witterung, von der bereits in den antiken Schriftquellen bei Appian berichtet wird. Die bisher vorliegenden Pläne wurden von dem Topographen Adolf Lammerer während den Ausgrabungen angefertigt, die in den Jahren 1909 bis 1912 unter der Leitung des Erlanger Althistorikers Adolf Schulten durchgeführt wurden (Schulten 1929; Blech 1995). Die zeichnerische Klarheit der Pläne beeindruckt auch heute noch.

Die Errichtung der Lager bei Renieblas hängt eng mit den kriegerischen Auseinandersetzungen um das nahegelegene Oppidum von Numantia zusammen, das in den Jahrzehnten nach 195 v. Chr. bekanntlich Ziel zahlreicher römischer Feldzüge war und schließlich erst durch eine neunmonatige Belagerung 134/133 v. Chr. erobert werden konnte. Über die wichtigsten bisherigen Forschungen zu den römischen Lagern um Numantia und die Neubearbeitung der Kleinfunde aus diesen Lagern wurde auf dem 17. Internationalen Limeskongress 1997 in Zalǎu, Rumänien berichtet (Luik 1999; vgl. 1997a; 1997b).

Bei den Ausgrabungen wurden insgesamt fünf römische Lager nachgewiesen. Die Zeitstellung der römischen Militäranlagen bei Numantia wurde von Schulten auf Grund historischer Überlegungen ermittelt, die im Folgenden kurz zusammengefasst werden sollen (Luik 1997a: 227ff. Nr. 7; ders. in: Breuer, Luik & Müller 1999: 129f.). Das zahlreiche Fundmaterial der Ausgrabungen spielte bei diesen Überlegungen keine Rolle. Das Lager I wurde von Schulten mit dem Feldzug des M. Porcius Cato 195 v. Chr. in Verbindung gebracht (1929: 33ff.). Den einzigen Hinweis darauf in den schriftlichen Quellen bildet eine kurze Erwähnung bei Aulus Gellius, wo eine Rede erwähnt wird, die Cato vor Numantia gehalten habe (Gellius XVI.1.3). Die Zeitstellung des Lagers II ist völlig offen.

Renieblas befindet sich am Ostrand der Hochebene von Soria, die halbkreisförmig von hohen Gebirgszügen eingerahmt wird (Abb. 1). Hier treffen gleich mehrere Verkehrswege von Norden und Osten aufeinander, von denen die alte Ost-West-Verbindung die wichtigste ist, die

771

Limes XVIII

Das Lager III wird von Schulten nach einer Textangabe Appians datiert (1929: 41ff.; Morillo Cerdán 1991: 153f.). Danach habe Q. Fulvius Nobilior sein Standquartier in den Jahren 153/152 v. Chr. 24 Stadien vor Numantia aufgeschlagen, was allerdings der Entfernung nach Renieblas keineswegs entspricht (Appian Iber. 46f., 188197).

weiteren ist der Cotillo im Süden c.50m zu weit östlich eingetragen. Auch differieren die Höhenlinien um c.1215m. Offenbar liegen diese Abweichungen und offenkundigen Fehler mit dem damaligen Stand der Vermessungstechnik begründet. Mit den Methoden der Zeit um 1910 war wohl keine größere Genauigkeit zu erzielen. Auch wurden offenbar damals die Pläne aus zeitlichen Gründen in mehreren Arbeitsschritten erstellt, von denen der letzte Schritt die Montage der zahlreichen Einzelpläne umfasste, die wohl erst in Deutschland erfolgte.

Die Datierung des Lagers IV wurde über die Umrissform des Lagers festgesetzt, welche derjenigen des Lagers V sehr ähnlich sei (Schulten 1929: 230; Morillo Cerdán 1991: 154). Beim Lager V glaubte schließlich Schulten nach der Anordnung der Kasernen eine Lagerung der Truppen nach Kohorten und damit eine Folge der so genannten Heeresreform des Marius erkennen zu können (1929: 231). Aus diesen Gründen wurde hier das Winterlager des Titurius 75/74 v. Chr. vermutet, in dem 15 Kohorten untergebracht waren, womit das Lager V in den historischen Zusammenhang des Sertoriuskrieges gehören würde. Vor einigen Jahren hat H.-J. Hildebrandt auf Grund einer erneuten Auswertung der Fundmünzen eine Zeitstellung des Lagers V schon im 2. Jahrhundert v. Chr., kurz nach 133 v. Chr., vorgeschlagen (1979: 244). In dieser Arbeit bildeten die Periodenzuweisungen durch Schulten die Grundlage für die Auswahl der Fundmünzen. Dass diese Vorgehensweise problematisch ist, wird weiter unten noch zu erörtern sein.

Im Verlauf von Geländebegehungen gelang uns auch die Neuentdeckung und Dokumentation von weiteren Wallzügen (Breuer, Luik & Müller 1999: 139, Abb. 7; 142ff.). So zeigte sich, dass der Bergrücken La Mesta im Osten wohl vollständig von einem Wall bogenförmig umschlossen war. Im Nordwesten weist dieser Wall ausgerechnet an der Stelle eine markante Lücke auf, wo ein Geländeeinschnitt den leichtesten Weg auf die Hochfläche ermöglicht, wo man aber am ehesten eine künstliche Befestigungsanlage erwarten würde. Man wird diese auffällige Diskrepanz nur durch Ausgrabungen lösen können. Von diesem Randwall zieht ein weiterer Wall hangabwärts, um unterhalb der Kuppe des Cotillo zu enden. Alle diese Wallanlagen haben sicherlich eine römische Zeitstellung und sind wohl zum Annexlager des Lagers III zu rechnen, in dem die Unterbringung von Auxiliartruppen vermutet wird (Luik 1997a: 230).

Die auf die beschriebene Weise von Hildebrandt zusammengestellte Münzreihe umfasst lediglich sechzehn Gepräge, worunter sich immerhin sechs iberische Münzen, die sich nur unscharf datieren lassen, und ein römisches Gepräge der Jahre 108/107 v. Chr. befinden. Derzeit liegt zu wenig Fundmaterial vor, um die Zeitstellung des Lagers V auch nur einigermaßen verläßlich angeben zu können (Kritisch bereits Fabricius 1911: Sp. 37f.).

Ferner zeigte sich jetzt, dass die Kuppe des Cotillo durch eine Böschung von c.2m Breite umzogen war. Fundmaterial zur näheren Datierung dieser Wallanlagen konnte zwar nicht geborgen werden. Jedoch macht die Bauweise der Wälle insgesamt einen prähistorischen Eindruck. Damit könnte auch das aus den alten Grabungen bekannte Fundmaterial der jüngeren vorrömischen Eisenzeit zusammenhängen.

Vor allem wegen dieser unterschiedlichen Datierungsansätze, aber auch angesichts des hervorragenden Erhaltungszustandes der Befunde erschien also eine Neubeschäftigung mit diesem römischen Lager als besonders lohnend.

Beim Lager V waren vor allem die Befunde im gesamten, oben auf der Geländekuppe errichteten Nordteil noch gut erhalten. Auch die Süd- und die Ostumwehrung des Lagers waren stellenweise gut zu erkennen. Die Südmauer war als Steinwall vorhanden, der im Westen bis zu 2.5m hoch war und nach Osten allmählich auslief. In der Verlängerung konnte eine leichte Geländewölbung nachgewiesen werden, die auffällig dicht mit Steinen durchsetzt ist. Vermutlich gelang hier der weitere Nachweis der Südmauer sowie der abschließenden Südostecke des Lagers.

Die Geländearbeiten der vergangenen Jahre umfassten mehrere Arbeitsschritte: An erster Stelle sind Kontrollmessungen mit satellitengestützter GPS-Technik zu nennen, die an markanten Punkten der Lager III, IV und V durchgeführt wurden, um die Genauigkeit der Aufnahme durch Lammerer zu überprüfen. Dabei zeigten sich über größere Entfernungsdistanzen sowohl in der Länge als auch in der Breite erhebliche Differenzen mit seinen Angaben auf. Nimmt man die Nordfront des Lagers V als gegebene Bezugsgröße an, so stimmen die Nord- und die Ostseite des Lagers III mit einer maximalen Differenz von 10m noch recht gut überein. Jedoch ergeben sich zur Nordfront des Lagers IV Abweichungen von bis zu 20m. Des

Im Norden winkelte die Ostmauer von der Nordostecke gerechnet nach c.130m ab, wie es auch auf den alten Plänen dargestellt wird. Wegen des dichten Buschwerkes gelang es allerdings jetzt im Gelände nicht, diese Stelle auch genau festzulegen. Überhaupt zeigte sich das Gelände jetzt stark bewachsen, während es auf den Fotoaufnahmen der alten Ausgrabungen viel kahler abgebildet ist. Stellenweise ließ sich die Breite der Nordumwehrung sehr gut bestimmen, die bis zu 4m betrug. In diesem Bereich 772

Martin Luik: Die römischen Lager bei Renieblas, Prov. Soria (Spanien) stellen die alten Pläne eine starke Vereinfachung der tatsächlich vorhandenen Befunde dar, deren Darstellung teilweise deutlich korrigiert werden muss.

den ersten Blick merkwürdig und bedarf weiterer Nachforschungen. Noch wichtiger für die Gesamtinterpretation des Lagers ist die Beobachtung, dass das untere Drittel des Steilhangs mit Terrassen versehen ist (Abb. 4). Derartige Terrassierungen lassen sich über die gesamte Ausdehnung des Lagers, also rund 1km weit, beobachten. Die Zeitstellung ist zunächst völlig offen. Als Gründe für ihre Anlage kommen Besiedlungsvorgänge oder Erfordernisse der landwirtschaftlichen Nutzung in Frage, dann wenn das übrige, in der Ebene gelegene, gute Ackerland knapp wurde. Zu keinem Zeitpunkt ist jedoch eine besonders dichte Besiedlung der Hochebene von Soria zu erkennen. Seit dem Mittelalter wurde vorwiegend Schafzucht betrieben. Die dringende Notwendigkeit, an Hanglagen meterhohe Terrassen anzulegen und dort Ackerbau zu betreiben, bestand also zu dieser Zeit wohl kaum, während der Anbau von Sonderkulturen, z.B. von Wein, aus klimatischen Gründen ausscheidet. Auf den Terrassen wurden auch mehrfach Gebäudereste beobachtet, die von Schulten als römisch bezeichnet wurden und genauere Aufschlüsse über die Zeitstellung dieser Anlagen geben könnten. Hier helfen aber nur weitere Ausgrabungen weiter.

Bei der Nordmauer vermitteln die alten Grabungspläne einen vollständig geradlinigen Verlauf dieses Mauerzuges, was durch die jetzt durchgeführte Geländeaufnahme in Frage zu stellen ist. Beim derzeitigen Stand der Auswertungsarbeiten ist vielmehr von mehreren geradlinig geführten Teilabschnitten auszugehen, die in stumpfen Winkeln miteinander verbunden sind. Stellenweise waren die alten Grabungsschnitte an den Türmen noch gut zu erkennen. Von der Innenfläche des Lagers zeigte sich der gesamte Südteil als riesige, durch die langjährige Feldbestellung verebnete Ackerfläche, wo kaum Befunde sichtbar waren und wo auch nur wenige Funde geborgen werden konnten. Bereits die archäologischen Untersuchungen des Jahres 1912 waren unter großem Zeitdruck durchgeführt worden. Dagegen sind im Norden noch zahlreiche Spuren der Ausgrabung zu erkennen. Die Grabungsschnitte blieben dort größtenteils offen liegen, im lockeren Erdreich konnten sich Bäume und Sträucher entwickeln. Klar und deutlich zeigt sich die Untersuchungsmethodik, die bei den alten Grabungen angewandt wurde und von der man sich beim Studium der summarisch gehaltenen Tagebücher und bei der Lektüre der Vorberichte kaum eine genauere Vorstellung verschaffen kann. Offenkundig wurde damals auf die Aufdeckung größerer Geländeflächen ganz verzichtet, auch unterblieb eine Untersuchung der Innenräume häufig. Statt dessen legte man entlang der gut sichtbaren Mauerzüge schmale Suchschnitte an, deren Aushub in das Innere der Gebäude geworfen wurde.

Für eine römische Zeitstellung scheint auch zu sprechen, dass nach dem derzeitigen Stand der Auswertungsarbeiten die Terrassenmauern jeweils an der westlichen und der östlichen Lagermauer zu enden scheinen. Diese Ergebnisse lassen Fragen nach der Belegungsdauer des Lagers in neuem Licht erscheinen. Sicherlich erforderte die Errichtung solcher Geländeterrassen umfangreiche vorbereitende Maßnahmen und eine längere Bauzeit. Folglich scheidet die bisher von der Forschung stets erwogene Nutzung als kurzfristig genutztes Winterlager wohl aus (Luik 1997a: 231).

Nach der erneuten Planaufnahme der Befunde läßt sich feststellen, dass die Mauerfluchten Abweichungen von den Angaben Lammerers aufweisen. Auf den alten Plänen sind die Mauerzüge oft viel geradliniger angegeben als es in Wirklichkeit der Fall ist. Ferner konnten weitere Mauerzüge dokumentiert werden, die auf den alten Plänen fehlen.

Grundsätzlich wäre zu bemerken, dass im Norden, im Bereich der Überlagerungen der Lager III, IV und V, keine kompakten überdeckenden Bodenschichten vorhanden sind. Wegen der starken Bodenerosion konnte sich hier nur wenig Erdreich ablagern, klare stratigraphische Verhältnisse fehlen. Eine eindeutige Trennung der einzelnen Lagerperioden wird so schwierig. Vor allem ist aus diesem Grund auch die Zuweisung von Funden an einzelne Lagerperioden als problematisch einzustufen, so vielversprechend solche Trennungsversuche für feinchronologische Überlegungen bei einzelnen Materialgruppen wie den Fibeln, den zahlreichen Waffen und der Keramik auch sein mögen (Luik 1997b: 463ff.; ders. 1999: 261ff.).

Bei einem auffälligen rechtwinkligen Gebäude nahe der nördlichen Umwehrung passen die ermittelten Abmessungen gut zu den bisher vorliegenden Angaben über Größe (c.28m x 15m) und Struktur des Steinbaus, der im Inneren von längs verlaufenden Mauerzügen (B. 0.750.8m) durchzogen wird (Abb. 2-3). Von Schulten wurde dieses Gebäude sicherlich zu Recht als Horreum bezeichnet (Schulten 1929: 158f.). Zusammen mit den im Lager Castillejo ergrabenen Horrea handelt es sich um die ältesten bekannten derartigen Magazingebäude aus römischen Lagern. Im Verlaufe unserer Arbeiten konnte jetzt festgestellt werden, dass dieses Horreum auf einer künstlichen Geländeanschüttung errichtet wurde, deren Höhe immerhin c.2m betrug. Eine derartig exponierte Lage ausgerechnet an der stark befestigten und daher wohl besonders gefährdeten Nordseite des Lagers erscheint auf

Im Frühjahr 2001 findet eine Abschlusskampagne statt, die der Zusammenführung der zehn Einzelpläne und einer letzten Gesamtkontrolle dienen soll. Eine ausführliche gemeinsame Abschlussveröffentlichung ist vorgesehen.

773

Limes XVIII

Literatur Blech M. 1995 Schulten und Numantia. Madrider Mitteilungen 36: 38-47. Breuer P., Luik M. & Müller D. 1999 Zur Wiederaufnahme archäologischer Forschungen in den römischen Lagern bei Renieblas (Prov. Soria). Madrider Mitteilungen 40: 125-145. Fabricius E. 1911 Über die Ausgrabungen in Numantia. Archäologischer Anzeiger: Sp. 370-382. Hildebrandt H.-J. 1979 Die Römerlager von Numantia. Datierung anhand der Münzfunde. Madrider Mitteilungen 20: 238-271. Luik M. 1997a Die römischen Militäranlagen der Iberischen Halbinsel von der Zeit der Republik bis zum Ausgang des Prinzipats. Ein Forschungsüberblick. Jahrbuch des Römisch-Germanischen Zentralmuseums 44: 213-275. Luik, M. 1997b Fibeln vom Typ Alesia aus den römischen Lagern um Numantia. Archäologisches Korrespondenzblatt 27: 463-480. Luik M. 1999 Neue Forschungen zu den römischen Lagern bei Numantia, Provinz Soria (Spanien). In N. Gudea (ed.) Roman Frontier Studies. Proceedings of the XVIIth International Congress of Roman Frontier Studies. (Zalău): 261-266. Morillo Cerdán A. 1991 Fortificaciones campamentales de época romana en España. Archivo Español de Arqueología 64: 135-190. Schulten A. 1929 Die Lager bei Renieblas. Numantia IV. (München).

774

Martin Luik: Die römischen Lager bei Renieblas, Prov. Soria (Spanien)

Abb. 1. Renieblas. Gesamtplan der römischen Lager. - M. 1. (nach Schulten 1929, Planbeilage 1).

775

Abb. 2. Renieblas, Lager V. Neuaufnahme von 1999 mit dem Magazingebäude und der Nordumwehrung. - M. 1: 500. (Vorlage D. Müller).

Limes XVIII

776

Martin Luik: Die römischen Lager bei Renieblas, Prov. Soria (Spanien)

Abb. 3. Renieblas, Lager V. Magazingebäude im Nordteil des Lagers. Blick von Norden. (Aufnahme M. Luik).

Abb. 4. Renieblas, Lager V. Terrassierungen wohl römischer Zeitstellung im unteren Drittel des Steilhangs. Blick von Westen. (Aufnahme D. Müller).

777

Limes XVIII

778

Twenty years of Roman military archaeology in Spain Angel Morillo & Victorino García-Marcos Spanish research on Roman camps is less developed in comparison to countries like Germany or Britain. Until a few decades ago the military establishments excavated were few and their publication idiosyncratic. A.Schulten’s studies at the beginning of the C20th on the Republican period camps at Numantia were not continued. His large monographs, written in German, were never translated and have remained largely inaccessible to the Spanish archaeological community. From the 1960s, coinciding with a resurgence in Spanish archaeology, new military sites have been discovered. These were identified by means of field survey and aerial photography, among them Rosinos de Vidriales (Zamora). In turn they have been subject to sporadic archaeological excavation, the best known being the camp of legio VII Gemina at Leon. In 1974, J.M.Roldan published a monograph on the Roman army in Hispania. In 1982, P.Le Roux wrote an important study on the same topic. These studies demonstrated the importance of the army in the (Roman) administration of the northern Iberian peninsula. Ulbert’s 1984 publication of the Republican camp at Cáceres el Viejo inaugurated a new phase in the Spanish military archaeology. This resulted in a total overhaul in methodology and a better knowledge of the military enclosures. Nevertheless it is a pity that this innovation has not been adopted in a greater number of publications In this paper we want to report the results of archaeological activity in Spain’s military context over the past two decades. The most important developments concern understanding of the legionary camps dating from Augustus and the Julio-Claudian emperors which originated in the Cantabrian War (29-19BC to C1st AD) or as direct consequence of that war. In identifying these encampments there are many problems. The camps are located in the north-west of the peninsula, in the contact area between the Iberian plateau and the Cantabrian Mountains. Excavations in the camps of Herrera de Pisuerga (legio IIII Macedonica), Leon (legio VI Victrix and legio VII Gemina), Astorga (legio X Gemina) and Rosinos de Vidriales (legio X Gemina) have demonstrated that they were permanent bases of the hispanicus exercitus during Julio-Claudian times. These establishments are earlier than the first camps of the German limes, although their construction and the stratification of some of them show many similarities with their German counterparts. These excavations have allowed an insight into their internal structures, defensive systems, their foundation dates and chronological evolution. With regard to Leon, the headquarters of the legio VII Gemina from AD74, the excavations have shown three earlier camps. These belonged to the legio VI Victrix, the earliest dated to the Augustan period. At Herrera de Pisuerga, over the Augustan-Tiberian camp of legio IIII Macedonica, was constructed another camp. This camp is associated with a previously unknown unit in Hispania, the ala Parthorum, whose camp is dated between the 60s and the beginnings of the C2nd. Other new discoveries include the temporary forts associated with the Cantabrian Wars or earlier. They are Andagoste (prov. Alava) and Espina del Gallego (prov. Cantabria). The layout of these camps are very important in understanding the techniques used at the beginning of the Augustan period. In Galicia excavations have continued at Aquis Querquennis (prov. Orense) and at A Cidadela (prov. La Coruña). Excavation in the Republican camps at Numancia has also re-commenced. There have also been discovered new auxiliary establishments at Villalazan (prov. Zamora) and Burgo de Osma (prov. Soria). On the other hand, the increase in evidence of Augustan date in regions like Lusitania and the valley of the Ebro confirms the important work undertaken by the army in the reorganization of these territories after the conquest of Hispania. This labour is much more active than was previously thought and we should relate it with the new Augustan foundations of Augusta Emerita (Merida) and Caesaraugusta (Zaragoza).

Research on the Roman forts of Spain has progressed slowly compared to that in countries such as Germany and Great Britain. Upto a few decades ago, few military camps had been excavated in Spain and the results rarely published. Schulten’s (1927; 1929) work on Republican forts, which he conducted in the first decades of the C20th, have not been continued. His extensive monographs concerning the camps at Numancia have not been translated into Spanish from their original German and have remained largely unexamined by the Spanish scientific community. New forts and military camps were not discovered until the 1960s, coinciding with the development of Spanish scientific archaeology. They were identified by surveys and aerial photography, as was the case with Rosinos de Vidriales (Zamora; Martín Valls et al 1975). Likewise, archaeological excavations were made in already well-known camps and forts such as the legio VII Gemina’s in León (García y Bellido 1970).

Following the trend begun by García y Bellido (1961), J. M. Roldán published his 1974 monograph about the Roman army in Hispania and in 1982, P. Le Roux introduced his study on the same subject. Through these monographs, the importance of the army in the territorial articulation of the northern region of the Iberian peninsula was established. G. Ulbert’s (1984) publication on the Republican camp of Cáceres el Viejo opened a completely new phase in Spanish Roman military archaeology. Nevertheless, its importance has been recognized only in the last decade. In the intervening period, a total renovation of methodological approaches and qualitative advances in the knowledge of the military bases has been made. However, this renovation has still not been accompanied by the publication of the results in accordance with this effort. In this paper, we aim to present the principal results of

779

Limes XVIII

archaeological activity in military contexts in Spain in recent years.

Until recently the Augustan and Julio-Claudian fortresses remained the least known feature of the Roman military presence in Spain. These camps arose during the Cantabrian wars (29-19BC) and as a direct response to that conflict. They are concentrated in the north-west Iberian peninsula, specifically in the region between La Meseta and the Cantabrian mountain range. The Augustan and Julio-Claudian military settlements present special problems of identification because the majority of them are beneath later cities which have altered the archaeological evidence, sometimes in such a way that archaeological interpretation is at best difficult. On the other hand, the camps’ building technique was still in an experimental period during Augustus’ and Tiberius’ reigns. The camps consist of temporary timber structures not yet standardizied or orthogonal in plan. Therefore, the common identification of structures in a military settlement is complicated. Moreover, we must consider how the rocky terrain and extreme fluctuations in temperature of northern Spain make difficult the preservation and subsequent identification of the camps.

Over the past 15 to 20 years, the emphasis of research in Spain has been focused largely on fieldwork. Effort has been directed especially towards a double objective. On the one hand, there exists the study of the construction and materials of auxiliary forts of the imperial age (eg. Rosinos II, A Cidadela, Aquae Querquennae), which is without doubt best documented from the structural point of view. On the other hand, there are the archaeological identifications and descriptions of the legionary fortress of the Augustan and Julio-Claudian periods (eg. León, Astorga, Herrera de Pisuerga, Rosinos I), which have traditionally caused formidable difficulties in the documentation of military structures. Aerial photography and fieldwork have aided in the discovery of forts and camps such as Valdemeda, Uxama, Villalazán, Andagoste and the siege works at La Espina del Gallego. The results have provided new perspectives about the distribution of the military units and the main objectives of the army in the Iberian peninsula in the Roman period.

Bearing in mind these difficulties, the identification of military camps has been made by the analysis of the most ancient archaeological records of these settlements as a starting point and by noting their unmistakable military elements (TSI; Vogelkopflampen and volute lamps of the older types; metal elements that are traditional elements of military kit; military coins with caetra, etc.; Morillo 1996: 79-80). Comparison of the archaeological material with the oldest Rhineland camps is another device. However the identification of structures has usually been made from a misapplied older typological definition of the military character of a settlement. As a result, we still know little about the plans and inner arrangements of the Spanish camps of this period.

The long duration of the Roman conquest of Hispania, which lasted two centuries (218-19BC), and the significance of certain historical episodes such as Celtiberian and Sertorian wars have traditionally attracted research attention to Republican period camps and forts. Spain preserves the best collection of Republican Roman military establishments as Schulten’s surveys and excavations revealed at sites such as Numancia, Renieblas, Cáceres el Viejo and Aguilar de Anguita (Schulten 1927; 1928; 1929). Nevertheless, important discoveries in Republican forts have not been made in the past 20 years, with the sole exception of Ulbert’s excellent publication about Cáceres el Viejo (Ulbert 1984). This camp was a stone wall construction with corners at acute angles and a double ditch system. It was founded during the Sertorian wars and destroyed c.80BC.

The Cantabrian wars necessitated the concentration of a large number of military units - nearly 7 legions - in the northern part of the peninsula. The camps of this period are yet little known. However, in past years, research on the first military settlements that we can directly associate with the Cantabrian wars has been published (Fig. 2). The forts are located in the interior of Cantabria, on the southern slope of the Cantabrian mountain range. Thus they controlled a natural corridor connecting the interior to the coast. They were located within a mountainous zone, upon extremely uneven hillsides, and were arranged around an important native settlement, the castro, so-called La Espina del Gallego (Cantabria). They formed a complete siege system reminiscent of the fortification works around Alesia. The most important of the group is the legionary fortress of Cildá. It was 25ha in size and was surrounded by several fortification lines and was rectangular in shape while conforming to the slopes of the hills. It had a double ditch with a V-shaped profile (fossa fastigata) and an inner earthen bank enclosing a large stone core. The fortress also had several gateways defended by claviculae and tituli. The El Cantón fort is situated some distance away. It is smaller and oval in shape with a defensive agger and

Archaeological excavations have not continued at Cáceres el Viejo nor at the other Republican sites documented by Schulten. Excavation of camps and forts around Numancia has merely made possible the reinterpretation of how some objects found by Schulten clarify the chronology of the settlements (Hildebrant 1979; Sanmartí 1985; 1992; Romero 1989; Sanmartí & Principal 1997; Luik 1997a). Certain structures also have been re-examined (Pamment Salvatore 1996). In the past few years, archaeological exploration has resumed in the Renieblas camps as part of a research project directed by the German Archaeological Institute in Madrid (Breuer et al 1999; Luik 1999). However as of now, these interventions have yielded no significant results. Additionally, the study of the Escipion’s circumvallatio around Numantia has also been resumed (Morales Hernández 1998 - forthc.) (Fig. 1). Likewise, few articles about Spanish military weapons of the Republican period have been published recently (Feugere 2000: passim). 780

Angel Morillo & Victorino García-Marcos: Twenty years of Roman military archaeology in Spain gateways with claviculae. Both bases, and maybe one more, are perfectly identifiable in situ and are the subject of current archaeological research (Peralta 1999a; 1999b). These settlements were built during campaigns to subjugate the natives, called transmontani, and to control access to the Cantabrian coast.

Tarraconensis province that consisted of three legions chosen from those that had taken part in the war: the IIII Macedonica, the VI Victrix and the X Gemina. Proof of the continuation of three units as a stable garrison in the peninsula is in a well-known passage of Strabo, in which he describes a commander in charge of two legions in Asturia and another commander, with a single legion, situated in Cantabrian territory. During the period subsequent to the war and that generally continues through the Julio-Claudian period, the archaeological and epigraphic testimonies relating to the conduct and the presence of the army in the peninsula gradually increases. It is from this time that a series of ‘permanent’ bases was created for the troops stationed in the northern peninsula. The sites of Herrera de Pisuerga, Astorga, León and Rosinos de Vidriales are the first permanent Augustan legionary fortress as verified by archaeological evidence. They shared similar strategic geographic and topographic locations, in contact with the natural communication routes that connected the north-west rim of La Meseta with the coastal regions of Galicia and the Cantabrian coast (Fig. 2). They configured a protective barrier to the south of the Cantabrian mountain range, one that we have called ‘a limes without a border’ and they give us an idea of the plan applied in the northern boundaries of the empire some years later (Morillo 1996: 81). The deployment of the army along the southern slope of the Cantabrian mountain range and to the east of the mountains of León was maintained until the end of the Julio-Claudian period and with much smaller dimensions throughout the imperial period.

We must also mention the recent discovery of a military site at Andagoste (Cuartango, Alava), along a corridor through the Ebro valley to the Cantabrian coast. It is situated in the midst of a valley with a gradual elevation in terrain. The structural evidence consists of a simple flatbottomed ditch, that surrounds the settlement cut into the rocky ground. In some places its bank survives. The structure is as significant as the archaeological remains which have also been discovered in situ. Among some of the most remarkable finds are large tacks, sling projectiles and light weapons, set in wide circles around the fortification. The site appears to have been fortified rapidly and was defended against a siege by the natives. The numismatic finds do not allow us to date the structure to the Cantabrian wars, but rather indicate an earlier date, between 40-30BC (Ocharan & Unzueta 1998: forthc.; Unzueta & Ocharan 1999). This dating, in addition to the placement of the fort outside the territory where the campaigns of the Cantabrian wars took place, represents the first proof of tactical military movements prior to the war in this region. Such actions may have smoothed the way for the conquest of the Cantabrian and Asturian peoples (Morillo 1998a - forthc.). Neither of the legionary fortresses of the subsequent period present surviving remains that belong to the first phases of the Cantabrian wars. The first ‘permanent’ legionary fortress was related to the campaign of Agrippa in 19BC; the camp of Herrera de Pisuerga, located in the zone between the plains of La Meseta and the foothills of the Cantabrian mountains. The archaeological data indicates that it was founded between 20/15BC (Pérez González 1989: 218; Morillo 1992: 166; 1999: 333). Certain potsherds could even provide an earlier chronology for perhaps as early as 25/20BC that would coincide with the first campaigns against the Cantabrians. The supposed camp of Lucus Augusti, which is actually the city of Lugo, creates a similar problem. Recent fieldwork has indicated the existence of an initial military phase during the Cantabrian wars. This fort would have been occupied by the legio VI Victrix, whose abbreviation is carved on a stone block (Rodríguez Colmenero & Carreño 1992; 1996). However, we have no military structures at the moment and the material evidence is contradictory (Morillo 1998a - forthc.).

The origins of the legionary fortress of Herrera de Pisuerga (Palencia) might lie towards 20/15BC or maybe some years earlier (Pérez González 1989: 218; Morillo 2000a: 617). It was located next to the River Pisuerga, the principal corridor into the interior mountain range and the sea. The camp housed the legio IIII Macedonica, a unit which stayed at this site throughout the Augustan-Tiberian period until its departure in 39 to Mogontiacum (Mainz). We do not have enough data about the structure of this fortress which had without doubt a timber defensive circuit (agger). In the later C1st BC, the camp underwent a major transformation, when its older timber buildings were replaced by firmer and more stable structures, in which adobe and stone, besides timber, were used for the lower parts of the walls (Pérez González 1996: 91; 1998). This consolidation phase can be seen in the barrack blocks. Several kinds of pottery were made in the fortress. We might note the italic terra sigillata vessels signed by the legion figlinarius, L. Terentius (Pérez González 1989: 216-217), as well as lamps (Morillo 1992: 167-168; 1993b) and common and thin-walled pottery. Objects of worked bone, metal and possibly textile and leather goods were also made (Pérez González 1995a; 1995b; Fernández Ibáñez 1998 - forthc.). The re-design of the defences and inner structures took place at the beginning of Tiberius’ reign.

The end of the war and the subsequent departure of the majority of the troops to the empire’s northern provinces clearly initiated a new period in the relationship between the Roman army and the peninsula. It is from this time that the basis of a military policy of long-term territorial occupation was established. The basis of that policy was the configuration of an exercitus hispanicus attached to the

Fieldwork carried out in Astorga (León), ancient Asturica Augusta, capital of the conventus asturum, has allowed us to identify an initial military phase. Nowadays, remains of 781

Limes XVIII

negative constructive structures with several foundation trenches and postholes, that are common examples of the timber military architecture, are well-known (García Marcos & Vidal 1995b: Morillo & García Marcos 2000a: 598). The recent discovery of a V-shaped double ditch (fossa fastigata) belonging to the defensive system of the camp (González Fernández 1996), constitutes a decisive argument for the military character of the early occupation of Astorga. The material remains from this site allow us to date its foundation to the later C1st BC, towards 15/10BC, but after the Cantabrian wars (Morillo 1999: 335; Morillo & García Marcos 2000a: 598-599). The find of several epitaphs of soldiers of the legio X Gemina, as well as two big stone blocks reused in a subsequent construction, which present the inscription L.X.G. carved in large square capital letters (García Marcos & Vidal 1995: 115; Morillo & García Marcos 2000a: 598, n. 91), allows us to identify the legio X as the occupying force at the camp.

agger was dismantled and a new one built on top. On this occasion, the rampart was built with blocks of turf setting up two parallel walls, filled with soil mixed with river pebbles, to a total thickness of c.4m. The outer side of this rampart, as well as the ditch or ditches that there will undoubtedly have been, were destroyed in the construction of the fort wall of the legio VII Gemina that was placed almost exactly on top of the defences of the earlier fort. An intervallum lay on the inner boundary of the fortification and was bordered entirely by a road (via sagularis). A barrack block (centuria) ran parallel to the new defences, whose lower wall were made with stone. These structures have become very distorted because the contubernia of the legio VII Gemina were built on top of them (García Marcos 1998 - forthc.). The existence of military levels of the same period is verified at other locations in León (Miguel & García Marcos 1993; García Marcos & Miguel 1996; Campomanes et al 1998 forthc.). With respect to the unit occupying the camp, several testimonies point to the legio VI Victrix (Morillo 1999: 296297, nº 40, fig. 170; Morillo & García Marcos 2000a: 600). In accordance with the evidence, we must accept that the legio VI Victrix was stationed in León at least between the C1st BC – C1st AD and the evacuation of the Iberian peninsula in 69/70 (García Marcos & Morillo, this volume).

The final moment of the legionary fortress of Astorga and its change into a town might have been towards AD15/20, coinciding with the beginning of the exploitation of the gold ore deposits in the region and which required a urban centre for its administration and control (Morillo 1999: 335). A large urban redesign with levelling of the site later occupied by the city, has been identified. At that time, the legio X Gemina moved to a new fortress at Rosinos de Vidriales (Zamora), 50kms to the south. Thanks to aerial photography, the existence of a 17.5ha rectangular camp with rounded corners was discovered some decades ago. The defences of this camp seem to have consisted a strong wall made with quartzite and mortar blocks, as well as a double ditch on some of its sides (Martín Valls et al 1975; Carretero & Romero 1996: 9-21). Nevertheless, systematic archaeological excavation in this camp has not yet been carried out. The retrieved materials point to a late AugustanTiberian foundation (Morillo 1998a - forthc.). The final dating of the fort would coincide with the departure of the legio X Gemina to Carnuntum in 63. This unit returned to Spain for a brief stay between 68 and 70, and there are some signs of its return in its old fort at Rosinos de Vidriales again. The fortress produced tiles on which appear the mark of the legion (Martín Valls et al 1998 - forthc.).

The legionary camps of the Augustan and Julio-Claudian period set up in the north of Spain should have been accompanied by an indeterminate number of forts. Aerial photography has allowed us to identify some of them. This is the case at Valdemeda, a rectangular base of 4.5ha, protected by a ditch and a rampart, with a gateway defended by a clavicula (Sánchez Palencia 1986). Two more forts have been discovered at Villalazán (Zamora) (Del Olmo 1994/95) and Burgo de Osma (Soria) (García Merino 1996). Archaeological fieldwork has not yet been carried out at either of them. Over the old camp of the legio IIII Macedonica at Herrera de Pisuerga, a fort belonging to a cavalry auxiliary unit, the ala Parthorum, was set up in the time of Nero. The presence of this unit in Hispania was previously unknown. Although its base has not yet been located, the archaeological evidence and the find of bricks with the units’ stamp leave little room for doubt regarding its presence between the time of Nero and the early years of the C2nd (Pérez González 1996: 91-92).

One of the biggest events in Spanish Roman military archaeology in recent years has been the progress in our knowledge about the military settlement in the city of León (García Marcos 1998 - forthc.). Towards 74, the legio VII Gemina settled at this site and remained as the unit in garrison until the end of the empire. Archaeological fieldwork in the last few years by V. García Marcos has allowed us to verify the existence of a legionary fortress previous to the VII Gemina. This base might have been founded in Augustan times, specifically towards the end of the C1st BC. The fort consists of an agger with a V-shaped profile ditch and rampart of turf, but there are no remains of the palisade. Outside the defences, there was a road paving with a thickness of 7m, paved with small stones. During the final years of Augustus’ reign or at the beginning of Tiberius’, the camp underwent profound changes. The earlier

The increase in evidence of Augustan and Tiberian date in regions such as Lusitania and the Ebro valley confirms the important work of the army in the reorganization and exploitation of these territories in the aftermath of the conquest of Hispania (García-Bellido 1994/95; 1998 forthc.; Morillo 1998 - forthc.). That work was much more extensive than was once thought so we should relate it to the new Augustan foundations of Augusta Emerita (Mérida) and Caesaraugusta (Saragossa). The early imperial camps post-74/75 also show remarkable results (Fig. 3). The standardization of their ground plans and the use of stone as a construction material makes it easier to identify the bases of this period. The existence of a legionary base - León - and some forts has been verified. 782

Angel Morillo & Victorino García-Marcos: Twenty years of Roman military archaeology in Spain By 74, the legio VII Gemina set up a camp in León, the same location where the legio VI Victrix had stayed for several decades, remaining as its garrison until the end of the empire. The camp encloses 20ha and preserves most of its walls. In 1968, A. García y Bellido’s excavations revealed that the walls were built with two walls. A second wall was placed against the outer face of the inner wall of 1.8m. width in the later Roman period, to make a complete width of 5.25m (García y Bellido 1970: 574-575). The recent excavations have allowed us to identify the inner wall as the defensive system of the Flavian camp. It has an outer face of opus vittatum with a concrete core, but the inner rampart has disappeared. Two towers, projecting slightly into the exterior as much as the interior, as well as one of the gates of the base, the porta principalis sinistra with double portal and flanked by two monumental towers rectangular in shape, have been documented. Gradually, data about the internal constructions of the legionary camp – its roads, the barrack blocks and baths - as well as data relating to the neighboring cannabae is being accumulated (García Marcos & Vidal 1995b; Campomanes 1997; García Marcos 1989/90; 1998: e. p.; Fernández Freile 1998: e. p.). Material about the camp is also beginning to be published (García Marcos 1989/90; Morillo 1999; Morillo & García Marcos 2000b - forthc.).

shaped profile ditch 4m thick. The porta decumana of a single portal protected by two square towers has been completely excavated. The porta praetoria has a double portal, also between square towers. At the moment, there has not been excavation in the fort’s interior. The construction of the defences took place between the final years of the C1st and Hadrian’s reign (Carretero 1993; Carretero & Romero Carnicero 1996; Carretero et al 1999). Some material remains from the fort have been published (Romero Carnicero & Carretero 1997; Martínez García 1999). At Atxa (Vitoria, Alava), fieldwork has documented a military settlement, with a barrack block and part of the principia and the horrea (Gil Zubillaga 1995). Nevertheless, the lack of the defences - possibly dismantled in modern times - raises some doubts about this site (Gil Zubillaga 1998 - forthc.). The existence of military bases in Spain during the late empire is traditionally based on the information given by the Notitia Dignitatum (Or. XLII.1.25). This document, dated to the beginning of the C4th, describes the presence of several military forces at a number of locations in northern Hispania: the legio VII Gemina in León, the cohors Lucensis at Lucus Augusti (Lugo), the cohors II Flavia Pacatiana in Paetaonio, (Rosinos de Vidriales) the cohors Celtiberae in Iuliobriga, the cohors I Gallica in Veleia (Iruña, Alava) and the cohors II Gallica at an undetermined location called ad cohortem Gallicam. The data of the Notitia, along with certain archaeological remains found in the area, has led some researchers to consider the existence of a limes in the C4th in northern Spain. This hypothesis has been rejected (Arce 1982: 6772).

The Flavian defences at León represent without doubt the oldest example of a stone-built camp in Spain. The example of León is imitated by auxiliary forts. The forts at A Cidadela (A Coruña), Aquae Querquennae (Ourense) and Rosinos de Vidriales II (Zamora) have the same kind of defences. A Cidadela was identified some years ago (Caamaño 1984). Archaeological fieldwork has verified the existence of stone walls with an outer bank which enclosed 2.4ha. One of the gateways to the base has been excavated, as well as the buildings corresponding to the principia and the praetorio (Caamaño 1991b; 1996, 1997; 1998). The excavated material confirm that the fort was occupied by the cohors I Celtiberorum between the early C2nd and the beginning of the C4th (Caamaño 1989; 1990). The fort at Aquae Querquennae, 3ha in size, was rounded by walls of the same type as A Cidadela, with battlements on top, square towers spaced at regular intervals and an outer ditch with a V-shape profile. The porta principalis sinistra with a double portal and flanked by rectangular towers of the same type as those wellknown at León have been completely excavated. Several barracks and the horrea also have been excavated (Rodríguez Colmenero & Herves 1992; Rodríguez Colmenero 1998 - forthc.). Several find reports are wellknown (Rodríguez Colmenero & Vega 1999). The base was in use between the end of the C1st AD and the midC2nd.

It is significant that three forces of limitanei mentioned in the Notitia were stationed in cities that have strong late empire defensive systems: León, Lugo and Iruña (Fig. 4). This coincidence obviously guaranteed the close link between the permanent troops and the walled cities of north Hispania. In the past few years, there has been study of the construction systems and the foundation dates of the late imperial walls of the northern region: at Braga, Lugo, Astorga, León, Gijón e Iruña (Fernández Ochoa & Morillo 1991; 1992; 1997a; 1997b; Fernández Ochoa 1997; García Marcos et al 1998; González Fernández et al 1998: e. p.; Sande Lemos et al 1998 - forthc.; Campomanes 1998). This group of small to medium-sized cities was all provided with powerful defensive systems in the late Roman period. Their walls seem to have clear structural similarities. The features of this style were defined by Richmond (1931). The walls also give a very similar date: the end of the C3rd or the beginning of the following century. We can confirm the existence of a regional style or program of urban fortification (Fernández Ochoa & Morillo 1992: 345). This hypothesis had been defended previously (Richmond 1931; Balil 1960; 1970). In our opinion, such a fortification style or program was without doubt military in origin. However we still have to search

At the end of the C1st AD, a 4.7ha fort that housed an auxiliary unit, the ala II Flavia, was built on top of the old legionary base at Rosinos de Vidriales. This fort shows very similar elements to Aquae Querquennae. It had stone walls with intermediate and angle towers, as a replacement for a vallum. The wall had an outer berm of 2m and a V783

Limes XVIII

for the reasons for the new strategic necessities that forced the fortification of the main regional communication roads for the cereal tax collection (annona militaris) and its relocation towards the northern frontiers of the empire (Fernández Ochoa & Morillo 1997b: 740-742; 1998 forthc.).

Blázquez J.M. & Corzo R. 1986 Luftbilder römischer Lager aus republikanischer Zeit en Spanien. In ed. Studien zu den Militärgrenzen Roms III. 13 Int. Limeskongress. (Stuttgart): 681-691. Breuer P., Luik M. & Müller D. 1999 Zur Wiederaufnahme archäologischer Forschungen in den römischen Lagern bei Renieblas (prov. Soria). Madrider Mitteilungen 40: 125-145. Caamaño J.M. 1984 Excavaciones en el campamento romano de Cidadela (Sobrado dos Monxes, Coruña). Noticiario Arqueológico Hispánico 18: 233-254. Caamaño J.M. 1989 Estampillas de la Cohors I Celtiberorum, halladas en el campamento romano de Cidadela. Gallaecia 11: 209-229. Caamaño J.M. 1990 Vidrios romanos del campamento de Cidadela. Gallaecia 12: 177-190. Caamaño J.M. 1991a El ejército bajoimperial romano en Hispania: la Cohors I Celtiberorum. In J.M. Bello & A. Vigo (edd.) Ciudad y Torre. Roma y La Ilustración en La Coruña. (La Coruña): 19-22. Caamaño J.M. 1991b O campamento romano de Cidadela (Sobrado dos Monxes, A Coruña). Larouco 1: 119-122. Caamaño J.M. 1996 Los campamentos romanos de Galicia. In C. Fernández Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 113-120. Caamaño J.M. 1997 Sondeos arqueológicos en la muralla del campamento romano de Cidadela. Gallaecia 16: 265280. Caamaño J.M. 1998 El urbanismo del campamento romano de Cidadela. In A. Rodríguez Colmenero (coord.) Congreso Internacional: los orígenes de la ciudad en el Noroeste hispánico. (Lugo): 1053-1064. Campomanes E. 1997 Algunas cuestiones en torno a la primera muralla de la Legio VII Gemina. Lancia 2: 129-148. Campomanes E. 1998 Nuevas perspectivas sobre el recinto amurallado de León. In A. Rodríguez Colmenero (coord.) Congreso Internacional: los orígenes de la ciudad en el Noroeste hispánico. (Lugo): 1057-1076. Campomanes E., Muñoz Villarejo F. & Alvarez Ordas J.C. 1998 – forthc. Ocupaciones militares anteriores la llegada de la Legio VII Gemina a la ciudad de León. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Carretero S. 1993 El cuadrante noroeste peninsular en época romana: los efectivos militares y sus establecimientos. Brigecio 3: 47-73. Carretero S. 1997 Los campamentos romanos y su implantación en Hispania. In La Guerra en la Antigüedad. Una aproximación al origen de los ejércitos en Hispania. (Madrid): 333-446. Carretero S. & Romero Carnicero Mª V. 1996 Los campamentos romanos de Petavonium (Rosinos de Vidriales, Zamora). (Zamora). Carretero S., Romero Carnicero Mª V. & Martínez García A.B. 1999 Les estructuras defensivas del campamento del Ala II Flavia en Petavonium (Rosinos de Vidriales, Zamora). In R. De Balbín & P. Bueno (eds.) II Congreso de Arqueología Peninsular. IV. Arqueología romana y medieval. (Zamora): 183-194. Domínguez Bolaños A. & Nuño J. 1997 Reflexiones sobre los sistemas defensivos tardoantiguos en la Cuenca del Duero. A propósito de la muralla de El Cristo de San Esteban, Muelas del Pan (Zamora). In Congreso Internacional “La Hispania de Teodosio”. (ValladolidSegovia): 435-450.

We have in recent times obtained more or less a general impression about the castella or late Roman forts such as Tedeja (Traspaderne, Burgos) (Lecanda 1998), El Cristo de San Esteban (Muelas del Pan, Zamora) (Domínguez Bolaños & Nuño, 1997), Bernardos (Segovia) (FuentesBarrios 1999) and Roc d’Enclar (Andorra) (Yañez et al 1997). All of them are currently in the process of excavation. In short, we must mark again the spectacular progress of Roman military archaeology in Spain during the last two decades, although we are not up to the same standard of publication as other countries. There has been significant progress in our knowledge of the camps and forts of the early empire, particularly Augustan and Julio-Claudian, whereas there is not a qualitative advance for Republican sites. Study has been focused on defensive constructions and materials, whereas the internal buildings have received less attention. Likewise, the features of the canabae and military vici are almost unknown. With regard to the late Roman period, the military horizon is not yet defined. This progress has become reality in several publications, from simple up-to-date studies to interesting thoughts about specific regions or themes (Morillo 1991; 1993a; 1996; 1998a - forthc.; 2000b - forthc.; Carretero 1993 y 1997; Luik 1997; Gil Zubillaga 1998 - forthc.; Le Roux 1999). In 1998, the first conference on Roman military archaeology in Spain was organised which offered new important observations and is a sample of the significant advances in Spanish research (Morillo 1998b - forthc.). Finally, we should also mention an important research project which began in 2000. We refer to the project directed by Mª P. García-Bellido from the Consejo Superior de Investigaciones Científicas (Madrid), in which we take part. It is about to investigate the numismatics of the main early empire fortress in order to give data about their chronology and the monetary circulation in military contexts. Bibliography Arce J. 1982 El último siglo de la España romana: 284-409 (Madrid). Aurrecoechea J. 1998 – forthc. Aproximación al conocimiento de los cinturones militares de época altoimperial en Hispania, a través de sus accesorios metálicos. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Balil A. 1960 La defensa de Hispania en el Bajo Imperio. Zephyrus XI: 179-197. Balil A. 1970 La defensa de Hispania en el bajo Imperio: amenaza exterior e inquietud interna. Legio VII Gemina. (León): 603-620.

784

Angel Morillo & Victorino García-Marcos: Twenty years of Roman military archaeology in Spain Fernández Freile B.E. 1998 – forthc. Aportaciones del estudio arqueológico de un solar extramuros al recinto campamental de Legio. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Fernández Ibañez C. 1998 – forthc. Metalistería bélica de la Legio IIII Macedonica procedente de su campamento en Herrera de Pisuerga (Palencia, España). In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Fernández Ochoa C. 1997 La muralla romana de Gijón (Asturias). (Gijón). Fernández Ochoa C. & Morillo A. 1991 Fortificaciones urbanas de época bajoimperial en Hispania. Una aproximación crítica (primera parte). Cuadernos de Prehistoria y Arqueología de la Universidad Autónoma de Madrid 18: 227-259. Fernández Ochoa C. & Morillo A. 1992 Fortificaciones urbanas de época bajoimperial en Hispania. Una aproximación crítica (segunda parte). Cuadernos de Prehistoria y Arqueología de la Universidad Autónoma de Madrid 19: 319-360. Fernández Ochoa C. & Morillo A. 1997a Urban fortifications and land defence in later Roman Spain. In W. Groenman-van Waateringe, B. L. van Beek, W. J. H. Willems & S. L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 343-346. Fernández Ochoa C. & Morillo A. 1997b La muralla de Iruña en el contexto de las fortificaciones urbanas bajoimperiales de la región septentrional de la Península Ibérica. In I Coloquio Internacional sobre la Romanización de Euskal Herria. Isturitz 9. (San Sebastian): 735-742. Fernández Ochoa C. & Morillo A. 1998 – forthc. Entre el prestigio y la defensa: la problemática estratégico-defensiva de las murallas tardorromanas en Hispania. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Fernández Ochoa M.C. & Morillo A. 1999 La tierra de los astures. Nuevas perspectivas sobre la implantación romana en la antigua Asturia. (Gijón). Feugere M. (dir.) 2000 L’equipement militaire et l’armement de la République (IVe-Ier s. av. J.-C.). Journal of Roman Military Equipment Studies 8. Fuentes A. & Barrios J. 1999 Proyecto de investigación arqueológica en el Cerro de la Virgen del Castillo de Bernardos (Segovia). In R. De Balbín & P. Bueno (edd.) II Congreso de Arqueología Peninsular. IV. Arqueología romana y medieval. (Zamora): 441-450. García y Bellido A. 1961 El “Exercitus Hispanicus” desde Augusto a Vespasiano. Archivo español de Arqueología 34: 114160. García y Bellido A. 1970 Estudios sobre la legio VII Gemina y su campamento en León. In Legio VII Gemina. (León): 569599. García-Bellido Mª P. 1994/95 Las torres-recinto y la explotación militar del plomo en Extremadura: los lingotes del pecio de Comacchio. Anas 7-8: 187-218. García-Bellido Mª P. 1998 – forthc. Labores mineras militares en Hispania: explotación y control. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). García Marcos V. 1989/90 Marcas de alfarero en sigillata hispánica halladas en la ciudad de León. Tierras de León 77-78: 89-114. García Marcos V. 1998 – forthc. Novedades acerca de los campamentos romanos de León. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania.

(Madrid). García Marcos V. & Miguel F. 1996 A new view on the military occupation in the north-west of Hispania during the first century: the case of León. In W. Groenman-van Waateringe, B. L. van Beek, W.J.H. Willems & S. L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 355-360. García Marcos V., Morillo A. & Campomanes E. 1998 Nuevos planteamientos sobre la cronología del recinto defensivo de Asturica Augusta (Astorga, León). In Congreso Internacional “La Hispania de Teodosio”. (ValladolidSegovia): 515-531. García Marcos V. & Vidal J. 1995a Asturica Augusta y Castra Legionis VII Geminae. In Catálogo de la Exposición Astures. (Gijón): 113-128. García Marcos V. & Vidal J. 1995b Recent archaeological research at Asturica Augusta. Proceedings of the British Academy 86 (Social complexity and the development of towns in Iberia from the Copper Age to the second century AD): 371-394. García Merino C. 1996 Un nuevo campamento romano en la Cuenca del Duero: el recinto campamental de Uxama. Archivo Español de Arqueología 69: 269-274. Gil Zubillaga E. 1995 Atxa. Memoria de las excavaciones arqueológicas 1982-1988, Memorias de yacimientos alaveses. (Vitoria). Gil Zubillaga E. 1998 – forthc. Testimonios arqueológicos en torno al mundo militar romano en Vasconia/Euskal Herria. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). González Fernández E., Ferrer Sierra S., Herves F. & Alcorta E. 1998 – forthc. Muralla romana de Lucus Augusti: nuevas aportaciones a su estudio y conocimiento. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). González Fernández Mª L. 1996 Consideraciones sobre el origen militar de Asturica Augusta. In C. Fernández Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 85-89. Hildebrant H.J. 1979 Die Römerlager von Numantia. Datierun anhand der Münzfunde. Madrider Mitteilungen 20: 238271. Lecanda J.A. 1998 Arquitectura militar tardorromana en el norte de España: la fortaleza de Tedeja (Traspaderne, Burgos). Un ejemplo de recinto no urbano y no campamental. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Le Roux P. 1982 L’Armée romaine et l’organisation des provinces ibériques d’Auguste a l’invasion de 409. (París). Le Roux P. 1999 Briques et tuiles militaires dans la Pèninsule Ibérique: Problèmes de production et de diffusion. In M. Bendala, C. Rico & L. Roldán (eds.) El ladrillo y sus derivados en la época romana. (Monografías de Arquitectura Romana 4, Madrid): 111-124. Luik M. 1997a Fibeln vom Typ Alesia aus dem römische Lagern um Numantia. Archäologische Korespondenzblatt 27: 463-479. Luik M. 1997b Die römische Militäranlagen der Iberischen Halbinsel von der Zeit der Republik bis zum Ausgang des Prinzipats. Ein Forschungüberblick. Jahrbuch RömischGermanischen Zentralmuseum Mainz 44: 213-275. Luik M. 1999 Neue Forschungen zu den römischen Lagern bei Numantia provinz Soria (Spanien). In N. Gudea (ed.) Roman Frontier Studies XVII. Proceedings XVIIth.

785

Limes XVIII

International Congress of Roman Frontier Studies (Zalău): 261-266. Martín Valls R., Delibes de Castro G. & Mañanes T. 1975 Sobre los campamentos de Petavonium. (Studia Archeologica 36: Valladolid). Martín Valls R., Romero Carnicero Mª V. & Carretero S. 1998 – forthc. Marcas militares en material de construcción de Petavonium. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Martínez García A.B. 1999 El vidrio en el campamento romano del Ala II flavia hispanorum civium romanorum en Petavonium. (Zamora). Miguel F. & García Marcos V. 1993 Intervención arqueológica en el patio del Centro Cultural Pallarés (León). Numantia 4: 175-206. Morales Hernández F. 1998 – forthc. La circunvalación escipiónica de Numancia: viejos y nuevos datos para una interpretación. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Morillo A. 1991 Fortificaciones campamentales de época romana en España. Archivo Español de Arqueología 64: 135-190. Morillo A. 1992 Cerámica romana en Herrera de Pisuerga (Palencia, España): las lucernas. (Santiago de Chile). Morillo A. 1993a Campamentos romanos en España a través de los textos clásicos. Espacio, Tiempo y Forma. Historia Antigua, serie II, 6: 379-397. Morillo A. 1993b Una nueva producción de lucernas en la península ibérica: el taller militar de Herrera de Pisuerga (Palencia, España). Trabalhos de Antropologia e Etnologia 33 (1-2): 351-361. Morillo A. 1996 Los campamentos romanos en la Meseta Norte y el Noroeste: ¿un limes sin frontera? In C. Fernández Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 77-83. Morillo A. 1998a – forthc. Conquista y estrategia: el ejército romano durante el periodo augusteo y julio-claudio en la región septentrional de la península ibérica. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Morillo A. (ed.) 1998b – forthc. I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Morillo A. 1999 Lucernas romanas en la región septentrional de la Península Ibérica. Contribución al conocimiento de la implantación romana en Hispania. (Monographies Instrumentum 8: Montagnac). Morillo A. 2000a La Legio IIII Macedonica en la Península Ibérica: el campamento de Herrera de Pisuerga (Palencia). In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les legions de Rome sous le HautEmpire. (Lyon): 609-624. Morillo A. 2000b – forthc. Neue Forschungen zu römischen Lagern der iulisch-claudischen Zeit in Nordspanien. Bonner Jahrbücher 200. Morillo A. & García Marcos V. 2000a Nuevos testimonios acerca de las legiones VI Victrix y X Gemina en la región septentrional de la Península Ibérica. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l’armée romaine: Les legions de Rome sous le Haut-Empire. (Lyon): 589-607. Morillo A. & García Marcos C. 2000b – forthc. Producciones cerámicas militares de época augusteo-tiberiana en Hispania. In Rei Cretariae Romanae Fautorum XXII. (Lyon). Ocharan J. A. & Unzueta M. 1998 – forthc. Andagoste (Cuartango, Alava): un nuevo escenario de las guerras

de conquista en el norte de Hispania. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Del Olmo J. 1994/95 Arqueología aérea en tres núcleos campamentales romanos de Zamora y León. Brigecio 4-5: 109-118. Pamment Salvatore J. 1996 Roman Republican castramentation. A reappraisal of historical and archaeological sources. British Archaeological Reports Int. Ser. 630. (Oxford). Peralta E.1999a Los castros cántabros y los campamentos romanos de Toranzo y de Iguña. Prospecciones y sondeos (199697). Las Guerras Cántabras. (Santander): 201-276. Peralta E. 1999b El asedio romano del castro de La Espina del Gallego (Cantabria) y el problema de Aracellium. Complutum 10: 195-212. Pérez González C. 1989 Cerámica romana de Herrera de Pisuerga (Palencia, España): la terra sigillata. (Santiago de Chile). Pérez González C. 1995a Una testera militar de caballo en bronce de época augustea en España. Anuario de la Universidad Internacional SEK 1: 53-68. Pérez González C. 1995b Un taller de útiles óseos de la Legión IV Macedónica. Revista de la Universidad SEK: 90-102. Pérez González C. 1996 Asentamientos militares en Herrera de Pisuerga. In C. Fernández Ochoa (co-ord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 91102. Pérez González C. 1998 Pisoraca (Herrera de Pisuerga): Urbanismo militar y civil de época romana. In A. Rodríguez Colmenero (co-ord.) Congreso Internacional: los orígenes de la ciudad en el Noroeste hispánico. (Lugo): 535-558. Richmond I.A. 1931 Five town walls in Hispania Citerior. Journal of Roman Studies 21: 86-100. Rodríguez Colmenero A. (co-ord.) 1996 Lucus Augusti I. El amanecer de una ciudad. (La Coruña). Rodríguez Colmenero A. 1998 – forthc. El campamento auxiliar de “Aquis Querquennis”, Baños de Bande (Ourense). In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Rodríguez Colmenero A. & Carreño C. 1992 Sobre Paulo Fabio Máximo y la fundación de Lucus Augusti, In Finis Terrae. Estudios en Lembranza do Prof. Dr. Alberto Balil (Santiago de Compostela): 389-416. Rodríguez Colmenero A. & Herves F.M. 1992 Aquis Querquennis. Campamento romano y ciudad-mansión viaria. Porto Quintela - O Baño – Mugueimes (Ourense). Rodríguez Colmenero A. & Vega Avelaira T. 1997 Equipamiento militar del campamento romano de Aquæ Querquennæ (Portoquintela, Orense, España). Journal of Roman Military Equipment Studies 7: 21-36. Roldán J. M. 1974 Hispania y el ejército romano. (Salamanca). Romero Carnicero Mª V. 1989 Las lucernas republicanas de los campamentos de Numancia. Boletín del Seminario de Estudios de Arte y Arqueología LV: 257-290 Romero Carnicero M.V. & Carretero S. 1997 Cerámica del campamento romano del Ala II Flavia en Petavonium (Rosinos de Vidriales, Zamora, España). In Acta RCRF 35: 55-61. Sánchez-Palencia J. 1986 El campamento romano de Valdemeda, Manzaneda: ocupación militar y explotación aurífera en el Noroeste peninsular. Numantia II: 227-231. Sande Lemos F., Martins M., Oliveira Fontes L. F., Freitas Leite J. M. & Cunha A. 1998 – forthc. A Muralha

786

Angel Morillo & Victorino García-Marcos: Twenty years of Roman military archaeology in Spain Romana de Bracara Augusta. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Sanmartí E. 1985 Las ánforas romanas del campamento numantino de Peña Redonda (Garay, Soria). Empúries 47: 130-161. Sanmartí E. 1992 Nouvelles donées sur la chronologie du camp de Renieblas V à Numance. Documents d’Archéologie Méridionale 12: 417-430. Sanmartí E. & Principal J. 1997 Las cerámicas de importación, itálicas e ibéricas, procedentes de los campamentos numantinos. Revista d’Arqueología de Ponent 7: 35-75. Schulten A. 1927 Numantia. Die Ergebnisse der Ausgrabungen 1905-1912. III. Die Lager des Scipio. (München). Schulten A. 1928 Campamentos romanos en España. Investigación y Progreso 5: 34-36. Schulten A. 1929 Numantia. Die Ergebnisse der Ausgrabungen 1905-1912. IV. Die Lager bei Renieblas. (München). Ulbert G. 1984 Cáceres el Viejo. Ein spätrepublikanisches Legionslager in Spanish-Extremadura. Madrider Beitrage XI. (Mainz). Unzueta M. & Ocharan J.A. 1999 Aproximación a la conquista romana del Cantábrico oriental: el campamento y/o campo de batalla de Andagoste (Cuartango, Alava). In Regio Cantabrorum (Santander): 125-142. Vidal J. & García Marcos V. 1996 Novedades sobre el origen del asentamiento romano de León y de la Legio VII Gemina. In C. Fernández Ochoa (co-ord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 147-155. Yañez C., Bosch J., Ruf M., Mª. A., Solé X., Vila A. & Llovera X. 1997 El Roc d’Enclar. Un ejemplo de las influencias del mundo romano en los Pirineos. Siglos IV-VI. In Congreso Internacional “La Hispania de Teodosio”. (Valladolid-Segovia): 739-751.

787

Limes XVIII

Fig. 1. Military settlements during the Republican period in Hispania.

Fig. 2. Legionary fortress and auxiliary forts during the Cantabrian Wars and the Julio-Claudian period in Spain (29BC - AD 69/70).

788

Angel Morillo & Victorino García-Marcos: Twenty years of Roman military archaeology in Spain

Fig. 3. Legionary fortress and auxiliary forts between AD 74/75 and the mid-3rd century in Spain.

Fig. 4. Military troops and urban fortifications in northern Hispania during the late Roman period.

789

Limes XVIII

790

The legionary fortress of VI Victrix at León (Spain). The new evidence (1995-2000) Victorino García-Marcos & Angel Morillo Archaeological excavation over the past few years in León has demonstrated the origins of the Roman establishment. It confirms the hypothesis that the camp of legio VI Victrix preceded that of legio VII Gemina. Legio VI fought in the Cantabrian Wars (29-19BC), remaining in Hispania thereafter. The archaeological evidence proves the existence of two earlier camps. The earliest dates to the end of the C1st BC and the beginning of the following century. The second camp appeared soon after and lasted until nearly the arrival of the legio VII Gemina. The Augustan camp The remains only partially survive, since it was razed by the later constructions. Nevertheless, it is possible to recognize two types of defenses. In the first there was dug a wide 'V' shaped ditch, between 5m and 6m wide and in a east-west direction (fossa fastigata) and into the natural gravel levels. The vallum, maybe of square section, was carried on the gravel levels. It was located behind the ditch and, as preserved, measures c.2.7m wide x 0.75m high. It was built between two walls formed by horizontal wooden planks and vertical posts. The core was filled with gravel extracted in the digging of the ditch. This type of defence ('Holzerdemauer' in the German and 'box rampart' in the English literature), is typical of Roman camps of the late C1st BC and the first half of the following century. In the vallum there are no remains of a wooden lattice. Such a lattice is usually characteristic of this type of construction, because it creates stability. Finally, the upper part of this wall should have had a parapet with an appearance similar to that of the most common models in this type of structures. The Julio-Claudian camp The defences of the first camp were sealed by a series of deposits which were to level the site. Upon them a new vallum was placed of turf blocks (caespites) which were easily obtained from the water meadows that surrounded the camp. This type of construction, in use from Republican times, is documented at military establishments on the Germanic and British limes in the C1st and C2nd AD. Little is known about the external face of the defences as it was completely destroyed when legio VII built the defences of its camp at the end of the C1st AD. Nevertheless, we think that they were of the 'walls of double lining' type. Their width varies between 3.5m and 4m, dimensions found at examples in Germany and Britain. Inside the vallum, evidence of a wooden lattice does not exist, at least at the moment. The core was filled with material dug out from the previous phase, as shown by the numerous ceramic fragments, fauna, bricks and tiles.

It is known that the reasons which initiated the wars against the Cantabrian and Asturian peoples were linked to the policy undertaken by Augustus – one designed to establish natural frontiers for all the empire - after the defeat of Mark Antony at Actium. In the case of Hispania, we could also add the exploitation of gold deposits which became necessary in order to maintain Augustus’ new monetary policy based around the aureus. Nevertheless, the profit of this resource took years to realise, needing at the beginning surveys and explorations of primary and secondary sites, as well as the planning of a road system and hydraulic infrastructure, necessary to begin its systematic exploitation (Domergue 1986: 32-33; Fernández-Posse & Sánchez-Palencia 1988: 152-176, 218222).

during the years immediately following. Then a new situation arose, one which might even constitute a real “limes without frontier”, an attempt at what Rome would apply a few years later in the northern limits of the empire (Morillo 1996: 81). To achieve this goal, the Romans ensured control of the passes into the mountains and established three permanent legionary fortress in an area which connected the plateau with the mountains. In addition to these fortresses there were constructed a number of forts belonging to several auxiliary troops. Thanks to the accounts of Strabo (Geog. III.4.20) and Tacitus (Anns. IV.5.1), we have proof of the condition of this zone in the years immediately subsequent to the war. In fact, we know that the exercitus hispanicus was already reduced to three legions by this time. Two legions were stationed in Asturia, under one commander, whereas the rest was deployed in the Cantabria, under another commander. Numismatic and epigraphic finds, especially those of the later years, have allowed identification and location of these legionary contingents.

The Cantabrian wars forced the movement of an important contingent of troops to the peninsula. The legions I Augusta?, II Augusta, IIII Macedonica, V Alaudae, VI Victrix, IX Hispana and X Gemina were doubtless among those legions (Schulten 1943: 202-205; García y Bellido 1961: 116-128; Roldán 1974: 188-209; Le Roux 1982: 61). Once the conflict ended, the majority of these units left Hispania for other frontiers.

The legio IIII Macedonica would have had its permanent base at Herrera de Pisuerga, guaranteeing control of the eastern part of the Cantabrian mountain range (Pérez González 1989; 1996: 91-102; Morillo 1992; 1996: 77-78; 1999a: 333; 2000). It remained there until the end of Caligula’s reign and then moved to Germania Superior, a

Although the conflict officially ended in 19BC, the consolidation and real control of the region took place 791

Limes XVIII

fact which is already accounted for in AD43 at Mogontiacum (Mainz).

that locates the legio VI in a double camp with the legio X (Morillo & García Marcos 2000a: 591).

Asturica Augusta is a singular case. Its first phase of occupation was determined by the camp of legio X Gemina, whose existence has been recently proved without doubt, since two fossae fastigatae belonging to the defensive system and several building remains and inscriptions have been found, as well as an important number of objects of clear military affiliation (García Marcos 1996: 69-70; García Marcos & Vidal 1995b: 373, 1996: 135; Fernández González 1996: 85-90; Morillo 1996b: 78-80; 1999: 333; Burón 1997: 15-16, 71). The presence of this legion would guarantee both penetration into the mountains of León and El Bierzo and the preparation of the necessary infrastructure for mining. Once its stay at Asturica ended, the legio X had its only permanent camp at Rosinos de Vidriales until it left Hispania for the first time in AD63 for Carnumtum (Morillo 1991: 164-166; Carretero 1993: 47-73; Carretero & Romero 1996). It returned to the peninsula in 68, almost certainly to its old camp, only to leave finally in 69/70, together with the VI Victrix and I Adiutrix, for the northern limes, settling first at Arenacum and then Noviomagus at the start of the C2nd.

Of all the legionary units that took part in the bellum cantabricum, the legio VI would be the one that remained settled in Hispania for the longest period of time, seeing the IIII leave first, followed later by the X, although the latter briefly came back not long afterwards. Legio VI had firmly supported Galba against Nero, and their contingents might have had a predominant rôle in the formation of the legio VII Galbiana, in honor of its founder first and then Gemina. In 69/70, it definitively left Hispania with the X Gemina and the I Adiutrix for the Rhine, settling at Novaesium at the end of 70 (Tacitus Hist.V.22). Alföldy (1969: 115) and Le Roux (1982: 106), followed García y Bellido (1961: 125) in considering the possibility of a legionary base at León prior to the stationing of the legio VII Gemina around AD74/75. This line of argument was essentially based on the text of two inscriptions. The first one, found in Rímini and dedicated to a primipilus called Marcus Vettius Valens, refers, for the first and only time, to legio VI Victrix participating in the repression of an Asturian uprising during Nero’s reign (CIL XI 395; Roldán 1974: 450-451, n.543). The second of the texts, now lost, was discovered in the walls of León. It was provided by the commader L. Pupius Praesens (CIL II 2666; Diego 1986: n.70). In the inscription mention is made of a legion whose name ended in –trix. It could only refer to the I Adiutrix or, or with more probability, the VI Victrix. To support this point, Le Roux relied on a series of ceramics dated to the Claudian discovered by García y Bellido during his excavations in the 1960s (1970: 580581) and on the Tabla I Itinerario de Barro, a document that presents certain problems of authenticity.

The location of the fortress of the second Asturian legion the VI Victrix - according to Strabo, seems to have been resolved in recent years thanks to the discovery of a succession of traces in León, which show that this unit had been settled a long time before the arrival of the legio VII Gemina, on a smooth hillock located in the watershed between the Bernesga and the Torío rivers. It is a shelf, the last terrace formed by fluvial activity, which slightly dominate the river banks. In fact, the site where the modern city is situated, a genuine crossroad between the table-land plateau and the León central mountain range, offered topographic and spatial conditions with a high strategic value which was not unnoticed by the Romans. These features coincide with those of the contemporary sites of Astorga and Herrera de Pisuerga, which indicates that the morphology of the site as well as its geo-strategic value were deciding factors for its choice.

To the material retrieved by García y Bellido in his excavations at León, one can add those from several interventions carried out in the 1980s, such as those at La Abadía Street (Avello 1985: 132) or at the church of San Salvador de Palat de Rey (Miguel 1988). These show that the stratigraphic sequence ascribed to the occupation of legio VII lay on top of others with a clearly early chronology. The archaeological excavations that we carried out at the beginning of the last decade - Edificio Pallarés and Casa Botines, Polígono de La Palomera, Serranos Street and Vizconde Square, among others corroborate once and for all an earlier pre-legio VII occupation of a military character, although its main features still have to be defined (García Marcos 1996: 7680; 1997: 303-314; García Marcos & Miguel 1996: 355360; Vidal & García Marcos 1996: 150-151; Morillo 1999a: 334) (Fig. 1).

According to Roldán (1974: 200, 206), legio VI Victrix would not have taken part in the whole campaign against the Cantabrian and Asturian peoples. It just made an appearance in the war in 25BC. Roldán based this assumption on the fact that its veterans were not among those who took part in the founding of Emerita, although the legion does appear together with the legions IV and X in the coins of the founding of Caesaraugusta. Once the war ended, there is absolutely no doubt about legio VI's constant presence in the Astur sector. Some detachments of it (vexillationes) might have been situated during at one time at Lucus Augusti, as a recently discovered inscription in the city reveals (Rodríguez Colmenero 1996: 130). We have recently expressed some doubts about the Roldán’s hypothesis which has never been proved archaeologically,

However, the excavations between 1997 and 1999 of a vast plot against the inner face of the north of the late Roman wall, very close to Puerta Castillo - the porta decumana of the camp of legio VII - has provided conclusive evidence about the features of the beginnings of the Roman presence

792

Victorino García-Marcos & Angel Morillo: The legionary fortress of VI Victrix at León (Spain). The new evidence in the city of León (Fig. 2).1 Given these facts, the existence of two bases prior to the legio VII’s has been recognized, which cover, without interruption, the period between the final years of the C1st BC and the beginning of the Flavian period.

During the excavation of two sites located around Vizconde Square, also situated in the area close to Puerta Castillo, a series of trenches and post-holes turned up. They were carved in to the natural sub-stratum and were doubtless intended for fixing wooden buildings. In Serranos Street, several levels of dirt-fill were discovered which may be interpreted as open spaces. Traces of this earliest occupation have also appeared at the other end of the city, where the excavation of the Palat de Rey church revealed charred timbers, that could have belonged to hearths (Miguel 1988).

The Augustan camp The remains of this camp have been partially documented during the course of the archaeological field work because there are some structures on top surviving beneath later phases of occupation. The existence of part of the camp defences, situated around 15m of the inner side of the late Roman wall, has been recognized. First of all, a wide ditch in an east-west direction with a V-shaped profile (fossa fastigata) was dug into the natural gravel with a width ranging from 5 to 6m. This variation was because the upper part of the northern edge of the ditch did not offer a rectilinear profile. It opened up slightly from its eastern half.

The field work carried out at Edificio Pallarés and the nearby Casa Botines has shown that, from its beginnings, the Roman military establishment was not only confined to the upper part of the hill, but also encompased the eastern side of the left meadow of the River Bernesga, very close to the shelf of the fluvial terrace where the walled city is situated. One or several timber constructions whose remains occupied the whole excavated area were built on a level surface of pebbles that could be interpreted as a first paving, although their degree of preservation differs. The limited scope of the survey undertaken in 1990 led us to interpret those remains as a raised floor (García Marcos & Miguel 1996), the systematic field work allowed visualization of a timber platform that possibly paved a set of rooms whose boundaries exceeded the area presently being excavated. This plank floor was supported by horizontal timbers linked with timber uprights set into foundation trenches into the natural soil. Other kinds of wooden buildings were constructed with woven wattle panels plastered with daub. There is an almost complete absence of (iron) nails, but the existence of some traces in the timbers allows us to suppose a system of joints similar to those documented in contemporary Roman camps (Johnson 1983: 97ff.).

The vallum, with an apparently quadrangular section, 2.7m wide x 0.75m high, was built over a natural layer of gravel, immediately behind the slope of the ditch. For its construction, two timber walls were built at either end, in which planks arranged horizontally were combined with vertical reinforcing posts situated some distance away from each other; the inner space was filled with the gravel extracted from the ditch (Fig. 3). This kind of defence, a so-called box-rampart (known in German as Holzerdemauer) a common system in Rhineland forts and camps dating between the late C1st BC and the first half of the subsequent century. Their thickness fluctuates around ± 3m in the majority of the cases (Jones 1975: 14-24; Johnson 1983: 62-63).2 At present, the only excavation trench that has been cut through the remains of the vallum has not allowed proof of the existence of a timber framework, a feature present in most of these constructions meant to guarantee their stability. Last, the upper part of this wall would have had a parapet (lorica) with a layout analogous to that of the models more common in this kind of structure (Jones 1975: 79, fig. 16; Johnson 1983: 61, fig. 36).

Therefore, we are faced with provisional structures that continue the tradition of military construction used on all the limes. We have to resort to contemporary sites on the Rhine and to the subsequent sites on the Danube and in Britain, especially those situated around Hadrian’s Wall, to find the closer parallels that supplement or clarify the interpretation of the remains from León (v. Petrikovits 1975; Jones 1975: 182-201; Johnson 1983: 97 and ss., et al).

Outside the defences, there was a road surface for whose construction there was a clear manipulation of the natural ground level. The surface for vehicular traffic had a lining of slightly curved profile and 6.7m in width, and it was possible to see the remains of several traces from the carts in some places of its route. A small drain, responsible for draining waters from the road, was placed in its northern end (Fig. 4).

In our current state of knowledge it is impossible to specify if this early enclosure belonged to a legion or an auxiliary unit. By the way of hypothesis, we might advance that the beginning of the military occupation in León would have taken place in the final years of the C1st BC. At the moment, it is not possible to establish a relationship between the foundation of the Augustan camp of León and historic events, such as the provincial reorganization of the north-west territories carried out by Augustus after the end of the Cantabrian wars and the redistribution of military contingents of which Strabo and Tacitus inform us (Morillo 1998a: forthc.; Morillo & García Marcos 2000: 590). Only further investigation will clarify this and other

The majority of the finds belonging to this phase have been recovered in the northern sector of the current city walls. 1 It is not possible to present the definitive plan of these remains here because the excavations only recently finished 2 10 Roman feet as Vegetius mentions (De Re Militari III.8): “Opus vero centuriones decempedis metiuntur”.

793

Limes XVIII

questions that arise concerning the rôle of this early Roman site in the pacification and administration of the north-west peninsula.

horizontal courses, of which 12 could be distinguished (Fig. 5). Each layer was separated from the other by stripes of organic and black hue earth, 0.02 - 0.03m thick, very possibly the remains of the grass. Pebbles of medium and small size appeared sporadically as well.

The Julio-Claudian camp The final years of Augustus’ and Tiberius’ reign seem to coincide with significant changes in the military occupation of the north-west (Morillo 1998a - forthc.), a process which might include the beginnings of the exploitation of gold deposits which, according to Domergue, began at this time (1996: 32-33). The camp of legio IIII at Herrera de Pisuerga was modified in ground plan while, at the same time, its early defences were substituted, and the timber constructions replaced with others more permanent elements (Pérez González 1996: 91; Morillo 2000: 617-618). The camp of the legio X at Astorga was dismantled and the first civilian buildings were constructed, starting in this way its activity as an urban nucleus (García Marcos & Vidal 1996: 135; Burón 1997: 27ff; Morillo 1999a: 333ff.; Morillo & García Marcos 2000: 598).

The internal fill of the vallum, with a preserved width of 0.8m, was made up of an earth of brownish-gray hue, several small-sized pebbles and remains of charcoal. The excavation resulted in the find of many fragments of ceramics, fauna and some brick remains, which seem to point to a use of earth from the early enclave for its construction. Just as with the vallum, there were no traces of a wooden framework. So, we see that the maximum preserved width of this defence at its base was 1.8m, measurements that must come close to half its original dimensions. Thus we would have a wall whose width might range from 3.5 to 4m. These dimensions agree with other examples from Germania and Britain (Jones 1975: 18, 69-74; Johnson 1983: 58-59).5 We can say very little about the form of the outer side, since, as stated above, it was completely destroyed when the legio VII built the stone defences. However, we can, without doubt, say that part of the defensive ditches of the second enclosure was used to place their foundations, an aspect that would make their construction easier. That makes it difficult to determine precisely the scheme of the wall superstructure. Taking into account all this data, we can suppose that both faces had a vertical facing, as at the Claudian site of Hod Hill (Johnson 1983: fig. 36) or, more probably, an oblique orientation as at Flavian Chester (Jones 1975: fig. 16).

The camp of León underwent similar transformations, where its defensive ditch and the exterior road paving of the first phase were sealed by a series of layers which served as a base for levelling deposits approximately 0.5m deep. From this level, a new vallum was built with regular blocks of turf (caespites). These materials were easily obtained in the semi-marshy meadows that surrounded the camp. This kind of construction, well described by the author of De munitionibus castrorum (50)3, is depicted in scenes on Trajan’s Column (Hobley 1988: figs. 2.2, 2.3, 2.4). It was already used in Republican times, as seen at the site of Alesia (Bénard 1996: 56; Reddé & v. Schnurbein 1997: 177), being perfectly documented in military sites of the German and British limes in the course of the C1st and C2nd AD. In Hispania, its existence has been recognized at Rosinos de Vidriales where, originally, the camp of the ala II Flavia was surrounded with defences of this kind, which were subsequently replaced by a stone wall (Romero Carnicero & Martínez: 1996), as also happened at other sites, such as Inchtuthil (Pitts & St. Joseph 1985: 59-71).

The internal fort boundary is entirely occupied by the via sagularis. With a thickness of 16.5m, it was built over a few levelling layers, showing a clear horizontal profile and a smooth east-west slope. Its surface, with a wellmaintained appearance, was made up of tamped gravel intermingled with earth.

We have discovered a stretch over 20m long and as high as 1.4m in some places. Although its northern half was destroyed by the trench for the foundations of the wall to the legio VII camp, we have an example of the so-called walls of “double turf revetement” (Jones 1975: 68ff.; Johnson 1983: 59-63). The inner face was vertical or with a slightly oblique tendency towards the interior and had a width that ranged from 0.85 to 1m. It was built with quadrangular turves (0.3 x 0.3 x 0.12m. or 0.4 x 0.3 x 0.12m)4 - known in Spanish as tapines - placed in

More radical changes occurred later in the area occupied by the ancient vallum, with the building of a barrack block (centuria) parallel to the defences of the new enclosure (per scamna), a development which in turn necessitated the construction of important terraced works. In fact, the steep slope resulting from the defensive ditch and the wall of the first settlement caused the north end of the barracks, which was situated in the path of the first, to be on top of a wide base of pebbles placed in its interior. Over it, a second stretch of small ashlar limestone was erected, with a good bond and clay, reinforced with small pebbles. Several of the blocks had iron clamps in their sides in order to reinforce the links. With a width of 0.5m and a height close to 1m, there were the remains of four courses. But only the

3 Pseudo-Hyginus: De munitionibus castrorum, (50): “Vallum loco suspectiori extrui debet cespite aut lapide, saxo sive caemento...” 4 These dimensions are not too far from those prescribed by Vegetius (III.8): “caespes autem circunciditur ferramentis, qui herbarum radicibus

continet terran, fit altus semissem, latus pedem, longus pedem semis”. 5 The measurments established by the author of De munitionibus castrorum (50) are slightly lower: “Sufficit latum pedes VIII, altum pedes VI et lorica parva fit.”

794

Victorino García-Marcos & Angel Morillo: The legionary fortress of VI Victrix at León (Spain). The new evidence upper one would go to the exposed wall, since its remaining elevation was hidden by the deposits which were used as a base to the via sagularis and as a levelling to the interior of the barrack hut. For their part, the partition walls which separated the several internal rooms (contubernia) and which had a slightly smaller width and were placed in most of its trajectory over the old vallum, showed just two or three courses, with a height ranging from 0.5 to 0.6m. This arrangement was different in zones of joint with the north wall, where the need for saving the unevenness caused by the ditch made them achieve the same depth (Fig. 6). With respect to the stretch of wall at the south end, we were only able to discern the upper side, although everything suggests that its construction features were similar. What in reality was the remainig elevation of the wall (consisting of mud walls which might have also been combined with adobe in some places) was set on these bases of small ashlar.

course of the excavations shows. It was at that time that the problem seems to have been handled, beginning with the deliberate raising of this area through dumping, especially clays of waterproof material. Thus, by the end of that century, the ground surface was raised more than 2m above the natural level, located at -4.7m below the present level. The excavation of the layer formed by the decomposition of the construction timbers of the previous phase provided a large quantity of materials, among which a good sample of terra sigillata italica vessels stands out. The chronology of these materials seems to cover the period from the end of Augustus or the beginning of Tiberius to Claudius’ reign. A group of pieces which seem to be a local product, although they have common formal features of sigillata italica, deserves mention. In this group there are plates and different kind of cups of poor firing and low quality glosses. They were made by a potter called C. Licinius Maximus, unknown until now, and who supplied his products to the unit settled at León.

The remains of the four contubernia, although significantly altered by the super-imposition of the legio VII constructions, show shared characteristics. They were rectangular in shape (8.3 x 3.77m) with an inner partition wall of limestone ashlar that separated the accommodation area (papilio) from the space used for storing the soldiers’ impedimenta (arma). The first unit shows slightly larger dimensions than the other three; a very plastic and decanted clay was used for the various floors. An open space in their western wing, which was opposite to another similar wing through which it was possible to gain direct access to the via sagularis, was the communication link between them. There is no evidence of the arcade porch typical of contubernia after the Augustan-Tiberian period (v. Petrikovits 1975: 36ff.; Davison 1989: I. 4-5, 18-20; II. figs. 1.2 to 1.8). Because of the absence of this element and in the light of the interior layout of the contubernia, we may consider them to belong to Davison’s type B, which describes chronologically proximate forts such as Oberaden, Rödgen and Dangstetten (Davison 1989: I. 1819, II. fig. A). The centurion’s rooms, not found during the excavation, most likely were located on the west side.

At the foot of the southern side of what subsequently became the fortified enclosure of the legio VII, close to the porta praetoria, there was a complex devoted to metallurgical activity - for the treatment of iron to be precise - as the features of several working rooms, as well as the materials associated with them, seem to show. Its period of activity seems to be relatively brief, covering the second-third of the C1st AD (Campomanes 1999: 269279). A large rectangular structure - 35m long and 12m wide has been discovered to the south-east of the camp, outside its wall. It had opus caementicium walls, planked with timber. The width was 0.6m, the total elevation 1.6m and its exposed side 1.3m, which corresponds to the rest of the foundations. The interior had two superimposed mortar pavings that guaranteed its impermeability. The function of the floors is still problematic, although all traces point to their being used as part of a water tank. Materials associated with the layers which appeared when recovering the interior, show that it stopped being used towards the mid-C1st AD and the dumping continued until the Flavian period or beginning of the C2nd (Vidal & García Marcos 1996: 151).

Although, at present, the finds of Puerta Castillo are the clearest example of this second camp phase, several excavations undertaken in the interior of the walled enclosure - Vizconde, Santo Martino and Conde Luna Squares; Serranos, San Pelayo or Cardenal Landázuri Streets - have repeatedly corroborated the continuity of the Roman occupation at León. Nevertheless, as happened in the Augustan period, the preserved construction remains are fragmentary and scattered, which in many cases makes it difficult to speculate about their function.

The central years of the C1st AD seem to coincide, as archaeology is repeatedly showing, with changes of some significance to both the interior and the exterior of the walled enclosure. This involved the alteration of some structures and the destruction of others. We do not yet have proof of the real meaning of these alterations nor the reasons which caused them, although we cannot forget that the Claudian-Neronian period, followed by the Civil War of 68-69, coincided with a reorganization of great significance that affected the arrangement of the military contingents stationed in north-west Spain. The legio IIII left for the Rhine between 39 and 43 and a cavalry auxiliary unit, the ala Parthorum, took its place some years

Occupation continued, likewise, in the eastern end of the left meadow of the River Bernesga (the area of the Pallarés and Botines buildings). The use of this site was conditioned over a long period by superficial (flooding) or subterranean (water-table) streams. These difficulties were to carry on until the mid-C1st AD, as the sedimentation, in a humid medium, of several deposits discovered during the 795

Limes XVIII

later and stayed at Herrera de Pisuerga for the rest of the C1st and the beginning of the C2nd, dependent on the legio VI first and then on the VII (Pérez González 1996: 91-102, Morillo 2000: 619). The legio X suffered a similar fate some years later, going to the Danube in 63, although it came back to the peninsula in 68. Its definitive departure together with the legio I Adiutrix (formed few years before and sent by Vitellius to Hispania, although at present it is unknown where it could have been settled during its brief stay) and the VI Victrix, occurred in 69/70 after the serious events of the Batavian revolt. One must not forget the uprising of the Asturian peoples in the 50s, whose only proof is the already mentioned inscription of the primipilus of the legio VI, M. Vettius Valens. Its intensity and duration are not known.

who supplied both the military and the civilian population settled in the canabae. This evidence has not been reinforced by the appearance of brick marks, so frequent with the arrival of the legio VII. This fact is not, however, an unusual phenomenon, since the first marks on military construction material appear in the Rhine around the middle of the C1st AD (Szylagyi 1972: col. 433; Le Bohec 1992: 45). Bibliography Alföldy G. 1969 Fasti Hispanienses, Senatorische Reichbeamte und Offiziere in den Spanischen Provinzen des Römischen Reiches von Augustus bis Diokletian. (Wiesbaden). Avello J.L. 1985 Calle de La Abadía y Plaza de Santo Martino, León. Arqueología 83. Memoria de las Excavaciones Programadas en el año 1983. (Ministerio de Cultura, Madrid 9: 132. Benard J. 1996 Alésia: du mythe a l’archéologue. In M. Reddé (ed.) L’Armée romaine en Gaule (Paris): 40-65. Burón M. 1997 El Trazado Urbano en las proximidades del Foro en Asturica Augusta. La casa del pavimento de opus signinum. (Arqueología en Castilla y León 2: Valladolid). Campomanes E. 1999 Hallazgo de un complejo metalúrgico romano en la ciudad de León. Excavación en la Calle Plegarias con vuelta a la Calle Ramiro III en la ciudad de León. Lancia 3: 269-279. Carretero S. 1993 El cuadrante noroeste peninsular en época romana: los efectivos militares y sus establecimientos. Brigecio 3: 47-73. Carretero S. & Romero Carnicero Mª V. 1996 Los campamentos romanos de Petavonium (Rosinos de Vidriales, Zamora). (Zamora). Carretero S., Romero Carnicero Mª V. & Martínez García A.B. 1999 Les estructuras defensivas del campamento del Ala II Flavia en Petavonium (Rosinos de Vidriales, Zamora). In R. De Balbín & P. Bueno (edd.) II Congreso de Arqueología Peninsular. IV. Arqueología romana y medieval. (Zamora): 183-194. Davison D.P. 1989 The barracks of the Roman army from the 1st to 3rd centuries A.D. A comparative study of the barracks from fortresses, forts and fortlets with an analysis of building types and construction, stabling and garrisons. British Archaeological Reports Int. Ser. 472. (Oxford). Diego F. 1986 Inscripciones romanas de la provincia de León. (León). Domergue C. 1996 Dix-huit ans de recherche (1968-1986) sur les mines d’or romaines du nord-ouest de la Péninsule Ibérique. In I Congreso Internacional Astorga Romana II (Astorga): 7-101. Fernández Ochoa M.C. & Morillo A. 1999 La tierra de los astures. Nuevas perspectivas sobre la implantación romana en la antigua Asturia. (Gijón). Fernández-Posse Mª.D. & Sánchez-Palencia J. 1988 La Corona y el Castro de Corporales II. Campaña de 1983 y prospecciones en La Valdería y La Cabrera. (Excavaciones Arqueológicas en España 153, Madrid). Figuerola M. & González E. 1999 Las contramarcas monetarias de la Legio VI en Hispania. Lancia 3: 127140. García y Bellido A. 1961 El “Exercitus Hispanicus” desde Augusto a Vespasiano. Archivo Español de Arqueología 34: 114-

New questions have arisen concerning the Julio-Claudian fortress of León as a result of the current investigation.6 We do not know the entirety of its extent and cannot ascertain whether all of the data found to date comes from the camp defences. Thus, we must consider how the typology of older camps (eg. Haltern, Oberaden, Anreppen or Vindonissa) deviates from the regularity that characterizes settlements built subsequent to the mid-C1st AD. Nonetheless, the distribution of the finds in León seems to show a layout sufficiently similar to that of the later castra of the legio VII. The topographic conditions would determine that the site occupied for both camps was the ideal place. Archaeological proof seem to have solved once and for all the question concerning the military unit that built and occupied this camp. To the scarce evidence that points to the legio VI Victrix as the occupying unit, we should add several finds made in recent years. A recently discovered coin of Claudius counter-marked by this unit is remarkable and we would have to add it to another coin of unknown origin from the collection of the Museum of León. Both are among the few examples discovered in the peninsula (Figuerola & González 1999: 128, nos. 1, 7; Morillo 1999b). There is a lamp from the excavations of the Edificio Pallarés that has an incomplete mark in relief in which, with square capital letters and separated by punctuation marks, is the legend L.V.¿I? (Fig. 7). Lamps that have legionary marks are few, although we have some examples that show very similar features to the ones on our lamps (Morillo 1999a: 296-297, no. 40, fig. 170). In close connection with the lamp, we must include the sigillata vessels signed by C. Licinius Maximus, of which at present 15 examples have been documented, always of plain ware shapes (Morillo & García Marcos 2000 forthc.). This potter, unlike L. Terentius with the legio IIII Macedonica (Pérez González 1989: 217), did not accompany his marks with the military unit settled here, so that more than his figlinarius we can think of a civilian 6 As we have pointed out already, the temporal proximity of these excavations does not allow us yet to present here the definitive plan of the structures.

796

Victorino García-Marcos & Angel Morillo: The legionary fortress of VI Victrix at León (Spain). The new evidence 160. García y Bellido A. 1970 Estudios sobre la legio VII Gemina y su campamento en León. In Legio VII Gemina. (León): 569599. García Marcos V. 1996 La romanización urbana: Asturica Augusta y la implantación romana en León. Arqueo León. Historia de León a través de la arqueología. (León): 69-81. García Marcos V. 1997 Excavaciones arqueológicas de urgencia en la ciudad de León (1993-95). Lancia 2: 303-314. García Marcos V. 1998 – forthc. Novedades acerca de los campamentos romanos de León. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). García Marcos V. & Miguel F. 1996 A new view on the military occupation in the north-west of Hispania during the first century: the case of León. In W. Groenman-van Waateringe, B. L. Van Beek, W. J. H. Willems & S. L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 355-360. García Marcos V. & Vidal J. 1995a Asturica Augusta y Castra Legionis VII Geminae. In Catálogo de la Exposición Astures. (Gijón): 113-128. García Marcos V. & Vidal J. 1995b Recent archaeological research at Asturica Augusta. In Proceedings of the British Academy 86 (Social complexity and the development of towns in Iberia from the copper age to the second century AD). (London): 371-394. García Marcos V. & Vidal J. 1996 Asturica Augusta: recientes investigaciones sobre su implantación y desarrollo urbano. Los Finisterres Atlánticos en la Antigüedad. Época Prerromana y Romana. (Gijón): 135-145. González Fernández Mª L. 1996 Consideraciones sobre el origen militar de Asturica Augusta. In C. Fernández Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 85-89. Hobley B. 1988 The evidence for the form and appearance of turf and timber defences of Roman forts in the late first century, based on experiments at the Lunt Roman fort. In Portae cum turribus. Studies of Roman fort gates. British Archaeological Reports Brit. Ser. 206 (Oxford): 25-61. Johnson A. 1983 Roman forts of the 1st and 2nd centuries AD in Britain and the German provinces. (London). Jones M. J. 1975 Roman fort-defences to A.D. 117, with special reference to Britain. British Archaeological Reports Brit. Ser. 21. (Oxford). Lander J.1984 Roman stone fortifications. Variation and change from the first century A.D. to the fourth. British Archaeological Reports Int. Ser. 206. (Oxford). Le Bohec Y. 1992 Les estampilles de l’armée romaine sur briques et tuiles. Epigraphica 54: 43-62. Le Roux P. 1982 L'Armée romaine et l'organisation des provinces ibériques d'Auguste a l'invasion de 409. (París). Miguel F. 1988 Informe de la Segunda Fase de Restauración de la iglesia de San Salvador de Palat de Rey: Excavación extensiva (Valladolid). Morillo A. 1991 Fortificaciones campamentales de época romana en España. Archivo Español de Arqueología 64: 135-190. Morillo A. 1992 Cerámica romana en Herrera de Pisuerga (Palencia, España): las lucernas. (Santiago de Chile). Morillo A. 1996 Los campamentos romanos en la Meseta Norte y el Noroeste: ¿un limes sin frontera?. In C. Fernández

Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 77-83. Morillo A. 1998a – forthc. Conquista y estrategia: el ejército romano durante el periodo augusteo y julio-claudio en la región septentrional de la península ibérica. In A. Morillo (ed.) I Congreso de Arqueología Militar Romana en Hispania. (Madrid). Morillo A. 1998b Asentamientos militares y civiles en el origen del fenómeno urbano en el noroeste peninsular. In A. Rodríguez Colmenero (coord.) Congreso Internacional: los orígenes de la ciudad en el Noroeste hispánico (Lugo): 339-354. Morillo A. 1999a Lucernas romanas en la región septentrional de la Península Ibérica. Contribución al conocimiento de la implantación romana en Hispania. (Montagnac). Morillo A. 1999b Contramarcas militares en monedas de la submeseta norte. Algunas consideraciones generales. In Anejos Archivo Español de Arqueología XX (Madrid): 71-90. Morillo A. 2000 La Legio IIII Macedonica en la Península Ibérica: el campamento de Herrera de Pisuerga (Palencia). In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l'armée romaine: Les legions de Rome sous le HautEmpire (Lyon): 609-624. Morillo A. & García Marcos V. 2000 Nuevos testimonios acerca de las legiones VI Victrix y X Gemina en la región septentrional de la Península Ibérica. In Y. Le Bohec (ed.) Deuxième congrès de Lyon sur l'armée romaine: Les legions de Rome sous le Haut-Empire (Lyon): 589-607. Morillo A. & García Marcos C. 2000 – forthc. Producciones cerámicas militares de época augusteo-tiberiana en Hispania. In Rei Cretariae Romanae Fautorum XXII. (Lyon). Pérez González C. 1989 Cerámica romana de Herrera de Pisuerga (Palencia, España): la terra sigillata. (Santiago de Chile). Pérez González C. 1996 Asentamientos militares en Herrera de Pisuerga. In C. Fernández Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 91102. Petrikovits H. von 1975 Die Innenbauten römischer Legionslager während der Prinzipatszeit. (Abhandlungen der Rheinisch-Westfälischen Akademie der Wissenschaften 56: Opladen). Pitts L.F. & St. Joseph J. K. 1985 Inchtuthil. The Roman legionary fortress. Britannia Monograph Series 14. (London). Reddé M. & Schnurbein S. von 1997 Les nouvelles fouilles d’Alesia (1991-1995). In W. Groenman-van Waateringe, B. L. Van Beek, W. J. H. Willems & S. L. Wynia (edd..) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 175-185. Rodríguez Colmenero A. 1996 Lucus Augusti, capital de la Gallaecia septentrional. In C. Fernández Ochoa (coord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 129-133. Roldán J.M. 1974 Hispania y el ejército romano. (Salamanca). Romero Carnicero M.V. & Carretero S. 1996 Cerámica del campamento romano del Ala II Flavia en Petavonium (Rosinos de Vidriales, Zamora, España). In Acta RCRF 35: 55-61. Schulten A. 1943 Los Cántabros y Astures en su guerra contra Roma. (Madrid).

797

Limes XVIII

Szylagyi J. 1972 Ziegelstempel (Militärziegelstempel). In Realencyclopädie X A: 433-446. Vidal J. & García Marcos V. 1996 Novedades sobre el origen del asentamiento romano de León y de la Legio VII Gemina. In C. Fernández Ochoa (co-ord.) Coloquio Internacional sobre los Finisterres Atlánticos en la Antigüedad. Epoca Prerromana y Romana. (Madrid): 147-155.

798

Victorino García-Marcos & Angel Morillo: The legionary fortress of VI Victrix at León (Spain). The new evidence

Fig. 1. Legionary fortress and auxiliary forts during the Cantabrian wars and the Julio-Claudian period in Spain (29 BC - AD69/70).

Fig. 2. Hypothetical plan of the camp of the legio VII Gemina at León with an indication of the accomodation blocks in the fortress of legio VI Victrix.

799

Limes XVIII Fig. 3. León. The Augustan camp. The remains of the rampart (vallum).

Fig. 4. León. The Augustan camp. The ditch (left) and the external road.

Fig. 5. León. The Julio-Claudian camp. A general view of the turf rampart.

Fig. 7. A fragment of a pottery lamp from the excavations of the Edificio Pallarés with a mark in relief: L.V. I?. We propose its identification with the mark of legio VI (Museo de León). Fig. 6. León. The Julio-Claudian camp. Contubernia of the Legio VI and Legio VII superimposed.

800

Les castra et les castella aux extrémités de l’Empire après la fin de la domination romaine: le Nord-ouest Ibérique et le Haut Rhin au Ve. siècle Jorge López Quiroga & F.Germán Rodríguez Martín L’effondrement du système politique et militaire au début du Ve siècle dans la Gallaecia et les Agri Decumates, deux régions périphériques aux extrémités de l’Empire, à donné lieu a des transformations importantes dans l’évolution des structures du peuplement antique. Néanmoins, les habitats fortifiés en hauteur ne disparaissent pas dans ces deux régions. En revanche, ils semblent se renforcer et même ils peuvent arriver à montrer une grande vitalité au début du Ve siècle. Malgré les différences entre ces deux régions périphériques, ces habitats ont joué un rôle important dans la hiérarchie du peuplement tardo-romain et ils possèdent une double signification: - D’une part, non seulement ils ont joué un rôle géo-politique évident, mais aussi ils ont favorisé le maintien de l’activité économique (et surtout des circuits commerciaux), qui ne peut pas s’expliquer exclusivement dans le cadre local ou régional, et qui est en rapport avec l’évolution des structures du peuplement durant l’Antiquité tardive en Occident. - D’autre part, du point de vue de l’évolution des structures sociales, la fin de la domination romaine donnerait lieu dans ces deux régions à un éclatement du pouvoir, qui aurait comme conséquence le “réveil” des élites locales et la formation des petits centres de pouvoir autonommes aussi bien au niveau local que supra-local. In Gallaecia and in the Agri Decumates, two outlying regions of the Roman empire, the declining political and military system at the beginning of the C5th AD gave place to important transformations in the evolution of the structure of ancient settlement. However, fortified establishments did not disappear in these two regions. On the contrary, they were reinforced and they show a great vitality at the beginning of the C5th. These castra and castella played an important part in the hierarchy of late Roman settlement, possessing a double meaning: - on one hand, they not only had an evident geo-political rôle but rather they favored the maintenance of economic activity (and mainly of the commercial circuits). This does not explain exclusively in a local or regional mark, and it is in connection with the evolution of the settlement structures during the late antiquity in the west. - on the other hand, from the point of view of the evolution of the social structures, the end of the Roman domination gave place in the two regions to a fragmentation of power. The consequences of this was an awakening of the local elites and the formation of a small autonomous centres of power as much at local level as supra-local.

Aussi bien dans le Nord-ouest ibérique que dans la région du Haut-Rhin, le passage du IVe au Ve siècle est marqué par de profonds changements et bouleversements du point de vue de l’évolution des structures socio-politiques. Vers la fin de 406, les Vandales, les Alanes et les “Suèves” traversent la frontière du Rhin et, en 409 ils arrivent en Espagne.1

du peuplement tardo-romain. Plus précisément, l’occupation ou la réoccupation des sites en hauteur durant la première moitié du Ve siècle va devenir l’une des caractéristiques du peuplement rural dans cette période, aussi bien au Nord-ouest ibérique que dans le Haut-Rhin. Bien évidemment, le contexte socio-politique dans ces deux régions n’est pas tout à fait le même. Néanmoins, leur caractère frontalier et périphérique par rapport aux zones centrales de l’Empire, en même temps qu’une géographie assez montagneuse, vont favoriser le développement d’un peuplement en hauteur dès la fin du IVe et surtout au début du Ve siècle avec des différences, bien sûr, mais aussi avec des similitudes qu’il faut expliquer au-delà du cadre régional pour les situer dans un contexte européen.

En ce qui concerne la région frontalière du Rhin et du Danube, ceci est précédé par de violents affrontements entre les Romains, commandés par le Général-Empereur Julien l’Apostat (361-363) et les “Germains” (des Alamans et des Francs).2 Cette situation de conflit avec les Alamans continue avec Valentinien (364-375) et pendant toute la deuxième moitié du IVe siècle autour de la frontière du Rhin et du Danube.

Les textes, aussi bien les récits d’Hydatius pour la Gallaecia que ceux d’Ammianus Marcellinus pour les Agri Decumates, mais aussi la Notitia Dignitatum4, ne sont pas très explicites en ce qui concerne le peuplement et les formes d’habitat à la fin du IVe et au début du Ve siècle. Dans le cas d’Hydatius, nous avons trois références à la présence des castella5 et des castra6 dans lesquelles, selon l’auteur galicien, était “concentrée” la population locale

Dans la Gallaecia (au Nord-ouest ibérique), la première moitié du Ve siècle est truffée d’épisodes de conflits: entre 419 et 429 avec les Vandales Hasdings (avant leur départ en Afrique) et entre 429 et 455 avec les Galaico-Romains.3 Les conséquences de ces faits ont donné lieu à des changements et à des transformations dans les structures

4

Texte qui nous informe de l’organisation militaire de l’Empire au début du Ve siècle. Vers 411: Hispani per civitates et castella residui a plagis barbarorum per prouincias dominantium se subiciunt seruituti (Hydace Chronique). Vers 430: Suebi, sub Hermerico rege, medias partes Gallaeciae depraedantes, per plebem, quae castella tutiora retinebat suorum partim caede partim captivitate, pacem quam ruperant familiarum qua tenebatur redhibitione restaurat (Hydace Chronique p. 130). 6 Vers 457: Unum Couiacense castrum tricesimo de Asturica miliario (Hydace Chronique).

1

5

Hydace Chronique: Alani et Vandali et Sueui Hispanias ingressi (A. Tranoy (ed.), 1974: 116). 2 Selon le récit d’Ammianus Marcellinus, officiel de l’armée romaine qui raconte la bataille d’Argentorate (Straßbourg) et la sévère déroute des Alamans en 357. 3 Hydace Chronique. Pour cette période et concernant les rapports entre les “Germains” et les Romains: Garcia Moreno 1976; López-Quiroga & Rodríguez Lovelle 1995-96; 1997.

801

Limes XVIII celle de Hessen15, 4 pour celle de Thüringen16 et 7 dans d’autre régions17) et 31 castros (Fig. 1) dans le Nord-ouest ibérique (16 pour la région entre la mer Cantabrique et le fleuve Miño18 et 18 pour la zone située entre les fleuves Miño et Douro19). Dans ces deux régions, l’occupation du début du Ve siècle est complètement différente de celle que l’on constate sur les mêmes lieux ou sur d’autres lieux de hauteur durant le VIe et le VIIe siècles.20 Voyons donc, quels sont les traits caractéristiques de ce peuplement en hauteur, et essayons de déterminer les raisons de ce phénomène au-delà du cadre régional. Puisque, n’oublions pas, des Höhenstationen (des stations de hauteur) sont trouvées également en Suisse, dans le Nord de l’Italie, ou encore dans les Balkans.21

galaico-romaine aux premiers moments de l’installation des “Suèves” dans la Gallaecia, c’est-à-dire, durant la première moitié du Ve siècle. D’autre part, le Ier Concile de Tolède (397-400) montre des castella comme l’une des unités du peuplement dans laquelle il pouvait y avoir une église en dehors des civitates.7 En ce qui concerne le HautRhin, Ammianus Marcellinus nous parle des colles8 et de montem9, comme des lieux où étaient installés les “Germains” (les Alamans, dans ce cas). Pour les deux régions, la Notitia Dignitatum nous donne aussi la liste des castella et des troupes qui étaient au Nord de la Péninsule Ibérique10 entre la fin du IVe et le début du Ve siècle. Donc, les textes montrent la présence d’une série de lieux apparemment fortifiés et en hauteur, qui sont nommés comme castra et castella pour le Nord-ouest ibérique et colles et montem pour le Haut-Rhin. Il n’est pas très difficile, en ce qui concerne la Gallaecia, de faire le rapport entre ces lieux en hauteur et les nombreux castros caractéristiques du paysage de cette région.11 Non plus, en ce qui concerne le Haut-Rhin, puisque les caractéristiques géographiques de cette région sont très favorables au développement d’une occupation des sites en hauteur, même s’il ne s’agit pas ici des castros mais des stations en hauteur, dans lesquelles (à l’égal du Nord-ouest ibérique, d’ailleurs) l’occupation que l’on observe au début du Ve siècle est fréquemment précédée d’une autre occupation antérieure qui remonte souvent à l’époque pré-romaine.

Breisgau-Hochschwarzwald), Ailenberg bei Rüdern (Esslingen), Dreifaltigkeitsberg bei Spaichingen (Tuttlingen), Kügeleskopf bei Ortenberg (Ortenaukreis), Heiligenberg bei Heidelberg, Calverbühl bei Dettingen (Reutlingen), Lehenbühl bei Fridingen (Tuttlingen), Schalksburg bei Laufen (Zollernalbkreis), Geißkopf bei Berghaupten (Ortenaukreis), Hals bei Bodman, Ludwigshafen (Konstanz). (Fig. 2). 14 Mainschleife bei Urphar (Kreuzwertheim, Main-Spessart), Gelbe Bürg bei Dittenheim (Gunzenhausen), Houbirg bei Happurg (Nürnberg Land), Gangolfsberg bei Oberelsbach (Rhön), Staffelberg bei Staffelstein (Lichtenfeld), Turmberg bei Kasendorf (Kulmbach), Reißberg bei Scheßlitz (Bamberg), Ehrenbürg bei Kirchehrenbach (Forchheim), Michelsberg bei Kipfenberg (Eichstätt), Schwedenschanze bei Michelbach, Hammelberg bei Hammelburg (Kissingen), Saupürzelberg bei Karlstadt, Marienberg bei Würzburg, Schwanberg bei Iphofen (Kitzingen), Bullenheimer Berg (Kitzingen), Hesselberg bei Gerolfingen (Dinkelsbühl), Wiesentfeld (Hollfeld, Bayreuth), Eggolsheim-Drügendorf (Forchheim), Giechburg bei Scheßlitz (Bamberg), Hoher Knock bei Hollfeld-Loch, Heidelberg bei Egloffstein-Schweinthal, Schloßberg bei Geisfeld (Bamberg), Wartberg bei Friesen (Bamberg). (Fig. 2). 15 Dünsberg bei Biebertal-Fellinghausen (Gießen), Glauberg (Wetteraukreis), Burgberg bei Königstein (Hochtaunuskreis), Alte Burg auf dem Hoppesberg (Gelnhausen), Heuneburg auf der Altscheuer bei Lichtenberg (Darmstadt-Dieburg), Dornburg bei Frickhofen (LimbachWeilburg). (Fig. 2). 16 Kleiner Gleichberg bei Römhild, Großer Gleichberg bei Römhild, Alte Burg bei Arnstadt, Hassenburg bei Haynrode (Worbis). (Fig. 2). 17 Greinberg bei Miltenberg (Unterfranken), Burgstall bei Oberhöchstädt (Neustadt, Aisch), Sulzbürg (Neumarkt, Oberpfalz), Gaulskopf bei Ossendorf (Warburg, Höxter), Kirchberg bei Bornstedt (Eisleben), Schloßberg bei Seeburg (Eisleben), Kirchberg bei Bösenburg (Eisleben), ces trois dernièrs dans la région de Sachsen-Anhalt. (Fig. 2). 18 Ces castros sont: celui de Santa Tecla (A Guardia), de Tui, de A Lanzada (Noalla, Sanxenxo), de Taboaexa (As Neves), de “O Calvario” (Santa Marina do Rosal), de San Cristóbal (Arealonga, Villagarcía de Arousa), tous dans la province de Pontevedra; celui de Santa María de Britonia (A Pastoriza), de Viladonga (Castro de Rei), de Fazouro (Foz), de San Vicente de Castillóns (Pantón), de Penadominga (San Marcos de Paradaseca, Quiroga), de Piñeira (Ribadeo) , de Calvor (Sarria) , de Proendos (Sober) , dans la province de Lugo; et celui de San Cibrán das Lás (San Amaro, Orense). Une description et une bibliographie complète de ces lieux dans López Quiroga & Rodríguez Lovelle 1999 (Fig. 1). 19 Ces castros sont: celui de Saceda (Lucenza, Cualedro), de Novás (Xinzo de Limia), de Sacoias (Baçal, Bragança), de Guifoões (San Martín de Guifões, Matosinhos), de Santomé (Orense), de São Brás (Frazão, Paços de Ferreira), de Monte Mozinho (Penafiel), de Fontes (Santa Marta de Penaguião, Vila Real), de Castelo (Villamartín de Valdeorras), de Curralha (Chaves), de Falperra (Braga), d’Eiros (Aboim das Chozas, Arcos de Valdevez), de “Monte do Castro” à Gomdomar (Porto), de “Monte de Sto.Ovidio” (Santa Marina de Arcozelo, Ponte de Lima), de São Fins de Ferreira (Paços de Ferreira), de “Alto da Fonte do Milho” à Canelas de Douro (Pesso da Régua, Vila Real), de Lanhoso (Póvoa de Lanhoso, Braga). Une description et une bibliographie dans López Quiroga & Rodríguez Lovelle 1999: 357-360. (Fig. 1). 20 Steuer 1994b; López Quiroga sous-presse. 21 Steuer 1990: 141-143: notes 11 à 16, avec toute la bibliographie.

L’archéologie permet aujourd’hui de comptabiliser un total de 62 Höhensiedlunge (des peuplements en hauteur) (Fig. 2) possédant des traces d’occupation ou de réoccupation entre la fin du IVe et le début du Ve siècle dans le Sudouest de l’Allemagne12 (19 pour le Land de BadenWürttemberg13, 23 pour la région de la Bavière14, 6 pour 7 uel in loco in quo est ecclesia aut castelli aut uicus aut uillae (J. Vives (ed.) Concilios visigóticos e hispano-romanos. (Madrid-Barcelone 1967)). 8 Lentienses...impetu celeri obsessos pertiuerunt inuiis cautibus colles, abruptisque per ambitum rupibus insistentes, rebus caritatibusque suis, quas secum conduxerant, omni uirium robore propugnabant (Ammianus Marcellinus XXXI.10.12). 9 Qui nullam ad tuendam salutem uiam superesse cernentes, ni se celeri defendissent occursu, locorum gnaritate confisi, unum spirantibus animis, montem occupare praecelsum, per confragosos colles undique praeruptum et inuium, absque septentrionali latere (Amminaus Marcellinus XXVII.10.9). 10 In prouincia (Hispania) Gallaecia: Praefectus legionis septimae geminae, Legione. Tribunus cohortis secundae Flauiae Pacatianae, Paetonio. Tribunus cohortis secundae Gallicae, ad cohortem Gallicam. Tribunus cohortis Lucensis, Luco. Tribunus cohortis Celtiberiae, Brigantiae, nunc Iuliobriga... (O.Seek (ed.) Notitia Dignitatum, Frankfurt-am-Main, p.25ff.). Sur la Notitia et l’armée romaine en Hispania: Arce 1980; Le Roux 1982: 385ff. 11 López Quiroga & Rodríguez Lovelle 1999; Díaz Martínez 1987. 12 Une description et une bibliographie sur ces sites dans: Steuer 1990: 173-196; Haberstroh 1996; Brachmann 1993. 13 Runder Berg bei Urach (Reutlingen), Zähringer Burgberg bei Gundelfingen (Breisgau-Hochschwarzwald), Rosenstein bei Heubach (Scwäb.Gmünd), Achalm bei Reutlingen (Reutlingen), Lochenstein bei Hausen am Tann (Zollernalbkreis), Hohenkrähen bei Deuchtlingen (Konstanz), Goldberg bei Goldburghausen (Aalen), Waldenbühl bei Donzdorf (Göppingen), Heuneburg bei Hundersingen (Sigmaringen), Lenensburg bei Betznau (Kreßbronn (Bodenseekreis), Kegelriß bei Ehrenstetten (Breisgau-Hochschwarzwald), Feimlisburg (Kirchhofen,

802

Jorge López Quiroga & F.Germán Rodríguez Martín: Les castra et les castella aux extrémités del l’Empire romaines (Figs. 8 & 9)26 et des pièces en bronze (des fibules, des boucles de ceinture, des harnois de cheval) typiquement tardo-romaines, les matériaux qui permettent d’attester une occupation de certains castros entre la fin du IVe et le début du Ve siècle. Les importations de céramiques en provenance du Nord de l’Afrique et du Moyen Orient montrent la vitalité et l’importance du commerce à ce moment-là dans la région côtière au Sud du fleuve Câvado (Figs. 8 & 9) (autour du Port et du castro de Oporto, sur la rive droite du Douro) et de l’espace côtier au Nord du Miño (autour de la via per loca maritima), qui ne s’interrompt pas avec l’arrivée et l’installation des “Suèves” au Nord-ouest ibérique, et qui va même continuer durant le VIe siècle.27 La présence des pièces en bronze est aussi caractéristique dans ces castros occupés ou réoccupés entre la fin du IVe et le début du Ve siècle. Dans ce sens, le castro de Viladonga, placé à 23kms au NE de Lucus Augusti-Lugo, la capitale politique et administrative du conventus lucensis, constitue l’un des exemples paradigmatiques de ce type d’occupation (Fig. 3). La phase d’occupation principale de ce castro (à 553m d’altitude) est située entre le IIIe et le VIe siècle.28 Possédant une superficie de 40,000m carrés, Viladonga montre un complexe système défensif avec plusieurs remparts (jusqu’à 4 dans sa partie Est) et une monumentalité considérable. On trouve ici des éléments typiquement romains: de la terra sigillata hispanique et de la sigillée hispanique tardive29, de la céramique estampillée tardive, des verres, des parois fines30, des fibules en omèga, de boucles de ceinture, et des monnaies du IVe-Ve s. Mais, ce qui nous intéresse de souligner ici, c’est la présence de matériaux en provenance de la Méditerranée orientale: des bronzes (deux harnois de cheval datés entre le IVe et le VIe siècles31 et un exagium, Fig. 4), de la céramique estampillée tardive, des morceaux d'amphores de type oriental datées du Ve-VIe s. (Fig. 5), des graffiti sur des fragments de céramique ou des tuiles avec une filiation orientale (Fig. 5), des anneaux, une ligula en argent avec des marques en caractères grecs ou encore des comptes de collier (Fig. 4). Tous ces matériaux nous indiquent l’existence de rapports commerciaux entre le Nord-ouest ibérique et la Méditerranée orientale. Soit par l’intérieur à travers la Meseta soit surtout à partir des nombreux ports de la côte atlantique. Ce commerce n’est pas sans rapport avec la présence des graeci dans la Gallaecia, comme les textes de l’époque l’attestent32,

Si l’on observe les zones où l’on trouve ces castros et ces Höhensiedlungen (des peuplements de hauteur), nous voyons qu’il y a toujours un rapport avec le réseau routier romain (s’agisse-t-il du réseau routier principal ou des voies secondaires), le réseau hydrographique et avec le peuplement tardo-romain (Figs. 1 & 2). Plus précisément, dans le Nord-ouest ibérique la plupart de ces castros se situe entre les fleuves Ulla et Duero, autour de la voie romaine qu’on appelle via per loca maritima (puisqu’elle parcourt toute la côte depuis Porto jusqu’à La Corogne). Dans le Haut-Rhin, on les trouve autour du Rhin, du Danube, du Neckar et du Main; c’est-à-dire, des fleuves sur lesquels se situe la frontière du limes et d’autres qui sont des affluents du Rhin (comme le Neckar et le Main). Par conséquent, aussi bien dans une région que dans l’autre, il y a un rapport entre les castros et les Höhensiedlungen, et les voies de communication fluviales ou terrestres. Ces peuplements de hauteur se placent aussi autour de centres politiques et administratifs (Braga et Lugo), des Ports (Oporto ou Iria Flavia), des castella du limes (Castra Regina-Regensburg, Augusta VindelicorumAugsburg, Argentoratum-Straßburg) (Figs. 1 & 2). Sans oublier, en ce qui concerne le Sud-ouest de l’Allemagne, la présence d’Augusta Treverorum-Trèves, placée sur la Moselle (elle aussi un affluent du Rhin), qui est un centre politique et administratif de premier rang pour l’ensemble de l’Empire Romain. Ces peuplements de hauteur sont toujours autour de noyaux possédant une grande importance géo-politique, parce qu’ils sont des lieux où l’on réside et à partir d’où l’on exerce le pouvoir. Il est fort probable que ces castros et ces Höhensiedlungen aient pu jouer aussi un rôle en tant que lieux centraux pour les zones où ils se situent. De plus, on peut supposer qu’ils occupent une position hiérarchique par rapport au réseau du peuplement tardo-romain dans lequel ils vont s’intégrer. Il pourrait s’agir, dans le cas de la Gallaecia, d’unités intermédiaires entre la ciuitas, les uici et les uillae, qui font partie du hinterland de celle-là.22 Mais, tandis que dans le Haut-Rhin ces Höhensiedlungen seraient au sommet de la hiérarchie (au-dessus des uillae23 et des uici)24, dans le Nord-ouest ibérique les castros seront par contre au début du Ve siècle toujours sous l’autorité de la ciuitas. En ce qui concerne les matériaux archéologiques que l’on trouve et qui permettent de dater l’occupation de ces sites de hauteur entre la fin du IVe et le début du Ve siècle, il y a aussi des ressemblances et des différences, comme l’on a signalé, en fonction du contexte socio-politique de chaque région.

Arcocelo), “Monte do Castelo” (Santiago de Carreiras), Viladonga (Castro do Rei) et Calvario (Santa Marina do Rosal); cf. Bouza Brey 1953; Fariña Busto 1974. (Fig. 7). 26 Naveiro López 1991. 27 Naveiro López 1991; comme le montrent aussi les trouvailles de plus en plus fréquentes des importations des productions céramiques tardives dans tout le Nord-ouest ibérique (à La Corogne, à Vigo, à Tui, à Braga, etc.) (Figs. 7 & 9): López Quiroga & Rodríguez Lovelle 1998a; 1998b; 1998c. 28 Avec peut-être une occupation précédente entre le IIe et le Ie siècle avant J.C. 29 Caamaño Gesto & López Rodríguez 1984. 30 Vila Martinez 1994. 31 Palol 1952: 319; Fuentes Domínguez 1990: 120. 32 Vallejo Girvís 1993.

Pour le Nord-ouest ibérique, ce sont les trésors monétaires avec des pièces d’Arcadio et d’Honorio (395-408) (Figs. 6 & 7)25, les trouvailles de céramiques importées tardo22

Revuelta Carbajo 1997. Plus de 1000 uillae rusticae pour le Sud-ouest de l’Allemagne: Nuber 1990: 277-286; aussi: Asskamp 1989. 24 Steuer 1994b: 90. 25 Dans les castros de: Eiros (Aboim das Chozas), “Outeiro de San Sebastião” (Palmeira), “Monte do Castro” (Gondomar), “Monte Mozinho” (Penafiel, “Monte de Santo Ovidio” (Santa Marina de 23

803

Limes XVIII

possédant un but commercial. D’autre part, le castro de Viladonga occupe une position intermédiaire (Fig. 1) entre la civitas de Lucus Augusti (qui joue un rôle de redistribution vers l’intérieur des produits importés qui arrivent par voie fluviale ou terrestre à partir de la côte atlantique) et le réseau de villae ou de vici, dans ce cas précis avec la villa de Doncide, placée à 1.5kms du castro de Viladonga.33

Dans le sud-ouest de l’Allemagne (Fig. 2), les matériaux que l’on trouve sur les Höhensiedlungen, nous montrent une population mixte composée d’hommes et de femmes, mais pas exclusivement associée à des éléments militaires.40 Il s’agit, bien sûr, de populations “germaniques” (des Alamans), mais avec un style de vie tout à fait romain ou d’influence romaine, comme le prouve la présence majoritaire de la céramique41 tardoromaine dans ces Höhensiedlungen. Par exemple, dans le Zähringer Burgberg (Freiburg in Breisgau), où l’on trouve de la céramique sigillée tardive et de la terra nigra à coté de la céramique et d’autres éléments typiquement “germaniques” (des anneaux, des colliers, etc.).42 Nous sommes donc en train de parler, de peuplements de hauteur qui ont surtout une fonction d’habitats. Néanmoins, certains ont aussi un évident caractère militaire, en tant que lieux de résidence des garnisons des foederati (des Alamans dans le Haut-Rhin ou des Burgunds autour du cours moyen du Main) autour de l’ancien limes. C’est le cas du Runder Berg (Urach, Reutlingen, BadenWürttenberg), ou peut-être, de celui de Geißkopf bei Berghaupten (Ortenaukreis, Baden Württenberg), où l’on me trouve pas de traces d’habitat mais des armes et des éléments de ceinture militaire tardo-romaine.43 Une autre des caractéristiques spécifiques à ces Höhensiedlungen (et qu’on ne trouve pas dans les castros ibériques), c’est qu’ils sont aussi des lieux où l’on fabrique des armes et où l’on produit des pièces métalliques (surtout en bronze).44 Ce sont donc parfois des habitats, parfois des camps militaires (comme les anciens castella du limes), mais dans la plupart des cas, il s’agit des lieux où l’activité économique (d’échange et de production) est aussi fondamentale.45 C’est-à-dire, des lieux multifonctionnels qui se situent au sommet de la hiérarchie du peuplement tardo-romain, mais sans rien y changer: le peuplement et l’occupation du sol typiquement tardo-romaines paraît continuer (des uillae rusticae et des uici) et se maintenir jusqu’au VIe siècle.46

Donc, d’une part, on constate que l’occupation de ces castros est antérieure à l’arrivée des “Germains” en 409. D’autre part, il est possible qu’il y ait eu aussi des raisons de type économique (à part des éléments de stratégie géopolitique) qui expliquent l’occupation ou la réoccupation de ces sites de hauteur. C’est-à-dire, dans le but d’assurer le maintien et la protection des circuits commerciaux à partir d’une série de castra et de castella qui permettent l’arrivée des produits de la côte vers l’intérieur à travers les fleuves et le réseau routier principal et secondaire (Fig. 1). Cela nous montre aussi (comme Hydace le souligne, d’ailleurs) que la population qui habite ces castros pendant la première moitié du Ve siècle est composée de Galaicoromains. Cela ne veut pas dire, qu’ il n’ y ait pas eu des castros occupés par des populations d’origine “germanique”, dans ce cas, des “Suèves”. Peut-être, est-ce le cas du castro “da Curralha” (à coté de la ville d’Aquae Flaviae-Chaves), qui semblerait montrer un système défensif très performant vraisemblablement dû à l’activité militaire des Suèves dans les régions intérieures de la Gallaecia34, du Portucalem castrum novum (le site d’Oporto sur la rive droite du Douro) où l’on constate la présence d’une enceinte fortifiée dès le début du Ve siècle35, ou encore du castro de Falperra (à coté de Bracara Augusta-Braga), où l’on a trouvé la seule pièce en bronze attribuable aux “Suèves” dans tout l’ensemble territorial du Nord-ouest ibérique.36 De plus, nous sommes en présence de castros qui se placent à coté des centres politiques, administratifs et économiques d’époque romaine (Braga, Chaves et Oporto) (Fig. 1). Également, Braga a été choisie comme la première capitale politique des “Suèves” jusqu’en 455; puis c’est le tour d’Oporto jusqu’en 555.37 En ce qui concerne la ville de Chaves, elle a été le noyau de la résistance à l’expansion politique et territoriale suève, à la tête de laquelle se trouvait l’évêque galicien Hydace.38 Cela ne nous empêche pas d’insister (comme nous l’avons déjà souligné) sur le fait évident que la plupart des habitants de ces castros est formée par une majorité de population galaico-romaine, ce que confirment aussi bien les textes que l’archéologie.39

Ces Höhensiedlungen, si différents soient-ils des castros du Nord-ouest ibérique par beaucoup d’aspects, ont aussi un rôle important dans la hiérarchie du peuplement tardoromaine, et ils possèdent une signification géo-politique et économique qui ne s’expliquerait pas exclusivement dans le cadre local ni régional, mais qui serait en rapport avec l’évolution des structures du peuplement et les structures sociales pendant l’Antiquité tardive en Occident. Dans ces deux régions aux extrémités et dans la périphérie de l’Empire, l’effondrement de la structure politiqueadministrative et militaire romaine au début du Ve siècle, donnerait donc lieu à un développement d’un peuplement de hauteur: - dans le Haut-Rhin, les Höhensiedlungen (qui sont au sommet de la hiérarchie du peuplement tardo-

33

Arias Vilas 1993: 208. Ferreira De Almeida 1976; 1978; 1992: 377; Barroca sous-presse; Medeiros Freitas & Dos Santos Junior 1981. 35 Pinto Osorio 1993:15-24. 36 Rigaud de Sousa 1968-1970: 57-64; Schlunk & Hauschild 1978. 37 López Quiroga & Rodríguez Lovelle 1995-96; 1996-97; 1997. 38 López Quiroga & Rodríquez Lovelle 1997; López Quiroga sous-presse 39 Les fouilles des castros de Viladonga (Castro do Rei), de Castillóns (Pantón) et de Penadominga (Quiroga), tous dans la province de Lugo, montrent sans aucun doute le contexte tardo-romain de ces lieux; cf. Chamoso Lamas 1977; 1958-59; Arias Vilas 1983. 34

40

Steuer 1990: 169; 1994b: 88. Sur la céramique dans la région de Breisgau: Bucker 1999. Steuer 1989. 43 Koch 1982; 1984; Hoeper & Steuer 1999. 44 Steuer 1994a. 45 Steuer 1990; 1994a. Aussi Brachmann 1993. 46 Steuer 1994b: 92. 41 42

804

Jorge López Quiroga & F.Germán Rodríguez Martín: Les castra et les castella aux extrémités del l’Empire

romaine), paraissent devenir le lieu de résidence des puissants locaux (des Alamans).47 Ceux-ci se situent également au sommet de la hiérarchie sociale dans cette région, tandis que les Romains (qui habitaient auprès d’eux et même avec eux) ser ont en bas de la Pyramide. Il s’agit, du point de vue de l’évolution des structures sociales dans l’Antiquité tardive, d’un changement majeur. - dans le Nord-ouest ibérique, les castra et les castella (les anciens castros), qui se placent au milieu de la hiérarchie du peuplement (entre les civitates, les uici et les uillae)48, sont probablement aussi devenu des lieux de résidence des puissants locaux (des Galaico-romains49) qui étaient déjà au sommet de la hiérarchie sociale au début du IVe siècle, tandis que les Germains (les Suèves), qui détiennent le pouvoir, ne peuvent pas empêcher (vu leur faible nombre et ses perpetuelles divisions internes) la fragmentation politique et territoriale de la Gallaecia et la formation des ces cellules de pouvoir autonomes. Le “retour aux sommets” entre la fin du IVe et le début du Ve siècle dans le Nord-ouest ibérique50, si l’on peut l’appeller ainsi, d’une certaine façon peut-être aussi vu comme un signe de changement dans les structures sociales de la Gallaecia de l’Antiquité tardive.

extrémités de l’Empire, l’éclatement du pouvoir et le morcellement politique et territorial deviendrait dès le Ve siècle l’une des caractéristiques les plus importantes, du point de vue socio-politique, après la fin de la domination romaine. Bibliographie Arce J. 1980 La ‘Notitia Dignitatum’ et l’armée romaine dans le diocesis Hispaniarum. Chiron 10: 593-608. Arias Vilas F. 1983 Algunhas cuestións sobre os asentamentos na Galicia baixo-romana. In II Seminario de Arqueología del Noroeste. (Santiago de Compostela): 263-271. Arias Vilas F. 1993 Apuntes sobre a ocupación do territorio na Galicia baixo-romana. In Galicia: da Romanidade á Xermanización (Homenaxe a Fermín Bouza-Brey). (Santiago de Compostela): 201-208. Asskamp R. 1989 Das Südliche Oberrheingebiet in Frührömischer Zeit. (Stuttgart). Barroca M. sous-presse Fortificações e Povoamento no Norte de Portugal (séc.IX-XI). In A. Fuentes Domínguez & J. López Quiroga (edd.) Poblamiento rural en el Norte de la Península Ibérica (ss.V-X): continuidades, rupturas, transformaciones. (Madrid). Bouza Brey F. 1953 Los tesorillos de monedas de Tremoedo y de Sarandón y su significado histórico en Galicia. In III Congreso Nacional de Arqueología. (Saragosse): 375391. Brachmann H. 1993 Der frühmittelalterliche Befestigunsbau in Mitteleuropa. (Berlin). Bucker C. 1999 Frühe Alamanen im Breisgau. Untersuchungen zu den Anfängen der Germanischen Besiedlung im Breisgau während des 4. und 5. Jahrhunderts n.Chr. (Sigmaringen). Caamaño Gesto J.M. & López Rodríguez J. 1984 Sigillatas del Castro de Viladonga. Gallaecia 7-8: 158-178. Chamoso Lamas M. 1958-59 Sobre hallazgos arqueológicos en Castillones. Boletín de la Comisión Provincial de Monumentos de Lugo IV: 213-215. Chamoso Lamas M. 1977 Las excavaciones del castro de Viladonga y la problemática que plantean sus resultados. In Actas del coloquio sobre el Bimilenario de Lugo. (Lugo): 41-46. Diaz Martinez P.C. 1987 Estructuras de gobierno local en la Antigüedad Tardía y en la Alta Edad Media. Un estudio regional: el NO de la Península Ibérica en el s.V. Studia Zamorensia VIII: 234-245. Fariña Busto F. 1974 Algunos aspectos de la circulación monetaria en Galicia durante el s.IV d.C. In Congreso Nacional de Arqueología. (Saragosse): 105-128. Fernández Mier M. 1999 Arqueología del paisaje y evolución histótica en la montaña asturiana: el valle del río Piguela. (Oviedo). Ferreira C.A. 1976 A propósito de ‘Galicia Sueva’ de Casimiro Torres. Gallaecia 3-4: 115-123. Ferreira C.A. 1978 Castellologia Medieval de Entre-Douro-eMinho. (Oporto). Ferreira C.A. 1992 Castelos medievais do noroeste de Portugal. In Finis Terrae. Estudios en lembranza do Prof.Dr. Alberto Balil. (Saint-Jacques de Compostelle). Fuentes Domínguez A. 1990 Los bronces imperiales en Hispania. In Los bronces romanos en Hispania. (Madrid): 117-135.

En définitive, même si le contexte local et la spécificité de chaque site ne permet pas de généraliser un modèle explicatif uniforme, il faut néanmoins souligner que ces occupations et réoccupations des hauteurs dans la périphérie de l’Empire, ont eu un rapport avec l’effondrement de la superstructure politique, administrative et militaire tardo-romaine: - d’une part, ce fait aurait donné lieu à des changements et à des transformations dans les structures du peuplement sans qu’il y ait eu une rupture mais une transformation interne aussi bien dans le Haut-Rhin qu’au Nord-ouest ibérique, dans un contexte évolutif qui frapperait l’ensemble des provinces de l’Empire Romain d’Occident. - d’autre part, du point de vue de l’évolution des structures sociales et politiques, cet événement majeur au niveau de l’Empire, serait à l’origine d’un éclatement du pouvoir qui a pour conséquence le “réveil” des élites locales et la formation des petits centres de pouvoir autonomes au niveau local ou supra-local. Dans le Haute Rhin, ce sont les Höhensiedlungen où l’on trouve les centres de pouvoir des petits roitelets Alamans au IVe-Ve siècles.51 Au Nord-ouest ibérique, on trouve sur les anciens castros une partie considérable de la population et des élites locales qui essaient de maintenir le statu quo existant avant l’arrivée des “Germains”. C’est ainsi qu’aux 47

Steuer 1994a Revuelta Carbajo 1997. 49 Et peut-être (surtout au VIe et VIIe siècles) des élites indigènes qui ont profité de la faiblesse numérique et de l’incapacité politique des Suèves pour assurer le control de toute l’ensemble territorial de la Gallaecia: López Quiroga sous-presse. 50 Mais aussi dans d’autres régions du Nord de la Péninsule Ibérique: Fernández Mier 1999; Orejas Saco del Valle 1996; Fuentes Domínguez & López Quiroga sous-presse; Díaz Martínez 1987: 234-245; Martín Viso 1995. 51 Comme le récit d’Ammianus Marcellinus le montre bien. 48

805

Limes XVIII

Fuentes Domínguez A. & López Quiroga J. (édd.) sous-presse El poblamiento rural en el Norte de la Península Ibérica(ss.V-X): continuidades, rupturas, transformaciones. (Collection Séminaires et Rencontres de la Casa de Velázquez 1). (Madrid). García Moreno L.A. 1976 Hidacio y el ocaso del poder impérial en la península ibérica. Revista de Archivos, Bibliotecas y Museos 79: 27-42. Haberstroh J. 1996 Germanische Funde der römischen Kaiserzeit und Völkerwanderungszeit aus Oberfranken. Arch. Nachrbl. 1: 338-345. Hoeper M. & Steuer H. 1999 Eine völkerwanderungszeitliche Höhenstation am Oberrhein -der Geißkopf bei Berghaupten, Ortenaukreis. Höhensiedlungen, Kultplatz oder Militärlager? Germania 77: 185-246. Koch U. 1982 Die frühmittelalterlichen Funde vom Runden Berg bei Urach Kr.Reutlingen. Archäologisches Korrenpondenzblatt 12: 81-92. Koch U. 1984 Handwerker in der alamanischen Höhensiedlungen auf dem Runden Berg bei Urach. Archäologisches Korrenpondenzblatt 14: 99-109. Le Roux P. 1982 L’armée romaine et l’organisation des provinces ibériques d’Auguste à l’invasion de 409. (Paris). López Quiroga J. sous-presse El final de la Antigüedad en la Gallaecia meridional (ss.V-X): la transformación de las estructuras de poblamiento entre el Miño y el Duero. (a Coruña). López-Quiroga J. & Rodríguez Lovelle M. 1995-96 De los Vándalos a los Suevos en Galicia: una visión crítica sobre su instalación y organización territorial en el noroeste de la Península Ibérica en el s.V. Studia Historica. Historia Antigua 13-14: 421-436. López Quiroga J. & Rodríguez M. 1996-97 De los Suevos a los Visigodos en Galicia (537-711): Nuevas hipótesis sobre el proceso de integración del noroeste de la Península Ibérica en el reino visigodo de Toledo. Romano-Barbarica 14: 259-298. López-Quiroga J. & Rodríguez Lovelle M. 1997 De los Romanos a los Bárbaros: la instalación de los Suevos y sus consecuencias sobre la organización territorial en el Norte de Portugal (411-469). Studi Medievali (3a Serie-Anno XXXVIII, Fasc.IIDicembre): 529-560. López Quiroga J. & Rodríguez Lovelle M. 1998a Consideraciones en torno al modelo de ciudad entre la Antigüedad tardía y la Alta Edad Media en el noroeste de la Península Ibérica (ss.V-XI). In A. Rodríguez Colmenero (ed.) Los orígenes de la ciudad en el noroeste peninsular. (Lugo): 1319-1347. López Quiroga J. & Rodríguez Lovelle M. 1998b De la cité antique aux évêchés du haut moyen âge en Galice et Nord du Portugal (IVe-Xe s.): continuités, ruptures, transformations. In N. Coulet (ed.) La ville au Moyen Age. (Paris, vol.II): 15-40. López Quiroga J. & Rodríguez Lovelle M. 1998c La relation ville-campagne au Nord-Ouest de la Péninsule Ibérique: un régard depuis le Moyen Age. In P. Gros (ed.) Villes et campagnes en Occident romain. (Paris): 189-205. López Quiroga J. & Rodríguez Lovelle M. 1999 Castros y castella tutiora de época sueva en Galicia y Norte de Portugal. Ensayo de inventario y primeras propuestas interpretativas. Hispania Antiqva XXXIII: 360-364. Martín Viso I. 1995 Poblamiento y sociedad en la transición al Feudalismo en Castilla: castros y aldeas en la Lora

Burgalesa. Studia Historica.Historia Medieval 13: 345. Medeiros Freitas A. & Dos Santos Junior J.R. 1981 O castro da Curralha-6a Campanha de excavações-1980. Trabalhos de Antropologia e Etnologia XXIV: 59-100. Naveiro López J. 1991 El comercio antiguo en el N.W. peninsular. Lectura histórica del registro arqueológico. (A Coruña). Nuber H.U. 1990 Das Ende des Obergermanisch-Raetischen Limes- eine Forschungsaufgabe. Archäologie und Geschichte des ersten Jahrtausends in Südwestdeutschland (Sigmaringen): 51-68. Orejas Saco del Valle A. 1996 Estructura social y territorio: el impacto romano en la cuenca noroccidental del Duero. (Madrid). Palol P. 1952 Algunas piezas de adorno de arnés de época tardo-romana e hispanovisigoda. Archivo Español de Arqueología 25: 297-319. Pinto Osorio M.I. 1993 A cidade: Lugares primeiros, Rúa D.Hugo, n°5. In O Porto das Mil Idades. Arqueologia na Cidade. (Oporto): 15-24. Revuelta Carbajo R. 1997 La ordenación del territorio en Hispania durante la Antigüedad Tardía. (Madrid). Rigaud de Sousa J.J. 1968-1970 A estação arqueológica da Falperra. Notas para a sua história. Arquivo de Beja XXV-XXVII: 57-64. Schlunk H. & Hauschild T. 1978 Die Denkmäler der frühchristlichen und westgotischen Zeit (Hispania Antiqva). (Mainz). Steuer H. 1989 Der Zähringer Burgberg bei Freiburg im Breisgau, eine Höhensiedlungen des 4./5. Jahrhunderts. Archäologisches Korrenpondenzblatt 19: 169-184. Steuer H. 1990 Höhensiedlungen des 4. und 5. Jahrhunderts in Südwestdeutschland. Einordnung des Zähringer Burgberges, Gemeinde Gundelfingen, Kreis BreisgauHochschwarzwald. Archäologie und Geschichte des ersten Jahrtausends in Südwestdeutschland. (Sigmaringen): 173-196. Steuer H. 1994a Handwerk auf Spätantiken Höhensiedlungen des 4/5 Jahrhunderts in Südwestdeutschland. The Archaeology of Gudme and Lundeborg. (Arkaeologiske Studier X), (Copenhague): 128-144. Steuer H. 1994b The hierarchy of Alamannic settlements in the former limes region of south-western Germany to AD 500. Journal of European Archaeology 2: 82-96. Vallejo Girvés M. 1993 Bizancio y la España tardoantigua (ss.V-VII): Un capítulo de historia mediterránea . (Alcalá de Henares): 9-15. Vila Martínez M. 1994 Cerámica de paredes finas no Castro de Viladonga. Croa 4: 12-13.

806

Jorge López Quiroga & F.Germán Rodríguez Martín: Les castra et les castella aux extrémités del l’Empire

Fig.1. Répartition de castros avec des traces d’occupation entre la fin du IVe et le début du Ve siècles et son rapport avec les agglomérations principales et secondaires d’époque romaine.

807

Limes XVIII

Fig. 2. Höhensiedlungen des IVe – Ve siècles dans le Sud-ouest de l’Allemagne. 1- Höhensiedlungen sur lesquels ont été éffectuées des fouilles archéologiques: 2 – Höhensiedlungen avec des trouvailles isolées: 3 Castella du limes tardo-romain (extrait de M. Hoeper & H. Steuer 1999: 187, Abb.1).

Fig. 3. Castro de Viladonga (Lugo, Espagne), IIIe – VIe siècles. Plan general du castro avec les structures d’habitation découvertes à partir des fouilles archéologiques (extrait de F. Arias Vilas Materiales del Meiterráneo oriental en el Castro de Viladonga : 341).

808

Jorge López Quiroga & F.Germán Rodríguez Martín: Les castra et les castella aux extrémités del l’Empire

Fig. 4. Divers matériaux provenants de la Méditerranée orientale trouvés dans le castro tardo-romain de Viladonga (Lugo); deux arnois de cheval en bronze (No. inventaire A 70-132 et 83-9598BIS), un exagium aussi en bronze (No. inventaire A 70-334), trois contes de collier (No. inventaire A 70-349, A 70-347 et A 70-348) et une ligula en argent avec des caractères grecs (No. inventaire A 70-602) (extrait de F. Arais Vilas Materiales del Mediterráneo oriental en el Castro de Viladonga: 343.

809

Limes XVIII

Fig. 5. Divers matériaux provenants de la Méditerranée orientale trouvés dans le castro tardo-romain de Viladonga (Lugo): deux morceaux de la céramique commune romaine (No. inventaire A 70-596 et A 70-597), grafitti d’un fond de terra sigillata hispanique tardive (No. inventaire A 70-517), des grafitti en caractères grecs sur deux morceaux de tegulae (No. inventaire A 70-1 et A 70-2) (extrait de F. Arias Vilas Materiales del Mediterráneo orientsl en le Castro de Viladonga).

810

Jorge López Quiroga & F.Germán Rodríguez Martín: Les castra et les castella aux extrémités del l’Empire

Fig. 6. Répartition des trésors monétaires dans le Nord de la Gallaecia (entre la mer Cantabrique et le fleuve Miño).

Fig. 7. Répartition des trésors monétaires dans le Sud de la Gallaecia (entre les fleuves Miño et Douro).

811

Limes XVIII

Fig. 8. Répartition des trouvailles de la céramique sigillée claire D dans le Nord de la Gallaecia (entre la mer Cantabrique et le fleuve Miño).

Fig. 9. Répartition des trouvailles de la céramique sigillée claire D dans le Sud de la Gallaecia (entre les fleuve Miño et Douro).

812

New intervention in the surroundings of the gate of the Roman wall of Gijón (10 Recoletas Street) C. Fernández Ochoa, V. García Entero, F. Gil Sendino & C. Valenciano Prieto Because of the renovation of an old house located at 10 Recoletas Street in the Cimadevilla neighborhood of Gijon (Asturias, Spain), we carried out a salvage excavation there in August 1999. In this paper we present the results. The site was in the middle of an historic district where archaeological ruins from the Roman city gate complex, and more specifically, other structures located within the city wall (eg. a guard post), were supposed to lie as demonstrated by the first systematic excavations, directed by Carmen Fernandez Ochoa, in the 1980s. The conclusions from these excavations have been published in the monograph Muralla Romana de Gijon. The area excavated lay exactly along the part of the main entrance to the Roman city where the remains of one of the two towers that flanked and protected the city were found. The results of the excavations allow us to pose some questions in order to clarify the architectural features of the Roman wall of Gijon as well as to learn about the history of this monument that covers a time-span from the late Roman period to the end of Medieval times. The foundations and base of the wall (Area B) were the only documented structures. It is the most interesting area belonging to the main gate of the city entrance of Gijon. Out of the architectural aspects related to the area east of the Roman wall gate, the last intervention has provided us with information about the length of the inner wall and adjacent zones. In particular, the quadrangular structure seems to be a small enclosure which was a guard post. The arrangement of this structure and its relationship with the East tower has parallels in the Roman camps and city walls, as documented in other Spanish Roman cities such as León, Inestrillas, Coria, etc.

exploring these sterile levels in the south-western area adjacent to the wall. They can only be fully documented by future, more comprehensive excavations.

Introduction During the months of July and August 1999, an archaeological excavation was carried out in the building at number 10 Recoletas Street in Gijón. The site is located in the city’s historic fish market, an urban area now declared Bien de Interés Cultural (an area of cultural interest) by Real Decreto 753/1987 of March 8th (Fig. 1).

The survey dimensions were determined by the site area, around 60m2, and by the necessity of including safety measures such as scaffolding the eastern wall of the adjacent building because of its precarious state and danger presented by the dwelling. Thus, a grid of 7.5m x 2.5m running north-south was established resulting in the total survey dimensions of 18.75m2 .

The area occupied by the site was likely to contain remains of the Roman city gate, more precisely, of the structures located in-wall and known as a guard post. This was revealed by the systematic excavations carried out at this location in the 1980s under the direction of one of us, and published in Muralla Romana de Gijón, a work that will be repeatedly mentioned (Fernández Ochoa 1997a). The surveyed area corresponds to ‘Area B’, where the remains of one of the towers flanking the gate were excavated (Fernández Ochoa 1997a: 58). According to the remains of foundations observed in the western part of the city wall, an interpretation of the internal face in the eastern sector was put forward (Fernández Ochoa 1997a: 240). The archaeological work performed largely fulfilled the resolution of the original proposal.

The stratigraphic sequence In the survey, a maximum depth of approximately 2m was reached and 44 stratigraphic units, structured in 7 strata, were recorded (Fig. 2). Strata dating from the modern era (Strata I to III), although they could be part of the same stratigraphic group, have been divided into three exclusive phases following the criteria of stratigraphic relation and the impossibility of determining their precise chronology. Stratum I belongs to a clearly contemporary chronology formed by the pavement sub-floor of the house being rebuilt and the modern trenches for the sewer system that cut and altered the underlying strata. In the base of this stratum were found to emerge the first courses of the internal side of the wall as well of a second wall, almost parallel to it and oriented slightly north-east - south-west (WSU = Wall Stratigraphic Unit 3) (Fig. 3). This fact reveals the level of devastation experienced by the Roman structures in this area of the wall and confirms it as the base of the dwelling under reconstruction.

Archaeological excavation Methodology The work methodology was based on the exposure of strata, both natural and human, described as stratigraphic units (SU), relying on the location of significant material elements by X, Y and Z co-ordinates using a Total Distance Station (TDS). Archaeological strata were documented in the excavated area down to the geological stratum, although space limitations created by the current sewer network impeded

Stratum II also dates to the modern era. It was located immediately beneath the house sub-floor and was made up 813

Limes XVIII

of two channels, one pavement and one structure of undefined function. The following modern stratigraphic group, Stratum III, appeared to be cut by later human activities identified in Strata I and II. It sealed Stratum IV. SU 24 was associated with Stratum III. It was located only at the north-western end of the grid, consisting of the filling of a hollow dug into SU 11 of Stratum IV and interpreted as the disturbance of that stratum during the modern era (Fig. 4).

western and northern walls of the cited structures. To excavate it, the virgin clay and stratigraphic units forming Stratum VII, dating from the early empire, were cut. The walls were inserted into this trench, leaving a narrow space 20cms wide which was filled with medium-sized and irregularly shaped stones and mortar. A small rim sherd from a red-glazed Roman common ware bowl is associated with this stratigraphic unit. This type of pottery is well documented in the levels related to the wall construction in other stretches and is typical of Gijón strata from the C4th (Fernández Ochoa 1997a: 89).

Stratum IV was located across the grid surface, although cut by a modern trench that impeded the connection between its north and southern areas. We associate a pebbled pavement with this stratum (WSU 25). This is well documented in the northern half, cut by the pit of a modern trench and by SU 24 in the area adjacent to the wall (Fig. 4). The first courses of the western wall (WSU 40) surfaces in this stratum. Its superficial alignment is partially integrated with the lime mortar and pebble pavement setting on a large preparation and levelling layer about 0.46m thick.

The stratigraphic units documented in the south-western end of the grid are also linked to Statum VI. These stratigraphic units seem to be associated with the city wall construction phase, although archaeological material that would allow more precise data is lacking. The last stratum documented in the survey is Stratum VII, made up of the stratigraphic units exclusively located in the northern area of the grid and with which fragments of Roman brick building material, an early empire module coin and a pottery sherd of uncertain date were associated. Strata VI and VII were cut in the eastern end by the trench excavated to insert the foundations of the different structures of the wall gateway that were unearthed.

Several fragments of medieval pottery, one sherd of Roman common ware, a Roman fishing or loomweight and some shapeless metal fragments were associated with Stratum IV. Despite the difficulty in establishing a precise date for the deposition of this stratum due to modern alterations and limitations of the excavated area, it can be assigned to the period when the area was frequented after the wall construction. At that time, structures WSU 40, WSU 19 and WSU 3 had been destroyed but the inner side of the wall in this area was still in use. The guard post was supported on the wall footing and stood until the siege of the city by Enrique III in 1395. After Gijón’s surrender the city burned and the wall was destroyed in order to prevent further uprisings.

The geological sub-stratum of the Cimadevilla peninsula geological sub-stratum surfaces below Stratum VII, believed to have been used during the early empire. With regard to the walls found in the excavation (Fig. 3), they defined a quadrangular in-wall structure belonging to the wall gate area, documented only in its foundations. The structure reached 5.5m long, that is 19 Roman feet (3.7m or 13 Roman feet excluding the walls). The width of the walls forming this structure varied from 93cms for the southern wall to 84cms for the northern wall. The walls were made of medium and large-sized rough-cut lime and sandstone blocks bedded with whitish-yellowish and very strong mortar of medium and large-sized caementa. The regular construction of the south and western walls contrasts with the technique used in the more irregular northern wall.

Immediately under Stratum IV and at the south-western end of the grid, that is in the area adjacent to the city wall and walls WSU 19 and WSU 3, the stratigraphic units associated with Stratum V were excavated. Stratum V was interpreted as the disturbance or restructuring of the area of the city wall gate, and the dismantling of part of the wall WSU 19. The very restricted space where the stratigraphic units associated with this stratum were documented (barely 1m2 limited by the wall itself and one of the still-in-use channels), together with the lack of interesting material associated with the stratum (except for a rim sherd of a TSHT bowl -terra sigillata Hispanica Tardía- shape 8 (Palol 10), make it impossible to offer greater accuracy.

The documented maximum height is located at the southern wall, reaching up to 2m, an area where the dismantling and reuse of some of the ashlar blocks that made up the structure has been identified. From the western wall it surves to 90cms remains, composed of large sandstone ashlar blocks and fairly regularly shaped medium and large-sized pieces of limestone.

Stratum VI is identified with the labours in constructing Gijón’s city walls attributed to the end of the C3rd or beginning of the C4th AD (Fernández Ochoa 1997a: 69ff.) by archaeological excavations carried out at other areas. The trench dug for the insertion of walls WSU 40 & WSU 19 makes up this stratum. It was a 1.10m-wide trench excavated into the geological sub-stratum, parallel to the

The section of the inner part of the wall unearthed is located in the in-wall area to the north of the Tower 9 that flanked, to the east, the main gate (Fernández Ochoa 1997a: 205). The nucleus of the excavated area acted, in part, as the foundation of both the southern wall of the (modern) dwelling being rebuilt and the house located in the adjacent site. It has been not possible to determine the 814

C. Fernández Ochoa et al: New intervention in the surroundings of the gate of the Roman wall of Gijón total width of the wall in this area. Only 1.6m of the 4.6m (16 Roman feet) that it might have reached, as indicated by the width of the wall to the western part of the gate, was seen. The length verified in the survey is 2.5m, although cleaning of the site surface towards the east and west of the grid allowed us to verify that the wall stretched along the 4.35m reached by the site at the south end. The construction nucleus is opus caementicium made of a highly consistent mortar mixed with medium and largesized limestone. The degree of devastation of this area of the wall barely allows the documentation of 2 or 3 courses (nearly 40cms). According to the reconstruction hypothesis of one of the authors, following Philon of Byzantium, the elevation must have risen 9.2m, 32 Roman feet, (Fernández Ochoa 1997a: 239ff.). This inner side is made up of small and medium rough-cut limestone blocks bedded with whitish mortar of small and very hard and consistent lime caementa. The wall foundation or footing reached a maximum height of 1.35m and projected out 20cms. It was made of small, medium and large limestone bedded with whitish and hard mortar. The upper part of the footing presented a regular and horizontal surface and was covered by a mortar layer 4-6cms thick. Both this layer and its regularity disappear at the eastern end of the survey, an area where the foundation juts out up to 40cms from the exposed side of the wall and connects to the guard post south wall (WSU 3) in an irregular medium-sized limestone conglomerate choked with looser mortar. The base of this foundation directly bedded onto the geological sub-stratum except for the part next to WSU 3, that bedded on to what seemed to be the remains of the dismantling of this structure.

1988; Bidwell et al 1988; Morillo 1991), and in the urban scope, where there are many examples of late empire date (Fernández Ochoa 1997a). However, it is worth noting that the excavated area has provided information on a hitherto undocumented area: the quadrangular structure located in the inner wall seems to belong to a small enclosure that was used as a guard post. The arrangement of this structure also has parallels in other Hispanic enclosures such as the late wall at Inestrillas (Hernández Vera 1982; Fernández Ochoa & Morillo 1991; 1992), circular towers projecting to the exterior, or the not so-well studied Caurium (Thouvenot 1961; Fernández Ochoa & Morillo 1991; 1992), with quadrangular bodies to the inner part of two of its gates. The guard post discovered at Gijón is in accordance (with some slight variations in its orientation) to the proposal posed by one of the authors, which observed the existence of quadrangular in-wall enclosures whose function was to lodge the guard post (Figs. 5 & 6). Bibliography Bennett J. 1988 The principal gateways of masonry forts on the Hadrianic frontier in England: aspects of their construction, planning and probable appearance. Portae cum turribus. Studies of Roman fort gates. British Archaeological Reports Int. Ser. 206. (Oxford): 113141. Bidwell P., Miket R. & Ford B. 1988 The reconstruction of gate at the Roman Fort of South Shields. Portae cum turribus. Studies of Roman fort gates. British Archaeological Reports Int. Ser. 206. (Oxford): 151231. Fernández Ochoa C. 1994 Una industria de salazones de época romana en la Plaza del Marqués (Gijón, Asturias). (Gijón). Fernández Ochoa C. 1997a La muralla romana de Gijón (Asturias). (Gijón). Fernández Ochoa C. 1997b Gijón en el Bajo Imperio: la evidencia arqueológica. Actas Congreso Internacional La Hispania de Teodosio, (Segovia-Coca 1995). Volume 2. (Salamanca): 451-464. Fernández Ochoa C. & Morillo Cerdán A. 1991 Fortificaciones urbanas de época bajoimperial en Hispania. Una aproximación crítica. I. Cuadernos de Prehistoria y Arqueología de la Universidad Autónoma de Madrid 18: 227-259. Fernández Ochoa C. & Morillo Cerdán A. 1992 Fortificaciones urbanas de época bajoimperial en Hispania. Una aproximación crítica II. Cuadernos de Prehistoria y Arqueología de la Universidad Autónoma de Madrid 19: 319-360. Fernández Ochoa C. & Morillo Cerdán A. 1997 Urban fortifications and land defence in later Roman Spain. In W. Groenman-can Waateringe, B.L. van Beek, W.J.H. Willems & S. L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 343-346. Gros P. 1996 L´Architecture romaine du début du IIIeme. Siècle av. J.C. à la fin du Haut Empire I. Les Monuments publiques. (Paris). Hernández Vera J.A. 1982 Las ruinas de Inestrillas. Estudio Arqueológico. Aguilar del Río Alhama, La Rioja. (Logroño).

Interpretation and final considerations This excavation has allowed us to move forward in the analysis of late imperial Gijón, when the city enjoyed a degree of economic prosperity (Fernández Ochoa 1997b). The construction of the wall is associated with this period. The city wall has links with other cities such as Braga, Lugo, Astorga, Iruña or León. It was a response to a new imperial strategy that, from the last quarter of the C3rd AD onwards, fortified some regional centres in north-west Hispania and south Aquitania, sites that acted as strategic protection points for the annona militaris that, through Bourdeaux, was transferred to the limes. The provisional results of the salvage excavation at the 10 Recoletas site in Gijón also permits us to pose some very interesting questions about the construction features of the Roman wall and its history which covers a time-span from the late Roman period (the end of C3rd - beginning of the C4th AD) to the end of medieval times (C14th). The remains of the foundations and the first courses of the wall located in Area B are the only documented structures. It is a most interesting area corresponding to the main gate to the urban enclosure entrance. The arrangement of this gate, consisting of two flanking quadrangular towers flanking projecting into the exterior, has numerous parallels (Fernández Ochoa 1997a: 239), both in the military area of the limes, in Britain and Spain (Lander 1984; Bennett 815

Limes XVIII

Johnson S. 1983 Late Roman fortifications. (London). Lander J. 1984 Roman stone fortifications. Variation and change from the first century AD to the fourth. British Archaeological Reports Int. Ser. 206. (Oxford). Morillo Cerdán A. 1991 Fortificaciones campamentales de la época romana en España. Archivo de Español Arqueologia 64: 135-190. Thouvenot R. 1961 Les remparts romains de Coria d´Estremadoure. R.E.A. 63.

816

C. Fernández Ochoa et al: New intervention in the surroundings of the gate of the Roman wall of Gijón

Fig. 1. Location of the salvage excavation at number 10 Recoletas Street in the Cimadevilla neighbourhood (Gijón, Asturias, Spain).

Fig. 2. Stratigraphic matrix.

817

Limes XVIII

Fig. 3. Plan of documented structures.

Fig. 4. Different stratigraphic sequences of the salvage excavation.

818

C. Fernández Ochoa et al: New intervention in the surroundings of the gate of the Roman wall of Gijón

Fig. 5. New documented structures according to the proposal of C. Fernandez Ochoa (1997).

819

Limes XVIII

Fig. 6. Reconstructed hypothesis. Extra-walls elevation and floor gate.

820

The material culture of Hadrian’s Wall Lindsay Allason-Jones A considerable amount of work has been carried out in the past 20 years on the study of the material culture of Hadrian’s Wall. This paper sets out to examine some aspects of this work, in particular looking at some of the questions and answers which have arisen during the course of the work and suggesting areas which would be worthy of further investigation, either on the north-west frontier itself or on the other borders of the Roman empire

At the beginning of the 1970s, for all the excavations which had been carried out on Hadrian’s Wall, only about 2% of the small finds from the frontier zone had been published and most of that 2% could be described as exotica, such as the Aesica hoard of jewellery (Charlesworth 1973). In the past 20 to 30 years, however, the picture has changed dramatically and the time has come to take stock. This paper sets out to examine some aspects of the work, looking particularly at some of the questions and answers which have arisen during the course of the investigations, and suggesting areas which would be worthy of further study, either on the north-west frontier itself or on the other borders of the Roman empire.

include domestic, luxury and religious artefacts, can we now anticipate what we might find if we excavate a barrack block or a headquarters building in the future? The recent work on forts in Scotland has provided a comparable data base and it has become increasingly clear that a Scottish fort was very different to a fort on Hadrian’s Wall in the quantity, quality and type of small finds present in their assemblages, although they may not necessarily resemble each other. In neither area are we yet able to identify a military unit from the material it leaves behind. We know that some auxiliary units, such as the Dacians or the Hamian archers, had unique weapons or armour, but the excavations at Birdoswald, where we know the Dacians were stationed, have not produced a Dacian falx (Wilmott 1997; CSIR I.6.266) nor have fragments of any of the distinctive Hamian helmets been found anywhere on the Wall (CSIR I.6.198). From the evidence of the finds we might conclude that most auxiliaries looked depressingly similar, except possibly on special occasions when they might wear their ceremonial kit - such items would be well looked after, however, and are unlikely to have been lost, worn out or discarded when a fort was abandoned (Bishop 1989).

The discussion will be divided into four main groups: military studies, gender studies, industry and technology, and domestic artefacts. It is probably true to say that, other than pottery, it is the military artefacts which have received the most attention in the field of material culture studies on Hadrian’s Wall. Even if many of the finds specialists working in the area were not themselves keenly interested in Roman military equipment, the growth in re-enactment societies would have ensured that spearheads, lorica segmentata and the like, would be studied in depth. Specific areas of research have included Jon Coulston’s work on archery equipment (Coulston 1985) and Bill Griffith’s study of slings and other airborne weaponry (Griffiths 1989). The publication of the Corbridge Hoard has also added greatly to our understanding of Roman armour and weapons (AllasonJones & Bishop 1988). All these three studies provided detailed analyses of the objects and used sculpture and classical texts in the discussion of the artefacts, combining classics and archaeology to provide an enhanced picture of the equipment of the Roman army.

Earlier generations of archaeologists considered it a simple matter to identify which forts had legionaries and which auxiliaries in the C1st AD, working from the basic premise that only legionaries wore lorica segmentata. This premise and the consequent conclusions which were developed from it were disproved in 1986 by Valerie Maxfield (Maxfield 1986). The average fort also rarely produces enough harness to indicate whether it had cavalry as well as infantry troops. Again, this is not too surprising. A cavalryman would have taken great care of his harness or would have had a groom to do so for him (Speidel 1989), and each fort, whether cavalry or infantry, would have had a large number of horses in residence or passing through.

To many people, of course, weapons and armour are the Roman army, but weapons and armour are not the only artefacts to be found on a fort on Hadrian’s Wall. Indeed, if one looks at fort assemblages as a whole, the proportion of objects which can be called ‘military equipment’ is actually very small and this reflects the fact that an average soldier’s life was not entirely taken up with fighting. To the individual soldier his personal comfort, social life and religious beliefs may have been of more immediate concern than his long-term military career. So what can one expect to find on a fort on Hadrian’s Wall? Objects such as spears, shield binding and chapes may be present but one is just as likely to find votive figurines, jewellery, or even wooden chair legs. So if we widen our remit to

An unexpected bonus for military studies throughout the empire has been the discovery of the Vindolanda writing tablets (= TV). These are generally regarded as inscriptional evidence rather than small finds, but their interest for small finds specialists lies in the information the tablets offer about the supply and cost of some of the objects. References to 12 shoemakers in TV 155 confirm the theory that cobblers were working in the fabricae so hobnails and shoemaking equipment could be expected if the sites were excavated (Bowman & Thomas 1994: no. 155). One letter refers to a dealer called Gracilis sending hobnails worth 2 asses, which gives a clue as to the

821

Limes XVIII

probable cost of a pair of soldier’s boots (Bowman & Thomas 1994: no. 186). TV 309, on the other hand, implies that some requisites, which we might have expected to have been made on site, were being imported ready-made in kit form as Metto tells Advectus that he has sent him wooden items which include 34 hubs, 38 cart axles, 300 spokes, 26 bed planks, 8 seats, and 6 benches as well as 6 goat skins (Bowman & Thomas 1994, no. 309).

indicated that most of the bronze artefacts produced in the area in the C2nd and C3rd were made of leaded bronze or gunmetal (Dungworth 1995; Allason-Jones & Dungworth 1995). This result differs from the analysis of the C1st military equipment, which has indicated that the latter objects were invariably of brass. This difference may be because of the different manufacturing techniques employed; lorica segmentata fittings, for example, were made by hammering and cutting sheet metal while openwork mounts of the sort made on the Hadrian’s Wall sites were cast in moulds.

Most of the items mentioned in the Vindolanda tablets could have been used by soldiers or civilians and recent studies have revealed that the use of the words ‘military’ and ‘civilian’ can be misleading or simplistic. The finds from the excavated turrets on Hadrian’s Wall included nail-cleaners, needles and tweezers despite the fact that the turrets were only occupied for 40 years in two clear-cut stages and then only for military purposes (Allason-Jones 1988). These items when found in other contexts are traditionally catagorized as being evidence for a civilian or even a female presence, but civilian and military populations have many of the same needs: C2nd soldiers would have had to mend clothes, clean their nails and remove splinters, while civilians may have required arrows and knives for hunting, hobnails for their boots, and chapes for their knife scabbards. Work over the past 20-30 years may have cleared up many detailed questions about equipment in the military zone but has also shown that the usual separation of finds into ‘military’ and ‘civilian’ is invalid.

The presence of zinc in the C2nd and C3rd objects also suggests that they were being made from scrap-metal, as this type of alloy is likely to have been the result of melting brass and bronze together. Fragments of scrapmetal, neatly folded into little bundles to fit into crucibles, have been found at Turret 26A and Newcastle fort, but enough survives elsewhere in the military zone to suggest that this recycling was a widespread practice. Analysis also suggests that the artisans were carefully adding a small quantity of lead so that the molten metal was able to flow easily into the moulds. There was a high level of knowledge of metallurgy and skill in working the alloys being used - more than an individual soldier might be expected to have without extensive training; yet, despite this, much of the metalworking on the Wall was on a small scale, even though the products and sites were military. Possibly the authorities provided the raw materials and the ‘recipe’ for the alloys to be used for each type of object but the work was then farmed out to individual specialists. The most important aspect of this study is the inference that even the most unpromising scrap of metal or clay found on a site can produce useful evidence for activity on that site and contribute to our discussion of the Roman army - a topic which some archaeologists might claim to be already exhausted. It is always horrifying to imagine the amount of evidence which our predecessors threw away because they lacked the techniques for detailed analysis - and no doubt our successors will be equally disparaging about our profligacy.

Equally, the traditional confidence that it is possible to tell which artefacts were used by women and which by men in the Roman period has come under attack (Allason-Jones 1995). The excavation of Barrack Block 13 at Housesteads has been of enormous value in this discussion about objects and gender (Daniels & Rushworth forthc.). The recording method allowed the artefacts to be attributed to a specific room or, in the later periods, to a specific chalet. This revealed that, of the ambiguous material found, only brooches were used, or rather lost, in barrack blocks or chalets, whilst the artefacts which are likely to have been worn by women alone, such as hairpins, are confined to the centurion’s quarters. There is no evidence to suggest that centurions were not allowed to marry during the imperial period and there are many inscriptions around the empire which attest to the presence of centurions’ wives in forts (Allason-Jones 2000); where there has been debate is in the location of these dependants’ accommodation. The Housesteads finds confirm the theory that they were housed within the centurions’ quarters.

Small finds are often the best evidence for revealing the details of everyday life in the past. Recent excavations at Vindolanda have produced brushes and brooms - rare examples of the archaeology of housework - while scraps of textile found on the site have been shown to have been dyed with the local bedstraw, as opposed to French madder which the classical writers, such as Pliny, tell us was used for dying military tunics (Taylor 1983; 1990; Pliny NH XIX.48). This would imply that the local soldiery were being clothed in locally produced garments. Research by the re-enactment societies have proved that a soldier’s tunic does not last very long, so a local source of supply would seem sensible.

Less glamorous than military equipment research or gender studies, work on Roman technology and industry has not hit the headlines but this is where the most interesting results have been obtained. Evidence for metalworking has been found at Turrets 26A and 18B, at Sewingshields milecastle, and in the forts of Housesteads, Vindolanda, Stanwix, Carrawburgh, South Shields, Newcastle and Piercebridge. As was reported at a recent Limes Congress the analysis of the metalwork by David Dungworth has

The damp conditions at Vindolanda and Carlisle have also produced some outstanding leather objects. Both sites have provided the evidence which has allowed Carol van Driel Murray to reveal the pattern and shape of a Roman tent, 822

Lindsay Allason-Jones: The material culture of Hadrian’s Wall whilst at Vindolanda the shoes have survived so well that Dr van Driel Murray has been able to trace the footwear of the Commanding Officer’s family to the extent that we now know that his son had a problem with his feet (van Driel Murray 1993; 1995)

assemblage from a Scottish fort, but how does it compare to material from Spain or Germany? Can the objects start to suggest units? Do the finds on a fort occupied by the 1st cohort of Vardulli, for example, compare in any way with finds from the area of north-west Spain where the unit was first recruited? These seem obvious questions but at the moment they are unanswerable.

In looking at domestic artefacts the native sites have not been ignored. Hadrian’s Wall is unique in the Roman world in that excavations by Prof. George Jobey have revealed a considerable body of evidence for the lives of the native population on both sides of the frontier; this work was published mostly in Archaeologia Aeliana between 1955 and 1992. The evidence does not suggest a population eager to adopt Roman practices but seems to hint at a shrewd group of people providing trinkets specially aimed at Roman military taste (Allason-Jones 1991). In other words, the invading army may not have been exploiting the downtrodden locals in quite the way that the Carvoran Modius implies from its misleading inscription (RIB 2415.56); the canny Brigantian and Votadinian tribesfolk may well have been fully aware of the marketing potential of a long-term military presence.

Secondly, we know very little of the funerary habits of the military or associated civilian populations. The cemeteries of Petty Knowes in Northumberland and Lanchester in County Durham were disappointing in their finds, the soldiers having been apparently buried only in their tunics and boots (Charlton & Mitcheson 1984; Turner 1990). Recent excavations at South Shields suggest that there is much to be learned about the religious beliefs and burial customs of the frontier population (Snape 1994). It is still impossible to identify the ethnic origins or religious beliefs of people from the cemetery evidence, largely because we don’t know what we are looking for or the significance of that evidence. Thirdly, little excavation has been carried out in the vici of Hadrian’s Wall. Recent years have seen an increase in the use of geophysics and other surveying techniques which have shown that these vici were much larger than previously thought (Biggins & Taylor 1999). Were the vici abandoned in the C3rd and C4th - a theory which led to the idea of fort chalets being married quarters - or did occupation simply retreat away from the fort walls? (Daniels 1980).

Even so, the lack of coins and material goods on the native sites suggests that the indigenous peoples to the north of the Wall may not have judged success in monetary terms. Similar settlement sites to the south of the Wall, such as Thorpe Thewles in Cleveland, have produced far richer assemblages in modern terms (Heslop 1987). Was this the result of the Wall being built or does it stem from different tribal attitudes to material goods which has nothing to do with the presence of the Roman invaders? Possibly the northern tribes displayed status by intangible indicators, although exactly what an individual object meant to its owner remains one of the great imponderables of small finds research.

We start the new millennium in good order as far as the Roman small finds from Hadrian’s Wall are concerned. From a 2% publication record the latest estimates suggest that 92% of the material is now either published or in press. With this plethora of publication contradictions, paradoxes and oddities are beginning to emerge and it is now important for syntheses to be produced. But equally we must also continue to re-assess the material. New scientific techniques are being applied all the time which are leading to new ideas about sites and the dating of those sites. Small find specialists need to take note of the changes and look again and again at their evidence for dating a particular type of brooch to the C4th or calling one sort of artefact a foreign import. We need to keep an open mind all the time. As Mr Mailey says in one of Arthur Conan Doyle’s Professor Challenger stories: ‘We must fit our theories to the facts. Up to now we have fitted the facts to our theories’ (Doyle 1952).

So what needs to be done next? Firstly, despite the fact that many archaeologists consider that Roman military equipment has had more than its fair share of attention, more needs to be done. The bulk of the material from the area is now catalogued and accessible and the Russell Robinson Archive is being transferred by Mike Bishop onto a web site called Armamentarium so that researchers all over the world can contribute parallels and dating evidence (www.ncl.ac.uk/antiquities). This is important as Hadrian’s Wall’s military equipment has tended to be studied in isolation. We need to compare our assemblages with those found elsewhere in Britain and in the other provinces; only by doing this can we build up our knowledge of the Roman army and the frontier. The regular meetings of the Roman Military Equipment Conference are helping enormously here.

Finally, the standards which have been achieved must be sustained. It would be regrettable if the progress which has been made in the past 20 to 30 years was to lose momentum because of lack of space in excavations reports or if it was decided that certain types of object had been done, ticked off, and need not be given further attention. To describe a particular object by its type number alone is not enough - a proper description, including measurements and an illustration must still be included so that not only

Can a fort assemblage from Hadrian’s Wall be considered to represent the norm for the whole empire or should we conclude that its geographical location, its political raison d’être and/or its history make it likely that it represents the exception to the rule? We already know that the material from a Hadrian’s Wall fort is likely to be different to an 823

Limes XVIII

can our initial identification be tested but so that the full information, which never goes out of date, continues to be accessible to everyone. If full publication of all the artefacts from sites throughout the Roman empire continues, the next 20 or 30 years should see some remarkable developments in our knowledge of how that Empire worked.

northern Britain. Unpublished PhD thesis, Durham University. Griffiths W.B. 1989 The sling and its place in the Roman imperial army. In C. van Driel-Murray (ed.) Roman military equipment: the sources of evidence. British Archaeological Reports Int. Ser. 476. (Oxford): 255279. Heslop D.H. 1987 The excavation of an Iron Age settlement at Thorpe Thewles, Cleveland, 1980-1982. CBA Research Report 65. (London). Maxfield V.A. 1986 Pre-Flavian forts and their garrisons. Britannia XVII: 59-72. McCarthy M. 1990 A Roman, Anglian and Medieval site at Blackfriars Street, Carlisle. Cumberland & Westmorland Antiquarian & Archaeological Society Research Series 4. (Kendal). McCarthy M. 1991 Roman waterlogged remains at Castle Street, Carlisle. Cumberland & Westmorland Antiquarian & Archaeological Society Research Series 5. (Kendal). Snape M.E. 1994 An excavation in the Roman cemetery at South Shields. Archaeologia Aeliana 5th Ser. 22: 4366. Spiedel M.P. 1989 The soldiers’ servants. Ancient Society 20: 239-247. Taylor G. 1983 Detection and identification of dyes on preHadrianic textiles from Vindolanda. Textile History 14(2): 115-124. Taylor G. 1990 Ancient textile dyes. Chemistry in Britain 26 (12): 1155-1158. Turner R.C. 1990 A Romano-British cemetery at Lanchester, Durham. Archaeologia Aeliana 5th Ser. 18: 63-77. Van Driel-Murray C. 1993 The leatherwork. In R. Birley (ed.) Vindolanda Vol. III: The early wooden forts. (Bardon Mill): 1-75. Van Driel-Murray C. 1995 Gender in question. In P. Rush (ed.) Theoretical Roman Archaeology: Second Conference Proceedings. Worldwide Archaeology Series 14. (Aldershot): 3-21. Wilmott T. 1997 Birdoswald: Excavations of Roman fort on Hadrian’s Wall and its successor settlements, 1987-92. English Heritage Archaeological Report 14. (London).

Bibliography Allason-Jones L. 1988 ‘Small finds’ from turrets on Hadrian’s Wall. In J.C. Coulston (ed.) Military equipment and the identity of Roman soldiers. British Archaeological Reports Int. Ser. 394 (Oxford): 197-233. Allason-Jones L. 1991 Roman and native interaction in Northumberland. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 1-5. Allason-Jones L. 1995 ‘Sexing’ small finds. In P. Rush (ed.) Theoretical Roman Archaeology: Second Conference Proceedings. Worldwide Archaeology Series 14 (Aldershot): 22-32. Allason-Jones L. 2000 Women and the Roman army in Britain. In A. Goldsworthy & I. Haynes (edd.) The Roman army as a community. Journal of Roman Archaeology Supplementary Series 34: 41-51. Allason-Jones L. & Bishop M.C. 1988 Excavations at Roman Corbridge: The hoard. English Heritage Archaeological Report No. 7. (London). Allason-Jones L. & Dungworth D. 1995 Metalworking in Hadrian’s Wall. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 317-322 . Biggins J.A. & Taylor D.J.A. 1999 A survey of the Roman fort and settlement at Birdoswald, Cumbria. Britannia XXX: 91-110. Bishop M.C. 1989 O Fortuna: a sideways look at the archaeological record and Roman military equipment. In C. van Driel-Murray (ed.) Roman military equipment: the sources of evidence. British Archaeological Reports Int. Ser. 476 (Oxford): 1-12. Bowman A.K. & Thomas J.D. 1994 The Vindolanda Writing Tablets. (London). Charlesworth D. 1973 The Aesica hoard. Archaeologia Aeliana 5th ser. I: 225-234. Charlton D.B. & Mitcheson M. 1984 The Roman cemetery at Petty Knowes, Rochester, Northumberland. Archaeologia Aeliana 5th ser. 12: 1-31. Coulston J.N.C. 1985 Roman archery equipment. In M.C. Bishop (ed.) The production and distribution of Roman military equipment. British Archaeological Reports Int. Ser. 275. (Oxford): 220-236. Daniels C.M. 1980 Excavation at Wallsend and the fourth century barracks on Hadrian’s Wall. In W.S. Hanson & L.J.F. Keppie (edd.) Roman Frontier Studies 1979. British Archaeological Reports Int. Ser. 71. (Oxford): 173-194. Daniels C.M. & Rushworth A. forthc. Report on Housesteads Roman fort. Doyle A.C. 1952 The Land of Mist. In The Professor Challenger Stories. (London). Dungworth D. 1995 Iron age and Roman copper alloys from

824

A revised programme and chronology for the building of Hadrian’s Wall1 Julian Bennett Hadrian's Wall is one of the most impressive engineering achievements of the Roman army and has attracted the attention of scholars and the curious since at least the C8th. Serious study of the remains began in 1851 and continued in fits and starts until the 1980s since when various measures have been initiated to implement a comprehensive archive and full record of the surviving remains, as well as an effective management plan. As a result of this long period of investigation, a consensus of opinion has developed about the overall design and chronology of the Wall, although much of the evidence involved is circumstantial at best, and based on conjecture alone at worse. Moreover, the logistical problems involved in the construction of the Wall have hardly ever been addressed, even though these have a direct bearing on both design and chronology. This paper is intended to redress these matters, by reviewing the existing evidence for the Wall's design and chronology, and combining this with logistical practicalities to arrive at a set of alternative conclusions with regard to both issues.

fourth revised edition of Breeze and Dobson’s Hadrian’s Wall (2000: 72-87). A comprehensive review of the structural and other evidence for the ‘Birley Model’, however (Bennett 1990: Chs. 4-8), produced a body of data suggesting the accepted version of the Wall’s building programme and chronology required revision. In particular, it appeared that there were in fact three and not two major phases in the frontier’s construction, and that a far simpler work programme could be suggested than the accepted complex scheme of alternating ‘legionary lengths’, each the work of a single legion. Moreover, the evidence also indicated that the building of the Wall began before 122, and probably continued more or less non-stop until 138, when the frontier became redundant after the reoccupation of Caledonia. The purpose of this paper is to present these conclusions to a wider audience, but a convenient place to start is with an overview of the ‘Birley Model’.

Introduction Any attempt at revising the accepted building programme and chronology of Hadrian's Wall requires some indication of how current opinions have developed. By the early C18th, for example, it was clear that what was then known as ‘Severus’ Wall’ was built by the legiones II Augusta, VI Victrix and XX Valeria Victrix, and the work apparently equally apportioned amongst them (Horsley 1732: 130). Yet it was not until the late 1920s that the subject was given any serious consideration. The stimulus was the recognition that the Antonine Wall was built by all three British legions, each assigned a series of discontinuous ‘building lengths’ marked by ‘Distance Slabs’ indicating the unit responsible (MacDonald 1921: as now understood, the 37 miles of rampart were divided into 15 such sectors, five for each legion). R.G. Collingwood and F.G. Simpson began an investigation of Hadrian’s Wall to see if similar evidence for ‘legionary lengths’ could be isolated there. Lacking anything comparable to the Antonine Wall’s ‘Distance Slabs’, they concentrated on examining the physical characteristics of Hadrian’s Wall and its interval structures to determine if these might provide the required evidence. It was, after all, already known that the milecastle gateways in the eastern sector displayed certain structural variations implying design and construction by different groups of builders (Gibson & Simpson 1911: 406-408). Their results seemed conclusive, that there were indeed a series of structural variations in both the milecastles and turrets which might be assigned to three different groups of builders, namely the three British legions (Collingwood 1931: 86-87; Simpson 1931: 309310). Subsequent discoveries seemed to confirm the belief (eg. Richmond 1950), but it was Eric Birley’s mastery of both the epigraphic and archaeological evidence which allowed the presentation of a probable scheme for the design, building programme and chronology of the frontier (E. Birley 1961: 94-95).

Part I: The ‘Birley Model’ The original scheme (cf. Breeze & Dobson 2000: 84)2 Hadrian’s Wall was initially planned to be 76 Roman miles long from the lowest practicable crossing-point on the River Tyne, to a like point on the Solway (Fig. 1). It was to consist of a continuous V-sectioned ditch, 30 pedes Monetalis wide and 10 p.M. deep, fronting a 10 p.M. (2.95m) wide stone wall from the Tyne to the Irthing, and a 20 p.M. (5.9m) wide rampart of ‘turf’ - in fact usually of turf and earth - from the Irthing to the Solway: whether of stone or ‘turf’, the two barriers correspond in form and nature to Roman military building practices of the time (see now Lander 1984: 35-37, 43; Jones 1975: 80). Every Roman mile of curtain (1479.55m) was marked by a ‘milecastle’, masonry-built on the Stone Wall, and of ‘turf’ and timber on the Turf Wall, while between each

What may be conveniently termed the ‘Birley Model’ isolated two main phases in the building of Hadrian’s Wall, the ‘original scheme’, and its modified version, the ‘second scheme’. Subsequent fieldwork, epigraphic analysis and excavation led to the modification of certain of Birley’s original conclusions (eg. Hooley & Breeze 1968; Moss 1969), but the basic scheme and chronology he devised was not significantly altered, and it became effectively crystallised in 1976 with the publication of the

1

The arguments detailed here were first reviewed by James Crow, the late Charles Daniels and Stephen Johnson; certain details were later discussed with Paul Bidwell, David Breeze, Nick Hodgson and Tony Wilmott; and Alan Greaves provided material assistance in completing this paper: I am grateful to all for their comments and help, but they are not, however to be blamed for the views I have chosen to express. 2 As many of the structural features relating to the form and building of the Wall are well known, references are provided only for those works strictly pertinent to the argument presented here. These primary references can be found in the relevant pages of Birley 1961 and Daniels 1978.

825

Limes XVIII

milecastle were two square masonry ‘turrets’. A system of almost identically-planned and regularly-spaced interval structures continued beyond Bowness-on-Solway for an uncertain distance along the Cumbrian coast (Fig. 1), suggesting they formed part of the original scheme, although it seems these coastal ‘milefortlets’ and ‘towers’ were never linked with a continuous barrier.

decision’. Once the ‘fort decision’ was implemented, however, work resumed on the Stone Wall and its interval structures. The curtain and the side walls of the unfinished milecastles were now completed to a narrower standard than first designed, roughly 2.20m, and significantly less than the 10 p.M. (2.95m) of the ‘original scheme’. This Narrow Gauge Wall was certainly butted against the perimeters of the forts at Housesteads and Great Chesters, and apparently at Birdoswald as well. It was also used for the spur-wall linking the bridgehead at Newcastle to a new fort at Wallsend, and for a re-building of the Turf Wall in stone from the Irthing to the Burtholme Beck. The Narrow Gauge Wall and these elements of the frontier system were thus broadly contemporary, yet belonged to a later stage in the ‘second scheme’. It was also probably now that the fort at Carrawburgh was added to the frontier system, for it was built over the vallum, while the Flavian-Trajanic fort at Carvoran - if indeed empty at the beginning of Hadrian’s reign - was recommissioned, thus making a total of 14 forts along the Tyne-Solway line (Fig. 2).

The evidence of inscriptions and historical circumstances indicated to Birley that work on the frontier was initiated during Hadrian’s visit to the province in 122, following his review of the military dispositions in the Germanies (see now A. Birley 1997a: 123, 131-133). By 17 July 122, A. Platorius Nepos, legate of Germania Inferior until at least 121, had been made governor of Britain: as his name appears on inscriptions from certain of the Stone Wall milecastles, and apparently on one from the Turf Wall, it was suggested that Hadrian had personally placed Nepos in charge of the project. Furthermore, at about this time, the VI Victrix was dispatched from Vetera in Germania Inferior to Britain to replace the IX Hispania, transferred by Trajan to (?)Noviomagus, thus restoring the province’s garrison to its notional 3-legion strength (see now A. Birley 1971: 81-82). As Breeze and Dobson (2000: 64) observed, it is “...certainly tempting to connect all three movements [of Hadrian, Nepos and the VI Victrix] from Germany to Britain, and place them in the same year” (cf. A. Birley 1981: 101, 104; 1997a: 124). Given that the structural pointers on the Stone Wall - specifically the axes of individual milecastles and the form of their gateways allowed the isolation of three distinct building ‘signatures’, Birley deduced that the construction of the masonry sector of Hadrian’s Wall had been equally divided into a series of ‘legionary lengths’ equally apportioned between all three British legions (eg. Fig. 3, K: current opinion is that these were of 5 millia passum east of the North Tyne, and mainly of 6 m.p. between the North Tyne and the Irthing: cf. Breeze & Dobson 2000: 76). The logical conclusion which followed was that the building of the Turf Wall and the Cumbrian coast structures was apportioned in a like fashion, although secure proof was (and remains) lacking.

A chronology for Birley’s ‘second scheme’ was again provided by epigraphy and known historical events. To begin with, inscriptions from the forts at Benwell and Halton Chesters (RIB 1340; 1427) indicated they were under construction when Nepos was still governor (ie. after July 122 and, as we now know, before 20 August 127, by when he had been replaced by L.Trebius Germanus: Nollé 1997: 269-275). But while the regular spacing of the primary forts on the Wall - including that at Wallsend suggested their locations were determined by Nepos, their actual construction may well have lasted until at least 128, the earliest date of a building inscription from Great Chesters (RIB 1736). Moreover, as the north-west angle tower of Great Chesters appeared to have been deliberately offset to allow direct bonding with the Narrow Gauge curtain, this meant that the construction of the Narrow Wall was probably begun while Great Chesters was being built, ie. c.128. Such seemed confirmed by a further inscription from the secondary-phase fort at Carrawburgh recording C. Julius Severus as governor (RIB 1550), indicating the fort was added between c.130-132. In other words, it was in the 130s that the Stone Wall took its final form, but a building inscription from Carvoran of 136/137 (RIB 1778: note also RIB 1808, almost certainly Hadrianic), indicates that work carried on in this area until the end of Hadrian's reign. Finally, as the programme of replacing the Turf Wall with the Narrow Gauge curtain had not extended beyond the Burtholme Beck before being halted, it seemed reasonable to suggest that this work was in progress and then abruptly halted in c.138/139, when Antoninus Pius made the decision to re-occupy Caledonia.

The second scheme (cf. Breeze & Dobson 2000: 85) The archaeological evidence indicated that work was abruptly suspended on building the Stone Wall before it was completed. In certain lengths of the curtain, both its superstructure and that of several milecastles and many turrets, were left standing to a variety of heights (see now Hill 1991: 35-36); in other areas, work had not yet begun on laying the foundations for either curtain or milecastle perimeter walls. Birley deduced that the reason for the stoppage was the resolution to build a series of forts on the line of the curtain itself, as at least five of these - Halton Chesters, Chesters, Housesteads, Great Chesters and Birdoswald - were erected over earlier features associated with the ‘original scheme’. At about the same time, the decision was made to institute the vallum: it clearly deviated to avoid certain of the forts, and was thus either contemporary with or immediately followed the ‘fort

Part II: A revised building programme Phase I: The blueprint There is little reason to doubt the broad outline of Birley’s 826

Julian Bennett: A revised programme and chronology for the building of Hadrian’s Wall ‘original scheme’ for the design of the Wall (but see Bennett 1998: 31, on the spacing and number of the interval structures from Newcastle to Denton; and Crow 1995: 18-12, for the suggestion that the Wall was originally planned to include the forts). Analysis of the structural pointers in the Stone Wall, however, suggests its construction was more likely divided into two distinct building sectors, and not a succession of ‘legionarylengths’ distributed between three legions. This can be deduced from the amount of work actually completed before the ‘fort decision’ was made, and the corresponding indications of the order in which the interval structures and lengths of connecting curtain were in fact built (Fig. 3). Between the Tyne and the North Tyne, for example, the Broad Gauge foundation is continuous, and several lengths of Broad Gauge curtain were erected above it (Fig.3, A and B). All the excavated turrets were apparently built with Broad Gauge wing-walls to allow them to be bonded to a Broad Gauge curtain, but not so the milecastles (Fig 3, D, E and F). Of the 14 excavated Stone Wall milecastles, 8 equally divided between the east and west ends of this sector (4, 9, 10, and 23-27) - are entirely or substantially built to Broad Gauge standard; five others (13, 17, 18, 20 and 22) appear to have been completed to Narrow Gauge standard; but only one, 19, has its perimeter certainly constructed in this way. From the North Tyne to the Irthing, by contrast, Broad Gauge foundation is not only discontinuous, but little - if any - Broad Gauge curtain was built (Fig 3, A and B). At least two if not three of the turrets at the west end of this sector (39a, 44b and possibly 45a (Crow 1991: 62-63)) are coterminous with the Narrow Gauge curtain (Fig. 3, G); while of the excavated milecastles, only three (35, 47 and 48) were completed to Broad Gauge, six (29, 30, 31, 38, 42 and 43) have a Narrow Gauge perimeter, and four (33, 37, 39 and 40) were built throughout to the Narrow Gauge standard (Fig. 3, D, E and F).

progressed. An obvious objection to this suggestion is that three discrete groups of associated structural features - ‘building signatures’ - have been identified along the Stone Wall, distinguishing the work of three separate building parties. These ‘building signatures’ are based on particular combinations of milecastle gateway-types and milecastle axis; the position of turret doorways (at the east or west end of the rear wall); and whether or not an offset existed at the rear of the Wall curtain (Breeze & Dobson 2000: 6768, 73). With regard to the milecastles, one of these combinations, Birley’s gateway type I/short axis (Breeze & Dobson 2000: 73: Type A), is found at milecastles 37, 38 and 42, all known from epigraphic evidence to have been built by the legio II Augusta. It has thus been assumed that the other two combinations should represent the work of the VI Victrix and XX Valeria Victrix (but note now Breeze & Dobson 2000: xvii, 68), and confirmation that all three legions were involved in building the Stone Wall. The idea is certainly a seductive one, yet it must not be forgotten that its ultimate inspiration was the knowledge that the Antonine Wall was equally apportioned amongst the three British legions (MacDonald 1921: in one sense the wheel has come full circle, with the identification of ‘legionary-types’ of building styles and interval structure on the Antonine Wall, as, eg. Hanson & Maxwell 1983: 121-131; also Breeze & Dobson 2000: 99). However, as was implied by Charles Daniels (1979: 361), “...the playing of this ‘Wall Game’”, using specific structural types to determine who built what on the Wall, is an exercise bedeviled by subjectivity - and, we might add, a lack of secure evidence. It also relies on three assumptions: that there were three legions in the province when the Wall was built; that only three structural combinations exist, representing the work of the three legions; and that the work was equally shared amongst all three. For the first we cannot say either way whether or not the legio VI Victrix arrived in the province before 125: any year between 121 and 125/6 is possible (cf. A. Birley 1981: 101, 104). For the second we might note that we have reliable details concerning gatewaytype and axis for only 18 of the 48 milecastles between the Tyne and North Tyne, and at least four appear to be ‘hybrids’, in that they do not conform to the accepted gateway-type/axis combinations (eg. milecastles 18, ?19, 33, and 47). Following Stevens (1966, passim; cf. Breeze & Dobson 2000: 68), they are generally discounted as ‘exceptions which prove the rule’, one legion having begun the work, another finishing it after the ‘fort decision’; or, as with 47, on the grounds that the original excavators misidentified the gateway-type (Hooley & Breeze 1968: 100; cf. Breeze & Dobson 1987: 69, a suggestion not present in Breeze & Dobson 2000). We might also note that at least six turrets (29a, 29b, 34a, 35a, 36a and 40b) have their doorways towards, rather that at the end, of the rear walls, and thus in theory could allow the isolation of one of more additional ‘building signatures’ (the doorway position is noted in Hooley & Breeze 1968: 101).

This evidence permits of four inferences. To begin with, the respective amounts of Broad Gauge foundation and curtain in the two sectors suggest that work had progressed significantly in the Tyne-North Tyne sector before it had even begun between the North Tyne and the Irthing. Then, it suggests that the order of works in the Tyne-North Tyne Sector was to build the structures simultaneously from both the east and the west end, working towards a central point in the vicinity of milecastles 16-17. The work programme in the North Tyne-Irthing sector, however, was arranged in a very different way: instead of starting from either end of the sector simultaneously and working towards the middle, work progressed concurrently from either end of two roughly equal lengths, Wall-mile 38-39 marking the division between the two sub-sectors. It seems, therefore, that each of these two sectors was a single construction block built under the supervision and direction of their own ‘works constructor’. Such would certainly seem a logistically more efficient approach than to sub-divide the whole of the Stone Wall into a series of eight disunited ‘legionary-lengths’, the building parties from each legion ‘leap-frogging’ one another as the work 827

Limes XVIII

Regarding the third assumption, it can be unequivocally stated that no single legion can be securely associated with the initial building of the Stone Wall curtain and its interval structures. The three virtually identical inscriptions of Hadrianic date from milecastles 37 (RIB 1634), 38 (RIB 1637) and 42 (RIB 1666), recording work by the II Augusta, are all from structures completed under Nepos to the Narrow Gauge standard, and need not necessarily record work in the ‘original scheme’ (Hill 1991; Breeze & Dobson 2000: 68). The building stone of the VI Victrix found in the recessblocking of turret 33b likewise cannot be securely attributed to this phase: not only is it stylistically more likely to be Antonine than Hadrianic (Wright, quoted in Maxfield & Miket 1972: 159), but it need not even come from the turret or its immediate vicinity (compare, for example, the recessblocking at turret 39a, which includes a hypocaust pila and several elongated blocks identical to those used in a postHadrianic context at Housesteads: Simpson 1976: 100; Charlesworth 1975: 23 and pl. liv, 2). As for RIB 1852, found re-used in an C18th context west of milecastle 47, and recording building work under Hadrian by the XX Valeria Victrix, while it is usually assigned to that milecastle, it carries Hadrian's filiation and titulature, as on more grandiose public inscriptions (eg. RIB 288: Wroxeter forum; RIB 739: Bowes fort; and Wright 1965: Hardknott fort): it is thus perhaps more likely to come from either the nearby fort at Carvoran, or even from the presumed temple at nearby Rose Hill, Gilsland (Coulston & Phillips 1988: 105-106). Finally, even if RIB 1935 from milecastle TW50 is the remnant of a wooden building inscription of Nepos (on which considerable doubt might be cast: Bennett 1990: 470472), there is no indication as to which legion was recorded.

Cumbrian coast structures combined (Bennett 1990: 484). Might it be, therefore, that the Stone Wall was the quantified building requirement of one legion, the Turf Wall and Cumberland coastal system the responsibility of another? And that it was only after the enormity of the task involved in building the Stone Wall became apparent that contingents from other legions were drafted-in to help? Phase II: Dislocation and the fort decision - and the vallum As can be determined from Fig. 3, H and I, and Figs. 4 and 5, perhaps as little as 3% of the Stone Wall and its associated structures was completed to the original Broad Gauge standard before work was abruptly interrupted by the ‘forts decision’. This resolution was evidently made some time after construction of the Stone Wall had begun. At Chesters, the foundations of the fort’s perimeter were cut into a 1m deep silty-peat deposit in the base of the Wall ditch, suggesting it had already been open to the elements for at least two if not three or more years (Haverfield 1901: 85-87; 1902: 15-17). Moreover, it is equally clear that the primary forts on the Stone Wall at Chesters, Housesteads and Great Chesters, were originally built as free-standing structures, designed and built with no allowance being made for their later incorporation into the curtain. In other words, at the time they were built, it may not have been intended to complete the curtain. The main evidence for this is the lack of wingwalls on either side of the forts at Chesters and Housesteads to facilitate an appropriate junction with the Stone Wall curtain: instead the curtain, when completed to Narrow Gauge standard - was butted against the existing perimeter walls of these two forts. At Great Chesters, the evidence is not so clear, but the claimed bonding of the Narrow Gauge Wall with the fort’s north-west angle proves on re-examination to be a later rebuild, and not, as has been stated, evidence for the two being built together. On the other hand, as the Narrow Gauge curtain immediately west of the fort clearly runs over the fort’s ditches, it seems quite clear that this fort also was originally designed to stand alone.

In conclusion, while it is not disputed that the various structural combinations identified among the milecastles on the Stone Wall indicate the work of different building parties, it is not at all certain that there were only three, or that they necessarily represent the work of three separate legions. The evidence provided by the overall structural analysis indicates significant discrepancies in the amounts of work begun and completed in the Tyne-North Tyne and North Tyne-Irthing sectors, and evident differences in the way the building programme was actually organised in each. It might therefore seem more plausible that each of these two sectors was assigned as a complete building block to two quite separate commands. This might also explain why there are apparent differences in the amount of work required for the building blocks in the Tyne-North Tyne and North Tyne-Irthing sectors.

As for the vallum, there is no doubt that it was initiated at the same time the ‘fort-decision’ was implemented, although in some sectors at least it was under construction before it had been decided on exactly where to site the forts themselves (eg. Rudchester: Bowden & Blood 1991: 30). Its purpose, however, remains obscure. Generally, it is considered to be some form of security device and/or a boundary marker defining the Wall zone (eg. Stukeley 1776: 59; cf. Breeze & Dobson 2000: 57; Dobson 1986: 18). A boundary marker in the form of ditch, the fossa Regia, certainly existed in Republican Africa Proconsular, defining the southern limit of the province (Daniels 1987: 236), but the very form of the vallum excludes such an explanation. The limits of the Wall zone could have been marked much more easily and as emphatically by laying a hedge or digging a simple ditch, rather than creating this 120 p.M.

As noted, a division of the work between two separate commands would be a more efficient - and logical approach to the matter rather than subdividing each legion in Britain into separate working parties working simultaneously or even successively in a series of blocks greatly removed from each other. Indeed, some support for the idea might be found in a logistical analysis of the actual work involved in building the Wall: the man-hours needed for building the Broad Gauge Stone Wall are more than twice those required for building the Turf Wall and 828

Julian Bennett: A revised programme and chronology for the building of Hadrian’s Wall (35.5m) wide obstacle course, with its parallel 20 p.M. (5.92m) wide mounds flanking a flat-bottomed ditch of the same width and 10 p.M. (2.96m) deep - a ditch, moreover, which in some places was rigorously cut through the most intractable of rocks. As Williams (1983: 33) observed, the vallum is akin to a “tank-trap”, an obstacle virtually impassable to man, and certainly so for animals or any form of wheeled transport. As such both its form and is probable function find a parallel at Alesia, where sections of Caesar’s siege works incorporated a trapezoidal ditch (Caesar Bell. Gall. VII.72), apparently to prevent the line being crossed when there was a limited body of men available for guard duty. On the other hand, it is also clear that whatever purpose the vallum served, it had become redundant by perhaps as early as c.130, when Carrawburgh was built, and almost certainly by the end of Hadrian's reign. It was during this period that substantial sections of its mounds were levelled to form ‘crossings’ as a preliminary to the eventual complete filling its ditch (Bennett forthc.).

Phase III - The Narrow Wall After the perimeter walls of the primary forts of at least Chesters, Housesteads, Great Chesters and (probably) Birdoswald had been completed, it was decided to complete the Stone Wall curtain and its interval structures to the Narrow Gauge standard. Given that the forts at Chesters, Housesteads were built as free-standing structures, it seems reasonable to suggest that the completion of the Wall curtain was an afterthought to the ‘fort decision’, and not coeval with it. It is not entirely clear how long a period may have elapsed between the completion of these fort perimeters and work starting on the Narrow Gauge curtain, but it was evidently of some duration. Such is indicated in the way that in several sectors, Narrow Gauge curtain was built in a construction trench which cuts quite deeply into existing lengths of Broad Gauge foundation and footings, as if to ensure a secure base over what was a suspect if not unstable structure (eg. at Planetrees and at turret 33b). It is also demonstrated in the way the Narrow Gauge curtain sometimes takes an entirely different alignment from existing lengths of Broad Wall foundation (eg. at ‘Mons Fabricius’, Crow 1991: fig. 1), suggesting that the latter was either overgrown and/or seriously eroded at the time the Narrow Wall was built.

The nature of the obstacle and the evidence for its short life, along with the determination to cut it in sectors where the Wall ditch was left incomplete, as at Limestone Corner, combine to suggest an alternative explanation of the vallum’s purpose. As originally designed, the Wall was most likely intended to allow for controlled movement via the milecastles to both its north and south (Breeze & Dobson 2000: 40, 62; contra Bidwell 1999b: 59). This is indicated by the existence of the opposed gateways in the milecastles and the evidence of primary crossings over the Turf Wall ditch at milecastles TW50 and TW54 (Simpson et al 1935: 225; Simpson & Richmond 1935: 243): at the latter site, later re-cutting of the Wall ditch had almost totally removed the evidence, and probably explains why they are apparently absent opposite other milecastle gateways on the Wall. At the time of the ‘fort-decision’, however, substantial sections of at least the Stone Wall were evidently incomplete or had not yet even been started, while lengths of the Wall ditch remained un-dug. In the absence of a continuous barrier (the Wall) to control unauthorised civilian movement across the formal provincial boundary at the time attention was focussed on building the forts, the vallum could have been intended as a substitute for the Wall itself. It was, after all, an equally effective obstacle, and moreover one which could be quickly built. Even if it was only the four auxiliary units suspected to have been involved in its construction who were responsible for the entire work, logistical analysis reveals it could have been built in its entirety in roughly 100 days (Bennett 1990: 496; but note R. Rawlinson’s estimate in 1850, for a period of 46 days (evidently with a larger force): Breeze & Dobson 2000: 83). Once the decision was made to complete the Wall curtain, however, the vallum was no longer required, hence its levelling before c.138. Indeed, it could even be that some of the ‘crossings’ marking the first stage of its systematic obliteration, also served as access routes to permit access to the curtain during its completion to the Narrow Gauge standard (cf. Collingwood 1931; Gibson 1932: 5).

The indication that the Narrow Wall was initiated some considerable if uncertain time after the ‘fort decision’ has an important bearing on the date assigned to the forts at Wallsend and - probably - Carrawburgh. The first of these is the only one certainly bonded with the Narrow Gauge Wall. It cannot, therefore, belong to the initial implementation of the Phase II ‘fort decision’, but to a later modification of the system. Likewise apparently with Carrawburgh. The fort’s east perimeter wall meets the Narrow Gauge Wall curtain at right angles, a relationship best explained if the two were built conterminously. Furthermore, the decision to replace Turf Wall with a masonry structure was quite possibly taken at about the same time, for Narrow Gauge curtain, with the appropriate interval structures, replaced the original Turf curtain from the Irthing to the Burtholme Beck, some 8kms distant. The initial incentive for this seems to have been a substantial deterioration of the Turf Wall itself, which in one place at least - Garthside - had already led to the collapse and subsequent rebuilding of a turret (54a(i)) and a lengthy section of Turf Wall curtain after their original construction. That apart, the Narrow Wall in this sector generally follows the earlier alignment to incorporate the existing turrets, although in the vicinity of Birdoswald it follows an entirely new alignment for about 3.2kms in order to include the fort’s north wall. West of the Burtholme Beck, however, the Turf Wall was replaced with the so-called Intermediate Gauge curtain, which seems to be of late Antonine or - more likely - Severan date (Bennett forthc.).

829

Limes XVIII

On the other hand, the evidence that certain - if not all - of the primary forts were built as free-standing structures before the Narrow Gauge Wall was begun, together with the fact that milecastles of Narrow Gauge standard have produced inscriptions with Nepos’ name, reveals that Phase III began before he left Britain in 126/7. To this we might add that while logistical analysis indicates the building of the forts would have taken a minimum of two full seasons (Bennett 1990: 493-494 - assuming they were constructed simultaneously), the work at Birdoswald at least was apparently done on a desultory basis (Bidwell 1999a: 147), perhaps over a period of years. Thus it is quite possible that work on the Narrow Gauge curtain and its associated elements did not begin until towards the very end of Nepos’ term of office, in 126 or 127. Such indeed seems confirmed by the evidence that the builders of the Narrow Gauge Wall found the existing Broad Wall foundations in too poor a condition to permit their re-use: if the Wall had indeed been initiated perhaps as early as 120, they would by then have been exposed to the elements for some six or seven years.

Part III: A revised chronology A revised chronology for the building of Hadrian’s Wall is made possible by re-assessing the archaeological and historical evidence in the light of the indications that its construction can be divided into three distinct phases. We might first stress that the evidence from Chesters reveals that a substantial interval elapsed between the digging of the Wall ditch in Phase I, and the ‘fort decision’ of Phase II. Certain of the Phase II forts were evidently under construction when Platorius Nepos was governor (Benwell and Halton Chesters certainly, and probably Birdoswald also, on the basis of the two coin-hoards from primary deposits both of which lack Hadrian's second coinage, issued in 125: Richmond 1931; Gillam 1950: 69: cf. RIC p.137), and some at least may even have been designed as free-standing structures. Finally, in Phase III, but before 127, while Nepos was still in office, the decision was made to complete the Stone Wall curtain using the Narrow Gauge standard, as is indicated by those milecastles completed in this way with inscriptions bearing his name. Then, after a short but certain interval, it was decided to add the forts at Wallsend and Carrawburgh, and finally to rebuild the Turf Wall in stone.

Regarding the completion of Phase III, it is probably relevant that those milecastles which have produced inscriptions of Nepos are all found in the central part of the North Tyne-Irthing sector, where work had barely begun on the Stone Wall before it was abruptly stopped by the fort decision. As logistical analysis suggests it would have taken quite possibly another seven seasons to complete the Stone Wall to Narrow Gauge (Bennett 1990: 496-497), we can suggest this work occupied the years from c.126/7134, which would accord with the evidence for Carrawburgh - constructed between 130-132 - apparently being bonded with Narrow Gauge curtain. This in turn would allow for work to begin on rebuilding the Turf Wall to Narrow Gauge standard as far as Burtholme Beck in the years 135-138, after which all work was abandoned on Hadrian’s Wall following the decision to re-occupy southern Caledonia and construct a frontier at a more convenient place.

The clear evidence from Chesters for a substantial interval between the decision in Phase I to build the Stone Wall, and the construction of the primary fort there in Phase II, should mean that Phase I was initiated by Q. Pompeius Falco, Nepos’ predecessor, at some date well before 122 (as was suggested by Stevens 1966: 39). Possible support for the idea is the complete absence of any inscriptions with Nepos’ name from any masonry-built Phase I structures. While it is true that we have no certain primary building inscriptions from those parts of the Stone Wall completed to Broad Gauge, the absolute lack of any at all recording Nepos seems somewhat inexplicable if he was involved in its construction. After all, this was a man with a high degree of self-pride and self satisfaction, as demonstrated in the relative abundance of his inscriptions from the Phase II and Phase III structures of the Wall, and his later fall from grace through - apparently - the sin of lèse majesté (cf. A. Birley 1997a: 281).

Bibliography Bennett J. 1984 Hadrian, and the title pater patriae. Britannia XV: 234-235. Bennett J. 1990 The setting, development and function of the Hadrianic frontier in Britain. (unpublished Ph.D thesis, University of Newcastle-upon-Tyne). Bennett J. 1998 The Roman frontier from Wallsend to Rudchester Burn reviewed. Archaeologia Aeliana5 26: 17-37. Bennett J. forthc A transect of Hadrian’s Wall and Vallum at Crosby-on-Eden. In A. Wilmott (ed.) Recent English Heritage work on Hadrian’s Wall. Bidwell P.T. (ed) 1999a Hadrian's Wall, 1989-1999. (Carlisle). Bidwell P.T. 1999b Hadrian’s Wall 1989-1997. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 55-65. Birley A.R. 1971 VI Victrix in Britain. In R.M. Butler (ed.)

If the Wall really was begun under Pompeius Falco, this would allow for Hadrian himself to make the abrupt and major decision to change the existing plan of works to include a regular series of forts (cf. SHA Had. 10.6: locum castris caperet): after all, his presence in the region seems confirmed by Tab. Vindol 2.344 (A. Birley 1997a: 134135; 1997b). Hence also, perhaps, the decision to post the leg. VI Augusta to the province: its earliest secure record in Britain is the building inscription from Halton Chesters with Nepos’ name. The suggestion that the ‘fort decision’ was perhaps made as early as 122/123 is by no means disproved by the Great Chesters inscription recording Hadrian as pater patriae: as has been demonstrated elsewhere (Bennett 1984), this title does occur on inscriptions from before Hadrian’s formal adoption of the title in that year. 830

Julian Bennett: A revised programme and chronology for the building of Hadrian’s Wall Soldier and civilian in Roman Yorkshire. (Leicester): 8196. Birley A.R. 1981 The fasti of Roman Britain. (Oxford). Birley A.R. 1997a Hadrian, the restless emperor. (London). Birley A.R. 1997b Supplying the Batavians at Vindolanda. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 273-280. Birley E.B. 1961 Research on Hadrian’s Wall. (Kendal). Bowden M.C.B & Blood K. 1991 The Roman fort at Rudchester: an analytical field survey. Archaeologia Aeliana5 19: 25-31. Breeze D.J. & Dobson B. 1987 Hadrian’s Wall. (3rd rev. edit.). Breeze D.J. & Dobson B. 2000 Hadrian’s Wall. (4th rev. edit.). Charlesworth D. 1975 The Commandant's house, Housesteads. Archaeologia Aeliana5 3: 17-42. Collingwood R.G. 1931 Ten years work on Hadrian's Wall, 1920-1930. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 30: 87-110. Coulston J.C.N. & Phillips E.J. 1988 Hadrian’s Wall west of the North Tyne, and Carlisle (Corpus Signorum Imperii Romanii, 1, 6: Oxford). Crow J.G. 1991 A review of current research on the turrets and curtain of Hadrian's Wall. Britannia XXII: 51-63. Crow J.G. 1995 Housesteads. (London). Daniels C.M. 1978 Handbook to the Roman Wall. (13th. Edit.: Newcastle-upon-Tyne). Daniels C.M. 1979 Fact and theory on Hadrian’s Wall. Britannia X: 357-364. Daniels C.M. 1987 Africa. In J.S. Wacher (ed.) The Roman World. (London): 223-265. Dobson B. 1986 The function of Hadrian’s Wall. Archaeologia Aeliana5 14: 1-30. Gibson J.P. & Simpson F.G. 1911 The milecastle on the Wall of Hadrian at the Poltross Burn. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 11: 390-461. Gibson W.W. 1932 The Vallum Crossings (Newcastle-uponTyne). Gillam J.P. 1950 Recent excavations at Birdoswald. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 50: 63-69. Hanson W.S. & Maxwell G. 1983 Rome's north-west frontier. (Edinburgh). Haverfield F. 1901 Report of the Cumberland Excavation Committee for 1900. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 1: 75-92. Haverfield F. 1902 Excavations at Chesters in September, 1900. Archaeologia Aeliana2 22: 9-21. Hill P.R. 1991 Hadrian’s Wall: some aspects of its execution. Archaeologia Aeliana5 19: 33-39. Hooley J. & Breeze D.J. 1968 The building of Hadrian’s Wall: a reconsideration. Archaeologia Aeliana4 46: 97-115. Horsley J. 1732 Britannia Romana, or the Roman Antiquities of Britain. (London). Jones M.J. 1975 Roman fort defences to AD 117, with special reference to Britain. British Archaeological Reports Brit. Ser. 21. (Oxford). Lander J. 1984 Roman stone fortifications: variation and change from the first century AD to the fourth. British Archaeological Reports Int. Ser. 206. (Oxford). MacDonald G. 1921 The building of the Antonine Wall: a fresh study of the inscriptions. Journal of Roman Studies 11:

1-24. Maxfield V. & Miket R. 1972 The excavation of Turret 33B (Coesike). Archaeologia Aeliana4 l: 145-178. Moss J. 1969 The milecastles and turrets of Hadrian's Wall and their allocation to legionary construction teams. (unpublished MA Thesis, University of Durham). Nollé J. 1997 Neue Militärdiplome. Zeitschrift für Papyrologie und Epigraphik 117: 227-276. Richmond I.A. 1931 Excavations on Hadrian's Wall in the Birdoswald-Pike Hill Sector, 1930. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 31: 122-134. Richmond I.A. 1950 Hadrian's Wall, 1939-49. Journal of Roman Studies 40: 43-56. Simpson F.G. 1931 Excavations on Hadrian’s Wall between Heddon-on-the-Wall and North Tyne in 1930. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society4 8: 305-327. Simpson F.G. 1976 Watermills and military works on Hadrian's Wall. (edited by G. Simspon; Kendal). Simpson F.G. & Richmond I.A. 1935 Randylands Milecastle. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 35: 236-244. Simpson F.G., Richmond I.A. & St Joseph J.K. 1935 The Turf Wall milecastle at High House. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society2 35: 220-228. Stevens C.E. 1966 The building of Hadrian's Wall. (Kendal). Stukeley W. 1776 Itinerarium Curiosum. (Oxford). Williams D. 1983 The vallum’s original intention: a multipurpose work of frontier support. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 83: 33-39. Wright R.P. 1965 A Hadrianic building inscription from Hardknott. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 65: 169-175.

831

Limes XVIII

Fig. 1. The ‘Original Scheme’ for Hadrian’s Wall.

Fig. 2. The ‘Second Scheme’ for Hadrian’s Wall.

832

Julian Bennett: A revised programme and chronology for the building of Hadrian’s Wall

Fig. 3. The structural evidence for the Building of Hadrian’s Wall.

833

Limes XVIII

Fig. 4. The state of work on the Wall curtain at the time of the ‘fort decision’.

5. The Wall at the time of the ‘fort decision’. Key: as for Figs. 1&2.

834

Recent research on Roman camps in Wales J.L. Davies & R.H. Jones The paper summarises the results of one component of a one-year research project funded by the Board of Celtic Studies of the University of Wales, namely Roman camps and communications in Wales. A review of aerial photographic and ground survey data has shed new light on both marching camps and practice camps in respect of numbers, morphology and broader relationships. A re-appraisal of the dimensions of some of the marching camps raises fresh issues concerning campaigning routes together with the size of the forces involved. The paper will also discuss the unusual character of practice camps in Wales against the wider background of Roman military training works.

The study of Roman camps in Wales has been a neglected subject, both in respect of marching camps and practice works. The last summary treatment of the former being by St Joseph in 1973 (233-244), and of the latter by R.W. Davies in 1968. In 1999 the Board of Celtic Studies of the University of Wales initiated a one-year research programme to investigate Roman camps and communications in Wales, and appointed Ms. Rebecca Jones as Research Assistant under my direction. The project was completed at the end of March 2000. Not only did it result in the discovery of new sites in the generic class of practice camps, but also led to a reassessment of the size and probable relationships of some of the known marching camps, inviting some debate as to their role in the context of military operations and campaigning routes.

within a later Neolithic enclosure. This new camp lies immediately to the east of the large camp (17.6ha/43.5a) of Hindwell Farm I. Part of the west and south sides of this new camp are now known, together with the south-western corner. It is of uncertain size, though topography suggests that it enclosed a minimum area of 1.6ha (4a) and a maximum of 8ha (19.8a). The camp’s proximity to the early Neronian auxiliary fort site at Hindwell Farm suggests that it is either earlier or possibly contemporary with it. In addition, to the south of the fort the three camps at Walton, hitherto interpreted as practice-camps (St Joseph 1969: 121), we would now re-interpret on the basis of aerial photographic evidence as small marching-camps possibly constructed in the same campaigning season by three separate units who, arriving consecutively, affected a concentration.

Regarding methodology, all existing paper and photographic records, particularly aerial photographs, were studied and subject to re-evaluation as and when necessary. Hence, a number of possible camp sites were rejected. We must emphasize the fact that no primary aerial reconnaissance was undertaken as part of the project.

The vicinity of the large auxiliary fort at Forden Gaer in the upper Severn valley has consistently revealed a complex of cropmarks - some clearly non-Roman - for over 50 years (Crew 1980). However, some cropmarks are markedly linear in character and possibly of Roman military origin. Indeed it was as long ago as 1953 when Prof. St. Joseph stated that he had noted a camp or camps close to the fort, though he never provided further details (1953: 85). It was not therefore, altogether surprising that analysis of aerial photographs taken over this long timespan has shown that at least one of these linear cropmarks represents a camp of some 1.4ha (3.5a). This lies about 300m north-east of the fort, though only two sides can be identified with confidence, whilst another side is overlain by a linear cropmark with a rounded angle - putatively a further camp.

Special attention was given to searching areas where camps might be expected to be found, specifically the vicinity of forts, fortresses and other military sites, together with the course of known, or suspected Roman roads which were considered to be of significance in the context of the routes utilised by forces engaged in campaigning both in the pre-Flavian period and in the subsequent conquest phase. All known sites, whether represented by earthworks or cropmarks were visited and were either planned or reassessed on the basis of aerial photographic evidence and ground observation, which in some instances has led to a recension of their size, aspect and possible relationships.

Another site in this complex, whose precise character is uncertain, indeed it seems never to have appeared as a crop-mark, lies about a kilometre north-east of the fort. It appears on a RAF photograph taken in August 1946 as a rectangular enclosure with two sides defined by trees planted along a bank. It seems to enclose an area of about 3.2-3.6ha (8-9a), a size which is very similar to that of the visible fort enclosure at Forden. We have been unable to visit this site to date and both its precise character, let alone its date, must remain tentative. However, either a camp, or very possibly a fort pre-dating Forden is probable.

Marching camps (Fig. 1) The project did not produce an extensive crop of new sites. It would have been remarkable had it done so. Photographic coverage frequently proved to be less than satisfactory, either because it was not available for certain critical areas, was seasonally inappropriate for the revelation of crop-mark sites, or was sometimes too smallscale. Occasionally all three factors came into play. Recent discoveries include a new camp at Hindwell Farm (Camp II), found by accident through geophysical survey

Other possible sites include that already mentioned near 835

Limes XVIII

Forden; a possible small camp at Dolwen in the upper reaches of the Severn, 2kms west of the road from Castell Collen to Caersws; and, finally, two possible camps at Brompton overlapping those already identified close to the Pentrehyling fort, just across the border in Shropshire.

westerly direction. Finally, in north-west Wales four marching-camps are known in this dramatic upland landscape. Of these only one lies in the heart of the Snowdon massif itself. This encompasses the well-known climbers' mecca of the Pen y Gwryd Hotel. At 4ha (9.9a) it is small but occupies a commanding blocking position straddling three passes; the camp itself facing west towards the Llanberis pass. The camp has hitherto been considered in isolation, or at least as the last vestige of operations in this mountainous zone. However re-assessment of the camp at Derwin Bach, about half a kilometre from the Flavian fort at Pen Llystyn, which Prof. St. Joseph stated could have occupied an area of up to 6ha (15a) (1973: 241), suggests that this is far too generous and that a more realistic size would be of a minimum of 2.5ha (6.3a) to a maximum of about 3.7ha (9.1a) (the latter can now be confirmed on the basis of a reassessment of the aerial photographic evidence pertaining to the site). It is thus tempting to link Pen y Gwryd and Derwin Bach on grounds of size. Such modest-sized camps would be perfectly in keeping with the numbers of men involved in operating in such difficult country. Tempting though it may be to assign these sites to the autumnal campaign of Agricola in 77, prior to the seizure of Anglesey and the establishment of a chain of forts in and on the margins of Snowdonia, such an association must be entirely speculative.

Now we turn to a re-assessment of the size and probable relationship of marching camps in the context of military operations and campaigning routes. Firstly, in mid-Wales three camps lie in relatively close proximity to one another in the upper reaches of the Wye: Cwm Nant (c.7.3ha/18a), Esgairperfedd (6.3ha/15.6a) and Trefal (c.7.6 ha/18.7a, or possibly as much as 8.5ha/21a). The first two are characterised by the possession of clavicular gates; the gate type at Trefal being unknown. Prof. St. Joseph postulated, prior to the discovery of Trefal, that Cwm Nant and Esgairperfedd represented the passage of a battle-group of approximately legionary size moving west, climbing out of the Wye valley and heading for the watershed that would eventually provide ready access to the lowlands along Cardigan Bay (1969: 123-6). Whilst such a route has much to commend it, we would interpret the evidence somewhat differently. Esgairperfedd faces east and probably represents the movement of troops in the opposite direction to that suggested by St. Joseph, whilst Cwm Nant faces north and is now known to lie close to a newly-discovered Roman road running north from the fort at Castell Collen to Caersws, and an early campaigning route linking the upper Wye and upper Severn valleys seems highly probable. The proximity of and the similar size of the camps at Cwm Nant and Trefal could either be interpreted as a force moving north and returning south (or vice versa) or two different operations.

The two camps which partly survive south-east of the fort of Tomen y Mur enclose minima of 1.5ha (3.6a) and 2.2ha (5.3a), but could enclose maxima of 9.2ha (22.8a) and 10.2ha (25a) on the assumption that the Roman road heading south-east of the fort cuts through the south-east gates of the camps (Davies 1999: 67-8). We consider that the larger size to be more probable, commensurate with the much more open landscape in which they are set and the ease of employing much larger bodies of men on campaign in this area.

The evidence suggests that the routes leading both west and north from the upper Wye were highly significant in the context of military operations in mid-Wales. To add a footnote here. It is distressing to recall that the last visible elements of the camp at Cwm Nant, the southeast with a well preserved rampart and clavicular gate, was wantonly destroyed by ploughing within the last five years despite its status as a Scheduled Ancient Monument.

Practice camps A number of additions may be made to some of the known clusters: firstly, at Gelligaer Common in south Wales, north of the fort at Gelligaer (Fig. 3), the possible remains of a small camp can be observed on RAF aerial photographs taken in 1947 lying on open moorland about 70m east-south-east of another camp. The site has been largely obscured by vegetation and drainage and encloses some 600sq m with only a single entrance gap visible (Fig. 4). Secondly, at Pant-teg Uchaf in south-west Wales, another camp measuring approximately 40m sq has been discovered close to the Roman road south of the fort of Llanio. Thirdly, there are additions to the largest group of practice camps known to date on Llandrindod Common, south of the fort of Castell Collen in mid-Wales. Twentyone camps are now identified, two of these being recent discoveries (Fig. 5). Camp XX measures about 27m sq, with three visible entrances, one protected by a titulus, and Camp XXI is about 24m sq. The gaps between the known

Secondly, in the upper Usk valley the two splendidly preserved camps of Y Pigwn, on the summit of Trecastle Mountain, are certainly the best known in Wales; Camp 1 encloses 15ha (37a), Camp 2, clearly later, 10ha (25a). Some 5.5kms distant to the south the large camp at Arosfa Gareg, 17.8ha (44a) in extent (Fig. 2), has been divorced from the above on grounds of size. However, one critical feature of this site has been largely ignored, a substantial ravine, giving rise to two streams which, together with other irregularities within the defences, would have reduced the usable area by at least 2ha (5a) to some 16ha (39a). It is thus possible that Y Pigwn I and Arosfa Gareg represent the passage of broadly equivalent campaigning forces along the high ground south of the upper Usk, but probably on different occasions, since both camps face in a 836

J.L. Davies & R.H. Jones: Recent research on Roman camps in Wales sites on Llandrindod Common suggests that many others may have been destroyed by agriculture and building operations and that the total number may have possibly reached 30 or more. Finally, a new camp has been identified at Llwyn-crwn near the fort of Tomen-y-Mur in north-west Wales. It lies nearly 2kms south-east of the fort, close to the Roman road, encloses an area of some 200sq m and has two opposed entrances. The aerial photographic evidence suggests that there may be two overlapping camps here, but it appeared unitary from ground level survey. This discovery brings the total number known in the vicinity of the fort of Tomen-y-Mur to at least 14, making it the second largest group in Wales.

The majority are less than 0.2ha (0.5a) in size, with one or two probable examples of just over 0.4ha (1a). Wales has by far the largest surviving numbers of this class of monument, which has received scant attention since the late Roy Davies' discussion of the type and its importance in the context of the study of the Roman army in peacetime or preparing for war (1968). It might be considered that the phenomena of practice camps is prevalent in Wales because the conditions for the preservation of such slight earthworks was more favourable insofar as the majority have been noted on rough pasture and moorland. However, this cannot be the sole explanation for their absence elsewhere given similar conditions in parts of northern England and Scotland. Indeed the best preserved temporary camp earthworks in Britain are in Northumberland, and a group of small camps, some of which may be practice camps, survives in the vicinity of Hadrian’s Wall.

The definition of practice camps is not always straightforward, although the concentration of these small camps in Wales means that the interpretations are usually clearer. Such camps are usually recorded in clusters, such as those noted above, anywhere between 100m and 4.5kms away from what is generally regarded as their ‘parent fort’. They are normally too small to hold substantial bodies of men, the majority encompassing an area of less than 0.2ha (0.5a) overall, the largest as much as 0.4ha (1a). They are often square or almost square, and show a concentration upon specific elements of camp building, namely the turning of rounded corners and the construction of gates. Some of the entrances are protected by tituli, others by claviculae. Two Welsh examples are protected by both, although it is not possible to be certain about the entrance types provided at over 30% of the practice camps in Wales, largely due to their state of preservation.

Is it a matter of chronology then, the Welsh auxiliary forts being largely early Flavian foundations with the great majority being abandoned by the mid-120s; the camps belonging to the Flavian-early Hadrianic period? The chronology of the Welsh practice camps is itself uncertain, although five of the 9 clusters contain at least one example with claviculae, a device for protecting a camp gate which it is thought may have been discontinued after the reign of Hadrian. For example, camps with claviculae are known in the groups on Llandrindod Common and Tomen y Mur, the former being associated with a fort occupied up to Severan times and later (that at Castell Collen already mentioned), the latter certainly abandoned under Hadrian. Thus the timeframe suggested by the gate types broadly conforms to the fort occupation dates. The impression gained is that the camps are most likely to be early rather than late; the frequency of camps with clavicular gates favouring a date within the last quarter of the first and the first decade or so of the C2nd. This does not preclude the possibility of training taking place over a longer time-span. Evidence of overlapping camps have been recorded at Allt yr Hafod Fawr near Y Pigwn (Davies 1999: 69) and possible reuse recorded during the excavations at Dolddinas (Jones & Knowles 1960: 399) and also noted as a possibility during ground survey on Gelligaer Common.

To conclude, we shall briefly consider the significance of the Welsh practice camps, of which 56 are now recorded. No less than 9 auxiliary forts in Wales have practice camps associated with them. Paul Holder suggested some years ago that only legionaries practised the art of entrenchment, and that legionary cohorts and new recruits would have occasionally been deliberately marched to an otherwise abandoned fort site to practice the craft in their vicinity (1982: 90). The evidence would belie this view. We may note, for example, that the fort at Castell Collen, north of the camp cluster on Llandrindod Common, though founded in the Flavian period remained in occupation through the Severan period and beyond. Oddly no practice-camps are known in the vicinity of the legionary fortress at Caerleon though possibly as many as sixteen are now believed to exist to the east of Chester, flanking the roads to Manchester and Wilderspool (Philpott 1998). These are all larger than their Welsh counterparts. In addition, two are known north of the fortress at York (Welfare & Swan 1995: 135-6), whilst on the Continent a celebrated cluster lie south of Vetera in the lower Rhineland (Horn 1987: 333). They are exceedingly scarce elsewhere in Britain though Welfare and Swan note some examples in the Hadrian's Wall zone (1995: 24).

Moreover, the frequency with which practice camps are associated with Welsh auxiliary forts suggests that special circumstances may have prevailed to involve auxiliary troops in such large-scale practice in field-entrenchment. An earlier suggestion that the training may have taken place during Hadrian’s visit to Britain (Frere & St Joseph 1983: 137) seems unlikely. However, another theory that they may have been constructed under a training initiative by Frontinus, given his manifest interest in tactics and training methods, is intriguing (Davies 1968: 104). Presumably if such an initiative had been attributed to Agricola then Tacitus would have obliged us with a reference. Some kind of special circumstances is, therefore, a preferred option to explain this involvement of

What is curious about the Welsh practice camps is their sheer frequency and, for the most part, diminutive size. 837

Limes XVIII

auxiliary soldiers in such large-scale practice in fieldentrenchment, rather than simply suggesting that a high proportion of the tribunes and prefects of the FlavianHadrianic era were disciplinarians of the Corbulo mould, or that parts of the Welsh uplands would have formed a sort of military reserve or training-ground, analogous to modern-day parts of the Brecon Beacons or Salisbury Plain.

72. Journal of Roman Studies 63: 214-246. Welfare H. & Swan V. 1995 Roman camps in England: the field archaeology. (RCHME, London).

One factor which springs to mind is the necessity of instructing newly-raised units by training cadres, or a large influx of recruits into existing units, one or both occurring within a relatively short time-span. Such a situation appears to have existed in the 70s of the C1st with the suppression of the Batavian revolt, the raising of a large number of fresh regiments from the defeated tribes and surrounding communities and their despatch to provinces where reinforcements were urgently required and where they could prepare for future offensives; for example into Scotland. ‘Diploma’ evidence indicates that no fewer than 18 cohorts and one ala of ethnically lower German origin were sent to Britain in the post-Civilis era, together with another three cohorts of non-German origin. Of these no fewer than 12 units are known to have been serving in or close to Wales in the period 98-105. It is not impossible, therefore, that a high proportion of the practice camps in the vicinity of Welsh forts were the product of specific auxiliary training in the 70s, probably in the immediate aftermath of the Welsh conquest complete by 77 - not necessarily at the specific behest of Frontinus or Agricola but around this time and later. Such intensive training could have been in anticipation of a much greater auxiliary involvement in field operations in northern England and Scotland. Bibliography Crew P. 1980 Forden Gaer, Montgomery. Bulletin of the Board of Celtic Studies 28: 730-741. Davies J.L. 1999 The Roman frontier in Wales: recent research. In N. Gudea (ed.) Roman Frontier Studies 1997: Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalãu): 67-76. Davies R.W. 1968 Roman Wales and Roman military practice camps. Archaeologia Cambrensis 117: 103-120. Frere S.S. & St Joseph J.K. 1983 Roman Britain from the air. (Cambridge). Holder P. A. 1982 The Roman army in Britain. (London). Horn H.G. 1987 Die Römer in Nordrhein-Westfalen. (Stuttgart). Jones G.D.B. & Knowles W.A.C. 1960 Roman Merionethshire: The Dolddinas Camps. Bulletin of the Board of Celtic Studies 18: 397-402. Philpott R.A. 1999 New evidence from aerial reconnaissance for Roman military sites in Cheshire. Britannia XXIX: 341-352. St Joseph J.K. 1953 Air reconnaissance of southern Britain. Journal of Roman Studies 43: 81-97. St Joseph J.K. 1969 Air reconnaissance in Britain, 1965-68. Journal of Roman Studies 59: 104-128. St Joseph J.K. 1973 Air reconnaissance in Roman Britain, 1969-

838

J.L. Davies & R.H. Jones: Recent research on Roman camps in Wales

Fig. 1. Marching camps in Wales and the Marches.

Fig. 2. Aerial view of the marching camp at Arosfa Gareg from the east. (Crown Copyright: Royal Commission on the Ancient and Historical Monuments of Wales Ref. No. 8818994).

839

Limes XVIII

Fig. 3. Military sites of the Flavian era in Wales and the Marches.

Fig. 4. Aerial view of Gelligaer Common I and II practice camps from the north-west. (AWZ 68. Cambridge University Collection of Air Photographs: copyright reserved).

840

J.L. Davies & R.H. Jones: Recent research on Roman camps in Wales

Fig. 5. Practice camps on Llandrindod Common showing location of new camps XX and XXI.

841

Limes XVIII

842

Supplying the frontier zones: the rôle of the East Anglian Fens Alison Taylor This paper is dedicated to the memory of Tim Potter, who died suddenly at the end of 1999. It draws heavily on the work, practical and academic, that he carried out from the 1960s and which has informed much of our growing understanding of the area. It looks in particular at the evidence for the ways in which the army, particularly at times of stress, developed and protected supply lines to the agricultural regions of eastern England. Hadrian not only built and manned his Wall, he went to great lengths to ensure the men stationed there would have food and other provisions. These measures included reclamation works and land settlements on large areas of under-populated fenland in order to produce the leather, wool and salted meat the army needed, and also construction of sufficient stretches of canal to provide continuous water transport from the rich grain-producing areas of southern Cambridgeshire to northern Britain. Work by Tim Potter and his British Museum team elucidated much of this development in the 1980s, and in the past three years excavations by the Cambridgeshire Archaeological Field Unit have uncovered more details about the history and use of the Cambridgeshire Car Dyke. This included excavations through the dyke (where beetles in the silt can be related to military occupation), of adjacent kilns making pottery (which has now been found at Wallsend) to store and transport grain from this damp area, of industrial-scale grain processing and of storage depots sited where the Car Dyke meets the River Cam. For the end of the Roman period, post-excavation work on the Roman town of Cambridge (published March 2000) has demonstrated the relevance of late C4th town defences in this area to the protection of eastern England because of its importance in supplying grain to the army in Gaul and the Rhineland, archaeological evidence being related to texts by Ammianus Marcellinus, Eunapius and the emperor Julian.

and areas of silt fen in the north of the region were drying out, becoming particularly suitable for salterns and summer grazing (Hall & Coles 1994). It was therefore an area of considerable agricultural potential and on the right communication networks.

Introduction This paper has to move far from the sunshine and handsome monuments of the eastern frontier into the dark mud of the East Anglian Fens. The move must also have been a shock to the Roman army but they too soon understood that the results could be fruitful for, as we are reminded many times in this volume, soldiers must be fed. East Anglia, from Iron age times to the present day, has been recognised as a bread-basket area (one reason for the Claudian conquest was perhaps a misunderstanding that all Britain resembled this rich grain exporting region, which already had many links and friendly relationships with Gaul and the Roman world) and a tactician like Hadrian would have quickly realised its importance for supplying the Wall.

Part of the area had been extremely troublesome during the C1st AD, when the Iceni under Boudicca revolted so fearsomely. The Iron age fortifications at Stonea Camp appear to have been one of the significant centres of Icenian power and, it is argued (Jackson & Potter 1996) the site of the final defeat of Prasutagus in AD47. Forts such as Grandford were rapidly built once the Boudiccan revolt was over, but by the beginning of the C2nd peace was guaranteed and hardly any troops remained. Romanisation in the form of towns, roads, potteries and other manufactures and villa estates was advancing fast in the region round about, at centres such as Durobrivae, Godmanchester and Cambridge. Some roads, notably the Fen Causeway, are attributed to the army on account of their construction in layers of brushwood, timbers, clay and gravel metalling, and the potteries too originated as supplies for forts such as Longthorpe. Apart from such military bonuses it had become an area where investment in civilian infrastructure could be safely made, from where produce could be marketed and where the land’s emptiness made new settlement under imperial control reasonably trouble free. The Roman authorities would have realised too, from experiences in reclaimed marshy areas such as the Po valley, that agricultural returns could be high.

The Fens cover an area of nearly a million acres in the counties of Cambridgeshire, Lincolnshire and Norfolk, all extremely low-lying (much of it at or below sea level). It had been well populated in early prehistoric times but rising sea levels from the late Bronze age had led to inundations which made most of it unsuitable for agriculture and stable settlement during the Iron age. The land was still valuable in other ways, for it was rich in fish, eels and wild fowl, reeds and turves, and could be used for summer grazing as well as hunting trips, but its population was low (Coles & Hall 1998) and the fertility tied up in the black peat that had grown during centuries of waterlogging was untapped, a situation that would be replicated in the C17th when the Dutch engineer Vermuyden again drained Fens that nature reclaimed after the Romans left. This region had access by navigable water to the east coast of Britain (The Wash), and so to the North Sea and northwestern Europe or, following the coast to the north or south, to London etc. or northern Britain. It was also in fairly easy reach of both Colchester and London by road. The sea was already receding somewhat in the C1st AD

Within East Anglia, the wild Fens appear to have been tamed in Tim Potter’s words (Jackson & Potter 1996), “...as the result of a single initiative, conventionally regarded as Imperially-inspired”, within the reign of Hadrian. “The hand of government”, Potter states elsewhere (1989) “implicit in the drove-roads and canals that show so clearly from the air, is manifest”. Part of this 843

Limes XVIII

initiative was met by construction of an ambitiously planned new town at Stonea, adjacent to the Iron age fort (Jackson & Potter 1996). Here excavations by the British Museum have given evidence for a high stone tower with attached hall, fronted by a large piazza, which would have dominated the flat landscape for many miles in all directions. The town that was built here left evidence for bureaucracy (wooden writing tablets and styli), a military presence, and many amenities connected with a sophisticated life-style, again pointing to an official foundation in this unpromising landscape. An important part of the economy that Potter identified was large-scale export of joints of lamb, which would have been salted with the product of innumerable salterns located by the Fenland Survey and other fieldwork. As far as we can tell within the limits of archaeological dating it was also in this Hadrianic period that a wholly new road network that has been shown on several sites (Ellis et al 1998; T. Malim pers. comm.) to ignore any earlier farmsteads, was laid out. Many new settlements also date to this period (Salway 1970; Potter 1981; VCH VII; Cambridgeshire Sites & Monuments record). (Fig. 1).

The next crucial part of the plan for the Fens was the Car Dyke, and it is this monument which is the clue to the scale on which Hadrian was working. It acted as a catch water system to help maintain the lowered water table, though its flow must have been sluggish, to judge from environmental evidence and from levels taken along the Lincolnshire sector (Simmons 1979), and its function as a canal has often been questioned (ibid), but it is inconceivable that such stretches of water were not used in this way. Its life as a canal may not have been long for, although it was re-cut at the Waterbeach end in the C3rd, causeways were also built across it, rubbish was tipped in and silting was not regularly scoured out. When freshlyconstructed in the early C2nd however it was ideal for taking the produce of this newly developed landscape safely and cheaply to where the emperor needed it, feeding the legions on Hadrian’s Wall. This dyke (Figs. 4 & 5) was a flat-bottomed canal, 24m wide and 2m deep, originally with banks set back from either side so that barges could be pulled from the edge. Excavations by Clarke and more recently by the Archaeological Field Unit of Cambridgeshire County Council (Macaulay & Reynolds 1994) have shown that it was constructed in the early C2nd and was out of use in the C3rd and C4th, when causeways were built across it. At its southern end, where it joins the Cam at Waterbeach/Horningsea, it widens into a barge-turning basin, and it is this area, investigated in 1997/8, that produced more evidence for its uses. It has long been known that the kilns at Horningsea, excavated in the early C20th (McKenny Hughes 1902; Walker 1912) made huge crude pots that could only have found a market because of their contents, so it is assumed they were intended for storing and moving grain in a land where any pit would fill with water. Recent discovery of these Horningsea wares at Wallsend (P. Bidwell pers. comm.) help confirm this theory and recent excavations clarified the scale of the enterprise, for industrial-scale processing of the grain was apparent from the great number of pudding stone querns found near the dyke’s banks, vastly more than occur in domestic contexts, and there was also uncovered the raft foundations of a 20m sq building (Fig. 6) with a mass of pottery (Macaulay 1998). This was interpreted as a warehouse for produce waiting shipment, a close parallel for one found by the Thames at Southwark (Milne 1995: 64).

Partly as a result of these excavations Potter was able to take forward theories proposed by Peter Salway (1970) that the Fens were settled and farmed as an Imperial estate, with peasants placed in small settlements paying rent to the central authority, though there is still debate (eg. Potter 1981) over just how much planning and even centuriation we can see in the Fens. Some settlements are very regular (for example at Grandford, Flaggrass and Christchurch), but many others are quite as haphazard as any Iron age/Romano-British farmsteads. Perhaps one of the most relevant sites in this debate is Bullocks Haste, Cottenham which, somewhat miraculously in the prairie landscape of the modern Fens, is still partly preserved under pasture (Fig. 2). Here, excavations by Graham Clarke (1949) showed that Car Dyke cut through the random ditches etc of a C1st settlement. The subsequent village, surveyed by RCHME (1996), was far more regular (Fig. 3). Animal bones from these sites are predominantly sheep, followed by cattle, with a considerable number of horse. Another element of Hadrianic planning according to the arguments of Ernest Black (1995) is construction of mansiones at nodal points every 15 miles or so along the road network so that travellers on official business would have somewhere warm, safe, comfortable and clean to stay the night, change or at least stable their horses, and enjoy an evening with friendly companions. The location of these mansiones seems to be an important part of the development of small towns, of which Cambridge itself is a good example (Taylor 2000). Others can be identified by finds of numerous horse bones, corrals, and high quality buildings found during recent work at Wimpole (Horton et al 1994), and Sawtry (Ellis et al 1998), and possibly at Stretham. These played an important role in opening up the previously inhospitable regions of the Fens, with Cambridge in particular providing marketing outlets.

Waterlogged remains found in the canal included a wooden harpoon along with a leather sandal, and also peas and beans, unlikely to have been grown nearby but useful provisions to be shipped north. Yet more pottery kilns were found on this Waterbeach side of the Cam. Further indications of the scale of grain storage and transhipment and for the likelihood of military involvement came from environmental evidence, for amongst the insect remains three samples included oryzaephilus surinamensis, a beetle associated with large scale storage of de-husked grain, which was introduced by the Romans and occurs in towns and forts but not in rural contexts. This pest was present 844

Alison Taylor: Supplying the frontier zones: the rôle of the East Anglian Fens over a long period. Other unsavoury evidence included whipworm in Car Dyke resulting from sewage there, and samples of Oulimnius sp caused by mouldy hay and old thatch (Robinson 1998).

this part of the world. The results of several decades of excavation at Cambridge were published in 2000 (Alexander & Pullinger 2000), and this work made it possible to piece together the anomaly of a settlement with few urban characteristics and little wealth which, after three centuries of use, was defended with a rampart, ditch and stone wall 2-3m wide surrounding about 9ha. That this was not done for the benefit of the occupants was shown by the ways in which the wealthiest areas were excluded, and the defences cut through habitations in an arbitrary way that no town council could have achieved. As at Great Chesterford (Going 1996), a few miles south down the Cam, it seems most likely that we are seeing imperial power at work again, this time protecting key points in an area that was becoming vulnerable to piratical attacks from the North Sea coast and down its navigable rivers, the routes which in the next century were to let in trouble which lead to rapid and total collapse of Roman power in this particular part of Britain. If our hypotheses of the social and economic situation are correct we are also looking at a potentially hostile local population squeezed and taxed ever more harshly to support an army in a foreign country and a civilisation that gave them equally little, a scenario in which defended tax-points were required more than the traditional amenities of urban life.

The produce of the Fens according to evidence excavated at Stonea and elsewhere was largely lamb, cattle and salt, which means that leather for tents, boots, buckets, armour etc was being supplied, along with joints of salted meat. From the Cam valley south of Cambridge came arable crops of grain, peas and beans which seem to have been processed at Waterbeach before shipment north, for which there is evidence from the environmental data as well as the querns and warehouse. Development of the Fens and construction of a canal that linked natural waterways to give continuous water transport from Cambridge to the Humber therefore can be seen as going together to feed Hadrian’s army on its northern outpost, a vital part of his strategy until local production became reliable, after which Car Dyke does appear to have been neglected. The next time the empire urgently required food in this part of the world was in the late C3rd and C4th. Then the army on the Rhine temporarily lost control of Gaul and was under pressure on all sides, and several documentary sources point to the steps taken to ship grain from Britain, East Anglia being the obvious source on this route. Perhaps the clearest of the sources was the emperor Julian himself, who wrote in a Letter to the Athenians, AD379, “...all of the barbarians were driven out of the Gallic province, most of the cities were recovered, and large numbers of ships were brought over from Britain” (quoted in Ireland 1986). Libanius and Eunapius, also in the late C4th, and the accounts of Zosimus in the C5th-C6th similarly indicate that Julian built many ships to move grain from Britain to the Rhineland, and that this was not a new resource but had just needed reinstating.

At Cambridge (Fig. 7) the centre being defended commanded one of the few points in a marshy landscape where goods could be loaded and unloaded as well as being on a bridging point and cross roads. It could have served as a collection point for the corn tax and its despatch, by this time up the Cam and Ouse to The Wash and hence to the Rhineland rather than up-country to Hadrian’s Wall. Such a situation may well have been replicated at other centres where it has not yet been possible to look at the totality of evidence. If it is a correct reading of this period it perhaps helps explain how the Anglo-Saxons so easily took control of this area in the C5th, for their rule may not have seemed rapacious to peasants accustomed to losing much of their produce to feed a distant army that in the event did nothing to protect the region that fed it for so long.

When we look more generally at sites of the late Roman period in East Anglia, Chris Going (1996) argues that we can see widespread effects of this policy on the landscape. Small settlements were deserted at this time, to be replaced by much larger fields for both arable and pastoral agriculture. Agricultural wealth generated by the large estates in this area was not reflected in living standards apparent in the archaeological record. This especially applies to towns which, apart from the exceptional case of Colchester, are remarkable for their lowly nature. Up to the C4th the only walled towns in Norfolk were the civitas capital at Caistor by Norwich and a small industrial settlement at Brampton, there were none in Suffolk or Cambridgeshire except Godmanchester and Durobrivae (both west of the Fens), and Essex too was weak. Unlike parts of Britain where rich villas are known from this time impoverishment in the countryside arguably increased in the C4th (ibid), though it was the time of great treasure hoards (eg. Water Newton, Mildenhall and Hoxne), and also the sudden construction of stone defences around minor towns such as Great Chesterford and Cambridge, juxtapositions that point to exceptional circumstances in

Bibliography Alexander J. & Pullinger J. 2000 Roman Cambridge: excavations on Castle Hill, 1956-1988. Proceedings of the Cambridge Antiquarian Society 88. Black E.W. 1995 Cursus Publicus: the infrastructure of government in Roman Britain. British Archaeological Reports Brit. Ser. 241. (Oxford). Clarke J.G.D 1949 Report on excavations on the Cambridgeshire Car Dyke’. Antiquaries Journal 21: 145-163. Coles J. & Hall D. 1998 Changing landscapes: the ancient Fenland. (Cambridge). Ellis P. et al 1998 Excavations alongside Ermine Street, Cambridgeshire, 1996. British Archaeological Reports Brit. Ser. 276. (Oxford). Going C.J. 1996 The Roman Countryside. In O. Bedwin (ed.)

845

Limes XVIII

The archaeology of the Essex countryside: proceedings of the Whittle conference. (Essex County Council). Hall D. & Coles J. 1994 Fenland Survey: an essay in landscape and persistence. (English Heritage, London). Horton W. et al 1994 Excavation of a Roman site near Wimpole, Cambridgeshire 1989. Proceedings of the Cambridge Antiquarian Society 83: 31-74. Ireland S. 1986 Roman Britain: a sourcebook. (London). Jackson R.P.J & Potter T.W. 1996 Excavations at Stonea, Cambridgeshire 1980-85. (London). Macaulay S. 1998 Car Dyke Roman canal, Waterbeach, Cambridgeshire. (Unpublished assessment report). Macaulay S. & Reynolds T. 1994 Car Dyke: a Roman canal at Waterbeach. Cambridgeshire County Council Report 98. McKenny Hughes T. 1902 On a Roman potters field at Horningsea. Proceedings of the Cambridge Antiquarian Society 10: 174-194. Milne G. 1995 Roman London. (London). Potter T.W. 1981 The Roman occupation of the central Fenland. Britannia XII: 79-133. Potter T.W. 1989 Recent work on the Roman Fens of eastern England and the question of imperial estates. Journal of Roman Archaeology 2: 267-274. RCHME 1996 Royal Commission on the Historical Monuments of England. Bullocks Haste, Cottenham, Cambridgeshire Robinson M. 1998 Molluscs and insects from Car Dyke, Waterbeach, Cambridgeshire. (Internal report for Cambridgeshire County Council). Salway P. 1970 In C.W. Phillips (ed.) The Fenland in Roman Times. Royal Geographical Society Memoir 5. (London). Simmons B.B. 1979 The Lincolnshire Car Dyke; navigation or drainage ? Britannia X: 183-196. Taylor A. 2000 Discussion and conclusions. In J. Alexander & J. Pullinger Roman Cambridge: excavations on Castle Hill, 1956-1988. Proceedings of the Cambridge Antiquarian Society 88: 82-83. VCH VII 1978 Victoria County History of the County of Cambridgeshire and the Isle of Ely. (Oxford). Walker F.G. 1912 Roman pottery kilns at Horningsea, Cambridgeshire. Proceedings of the Cambridge Antiquarian Society 17: 14-69.

846

Alison Taylor: Supplying the frontier zones: the rôle of the East Anglian Fens

Fig. 1. Roman Cambridge.

Fig. 2. Present Day Car Dyke.

847

Limes XVIII

Fig. 3. Bullocks Haste, Cottenham, from C.W. Phillips 1970, The Fenland in Roman Times.

Fig. 4. Car Dyke, Waterbeach, in the flat landscape of the Cambridgeshire Fens. Excavations at the southern end (now infilled) show where it widens out near to its original junction with the Cam.

848

Alison Taylor: Supplying the frontier zones: the rôle of the East Anglian Fens

Fig. 5. Sections through the Car Dyke at Cottenham, excavated by Graham Clark.

Fig. 6. Plan of the warehouse excavated near the Cambridge at Waterbeach. © Cambridgeshire County Council Archaeological Field Unit.

849

Limes XVIII

Fig. 7. Roman Cambridgeshire. © Cambridge Antiquarian Society.

850

Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999 Tony Wilmott The first Limes Congress in 1949 occurred only 20 years after the scheme of Wall periods had been formulated, following the well known excavations of Simpson, Richmond and Birley at Birdoswald in 1929 (Richmond & Birley 1930). This concept combined the evidence of stratigraphy and finds with the text of two inscriptions found reused in a C4th floor during these excavations to advance a chronological framework which was thought to be applicable not only to Birdoswald but to Hadrian’s Wall at large (Wilmott 1997a: 8-14). Also in 1949 the walls and gates of the fort, which had been taken into state guardianship 10 years earlier, were the subject of a programme of consolidation undertaken by the then Ministry of Works, with some archaeological cover by John Gillam, who excavated on several of the interval towers in 1950 (Gillam 1950), and consolidated knowledge on the principal east gate which had been first excavated by Potter in 1850 (Potter 1855). In 1959-60, ploughing two fields west of the fort revealed a cremation cemetery, which was partially recorded by Robert Hogg (Wilmott 1993).

(Whitworth 1992). The section was recorded in 1979 and 1989, when it was also sampled for environmental evidence (Wiltshire 1992; 1997). In 1999 the section was extended to include the turf wall ditch, and counterscarp and the vallum, and further environmental sampling was carried out. I would like now to summarise the present state of knowledge on a selection of specific problems, with particular emphasis on the results of work from 1997-99. The existence of a timber fort phase on the site was suggested in 1931, when a broad stone rampart base was found to the south-east of the fort (Simpson & Richmond 1932). In 1929, an early level (Level 0) comprising a drain and wood chips was identified, and a similar level was found in 1989 (Wilmott 1997a: 47-8). The first real evidence came in 1998, however, when two pits were found beneath the earliest stone fort strata in the praetentura, containing Roman shoe and tent leather, as well as an enamelled legionary belt plate, which has two exact parallels at Carleon. This suggests that there was indeed an early fort, but contra Wilmott (1997a: 401), this probably projected to the north of the Wall like its later Hadrianic stone successor.

Research excavation on a large scale was resumed in 1987 and ran until 1992, revealing a large area in front of Birdoswald farm, comprising the principal west gate and ditches, and a number of buildings including granaries, a fabrica and a basilica exercitatoria which I reported to our congress in Rolduc. Smaller areas included part of the north-west rampart buildings, the east porta quintana, where the fort wall had totally collapsed, and the southeast angle tower. This work was published in 1997 (Wilmott 1997a). In 1996, meanwhile, the 1930s work on the cliff edge to the south of the fort was re-visited (Fig. 2; cf. Young, Friell & Wilmott 1999: 269-70).

It still seems likely that the vallum was built to respect the timber fort. Fort building involved the demolition of part of the Turf Wall, and the vallum line deviated southwards to respect the south wall of a fort. Despite this, it was clearly shown at Appletree that the mounds of the vallum were built on ground previously stripped of turf for the Turf Wall before any substantial regeneration of vegetation took place. The time lapse is problematical, as the potential speed of such regeneration cannot be assessed, although during excavation it was noted that truncated rhizomatous plants began to sprout again after only six weeks.

In 1997 a geo-physical survey revealed the extent of stone founded vicus buildings to east and west of the fort (Fig. 3; cf. Biggins & Taylor 1999; Biggins, Robinson & Taylor 1999), and Cumbria County Council received a grant to convert the farm buildings in the north-west corner of the fort into a residential study centre for the Wall. The corollary of this was excavation beneath the floors and yards of the farm during 1997 and 1998, linking the stratigraphy of previously excavated areas and completing the excavation of the western half of the praetentura. In fact, in the last 13 years, 21% of the fort interior has been excavated.

The 1999 section proved that the so-called marginal mound of the vallum had the identical relationship to the underlying strata as did the main north and south mounds and was constructed of the same clean material as the other mounds; a fact demonstrated also in Wall mile 29 in 1997 (Wilmott 1999). The mound sits on the southern lip of the vallum ditch, and, though smaller than the main north and south mounds, presents a considerable earthwork along much of the line of the vallum. Now in two places 20 miles apart, the evidence from stratigraphy and composition shows that the mound was part of the original concept of the vallum. This goes against the accepted idea of the marginal mound being the result of the cleaning out of the vallum ditch after the retreat from the Antonine Wall. We should now accept that in its primary form the vallum had two mounds to the south and one to the north, almost like a linear multivallate fortification. The implication of this is to bolster a primarily military interpretation for the work

In 1999 the television archaeologists of Time Team excavated trial trenches in the western vicus and in the cemetery beyond it, produced a TV programme as a result, and the new centre was opened by the Princess Royal. Elsewhere in the Birdoswald sector, the well-known Appletree section (Fig. 1) was opened-up every 10 years to reveal the Turf Wall to the Pilgrims of Hadrian’s Wall

851

Limes XVIII

which now emerges as having been stronger against the south than previously thought.

construction of the drain took place during the occupation of Scotland, and may have been carried out by a holding garrison. The rebuilding of the Turf Wall in stone to the north corners of the fort might then be considered in context with the work attested on the curtain elsewhere in 158. The cutting of the ditches after c.160-170 might thus be an aspect of reconditioning if the fort received a full garrison on the withdrawal. Clearly more work is required.

As to the Hadrianic stone fort of Birdoswald; at Rolduc (Wilmott 1997b) I mentioned that the discovery of the basilica must have had implications in the way that the praetentura was laid out. We now know what these were. On the street front was a narrow store, later converted to use as a fabrica, behind it the basilica, and behind that a pair of barracks with 8 contubernia and projecting officer’s quarters at the western end (Fig. 4). The fact that there were only 8 contubernia in these buildings, and those identified through geophysics in the eastern reminds us that we do not know for certain what the primary garrison was. The only evidence from the site is a graffito on a samian sherd recording the decurio Martinus (Tomlin 1997: 356). This was found in the primary rampart immediately to the north of the 8 contubernia barracks. It seems at least possible that the barracks each housed two turmae, perhaps belonging to the cavalry element of a cohors equitata, and that this was the context for this graffito on the inside wall of the basilica, but it really is to soon to say. Although little of the barracks survived, we can be sure that pits of the sort found in the combined stable barracks of Wallsend, Oberstimm and Dormagen were not present in these buildings.

The Severan remodelling of the fort (Fig. 5) included the construction of the granaries, and the reconstruction of the principal west gate, the conversion of an interval tower on the north wall into a cookhouse, and the alteration of the barrack accommodation in the west praetentura. The northern barrack was partially rebuilt to create a separate centurion’s quarters. This was provided with a private latrine and drain, and one room at least was provided with a channel-hypocaust. It is difficult not to conclude that this burst of activity coincided with a change of garrison, and the mention of cohors I Aelia Dacorum on the inscription relating to the construction of a granary found in 1929 (RIB 1909) may show that it was their arrival which prompted the work. The C3rd was one of continued occupation by the Dacian cohort. Building plans were changed little, though some gate portals, certainly the south portal of the principal west gate, were walled up. This is the period for which we now have increasing evidence from beyond the fort walls. The Time Team trenches produced C3rd cremations, contemporary with those which had been ploughed up in 1959-61. The trenches in the west vicus showed the presence of substantial stone-founded buildings, some with apparent industrial functions, during the late C2nd and C3rd. There is no doubt from the geophysics that the vicus to the east of the fort was similar in character. The only dating evidence for the fort vici derives from these trenches. The pottery shows strong representation of late C2nd and C3rd wares and forms, but not a single piece of Crambeck greyware, nor a sherd of East Yorkshire calcite gritted ware. Given the reasonable size of the assemblage, these would be expected if the vicus had been occupied beyond the end of the C3rd, and are a common find from within the fort walls. It seems reasonable to conclude that the vicus was, therefore abandoned during the later C3rd or earlier C4th. This seems to be becoming a general pattern for Hadrian’s Wall and its hinterland; vici with C2nd and C3rd occupation, coming to an end by the end of the C3rd, can be seen at Greta Bridge, Binchester, Housesteads, and Vindolanda vicus 1 (Bidwell 1999: 29).

It is clear that the stone fort is broadly Hadrianic in date, and that, as noted in 1929 (Richmond & Birley 1930), there really is a major reconstruction phase during broadly the Severan period. The period between, however is a serious problem in the history of the site. In 1929 it was not claimed that Wall Period 1B, which conventionally reflected the return from Antonine Scotland, existed at Birdoswald, and this lack of resolution to the period has been mirrored in every excavation since. Within the fort it has not been possible to identify any differentiation in structures or their use between these dates. Outside the fort, however, the sequence of vallum and fort ditches has been suggestive, particularly in the light of work on the pottery by Drs Jerry Evans and Steven Willis. The deliberate vallum fill towards the south-west corner of the fort contained a large pottery assemblage of late HadrianicAntonine date, the group being closed shortly after 150. The back-filled vallum was then cut by a stone-lined drain which ran westwards from the fort, and this was in turn cut by the primary triple ditch system of the stone fort. A small group of pottery from the fill of the drain gives a terminus post quem of 160/70 for the ditches. The fort was clearly not provided with ditches at all until this date. Furthermore, a reconsideration of the dating evidence for the replacement of the Turf Wall with stone in the Birdoswald area is beginning to suggest that this did not occur late in the reign of Hadrian as previously thought, but later, in the 150s. This evidence, which is more redolent of an active continuity of occupation on some scale through the Antonine occupation of Scotland than of complete desertion is confirmed by Dr David Shotter’s work on the total coin assemblage from the site. It seems likely that the back-filling of the vallum and the

An inscription found in 1929 (RIB 1912) states that the praetorium was collapsed and covered in earth, and rebuilt around 297-305, and has been held to be evidence of a short-term desertion of the fort. Although accepting with Thomas and Witschel (1992) that such texts are often at the very least misleading, there is further evidence for change, possibly for desertion. In the late C3rd the fabrica in the west praetentura collapsed and was not rebuilt for some time. When it was rebuilt it was no longer used for 852

Tony Wilmott: Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999 metalworking. The inner fort ditch on the west side silted completely, silt being deposited against the fort wall. It is probable that this would have affected the drainage system of the fort, which would have created flooding and silt deposition in the lowest lying part of the fort which is the site of the praetorium. Though this is far from conclusive it is suggestive evidence. Whatever the truth of the theory of the fort desertion, the rebuilding of the praetorium attested in the inscription synchronises well with the construction of the building which succeeded the fabrica, and the reassertion of the ditch and drainage system including a new culvert at the west gate (Wilmott 1997a: 199-201). In the praetentura truncation has removed most evidence for the C4th, but there is evidence for the remodelling of the north-west barrack, perhaps as a series of so-called chalets (Fig. 6). These buildings partly overlie and partly reuse the walls of the earlier barrack blocks. In the centurion’s quarters of the north barrack a pair of rooms became one, and were provided with a western apse. There is some evidence to show that the apsidal building continued in use longer than any other structure in this part of the praetentura, as an access path was laid over the demolished walls of the late chalet style barrack suggesting that it may have had some significance out of proportion to its size.

what it is worth) in the assemblage from the trenches in the west vicus. When trenches were opened south of the fort, however, this material started to turn up again, scattered through at least 10 different features in two separate excavation areas, all dating to the C3rd or later. At least two sherds were found in beam slots of the timber buildings, and the two complete vessels found in the 1930s come from a pit associated with a hearth which may be associated with the timber buildings. The timber buildings cut the back-filling of the outer fort ditches which contain pottery of C3rd date. On the other two sites where this ware has been found, Housesteads and Vindolanda, it is also limited to outside the fort walls (saving three sherds from Housesteads). This pattern is very strange and we can find no good parallels for restriction of a material class to the vicus in contemporary occupation with a fort. There can be no doubt that the patterning at Birdoswald and probably also at Housesteads and Vindolanda is significant. The context of the material becomes even stranger when the contemporary finds assemblage from the Birdoswald Spur is considered. There is no later C3rd glassware, little contemporary Roman pottery, and only one C3rd radiate coin; hardly the finds assemblage to be expected from a set of vicus buildings housing numeri (and their families). Why should numeri, or their pots, be barred from the fort interior? And if this were the case why build a new, unique, timber vicus when the west stone vicus was apparently being abandoned at this time?

The C4th saw the collapse of the north granary, and the continued use for a different function of the southern building. This was apparently followed in the C5th by a succession of timber buildings which form a stratigraphic continuum from the C4th, and possibly reflect the rise of a new social order, descended directly from the Roman system.

Given the exceptional nature of these patterns the other possibility which might be considered is that the pottery is early post-Roman as are the Birdoswald timber buildings. This would make sense of the association, and would either bring the timber buildings into contemporaneity with those inside the fort, or make them even later, conforming to the typical wic pattern, where post-Roman activity appears just outside the preceding Roman settlement. The latter interpretation is strengthened by the presence in similar contexts of several sherds of early Saxon pottery, possibly from the Charnwood forest area, another ware which does not appear in the fort itself.

A new and intriguing question has emerged from the analysis of the evidence from the 1996 excavations to the south of the fort, particularly Jerry Evans’ pottery analysis. It has become clear that the sequence deduced from trenching in this area in the 1930s was erroneous. Stratigraphic relationships and new dating evidence from groups of pottery have clearly shown that previous interpretations are highly suspect. A startling example is the large quadrangular enclosure, thought to be pre-fort. Pottery from the primary fill of this feature shows that it is in fact C3rd in date. Its recorded relationship with the vallum is therefore wrong. A basic error in the interpretation of the relationship between two such major cut features opens up the whole area to reinterpretation based on the new evidence. There are two particular oddities in the archaeology of this area; the presence of a large number of timber buildings, and that of the characteristic pottery known as Housesteads ware. Housesteads ware was first found in this area south of Birdoswald when two complete vessels were found. It has affinities with material from Frisia in the northern Netherlands, and, when found at Housesteads as well, was associated with the cuneus Frisiorum attested there in the C3rd (Jobey 1979). This association earned the ware its name. Not a single sherd of this ware was found in the excavated 21% of the fort interior at Birdoswald, or (for

The first antiquarian visitor to Birdoswald was Reginald Bainbrigg in 1599. As we enter the fifth century of archaeological interest in the site, and as work continues on the post-excavation analysis of the recent work, there is no shortage of outstanding questions to answer. Bibliography Bidwell P. (ed.) 1999 Hadrian’s Wall 1989-99. (Newcastle). Biggins J.A. & Taylor D.J.A. 1999 A survey of the Roman fort and settlement at Birdoswald, Cumbria. Britannia XXX: 91-110. Biggins J.A., Robinson J. & Taylor D.J.A. 1999 Birdoswald: geophysical survey. In P. Bidwell (ed.) Hadrian’s Wall 1989-99. (Newcastle): 157-160.

853

Limes XVIII

Gillam J.P. 1950 Recent excavations at Birdoswald. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 50: 63-68. Jobey I 1979 Housesteads Ware; a Frisian tradition on Hadrian’s Wall. Archaeologia Aeliana5 9: 127-143. Potter H.G. 1855 Amboglanna: Communicated to the Society of Antiquaries of Newcastle-upon-Tyne, Dec. 1852. Archaeologia Aeliana4 1: 141-149. Richmond I.A. & Birley E.B. 1930 Excavations on Hadrian’s Wall in the Birdoswald-Pike Hill Sector, 1929. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 30: 169-205. Simpson F.G. & Richmond I.A. 1932 1. Birdoswald. In Report of the Cumberland Excavation Committee for 1931: Excavations on Hadrian’s Wall. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 32: 140-145. Thomas E. & Witschel C. 1992 Constructing reconstruction; claim and reality of Roman building inscriptions from the Latin west. Papers of the British School at Rome 60: 135-177. Tomlin R.S.O. 1997 Inscribed stones and graffiti. In T. Wilmott Birdoswald: Excavations of a Roman fort on Hadrian’s Wall and its successor settlements: 1987-92. English Heritage Archaeological Report 14. (London): 355-357. Whitworth A.M. 1992 The cutting of the Turf Wall at Appletree. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 92: 49-56. Wilmott T. 1993 The Roman cremation cemetery in New Field, Birdoswald. Transactions of the Cumberland & Westmorland Antiquarian & Archaeological Society 93: 79-85. Wilmott T. 1997a Birdoswald: Excavations of a Roman fort on Hadrian’s Wall and its successor settlements: 1987-92. English Heritage Archaeological Report 14. (London). Wilmott T. 1997b A new type of building in an auxiliary fort; the Birdoswald basilica. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L.Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 581-586. Wilmott T. 1999 Black Carts. In P. Bidwell (ed.) Hadrian’s Wall 1989-99. (Newcastle): 120-122. Wiltshire P.E.J. 1992 Interim palynological report from Birdoswald Roman fort and Appletree Turf Wall, Cumbria. Ancient Monuments Laboratory Report 3/92. Wiltshire P.E.J. 1997 The pre-Roman environment. In T. Wilmott Birdoswald: Excavations of a Roman fort on Hadrian’s Wall and its successor settlements: 1987-92. English Heritage Archaeological Report 14. (London): 25-40. Young C., Friell G. & Wilmott T. 1999 The management, conservation and better understanding of Hadrian’s Wall. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău).

854

Tony Wilmott: Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999

Fig. 1. Location of Birdoswald and the Appletree section.

Fig. 2. Areas of excavation 1987-92.

855

Limes XVIII

Fig. 3. Primary structures in the north-west quarter of the stone fort.

856

Tony Wilmott: Research and development in the Birdoswald sector of Hadrian’s Wall 1949-1999

Fig. 4. Early C3rd structures in the north-west quarter of the stone fort.

857

Limes XVIII

Fig. 5. C4th structures in the north-west quarter of the fort.

858

Cawthorn Camps – 70 years after Richmond Pete Wilson Cawthorn Camps is one of the classic sites of RomanoBritish military archaeology. The site is located 40kms north-east of York (Fig. 1). It first enters antiquarian literature in Drake’s Eboracum (Drake 1736: 36) and was discussed and illustrated by Roy (1793: 65, pl. xi). However it was Professor Sir Ian Richmond’s excavations, undertaken following trial work by F.G. Simpson, that really drew the site to the attention of the archaeological profession and public (Richmond 1925; 1926; 1927; 1928; 1929; 1932; Simpson 1923). Contemporaneously with Simpson and Richmond’s excavations a local amateur, Dr J L Kirk, undertook work on aspects of the site, his results were incorporated in Richmond’s final report (Richmond 1932).

published plans only show those things that Richmond considered of interest. Richmond found evidence of two phases of occupation in earthwork A which he believed to have been extended by the addition of earthwork B. He suggested that the eastern rampart of A was ‘half-demolished’ to allow the creation of an extended ‘camp A/B’. Ovens were located within earthworks A and B, four of which Richmond assigned to the first phase of A and 8 the second phase. He also claimed to identify a first phase ballistarium on the southeast corner of earthwork A (Richmond 1932: 33-34, plate viii) and a further six on the north rampart of earthwork B belonging to the second phase (Richmond 1932: 57-58). The well-preserved claviculae, seen in earthworks A, B and C, and which in the case of A appear to have replaced straight traverses, or to use Richmond’s term tutuli, are consistent with Richmond’s proposed dating of AD80-120 for the whole complex.

The site consists of four well-preserved earthworks, rather confusing labelled by Simpson and Richmond from west to east D, C, A and B (Fig. 2), although in this designation they follow Young (1817: 694). In some ways the sites have had a sorry history, having been the subject of extensive afforestation, apparently on a commercial basis in the early part of this century, more recent woodland regeneration, the bulldozing of a firebreak, ploughing in the northern part of earthwork D, military training and unintentional bombing. Despite this they present a tremendous resource. They are Scheduled Ancient Monuments and were purchased in 1983 by the North York Moors National Park with the objective of preserving the site, not only because of its archaeological importance, but also for its nature conservation interest and the potential for developing appropriate public access. The work I describe below forms part of a joint English Heritage and North York Moors National Park project.

Richmond concluded that the complex had ‘no connexion with the permanent occupation of Yorkshire’ (Richmond 1932: 78) and represented a series of practice works. He suggested that they were constructed by legionaries from York, either of legio IX Hispana, if before c.108, or legio VI Victrix if after. This conclusion was largely based on the evidence of the ballistaria, as he suggested no other type of unit would have artillery at such an early date (Richmond 1932: 78). Richmond’s argument went contrary to the thinking of Young (1817: 698) who had concluded that both camp D and camp A must have been what he termed ‘standing camps’, that is forts. Despite this Richmond’s interpretations have stood unchallenged until relatively recently. However in 1982 Brian Hartley suggested that Fort D might have had a more permanent occupation (Hartley 1982: 211-12). Cawthorn only received further direct attention when it was resurveyed by the former Royal Commission on the Historic Monuments of England, now part of English Heritage, as part of the work that led to the publication of Roman Camps in England (Welfare & Swan 1995: 137-42). This work identified many additional features and provides the base plan for Fig. 2.

In addition to the Roman earthworks the site incorporates a probable Bronze age barrow within earthwork A. Richmond (1932: 62) suggests that the barrow was trenched by the antiquarian Thomas Bateman in 1860, although he names [Canon William] Greenwell on his plan (Richmond 1932: fig 15). The barrow was subject to further investigations by Simpson and Kirk in the 1920s leading to the suggestion that it had been modified during the Roman occupation to form a tribunal (Richmond 1932: 61-63). A middle Iron age square barrow containing a poorly preserved cart burial was excavated outside the south-east corner of earthwork B (Mortimer 1905: 361), and there is another square barrow south-west of earthwork A (Fig. 2). Further probable square barrows have been found to the south of the main earthwork complex.

Welfare and Swan proposed that earthworks A and D were both permanent forts, albeit occupied for relatively short periods; earthwork B was an annexe added to Fort A, presumably during Richmond’s second phase and that earthwork C was the only real camp in the group. Their discussion considers a number of anomalies that are outside the scope of this short paper, however their reevaluation is both incisive and convincing and opens up many questions regarding the site. One key suggestion that

The work in the 1920s concentrated on earthworks A and B, with only limited trenching being undertaken on earthworks C and D. As with many excavations of that time we are woefully ignorant of the extent of the interventions, there being no extant trench plan and the

859

Limes XVIII

they made was to propose that Fort D, as it is seen now, had a predecessor in the same general position that was in occupation when Camp C was constructed. It is clear from the existing earthworks that D, in its present form, is later than C as the latter’s defences are cut away by the outer ditch of D. An earlier single-ditched phase of D could necessitate the truncation of the north-west corner of C to retain access to the east gate of the fort.

the archaeological record. The turf-walled building had a shallow foundation-trench, presumably representing part of another structure, underlying its southern wall. The structure investigated in Trench 1 (Figs. 3 & 4) was part of a sub-square earthwork that Richmond suggested was the site for a pair of centurion’s tents. It was 6.25m square internally and better preserved than the earthwork in Trench 4. The southern wall (102) was constructed of well-defined turves (117), sealed by a capping of a more homogenous layer (113). Layer 113 appeared to represent degraded turf, similar to that eroded from the wall on the south side of the mound (160), but with layer 113 remaining in situ. Overall wall 102 survived to a height of 0.35m and was 1m wide. The north wall (101) was similar to wall 102, but less well preserved. Overall the wall was 0.3m high, but the well-defined turf component (142) was at most 0.12m thick, as opposed to 0.24m for layer 117 in wall 102. Again there was a layer of degraded turf forming the upper part of the wall (107/150) which was clearly seen to have been eroded off the wall to the north. The southern wall incorporated a post-pad consisting of flat-laid stones (126) 0.5m in diameter. There was no clear evidence of a post-pad in the corresponding position in the north wall, although the wall had suffered considerable erosion at that point and further evidence could easily have been lost. Another feature (122) found in the centre of the building appeared to represent a small hearth which consisted of a sub-circular ‘ring’ of clay filled with charcoal. The hearth was sealed by a group of flat-laid stones associated with further stones forming a near vertical eastern edge. There was no evidence that the stones had been burnt, therefore this later phase is unlikely to represent a further hearth and it is possible that they belong to a post-pad deriving from a second phase of occupation in the area. No clear floor levels were identified, although three layers (128, 135, 157) appeared to represent occupation surfaces within the structure, the latest of which had a shallow charcoal filled scoop (127) cut into it, which appeared to represent a hearth. To the north of the structure, which was clearly a building, a gravel/cobble spread (118) was located running east-west across the trench. This spread appeared to represent a deliberate attempt to produce a laid surface within the area of the ‘hollow-way’ (103) created by the surrounding earthworks.

There are problems with the site as a whole with regard to long term occupation, not least the absence of water given its location on the edge of a substantial escarpment. However until recently there were dew-ponds nearby and the Sutherland Beck is only some 300m away at the foot of the northern slope. For a confident field army in a period of expansion the dispatch of watering parties such a short distance would presumably have occasioned little concern. Equally it is possible that wells may exist within the complex, but if they do they are still to be found. A more recent reconsideration of the complex by Graham Lee (1997) has proposed a sequence for the earthworks based on the work of both Richmond and Welfare and Swan, which although logical requires validation. The earliest component is suggested to be Fort D with a single ditch. Camp C is then constructed while D is occupied, with C presumably acting as a construction camp for Fort A, as originally suggested by Richmond. A and D could then have been occupied at the same time or D abandoned. The final phases see D refurbished with a second ditch and an Annexe B added to A, although there is nothing to suggest the relative sequence of the final occupations of Forts A and D. Our excavations were conceived against this background, and were designed to elucidate the site sequence and it was hoped provide additional dating evidence. A total of 10 trenches have been excavated (Fig 2). In Annexe B Trenches 1 and 4 examined two turf-walled structures of the type that Richmond suggested: “…related to tents or rough buildings of the type that appear in not dissimilar stone mounds at Masada, … while on one hand it is clear that at Masada buildings were involved, it is not clear that at Cawthorn these turf mounds ever formed the walls of buildings. …it seems clear that the lines turf have to be considered much rather as screens against wind, or as dams against wet, than as the walls of regular buildings, and that they were arranged round tents or camp-fires, or even inside tents, serving as benches or tables. Even on this view they do not all now become coherent; but the south-east group clearly marks two lines of tents, back to back, with a pair of centurion’s tents next to the gate” (Richmond 1932: 59-61).

In Fort A a further earthwork was investigated in Trench 5. Overall this structure was c.7 x 44m and aligned east-west. A probable entrance in its southern side fronted onto a possible street identified during a recent topographic survey (not shown on Fig. 2). The turf structure was the earliest in a sequence of two or three buildings. The possible second phase was represented by in situ daub (Fig. 5) with clear evidence of stake-holes for wattles. It is currently not clear if the wattle and daub defines a separate construction phase, or was inserted as lining to the turf walls. The final building was represented by two shallow east-west foundation trenches filled with local stone and associated with a rough stone floor.

The structure in Trench 4 was sub-rectangular and aligned east-west. It was c.8 x 6.7m overall had rounded corners at the eastern end and the west end was either open, or had a wall of timber or other material that did not survive into 860

Pete Wilson: Cawthorn Camps – 70 years after Richmond

The defenses of Fort A were investigated at three points. Trench 2 was excavated to natural through the eastern rampart and ditch (Figs. 6 & 7). Although evidence of a palisade trench was found along the top of the rampart, no evidence came to light of the timbering Richmond claimed to have found at the front and back of the rampart. Our trench was 5m wide and, although only a 1.5m strip was taken to natural, enough of both the front and tail of the rampart were examined to be confident that some elements of Richmond’s front posts at 10 foot (3m) intervals and rear posts at 5 foot (1.2m) intervals (Richmond 1932: 2425, 27) would have been seen. What was revealed was a rampart of dumped material that 6.1m wide and survived to around 0.7m. There was no evidence of deliberate slighting, despite some rampart material being seen in the ditch (see below). The ditch itself was 4.5m wide at the level of the old ground surface and was c.2.15m deep, with a distinct cleaning slot. The presence of the cleaning slot suggests that the fort was occupied for some time, rather than being a practice work as suggested by Richmond. It would seem that the ditch was not backfilled on the abandonment of the site, but remained open to allow the development of a turf level (279) which was sealed by a layer with a high stone content (278). The stone layer probably represents decay, it certainly did not appear to represent an attempt to backfill the ditch and level the rampart to allow Fort A to be expanded to the east when Annexe B was added as Richmond suggested. This seems to be confirmed by the fact that the defences would have remained a substantial obstacle, as from the top of the stone layer to the top of the rampart as it exists now is c.2m. A further stone layer was encountered higher-up in the ditch fills and it is possible that this relates to the construction of the next feature to considered.

deep, and appeared to be of a single phase. Neither of them contained any silt, but were back-filled with rampart material, suggesting that they had been re-cut for the second phase of occupation which was probably of short duration. Both ramparts were constructed over a buried soil which sealed under the north rampart a substantial ditch (c.2.8m wide and 0.6m deep) and a large pit under the western. These features provide evidence of pre-Roman occupation to go with the probable barrow within the fort and the 95 residual flint objects from the 1999-2000 excavations. Two trenches were excavated within Fort D to test the level of preservation of internal features. It is clear that the interior of northern third of the fort had substantially damaged, given a total absence of features in Trench 8. Aerial photographs from 1925 suggest that the destruction was the result of C20th ploughing. No clearly Roman features were recorded in Trench 9, although the remnants of some four features survived cut into the natural subsoil. A spread of stone in the northern part of the trench suggested that stone-founded buildings, or stone floors similar to that seen in Trench 5 may have existed in the area, alternatively the stone could derive from an area of yardage.

Trench 2 was sited to partially take-in a feature cut into the rampart - Richmond’s ‘Officer’s dug-out’ number 2. Richmond (1932: 65-68, pl. xvi) identified five of these features which he dated to his second occupation (his Camp A/B) and suggested they were ‘officer’s ‘dug-outs’’ as they were located in what he regarded as the area of the praetorium. Although it is clear that he left ‘Officer’s dugout’ number 2 open to weather and fill gradually the rapid re-excavation of half of it demonstrated that it was in fact a grubenhaus, with post-holes observed by Richmond providing support for the roof. Unfortunately there were no finds from the grubenhaus.

Trench 10 was excavated across the defenses of Fort D where they cut away the western defenses of Camp C. As in the case of Fort A two phases of rampart were recognized, although without any evidence of palisade trenches, or other structural features. The earlier phase was of dump construction and largely consisted of ditch upcast. The second phase was constructed of turf and incorporated large numbers of clearly defined turves (Fig. 8). Overall the rampart survived to a height of 1.5m and was 5m wide. The inner ditch was of four phases up to 1.35m deep and 1.5m wide. The outer ditch was single phase and 1.1m deep and 3.4m wide. The lack of re-cuts suggests that the outer ditch related to the second phase of the fort. This supports for the suggestion that the first phase of Fort D may have remained in use when Camp C was constructed with access to the east gate of the fort being maintained (Welfare & Swan 1995: 140). Upcast from both ditches was used to heighten the berm between them turning the ditches into more formidable obstacles. The rampart of Camp C was shown to be of simple dumped construction and 3.3m wide and 0.6m high.

Trenches 6 and 7 were located on the northern and western ramparts of Fort A and were designed to compare the badly eroded northern rampart with the better preserved western one, the latter being 4.6m wide and 1.45m high. Both trenches produced clear evidence of two phases of rampart, an earlier one of dumped construction associated with a palisade trench, with a capping of clay representing the later phase. Three small oven bases were located cut into the rear of the northern rampart. They were apparently sheltered internally by a short turf bank aligned parallel to the rampart. The ditches were up to c.5.5m wide and 2m

The structures investigated in Fort A and Annexe B are, contra Richmond, clearly buildings. Although they do not appear as conventional Roman military buildings they are almost certainly contemporary with the military earthworks. The finds from them are consistent with occupation in the late C1st and C2nd as Richmond suggested. However the paucity of material from the site as whole (including Richmond’s work): less than 50 sherds of pottery; four melon beads; one nail; one lump of slag and four fragments of glass is surprising. The assemblage can be taken to suggest that the occupation was by Units that

861

Limes XVIII

were only there for short periods of time. These time periods might be measured in months, or perhaps extend to a couple of years. The lack of finds can be taken to suggest that the troops were operating in, what might be termed, ‘campaign mode’; not using much pottery or glass and presumably therefore cooking with and eating from patera.

Richmond I.A. 1928 The Roman camps at Cawthorn, near Pickering. Fourth interim summary. Yorkshire Archaeological Journal 29: 225-231. Richmond I.A. 1929 The excavation of the Roman camps at Cawthorn. Fifth interim summary. Yorkshire Archaeological Journal 29: 327-331. Richmond I.A. 1932 The four Roman camps at Cawthorn, in the North Riding of Yorkshire. Archaeological Journal 89: 17-78. Roy W. 1793 Military Antiquities of the Romans in North Britain. (Society of Antiquaries. London). Simpson F.G. 1923 The Roman camps at Cawthorn, near Pickering. Yorkshire Archaeological Journal 28: 25-33. Welfare H. & Swan V. 1995 Roman camps in England. (London). Young G. 1817 History of Whitby and Streoneshalh Abbey. (Whitby).

It is also possible to say with confidence that Fort A was not slighted to the extent that Richmond suggested. The lay-out of the earthworks in the south-eastern and northeastern parts of Annexe B can be seen as somewhat reminiscent of barrack blocks, with the structure in Trench 1 representing the centurion’s quarters, paralleling what Richmond (1932: 61) suggested had the accommodation been in tents. The complex in the south-eastern part of Annexe B seems rather over-long for a barrack block. It is c.90m, compared with the average of 40-50m suggested by Johnson (1983: 167) for auxiliary barracks and the 70-75m lengths of the York, Caerleon and Chester legionary barracks. It might be able to account for the latter discrepancy by the method of building, the thickness of the turf walls necessitating a lengthening of the structure. However the earthworks on the north side of the ‘street’ are much less regular and it seems that as things stand now we do not know what the buildings in Annexe B were used for. Acknowledgements Thanks are due to the North York Moors National Park, both as owners of the site and partners in the project, with particular gratitude due to Graham Lee their Archaeological Conservation Officer. The figures accompanying this paper were prepared by Vince Griffin of the English Heritage Centre for Archaeology Graphics Studio. Bibliography Drake F. 1736 Eboracum: or the History and Antiquities of the City of York. (London). Hartley B.R. 1982 Military activity. In D.A. Spratt (ed.) Prehistoric and Roman archaeology of north-east Yorkshire. British Archaeological Reports Brit. Ser. 104 (Oxford) = rev. ed. 1993 – Council for British Archaeology Research Report 87. Johnson A. 1983 Roman forts of the 1st and 2nd centuries AD in Britain and the German provinces. (London). Lee G. 1997 Cawthorn Roman military complex - an update. Archaeological Journal 154: 260-267. Mortimer J.R. 1905 Forty years researches in British and Saxon burial mounds of East Yorkshire. (London). Richmond I.A. 1925 The Roman camps at Cawthorn, near Pickering. Yorkshire Archaeological Journal 28: 332-339. Richmond I.A. 1926 The Roman camps at Cawthorn, near Pickering. Second interim summary, 1925. Yorkshire Archaeological Journal 28: 421-426. Richmond I.A 1927 The Roman camps at Cawthorn, near Pickering. Third interim summary, 1926. Yorkshire Archaeological Journal 29: 90-96.

862

Pete Wilson: Cawthorn Camps – 70 years after Richmond

Fig. 1. Cawthorn Camps – Site location

Fig. 2. Cawthorn Camps – Trench locations

863

Limes XVIII

Fig. 3. Cawthorn Camps – Plan of turf building in Annexe B (Trench 1).

Fig. 4. Cawthorn Camps – Section through turf building in Annexe B (Trench 1).

864

Pete Wilson: Cawthorn Camps – 70 years after Richmond

Fig. 5. Cawthorn Camps – In situ daub with stake-holes from the buildings in Fort A (Trench 5).

Fig. 6. Cawthorn Camps – Plan of trench through the eastern defences of Fort A (Trench 2).

865

Limes XVIII

Fig. 7. Cawthorn Camps – Section through the eastern defences of Fort A (Trench 2).

Fig. 8. Cawthorn Camps – Part of second phase turf rampart of Fort D, material forming the phase 1 rampart is visible at the bottom of the section (Trench 10).

866

The Roman Gask frontier: 1997-2000 D.J.Woolliscroft This paper offers an update on work on the system of military works on or around the Gask Ridge in north-east Scotland, a line which may now represent Rome's first fortified land frontier. Since the last report at the XVIIth Congress 3 seasons work have been completed and a number of significant discoveries can be reported, some of which have clarified aspects of the system. Conversely other discoveries have served to create new problems and further questions. The Project has also begun to explore the issue of the native/civilian population in the area of the system and its relationship with the Roman military and a number of observations might be offered.

At the 1997 Limes Congress the writer reported on the first two years of a project to study the system of military works on and around the Gask Ridge in north-east Scotland, a line which may now represent Rome’s first fortified land frontier (Woolliscroft 1999). Since then, a further three seasons’ work have been completed and the present paper will give a brief account of developments.

though, still remains open to question. If Tacitus is to be believed, the Romans did not even reach this area until Agricola's third campaign in 79 or 80, whilst Hobley (1989) has argued, from coin evidence, that they must have left again soon after 86/7. Much probably depends on how long tower uprights of up to 40cms in diameter could be expected to last before needing replacement which, in turn, depends at least partly on what they were made of. Hanson (1978) has pointed out that the Romans sometimes used far from ideal timber in military structures, alder for example, which tends to rot fairly quickly once set into the ground. Environmental analyses from a number of Gask sites have suggested that virtually the only trees in this landscape in Roman times were water loving species, such as alder, around the rivers (Ramsay forthc. a & b), but there are a number of wood fragments from the post holes of Raith and Roundlaw towers (Christison 1901: 28f.; Ramsay in Woolliscroft forthc.) which suggest that the structural timber actually used was oak which, even without preservative treatment, should have been reasonably durable. This sits well with evidence from the ditches of Shielhill South and Glenbank fortlet which had been redug (in the latter case repeatedly) after a considerable depth of silt had formed in the primary cuts. We might conclude, therefore, that we have either a single extended Flavian occupation, perhaps lasting for the full six to seven years of the known Flavian period, if not longer or, at the very least, that we have two distinct Flavian phases separated by such a short period of disuse that it is now archaeologically undetectable. Other scenarios are possible, for example a general destruction of the tower line by a severe storm, but the post-hole evidence would not appear to support this.

At the time of the XVIIth Congress, the Gask Project’s most important announcement was the discovery that two of the southernmost towers, Greenloaning and Shielhill South, had needed to be completely rebuilt at some point during their service lives. This was a hint that the system may have survived for longer than the very brief occupation than has often been envisaged (eg. Breeze 1982: 65) and similar results have since been obtained by a different team during Historic Scotland sponsored excavations at the tower of Blackhill Wood (Glendinning & Dunwell 2000). There remained, though, something a problem, because these three towers form part of a clear group of four, from Greenloaning to Shielhill North, at the southern end of the line (Fig. 1), which share one marked difference from the rest of the series, in that they are surrounded by two ring ditches rather than the usual one. Given a profound lack of dating evidence, for to date we have only a single sherd of late C1st glass between all four sites, there could be no guarantee that these towers shared the same history as the rest of the line. It remained conceivable, for example, that the southern end of the system was built first and then extended using a slightly different tower design. Moreover, if that was the case, it might also be possible that the southern towers were rebuilt when this extension was added, even though this might not yet have been strictly necessary, so that everything was new together. In other words, it remained possible that we might still be dealing with a relatively short chronology.

The last three years have done little for absolute dating on the Flavian line installations, for although two more pottery sherds have been recovered (one each from Huntingtower and a mysterious, site at Cuiltburn), these have not been more closely datable than that they are Roman. The forts have been better served, however. An analysis of field walking finds has finally been able to prove that (as has long been suspected) the fort of Bertha, like Ardoch and Strageath, was reoccupied in the midC2nd, presumably as an outpost to the Antonine Wall. More surprisingly, Dalginross, one of the line of so called ‘glen blocker’ forts to the north of the Gask, has also produced evidence for such reuse, although this series had previously shown only Flavian activity. Despite poor

In fact, there had long been evidence from earlier excavations that the towers on the longer, northern, single ditched, part of the line had also been rebuilt (Woolliscroft 1999: 294), but at the time of the last Congress this was not enough to be conclusive. In 1998, therefore, the Project excavated the northernmost known tower on the system, Huntingtower (Fig 1), and although the evidence was less clear than further south, this site does also seem to have been rebuilt, which means that we should now have a unified history for the entire frontier and probably a longer one than had previously been though. Just how long, 867

Limes XVIII

dating evidence, however, there would still seem little reason to doubt that the Gask towers were exclusively Flavian. This means that only the system’s three fortlets remain undated, for although the Gask Project conducted extensive excavations at the southernmost example, Glenbank, in 1999, no dating evidence has been recovered from any of these sites. It thus remains conceivable that the fortlets might be Antonine, and the possibility of a Flavian system of forts and towers and an Antonine one of forts and fortlets is certainly attractive. There is, however, one strong counter argument in that Glenbank, like the towers around it, had a double ditch, whereas the northern fortlets, like their towers, had single ditches, which does suggest integration between the two site types.

and 3-4m wide, whereas those of the southern group are often less than 50cms deep and only just over a metre wide and can have had little defensive value. Unlike the southern towers, this group shows no regularity of spacing. Finally, lying between the southern and eastern Ridge groups was an intermediate group, from Westerton to Roundlaw, whose ditches are noticeably smaller in diameter (average c.19m) and which have intermediate ditch volumes at around 1m deep and 2m wide. This group also shows no apparent regularity of spacing, but it does have another distinguishing feature in that, unlike the rest of the line, its rectangular towers have their long axes facing, rather than at right angles to, their ditch entrances. Moreover, at least one site, Westerton, had a pair of projecting slots running out towards the entrance which might be the foundation for a flight of steps to the tower top (Hanson & Friell 1995: 502ff.).

The fortlets have also produced a second conundrum. Excavations at the end of the C19th at Kaims Castle and Midgate (Christison 1901: 18ff., 32ff.) found traces of metalling but no internal buildings. This might have been dismissed as a reflection of the poorer field techniques of the period, but the Gask Project’s Glenbank excavations produced exactly the same results. The site was surrounded by two much re-cut ditches and had a turf rampart with a massive (apparently single phase) timber gate tower at its entrance, but nothing was found in the interior except for badly plough damaged metalling and a single small post hole which was not necessarily Roman. It would seem, therefore, that the Gask fortlets either never had internal buildings or, perhaps more probably, that their buildings are for some reason no longer archaeologically detectable. They may, for example, have been of cob, mud brick or turf walled construction or they may have been founded on sleeper beams which rested on, rather than cutting into, the surface and if this is the case their remains will have been ploughed away long ago.

The work at Huntingtower and a geophysical survey of its unexcavated western neighbour Peel, however, have provided clues that might suggest yet a fourth building sector, for both these sites have yielded ditch diameters of around 15.5m which are smaller even than some of the southern group’s inner ditches (which average c.16m). Huntingtower’s ditch sections were similar in volume to those of the intermediate group, but it had its tower set right at the back of the internal area, with its rear posts set into the turf rampart (Fig. 3), which may or may not prove to be a further diagnostic feature. There are also indications that this northern group might again have a regular spacing interval, this time of 2/3 of a Roman mile. All this seems even more clear cut and logical than it did in 1997 for, with the addition of a fourth building sector, the groups are not too dissimilar in length and they coincide reasonably well with what one might expect to be the spheres of influence of the forts. If we do have four groups, one can hardly resist noticing that there were also four legions in Britain at the time the system was built, which are the usual builder units and, as none of these designs repeat on further sectors, we may have an indication that the entire frontier was built in a single season. As before, however, we must continue to treat at least the three northern divisions with caution because there is, albeit just one, anomaly. For Parkneuk, at the western end of the Gask Ridge, which one would expect to belong to the intermediate group, has the widest ditch diameter of any tower known on the Ridge (Fig. 2) and a ditch volume in the eastern group range. Its excavator (Robertson 1973: 21ff.) thought that its tower was oriented in the intermediate group fashion, with its long axis facing the entrance, but not even this much is certain and so for the moment these groupings must be treated as suspect. There are in fact gaps in the spacing pattern which suggest that there are still quite a number of towers to be discovered on the Gask, especially in this area, and only further study will tell whether Parkneuk is really an isolated aberration or whether the model presented above is a figment of the current, incomplete data.

Both Glenbank and Huntingtower have, though, provided further evidence that the eventual end of the system saw a process of orderly demolition and abandonment, rather than military expulsion. For at both sites the main structural posts had been carefully dug out. In addition to progress on the frontier’s history, work has continued to understand its construction. At the last Limes Congress the writer (Woolliscroft 1999: 294f.) was able to suggest that the system had been built in three sections which each showed slightly different characteristics. The first was the southern group of four towers with its double ditches and a regular installation spacing of 3/5 of a Roman mile. At the time it was thought that the towers on this sector did not have the internal ramparts which have been found everywhere else on the line, but the recent work at Blackhill Wood (Glendinning & Dunwell 2000) has disproved this. Next came a group of sites from Kirkhill to Midgate towards the eastern end of the Gask Ridge which, although single ditched, have ditch diameters averaging c.22m (Fig. 2), only slightly smaller than the southern group’s outer ditches (which average c.24.5m). The ditches of this group are much more substantial than those of the southern towers. They range up to 1.8m deep 868

D.J.Woolliscroft: The Roman Gask frontier: 1997-2000 Around 70 large Roman temporary camps are known in northern Scotland, but at the 1997 Congress the writer mentioned aerial photographic indications that there might be a group of much smaller camps in the immediate vicinity of the Gask. At the time little had been done beyond mapping the sites from rectified photographs, but the Project has now been able to carry out trenching and geophysical work on six of the candidates. Four of these now seem unlikely to be Roman, but two (interestingly the two smallest) do still appear to be camps. Two (1 & 4) and just over 1 acre (0.4ha) in area. Each have standard military V-shaped ditches of up to 3.65m wide x 1.13m deep, but their dating remains less than secure. Easter Powside produced no datable material whatever and a carbon sample from East Mid Lamberkin’s ditch fill gave a calibrated date of AD585-700. This is only a terminus ante quem, however, which does at least rule out a medieval date, which had seemed the only other likely period of occupation. As the material came from some way up in the fill it should be considerably later than the cutting of the ditch which, coupled to the fact that the site’s morphology, and especially its three surviving entrances (the fourth has been destroyed by a modern road), are Roman in appearance, does make a Roman date appear likely. Moreover, one of the gates is reminiscent of, although not identical to, the so-called ‘Stracathro gates’ known from larger camps in the area and these do seem to be Flavian.

cultivation. This pattern contrasts with a more arable regime seen in the area in earlier times (eg. Romans & Robertson 1983) but, as the same picture can be seen in both immediately pre-Roman contexts and in Roman ditch silts, the Roman incursion may have had relatively little impact on agriculture.

The rôle of such diminutive camps presumably differed from that of the larger campaigning camps. They may reflect the movement of small bodies of troops, or they may be construction camps used during the building of the frontier. They might also be a product of training exercises and this does seem particularly feasible in the case of East Mid Lamberkin for, as well as having rather oversized ditches for its size, despite the fact that these are cut into an extremely hard rock and clay subsoil, the site’s three surviving entrances are each of a different, albeit recognisably Roman, type. As far as the writer is aware, this is a unique feature and it is tempting to wonder whether a group of Roman recruits may have been given practice here in constructing a range of alternative gateways. If so, they might have got the Stracathro type slightly wrong.

Finally, it is worth mentioning an unusual site at Cuiltburn, just to the south of Strageath. This was tentatively identified from the air as a new fortlet but, although at the time of the 1997 Congress only trial trenching and a resistivity survey had taken place, this identification had already come into question (Woolliscroft 1999: 284). Much larger scale work undertaken immediately after the Congress removed any remaining doubt for, although the site did produce a few sherds of Roman pottery (and nothing from any other period), along with Roman looking rectangular beam founded buildings, grouped around an internal courtyard and fronted by some sort of post hole based facade (Fig. 5), the defences are not those of a fortlet. The ditch goes around only three of the four sides and, although the site sits right beside and parallel to the Roman road, the open side faces away from the road not towards it. The ditch itself has a steep sided, flat bottomed profile, quite unlike the Roman military V-shape and there were no signs of an internal rampart or palisade. Nevertheless the site is quite probably Roman and it is certainly fascinating in its own right. Indeed we may be dealing with an entirely new site type, for we have yet to find a single excavated parallel, and the writer would be most interested to see if any similar features are known elsewhere in the Roman world.

As for the natives themselves and their interactions with the occupying power, we have as yet only tantalising glimpses. Three of the Gask installations, including Huntingtower and Glenbank, overlay native groove houses and it is tempting to conjure up a picture of local people being thrown out of their homes to make way for the army but, in the absence of dating evidence, it remains equally possible that these features may have been abandoned centuries before the Romans arrived. Aerial photographs of the area have revealed numerous examples of a variety of Iron age site types but, as always, it is difficult to date such features with any precision from the air to select promising targets for excavation and further study. As a compromise, the Project has been looking for multi-phased sites. These should be an indication of prolonged occupation which should, at least, provide a statistically higher chance of continuity into the Roman period, and an excavation is planned for the coming season at East Coldoch, near the Roman fort of Doune. This site has already been subjected to trial trenching and appears to show at least five separate phases.

Work on the military aspects of the frontier continues, but the Project now intends to give more attention to the native/civilian populations of the area. One large geophysical survey has already been conducted outside the fort of Strageath in the hope of locating vicus activity (as yet with inconclusive results) and more work is planned. For the moment the Project’s principle tool for studying the indigenous population has been environmental analysis, but we have been hampered here by the free draining and acidic nature of many of the local soils which tend to destroy both pollen and bone. The picture to date is thus far from complete, but what evidence we do have points to a virtually treeless landscape given over almost exclusively to grazing with, as yet, little sign of cereal

Bibliography The Gask Project’s research reports, both published and unpublished are all available on the WWW at www.morgue.demon.co.uk/Pages/Gask/index.html

869

Limes XVIII

Breeze D.J. 1982 The northern frontiers of Roman Britain. (London). Christison D. 1901 Excavations undertaken by the Society of Antiquaries of Scotland of earthworks adjoining the "Roman road" between Ardoch and Dupplin, Perthshire. Proceedings of the Society of Antiquaries of Scotland 35: 16-43. Glendinning B. & Dunwell A. 2000 Excavations of the Gask frontier tower and temporary camp at Blackhill Wood, Ardoch, Perth & Kinross. Britannia XXXI: 255-290. Hanson W.S. 1978 The Roman military timber supply. Britannia IX: 293-306. Hanson W.S. & Friell J.P.G. 1995 Westerton: a Roman watchtower on the Gask frontier. Proceedings of the Society of Antiquaries of Scotland 125: 499-520. Hobley A.S. 1989 The numismatic evidence for the postAgricolan abandonment of the Roman frontier in northern Scotland. Britannia XX: 69-74. Ramsay S. forthc. – a Pollen analysis from Parkneuk Roman road. In D.J. Woolliscroft & M.D. Davies Parkneuk Roman road, Perthshire, excavations in 1967 and 1997. Place of publication not finalised. Ramsay S. forthc. – b Pollen analysis from Huntingtower. In D.J. Woolliscroft The Roman Gask Series Tower at West Mains of Huntingtower, Perth & Kinross. Proceedings of the Society of Antiquaries of Scotland 130. Romans J.C.C & Robertson L. 1983 The general effects of early agriculture on the soil profile. In G.S. Maxwell (ed.) The impact of aerial photography on archaeology. CBA Research Report 49. (London). Robertson A.S. 1973 Roman ‘signal stations’ on the Gask Ridge. Transactions of the Perthshire Society for Natural Science Special Issue: 14-29. Woolliscroft D.J. 1993 Signalling and the design of the Gask Ridge system. Proceedings of the Society of Antiquaries of Scotland 123: 291-314. Woolliscroft D.J. 1999 The Roman Gask Project. In N. Gudea (ed.) Roman Frontier Studies 1997, Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 293-301. Woolliscroft D.J. forthc. The Roman Gask series tower at West Mains of Huntingtower, Perth & Kinross. Proceedings of the Society of Antiquaries of Scotland 130.

870

D.J.Woolliscroft: The Roman Gask frontier: 1997-2000

Fig. 1. The Roman Gask Frontier

Fig. 2. Gask Tower ditch diameters.

871

Limes XVIII

Fig. 3. Plan of Mains of Huntingtower.

Fig. 4. Small Roman temporary camps near the Gask system

Fig. 5. Plan of Cuiltburn. Excavations 1995-1997. Find spots: F - Flint fragment; N - Nail; P - Roman pottery; Lp - ?Post stance; RH - ?Round house.

872

The timing of marriage in the Roman army Sara E. Phang Due to the principate’s marriage ban for common soldiers and the circumstances of military service, Roman soldiers’ family formation differed from civilian men’s (Campbell 1978). I have carried out a survey of Latin epitaphs from Africa, the Danubian provinces, Rome (Praetorians) and Italy (the fleet). The first finding re-examines Saller & Shaw’s (1984) study which assumes that differences in soldiers’ family formation are regional. Instead, we find that dating the inscriptions shows the soldiers’ family formation to be temporally distributed; more soldiers never married in the C1st AD and the marriage rate increased in the C2nd and C3rd. I suggest reasons for this change; in the C1st legionaries were externally recruited and were less well paid; local recruitment and better pay in the later period fostered the marriages of soldiers. The second part of my survey examines soldiers’ age at first marriage, which is assumed to be young. This is a modern anachronism; Saller (1987) used a method of estimating civilian men’s age at marriage from tabulating commemorators (parents or wives/children) of these men; civilian men seem to have married in their late 20s. I used the same method on soldiers’ epitaphs and find that soldiers may have married in their mid-30s, later than civilians. I suggest reasons: military culture or selective tolerance of soldiers’ unions (cf. C19th western armies), or promotion and/or saving to support a household. Implications of soldiers’ later marriage include a re-examination of whether the army became a “hereditary military caste,” as the traditional view has it.

often married in their mid- to late 30s, a decade later than Roman civilian men. This finding is based upon Latin epitaphs; age at marriage is estimated from shifts in commemoration patterns in the epitaphs.

It is usually accepted that Roman soldiers were denied legal marriage during the principate; this ban was probably created by Augustus and was lifted by Septimius Severus in AD197.1 Despite the ban, many soldiers formed de facto families whose existence is recorded in military diplomas, papyri (both legal documents and private letters), and especially in Latin epitaphs. At what age did soldiers ‘marry’, to use the term in its de facto sense?2 Authors often assume that soldiers married young, shortly after enlistment.3 J.C. Mann asserts that “…[m]any of the (soldiers) began to cohabit with local women soon after service began, and by the time they were discharged had grown-up children” (Mann 1983: 62; cf. Wolff 1974: 479; Roxan 1986: 276; 1991: 465-466; Luttwak 1976: 125 but without explanation; Carrié 1986: 475). In fact the military diplomas only inform us that a proportion of the auxiliaries and fleet soldiers at discharge – thus close to age 50 – had wives and children; since the ages of children are not given, the diplomas do not show when the soldiers married.

The Latin epitaph is our main source of statistical evidence for Roman family relations below the aristocracy (Saller & Shaw 1984; Shaw 1991). The Latin epitaph often records the name of the deceased, his or her status (senator, artisan, freedman), origin, career, age at death, and relationship to the commemorator, the person who dedicates the tombstone. Proper burial and commemoration were a heartfelt duty in Roman society, usually carried out by the heir of the deceased person; the commemorator may be assumed to be a major social affiliate (family or friend) of the deceased. Patterns of commemoration can be tabulated from hundreds of these epitaphs. Thus, since we know almost nothing about the Roman military family from literary authors, studies of the Roman military family are based on these patterns of commemoration (Saller & Shaw 1984; Roxan 1991; Shaw 1991).

To know the soldiers’ usual age at first marriage would shed light on the demographics of the Roman military community and the rôle of marriage within it.4 Early and universal marriage is a given only in C20th western societies; in C16th-C18th north-western Europe, both sexes married relatively late (late 20s), and hence we should not import present ideas about the timing of marriage into past societies (Hajnal 1965). Furthermore, the military society may have differed from Roman civilians. Our study suggests that Roman soldiers most

Saller and Shaw’s (1984) study observed regional and occupational differences in the frequency of commemoration of the deceased by nuclear family members (parents, siblings, children), as opposed to nonkin; groups rarely commemorated by (or commemorating) family are assumed usually not to form families. Saller and Shaw found that the military population overall showed less family commemoration than civilians, and argued that the military population showed regional differences in family formation; soldiers in Africa frequently ‘married’ and had children, while soldiers in Britain, Gaul and the Germanies did so the least (Cherry 1989). This conclusion is flawed in that Saller and Shaw did not date the military epitaphs but assumed that all derived from the C2nd and C3rd AD.5 Though Latin epigraphy in general shows a peak in the late C2nd and early C3rd, in fact some provinces military epitaphs date mostly from the C1st AD, as in the Germanies; in other provinces, such as Africa, the military epitaphs date mostly from the C2nd and C3rd.

1

This study is adapted from a chapter in my monograph The marriage of Roman soldiers, 13 BC – AD 235: law and family in the Roman army (hereafter MRS). The legal aspects are treated in full in MRS: cf. Campbell 1978; Jung 1982. 2 Throughout this paper, ‘marriage’ will be used in the de facto sense of a long-term stable union with a woman; legal marriage will be denoted as such. 3 Age at enlistment is estimated from inscriptions; no clear statement exists in the literary authors. Forni 1953: 27 suggest an average age of 1823, refined statistically (correcting for age-rounding) by Scheidel 1996: 97-116. Scheidel argues that enlistment at age 18-20 was most common for legionaries. 4 ‘Community’ here includes soldiers and their non-combatant relatives and dependents: cf. Goldsworthy & Haynes 1999 for various ideas on what constituted a military community.

5

“…the inscriptions from the provincial areas are concentrated in the period from the mid-second to the first quarter of the third century” Saller & Shaw 1984: 131.

873

Limes XVIII

Thus Alföldy and Roxan have suggested that the Roman military showed more temporal difference in family formation than regional difference; C1st AD soldiers rarely formed conjugal families, while in the C2nd and C3rd the marriage rate increased (Alföldy 1987: 33, n.27; Roxan 1991).6 My survey in MRS confirms this. As a result, this study is based on inscription populations that date from the C2nd and C3rd AD, when the soldiers’ marriage rate was high enough to make a survey worthwhile.7

commemorated by wives, and almost none by children; it is logical that in these years few soldiers were married or had children. Commemoration by wives began to rise around age 30 and rises steadily through the 30 to 50 year old age range; that is, soldiers, who died in their 30s through 50s were increasingly commemorated by wives. The graph of commemoration by the birth family crosses with that of commemoration by wives around age 37, suggesting an average age at which soldiers entered their de facto marriages.12 Commemoration by children begins to rise for soldiers who die in their 30s, but only increases markedly for men who die after age 50, supporting the hypothesis of soldiers’ marriage in their mid- to late 30s, since children of such marriages would only be mature by the soldiers’ or veterans’ 50s. Figure 2 depicts the same population but plots soldier’s age at death against percent of epitaphs in each age bracket. Thus commemoration by the birth family is almost total, near 100%, in the soldiers 20s, while commemoration by wives and children is minimal in this period. The graphs of commemoration by the birth family and commemoration by wives cross (and are near 50% of all epitaphs in that age range) in the soldier’s late 30s.

Civilian men and women’s ages at marriage has been estimated from the epitaphs. Here too advances in method have occurred. Keith Hopkins (1965) collected epitaphs of women that gave both length of marriage and age of death; from these he suggested that Roman women married in their early teens. However, soldiers and their wives used this epitaph type both rarely and imperfectly.8 Subsequent studies have estimated age at marriage from shifts in commemoration throughout the life course (Shaw 1987; Saller 1987; 1994: 25-41). For example, a child who dies is usually commemorated by his or her parents; when a majority of subjects dying at a certain age are commemorated by their husband or wife, as opposed to parents or siblings, this gives our estimate of the age at which marriage usually occurs. These studies can use much larger numbers of epitaphs than the length-ofmarriage epitaph type. Thus Shaw has estimated women’s age at marriage to usually be in the late teens; the lengthof-marriage epitaphs seem to record unusually early marriages. Saller estimates civilian men’s age at marriage to be in the late 20s and states that “…all soldiers and those suspected to be of servile origins have been excluded on the grounds that service in the army (often lasting for decades) and slavery must have disrupted normal patterns of family life” (Saller 1987: 23).9

The general survey thus suggests that soldiers usually married in their late 30s. This is supported by regional surveys, which however show slight regional variations in the estimated age at marriage. Figure 3 comprises epitaphs from CIL VIII (Africa), chiefly legionaries. The picture is similar to the aggregate one (Figs. 1 and 2) in form, but the birth family and wife graphs cross at a slightly later age, early 40s, though the marriage rate begins to rise sharply in the late 30s. This suggests that soldiers in Roman Africa married at a slightly later age. Figure 4, based on military epitaphs from the Danubian provinces13, suggests that soldiers there married at a slightly earlier age than in general, the mid-30s. Figure 5 presents the graph of commemoration by wives for the aggregate survey, Africa, and the Danube, showing the slight differences in the curves and showing that they reach the 50% mark for slightly different age groups.

Figure 1 is based on a survey of military epitaphs in North Africa, the Danubian provinces, and the Praetorian Guard.10 All date from the C2nd/C3rd AD. All epitaphs are commemorations of soldiers by family members, commemoration by non-kin such as comrades (or rarely slaves) are numerous in all age groups and have been omitted for clarity.11 It is apparent that most soldiers who died in their 20s were commemorated by their birth families (parents or siblings). Few in this age range were

Figure 6 is based on epitaphs on fleet soldiers at Misenum and Ravenna; these epitaphs could not be dated by troop displacements, though it is thought that most are from the C2nd AD. The graph may thus be less accurate. But it shows a distinct variation from the marriage patterns of the land forces; commemoration of fleet soldiers by wives increases at a much earlier age and suggests that fleet soldiers ‘married’ younger than soldiers elsewhere.

6

The temporal pattern is observed for the Praetorians by Panciera 1993. Military epitaphs may be dated by unit displacements; legions - Mann 1983: 160; auxilia – Kraft 1951, updated with CIL XVI suppl and RMD IIII series. The changes in epitaph formulae are summarised by Calabi Limentani 1991; Le Bohec 1989: 64-65. 8 eg. CIL VI. 2470; 2482; 2496. Length of marriage may be given, but no age at death or vice versa. That this type of epitaph was unpopular with soldiers’ families supports our conclusion: their marriages tended to be short because the men married late. 9 Senatorial men married at a slightly younger age, in the early 20s: Saller 1987: 29-30; 1994: 39. 10 Sources of epitaphs are given in full in MRS Ch. 6. CIL and AE are the main sources. 11 The great majority of military epitaphs are commemorations of soldiers by other persons. Soldiers’ commemorations of family members (wives, children) form a much smaller population in which the soldiers are almost always officers. 7

In general the military epitaphs’ commemoration patterns strongly suggest that soldiers tended to marry relatively late, in their mid-to-late 30s, as much as a decade later than 12

Saller 1987: 23 warns that “… no precise mean or mode can be calculated”. However, what we require is a demographic ‘impression’, the term used by Parkin 1992. 13 The population includes both legionaries and auxiliaries; no difference in their family formation is apparent in this time period.

874

Sara E. Phang: The timing of marriage in the Roman army civilian men. A possible criticism of the commemorationshift method has been made by J.K. Evans, suggesting that the shift overestimates age at marriage, since a time might pass between the actual (earlier) onset of the marriage and the period when it was considered worthy of record (Evans 1991: 205, n.53). Saller considers this improbable in the mass (Saller 1994: 39). It is possible that since soldiers’ unions were not legal marriage, they took longer to obtain social ‘recognition’, but the very late appearance (in the soldiers’ 50s and 60s) of commemoration by children supports the hypothesis of actual marriage in the soldiers’ late 30s.

also in soldiers who were not promoted; possibly they accumulated savings to start a household, as such saving was a common motive for late marriage in C16th-C18th western Europe (Stone 1977: 46ff.). It is possible that poorer soldiers’ marriages were delayed because they had to support their birth families such as aged parents; Saller suggests that Roman parents tended to die by their sons’ mid-30s (Saller 1974: 51 – computer simulation). Roman soldiers late marriage suggests an army in which marriage, so far from being early and universal, was a privilege for older soldiers. Accounts of the soldiers’ marriage policy often describe a ‘civilianisation’ of the army, married soldiers being less effective as soldiers (eg. Parker 1928; Watson 1969: 133; Pflaum 1969); this judgement is based more on prejudice than reality.17 This study of soldiers’ age at marriage suggests that, since younger men (the demographic majority) were unattached, the Roman army was not ‘immobilised’ by soldiers’ marriages even in the C2nd and C3rd, and the degree of assimilation with the local population was limited even in this period.

It is probable that soldiers did indeed marry in their mid- to late 30s. This datum has interesting implications. The first, from the viewpoint of the family members, is that many soldiers’ wives would be widowed and their children fatherless at a relatively early age, due to an age gap as much as 20 years between husband and wife. The result is an exaggeration of the legal and social consequences analysed for civilian Roman families by Saller and Krause (Saller 1994: 161-180, 181-203; Krause 1994). In addition, if the father died while still in service, while the marriage was still illegitimate, his widow might not be able to recover her dowry14; if the father left no will, his children might not be able to inherit his property because they were illegitimate.15 The wife’s family may not have perceived a possible strategic advantage in marriage to an older soldier; a shorter time would pass between the onset of the union and its legalisation.

Bibliography Alföldy G. 1987 Das Heer in der Sozialstruktur des römischen Kaiserreiches. In G. Alföldy (ed.) Römische Heersgeschichte: Beiträge 1962-1985. (Amsterdam): 26-42 (revised and reprinted in G. Alföldy, B. Dobson & W. Eck (edd.) Kaiser, Heer und Gesellschaft in der römischen Kaiserzeit: Gedenkschrift für Eric Birley. (Stuttgart 2000): 33-58). Breeze D.J. 1971 Pay grades and ranks below the centurionate. Journal of Roman Studies 61: 130-135. Breeze D.J. 1974 The organisation of the career structure of the immunes and principales of the Roman army. Bonner Jahrbucher 174: 245-292 (republished in Roman officers and frontiers: 1993: 11-58). Calabi Limentani I. 1991 Epigrafia latina. (4th ed. Milan). Campbell J.B. 1978 The marriage of soldiers under the empire. Journal of Roman Studies 68: 153-166. Carrié J.-M. 1986 L’esercito: trasformazioni funzionali ed economie locali. In A. Giardina (ed.) Società romana e impero tardoantico: istituzioni, ceti, economie. (Roma): 449-488. Cherry D. 1989 Soldiers’ marriages and recruitment in Upper Germany and Numidia. Ancient History Bulletin 3: 128-130. Dobson B. 1972 The significance of the centurion and ‘primipilaris in the Roman army and administration. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.1. (Berlin): 392-434 (republished in Roman officers and frontiers: 1993: 143-185). Evans J.K. 1991 War, women and children in ancient Rome. (London).

What feature of the military society promoted later marriage ? It is possible that younger soldiers’ relationships with women were discouraged by the Roman army, and older soldiers were permitted to form families, even though they were not permitted legal marriage till discharge, as the diplomas conferring conubium show in the case of the auxilia, fleet and Praetorians. Permission to (legally) marry after a decade or so of service was given in some C19th European armies. However, there is no literary evidence on this point for the Roman army. Another possibility is that Roman soldiers’ late marriage is related to promotion. Officers of centurion rank and below were promoted on a fairly regular track, 10 to 15 years after enlistment (Dobson 1972: Breeze 1993). They enjoyed pay increases. The centurion, principalis or immunis had both prestige and wealth (Breeze 1971) and thus was a desirable husband.16 But later marriage is found 14 Pap. Cattaoui I.5-13; I.14-III.10; VI.18-23. This damaged her chances of remarriage. These legal problems are covered extensively in MRS. 15 Illegimate children bore no legal relationship to their biological father and hence could not claim part of his estate if he died intestate. Hadrian ameliorated this for children in an edict of 119 (BGU 140), granting soldiers’ illegitimate children a right of inheritance unde cognati (thus subordinate to any legitimate children) on the father’s intestacy. Illegitimate children were not excluded from testate inheritance: cf. Rawson 1989. 16 Whether the legal marriage ban applied to centurions is discussed in MRS forthc. Most authors believe that centurions were permitted legal marriage.

17 The prejudice is apparent in Herodian III.8.4-5; Libanius Or. 2.39-40. The permission of legal marriage by Septimius Severus in 197 has been used to construct a narrative of ‘civilianisation’ of the Roman military, which became largely hereditary and deteriorated into the peasant-militia of the limitanei. Isaac 1988 has questioned the existence of this peasantmilitia; in MRS I question whether the Roman army can ever have relied on soldiers’ sons as recruits.

875

Limes XVIII

Forni G. 1953 Il reclutamento delle legioni da Augusto a Diocleziano. (Rome). Goldsworthy A. & Haynes I. (edd.) 1999 The Roman army as a community. Journal of Roman Archaeology Supplementary Series 34. Hajnal J. 1965 European marriage patterns in perspective. In D.V. Glass & D. Everseley (edd.) Population in history. (Chicago): 101-141. Hopkins K. 1965 The age of Roman girls at marriage. Population Studies 18: 309-327. Isaac B. 1988 The meaning of the terms limes and limitanei. Journal of Roman Studies 78: 125-147. Jung J.H. 1982 Das Eherecht der römischen Soldaten. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II.14. (Berlin): 302-346. Kraft K. 1951 Zur Rekrutierung der Alen und Kohorten an Rhein und Donau. (Bern). Krause J.-U. 1994 Witwen und Waisen im Römisches Reich I: Verwitwung und Wiederverheiratung. (Stuttgart). Le Bohec Y. 1989 Le troisième légion Auguste. (Paris). Luttwak E.N. 1976 The grand strategy of the Roman empire; from the first century AD to the third. (Baltimore). Mann J.C. 1983 Legionary recruitment and veteran settlement during the Principate. (London). Panciera S. 1993 Soldati e civili a Roma nei primi trei secoli dell’impero. In W. Eck (ed.) Prosopographie und Sozialgeschichte: Studien zur Methodik und Erkenntnismöglichkeit der kaiserzeitlichen Prosopographie. (Wien): 261-276. Parker H.M.D. 1928 The Roman legions. (Cambridge). Parkin T. 1992 Demography and Roman society. (Baltimore). Pflaum H.-G. 1969 Forces et faiblesses de l’armée romaine du Haut-Empire. In J.-P. Brisson (ed.) Problèmes de la guerre à Rome. (Paris): 85-98. Rawson B. 1989 Spurii and the Roman view of illegitimacy. Antichthon 23: 10-41. Roxan M. 1986 Observations on the reasons for changes in formula in diplomas circa AD 140. In W. Eck & H. Wolff (edd.) Heer und Integrationspolitik: Die römische Militärdiplome als historiche Quelle. (Koln). Roxan M.M. 1991 Women on the frontiers. In V.A. Maxfield & M.J. Dobson (edd.) Roman Frontier Studies 1989. Proceedings of the XVth International Congress of Roman Frontier Studies. (Exeter): 462-467. Saller R.P. 1987 Men’s age at marriage and its consequences in the Roman family. Classical Philology 82: 21-34. Saller R.P. 1994 Patriarchy, property and death in the Roman family. (Cambridge). Saller R.P. & Shaw B.D. 1984 Tombstones and Roman family relations in the Principate: civilians, soldiers and slaves. Journal of Roman Studies 74: 124-155. Scheidel W. 1996 The demography of the Roman imperial army. In W. Scheidel Measuring sex, age and death in the Roman empire; explorations in ancient demography. (Ann Arbor): 93-118. Shaw B.D. 1987 The age of Roman girls at marriage: some reconsiderations. Journal of Roman Studies 77: 30-46. Shaw B.D. 1991 The cultural meaning of death; age and gender in the Roman family. In D. Kertzer & R. Saller (edd.) The family in Italy from antiquity to the present. (New Haven): 66-90. Stone L. 1977 The family, sex and marriage in England 15001800. (London). Watson G.R. 1969 The Roman soldier. (London).

Wolff H. 1974 Zu den Bürgerrechtsverleihungen an Kinder von Auxiliaren und Legionaren. Chiron 4: 479-510.

876

Sara E. Phang: The timing of marriage in the Roman army

Fig. 2. Military epitaphs from wives; comparison by region.

Fig. 1. Military familial epitaphs: all regions.

Fig. 3. Military familial epitaphs: Africa (n = 247).

Fig. 4. Military familial epitaphs: Danube (n = 240).

877

Limes XVIII

Fig. 5. Military familial epitaphs: all regions.

Fig. 6. Military familial epitaphs: Fleet (n = 131).

878

The Roman auxilia in the east – different from the west ?1 D.B. Saddington Auxiliaries come in for only incidental and brief comment in the historians of the early principate. Their few remarks can be supplemented by what can be deduced from the military diplomas issued to auxiliaries in conferment of Roman citizenship after 25 years of service. But these only start appearing under Claudius and Nero, and refer overwhelmingly to the armies of the west. Tombstones erected to deceased auxiliaries provide further information, but again they are rare in the east. Nevertheless such information as there is suggests that both the composition of, and the deployment of auxiliaries in the east followed a different pattern from the better-known position on the frontiers of the north-west provinces. In some cases the eastern auxilia appear to approximate more closely to the legions than to the peregrine-type requirements known from elsewhere. The anomalies do not seem to have been due to chance, but to a specific response to a particular situation. The available material will be analysed in an attempt to isolate specific features of the eastern auxilia.

There are small differences between the funerary monuments of auxiliaries in the east as compared with those of the west. To take an example, a Gaul in an ala Macedonica was buried in Thessalonica (which is in Macedonia): Catto Bellouaci f. domo Segusiauo equiti alae Macednicae ex testamento. T. Cornelius T. f. Fronto faciundum qurauit idemque probauit (ILGR 219). Elements commonly found on western stelae but missing here are trooper’s age and number of stipendia. His name and that of his father are Celtic, but the person who erected the tombstone has a full Roman name and was in all likelihood a Roman citizen, although he does not name his tribus or Roman voting constituency. The ethnic composition of the ala cannot be established. ‘Macedonica’ may refer to an association with the legio V Macedonica or, more likely, to being stationed in Macedonia (or both, since V Macedonica was in Macedonia from the Battle of Actium until AD6, when it went to Moesia, at the time not yet a formal province but probably still part of Macedonia; RE XII: 1573). The auxiliary was a Gaul, from the Segusiavians: the Roman colony of Lugdunum (Lyons) was in their territory, and it was presumably from it that he was recruited.

regiments as well as local forces from the outset. A bilingual inscription from Side (Selimiye) in Pamphylia records ‘[...coho]rte Apula Λεύκιoς Σάλβιoς Λευκίoυ υείòς Σεργία,’ (AE 1966: 478) the regiment being given in Latin, the name in Greek, with the filiation and tribus of a Roman citizen. The coh Apula carried out public works in the colony of Alexandria Troas (Eskistanbul) under the prefect Fabricius Tuscus (PME F 18; Saddington 1980: 11) on the orders of Augustus (AE 1973: 501). Alexandria Troas is a long way from Side. Was the auxiliary involved in the Galatian wars north of Side before the regiment went to the colony? But what is of interest is that a full Roman citizen was serving in an auxiliary regiment in the east (and that, like a legion, it was used for what today would be considered as non-military purposes). Furthermore the Apula belongs to a series of units called Bürgerkohorten (Kraft 1951: 82ff.; cf. Saddington 1982: 141ff.) or citizen cohorts, auxiliary regiments comprised not of peregrine provincials but Roman citizens. The Apula was originally recruited in Apulia (Puglia). Another such regiment serving in the east - it is recorded at Amastris (Amasra) in Paphlagonia - was the coh. Campana (originally from Campania). Again the auxiliary in the cohort had a Roman name, L. Sempronius, and was in all likelihood a citizen, like his centurion, L. Allidius (AE 1993: 1429; Marek 1993: 171, no. 52; Solin 1994: 113). The Ligurians in Cisalpine Gaul seem to have been treated as Italians in the auxiliary system (Saddington 1980: 146). A coh I Ligurum was sent to the east: it is attested by an epitaph of its commander in Ephesus: ‘L. Pompeio L. filio Fabia Marcellino Roma tri. coh. pri. Liguri uixit annos XXIII.3 ‘Liguri’ is presumably an error for ‘Ligurum’, and ‘pri.’ can only be ‘primae’. Pompeius Marcellinus specifies his tribus. What is of greater interest, apart from

A cavalryman in the ala Bosporanorum is recorded on the Euphrates near Europus (Salihiya; ILS 2510). Again his age and stipendia are not given. He was a Nantuatian, of a people between the Pennine Alps and Lake Geneva. He may have been recruited in the Pennine Alps - there were many Alpini in the Roman auxilia - or, more likely, from the colony of (Julia Equestris) Noviodunum (Nyon) across the lake. He bore the Roman name of Scaurus, but his father’s name was Celtic. A prefect of his regiment under Claudius is known, M. Licinius Rufus (PME L 16; Saddington 1980: 65); interestingly, he described it as ‘ala Bosphorana exe[rcitus qui est] in Syria, an ala of the army in Syria (AE 1969/70: 649, 653). That a Gaul was serving in a Bosporan regiment (of which its Claudian commander was proud) on the distant Euphrates far from both his and the regiment’s original home, shows that it (like the ala Macedonica) was completely professionalized by this date.2 Thus the east was being controlled by professional

auxiliary regiments in the east’ delivered at the VIIIth International Congress of Greek and Latin Epigraphy held in Athens in 1982. 2 For the concept of ‘professional’ auxiliary regiments in the early principate, cf. Saddington 1982: 2, cf. 154, 193. 3 IEph 2305B; PME P 60. Another tribune is known to have commanded the regiment, L. Faianius Sabinus (CIL XI 838; PME F 22; Saddington 1980: 33). For the probable stationing of the regiment in the east, cf. Pflaum 1971: 62.

1

This article incorporates most of the material in an as yet unpublished paper on ‘Roman auxiliary inscriptions and some special features of the

879

Limes XVIII

his coming from Rome itself, is his youth: most auxiliary commanders were in their mid-thirties (Birley 1949 (= 1976)), and that his title was not prefect but tribune. The rank of tribune was normally assigned to commanders of ‘citizen cohorts’, milliary regiments and sometimes apparently as an honour (Saddington 1982: 62, 214 n. 49). Marcellinus must have been regarded as commanding a citizen unit.

was in a vexillation of archers. If Claudius Lysias’ regiment was it, it may have been doubled in size under Nero as the security situation in Judaea deteriorated. A coh civium Romanorum is known in the east, the coh V Gemella civium Romanorum (IEph. 4112) whose commander, Flavius Juncus (PME F 52bis), who had the rank of tribunus, came from Judaea where he served under Trajan.8

Cohortes Italicae as such are also found in the east. Such was the ‘Italian band’, the σπειρα ‘Iταλική, of the New Testament. It was probably that known in AD69 as the coh II Italica. The inscription recording it reads as follows: ‘Proculus Rabili f. Col. Philadel. mil. optio coh. II Ital. c. R. 7 Faustini ex uexil. sagit. exer. Syriaci’. Proculus, in spite of his peregrine filiation, was probably a citizen before his enlistment.4 It is also possible that the coh [-] Italica commanded by T. Volumnius Varro (PME V 130) of Antioch towards Pisidia (Yalvaç) in the ClaudioNeronian period is the same regiment.5 If so, the regiment was entitled coh II Italica under Claudius and coh II Italica c.R. in the year of the four emperors, as it is on a Syrian diploma of 88 (CIL XVI 35). One of its commanders in the C2nd, L. Maesius Rufus (PME M 7 + Suppl. II), styled himself trib(unus) coh. mil(liariae) Italic. volunt(ariorum) quae est in Syria, omitting its numeral of II but specifying it as milliary and Voluntariorum: his rank of tribune suits both a milliary and a citizen cohort.6 Volumnius Varro only had the title of prefect, which suggests that it was only quingenary under Claudius. One wonders whether it was the regiment commanded by the Claudius Lysias (PME C 154; Saddington 1980: 74) who arrested Paul in Jerusalem in 57. As already noted, it was in Caesarea in Judaea two decades or so earlier. The size of the detachment (if it came from this single regiment) which Claudius Lysias, notably styled as χιλίαρχoς or tribune, detailed to take Paul from Jerusalem to Caesarea certainly suggests a milliary unit. It may even have included archers (the enigmatic δεξιoλάβoι of Acts), hence a cohors equitata sagittaria7: it may be noted that Proculus above

As P.A. Holder9 has argued, both Vespasian and Trajan seem to have created new citizen regiments in the Syrian area, the coh I Flavia c.R. (CIL XVI 35 of 88) and the coh II Ulpia c.R. (CIL XVI 106 of 156/7). The cohortes Classicae were included in the citizen regiments (cf. Saddington 1980: 142). There was one in Syria by AD6, the coh II Classica (ILS 2683) commanded by Q. Aemilius Secundus (PME A 90; Saddington 1980: 10) who used it in the census of Apamea (Qalaat elMoudiq) in Syria which he conducted. It was still in Syria in 88 (CIL XVI 35). The other regiment commanded by Aemilius Secundus was the coh Augusta I. Not to have an ethnicum and to have been named after the emperor in this early date argue for quality. The σπε²ρα Σεβαστή (Ac. 27.1) of the New Testament, in Judaea under Nero, may be compared as well as the σπε²ρα Αõγoυστα attested in the Hauran by a centurion with the Roman name of L. Obulnius (AE 1925, 121) and at Eitha (el-Hit) in southern Syria under Agrippa II (IGRR III 1136). Two eastern regiments were named by their weapons alone. Under Trajan, Pliny the Younger (Epist. X 106f.) handled a petition on behalf of a centurion of a coh VI Equestris in Bithynia. Possibly under Hadrian if not under Trajan there was a coh II Equitum from Syria which Valerius Lollianus included in a vexillation he led to Mesopotamia.10 Less needs to be said about the alae. As noted, the ala Bosporanorum, ‘professionalized’ early, was stationed on the Euphrates by the time of Claudius. Regiments named after their commanders seem to have had a high reputation from the outset (cf. Saddington 1982: 147ff.). The ala Antiana was in Syria also under Claudius: it was a composite force of Gauls and Thracians (CIL XVI 3) which points to a specially integrated unit. There was an ala Agrippiana in the west. It is, however, not clear

4

Ac. 10, 1: Saddington 1982: 51; ILS 9168; for Proculus’ citizenship, cf. Forni 1974: 347, accepting the expansion of Col. to Collina tribu, thereby making Proculus a citizen in his own right. This is more probable than ascribing his citizenship to his rank: although centurions of auxiliary units were sometimes citizens, those of lower ranks were usually not. Cf. Forni 1979: 212f. (= 1992: 187). 5 AE 1941: 142, where the lacuna is supplied as coh. [c.R.] Italica. However, as Kraft 1951: 195, has argued that emendation is unsatisfactory and a numeral such as II (coh. II Italica) is more appropriate. Volumnius Varro should be added to Saddington 1980: 40, as no. 88a. 6 CIL XI 6117; for voluntarii in the Roman army cf. Saddington 1982: 179. PME II (s.v. M 7) adopts a date under M. Aurelius and L. Verus for Maesius Rufus. But Augustorum in his inscription need not necessarily refer to joint emperors: the word can be used of successive emperors. In fact PIR2 M79 has suggested that, on military grounds, Maesius Rufus may be placed under the Flavians, if not even under Domitius Corbulo. 7 Ac. 21, 31; 23, 26; 24, 22; for χιλίαρχoς cf. Saddington 1982: 51, 210 nn. 109, 113 on the composition of the detachment, (23, 23) cf. Saddington 1982: 101, 210 n. 113; 1996: 2416; for Paul, RE Suppl. VIII: 431, no. 11. Barrett 1998 ad loc. (p.1018f.) on δεξιoλάβoι, however, draws attention to Jerome’s translation of the word as lancearii (spearmen): this recalls the personal guards (pedites singulares) of Roman

army commanders: Josephus (Bell. Jud. III 95) says they were armed with the λόγχη or lancea (RE XII: 618). 8 cf. the discussion in Eck 1999: 70ff. It may be noted that in the west in the small province of the Alpes Graiae there was a coh. I c. R.: the soldier attesting it (AE 1955: 113) was a citizen (he gives his tribus). 9 Holder 1980: 67. However, it should be noted that the element c.R. does not appear on the diploma: but it is found on inscriptions (eg. ILS 2724). Holder (p.68) also suggests a coh. I Ulpia c. R., likely indeed on numerical grounds. But it should be noted that the lacuna in CIL XVI 70 of 124, corrected to I VLP. C. R. by Wright 1964: 150, has now been filled to read ?I Celtiberor(um) (by Roxan RIB: 2401.6). 10 ILS 2724 + add.; for Valerius Lollianus, PME V 17; for discussion of the date, Kennedy 1997: 72ff.

880

D.B. Saddington: The Roman auxilia in the east – different from the west? whether the ignotus who died on returning from Germany in an ala Agrippiana at Eitha (IGRR III 1140 - άπÄ Γερµαvίας ’αvελθ´v καà ’εv εϊλη ’Αγριππιαv± απεθαvώv) belonged to this regiment, or whether it was the ala Agrippiana which accompanied Valerius Lollianus to Mesopotamia. An ala Flavia Agrippiana is attested by tiles at Tille, near Samosata (Samsat) on the Euphrates.11 Its relation to the Agrippiana attested on the Rhine early under Tiberius (AE 1992: 186; cf. Saddington 1994: 73ff.) or that in Britain (also entitled Miniata - CIL XVI 69 of 122) is unclear. The history of the ala Herculiana is also difficult to reconstruct. It first appears in the east in the cursus of C. Sappius Flavus who held the positions of praef. alae Thracum Herculaniae praef. ripae fluminis Euphratis under Trajan (ILS 2709; PME S 8). It remained in Syria accompanying Valerius Lollianus on his expedition.

One can concentrate too much on the legion as the basic unit of the Roman army: the small separate cohort must not be left out of account. One has to be cautious in view of the extreme paucity of the evidence. But it would appear that in the east there were several auxiliary regiments composed of men with Roman citizenship or of men recruited in Italy. This may point to the special deployment in the east of cohorts of men of similar status to legionaries, notable in such areas which were remote from the main concentration of the legions towards the Euphrates. Not only was the proportion of citizen regiments high in the east, but the number of citizens in apparently peregrine units was high.13 The interest of the Romans in citizen units for the east can also be deduced from an incident recorded by Tacitus (Hist. III 47): when the troops of the king of Pontus were incorporated into the Roman army they were given not only Roman standards and armour, but Roman citizenship as well. Not that this need cause surprise. Julius Caesar had formed a legion, the V Alaudae, in Transalpine Gaul; many of Antony’s recruits to his legions in the east were recently enfranchised locals and part of the army of the client kingdom of Galatia became the legio XXII Deiotariana.14

The ala Praetoria Singularium appears on a Syrian diploma of 88 (RMD 3). It has the title of Flavia on a diploma of 91 (RMD 4) but is called simply ‘Praetoria’ on the Lollianus inscription. Its relation to the ala Praetoria of the west is not clear (AE 92: 186; Saddington 1994: 73ff.). It was also commanded by Fabricius Tuscus (for whom cf. above). E. Birley (1978: 267) suggested that the ala Germaniciana was named after Germanicus, who was in the east in 18. It set up a dedication (AE 1914: 128; 1966: 472) to a dignitary in Antioch towards Pisidia (Yalvaç), possibly its commander, C. Julius Proculus (PME I 101), under Nero and with the added element of Augusta it is recorded (AE 1973: 539) in the career of a prefect, M. Sempronius Albanus (PME S 17; Saddington 1980: 137) at Attaleia (near Antalya) early under Vespasian.

Naturally normal or ‘standard’ peregrine regiments were stationed in the east. But it would appear that, with typical flexibility, the Romans were prepared to use citizens, and, where there were in short supply, to create citizens, to form smaller units than legions, either 500 or 1000 strong15, to bear responsibility for possible difficult conditions where there were insufficient legionaries to be stationed. In some areas special fighting methods, such as archery, might prove an added advantage.

Under Agrippa II an Herodes the son of Aumos (RE Suppl. II 166, no. 22; Saddington 1980: 124) described himself as στρατoπεδαρχήσας ίππέωv κoλωvειτ´v (IGRR III 1144 at Saura (Sur) in Syria. Stratopedarches usually represents praefectus castrorum but perhaps simply means ‘commander’ here. Herodes - the name is peregrine commanded cavalry who were from a colony, presumably one of those which Herod the Great had settled in the north of his territory, such as from Gaba (?Jaba; Josephus Bell. Jud. III. 36; RE VII: 408, no. 8), known as a city of cavalrymen. The relation of this force to the ala Colonorum recorded in Syria and Cappadocia is not clear. Two decurions in it are known, both with the tria nomina of Roman citizens and conceivably recruited from a colony.12 The ala is called ala Gemina Colonorum (AE 1926: 150) by one of its prefects, M. Sentius Proculus (PME S 25), in the Flavian period and ala I Augusta Gemina Colonorum (ILS 1400), when it served under Sex. Cornelius Dexter (PME C 234) in Judaea under Hadrian.

Eastern auxilia - Testimonia ILGR 219 Catto Bellouaci f. domo Segusiano equiti alae Macednicae ... T. Cornelius T. f. Fronto faciundum qurauit ... ILS 2510 Scaurus Ambitouti f. domo Nantuas eques ala Bosphoranorum ... AE 69/70: 649, 652f. M. Licinius Rufus praefectus alae Bosphoranae exer[citus] qui est in Syria ... AE 1966: 478 [--- coho]rte Apula Λεύκιoς Σάλβιoς Λευκίoυ ύειÄς Σεργία AE 1973: 501 C. Fabricio ... Tusco ... praef. cohort. Apulae et operum quae in colonia iussu Augusti facta sunt ... praet. equit. alae praet. ... AE 1993: 1429 L. Senpronio militi et tubicini cohortis Camp. 7 L. Allidi ... 13 Saddington 2000: 172; for a full discussion of Roman citizens in the diplomas, Dušanić 1996: 31ff. 14 Saddington 1982: 25; for Antony’s eastern legionaries, Cuntz 1929: 70ff. 15 For the early appearance of milliary regiments in the east cf. Saddington 1982: 174; Kennedy 1983: 253ff. Weber 1921: 119, was even prepared to entertain milliary alae as early as Nero.

11

Crow & French 1980: 905, who usefully discuss the alae entitled Agrippiana in the east. 12 MAMA VIII 94, 327. They both have the nomen of Aponius. (Aponii are known in Lycaonia). The inscriptions were found near Iconium in Lycaonia, an Augustan colony. For Iconium and the Aponii, cf. Mitchell 1993: 74, 77, 151.

881

Limes XVIII

IEph. 2305B L. Pompeio L. filio Fabia Marcellino Roma tri. coh. pri. Liguri ... ILS 9168 Proculus Rabili f. Col. Philadel. mil. optio coh. II Italic. c. R. 7 Faustini ex uexil. exer. Syriaci ... AE 1941: 142 T. Volumni Varronis ... praef. coh. [II] Italic. CIL XI 6117 Maesio Rufo ... trib. coh. mil. Italic. uolunt. quae est in Syria ... IEph 4112 [T. Flauium] Iuncum ... trib. cohortis V gemellae ciuium Romanorum ... ILS 2683 Q. Aemilius ... Secundus ... praefect. cohort. Aug. I praefect. coh. II classicae; idem iussu Quirini censum egi Apamenae ciuitatis ... AE 1925: 121 Λoύκιoς Όβoύλvιoς έκατovτάρχης σπίρης ’Αυγoύστης ... IGRR III 1136 Χάρητoςε’ ‘πα[ρχoς ---] σπείρης Αυ[γoύστης ... IGRR III 40 ... ’απÄ Γερµαvίας άvελθÆv και έv ‘εγλ› Άγριππιαv απoθαvώv ... Limes XII 1979 (1980) 905 Al·FL·AG ILS 2709 Sappio ... Flauo ... tribun. militum leg. XXI Rapacis praef. alae Thracum Herculaniae praef. ripae fluminis Euphratis ... AE 1992: 186 P. Cornelio P. f. Scipioni ... alae V ... Agrippiana ... Praetoria AE 1914: 128; 66, 472 C. Iulio ... Proculo ... ala Aug. Germanica ... AE 1973: 539 Μ°ρκoς Σεπρώvιoς ’ΑλβανÄς έπαρχoς ’ιππέωv ’ιλης Σεβαστ²ς ΓερµαvικιαvÂς ... IGRR III 1144 Ήρώδη Αύµoυ στρατoπεδαρχήσαvτι ι‘ππέωv κoλωvειτ´v ... MAMA VIII: 94, 327 Γ. Άπώvιov Λόγγov ... στρατευσάµεvov δεκαδάρχηv ειλης Κoλώvωv ... Γαίoς Άπώvιoς Φίρµoς στρατευσάµεvoς δεκαδάρχης εϊλης Κoλώvωv ... AE 1926: 150 M. Sentio ... Proculo ... praef. alae gem. col. ... ILS 1440 Sex. Cornelio ... Dextro ... donis militaribus a diuo Hadriano ob bellum Iudaicum ... praef. alae I Aug. gem. colonorum ...

nei primi tre secoli dell’impero. In H. Temporini (ed.) Aufstieg und Niedergang der Römischen Welt II. 1. (Berlin): 339-391. Forni G. 1992 L’anagrafia del soldato e del veterano. Actes du VIIe congrès international d’épigraphie, Constanţa 1977 (Bucaresti, 1979): 212f. (= Esercito e marina di Roma antica. Mavors 5 (Stuttgart 1992): 187ff). Holder P.A. 1980 Studies in the auxilia of the Roman army from Augustus to Trajan. British Archaeological Report Int. Ser.70. (Oxford). Kennedy D. 1983 Milliary cohorts: the evidence of Josephu BJ. III.4.2 (67) and of epigraphy. Zeitschrift für Papyrologie und Epigraphik 50: 213-263. Kennedy D. 1997 The special command of M. Valerius Lollianus. Electrum I: 69-81. Kraft K. 1951 Zur Rekrutierung der Alen und Kohorten an Rhein und Donau. (Bern). Marek C. 1993 Stadt, Ära und Territorium in Pontus-Bithynia und Nord-Galatia. (Tübingen). Mitchell S. 1993 Anatolia: Land, men and gods in Asia Minor I. (Oxford). Pflaum H.-G. 1971 Inscriptions de Chios. Zeitschrift für Papyrologie und Epigraphik 7: 61-63. PME 1976-1993 H. Devijver (ed.) Prosopographia Militiarum Equestrium. (Louvain). Saddington D.B. 1980 Prefects and lesser officers in the Roman auxilia of the early imperial period. Proceedings of the African Classical Association XV: 20-58. Saddington D.B. 1982 The development of the Roman auxiliary forces from Caesar to Vespasian. (Harare). Saddington D.B. 1994 A context for a dedication by five cavalry regiments to a Cornelius Scipio in Rome ? Zeitschrift für Papyrologie und Epigraphik 104: 73-80. Saddington D.B. 1996 Roman military and administrative personnel in the New Testament. In W. Haase (ed.) Aufstieg und Niedergang der Römischen Welt II.26.3. (Berlin): 2409-2435. Saddington D.B. 2000 The sorts of names used by auxiliaries in the early principate. In G. Alföldy, B. Dobson & W. Eck (edd.) Kaiser, Herr und Gesellschaft in der römischen Kaiserzeit: Gedenkschrift für E. Birley. (Stuttgart): 163-178. Solin H. 1994 Analecta Epigraphica (CL VIII, 1). Arctos XXXVIII: 113. Weber W. 1921 Josephus und Vespasian. (Berlin, republished Hildesheim, 1973). Wright R. P. 1964 Two new items in the army list of Roman Britain. Journal of Roman Studies 54: 150-151.

Bibliography Barrett K.C. 1994-98 A critical and exegetical commentary on the Acts of the Apostles. (Edinburgh). Birley E. 1949 The equestrian officers of the Roman army. Durham University Journal XLII: 8-19. = Roman Britain and the Roman army: Collected papers (Kendal, 1976): 133-153. Birley E. 1978 Alae named after their commanders. Ancient Society IX: 257ff. = The Roman army. Papers 19291986. Mavors 4. (Amsterdam 1988): 368-384. Crow J.G. & French D.A. 1980 New research on the Euphrates frontier in Turkey. In W.S. Hanson & L.G.F. Keppie (edd.) Roman Frontier Studies 1979. Proceedings of the XIIth International Congress of Roman Frontier Studies. British Archaeological Reports Int. Ser. 71. (Oxford): 902-909. Cuntz O. 1929 Legionäire des Antonius und Augustus aus dem Orient. Jahreshefte des Österreichischen Archäologischen Instituts XXV: 70-81. Dušanić S. 1996 Military diplomata for auxiliary soldiers from the Hellephone provinces. In A.D. Rizakis (ed.) Roman onomastics in the Greek east. (Leiden): 31-42. Eck W. 1999 Flavius Iuncus, Bürgus von Flavia Neapolis, ein kaiserliche Prokurator? L’antiquité classique XLII: 6775. Forni G. 1974 Estrazione etnica e sociale dei soldati delle legioni

882

Ditches for defence: towards a study of multiple fort ditches David J Breeze Most forts were protected by 2 ditches, but in some cases more occur, up to 6. It is usually argued that multiple defence systems occur when forts were overlooked by high ground or were in vulnerable locations or where the enemy had long-range missiles. Yet some forts with several ditches lie on elevated sites while the depth of some defences might be regarded as offering protection to attackers. These and related aspects will be examined though with little by way of hard conclusions offered apart from emphasising the army's inconsistency

The genesis of this discussion is a visit to the fort of Gilau during the Romanian Congress when it was suggested that the additional ditches on one side were provided because the fort was overlooked on that side from a neighbouring hill. I expressed my doubts at the time. These doubts were the result of my belief in the power of the Roman army, the efficiency of its intelligence gathering and the purpose of the Roman fort, which was not a defended garrison point but the base of a mobile unit which preferred to defend itself in the open and whose base only normally required reasonable protection: as Tacitus remarked in relation to Germany, “our presumption was that we would be attacking the Germans, not them us.” Thus, I did not consider that even an auxiliary unit would usually require extra protection in the form of additional ditches to its fort. I was also of the view that to argue that the presence of an overlooking hill would result in a multiplicity of ditches was too simplistic an approach. After more detailed consideration of the evidence, I confess to be little further forward. Indeed, if anything, I am more confused.

Various reasons have been offered for the existence of multiple ditch systems and I wish to examine these in turn. I restrict my period of study to the principate and my area of study to Europe. Weapons It is often suggested that units occupying forts with wide ditch systems might have contained “…soldiers skilled with other weapons such as archers and slingers” (Jones 1975: 113) but, as Jones went on to note, there is little evidence for the use of such weapons by normal auxiliary units and none for their use of ballistae. More recent study, however, has indicated that arrows and fragments of bows are more common finds in forts than hitherto supposed (Coulston 1985; Zanier 1988). Such items have been found at forts throughout the area of study (Coulston 1985: 22433, 264-6; Zanier 1988). Coulston concludes that “it would be an exaggeration to suggest that virtually all soldiers in the Roman army could have practised archery at some point in their careers. However, archery equipment was much more widely employed than might be expected judging solely by regimental titles” (Coulston 1985: 282). Yet, the number of forts at which archery equipment remains relatively few and Coulston goes on to note that “weapons-stores in forts must be envisaged as having had a few bows, bundles of arrows, bundles of light javelins, perhaps shaped stones for throwing and dropping, and a sack or two of caltrops to be scattered on causeways” (Coulston 1985: 295; my italics). Sling-shot is also found in Roman forts within the area of study, but in fewer numbers (Völling 1990), while Greep (1987: 197) has concluded that “it would seem that although both legionaries and auxiliaries would have been trained in the use of the sling, it was not, after the C1st AD, a common military weapon”. Ballistae was not generally available to auxiliary units in the period under discussion (Campbell 1984; 1986).

Our general assumption is that Roman forts were normally protected by two ditches. This was the conclusion of M. Jones (1975) in his study of fort defences in Britain before 117. He noted that within his period of study the range of types of ditch systems was wide, but no system in Britain before 117 had definitely more than three ditches and that “ditch systems commonly extended as far as 60ft (18m) beyond the rampart” though in some cases “there is no evidence of any ditch or other obstacle beyond about 20ft (6m) from the rampart” (Jones 1975: 110, 112). Single ditches are known and double ditches of wide variation in size and width of berms. Where multiple systems occurred, they often included one or more ditches of comparatively small size. After 117 the situation becomes more complicated. Although Hadrianic forts generally appear to have been protected by two ditches, early Antonine forts were more complex, some exhibiting 3, 4, 5 or even 6 ditches. Subsequently, there is a return to a more normal pattern.

Ian Richmond, on the basis of his work at Hod Hill, suggested that wide berms, when accompanied by an outer punic ditch might be intended to entice the would-be assailant into the range of effective fire-power (Richmond 1968: 69), but, as von Petrikovits pointed out (pers. comm.), one would have to be an Olympic athlete to hurl a javelin as far as the outer ditch. Nevertheless, the firepower of Roman soldiers and their enemies requires consideration. There is general agreement that the extent of the throwing distance of Roman spears, the usual weapon of defence as well as attack, was about 30m, well beyond

Along the northern frontier of Europe, one or two ditches have also been regarded as the norm, but multiple ditch systems do occur. First century Valkenburg (Germania Inferior), for example, had three ditches, as did C2nd Unterböbingen (Raetia), yet I can see no special reason for either. Both forts are located on slightly raised ground and neither are overlooked. Slaveni in the Olt Valley of Dacia had three ditches, but why was this particular fort singled out for special treatment?

883

Limes XVIII

the width of the normal ditch system of no more than 18m. The Quinta Society at South Shields has conducted a number of experiments with javelins, hand-thrown stones and slings: the range of the javelins was 15-20m (Griffiths 1993: 11). Baatz (1990) has considered the distance which shot of clay, stone and lead could be thrown. He concludes that while clay could be thrown up to 200m and stone a little further, the most effective ranges are 65m and 85m. Lead could be thrown further, over 300m, but is far more rare on military sites (Völling 1990). Vegetius (II.23) mentions shot hitting a target at 600ft. The self-confessed amateur sling throwers at South Shields managed 60-90m (Griffiths 1993: 11), while Griffiths (1989: 261) notes that at Burnswark the targets were placed 150m from the firing point. In summary, the 18m width of the ‘normal’ ditch system was within the effective range of Roman soldiers be they equipped with spears, sling or arrows. However, it is far from clear that all units were equipped with the weapons which covered the greater field of ground, that is bows and arrows and slings.

same distance from the adjacent hill. The suggestion that ditches provided additional defence on the side overlooked by higher ground, as at Whitley Castle, does not of course explain why some forts which were not overlooked were protected by extra ditches, such as Valkenburg, itself on a low hill. Furthermore, with so much land at its disposal, why would the army choose a dangerous site for a fort? This strongly implies that the inconvenience of being overlooked did not seem a serious consideration. Vulnerable locations Another reason offered for multiple ditch systems is the need to protect forts in exposed positions. This argument has been advanced to explain the multiple ditch systems at Ardoch north of the Antonine Wall and at Birrens, an outpost fort of Hadrian’s Wall. The argument has a tendency towards circularity: outpost forts require special protection and the fact that Ardoch is provided with five ditches demonstrates that outpost forts were specially protected. Another school of thought would see the fort at Ardoch contracting within an earlier, larger enclosure, filling the space between the early outer ditches and the new rampart by additional ditches (Breeze 1983). The two outer ditches, over 30m from the rampart, were then beyond the 30m killing range of Roman spears, though still within range of arrows and sling shot (no arrow heads nor fragments of bows have been found at either fort, but clay shot have been recovered from Ardoch and lead shot from Birrens). It could be argued that far from the commander being concerned about defence he was actually so relaxed about the power of the unit he commanded that he could afford to leave open ditches which should have been backfilled in order not to allow them to be used for the protection of attackers. An officer with careful consideration for defence might have been expected to provide sufficient ditches to cover only the killing ground. Any additional ditches could, we might believe, be used by the enemy with impunity. We might expect, therefore, that only two or three ditches to cover a killing ground of 30m would be provided if the conditions were unfavourable.

The same range of weapons was available to Rome’s enemies with the exception of ballistae. In northern Britain the weapons used by the enemies of Rome were spears and swords. There is very little evidence for the use of the bow or the sling, though bows have been found in bogs on the continent. Furthermore, it might be expected that if Rome’s enemies did use the bow and the sling, or any other weapon capable of being fired over a long distance, then all forts would be protected by a greater depth of defence than two ditches. Adjacent high ground The starting point for this review was the suggestion that forts required special protection on the side where they could be overlooked from higher ground. However, this higher ground is never immediately adjacent to the fort and so it is unlikely that the provision of extra ditches on the up-hill side had anything to do with defence against firepower as forts would normally be beyond the range of the weapons of the attackers. If there is an association between defences and adjacent hill it might be that the extra ditches were provided to stop a downhill rush, but surely that rush would be broken at the first ditch. It is possible that the purpose of the extra ditches would not only be to slow the attackers but to render them more vulnerable to the javelins and other weapons of the defenders.

Furthermore, at Ardoch there appears to be no special reason why there should be additional ditches. The land round the fort is relatively flat. This is also the case at Birrens where, inexplicably, as many as 6 ditches were provided beyond the north rampart of the fort facing relatively flat ground.

Whitley Castle in northern England is often cited as the exemplar of this type of situation, defence against the threat posed by rising ground a short distance beyond the defences. Certainly on one side there were as many as 7 ditches in an elaborate formation. (We may also note that a prominent upcast mound was left beyond the outer ditch. This is similar to the bank beyond the defences at Ardoch in Scotland and it seems not impossible that these were the result of later cleaning out of the ditch). These ditches form a system of defence 80m wide from the front of the rampart while the outer lip of the outer ditch is about the

Status We may consider that status of the unit may have played a part in dictating the number of ditches round a fort, but it is noteworthy that legionary fortresses generally only have one ditch so a multiplicity of ditches is, perhaps, unlikely to be related to status. The point here, however, is perhaps that the unit, a legion, was capable of defending itself and therefore did not require additional support in the form of ditches. This in turn suggests that where units or forts were considered to be vulnerable additional defences might be 884

David J Breeze: Ditches for defence: towards a study of multiple fort ditches provided.

Conclusions

It might not be the unit which required additional defence, but the contents of the fort. On this basis, an argument could be created for the extra ditches at Whitley Castle in north England. It is generally agreed that the fort may have had a function in relation to lead mining in the area. But to leap from that purpose of the fort to the requirement for additional defences requires a belief that the indigenous population had a requirement for lead or a desire to steal it, both of which are unproven.

We can see that extra ditches were provided when forts were overlooked, when they lay on flat ground and when they were situated in elevated position. Some ditches lay beyond the killing range of Roman spears indicative, perhaps, of Roman complacency rather than vulnerability. And we may also ask why, if fear of the enemies’ weapons had led to a multiplicity of ditches, were not all forts protected by 3, 4 or more ditches? So far as we can determine, both sides used similar weapons. Furthermore, although defensive weapons have been found at many Roman forts, they remain relatively few in number. It still remains probable that Roman soldiers under siege would soon have run out of defensive weapons and this reminds us that the fort was normally only lightly defended and in general that the defences were not to protect the unit, which was capable of defending itself.

Discussion at Congress and later During discussion at the Congress the old chestnut that commanders might have set soldiers to excavate ditches as a way of employing their energy was mentioned. More interestingly, A. Thiel suggested that ditches might have been dug to provide material for other activities, including the quarrying of gravel or clay. A primary function of ditches as drains was mooted. A. Thiel subsequently suggested that new ditches might be cut when the unit was changed, for example at Künzing and Straubing, giving rise to a multiple ditch system (cf. also Ardoch above); or at least the appearance of such a system to later excavators, for some ditches might have been back-filled when their successors were dug.

Finally, all our attempts to produce sensible explanations are thrown into disarray when the arrangements at forts on the Antonine Wall in Scotland are considered. Extra ditches were provided at several forts, with no special relevance to topography. At both Mumrills and Bearsden there were three ditches to the west, where the land is flat, two to the east where there is a slight slope down away from the fort and one double-sided ditch to the south where the land falls away more steeply! At Rough Castle there is an additional ditch half-way down the steep slope to the stream. Against rising ground at Old Kilpatrick, there were four ditches to the north, and three in a similar situation at Croy Hill, while Duntocher, with the land falling away from the fort on all sides, had three ditches to south-east and south-west. It is difficult to make any sense of these arrangements, apart to note that they seem to relate to the general individualistic nature of the forts on the Antonine Wall and the generally more defensive nature of this frontier system. (It may be noted that the two ditches of Hadrianic Künzing were extended to five in the Antonine phase, cf. Schönberger 1975).

One reason why some forts might have had fewer ditches was, accordingly to Dr. C.S. Sommer, the existence of a civil settlement preventing modification or expansion of the defences (cf. Sommer 1984: 17 for the argument that there was only one ditch on the south side of Great Chesters fort on Hadrian’s Wall as the civil settlement may have lain here). Dr. Sommer also pointed out that geophysical prospection in Germany is indicting that some forts had more than the two ditches previously identified and therefore it may be that three or more ditches may transpire to be more normal. In short, our ground plans are too often carelessly reconstructed by archaeologists from very little evidence.

In our endeavours to explain the past I suspect we are frequently too logical, seeing sensible explanations where there are none. We forget the capriciousness of human nature too much. Perhaps much did indeed depend upon the whim of the commanding officer or the human need for security through the provision of additional defences even when they were not strictly necessary.

Dr. S. Jilek also questioned whether two ditches would have been the norm. She pointed out that at the later stone fort at Carnuntum Dr. Stigliz had been able to demonstrate that there were three ditches to the south, two to west and east and none on the north side where lay not only the river but the enemy. The ‘undefended’ space today is 300-400m wide but is likely to have been broader in antiquity. In some locations the presence of woodland may have led to a desire for extra protection. For Britain, however, it has recently been argued that the British landscape had largely been cleared of trees within the frontier zone during the pre-Roman Iron age (Tipping 1994: 32), so this proposition would not appear to be relevant for Britain.

Acknowledgments I am grateful to Dr. B. Dobson, Mr. W. B. Griffiths, Mr. B. Jacobs, Dr. S. Jilek, Mr. K. Kleeman, Lt. Col. C. Mackinder, Dr. C. S. Sommer and Mr. A. Thiel for discussing various aspects of this paper with me.

885

Limes XVIII

Bibliography Baatz D. 1990 Schleudergeschosse aus Blei - Eine waffentechnische Untersuchung. Saalburg Jahrbuch 45: 59-67. Breeze D.J. 1983 The Roman forts at Ardoch. In D.V. Clarke & A.O’Connor (edd.) From the stone age to the FortyFive. (Edinburgh): 224-236. Campbell D.B. 1984 Ballistaria in first to mid-third century Britain: a reappraisal. Britannia XV: 75-84. Campbell D.B. 1986 Auxiliary artillery revisited. Bonner Jahrbücher 186: 117-132. Coulston J.C. 1985 Roman archery equipment. In M.C. Bishop (ed.) The production and distribution of Roman military equipment. Proceedings of the second Roman military equipment research seminar. British Archaeological Reports Int. Ser. 275. (Oxford): 220-366. Greep S.J. 1987 Lead sling-shot from Windridge Farm, St Albans and the use of the sling by the Roman Army in Britain. Britannia XVIII: 183-200. Griffiths W.B. 1989 The sling and its place in the Roman imperial army. In C van Driel-Murray (ed.) Roman military equipment: the sources of evidence. Proceedings of the fifth Roman military equipment conference. British Archaeological Reports Int. Ser. 476. (Oxford): 255-79. Griffiths W.B. 1992 The hand thrown stone. The Arbeia Journal I: 1-11. Griffiths W.B. 1993 Experiments with replica Roman javelins. The Arbeia Journal II: 1-13. Griffiths W.B. 1994 Reconstructing Roman slings. The Arbeia Journal III: 1-11. Jones M.J. 1975 Roman fort defences to AD 117. British Archaeological Reports Brit. Ser. 21. (Oxford). Richmond I.A. 1968 Hod Hill, Volume Two. Excavations carried out between 1957 and 1958. (London). Schönberger H. 1975 Kastell Künzing-Quintana. Limesforsuchungen 13. (Berlin). Sommer C.S. 1984 The military vici in Roman Britain. British Archaeological Reports Brit. Ser. 129. (Oxford). Tipping R. 1994 The form and fate of Scotland’s woodlands. Proceedings of the Society of Antiquaries of Scotland 124: 1-54. Völling T. 1990 Funditores im römischen Heer. Saalburg Jahrbuch 45: 24-58. Wheeler R.E.M. 1943 Maiden Castle, Dorset. (Oxford). Zanier W. 1988 Römische dreiflügelige Pfeilspitzen. Saalburg Jahrbuch 44: 5-27.

886

‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts at Wallsend and South Shields on Hadrian’s Wall Nicholas Hodgson This paper reports on the recovery in 1998-2000 of complete barrack plans at the Roman forts of South Shields and Wallsend at the eastern end of Hadrian’s Wall. At both forts excavation has shown that the barracks in the retentura in use in the C2nd had accommodated cavalrymen and their mounts in the same building. In each contubernium a drainage pit for animal waste occurred in the front room (arma), which could have housed three horses. These features were found within the unmistakable context of barracks wholly revealed in plan. These discoveries confirm the suggestion made in 1995 by C. Sebastian Sommer that buildings of this type represent not stables (the prevailing view based on fragmentary continental examples) but standard cavalry barracks, where, contrary to modern expectations, troopers and their mounts lived side-by-side. The commonest layman’s enquiry: ‘where did they keep the horses?’ and the long-standing difficulty in identifying stables in Roman forts, are simultaneously resolved. A single barrack of this type, with three horsemen and mounts in each of the contubernia would house the 30 known to be the approximate strength of a turma, along with their commanding decurion. The view favoured since the 1940s, that mounts were kept in separate stables, and that two turmae of c.30 would be accommodated in a single barrack, is decisively overturned. Instead the number of cavalry barracks in a fort corresponds directly with the number of turmae in its garrison.

In 1975-84 the late Charles Daniels revealed the whole of the interior of the Roman fort of Wallsend, at the eastern terminus of Hadrian’s Wall in north Britain (Fig. 1; from Daniels 1989). He interpreted the Hadrianic plan of the fort as possessing 8 barracks: six for the infantry centuries of a cohors quingenaria equitata, and two for the cavalry of the unit (with two turmae, or troops of cavalry, in each barrack). Two further buildings were interpreted as stables (VI and XII on Fig. 1).

Examples can also be recognised in the fortress of Usk in Britain (Marvell 1996). In the front rooms at Dormagen, environmental analysis revealed carbonised remains of horses’ hay and fodder; phosphate analysis suggested the deposition of animal waste. The pits, which would have been covered with boards, or, in a stone building, slabs, were for the collection of urine from stalled animals. The continental buildings were of C1st and C2nd date and only fragmentary plans were recovered. Until recently buildings of the Dormagen type were seen as the best evidence to date for the existence and appearance of stables in Roman forts (eg. Dixon & Southern 1992: 193).

In 1998 there was an opportunity to re-excavate parts of the fort interior in advance of the preparation of the remains for public display. Two buildings in the retentura were re-excavated in their entirety: IX (a barrack) and XII (interpreted by Daniels as a stable in its earliest form). It was found that both buildings had been barracks from the foundation of the fort through to the C3rd. The first (Hadrianic) phase of the barracks had been of timber (Figs. 2 and 3A). Each measured 45m by 7.2m and consisted of 9 contubernia and an officer’s house. Each contubernium, 3.6m wide, possessed, in its front room, a centrally placed, elongated pit, some 3m long and 0.8m wide. Corresponding with each front room pit was a hearth in the rear room, set up against the longitudinal partition. These arrangements were replicated (with stone-lined pits) when the buildings were replaced in stone in the Antonine period (Fig. 3B). By an extraordinary coincidence, excavations in 1999-2000 inside the neighbouring fort of South Shields, in the C2nd (Period 4) fort, have shown that three of the four barracks in the retentura (and presumably therefore all four) possessed an exactly similar arrangement (Fig. 4). Here the barracks had also originated in timber (Fig. 5) and had been rebuilt in stone (Fig. 3C). The stone-lined front room pits were often very well-preserved (Fig. 6) and the rear room hearths lay exactly opposite the pits, as at Wallsend (Fig. 7).

The ground conditions at Wallsend and South Shields were not suitable for the preservation of biological remains, but the preliminary results of phosphate analysis, and finds of equine equipment, confirm that horses were accommodated in the Wallsend buildings. The complete building plans now recovered at Wallsend and South Shields demonstrate beyond doubt that arrangements now unequivocally associated with the accommodation of horses occur in the context of regular barracks. The officers’ houses of these barracks contained larger pits and elaborate drainage systems, evidently serving the stabling area for a complement of horses. At Wallsend and South Shields the living space is typical of a barrack and reflects the normal hierarchy of the Roman army. There is thus no question of the buildings being stables with accommodation merely for grooms or servants (often the interpretation of the fragmentary Continental examples: eg. Gechter 1990: 75). As this building type occurs at both of these sites and may be recognised with certainty at 9 other forts partly or wholly garrisoned by cavalry, it is not possible to view it as an anomalous or unusual phenomenon. Indeed, the perspicacious suggestion made by C. Sebastian Sommer (1995), that buildings of this type represent not stables, but the standard form of a cavalry barrack, shared by troopers and mounts, stands triumphantly confirmed by the discoveries at South Shields and Wallsend. Sommer went on to suggest that a single barrack of this type, with three horsemen and mounts in each of the contubernia would house the 30 known to be

This disposition of features inside the rooms resembles exactly that found in certain Roman fort buildings on the Continent, notably at Dormagen in Germania Inferior (Müller 1979), and Ladenburg in Germania Superior (Sommer 1995: 159-60), but also at several other sites.

887

Limes XVIII

the approximate strength of a turma, along with their commanding decurion. This suggestion is now proved to be correct: the number of cavalry barracks in a fort corresponds directly with the number of turmae in garrison. The view favoured since the 1940s, that mounts were kept in separate stables, and that the men of two turmae – 60 in all – would be accommodated in a single barrack, is decisively overturned, and the mystery addressed by C.M. Wells at the 1977 Congress (Wells 1978), regarding the long-standing difficulty in identifying stables in Roman forts, solved. In C2nd Wallsend and South Shields there are no separate stables: the cavalry mounts were in the barracks of the troopers, and the only separate stabling that need be envisaged would be for baggage animals and supernumerary beasts such as remounts. As Sommer suggests, the roof-space of the barracks could have accommodated the calones, or servants, which cavalrymen, including those of the cohortes equitatae, are known to have possessed (Speidel 1989).

a similar position in relation to a cavalry barrack have recently been found in excavations at Heidenheim (abandoned in the 150s).2 The plans recovered at Wallsend and South Shields allow the following further observations to be made: 1. These barracks provide new evidence for the number of men in a cavalry turma. In each of these examples the number of contubernia is 9. Nine contubernia of three each can have held no more than 27. The number of 9 contubernia is widespread in cavalry barracks. As well as occurring at both of these Hadrian’s Wall forts, examples with this number can be found at Oberstimm and Carnuntum (Schönberger 1978: 110-15, Abb. 53; Stiglitz 1997: 24, Planbeilage 1). But to accept 27 as the strength of a turma is to fly in the face of the literary and papyrological evidence, which consistently suggests something in the range 28-30 (Tomlin 1998: 47). The solution is that in the case of Wallsend and South Shields the junior officers (principales) - duplicarius, sesquiplicarius and perhaps vexillarius - must have been accommodated with the decurion in the end building. Here the stabling arrangements seem designed to accommodate a maximum of four horses. Note that this leaves no space for the multiple mounts ascribed by Hyginus (16) to the decurion (3 in number) and junior officers (2 each). That three principales shared with the decurion seems inherently likelier than two, which would result in the turma being composed of an odd number of 29 equites.

The main objections that have been raised against Sommer’s suggestion concentrate on the effects that close confinement would have on horses, and on the benefits to horses of being kept in the open (Grönke 1997: 93-4; 1999). The accommodation of three horses in a stall 3.6m (12 Roman feet) wide is undoubtedly cramped by modern standards, but is paralleled in the ancient world, for example in a late Roman civilian context (possibly an accommodation building for pilgrims) at Tebessa in North Africa (Christern 1976: 91-4, 231-44).1 There is also a difference between saying that the entire cavalry arm of a unit was capable of standing at readiness overnight in a fort, and suggesting that they were actually confined for long periods. Sommer (1999a) has made the point that the horses would have enjoyed daily exercise, patrol or detachment from base. When they were at rest inside the fort they were by their rider and available for instant deployment, an advantage that would be completely lost if cavalry mounts grazed outside the fort. Horses could also be tethered outside the front of the barracks in the fort. This is possibly the explanation for a regular series of pits, each square in shape, 1m deep and between 1.5 and 2m across, found outside the South Shields barracks. One pit lay in front of each contubernium, and a larger pit outside the decurion’s house (Fig. 5). When first encountered it was thought that these pits must belong to a phase of temporary encampment earlier than the barrack, but it is equally possible that the pits functioned as plank covered soakaways beneath the street in the area where horses might have been tethered and groomed in the open in front of the building. If so, the pits had only operated during the earlier part of the life of the barrack: they were overlain by a rebuild of the frontage of the building, but by this time the street was supplied with a built drain. Pits occurring in

The suggested arrangement (27 in the contubernia, plus 3 principales in the end block) is important because it represents the highest number that could possibly have been accommodated in these barracks. The plans therefore give the maximum possible strength for the turma as 30 plus the decurion, and indeed confirm that this is likely to be the theoretical intended strength of a turma in a cohors quingenaria equitata. There is no reason to think the 9 contubernium barracks were designed for less than a full turma. The occurrence of this number of contubernia at a range of forts, and its persistence over a very long period of time, through timber and stone phases at Wallsend and South Shields, shows that this was not a temporary arrangement reflecting, say, the detachment of one contubernium on duty away from the fort. The figure of 30 for the strength of a turma in the cohors quingenaria equitata must now be preferred to that of 32 indicated by Arrian (in the context of an ala quingenaria) or the 33 given by Vegetius (II.6, in the context of legionary cavalry). 2. Here is new evidence for the way in which the cavalry 2

I am grateful to Dr C. Sebastian Sommer for drawing may attention to the Heidenheim pits at the Jordan Congress, and to the excavator of Heidenheim, Herr Gereon Balle, for supplying information about these features. At Heidenheim there were two distinct arrangements of pits outside the building. One was cut in places by the wall-trench at the front of the barrack, but it is possible that this wall-trench represents a rebuild.

1

EDITORS NOTE: Note Boscoreale villa, where three horses were found in a stable 3.1m wide: cf. S. Applebaum Roman Britain. In H.P.R. Finberg (ed.) Agrarian history of England and Wales 1: 2, AD43-1042. (Cambridge 1972): 3-277, at p.149.

888

Nicholas Hodgson: ‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts of a part-mounted unit could be placed within a fort plan. At Wallsend the fort certainly accommodated a partmounted cohort (cohors IV Lingonum equitata by the later C2nd or earlier C3rd).3 The C2nd garrison of South Shields is unknown, but the exact similarity of the fort plan suggests that this must also have been for a cohors quingenaria equitata. We can now present a revised view of the Hadrianic plan of Wallsend, where there are four barracks in the retentura and six in the praetentura. Here and at South Shields there is a striking division of the fort into zones for infantry and cavalry. A particularly distinctive feature of the plans of Wallsend and South Shields is an abnormally wide via quintana (the street or open strip running the width of the fort behind the central range). As this space in each case is adjacent to the retentura, assigned to the cavalry, it is tempting to interpret it as an open area for the movement and exercise of horses and the assembly of mounted troops.

Davies 1977) have supported the idea that there were usually four rather than five turmae in these units. Despite this there was always a concern that decurions might have been missing from the totals given in such documents. These concerns can be set aside in view of the unit structure revealed at Wallsend and South Shields, where the cavalry accommodation of the fort is clearly designed for four turmae. Acknowledgements I am grateful to the British Academy for a generous Overseas Conference Grant towards the cost of my travel to Jordan to give this paper. A large team of professional staff of Tyne & Wear Museums have been involved in excavation of the cavalry barracks at Wallsend and South Shields and have contributed to the discoveries described here. I owe particular thanks, for fruitful discussions about the significance of the results, to Paul Bidwell, and to the supervisors responsible for excavating the barracks: Gary Brogan and Jonathan McKelvey. Members of the Earthwatch Institute are thanked particularly warmly for their participation in the work at South Shields, without which much of this research could not have taken place.

A similar separation of cavalry and infantry zones may be suspected in other auxiliary fort plans. For example, the Flavian timber fort at Fendoch has been considered a typesite for a cohors milliaria peditata, apparently containing 10 barracks and suggested stabling sufficient only for baggage animals (Richmond & McIntyre 1939; Hassall 1983: 113-4). The barracks divide into six in the retentura and four in the praetentura. Those in the praetentura appear to have much larger front rooms than the six in the retentura; in fact their front and back rooms are fairly equally sized, as in nearly all cavalry barracks of the newly recognised type. It seems possible that the barracks in the praetentura at Fendoch were stable-barracks for the four turmae of a cohors quingenaria equitata, leaving the six barracks with a smaller arma than papilio for the six centuries of infantry.4

Bibliography Christern J. 1976 Das frühchristliche Pilgerheiligtum von Tebessa. (Wiesbaden). Daniels C.M. (ed.) 1989 The Eleventh Pilgrimage of Hadrian’s Wall. (Newcastle-upon-Tyne). Dixon K. & Southern P. 1992 The Roman cavalry. (London). Fink R. O. 1971 Roman military records on papyrus. (Princeton). Gechter M. 1990 Neue Erkenntnisse zur römerzeitlichen Besiedlung Dormagens. Archäologie im Rheinland 1989 (Rheinisches Amt für Bodendenkmalpflege): 7476. Grönke E. 1997 Das Römische Alenkastell Biricianae in Weissenburg I. Bay. Limesforschungen, 25). (Mainz). Grönke E. 1999 Grundsätzliches zur Pferdehaltung in römischen Kastellen – Die Ställe im Alenkastell in Weissenburg. In M. Kemkes & J. Scheuerbrandt (edd.) Fragen zur römischen Reiterei. Kolloquium zur Ausstellung ‘Reiter wie Statuen aus Erz. Die römische Reiterei am Limes zwischen Patrouille und Parade’ im Limesmuseum Aalen am 25./26.02.1998. (Stuttgart): 91-100. Haalebos J.K. 1999 Nederlanders in Roemenië. Westerheem (Tijdschrift voor de Nederlandse archeologie) jaargang 48 no. 6: 197-210. Hassall M. 1983 The internal planning of Roman auxiliary forts. In B. Hartley & J. Wacher (edd.) Rome and her Northern Provinces. (London): 96-131. Marvell A. G. 1996 Excavations at Usk, 1986-1988. Britannia XXVII: 51-110. Müller G. 1979 Ausgrabungen in Dormagen, 1963-1977. Rheinische Ausgrabungen, 20. (Bonn). Richmond I.A. & McIntyre J. 1939 The Agricolan fort at Fendoch. Proceedings of the Society of Antiquaries of Scotland 73: 110-154.

3. A long-standing uncertainty about the structure of the cohors quingenaria equitata is now decisively resolved by the discovery of these single turma cavalry barracks in the context of well-understood fort plans. For the number of turmae in a cohors quingenaria equitata, four is the generally accepted figure. But the failure of the text of Hyginus (27) to state the figure explicitly, and the records of cohors XX Palmyrenorum, which had five turmae (Fink 1971: nos. 1-4), have induced doubts (Hassall 1983: 97-8). Other strength reports mentioning four decurions in cohortes equitatae (Fink 1971: nos. 63-4; Thomas & 3

EDITORS NOTE – RIB 1303 records the coh. II Nerviorum at Wallsend. The inscription is undated but lacks the c. R. title taken by this unit in the C2nd (cf. R.W. Davies Some troop movements to Britian. Klio 60 (1978): 363-370 at p.368). This could be the Hadrianic garrison of Wallsend but no evidence if it was a coh. eq. (cf. J. Bennett The setting, development and function of the Hadrianic frontier in Britain. Unpubl. Ph.D thesis, University of Newcastle-upon-Tyne 1990: 389). 4 Recent geophysical surveys at Buch, in Raetia, and Tihau, in Dacia suggest that both of these forts were divided into cavalry and infantry quarters in this way. For Buch (infantry in praetentura, cavalry in retentura): Von der Osten-Woldenburg 1993; Sommer 1999b: 189-90 and this volume. For Tihau (infantry in praetentura, cavalry in retentura): Haalebos 1999: Afb. 12. At Tihau what is probably an open exercise space like those at Wallsend and South Shields occurs between the central range and the cavalry barracks.

889

Limes XVIII

Schönberger H. 1978 Kastell Oberstimm. Limesforschungen,18. (Berlin). Sommer C.S. 1995 Where did they put the horse? Überlegungen zu Aufbau und Stärke römischer Auxiliartruppen und deren Unterbringung in den Kastellen. In Provincialrömische Forschungen: Festschrift für Günter Ulbert zum 65. Geburtstag (Espelkamp): 149168. Sommer C.S. 1999a ‘Wohin mit den Pferden? Stallbaracken sowie Aufmarsch- und Übungsplätze in römischer Zeit’. In M. Kemkes & J. Scheuerbrandt J. (edd.) Fragen zur römischen Reiterei. Kolloquium zur Ausstellung ‘Reiter wie Statuen aus Erz. Die römische Reiterei am Limes zwischen Patrouille und Parade’ im Limesmuseum Aalen am 25/26.02.1998. (Stuttgart): 84-90. Sommer C.S. 1999b From conquered territory to Roman province: recent discoveries and debate on the Roman occupation of SW Germany. In J.D. Creighton & R.J.A. Wilson (edd.) Roman Germany: studies in cultural interaction. Journal of Roman Archaeology Supplementary Series 32: 160-198. Speidel M.P. 1989 The soldiers’ servants. Ancient Society 20: 239-247. Stiglitz H. 1997 Das Auxiliarkastell Carnuntum 1, 29. (Vienna). Thomas J.D. & Davies R.W. 1977 A new military strength report on papyrus. Journal of Roman Studies 67: 50-61. Tomlin R.S.O. 1998 Roman manuscripts from Carlisle: the inkwritten tablets. Britannia XIX: 31-84. Von der Osten-Woldenburg H. 1993 Neue Entwicklung der Geophysik – Beispiele aus den ProspektionsErgebnissen des Jahres 1992. Archäologische Ausgrabungen in Baden-Württemberg 18: 61-70. Wells C. M. 1978 ‘Where did they put the horses?, Cavalry stables in the early empire. In J. Fitz (ed.) Limes: Akten des XI. Internationalen Limeskongresses. (Budapest): 659-665.

890

Nicholas Hodgson: ‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts

Fig. 1. The Hadrianic plan of the fort at Wallsend according to Charles Daniels (Daniels 1989).

Fig. 2. A revised plan of Hadrianic Wallsend following excavations in 1998 which revealed the primary timber barracks.

891

Limes XVIII

B Fig. 3. A. Wallsend: Hadrianic timber cavalry barrack. B. Wallsend: Antonine stone cavalry barrack. C: South Shields: Antonine stone cavalry barrack. Shaded areas = hearths, stippled areas = urine-pits.

Fig. 4. The Period 4 (Antonine) fort at South Shields, showing cavalry barrack B6 as rebuilt in stone in the later-Antonine period.

892

Nicholas Hodgson: ‘Where did they put the horses?’ revisited: the recent discovery of cavalry barracks in the Roman forts

Fig. 5. South Shields: cavalry barrack B6 as first built in timber, c.160. Shading = hearths, stippling = urine-pits.

Fig. 6. View from the rear external wall (in foreground) of the interior of Barrack B6 at South Shields. In the left foregound can be seen a clay hearth with ashy layer spreading from it towards the viewer. Beyond this, centre left, can be seen a stone-lined urine-pit in the centre of a front room (with flagged floor). Centre right can be seen the urine-pit of the neighbouring contubernium, partly covered by a third century granary wall. Two pits lying outside the front of the barrack can be seen beyond the excavators.

893

Limes XVIII

Fig. 7. View along the long axis of Wallsend Barrack IX, from east to west, showing three contubernia. The stone wall in the centre of the picture is of a later period than the barrack. The stone front wall of the barrack is visible at the extreme left. Just to the right of this a urine-pit with a lining of stone slabs can be seen in the foreground. Immediately to the right of this is a hearth. The hollows of two further urine-pits can be seen in the centre and distance, with the later wall going over the top of them, and a hearth to the right of the central pit. 2m scales.

894

The rampart buildings of Roman legionary fortresses Birgitta Hoffmann The space between the defences proper and the via sagularis is in many fortresses filled with ovens, latrines and buildings. Their frequency and their position opposite the barrack blocks has in the past frequently led scholars to assume these structures are extensions of these barracks, providing either mess-huts or storage-space. Beginning with the excavations at the legionary fortress at Chester, which has currently the most extensively explored intervallum area, and combining it with the published evidence from the other fortresses in Britain and Germany, this paper reviews the evidence for the use of the intervallum as it presents itself to date.

In 1998 Chester Archaeology asked me to comment on the results from the excavations in the intervallum since the 1950s. Chester’s intervallum is probably by now the most extensively excavated of the earlier fortresses and therefore comparing it with the evidence from other sites in Britain seemed logical.

200). Webster mentions briefly rampart buildings and ovens from Wroxeter fortress (Webster 1988: 129, fig. 6.3, 6.19.), while the post-trenches from the East Bight site in Lincoln could be the remains of another timber building (Jones 1980: 7-9, fig. 8). Finally Exeter, has produced possible intervallum buildings at Friernhay Street and Bartholomew Street East. With apparently large stretches of empty space found in the south-eastern corner excavations (Henderson 1988: 105, 109; Fox 1966: 50, fig. 9; Bidwell 1980: 24, fig. 9).

Chester (Fig. 1) is one of the larger legionary fortresses. It was founded in the Flavian period and continued in use until the C4th, although it is unclear how much of it was still in use as a military installation by then. During the C2nd it experienced a prolonged period of neglect, with the rebuilding of the fortress in stone only really completed in the early C3rd. The intervallum has produced a series of stone built buildings of a uniform character, traces of earlier timber buildings and a series of ovens.

So far the evidence from Britain; and it is already becoming clear that it is far from uniform. It seemed therefore wise to assemble the evidence by building type and look for consistent patterns.

Caerleon, which is roughly contemporary but much smaller, has produced apart from a similar series of ovens and the famous ‘cook-houses’ or tower-extensions, a series of timber and stone buildings of very varying character, ranging from simple wind-breaks and sheds to complicated stone built latrines and the long stone building with weapon finds in Prysg Field (Boon 1987 with references: Caerleon gates - Evans & Metcalf 1992).

The earliest structures usually encountered in the intervallum of fortresses are usually rubbish pits and the latrines (eg. rubbish pits: Manning 1981: 110-114; Crummy 1988: 31; latrines: Nash-Williams 1931: 131ff.; Boon 1987: 39 n.129.), both dating back to the Augustan period, as at Oberaden (Kühlborn 1992: 83f.). The fact that this part of the fortress could be used to dump refuse and other debris, might indicate that the area was not a priori assigned as specific purpose and may instead have served as a convenient strip of essentially vacant land, which could be used in any number of ways, as and when the need arose.

The situation in York is by comparison much less well known, mainly due to the heavy disturbance by the medieval and modern town. Consequently, only rarely have entire buildings (apart from the towers) been found, but both timber and stone buildings are known to have existed, and it appears that some of the stone buildings may have touched the curtain wall of the fortress (eg. RCAHME 1962: 44, 45). Apart from these structures, ovens have been encountered at York Museum Street and York Davygate, with a third one being known from interval tower NE6, which suggests a similar frequency of these structures as in the other fortresses (Miller 1925: 182; Wenham 1962: 569f.; RCAHME 1962: 43; Ottaway 1996).

The most common feature is however, the ovens, especially in contexts pre-dating AD200. They occur in Usk, Gloucester and Inchtuthil (Manning 1981: 108-110; Heighway 1983: 23; Pitts & St.Joseph 1985: 195-200) and in all of the later fortresses, also they appear to be generally absent from C3rd+ AD contexts. Continental evidence supports early dates for oven structures, with ovens being reported amongst others from the Augustan Oberaden (Kühlborn 1992: 83f.).

The situation in the other British legionary fortresses is far less well understood: Colchester’s intervallum at Lion Walk contained only debris and pits, Gloucester has produced structures at several sites amongst them a lean-to structure and a free-standing building from the Northgate site and as well as several ovens (Crummy 1988: 31; Heighway 1983: 20-23; Hurst 1988: fig. 3.6).

To judge from the finds in the associated ashpits in front of the ovens, they were mainly used for the preparation of food. And their position is usually interpreted as a measure to minimise the fire risk to the rest of the fortress. As these ovens operate at the same time as countless hearths in the barracks, praefurnia for the bathhouses and the hearths and furnaces of the fabrica, this can hardly be the only reason for this position and other reasons need to be looked for.

Ovens and pits are also known from pre-Flavian Usk (Manning 1981: 108-115) and ovens and a bowl furnace are known from Inchtuthil (Pitts & St. Joseph 1985: 195-

It has been argued in the case of Inchtuthil (Pitts & St.Joseph 1985: 200) that there may have been one oven 895

Limes XVIII

for every century and that their location opposite the barrack buildings makes for short carrying distances. Several points argue against such a straightforward system of allocation: the evidence from Chester and Caerleon with clusters of ovens in particular areas (eg. Thompson 1967: 12; Boon 1987: fold-out plan), makes it clear that the siting of the ovens is far from regular and even at Inchtuthil, Chester and Usk the ovens were located opposite the granaries, whilst another of the Inchtuthil ovens cited in support for this theory was actually a bowl furnace, hardly a satisfactory cooking instrument for the century concerned.

All things being equal, it seems therefore that the case of assigning an oven directly to the century opposite, must at best be considered unproven, and there may well be other reasons for locating the bread ovens along the via sagularis. One of these may be institutional lethargy, as ovens in the intervallum can be found a number of camps in Scotland (eg. Alexander 1997: 8; Britannia XIX (1988): 425) and elsewhere, and it may just have been unnecessarily complicated to find a new place for such structures without pressing need. When the ramparts of some of the British fortresses were widened in the late C2nd AD room for the ovens was presumably found elsewhere.

Both arrangements of ovens, single and in batteries or clamps can be found, and there is so far no consistent pattern within one fortress or in comparison with the whole sample. They are also sometimes reported to be sandwiched between buildings as at Gloucester (Heighway 1983: 23, fig.12), or to have existed alongside longrectangular buildings as in Caerleon Gates Phase III (Evans & Metcalf 1992: 16), although the results from the Abbey Green excavation in Chester are problematic in this and several other contexts.

The structures reviewed so far are all rather flimsy and easily removable, this may suggest that the intervallum may have originally been a buffer zone, keeping the attackers away from the Roman encampment (Koenen 1903: 160). If this use of space was continued in the more permanent fortresses, one would expect that the builders would avoid creating additional risks to the occupants of the fortress, and would, therefore, avoid using this space for any vital installations, as far as this was possible. The choice of structures, oven, latrines and rubbish pits, would bear this out, as none of them is essentially military in nature, and their destruction - although deplored and inconvenient - would not have created any immediate crisis for the fortress as a whole. The intervallum may, therefore, have originally provided space for non-essential, often domestic facilities, thereby freeing up space for more important structures inside the cordon of the via sagularis.

At the Woolworth site in Chester, the ovens are dated to the Trajanic period, making them contemporary with other rampart buildings in the fortress (Thompson 1967: 12). If we return to the theory of one oven per century stationed opposite, the fact that some intervallum space was instead taken up by other buildings would have left some centuries (at least 4-6 in Chester and at least 2 in Gloucester) without a space to bake.

Following on from the early structures, more permanent buildings are occasionally found in pre-Flavian fortresses (eg. Fox 1966: 49, fig. 9), and are a common feature from the C2nd onwards.

Although this is rarely mentioned in the reports, the ovens do vary substantially in size: with the larger ones being up to twice the size of the smaller examples. This once again appears to run counter to any concept of standardisation or the provision and placing of ovens for the use and convenience of individual centuries. Indeed nothing in the current evidence would preclude an interpretation in which any one oven was being shared between different centuries, or indeed one in which the same century may make use of different ovens.

Most legionary fortresses in Britain have provided evidence for some form of long rectangular structures in the intervallum. In some cases these buildings are little more than flimsy sheds, while in other cases, eg. Caerleon Prysg Field (Nash-Williams 1931: 122-6) a large building, which may have been a weapon factory or store replaced smaller earlier buildings, which appear to have had residential character. Multi-roomed buildings are also known from York (RCAHME 1962: 44f.), but little more can currently be said about their use and plan.

A further point worth making is that at Chester (Thompson 1969: 11), Caerleon and York (Ottaway 1996: 117f.), these ovens went out of use by the end of the C2nd, in some cases earlier than that. If the rampart ovens are the only bread ovens available in the fortress at the time, we have to assume that bread - at least the oven-baked variety went out of use at that time. And even as chapati-style bread, which is baked on braziers is certainly delicious and may have been known to the Romans in one of its many varieties1, it is just as likely that baking was shifted to some other location within, or even outside the fortress.2

Chester’s rampart buildings stand out in the sample for several reasons, for one they are at the moment the largest sample from certainly any British fortress and they can actually be used to test some hypothesis on spacing. They are also unusual in that unlike the structures from the other fortresses, they appear to be all of a near uniform size and build. Their uniformity has in the past frequently attracted attention and the frequent depiction of Chesters’ northern intervallum has led to some interesting assumptions about possible regularities in function and spacing.

1

Compare the ceramic ovens found at Bearsden (information kindly provided by V. Swan and D. Breeze). 2 A possible bakery has been identified inside the legionary fortress at

Jerusalem (Guy Stiebel pers. comm.)

896

Birgitta Hoffmann: The rampart buildings of Roman legionary fortresses Currently 16-18 buildings are attributed to this group, of which only 3 have fully known dimensions (21m, 21.35m and 19.93m long and 7.3m wide), with the middle room being about 4.27- 4.88m long. The length of the two outer rooms seem to vary between 6.2 to more than 9.15m. Therefore, the regularity as visible on the large-scale plans is more apparent than real.

opposite. But could a more flexible system have been in operation? For example we could assume that there was a more general trend for the intervallum structures to be extensions of the barrack blocks, even though there was no close spatial relationship. But then we would have to expect that all buildings are of roughly the same date or at least share the same sequential history. Yet, instead we find radically different combinations, including stone rampart buildings with and without timber predecessors, stone buildings both being replaced and replacing ovens. The pattern is even within Chester not consistent. There appears to be no correlation between the rampart buildings and the ovens, nor does the dating material support distinct general trends (eg. an oven phase, a stone building phase). We must therefore conclude that even within the same fortress, the histories of these structures can be anything but uniform.

The rampart buildings survive in places to up to 1m in height, suggesting that they were stone built to a large extent and to judge from the debris encountered in their vicinity probably tiled. Some of these appear to have had timber predecessors, but too little survives to make any firm statements on their nature. Very little of the interior fittings survive and even less can be said with certainty about their function. A concrete floor in the middle room and some white wall plaster seem to constitute all that is known about their interior and only rare traces of later alterations survive.

If we then further adjust our hypothesis to take this into account, an obvious suggestion would be that instead of being assigned a specific structure, each century may have been given a particular part of the intervallum to use as they saw fit. This would certainly explain the divergent use of space, but can these divisions be detected on the ground, eg. by demarcation lines or boundary ditches? To date none have been detected.

The dates for their existence seem to cover the C2nd, with the end of the buildings c.200, when the rampart was extended, although there is little to suggest that they were still going concerns in the second half of the C2nd AD. Given their apparent similarity, although the data is far less cogent on this matter than is generally believed, it is interesting to view them in relation to their buildings opposite. As to be expected with the general plan of any fortress, the majority of rampart structures lie opposite barrack blocks, and it is therefore possible that they are associated with each other. Indeed on the northern defences it seems that the rampart buildings appear to lie one to every two centuries. Elsewhere in the fortress, rampart buildings lie opposite granaries, the bath block, and the scamnum tribunorum, which meant that if we are to assume a straightforward association between the function of the rampart building and the structures facing them, we have to envisage a number of very different functions for these buildings.

Moreover, although, a check for recurrent spacings suggests a plot length of c.24m along the fortress’s north side (allowing for non-allocated space around the gates and towers), this pattern appears to break down when applied on the western side of the fortress. Here whatever the putative interval tested, some of the stone ramparts buildings always proved to straddle the boundary lines. Worse still, the plots appear to have changed over time, for a number of the stone rampart buildings lie half over their timber predecessors and half over ovens. It could of course be argued that certain parts of the intervallum were not allocated, or that the allocation system changed between timber and stone phases, or even that the western side was allocated to the first cohort and therefore given more room. But although all of these explanations are technically possible, they introduce so many unknowns and variables that it is safer at this point simply to say that with our current evidence we cannot discern any pattern of building or space allocation with any confidence. It does have to be stressed however, that even at Chester, the available space in unexplored areas is still sufficient to allow space for 30 similarly sized rampart buildings (one for every two centuries) and 60 ovens, especially if one allows for a flexible spacing interval.

As most of these buildings are still opposite barrackblocks, at least by spacing, let us consider one possible identification. It has been argued that the structures may represent canteens/refectories or storerooms for the centuries opposite them. If such a rigid allocation had been in operation, we would expect to find a straightforward association of space wherever possible, to avoid confusion and to cut down on unnecessary traffic in an otherwise cramped environment. Ideally, the rampart structures should lie directly opposite their parent blocks. Yet, even in Chester, this is far from always the case, with some large empty gaps between the rampart buildings on the western defences, right opposite the barracks, even though they could have been easily clustered similarly to the situation on the northern defences. We have to conclude therefore that no strict and rigidly enforced system of spatial association existed, and that it may not be possible to deduce the use of any given structure from the building

To sum up, a few points stand out. Foremost their irregularity: Rampart buildings are not typical or standardised. Even within one province large difference occur, and if anything Chester is unusual in having to date such a small range of structures, while other fortresses 897

Limes XVIII

appear to use a much wider amount of buildings and structures in the intervallum, including buildings with a double range of rooms (at York), or with residential or industrial character (at Caerleon) or buildings that may only be encountered at this specific site (eg. the tower extensions at Caerleon). Differences can also be seen in the sequence of events: Not just between different fortresses, but also within the same fortress: we have discussed Chester, but the sequence of events at Caerleon Prysg Field (Nash-Williams 1931) and Caerleon Gates (Evans & Metcalf 1992) differs markedly as well. The difference of building type in Caerleon and York, make it also less likely that the structures had any uniform relationship with the internal buildings opposite them, a fact that may add some more weight to our analysis of the Chester situation.

Heighway C. 1983 The east and north gates of Gloucester and associated sites. Excavations 1974-81. (Western Archaeological Trust). Henderson C. 1988 Exeter (Isca Dumnoniorum). In G. Webster (ed.) Fortress into city. The consolidation of Roman Britain, first century AD. (London): 91-119. Hurst H. 1988 Gloucester (Glevum). In G. Webster (ed.) Fortress into city. The consolidation of Roman Britain, first century AD. (London): 48-73. Jones M.J. 1980 The defences of the upper Roman enclosure. The archaeology of Lincoln Vol. 7.1. (Lincoln). Koenen C. 1903 Bonn. Ausgrabungen im Legionslager, Funde ausserhalb des Lagers. In H. Lehner Augrabungs- und Fundberichte des Provinzialmuseums in Bonn. Bonner Jahrbücher 110: 152-165. Kühlborn J.S. 1992 Das Römerlager in Oberaden III. Bodenaltertümer Westfalens 28. (Munster). Manning W.H. 1981 Report on the excavations at Usk 1965-1976. The fortress excavations 1968-1971. (Cardiff). Miller S.N. 1925 Roman York: excavations of 1925. Journal of Roman Studies 15: 176-194. Nash-Williams V.E. 1931 The Roman legionary fortress at Caerleon in Monmouthshire. Report on the excavations carried out in the Prysg Field, 1927-9. Archaeologia Cambrensis 86: 99-158. Ottaway P. 1996 Excavations and observations on the defences and adjacent sites, 1971-90. The archaeology of York. The legionary fortress 3/3. (York). Petrikovits von H. 1975 Die Innenbauten römischer Legionslager während der Prinzipatszeit. (Opladen). Pitts L.F. & St. Joseph J.K. 1985 Inchtuthil. The Roman legionary fortress. (London). RCAHME 1962 Eburacum - Roman York. (London). Thompson F.H. 1967 Excavations on the site of Woolworths, Chester, 1959. Journal of the Chester & North Wales Archaeological Society 54: 9-19. Thompson F.H. 1969 Excavations at Linenhall Street, Chester, 1961-2. Journal of the Chester & North Wales Archaeological Society 56: 1-21. Webster G. (ed.) 1988 Fortress into city. The consolidation of Roman Britain, first century AD. (London). Wenham L.P. 1962 Excavations and discoveries within the legionary fortress in Davygate, York, 1955-8. Yorkshire Archaeological Journal 40: 507-587.

Also different is the continuation of use. It seems that in Chester the intervallum was abandoned and the rampart widened by the C3rd, whereas both Caerleon and York continue to use theirs. It may be that here we have to take space into account: Chester is markedly bigger than the other two and may have had more space within the defences. But other reasons, for example differing planning models may also have had a rôle to play. It therefore seems fair to say that in view of some of the evidence presented here, there may be good reason, that at least in Britain, and maybe elsewhere as well, each fortress may have developed its intervallum in response to its individual needs, rather than in accordance with any general blueprint, be it at provincial or imperial level. It would seem that similar to more recent army practice3 the intervallum was a space of opportunity into which the fortress could develop as and when the need arose, and from which building could serve a variety of purposes, perhaps both on a long-term basis or until better position could be found elsewhere. Bibliography Alexander D. 1997 A96 Kintore Bypass. In R. Turner (ed.) Discovery and excavation in Scotland 1996. (Edinburgh): 8. Bidwell P.T. 1980 Roman Exeter: fortress and town. (Exeter). Boon G.C. 1987 The legionary fortress of Caerleon-Isca. (Cardiff). Crummy P. 1988 Colchester (Camulodunum Colonia Victriciensis). In G. Webster (ed.) Fortress into city. The consolidation of Roman Britain, first century AD. (London): 24-47. Douet J. 1998 British barracks, 1600-1914. (London). Evans D.R. & Metcalf K.N. 1992 Roman gates – Caerleon: the ‘Roman gates’ site in the fortress of the Second Augustan legion at Caerleon, Gwent. (Oxford). Fox A. 1966 Roman Exeter (Isca Dumnoniorum). Origin and early development. In J.S. Wacher (ed.) The civitas capitals of Roman Britain. (Leicester): 46-51. 3

cf. the C19th plans of Stonehouse Marine barracks, the cavalry barracks at Christchurch and Wyvern Artillery barracks (Douet 1998: fig. 146, 60 and 74).

898

Birgitta Hoffmann: The rampart buildings of Roman legionary fortresses

Fig. 1. Chester in the C1st and C2nd AD.

899

Limes XVIII

900

Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica) Maureen Carroll in the province of Namur and in the Ardennes in Belgium (Breuer et al. 1952: 100; Brulet 1969: 57, fig. 5.3). The origin of this particular pottery with its characteristic tempering can be traced to this area, the former civitas Tungrorum. The soils of the region at the confluence of the Maas, Sambre and Lesse rivers are rich in lime and chalk from which the tempering could have been extracted (Plumier 1986: 9, 31).

Introduction Stratigraphic excavations in 1995/96 on the site of the Rhine fleet (Classis Germanica) in Köln-Alteburg clarified the chronology of the site, revealing multiple building phases and a much more complex history than previously had been realised (Carroll 1999; Carroll & Fischer 1999). After the initial military occupation of the Tiberian period, the site became the permanent operational base of the fleet probably under Claudius.

Group 2 The pottery of the second group consists of vessels made of dark grey to black clay with vegetable tempering and occasional inclusions of sherd grit and shell. The surface of the vessels is light brown or ochre, sometimes bright orange, and it is often burnished (Fig. 3). This pottery, appearing most frequently in Phases 1, 3 and 4, has parallels in fabric, appearance and form with pottery assemblages in Friesland. The commonest shape appears to be the medium-mouthed pots with an s-curved neck, but because of the fragmentary nature of the sherds it is uncertain whether the vessels are tall or low forms (Fig. 1.3-4). Horizontal ridges (Streepband) on these pots are common in Frisian pottery as well in the Frisian-influenced pottery assemblages from The Hague and Rijswijk in south Holland in the territory of the Cananefates (Taayke 1990: 132, fig. 14; Stuurman 1968: 171, fig. 3; Bloemers 1978: 362, fig. 166.80). Frisian or Frisian-type pottery has been found outside the Netherlands at Haltern on the River Lippe and in Xanten on the Rhine (Bloemers 1973; Goddard 1996: 106, pl. 28.40).

During the excavations, 198 fragments of native, handmade Roman pottery were found in the post trenches, layers and pit fills, ranging in date from the early C1st to the mid-C2nd. This may seem to be a negligible amount of pottery, however an analysis of it allows us to determine its provenances and better understand the interaction between the Roman military and indigenous societies. It also is a useful tool in understanding the rôle various regions played in supplying the fleet with provisions.1 The pottery The native hand-made pottery can be ascribed to the following occupation phases and comprises the following percentages of the total assemblage: Phase 1 (Tiberian): 3.3% Phase 2 (Claudian): 4.8% Phase 3 (Vespasianic): 2.4% Phase 4 (Domitianic): 0.9% Phase 5 (early C2nd): 0.4%

Group 3 Characteristic of the third group of pottery is the dark grey fabric tempered with organic matter which burnt out during the firing process, giving the fabric a pitted appearance (Fig. 4). The surface of the vessels is greyish brown to black and it is either polished or burnished. The vessels are small in size and range in shape from bowls and dishes to beakers and pots (Fig. 1.5). This type of pottery is relatively rare in Phases 1 to 3, increasing in Phases 4 and 5, and may be a local Germanic product. Similar pottery, mostly pots and bowls, with organic tempering and sherd grit occurs at other settlements and forts west of the Rhine such as Xanten, Nijmegen and Schayk, where it is considered to be local (Goddard 1996: 106-7, pl. 28.1-19; Stuart 1977: pl. 24, type 401; Modderman & Isings 1960/61: 342-43, fig. 10-11). It also appears regularly as part of the assemblage of grave goods in the Gallo-Roman cemetery at Maaseik in the Maas valley (Janssens 1977: 16, pl. 4-5). In areas east of the Rhine, for example at Leverkusen-Rheindorf and Cologne-Wahn, completely different vessel types are found which have nothing in common with the pottery from the fleet base (Von Uslar 1938: 224-38, pl. 1-4; Waugh 1993: 297-304), and the known hand-made Germanic pottery found in early

No hand-made pottery associated with contexts dating to the period after the mid-C2nd was found. The pottery can be divided into three main groups discussed below. Group 1 The first group consists of ovoid urns with an inverted rim, a form with La Tène origins (Cahen-Delhaye 1977) (Fig. 1.1-2). The fabric is grey to black and generously tempered with fine white lime particles (Fig. 2). The surface is light brownish grey to black and it is quite rough in the lower half of the vessels. The ubiquitous so-called Haltern cooking pots of Roman manufacture found at many Augustan to Claudian military sites in Germany are derived from the native ovoid urns (Loeschcke 1909: 294300, type 91). In contrast to the thick-walled Haltern cooking pots, the thin-walled native urns continued to be produced throughout the C2nd and into the C3rd. They appear in abundance in grave and settlement assemblages 1

A detailed study and analysis of all the hand-made pottery is now published in the Journal of Roman Archaeology (14 (2001): 310-324. My thanks go to C. Höpken for the pottery drawings and D. Godden for the plan.

901

Limes XVIII

Cologne or in Cologne’s hinterland also bears no resemblance to the Alteburg assemblage (Fremersdorf 1927: 260-61, fig. 7; Lenz & Schuler 1998). The question thus remains unanswered how local the production of this pottery was and who made it.

irregular troops in the C1st (Alföldy 1968: 73-74; Will 1987: 29-30). The Cananefates enlisted as auxiliary cohorts as did the Tungri, but whether this enlistment involved the fleet is unknown (Will 1987: 22-24; Vanderhoeven 1996: 221). The Batavians certainly served in the fleet as rowers and even as officers, and they supplied a number of Batavian cohorts (Will 1987: 15-20), however I cannot detect specifically ‘Batavian’ pottery in the fleet base assemblage which they might have brought with them. This could be related to the relatively wide distribution of handmade pottery with similar characteristics in the eastern Dutch river area, the Maas valley and the neighbouring German lower Rhine area (Hulst 1981: 359 “Keramik-Provinz”) which, in my opinion, may prohibit us from calling this pottery ‘Batavian’. The ascription of ceramic artefacts to specific tribal groups is often fraught with difficulties, particularly in the lower Rhine region, where there was much disruption of late Iron age populations and a subsequent demographic restructuring involving migrations and transplantations of new groups to the area under Augustus (Carroll 2001: 29-32).

Other wares A variety of other hand-made wares were found at the fleet base which cannot be attributed to the three main groups. This includes a vessel type with a polished neck and rows of incised dots on the shoulder which appears frequently in native Roman period pottery assemblages from the C1st to C3rd in Tournai in south-western Belgium and in the region north of Tournai between the Leie and Scheldt estuaries (Amand 1959: 118, fig. 6; Brulet & Coulon 1977: 45, pl. 31) (Fig. 1.6). This type of vessel of La Tène origin is apparently a product of the Scheldt region which in the Roman period was part of the civitas Menapiorum (Leman-Delerive 1984: 85-89, fig. 8-9). Finally, an extremely soft, organic-tempered hand-made pottery was found which is commonly known at sites in the western Netherlands and Belgium near the North Sea coast where salt was extracted from peat and sea water. This pottery was characteristically used for the packaging and shipping of salt, and most commonly appears in the form of cylindrical beakers (Bloemers 1978: 372-73, fig. 178; Van den Broeke 1995) (Fig. 1.7-8). The clay itself is grey, the chalky surface is either buff or pale green (Fig. 5). These cylinders appear only in contexts dating to the Vespasianic period (Phase 3), in particular in refuse pits where they were dumped after the salt cakes in them had been removed.

On the northern frontier of the empire, inscriptions and pottery found at Housesteads, Chesterholm and Birdoswald in possibly late C2nd and C3rd contexts have been cited as evidence that Frisii enlisted in the Roman army and were stationed at Hadrian’s Wall, either taking their pottery with them or producing it locally (Jobey 1979). Soldiers from distant provinces could indeed foster the production of ceramics which were an integral part of food preparation and cooking as they knew it, as the occurrence of clay braziers, casseroles and platters of North African type on the Antonine Wall in Scotland indicate (Swan 1997: 29193). The Frisian vessels from the Hadrian’s Wall sites have been classified as cooking pots, and many of the sherd surfaces have extensive traces of burning. But were the C1st and early C2nd Frisian pots in Cologne also used in cooking? The forms represented in the British group of pots are not present in the Frisian pottery at the Alteburg, nor do the vessels from the fleet base show signs of burning as a result of having stood on a hearth. Must they be related to cooking in the first place? These pots could have functioned in other ways, and it is worth considering whether they might have been equally suitable as containers.

Native pottery and ethnic origins of troops Pottery not native to the area in which a military base was located could be interpreted as an indication of the ethnic origins of the troops stationed at a fort, the troops having brought this pottery with them (Furger-Gunti 1979: 142). Alternatively, the presence of ‘foreign’ pottery at Roman military sites is often said to reflect the movement of troops and soldier-potters making pots in the tradition of their origins (Swan 1984: 85-87). Troop deployment in the Rhine fleet must be considered if we are to determine whether either of these alternatives apply to the Alteburg. Although many of the men serving in the fleet were of eastern origin, others were recruited from the western provinces. Saddington in his study on the composition of the provincial fleets suggested that the Classis Germanica used non-local expertise combined with local rowers who were analagous to the local auxiliaries used in fighting on land (Saddington 1990: 230). The native groups of the lower Rhine often appear to have been taxed in recruits for the military (Roymans 1996: 86-87), yet there is no firm indication that the regions which supplied the hand-made pottery to Cologne also supplied troops to the legions or the fleet at the Alteburg. There is no evidence that the Ubii of the Cologne region or the Frisii served in the fleet, although they did provide auxiliary troops and possibly

Native pottery as packaging: supplying the fleet The native pottery found at the fleet base may not reflect trade in ceramics or the origins of the troops. It may be that the vessels themselves were not traded items, but rather the contents of the vessels, even if it is not always clear what these contents were. At least some of the pottery, with the exception of the bowls, plates and beakers of Group 3 which must have been used as utilitarian items for table and hearth, could have arrived at the fort as packaging, for example for foodstuffs. Based on the native pottery (5% of the total assemblage) found at the Roman fort at Velsen on the lower Rhine frontier, it has been suggested that the 902

Maureen Carroll: Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica) native pots might have been delivered to the base because of their contents, but it was also considered possible that the pots were brought into the base by local recruits “who had their own kitchen outfit” (Brandt 1983: 135). The same possibilities exist in regard to the native pottery, apparently Frisian or Frisian-influenced pottery, found at the earliest fort of the C1st at Valkenburg on the Rhine frontier in Cananefatian territory (Glasbergen & Groenman-van Waateringe 1974: 35, pl. 9).

territory of the Nervii at Pommeroeul on the River Haine and at the fort at Woerden on the Rhine (De Boe 1976; Haalebos 1986: 170-72). On board the latter ship were grain from the Belgian loess belt and pottery vessels identified by the excavator as originating in Nervian and Menapian territory (Flanders). In this context one could ask whether these vessels were food containers for further distribution accompanying the grain transport or were merely the kitchen outfit of the men on the barge. Even if they were pots containing food just for the crew they are still food containers. Furthermore, the military bases in the lower Rhine area from which the fleet operated were in contact with each other and able to transfer goods from the coast or from the interior further afield. Near the forts were native civilian settlements, any of them able to supply provisions and goods of various kinds to the military. Moreover, beyond the frontier was a substantial nonRoman population of Frisians who lived in part on trade with the Roman empire. Finally, large areas of the GalloBelgic hinterland were certainly involved in the provisioning of the military in the lower Rhineland. In eastern Belgium, a town like Tongeren as the capital of the civitas Tungrorum, located on the one of the most important east-west overland routes, would have been an ideal collecting point for goods, among other things cereals cultivated in the southern Belgian villa belt, destined for military consumption on the Rhine (Vanderhoeven 1996: 221).

At the Alteburg, however, neither local foods nor local recruits can account for the native pottery of distant origin. Foods, and in particular speciality foods such as the famous Menapic salted hams from the De Panne area in coastal Belgium, could travel great distances (Strabo Geog. IV.4.3). The important discovery of an ovoid urn of the Belgic type from the Ardennes at the Roman fort on the Kops Plateau in Nijmegen containing the bones of at least 28 song thrush breasts (preserved in salt or honey?) points to one of the possible commodities that were transported in these vessels and to the distances these delicacies could travel (Lauwerier 1993). The same kind of pots found at the Alteburg (Group 1), for which an origin in the Namur area and the Ardennes is proposed, were coated with a black substance on the inside, possibly disinfectant pitch, perhaps indicating that the contents were indeed perishable foodstuffs that needed preservation during transportation (Rottländer 1974). The fact that the Belgic urn is generally referred to as a cooking pot, and yet it obviously could be used primarily as food packaging, should be a warning against simple one-dimensional interpretations of vessel functions. Whether or not these pots could have been used secondarily as cooking pots after arriving at their destination is irrelevant. Export products of the south Holland region included goose liver, fish sauce and dried fish, all of which are possible candidates for the contents of the Frisian vessels at the Alteburg, as are fish and dairy products from Friesland itself. Clearly salt was contained in the cylinders of northern coastal origin, the vessels themselves being nothing more than packaging with no intrinsic value. Inscriptions attest military involvement in the salt trade, referring to salt merchants on the Belgian and French North Sea coast (salinatores civitatis Menapiorum, salinatores civitatis Morinorum) and a centurion of the legio VI in Neuss on the Rhine who supervised this exchange during the reign of Vespasian (van Beek 1993: 8-9). Civilian salt merchants based in Cologne also dealt with North Sea salt producers (Hassall 1978: 45, table I).

Possibly negotiatores with military contracts to provide the military with supplies were involved in the procurement of goods from the north-west, on both sides of the frontier. In AD69 negotiatores, presumably with military contracts, were active everywhere in Frisian territory according to Tacitus (Hist. IV.15). Inscriptions and other written documents record economic exchange well beyond the frontier (Galestin 1997: 352). Of particular importance regarding native supplies to the military in Cologne is the wax tablet from Tolsum in Friesland recording the purchase of Frisian cattle by a military purchaser accompanied by two centurions from legiones I and V (Whittaker 1994: 113). The tablet may be as early as the early C1st, when legio I was stationed at the legionary fortress at the Alteburg, and the connection with this unit stationed in Cologne is significant. The military ties with the Menapian and Morinian salt merchants and the association of negotiatores at Colijnsplaat and Domburg with the German fleet have been recognised (Van Beek 1993: 9-11; Middleton 1979: 85).

The military, particularly the Rhine fleet, had regular contact and exchange with those peoples living in northern and north-western estuary systems and coastal regions (Fig. 6). Since the late C1st BC not only the North Sea and Atlantic coast, but also the delta area and numerous other tributaries deep into the hinterland behind the frontier were integrated into the Roman military, political and administrative system. These areas were an essential part of the supply hinterland for the fleet operating on the lower Rhine, as indicated by the finds of transport barges in the

Suppliers for the fleet vs. suppliers for civilians? It is noticeable that the native Roman period pottery under discussion here apparently is absent from the ceramic assemblages of the civil settlement at Cologne. Perhaps the presence of the Belgic and Frisian wares and the rest of the native handmade pottery in this military context is an indication that the military as a large consumer group could tap other sources and had different suppliers of particular products than the civilian population. Whittaker 903

Limes XVIII

argues convincingly that the military relied heavily on compulsory requisition of a wide variety of supplies, either for payment or as an extra tax indiction, rather than the general procurement of goods through the less dependable open market (Whittaker 1994: 106-8).

De Boe G. 1976 Binnenhafen und Schiffe der Römerzeit von Pommeroeul im Hennegau (Belgien). Archäologisches Korrespondenzblatt 6: 227-234. Fremersdorf F. 1927 Gräber der einheimischen Bevölkerung römischer Zeit in Köln. Prähistorische Zeitschrift 18: 255-293. Furger-Gunti A. 1979 Die Ausgrabungen im Basler Münster I. Die spätkeltische und augusteische Zeit. (Basel). Galestin M.C. 1997 Romans and Frisians: Analysis of the strategy of the Roman army. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.), Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 347-353. Glasbergen W. & Groenman-van Waateringe W. 1974 The preFlavian garrisons of Valkenburg (Z.H.). (Amsterdam). Goddard E. 1996 Colonia Ulpia Traiana. Die Ausgrabungen im Bereich des Hauses am kleinen Hafentor (Insula 38). (Munich). Groenman-van Waateringe W. 1989 Food for soldiers, food for thought. In J.C. Barrett, A.P. Fitzpatrick & L. Macinnes (edd.) Barbarians and Romans in north-west Europe. British Archaeological Reports Int. Ser. 471. (Oxford): 96-107. Haalebos J.K. 1986 Ausgrabungen in Woerden (1975-1982). In C. Unz (ed.) Studien zu den Militärgrenzen Roms III. 13. Internationaler Limeskongreß Aalen 1983. (Stuttgart): 169-174. Hassall M. 1978 Britain and the Rhine provinces: Epigraphic evidence for Roman trade. In J. du Plat Taylor & H. Cleere (edd.) Roman shipping and trade: Britain and the Rhine provinces. (London): 41-48. Hulst R.S. 1981 Einheimische Keramik aus römischer Zeit im geldernschen Flußgebiet - ein Klärungsversuch. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 31: 355-363. Janssens D. 1977 Een gallo-romeins grafveld te Maaseik (Archaeologia Belgica 198). (Brussells) Jobey I. 1979 Housesteads Ware. A Frisian tradition on Hadrian’s Wall. Archaeologia Aeliana 7: 127-143. Lauwerier R.C.G.M. 1993 Twenty-eight bird briskets in a pot; Roman preserved food from Nijmegen. Archaeofauna 1993/2: 15-19. Leman-Delerive G. 1984 Céramique laténienne domestique de la Région Lilloise (Nord). Gallia 42: 79-95. Lenz K.H. & Schuler A. 1998 Handgeformte Gefäßkeramik der frühen römischen Kaiserzeit aus Bornheim-Sechtem, Rhein-Sieg-Kreis. Archäologisches Korrespondenzblatt 28: 587-599. Loeschcke S. 1909 Ausgrabungen bei Haltern. Die keramischen Funde. MittWestfalen 5: 101-322. Middleton P.S. 1979 Army supply in Roman Gaul: an hypothesis for Roman Britain. In B.C. Burnham & H.B. Johnson (edd.) Invasion and response. The case of Roman Britain. British Archaeological Reports Brit. Ser. 73. (Oxford): 81-98. Modderman P.J.R. & Isings C. 1960/61 Een grafveld uit de Romeinse tijd op de Gaalse Heide, gem. Schayk (N.Br.). Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 10/11: 318-346. Plumier J. 1986 Tumuli belgo-romains de la Hesbaye occidentale: Seron, Hanret, Bois du Buis, Penteville. (Namur). Rottländer R.C.A. 1974 Keramik mit Randverpichung und Schwarzrandware. Archäologisches Korrespondenzblatt 4: 95-98.

What other goods were brought from the supplying regions into the base at Cologne is unknown, but those in the containers discussed in this paper would certainly have accompanied larger transports, and as such the native pots at the fleet base are an indicator of supply routes. Since the frontier zone will not have been able to begin to meet the food needs of the military at least until the later C1st AD, if not the C2nd (Groenman-van Waateringe 1989: 96-107), the commodities reaching the fleet base may have been grain from the Belgic hinterland (via Tongeren and the Maas River), but perhaps also cattle and hides from the lower Rhine (via the Rhine delta) and certainly salt from the coast (via the Rhine-Scheldt delta). The noticeable quantitative reduction in native wares from the late C1st, when the fleet base was rebuilt and the province reorganised under Domitian, may indicate that the unit could rely from then on more heavily on local supplies than on distant sources. Bibliography Alföldy G. 1968 Die Hilfstruppen in der römischen Provinz Germania Inferior. (Düsseldorf). Amand M. 1959 Céramique pré-Claudienne à Tournai, L’Antiquité Classique 28: 107-224. Bloemers J.H.F. 1973 ‘Fries’ aardewerk uit de Romeinse legerplaatsen de Haltern (Duistland). In Archeologie en historie, opgedragen aan H. Brunsting. (Bussum): 225231. Bloemers J.H.F. 1978 Rijswijk (Z.H.)., ‘De Bult’. Eine Siedlung der Cananefaten. (Amersfoort). Brandt R. 1983 A brief encounter along the northern frontier. In R. Brandt & J. Slofstra (edd.) Roman and native in the Low Countries. Spheres of interaction. British Archaeological Reports Int. Ser. 184. (Oxford): 129143. Breuer J., Roosens H., & Mertens, J. 1952 Le cimetière belgoromain de Cerfontaine Archaeologia Belgica 6: 95129. Brulet R. 1969 La nécropole belgo-romaine de Biesme. (Annales de la société archéologique de Namur 55). Brulet R. & Coulon G. 1977 La nécropole gallo-romaine de la Rue Perdue à Tournai. (Louvain). Cahen-Delhaye A. 1977 Seconde campagne de fouilles dans l’éperon barré de Cherain-Brisy. Archaeologia Belgica 196 (Conspectus MCMLXXVI): 33-36. Carroll M. 1999 New excavations at the base of the Classis Germanica in Cologne (Alteburg). In N. Gudea (ed.) Roman Frontier Studies. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 317-324. Carroll M. 2001 Romans, Celts and Germans: The German provinces of Rome. (Stroud). Carroll M. & Fischer T. 1999 Archäologische Ausgrabungen 1995/96 im Standlager der römischen Flotte (Classis Germanica) in Köln-Marienburg. Kölner Jahrbuch 32.

904

Maureen Carroll: Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica) Roymans N. 1996 The sword or the plough. Regional dynamics in the Romanisation of Belgic Gaul and the Rhineland area. In N. Roymans (ed.) From the sword to the plough. Three studies in the earliest romanisation of northern Gaul. (Amsterdam): 9-126. Saddington D.B. 1990 The origin and nature of the German and British fleets. Britannia XXI: 223-232. Stuart P. 1977 Gewoon aardewerk uit de Romeinse legerplaats ende bijbehorende grafveld te Nijmegen. (Leiden). Stuurman P. 1968 Roman period pottery from the Zichtenburg Town Development scheme, The Hague. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 18: 163-174. Swan V.G. 1984 The pottery kilns of Roman Britain. (London). Swan V.G. 1997 Vexillations and the garrisons of Britannia in the second and early third centuries: A ceramic viewpoint. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 289-294. Taayke E. 1990 Die einheimische Keramik der nördlichen Niederlande, 600 v. Chr. bis 300 n. Chr., Teil I: Westergo (Friesland). Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 40: 109-222. Van Beek B.L. 1993 Salinatores and Sigillata: The coastal areas of north Holland and Flanders and their economic differences in the 1st century A.D. Helinium 23: 3-12. Van den Broeke P.W. 1995 Southern sea salt in the Low Countries. A reconnaissance into the land of the Morini. In M. Lodewijckx (ed.) Archaeological and historical aspects of west-European societies. (Leuven): 193-205. Vanderhoeven A. 1996 The earliest urbanisation in northern Gaul: Some implications of recent research in Tongres. In N. Roymans (ed.) From the sword to the plough. Three studies in the earliest romanisation of northern Gaul. (Amsterdam): 189-260. von Uslar R. 1938 Westgermanische Bodenfunde des ersten bis dritten Jahrhunderts nach Christus aus Mittel- und Westdeutschland. (Berlin). Waugh K.E. 1993 The Germanic cemetery at Rheindorf. Problems with the identification of gender and status. In M. Struck (ed.) Römerzeitliche Gräber als Quellen zu Religion, Bevölkerungsstruktur und Sozialgeschichte. (Mainz): 297-304. Whittaker C.R. 1994 Frontiers of the Roman empire. A social and economic study. (Baltimore). Will W. 1987 Römische ‘Kleintel-Randstaaten’ am Rhein? Bonner Jahrbuch 187: 1-62.

905

Limes XVIII

Fig. 1. Hand-made pottery from the fleet base (Alteburg). 1-2 - Belgic urns from the Namur/Ardennes region, 3-4 Frisian or Frisianinfluenced pots, 5 - Germanic organic-tempered bowl, 6 - pot from the Scheldt region, 7-8 salt cylinders from the North Sea coast.

Fig. 2. Hand-made pottery from the Namur/Ardennes region.

Fig. 3. Hand-made pottery from Frisia or neighbouring areas.

906

Maureen Carroll: Native pottery, food packaging and the supply lines of the German fleet (Classis Germanica)

Fig. 4. Hand-made Germanic pottery.

Fig. 5. Hand-made salt cylinders from the North Sea coast.

907

Limes XVIII

Fig. 6. Northern Belgica and the Lower Rhine region showing the economic basis of both, the tribal civitates and their capitals. (Adapted from Whittaker 1994): 1. Castellum Menapiorum (Cassel), 2. Tarvanna (Therouanne), 3. Nemetacum (Arras), 4. Bagacum (Bavai), 5. Atuatuca Tungrorum (Tongeren), 6. Colonia Claudia Ara Agrippinensium (Cologne), 7. Colonia Ulpia Traiana (Xanten), 8. Ulpia Noviomagus Batavorum (Nijmegen), 9. Forum Hadriani/Municipium Aelium (Voorburg-Arentsburg).

908

Neue Grabungen an der Westseite des römischen Flottenlagers Köln - Alteburg Thomas Fischer

Im Stadtgebiet von Köln lag im heutigen Stadtteil Marienburg auf der Flur Alteburg ein großes Militärlager. Mit seinen c.7ha. Grundfläche dürfte es weit mehr als tausend Mann Platz geboten haben, wenn man als Vergleich Auxiliarlager mit bekannter Mannschaftsgröße heranzieht. Spätestens seit der Mitte des 1. Jhs. n. Chr. kann man das Lager von Köln Alteburg durch Ziegelstempel und einschlägige Inschriften von Flottenangehörigen, oft östlicher Herkunft, als das Standlager der Classis Germanica ansprechen (Schönberger 1985; Oschmann 1987; Bechert & Willems 1995; Hanel 1998a; Hanel 1998b). Das Flottenlager befindet sich c.3kms südlich der Koloniestadt auf einer c.16ha großen hochwassersicheren Terrasse über dem Rhein, dort hatte man es in der Grundform eines verschobenen Trapezes der Geländeform angepasst (Päffgen & Zanier 1995).

Man kannte aus den älteren Grabungen zwei Hauptbauphasen, eine ältere Holzbauphase der vorflavischen Zeit und Steinbauten ab der spätflavischen Periode, das Fundmaterial, soweit es den Krieg überstanden hat, ist mit Ausnahme der Inschriften und der Münzen noch nicht bearbeitet. Grabung 1995/96 In Zusammenarbeit des Römisch-Germanischen Museums der Stadt Köln und der Abteilung Archäologie der römischen Provinzen am Archäologischen Institut der Universität zu Köln kam es in den letzten Jahren zu neuen großflächigen Untersuchungen im Innenraum des Flottenlagers Köln - Alteburg. In den Jahren 1995/96 erzwang der Bau einer Tiefgarage im Osten des Lagers die Untersuchung eines Areals von c.1000m (Carroll-Spillecke 1999; Carroll & Fischer 2001).

Die Forschungen zum Lager von Köln Alteburg setzten im letzten Drittel des 19. Jh. ein, als bei Bauarbeiten erste Dokumentationen und Fundaufsammlungen zustande kamen. Größere Plangrabungen wurden von Fritz Fremersdorf 1927/28 durchgeführt (Hanel 1998a). Im zweiten Weltkrieg sind allerdings die Grabungsdokumentation und viele Funde dieser Untersuchungen zugrundegegangen. So war es leider lange Zeit kaum möglich, die Verhältnisse im Flottenlager Alteburg im Detail zu überblicken, denn der Publikationsstand war hier sehr bescheiden und nicht im mindesten der Bedeutung des Platzes angemessen.

Grabung 1998 Im Jahre 1998 ergab sich durch ein weiteres Bauvorhaben im Bereich des Flottenkastells erneut die Möglichkeit, in Zusammenarbeit des Römisch-Germanischen Museums der Stadt Köln mit der Abteilung Archäologie der römischen Provinzen am Archäologischen Institut der Universität zu Köln weitere Untersuchungen durchzuführen (Fischer 2001). Die durch die Baugrube einer Tiefgarage vorgegebene Grabungsfläche von 1998 maß c.1400m (20 x 70m). Sie lag insofern günstig, als sie fast parallel zu den übereinanderliegenden Kasernenbauten des Lagers verlief. In ihrer westlichen Schmalseite erreichte sie noch die via sagularis, am Ostende konnte kein Bau in ganzer Länge erfasst werden.

Es lag bisher nur ein sehr summarischer Plan der Grabungen aus dem letzten Jh. und vor dem 2. Weltkrieg vor, bei dem mehrere Bauphasen undifferenziert zusammengezeichnet worden sind, außerdem war in der Literatur die Größe des Lagers um die Hälfte zu klein angegeben (Oschmann 1987). Er zeigt Teile der Umwehrung in einer älteren Holzbau - und einer jüngeren Steinbauphase. Von den Toren sind das Südwesttor, vielleicht die porta praetoria, und das Nordtor gesichert. Die Innenbauten im Nordbereich, stellen eine kaum entwirrbaren Komplex verschiedener Steinbauphasen dar, die sich einer sinnvollen Interpretation weitgehend entziehen. Im Südwesten dagegen konnte Fremersdorf große mehrphasige Kasernenbauten nachweisen, die jetzt vorgelegt worden sind (Hanel 1998a). Bei einem im 19. Jh. nur notdürftig dokumentierter Baukomplex im Osten des Lagers dürfte es sich um die Reste der Lagerthermen handeln. Auch der vicus des Lagers ist nur in Ansätzen ergraben, Streifenhäuser und eine mansio der Grabung Fremersdorf hat Hanel 1998 publiziert. Die noch nicht näher bearbeiteten, in Ansätzen ergrabenen Gräberfelder des Flottenlagers scheinen sich in der SW - Ecke zu befinden, bzw. auf die westlich des Lagers von S nach N führende Limesstraße zu beziehen.

Die Grabung ergab mindestens sechs Bauperioden, deren Unterphasen allerdings noch zu erarbeiten sind. Sie reichen von der tiberischen Zeit bis in das 3. Jh. Dabei zeigte sich im Vergleich mit der Grabung 1995/96, daß die Periodisierung zwar in einigen Phasen (bis in die frühflavische Zeit hinein) im Wesentlichen gleich verlief, sich dann aber auch seit dem Ende des 1. Jhs. auch grundlegend unterschied. In der Periode I ergab sich für die Grabung 1998 wie bei der Grabung 1995/96 ein spätaugusteisch - tiberisches Anfangsdatum. Offensichtlich standen hier im gewachsenen lehmigen Sand noch keine stabilen Kasernenbauten, vielmehr fanden sich die Spuren von mehreren, sich überlagernden Bauten, die eher für ein wiederholtes periodisches Aufsuchen des Platzes sprechen als für eine feste Bebauung. Diese Bauperiode endete in claudischer Zeit. Auch die folgenden Holzbauphasen der Perioden II und III entsprechen denen der Grabung 1995/96, präsentieren sich allerdings in wesentlich

909

Limes XVIII

schlechterem Zustand. Dies rührt daher, daß im Westen des Lagers offensichtlich immer wieder das stärker reliefierte Areal abgegraben und so allmählich eingeebnet worden ist. Im Gegenzug dazu planierte man im Osten Material an, wenn eine Neubebauung anstand. In beiden Bauphasen errichtete man - im Gegensatz zur Periode I stabile Lehmfachwerkkasernen mit Schwellbalkenkonstruktionen. Periode II datiert in die claudisch - neronische Zeit, Periode III wurde in frühflavischer Zeit errichtet. Die Holzbauperiode IV gehört nach erster Sichtung des Fundmaterials dem frühen 2. Jh., also der traianischen Zeit an. Hier erfasste die Grabung die Ausschnitte zweier Kasernenbauten, die nahezu in O - W Richtung verliefen. Vor diesen Baueinheiten lag jeweils eine c.2m breite Veranda, sie waren durch eine c.4m breite Schotterstraße getrennt. Die beiden Bauten sind nach einem Grundmodul aufgebaut, das bisher aus keinem römischen Militärlager bekannt geworden ist: drei kleinere Räume von c.9.5m lassen sich jeweils zwei größeren rückwärtigen Räumen von c.17m zuordnen. Gut erhalten waren jeweils Feuerstellen in den größeren Räumen und den beiden äußeren kleinen, der mittlere Raum ohne Feuerstelle ist als Korridor zu interpretieren, der die vier anderen Räume erschließt. Die Kaserne besitzt auffälligerweise keinen Kopfbau an der Via sagularis, nur eine etwas andere Raumeinteilung. Vergleicht man ein solches aus 5 Räumen bestehendes Grundmodul von 62.5m mit den Räumen von Contubernien (jeweils papilio und arma) von Legionslagern und Auxiliarkastellen, so könnte man zwei Kontubernien darin unterbringen, denen jeweils etwas mehr als 31m zur Verfügung stünden.

erhalten großformatige Waffen und Geräte zutage: Drei Dolche, zwei Schilde, mehrere Äxte, Herdschaufeln und andere größere, vor der Restaurierung nicht zu beurteilende Eisengegenstände. Auch ganze Bronzegefäße kamen in den Boden und belegen so zusammen mit dem übrigen, zum Zeitpunkt des Brandes gebrauchsfertigen Gegenständen, daß hier vollständig eingerichtete Bauten vom Feuer erfasst worden sind. In mehreren hundert Exemplaren kamen aus der Brandschicht faustgroße Wurfgeschosse aus grob bearbeitetem Tuff zutage, die auch sonst das Fundgut der Alteburg prägen. Da sie nur in der Brandschicht, aber nie auf dem Estrich lagen, kann man erschließen, daß sie über den Wohnräumen auf dem Dachboden gelagert waren.

Die beiden Kasernenbauten der Periode IV gingen im frühen 2. Jh. n. Chr. durch ein lokales Schadensfeuer zugrunde. Daraufhin hat man den Brandschutt nicht abgetragen, sondern planierte ihn an Ort und Stelle ein, so entstand eine leicht von älteren und jüngeren Straten trennbare Brandschicht. Diese war mit außerordentlich großen Mengen an Fundmaterial durchsetzt, was auf eine rasche Ausbreitung des Feuers zu einem Zeitpunkt schließen lässt, als die Kasernen in voller Nutzung standen. Als Konstruktionsprinzip der abgebrannten Bauten ergab sich rutengeflechtgestütztes Lehmfachwerk auf überblatteten Schwellbalken, die Räume waren verputzt und mit Wandmalerei ausgestattet (weiße Felder mit Rahmung aus roter Streifenbemalung, dazwischen Kandelaber). Das völlige Fehlen von Dachziegeln läßt auf organische Dachdeckung (Stroh, Schilf, Schindeln) schließen.

Das Fundmaterial der beiden neuen Grabungen unterscheidet sich kaum von dem, was in römischen Lagern allgemein üblich ist: Keramik, Tierknochen, weit über 200 Münzen, mindestens 50 Fibeln, zahlreiche Waffenteile etc. Das Ende der Classis Germanica scheint am Ende des 3. Jhs gekommen zu sein, da wird jedenfalls das Lager auf der Alteburg aufgelassen. Die durchgehende Münzreihe der Grabung 1995/96 bricht mit barbarisierten Prägungen des Tetricus I oder II (270/273) ab, auch Inschriften und Ziegelstempel, Keramik, Fibeln und andere Kleinfunde erreichen das 4. Jh. nicht mehr.

Auf die Holzbauphase IV folgte nach den neuesten Forschungen durch N. Hanel eine Holzbauphase IVa, dann die auf Steinfundamenten errichteten Kasernenbauten der Perioden V bis VI. Bemerkenswerterweise entsprechen diese dann dem sonst üblichen aus zwei Räumen bestehenden arma - papilio - Schema römischer Legions oder Auxiliarbaracken. Die Bauten der Periode IVa erlitten um die Mitte des 2. Jhs ein unterschiedliches Schicksal: Möglicherweise brannte nur der Bau südlich der Lagerstraße ab, der andere blieb vom Feuer verschont. Die oberste Periode VI gehört schon dem 3. Jh. an. Allerdings scheint es gerade bei den Steinbauphasen möglich zu sein, bei der Auswertung der Grabungspläne noch weitere Unterphasen herauszuarbeiten. Fundmaterial

Innerhalb des Lagers von Köln - Alteburg lassen sich im Fundmaterial inzwischen Hinweise auf wirtschaftliche Aktivitäten ausmachen, die in ihrer Summe durchaus geeignet sind, das bisher gezeichnete Bild von der Funktion des Lagers als Stabsquartier und Truppenlager wesentlich zu ergänzen. Anscheinend stellte die Truppe wesentliche Teile ihres Bedarfes an Bewaffnung und Ausrüstung selbst her: So wurden offensichtlich die erwähnten Wurfgeschosse aus Tuff im Lager selber produziert. Sie sind nämlich gelegentlich mit Spuren von Kalkmörtel versehen, was belegen könnte, daß sie nicht in den Steinbrüchen des Brohltals aus bergfrischem Material gehauen worden sind. Vielmehr hat man hier erst in sekundärer Verwendung des Tuffgesteins abgebrochene Tuffbauten des Lagers - in Munition umgewandelt. In einem Contubernium der ersten oder zweiten

Bei den Kleinfunden war erwartungsgemäß der Anteil an Keramik besonders groß, besonders an stark verbrannten ganzen oder ergänzbaren Gefäßen. Bei der Reliefsigillata dominierte spätsüdgallische Sigillata, bei der glatten südgallischen Sigillata fiel das mehrfache Vorkommen der sonst so seltenen Tintenfässer auf. Als interpretierbares Fundensemble ergab sich eine komplette Kücheneinrichtung mit Herdstelle, Handmühle, Reibschüsseln, Krug und anderen Gefäßen. An Metallfunden kamen neben Münzen und Fibeln auch gut 910

Thomas Fischer: Neue Grabungen an der Westseite des römischen Flottenlagers Köln - Alteburg Steinbauperiode der Grabung 1998 trat eine Beinschnitzerei des 2. Jhs. zutage. In großer Menge fanden sich Rohmaterial, Halbfabrikate und Fertigprodukte, darunter Haarnadeln aus Bein. Das Halbfabrikat einer Soldatenfibel (Hakenfibel) aus Bronze weist zusammen mit anderen Indizien auf die Buntmetallverarbeitung im Lager hin. Auch die Herstellung von Melonenperlen aus Quarzkeramik ist durch entsprechende Funde, u. a. durch eine Anhäufung von mißratenen Stücken und ein Rohglasdepot, belegt.

Hanel N. 1998a Die Ausgrabungen im Lager der Classis Germanica in Köln - Marienburg (Alteburg) in den Jahren 1927/28. Kölner Jahrbuch 31: 351-400. Hanel N. 1998b Ziegelstempel aus dem Areal des Flottenlagers Köln - Marienburg (Alteburg). Kölner Jahrbuch 31: 401-415. Nuber E. 1984 Die Fundmünzen der Römischen Zeit in Deutschland (FMRD) VI 1. 1 Stadt Köln. (Berlin). Oschmann M. 1987 Köln - Marienburg. In H.-G. Horn (Hrsg.) Die Römer in Nordrhein-Westfalen. (Stuttgart): 516519. Päffgen B. & Zanier W. 1995 Überlegungen zur Lokalisierung von Oppidum Ubiorum und Legionslager im frühkaiserzeitlichen Köln. In W. Czysz (ed.) Provinzialrömische Forschungen. Festschr. f. Günter Ulbert zum 65. Geburtstag. (Espelkamp): 111-129. Schönberger H. 1985 Die römischen Truppenlager der frühen und mittleren Kaiserzeit zwischen Nordsee und Inn. Bericht der Römische-Germanischen Kommission 66: 321-497.

Eine Fundgattung, die bei den Altfunden und bei den laufenden Grabungen auftritt, fällt völlig aus dem Rahmen dessen, was in anderen Militärlagern an Fundgut auftrat. Es handelt sich dabei um große tönerne Webgewichte, die Bestandteile von Webstühlen für grobe Stoffe waren. In dieser Fundgattung, die in großer Menge ab der Mitte des 1. Jhs. bis zum Ende des 2. Jhs. belegt ist, zeigt sich anscheinend wirklich eine Spezialität der Flotte, die das Lager von Köln - Alteburg von anderen zeitgleichen Lagern und Kastellen der Landtruppen unterscheidet. Mir scheinen sie auch gut erklärbar: Die Antwort auf die Frage, was Textilherstellung in einem römischen Flottenlager zu suchen hat, geben zwei Grabsteine von der Alteburg, die für Segelmacher (velarii) der Classis Germanica gesetzt worden sind! Bei all diesen Belegen für Bein - Metall - und Glasverarbeitung sowie für Textilproduktion fällt auf, daß sie sonst bei zeitgenössischen Legions - und Auxiliarlagern eher nicht in den Lagern selber, sondern in den zugehörigen Zivilsiedlungen (canabae legionis und Kastellvici) anzutreffen sind. Insgesamt ergänzen diese außergewöhnlichen Belege für handwerkliche Produktion innerhald des Lagers von Köln - Alteburg das Gesamtbild des Flottenlagers in einem wichtigen Punkt: Neben seiner bisher postulierten Funktion als Stabsquartier und Truppenlager stellte es offensichtlich auch ein bedeutendes Zentrum für die Produktion und Verteilung von Nachschubgütern für die Classis Germanica dar. Literatur Bechert, T. & Willems W.J.H. 1995 Die römische Reichsgrenze von der Mosel und Nordseeküste. (Stuttgart). Carroll – Spillecke M. 1999 New excavations at the base of the Classis Germanica in Cologne (Alteburg). In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalău): 317-324. Carroll M. & Fischer T. 2001 Archäologische Ausgrabungen 1995/96 im Standlager der römischen Flotte (Classis Germanica) in Köln-Marienburg. Kölner Jahrbuch 32: 1999: 519-568. Fischer T. 2001 Neuere Forschungen zum römischen Flottenlager Köln - Alteburg In T. Grünewald & H. Schalles (Hrsg.) Germania Inferior. Reallexikon der Germanischen Altertumskunde Erġäblungs Band 28: 547-564.

911

Limes XVIII

Abb. 1. Köln-Alteburg, Plan mit Grabungsflächen 1995/96 und 1998. (Zeichnung A. Smadi, Arch. Institut Univ. Köln).

Abb. 2. Köln-Alteburg, Grabung 1998, Holzbauperiode IV. Gerastert: Grundmodul von fünf Raumeinheiten (Vier Wohnräume, ein Korridor) für zwei Contubernien. Schwarz ausgefüllt: Herdstellen. (Zeichnung A. Smadi, Arch. Institut Univ. Köln).

912

Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg (Alteburg) Norbert Hanel Excavations conducted in the second half of the C19th and at the beginning of the C20th in the headquarters of the German fleet at Alteburg near the Colonia Claudia Ara Agrippinensium (Cologne) uncovered interesting features which shed light on the development of its defensive system during the C1st AD. The first known fortification at the location was a common timber rampart; during the first half of the C1st AD structural difficulties caused the reinforcement of the rampart front. We can identify similar solutions to this problem at other legionary fortresses and auxiliary forts along the frontier of Lower Germany and in Raetia (eg. Valkenburg, Neuss Lager F, Munningen). Probably after the Batavian revolt in AD69/70, a stone wall was set in front of the timber rampart: this was detected during reexamination at the north and presumably also at the west front. It is the first time that this type of fortification has been documented for a fort in Germania Inferior. Possibly at the end of the reign of the emperor Domitian or at the beginning of the reign of Trajan, this intermediate defensive system was abandoned and replaced by a new stone wall which protected the fleet-base until its end in the last third of the C3rd.

At the XVIIth International Congress of Roman Frontier Studies in Zaláu I reported on the excavations carried out by F. Fremersdorf in 1927/28 in the fortress of the German fleet at Alteburg near Cologne in 1927/28 (Hanel 1999a).1 Since then excavation and investigation into the site continue (Carroll-Spillecke 1999: 317-324; Fischer this volume). The following paper summarizes the results of my investigations on the different fortifications (Hanel 1999b), which were recognised for the first time by Franz Wolf in 1887 (Wolf 1889: 12-19).

gateway in the north front (Fig. 2). The feature was built over, but not destroyed by the eastern gate tower of the intermediate phase, which will be discussed later. In the line of both foundation trenches a 6 post structure was encountered, which deviates at an angle of 30o from the axis of the rampart. The most northerly and the most southerly posts lie exactly in the line of the two trenches. The 6 postholes, which are arranged in two rows of three, were set at 1m apart, measured from centre to centre. In the longitudinal axis they have intervals of 1.2 to 1.4m. The pits are 0.4 to 0.5m in diameter. Measurements of the posts were not documented; they had lengths of approximately 0.25 to 0.3m. In the excavation area further postholes were found, but they cannot be connected with the 6-post structure (Hanel 1999b: 579). Their oblique alignment to the rampart does not fit into the scheme of Roman timber gateways (Manning & Scott 1979: 30-61). No doubt one can observe sporadically gates with recessed towers which show an oblique alignment contrary to the axis of the timber rampart. Such gateways are found in the legionary fortresses of Oberaden and Nijmegen-Hunerberg and at the municipal centre of Lahnau-Waldgirmes (Kühlborn 1992: 97 fig. 35; Brunsting 1961: 52, pl. 2; v. Schnurbein 1995: 306, fig. 5); however this oblique position never exceeds 10o. The interval between the posts is unusually close in comparison with other gateways in Roman military camps, but there is no reason to interpret them as a pile-grating. In such cases the piles are rammed in much closer and their number is much greater (Borrmann 1991: 24-25, 27). For these reasons I assume that this feature of the Alteburg must be connected to a gateway, even if the exact construction is uncertain. As mentioned above, the gateway of the intermediate phase was built over it. Moreover, a wooden drain passes northwards through the postulated gate (Lehner 1906: 254-255, pls. 13.2, 15.6; Hanel 1999b: 581-583, fig. 13). Perhaps a similar postsystem was completed symmetrically on the westside creating a funnel-shaped entrance.

During the examination of the Alteburg in the years 1905/06 Hans Lehner uncovered traces of the oldest fortification, especially the north front with a length of 110m, which can be reconstructed to 158m. The west front has a total length of about 220m. A connection with the north front could not be found (Lehner 1906: 257). The south front was detected for a length of about 87m. At the present stage one only knows the western area of the timber phase of the Alteburg; the whole eastern part remains unknown. It would be important to know if there were an eastern rampart running parallel to the Rhine, which has been proven for the stone phase. The oldest fortification consists of a single V-shaped ditch with an average width of 2.3m and a maximum depth of 1.65m. The slope rises at an angle of 45o. The bottom of the ditch – 2 to 2.5m in front of the rampart – was constructed squarely. The rampart is marked by two parallel trenches around 3m apart, which indicate that the front and back of the rampart were constructed in different ways. While Lehner observed in the foundation trench of the inner face impressions of posts standing at intervals of around 1.20m, no post-holes could be found in the outer foundation-trench, which means that the uprights were standing on a sill beam. Considering these features and using the analogy of other earth-and-timber ramparts (Holzerdemauer) I propose the following reconstruction (Fig. 1).

Concerning the date no finds were made which can be linked with the oldest defence-system of the Alteburg, but one has to consider that the earliest traceable rampart is not superimposed upon older structures. At present we can link

In the course of the oldest defence system there is only one 1

The author is grateful to Dr. C. Bridger, Prof. D. B. Saddington and B. Takats for their help.

913

Limes XVIII

the oldest finds from the site (coins, fibulae, samian ware) with this rampart, which can be dated to the second decade of the C1st AD (Päffgen & Zanier 1995: 127).

alteration would be more likely, as it prevents a collapse of the rampart into the ditch. Evidence for dating this reinforcement of the rampart front is missing; no finds can be assigned. In any case its construction can be dated before the Batavian revolt (AD69/70).

During Lehner’s excavations a remarkable structure was detected at the north and west front of the timber rampart (Fig. 3). At regular intervals of 1.2m there were features in the ground which looked like tongues filled up in part with traces of charcoal; they extend from the front into the berm. The length of these features varies between 0.6m and 1.2m, the width between 0.3 and 0.7m. Importantly these structures ascend at an angle of 8 to 10o towards the ditch. At the point of intersection with the front trench the lower edges of the cuttings lie 0.3m higher than the lower edge of the trench. Finally, the cuttings correspond with the positions of the vertical posts in the inner foundationtrench. The detailed analysis of these features leads to the conclusion that this was a buttress of the rampart front, which Lehner had recognised (Lehner 1906: 247-249, fig. 5). Nevertheless his reconstruction has to be modified in detail (Fig. 4) eg. the anchor beams were mortised with vertical posts of the rampart front above the sill beam. This is established by the fact that the lower edges of the anchor-beams lie 0.3m above the lower edge of the foundation trench; 2) the vertical uprights of the timberrampart are smaller and do not occupy the entire width of the trench, as the postholes at the inner face demonstrate; 3) in my opinion the slanting reached the level of the rampart-walk; 4) since the detached sloping buttresses would have represented weak points of the fortification, it is probable that the timbers were covered to prevent the enemy pulling them out. Therefore, I suggest a horizontal timber revetment; the space between the rampart front and this revetment was presumably filled with soil. Whether the revetment was also protected by turf is hypothetic.

The following period of the fortress of the classis Germanica at Alteberg forms a transition from the timber rampart to the stone fortification; for the first time we can detect this transition in Lower Germany. We know parallels in Britain, Upper Germany and Raetia, which will be discussed later. The transition period at Alteburg was documented in 1905/6 at the north and presumably at the west front; but its significance was not recognised by Lehner. Both the foundation trench and the tongue-shaped features of the rampart front were overlaid by a foundation of gravel 1-1.2m wide and 0.1m thick (Fig. 5). According to the drawings it was exposed over a length of 40m. A connection with the northern stone wall can be excluded, because the line of both features does not run parallel. The following reasons suggest that this rubble layer was the remains of a foundation of another defensive feature. Both structure and measurements correspond to the foundation of gravel of the younger stone-defensive feature. Moreover, both lower edges lie at the same level. Since the foundation of gravel of the intermediate phase overlies the rampart front and the tongue-shaped features, the front of the earth-and-timber rampart must have been demolished earlier. The extraction of the timbers would explain the different sizes of the cuttings. To sum up, a stone wall was built in front of the earlier fortification, while its core and rear were preserved. Because this added stone-facing follows the course of the earth-and-timber rampart, the fortress layout of the intermediate phase would correspond with the ground plan of the timber-phase; but one cannot exclude that the stone wall was built only in sections.

Timber ramparts with slanting buttresses have been found at different military sites in the northern provinces. The best parallel to the Alteburg in Lower Germany is the Tiberian camp at Novaesium (Neuss: v. Petrikovits 1960: 21-22, fig. 4). Here we find similar provisions at the rear of the rampart too. Furthermore traces of such buttressing of fortifications have been found at the Trajanic fort at Munningen in Raetia (Baatz 1976: 21, fig. 11).

More evidence for the construction of the intermediate phase can be detected at the northern gateway. A stone gateway was built in place of the previously mentioned, supposed timber gateway, of which only a few traces of the foundations were preserved. Nevertheless, these small remains make it obvious that the gate towers were not surrounded on all four sides by walls, but only on three lying next to the passage. The excavated feature can be explained best by the following presumption: during the construction of the intermediate phase only the terminals of the two earth-and-timber ramparts were surrounded on three sides by stone walls. Off-sets of the foundations at the abutments to the gate-towers with the walls indicate a certain interval in time between the building of the gateway and the wall of the intermediate phase. The recessed gate has a single carriageway 2.9m wide; central piers are missing. We have no indication how the upper portions of the gateway were reconstructed. This is not surprising, because the defence works of the intermediate phase must have been dismantled completely, when the defence system of the stone phase was built. The stone

At the Claudio-Neronian camp at Valkenburg (Period 2) the above mentioned problems occurred at the rear of the rampart; at the same time two slanting buttresses supported the rear posts (v. Giffen 1948: pl. 28). Lehner thought that he had found further examples of buttresses at Remagen and Vetera castra I, Xanten (Lehner 1906: 226-228, figs. 1-3; 1930: 31, fig. 19). A detailed study of these features demonstrates that his reconstructions are not valid in this respect (Hanel 1999b: 590-591). Based on the situation at Alteburg one cannot decide whether these struts of the rampart front were projected from the beginning or whether they were a later provision. No indications were found either at the fortress at Neuss or at the camp at Munningen. But I believe that a later 914

Norbert Hanel: Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg material was presumably reused to construct the new fortification.

of an irregular square or pentagon; the inner area measures some 7.1ha (Hanel 1999b: 616). In general the course of the stone wall varies only slightly compared with the older defences; the intersections indicate that for the building of the stone defence the former ramparts had to be demolished. For this period we have to mention some important new data. For this period the course of the eastern front could be located at several places on the slope of the Rhine. This means that the actual fortress area with its interior buildings was blocked against the riverfront. Since the C19th there have been different indications for connecting walls which run from the corner of the fortress neatrly right angled to the Rhine, protecting a harbour or better a landing-place for the fleet for a length of 300m (Wolf 1889: pl. 2; Höckmann 1998; 318-320; Hanel 1999b: 614).

As I noted before, in the northern provinces the phenomenon of an intermediate phase between the earthand-timber rampart and the stone defence stands not alone. At military sites such as Inchtuthil (Pitts & St. Joseph 1985: 60-73), Chester? (Frere 1984: 286-287, fig. 8), Gloucester (Hurst 1986: 104-106, fig. 43) and Lincoln (Webster 1949: 62, fig. 6) in Britain, Hofheim (Nuber 1974: 237-238, fig. 8) and Echzell (Baatz 1963-64: 35) in Upper Germany and Ellingen (Zanier 1992: 19-21, fig. 8) and Künzing (Schönberger 1963-64: 61, 74, fig. 11) in Raetia it could be demonstrated that the front of the earthand-timber rampart was substituted by a stone wall, while the rear of the defence and the agger of the previous phase were preserved. The intermediate phase of the fleet fortress at Cologne-Alteburg is the first traceable proof in Lower Germany.

The stone defence was built up as follows: on the outside a single V-shaped ditch with an average width of around 6m and a depth of 2.5-3m was excavated. The actual wall has a width of 80cms to 1m, which made an agger on the inner face necessary. The intervallum leaves sufficient spaces for an agger with an estimated width of 5m. Traces of a gateway were only found at the west front. In all probability a stone gateway was situated at the north front in the area of the gateways of the early periods, but during the excavations of 1905/06 no traces were found (Lehner 1906: 252; Klinkenberg 1906: 366). No towers could be observed in the course of the walls.

Concerning the dating of this defensive system different factors have to be taken into consideration. Beneath the northern gateway two samian cups of type Dragendorff 24 were found; their stamps with the name of Lucceius from La Graufesenque are assigned to Claudio-Neronian period and give a terminus post quem (Hanel 1995: 145). A second hint is provided by the layout of the gateway: although one has to keep in mind the unusual construction, the ground plan with the small off-sets in front of the walls shows that those gateways can be dated to the time of Vespasian. The best parallel to the north gate of the Alteburg is that of the stone defence of Wiesbaden, which was built in the time of Domitian (Ritterling 1909: pl. 2). Finally, one has to consider the general dating of the first defences made of stone in this region. Before the Batavian revolt in 69/70 no stone defences can be proved for Lower Germany; only after the reorganisation of the Rhine frontier under Vespasian did the rebuilding of the destroyed military camps start. In this way the first legionary fortress anywhere obtained their stone defences and one or two decades later the auxiliary forts. At the fleet fortress Alteburg one can perhaps reconstruct the following procedure: first of all the stone gateways were erected and later – presumably in a second step – the stone wall was placed in front of the old earth-and-timber rampart. Perhaps this already occurred in the time of Vespasian – if we take the samian ware as a terminus post quem – but in the time of Domitian at the latest. How long this intermediate phase of the fortification was in use cannot be estimated at the moment. I would suppose that it was planned as a short-term solution, but which was in use until the construction of the stone defence approximately at the end of the reign of Domitian or at the beginning of the reign of Trajan, ie. perhaps a duration of two decades.

While the building of the stone defences cannot be dated exactly – possibly during the 90s or at the beginning of Trajan’s reign, the latest coins found in the filling of the ditch place the end of the fortress of the classis Germanica in the last third of the C3rd. These coins are antoniniani of the Gallic usurpers Victorinus and Tetricus II in 270-274 and 272-280 respectively (Lehner 1906: 292). We know from the Historia Augusta (SHA Quadr. Tyr. 15) that in 281-282 the admiral of the classis, Bonosus, rebelled in Cologne against the Caesar Probus, but was defeated immediately. Perhaps one can connect this occurrence with the end of the fleet fortress at Alteburg (Fischer 1995-96: 90; Hanel 1999b: 624). In any case the collapse of the stone walls into the open ditch suggests that this event did not take place a long time after the withdrawal of the garrison (Fig. 6). It seems at the present state of investigation that no men were quartered in the Alteburg during the C4th. Bibliography Baatz D. 1963-64 Die Grabungen im Kastell Echzell 1962. Saalburg-Jahrbuch 21: 32-58. Baatz D. 1976 Das Kastell Munningen im Nördlinger Ries. Saalburg-Jahrbuch 33: 11-62. Borrmann M. 1991 Pfahlgründungen in der römischen Antike und im Mittelalter. In A. Hoffmann, E.-L. Schwandner, W. Hoepfner & G. Brands (edd.) Bautechnik der Antike Internationales Kolloquium in Berlin vom 15-17. Februar 1990. Diskussionen zur Archäologischen Bauforschung 5. (Mainz): 22-28.

In comparison with both predecessors, the course of the walls of the stone defences and consequently the ground plan of the fleet fortress in this period can be almost completely reconstructed; only the south-east corner is missing. The defence system encloses an area in the form 915

Limes XVIII

Brunsting H. 1961 De Nijmeegse castra. Resultaten van de opgraving in 1960. Numaga 8: 49-67. Carroll-Spillecke M. 1999 New excavations at the base of the classis germanica in Cologne (Alteburg). In N. Gudea (ed.) Roman frontier studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies 1997. (Zalǎu): 317-324. Fischer T. 1995-96 Grabungen im römischen Flottenlager Köln-Alteburg. Ein gemeinsames Forschungsprojekt des Römisch-Germanischen Museums und der Universität zu Köln. Universität im Rathaus 4: 83-93. Frere S.S. 1984 Roman Britain in 1983. Britannia XV: 286288. v. Giffen A. E. 1948 De Romeinse Castella in den dorpsheuvel te Valkenburg aan den Rijn (Z.H.) (Praetorium Agrippinae) I. De opgravingen in 1941. Jaarverslag van de Vereeniging voor Terpenonderzoek 25-28, (1940-1944): 1-316. Hanel N. 1995 Vetera I. Die Funde aus den römischen Lagern auf dem Fürstenberg bei Xanten. Rheinische Ausgrabungen 35. (Köln). Hanel N. 1999a Neue Forschungen zu den Grabungen im Flottenlager Köln-Alteburg der Jahre 1927-1928. In N. Gudea (ed.) Roman Frontier Studies 1997. Proceedings of the XVIIth International Congress of Roman Frontier Studies. (Zalǎu): 309-316. Hanel N. 1999b Die Umwehrungen der römischen Flottenlager Alteburg in Köln-Marienburg. Kölner Jahrbuch 32: 569-625. Höckmann O. 1998 Das Lager Alteburg, die germanische Flotte und die römische Rheinschiffahrt. Kölner Jahrbuch 31: 317-350. Hurst H. 1986 Gloucester. The Roman and later defences. Gloucester Archaeological Reports 2. (Gloucester). Klinkenberg J. 1906 Das römische Köln. In P. Clemens (ed.) Die Kunstdenkmäler der Stadt Köln. Die Kunstdenkmäler der Rheinprovinz VI.2 (Düsseldorf): 131-393. Kühlborn J.-S. 1992 Das Römerlager in Oberaden III. Die Ausgrabungen im nordwestlichen Lagerbereich und weitere Baustellenuntersuchungen der Jahre 19621988. Bodenaltertümer Westfalens 27. (Münster). Lehner H. 1906 Die Alteburg bei Cöln. Bonner Jahrbucher 114/115: 244-318. Lehner H. 1930 Vetera. Die Ergebnisse der Ausgrabungen des Bonner Provinzialmuseums bis 1929. RömischGermanische Forschungen 4. (Berlin). Manning W.H. & Scott I.R. 1979 Roman timber military gateways in Britain and on the German frontier. Britannia X: 19-61. Nuber H. U. 1974 Römisches Steinkastell Hofheim, MainTaunus-Kreis. Vorbericht über die Grabungen 19691970. Fundberichte aus Hessen 14: 227-250. Päffgen B. & Zanier W. 1995 Überlegungen zur Lokalisierung von Oppidum Ubiorum und Legionslager im frühkaiserzeitlichen Köln. In W. Czysz, C.-M. Hüssen, H.-P. Kuhnen, C.S. Sommer & G. Weber (edd.) Provinzialrömische Forsuchungen. Festschrift G. Ulbert zum 65 Geburtstag. (München): 111-129. v. Petrikovits H. 1960 Das römische Rheinland. Archäologische Forschungen seit 1945. Arbeitsgemeinschaft für Forschung des Landes Nordrhein-Westfalen Heft 86: (Köln). Pitts L. F. & St. Joseph J.K. 1985 Inchtuthil: The Roman legionary fortress. Britannia Monograph Series 6. (London).

Ritterling E. 1909Das Kastell Wiesbaden. Der Obergermanisch-Raetische Limes des Roemerreiches II B No. 31. (Heidelberg). v. Schnurbein S. 1995 Ausgrabungen in Waldgirmes. Bericht der Römisch-Germanischen Kommission 76: 304-307. Schönberger H. 1963-1964 Das Römerkastell Quintana-Künzig, Grabung 1962. Saalburg-Jahrbuch 21: 59ff. Webster G. 1949 The legionary fortress at Lincoln. Journal of Roman Studies 39: 57-78. Wolf F. 1889 Kastell Alteburg bei Köln. Geschichtliches Denkmal der ältesten Römerzeit am Rhein. (Köln). Zanier W. 1992 Das römische Kastell Ellingen. Limesforsuchungen 23. (Mainz).

916

Norbert Hanel: Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg

Fig. 1. Cologne-Marienburg (Alteburg): Rampart of the timber phase. Reconstruction.

917

Limes XVIII

Fig. 2. Cologne-Marienburg (Alteburg): Excavation 1905. a) Plan of the north front with gateways; b.) timber-phase; c.) intermediate phase. Scale 1: 200.

918

Norbert Hanel: Recent research on the fortifications of the headquarters of the classis Germanica: Cologne-Marienburg

Fig. 3. Cologne-Marienburg (Alteburg): Excavation 1905. The earth-and-timber rampart at the north front; ditch (c) and trenches of the rampart (a, b) with cuttings for the frontal buttresses (Bildarchiv Foto Marburg).

Fig. 4. Cologne-Marienburg (Alteburg): Rampart of the timber phase with frontal buttresses. Reconstruction.

919

Limes XVIII

Fig. 5. Cologne-Marienburg (Alteburg): Plan of the north front with defence-systems. 1) Stone phase (the outer face); 2) Intermediate phase (the outer face); 3) Timber phase (the outer face). Scale 1: 250.

Fig. 6. Cologne-Marienburg (Alteburg). Excavation 1927/28. Stone fortification at the south front. Sections of the stone wall have fallen into the ditch (Rheinisches Bildarchiv).

920

Now you see it, now you don’t. The British fleet in Vegetius IV.37 Boris Rankov A passage of Vegetius’ Epitome rei militaris (IV.37) has been used as evidence for the maintenance of a Roman fleet in British waters in the late C4th AD, in order to scout for coastal raiders. This fleet has also, on occasion, been associated with the defensive system of the Saxon Shore forts. This papers argues that Vegetius does not claim that Roman ships scouted off the coast of Britain, and that this would, in any case, have been impracticable. Comparison with Vegetius’ method of composition elsewhere suggests that his description of naval scouting activity, and his claim that scouting ships employed camouflage, were entirely his own invention, based on his own ignorance, the need to fill his own text, and unsubstantiated etymological deductions and extrapolations.

oars, the slightly bigger ones two ranks, those of the appropriate displacement three or four; sometimes they get five levels of oars. Nor should anyone think this abnormal, since it is recorded that much bigger ships fought in the Actian war, even of six ranks of oars or more. Scouting boats are, however, attached to the larger warships, with roughly twenty oarsmen on each side, which the Britons call picati [or ‘pitched’ or ‘painted’]. It is usual for surprise attacks to be carried and on occasion for convoys of enemy shipping to be intercepted by means of these ships, and for their approach or intentions to be detected by careful observation. Moreover, in order for the scouting ships to avoid being given away by the whiteness of their sails and rigging, these are dyed Venetian blue, which is like the colour of the sea-waves; the wax used to varnish ships is also dyed. And the sailors and marines wear Venetian blue uniforms so that they may the more easily remain undetected as they scout, not only by night but also by day.’

One of the best known pieces of evidence for a British fleet in the late Roman period occurs in Vegetius. Flavius Vegetius Renatus was a senior Roman official, possibly comes sacri stabuli, in charge of the procurement of horses for the emperor Theodosius the Great, who reigned from 379 to 395 (although some have dated him to the early years of Valentinan III, western emperor from 425 to 455). Vegetius’ military treatise, the Epitoma rei militaris, is essentially an armchair scholar’s thesis which argues that, in order to deal with the external threats to its existence, the empire must reform its army on traditional lines, in particular by reconstituting the old-style legion, the antiqua legio. Vegetius derives his knowledge of the ‘old’ Roman army from earlier military treatises going back to Cato the Elder in the C2nd BC, although his acquaintance with these may, in fact, have been entirely through epitomes, or abbreviated versions, of the many works he cites. There is no reason to believe that ever acquired any practical military experience himself. His work was nevertheless widely read as a practical manual in the medieval period, finally going out of fashion only in the C17th.1

The passage has regularly been taken as evidence that the late Roman navy in Britain maintained 40-oared scout ships to watch for and intercept sea-borne raiders. They have sometimes been associated with the Saxon Shore system of defences running from Brancaster on the Wash, round the coasts of East Anglia and Kent, and along the south coast of Britain as far as Portchester and the Isle of Wight (eg. Dove 1971; Johnson 1979: 132; Frere 1987: 210; Mann 1989: 10; Gauld 1990; Cotterill 1993: 232-234; Casey 1994: 160-161; Ireland 1996: 154 no. 262; etc.). These deductions are, I believe, wholly unjustified.

In the very last section of his work (IV.31-46), Vegetius discusses the Roman navy, and it is in one chapter (37) of this section that he supposedly makes reference to a British fleet patrolling (presumably) in the North Sea. The actual subject of the chapter in question, however, is types of warships (De modo liburnarum): Quod ad magnitudinem pertinet, minimae liburnae remorum habent singulos ordines, paulo maiores binos, idoneae mensurae ternos vel quaternos; interdum quinos sortiuntur remigio gradus. Nec hoc cuiquam enorme videatur, cum in Actiaco bello longe maiora referantur concurrisse navigia, ut senorum etiam vel ultra ordinum fuerint. Scaphae tamen maioribus liburnis exploratoriae sociantur, quae vicenos prope remiges in singulis partibus habeant, quas Britanni picatos [or picatas or pictas] vocant. Per has et superventus fieri et commeatus adversariorum navium aliquando intercipi adsolet, et speculandi studio adventus earum vel consilium deprehendi. Ne tamen exploratoriae naves candore prodantur, colore veneto, qui marinis est fluctibus similis, vela tinguntur et funes; cera etiam, qua ungere solent naves, inficitur. Nautaeque vel milites venetam vestem induunt, ut non solum per noctem sed etiam per diem facilius lateant explorantes.

It must first be pointed out that Vegetius does not say that such patrolling was carried out off the coast of Britain. He simply states, parenthetically, that the Britons call the 40oared scout vessels which he has just described picati. Whilst he may imply as much, it is not necessary to locate the sort of interceptions and the use of camouflage he describes specifically off the coast of Britain, and there is certainly no justification for associating them with the Saxon Shore. Small fleets may well have been attached to the Saxon Shore forts, as is indicated by the Pevensey Fleet (classis Anderitianorum) recorded as having been transferred to Paris in the late C4th document known as the Notitia Dignitatum (Occ. 42.23). But neither the text of Vegetius nor any other evidence implies that such fleets were composed of picati. Vegetius here is simply making general comments on the how scaphae exploratoriae are employed, and Britain should be left out of the discussion.

‘As for size, the smallest warships have single ranks of 1

On Vegetius in general, see Milner 1996: xiii-xliii, citing the most important earlier work.

921

Limes XVIII

Secondly, for practical reasons, the sort of operation described by Vegetius is inherently implausible. It is very difficult to conceive of how these ships are meant to blend into their marine surroundings and lie in wait, watching for raiders or enemy fleets, ready to pounce upon them out of the blue, as it were. The difficulty is that oared warships cannot cruise or patrol over any useful distance or even stay at sea for prolonged periods. The need for intensive manning of the long narrow hulls means that provisioning, sanitation and most of all watering become problematical after even half a day at sea. Nor can large numbers of oarsmen easily move around in such hulls even when under sail, both because of the limitations of space and the problem of keeping the boat trimmed when the crew are the main ballast. Moreover, after a few hours of continuous rowing, or even of resting under sail in cramped conditions, physical fatigue will limit any oared warship’s ability to pursue and engage an enemy, even if that enemy is himself suffering from the same conditions.

being viewed downsun (as for instance from the southern i.e. Gallic coast of the Channel). Otherwise, even white sails appear dark against the sky. Similarly, any hull, of whatever colour, appears dark against the sky, and as Tavender notes, human figures would be more or less imperceptible at the sort of distances where one might expect camouflage to work. Vegetius’ comments on the colouring of rigging, crew and hull would therefore appear to be nonsense, although Tavender actually argues for the sort of camouflage Vegetius describes to be most effective at night, in lightening a dark silhouette against a moonlit sky. At any rate, the potential effectiveness of Vegetius’ camouflage is dubious for most daytime conditions, and some of his details are certainly wrong. That what Vegetius describes may be wholly impractical should raise concerns about the accuracy of his account. But there are, thirdly, some good historiographical reasons for taking the whole passage with a very large pinch of salt. Vegetius’ method in writing his history is to make use of the epitomes of a limited number of earlier military manuals, which he expands rhetorically. This frequently involves repetition, either because his different sources were covering similar ground, or simply through his fondness for elaboration and abundantia in his writing. It is clear, moreover, that Vegetius does not present a clear, consistent and accurate picture of the antiqua legio, but puts together elements of the army of different periods drawn from fragmentary information in diverse sources, which themselves reached him at second hand. Nor was Vegetius particularly concerned about this, because his purpose was not historical but to advocate a model which would best serve, in his opinion, for the reform of the late C4th army. All this is well known to scholars of the Roman army, who make use of Vegetius with extreme caution and make allowances for his tendency to fill out his text with rhetorical and propagandistic invention. Furthermore, as his most recent editor, Nicholas Milner, says, ‘he largely supplemented gaps in the records available to him by intelligent conjecture based on etymologies of old military titles and knowledge of contemporary military institutions” (1986: xxviii). Thus he assumes, for instance (II.13), that infantry centuries in the antiqua legio were made up of 100 (rather than 80) men. He then claims that each century was divided into 10 man sections, each one of which was known as a manipulus because the ten fought in manus (handfuls of five) joined together. In reality, of course, a manipulus had been a grouping of two centuries.

In fact, very few naval engagements in the ancient world took place in the open sea; the vast majority occurred close to shore between fleets which had only just put to sea from bases close to each other (eg. Actium). This is because oared fleets simply cannot lie in wait for each other out at sea. Even if the object is not to make a sudden attack supervenire - or intercept - intercipere - but simply to observe - speculare - and detect enemy intentions consilium deprehendere - the same limitations apply to even a single oared warship either cruising up and down under oar or sail, or just sitting and waiting.2 A much more likely scenario for a fleet such as that sometime based at Pevensey would have been either to lie in wait for raiders to pass by, and attack them with fresh crews from harbour, or to go looking for them by moving in stages along the coast, perhaps as part of a combined naval and military operation. This, according to Sidonius Apollinaris, is what his friend, the admiral Namatius, did along the Atlantic coast of Gaul in search of Saxon raiders whilst in the service of his Visigothic overlords towards the end of the C5th (Tavender & Dove 1972).3 Vegetius’ description of the use of camouflage is also open to question. Vegetius appears to be most worried about the whiteness of sails and advocates colouring them ‘Venetian blue, which is like the colour of the sea-waves.’ As Tavender pointed out in Antiquity in 1972 (drawing on experience of aircraft camouflage, particularly in WWII: Tavender & Dove 1972), this would serve to conceal the sails against the sky (not the sea) only when the sails were

When discussing the navy, he assumes that the two fleets at Misenum and Ravenna were organised like legions, or at least his reconstruction of the antiqua legio. Thus, he claims that they were each commanded by a praefectus (which they were) and organised into 10 cohorts, each commanded by a tribune (which they were not). He also claims that each ship was commanded by a nauarchus, although in truth this was the title for a squadron commander, a ship’s captain normally being ranked as a trierarchus (IV.32, cf. II.12). Vegetius’ approach to the

2

On the logistical problems of travelling on the high seas under oar, see Green 1963: 103-113. For the limitations of naval operations and engagements under oar, see Gomme 1933: 16-24; Rankov 1996: 49-52. 3 Letters VIII.6.13: (nuntius) asseveravit nuper vos classicum in classe cecinisse atque inter officia nunc nautae, modo militis, litoribus Oceani curvis inerrare contra Saxonum pandos myoparones, quorum quot remiges videris, totidem te cernere putes arcipiratos...; it is only a few lines later, however (VIII.6.15), that Sidonius notoriously refers to the Saxons as ‘setting sail’ (vela laxantes) a couple of centuries before we have archaeological evidence for the Germanic use of sail.

922

Boris Rankov: Now you see it, now you don’t. The British fleet in Vegetius IV.37

navy is in fact entirely in line with his approach to the old Roman army as a whole, and just as inaccurate and conjectural in detail. There is therefore every reason to approach the chapter under discussion with the normal caution. That this has not been done is in part due to the relative ignorance of the army specialists, who normally deal with Vegetius, in naval matters. Even Milner, who is usually highly sensitive to Vegetius’ weaknesses, is prepared to accept the central portion of the naval section, from chapters 33 to 42, as relatively well-informed (Milner 1986: xxvii). Most of those who have used the passage have not, however, even been army specialists, but archaeologists and historians interested in Roman Britain and only too grateful to take on trust a rare, apparently priceless, glimpse of C4th Britain from a written source. I have already suggested that they are wrong to connect the details of scouting and camouflage with what is really just a parenthetical mention of the Britanni. I want now to go further and suggest that the whole chapter is essentially worthless.

ships (constratis navibus) and several smaller (minoribus) vessels which were either open, beaked ships (apertae rostratae) or unbeaked scout ships (sine rostris speculatoriae; XXXVI.42.8). During the first landing in Britain in 55BC, Caesar filled his ship’s boats and scout ships (scaphas longarum navium, item speculatoria navigia) with troops in order to reinforce those who had already got ashore (IV.26.4). And in the African War (XXVI.3-4), Caesar sends a catascopus from Ruspina on the African coast to take instructions to his lieutenants in Sicily, expecting it to return within two days. In only in one of these cases, the earliest historically, do speculatoriae carry out the function implied by their name, that is scouting or reconnaissance. In the other passages, they are merely light, fast and unarmed oared vessels. As for Vegetius’ reference to picati, I have already noted that in the text his comment appears to be added as an afterthought. The word itself is also a textual crux. Picatos, which only appears here, is the reading of five of the manuscripts, but picatas (‘covered in pitch’ or ‘resin’) is in another four and pictas (‘painted’) in four more. The last reading in particular has often found favour with modern interpreters, but in terms of textual criticism, picatos has much the strongest claim as the lectio difficilissima. The last two variant readings do, however, give a hint as to how to interpret the last part of the chapter. I have argued that the practicality of Vegetius’ remarks about patrolling and camouflage at sea is questionable. I have also demonstrated that he is as ignorant about the old Roman navy in detail as he is about the rest of the army, that it is a frequent part of his method to fill out his ignorance by etymological deduction, and that he would appear to be doing this even at the beginning of this chapter where he makes false deductions about levels of oars in earlier warships. What I therefore wish to suggest is that Vegetius in fact had no information whatsoever either about how scaphae exploratoriae operated or about what colour they were. All he knew was that in addition to the single-level and two-level major warships which still existed in his day, the Roman navy had in the past included threes, fours, fives, even sixes and bigger; also that these had had their own small boats (scaphae) and had sometimes been accompanied by light vessels known as catascopi, naves speculatoriae, or the like (as in Caesar). He also knew that the Britons used vessels known as picati.

That Vegetius is including some invention in this chapter is clear from the first section, where he suggests that proper warships - warships idoneae mensurae - have three, four, even five, ranks (ordines) or levels (gradus) of oars. He counters his readers’ scepticism about this by pointing out that ships of six ranks (ordines) and larger were recorded at the Battle of Actium. The reason why Vegetius has to make this point is that in his day, even three-level ships were probably wholly unknown. The last recorded use of triremes is in Constantine’s civil war against Licinius in 324 (Zosimus Nova Historia 2.22-28). By the end of the C5th, Zosimus can say that the secret of building such ships has been lost long since (ibid. 5.20). The sea-going warships which Vegetius knew - liburnae as they were then known - had only one or two ranks, so he deduces, presumably from the names he found in his sources and as many have deduced since, that triremes had three levels, quadriremes four, qinqueremes five, and so on, although it is actually impossible to design a viable oarsystem at more than three levels. This brings us to the key sentence in the chapter, the one which refers to scout boats - scaphae exploratoriae - being associated with the larger warships. The phrase is unparalleled in Latin literature as it stands, but it is not difficult to recognise a reference here to the vessels which appear Livy and the Caesarian corpus as (naves) speculatoriae, navigia speculatoria or catascopi (the latter term obviously derived from the Greek kataskopos, meaning a scout). Livy, in a passage derived from Polybius, speaks of two Massiliot speculatoriae sent by Cn. Scipio to operate ahead of the Roman fleet moving down the coast of Spain in 217BC, and of their finding the Carthaginian fleet anchored at the mouth of the Ebro (Livy XXII.19.5; cf. Polybius III.95.6 and VIII; 96.1). Later on, he describes the crossing of the praetor of 191BC, C. Livius Salinator, from the Piraeus to Delos with 81 decked

The rest of the chapter is merely Vegetius’ own extrapolation of the last two pieces of information. ‘Scout’ boats are made by Vegetius to operate like the superventores units of the late Roman army (Notitia Dignitatum Or. XXXIX.21; Occ. V.120 = 270 = VII.96; XXVII.18; XL.31; cf. Ammianus XVIII.9.3) and like the scouting cavalry and light troops which he discusses earlier in the Epitome (III.6). The job of the latter is ‘to scout really carefully…so that we may discover the enemy’s present and future plans’ and ‘to catch them in an unforeseen ambush while marching or seeking fodder and

923

Limes XVIII food;’4 ‘scouts, moreover, operate more safely at night than in daytime.’5 The idea of camouflage, on the other hand, is prompted by the word picatos (which may or may not be associated with the Picts6) which may have suggested to him both the treatment of the hull with ‘wax’ (picatas) and the painting itself (pictas). The fantasy would have been reinforced by the common knowledge that sailors did indeed (as it would appear) wear light or dark blue uniforms.7 A possible further indication that this was Vegetius’ own invention is that the shade of blue to which he refers, Venetian, is that which would have been best known to his metropolitan readers, namely the colour of the ‘Blue’ faction in the circus. ‘Venetian’ was their particular shade, just as the ‘Greens’ made use of a particular shade of leek-green (prasinus).

Rankov N.B. 1996 The Second Punic War at sea. In T. Cornell, B. Rankov & P. Sabin (edd.) The Second Punic War. A reappraisal. (London): 49-57. Tavender L. & Dove C.E. 1972 Camouflage in the first British navy. Antiquity 46: 320-322.

I conclude, therefore, that Vegetius’ account should be abandoned as mere invention, and that it is entirely worthless as evidence for the operation of the British or any other Roman sea-going fleet. Bibliography Casey P.J. 1994 Carausius and Allectus: the British usurpers. (London). Chadwick N.K. 1958 The name Pict. Scottish Gaelic Studies 8: 163-166. Cotterill J. 1993 Saxon raiding and the role of the late Roman coastal forts of Britain. Britannia XXIV: 227-239. Dove C.E. 1971 The first British navy. Antiquity 45: 15-20. Frere S.S. 1987 Britannia: a history of Roman Britain. (3rd ed., London). Fuentes N. 1987 The Roman military tunic. In M. Dawson (ed.) Roman military equipment: the accoutrements of war. Proceedings of the third military equipment research seminar. British Archaeological Reports Int. Ser. 336 (Oxford): 41-75. Gauld W.W. 1990 Vegetius on Roman scout boats. Antiquity 64: 402-406. Green C. 1963 Sutton Hoo. The excavation of a Royal ship burial. (London). Gomme A.W. 1933 A forgotten factor of Greek naval strategy. Journal of Hellenic Studies 53: 16-24. Ireland S. 1996 Roman Britain. A sourcebook. (2nd ed., London). Johnson S. 1979 The Roman forts of the Saxon Shore. (2nd ed., London). Mann J.C. 1989 The historical development of the Saxon Shore. In V.A. Maxfield (ed.) The Saxon Shore. (Exeter): 1-11. Milner N.P. (ed.) 1986 Vegetius: Epitome of military science. (2nd ed., Liverpool). Morris J. 1973 The age of Arthur. (London). 4

Vegetius III.6.34: sed explorare sollicite verum…ut, quid hostis moliatur in praesenti vel in futurum, possimus agnoscere…ambulantes eosdem vel pabula victumque quaerentes inproviso terrore decipere. 5 Vegetius III.6.12: tutius autem operantur exploratores noctibus quam diebus. 6 cf. Chadwick 1958. Morris 1973: 16-17 argues, quite rightly, that the C4th towers on the north Yorkshire coast should have acted as defences against raiders coming down the coast from the north, presumably Pictish sea-raiders. 7 The evidence comes from the Barberini and the Piazza Armerina mosaics: see Fuentes 1987: 41-75.

924

The Roman army in the Vindolanda Tablets A.R. Birley This paper provides a review of the following two themes: 1: The extent of the tablets so far discovered, their nature and the state of publication so far 2: Writing-tablets which throw light on Roman military practice will be discussed, in particular the strength report of coh. I Tungrorum; the renuntia of coh. VIIII Batavorum; the commeatus requests; the ratio frumenti showing the presence of legionaries at Vindolanda and other evidence for a mixed garrison.

The discovery of the Vindolanda writing-tablets, of which Robin Birley found the first example in March 1973, has been related often enough and there is no need to go into detail here.1 The excavations under his direction which produced this find had been in progress since 1969, from 1970 onwards under the aegis of the Vindolanda Trust, an independent, self-financing charitable institution,2 in the vicus on the west side of the stone fort that was garrisoned from the C2nd to the late C4th (and probably occupied in some form for a good while after that). The tablets proved to derive from some five successive periods of occupation preceding that stone fort. As is now well known, the tablets in question are different from the familiar thick wooden ones designed for writing by incision with a stylus on the wax with which the hollowed out part was filled. The Vindolanda ‘leaf-tablets’ are far thinner, mostly 1-2mm thick, almost all made of birch or alder, and were written on with ink, with an iron-nibbed pen, at least in some cases - several such nibs have been found, two with a wooden barrel still in place (these nibs have frequently been described in the past as ‘ox-goads’). Only a few examples of such leaf-tablets had previously been found. Happily, some significant examples have now been recovered - and published - from other sites, notably Carlisle-Luguvalium (Tomlin 1988).

earliest garrison. Notable is a strength report (TV II 154) unusually, of oak, and much larger than any other tablets of the Tungrians, found in the Period I fort west ditch (excavations in 2000 have now located the south ditch of the Period I fort, of which no structures have yet been found). It shows the size of the unit, commanded by the prefect Julius Verecundus, to have been 752 men on 18 May in an unnamed year. In other words, it was almost large enough to be a cohors milliaria peditata, although with only 6 centurions, as in a cohors quingenaria; but well over half the men, 456, including 5 centurions, were absent elsewhere, while only 295 men and one centurion were at Vindolanda. The cohort had perhaps just been upgraded to be a cohors milliaria. This fort’s date, largely based on a substantial quantity of terra sigillata found in its west ditch, can hardly be earlier than c.85.4 Cohors I Tungrorum was replaced c.92 by coh. VIIII Batavorum equitata, which cleared the site, filled in the western ditch of the Tungrians’ fort and erected above it buildings for its own central range.5 This should mean that the Batavians’ fort, soon to be at least partly reconstructed by the same regiment, was something like double the size of its Tungrian predecessor. The via principalis to the west of this range, and the south gate, were also excavated. The filling-in of the Tungrians’ ditch was not very successful and in this part of the site there were repeated attempts in successive periods, with boulders, gravel and puddled clay, to achieve a dry and stable foundation. These circumstances - wet material, with a series of impervious seals above it - largely contributed to the preservation of the wooden writing-tablets and of much other organic material.

First, a brief résumé of the dating. Four Batavian and two Tungrian cohorts played a key rôle in Agricola’s army at the Battle of the Graupian Mountain in AD83 (Tacitus Agricola 36.1). Two of the Batavian cohorts and one of the Tungrian cohorts are discovered, appropriately enough, to have been stationed in close proximity, or in close succession, to one another, at Vindolanda, over the next four decades or so. In the five periods of occupation from which the tablets derive,3 coh. I Tungrorum was the

and beginning of V remain a little certain: 120+. 4 This seems to be the earliest date to which one can assign Period I finds. It might be expected that material from at least 10 years earlier than c.AD 85 might have been found at Vindolanda, given that Carlisle-Luguvalium is now known (by dendrochronology) to have been founded c.72-3: cf. Caruana 1989: 25; 1992: 45, 103; forthc.; Groves 1990. The answer may be that an earlier fort at Vindolanda lay slightly to the north of the known ones, in an area not yet examined archaeologically: air photographs by the late Barri Jones and a resistivity survey carried out for the Trust in 2000 show significant traces in this area. 5 TV II 396, from a Period II level, reading c(o)hor(tis) VIIII[ ], is the clearest tablet evidence for VIIII Batavorum being the garrison. 505, also Period II, is evidently a letter from the decurion Masculus, who must be the same man as the decurion Masclus who wrote Inv. 1544 to his ‘king’ Cerialis in Period III (Bowman & Thomas 1996: 323ff., cf. below. Note also the inscription cut on a piece of oxhide found in the watertank of the Period II praetorium, CIXB, ie. c(ohortis) IX B(atavorum), matching three similar examples from Period III: R. Birley 1993: VRR II 92, no. 3.

1

See, for example, R. Birley 1977. Robin Birley’s subsequent excavations have of course led to various modifications in interpretation as well as to a great increase in the number of tablets. See for example the Vindolanda Research Reports (VRR), new series, I (1994), II-III (1993), esp. vol. I. Unpublished tablets are prefaced by ‘Inv.’ 2 The Trust has benefited over the years from grants from various public bodies, of which the Sir James Knott Trust deserves to be singled out. But its principal income derives from the admission fees to the site and from sales in the museum shop and café. Since 1998, the Trust has also taken over the running of the Roman Army Museum at Carvoran fort on Hadrian’s Wall. 3 For the occupation periods, see R. Birley 1994, with the slight modification on the end of Period III in the light of Inv. 1474A noted by Bowman & Thomas 1996: 311. I now incline to date the end of Period III, marked by the departure of the Batavians, to summer 105. The end of IV

925

Limes XVIII

Probably c.97, a few years after the Batavians first arrived, their Period II fort was substantially reconstructed, but on much the same plan. A great many tablets attest the presence of coh. VIIII, and over a hundred name its prefect in Period III, Flavius Cerialis; and several also name his wife, Sulpicia Lepidina. One of coh. VIIII’s sister units, coh. III Batavorum, may also have been based near, if not at, Vindolanda: it is named in two tablets from Period III. One, a letter from the decurion Vitalis to the prefect of coh. VIIII, Flavius Cerialis, refers to a previous letter delivered to Cerialis by Equester, centurion of coh. III Batavorum; the other is a letter to Paris, presumably a soldier, in coh. III Batavorum (TV II 263 and 311).6

There was a further reconstruction at about this time, labelled Period V, the barrack block being replaced by workshops and a large courtyard building. Excavations now in progress may modify some of these details, especially the relationship of Periods IV to V, as well as of Period V to that of the first stone fort, Period VIA. To summarise: Period I Period II Period III Period IV Period V

Coh. VIIII Batavorum remained here until, probably, at least until summer 105, when it was withdrawn from Vindolanda, and indeed from Britain - never to return evidently to take part in the 2nd Dacian War. Period III is dated partly by writing-tablets, one (TV II 225) naming the governor Neratius Marcellus, attested by a diploma as being in office in January 103, the other (Inv. 1474A), published by Bowman & Thomas (1996: 307-323) giving consular dates for 103 and 104 (cf. further below). Some months after the Batavians had departed, coh. I Tungrorum returned, thus inaugurating Period IV; and the Tungrians remained here probably until at least the 140s. The Batavians’ fort was drastically altered and a double barrack-block was erected where the praetorium of coh. VIIII Batavorum had been. One tablet from this 4th period, an account, is dated to AD 110-111 by the consuls of the latter year (TV II 186). During Period IV, perhaps near its end, there were other soldiers at Vindolanda apart from the Tungrians: legionaries7 and the equites Vardulli, presumably the cavalry section of the coh. I fida Vardullorum equitata.8 A draft letter (TV II 344) on the back of the account naming legionaries (who, it may be conjectured, were housed at Vindolanda in connection with the beginning of the Wall) allows the inference that a visit from the emperor - Hadrian, who came to Britain and started his Wall in 122 - was imminent (the writer appeals to ‘Your Majesty’, to whom he evidently hoped to hand his petition).9

coh. I Tungrorum, c.85-92. coh. VIIII Batavorum, c.92-97. coh. VIIII Batavorum, c.97-105. coh. I Tungrorum, plus elements of other units, c.105-120+. coh. I Tungrorum, c.120+-140s+.

The ink-tablets from the 1985-9 excavations have been (virtually10) all published, by A.K. Bowman and J.D. Thomas, with contributions by J.N. Adams (1994), for which the abbreviation recommended by the editors is: TV II. In the same volume they republished the tablets found in 1973-5, which had previously appeared in their Vindolanda: the Latin Writing-Tablets (1983). This represents a total of 426 texts.11 Naturally some revision to the edition of 1983 was possible in the light of new finds. Further modifications will obviously be needed.12 The stylus tablets, of which a good many were also found, most of them extremely difficult to read, have yet to be deciphered or at any rate to be published.13 A good many more ink- or leaf-tablets, from the excavations of 1991-1994, are still unpublished. The Inventory numbers run from 998-1617, at first sight 620 separate tablets. In some cases several numbers were assigned to pieces from one original; but conversely - and more frequently - separate tablets were taken at first to belong together. At first sight the initial total is thus 1997a: 135. 10 A fair number of the tablets from the 1985-9 excavations were not worth photographing or are so abraded that nothing legible remains. Over 80 stylus tablets were also found, a few of which have already been mentioned in preliminary publications but of which the editing is still outstanding. Of the ink-tablets from 1985-9 not included in TV II, one might note Inv. 218 (perhaps reading optio..Feli--, hence possibly from a renuntium); 251 (?tu...em); 438 (?val[ ]); 762 (? [ ] et..eni); 846 (?et primo et fere.r...); 952, some six lines are partially legible, eg. ?.co tibi...sed...ut; 953, ?rogo adsumes•ut eum[ |constrictum per..- this is an improved version of the reading offered by A. Birley 1999: 37 n.2. 11 The numbers assigned by the editors of TV II begin with 118 and run to 573. 12 For example, in TV I the Ninth Cohort of Batavians was taken to be the Eighth Cohort; and neither the renuntia nor the prefect Cerialis’ wife Lepidina could be identified for what they really were. As for modifications to TV II in the light of subsequent finds, note the formula in the renuntia, qui debunt (previously understood as q(ui) videbunt); and coh. VIIII can now firmly be identified as equitata because the decurio Masclus was under Cerialis’ command; and, two minor examples (both TV II 281), one can now eliminate Fl(avius) Petrus - instead one can restore a decurion of the ala Petriana as the writer - and the name of the recipient should be read as Cluvio Floro, not Cluvio Fabro. 13 At best some of the addresses, with names of persons and places, can be read without too much difficulty (but note the corrections offered in TV II 26 n.22 and 364f., to some of the present writer’s readings). Of the new stylus tablets, one may note that Inv. 1228 has an at least partially legible ink rim-inscription, registering that a soldier called Billo owed money.

6

The address in TV II 311, while fragmentary, might be read, as the editors suggest, P̣ạṛịḍị Ụḷục̣ịọ c̣ạṣ[tris coh]ọrtis III Batavorum. In that case, III Batavorum was stationed not at Vindolanda but at Ulucium, a place not yet identified but probably not far away, at which a soldier (probably of VIIII Batavorum) requested to spend his leave, TV II 174. All the same, the presence of TV 311 at Vindolanda suggests that its recipient, Paris, was there at some time. One might read the symbol for centurion, 7̣, after his name in the address, although the ink mark is perhaps part of the Ụ. The men greeted by the writer Sollemnis--Diligens, Cogitatus and Corinthus - may be taken as further soldiers in III Batavorum. The rank or status of Sollemnis was evidently given in the address, but is unfortunately not legible; the editors ‘doubt whether’, after a Solleṃṇị, ‘.......(con)s]ẹṛṿọ can be read’. Some military rank is another possibility. 7 TV II 180, a Period IV ratio frumenti eṃ[ensi], records distribution to militibus legionaribus (sic). Some of the men named in this account were probably legionary centurions. 8 TV II 181, an account from Period IV, names equites Vardulli among those owing money: the vexsillariustis Tagamatịṣ, whose name appears directly after the Vardulli, probably belonged to these cavalrymen. 9 Cf. for this interpretation (not shared by the editors of TV II), A. Birley

926

A.R. Birley: The Roman army in the Vindolanda Tablets increased, to just over 700. However, some 100 apparent tablets which had been given inventory numbers proved after conservation not to be writing-tablets, and some 125 of them are stylus tablets, only six of which have also some ink writing on them. This leaves 485 leaf-tablets, designed for ink, of which about 200 are blank, illegible or show only tiny traces of lettering. There remain over 250 leaf-tablets from 1991-1994, with varying amounts of text, of which, so far, only 7 have been published. For four of those found in 1991-2 (most of which were fragmentary: Inv. nos. 91/998-92/1206) see Birley & Birley 1994: 440ff., Inv. 91/1022, is a (?draft) letter from one Major to Cocceiius (sic) Maritimus (Period IV), dealing with matters involving the Caesariani, ie. imperial slaves and freedmen from the familia Caesaris, and written while the writer was ‘making the bed warm’. On the significance of the address, Vindoland(a)e, cf. remarks below 437ff.. 91/1091A (Period II), is a list of names from part of an account. 431ff., 92/1108 (Period II), is a partially legible letter about supply of bracis, a cereal used for making beer, including mention of ‘wagons of the Britons’, and the technical terms vecturae, ‘transport fees’ and velatura, ‘transport business’ (a rarely attested word). It includes two names assumed to be of members of the Period II garrison, coh. VIIII Batavorum. 435ff., 92/1187 (Period II), is page 2 of a letter (with most of the address on the back) from Ascanius corniclarius to a man whose name is missing, described as mensor, evidently referring to the conscientiam praefecti and greeting omnes cives et amicos, of whom four are named, one a woman.14

This tablet requires the departure of coh. VIIII Batavorum equitata, and thus the end of Period III, to be put slightly later than previously thought, probably in the summer 105. A number of other names of persons, some of them members of Cerialis’ household or of the garrison, are mentioned. 323ff., 93/1544 is a letter from the decurion Masclus to Cerialis, whom he addresses as regi suo, ‘his king’, and whom he asks to be mihi propitius, before adding a PS, ‘the comrades have no beer, I ask that you order some to be sent’, cervesam commilitones non habunt, quam rogó iubeas mitti. This usefully confirms that the Ninth Batavians drank plenty of beer, and that they were a cohors equitata. Further, Masclus’ usage habunt instead of the normal habent convinced the editors that qui debunt in the renuntia (the laconic reports by the optiones and curatores, omnes ad loca qui debunt et impedimenta, ‘all at their place who ought to be and the equipment’) was a variant for qui debent (rather than an abbreviation for q(ui) videbunt, as this phrase was interpreted in TV II). The editors take ‘king’ to be standard flattery; still, some have a residual wish to see Cerialis as ‘king of the Batavians’, cf. further below. Bowman and Thomas also offer some useful improvements, 326ff., to the text of the Major letter (Inv. 91/1022, cf. above), in particular reading si itá instead of spicá in line 7, which makes much better sense;15 and they will undoubtedly have further improvements to offer on the other tablets published by Birley and Birley. It was hoped that their work on the still unpublished ink-tablets would be complete by the end of 2000 and that TV III might appear in 2001 or 2002. (Latest information suggests that this was over-optimistic).

Of those found in 1993 (Inv. 1207-1566), three important examples (all Period III) have been published by A.K. Bowman and J.D. Thomas (1996: 300ff.), 93/1398 is an account on two sheets (the reverse of which is too abraded to read) with interesting commodities, including infiblatoria (cloaks with attached brooches, a word hitherto known only in Greek transliteration from a Trajanic inscription at Pessinus), corticia (thought to mean clothing made from tree bark), and a scordiscus; with prices both per item and totals; and an unexampled set of symbols for denarius fractions. No names of persons are included. 307ff., 93/1474A is the expensa of Flavius Cerialis, not named, but clearly identifiable from the entertainment of known friends and correspondents of his, fellow-officers, including (Aelius) Brocchus, (Oppius or Valerius) Niger and (Caecilius) September, and simply because it derives from Period III, indeed from a bonfire of a pile of rubbish representing a clear-out at the end of this period. In a total of 112 lines, spread over five sheets, four written on both sides, it gives two consular dates, AD103 (Traiano v) and 104 (Sex. Attio Subur[ano]); refers to the arrival of the governor, adventu consu[laris]; a legate; consumption of considerable quantities of poultry (pulli and anseres); at least one religious festival (the Matronalia, 1 March); a brewer (cervesar[io]); a veteran; and much else besides.

The editors of TV II have a very full discussion of what the tablets have revealed about the Roman army and many other matters, not least palaeographical. This has been supplemented by Adams (1995). He discusses in detail questions such as grammar, syntax, spelling, vocabulary, punctuation and accents (the apex). It is important to register his conclusion that the general literacy of army personnel, not least in spelling and grammar, was of a high standard. One exception, debunt and habunt instead of debent and habent, was produced by ‘a social class (probably that of the optiones) who regularly used [a] substandard form, yet were literate’. The standard of spelling is pronounced to be on the whole exemplary. (One may again stress the startling discovery that Virgil’s Aeneid was being read at Vindolanda, presumably as a school author for the son of Flavius Cerialis, TV II 118). Adams does note examples of Celtic loan-words and one 15 But their wish to take the recipient’s name as Coccelió instead of the ‘hypercorrect’ Coceiió-CocceIió with ‘I longa’- must be rejected. The l in the name of one ‘Coccelia’ Severa , the only parallel they can find for the nomen ‘Coccelius’, an inscription from Norba (Cáceres) in Spain, is a misprint for I longa, as noted in the edition they cite, J. Vives, ILER 4781. It is in fact another example of the hypercorrect double I in Cocceiius (and, incidentally, materterae on the Norba stone is not a second cognomen for Trebia Vegeta, the woman commemorated by Cocceiia Severa. It means ‘maternal aunt’). By mishap the correction of spicá to si itá was misreported in A. Birley 1997b: 275.

14 This is an improved version, owed to A.K. Bowman, of the names of the addressee and writer, and of one of the friends greeted - a woman, Verecunda (rather than a man called Verecundus).

927

Limes XVIII

indicating Celticized pronunciation, and the Octavius letter (TV II 343) is referred to repeatedly as being an exception to the general high standard. He discusses a range of linguistic items of interest for army life in the period just before the Wall was built, too numerous even to summarise here; but note the use of frater, collega and soror by correspondents of the same rank, cf. further below.16 I omit here further discussion of what the locative (or equivalent) place-names on the address-side of the letters refer to: place of writing or destination? But I note at least that while the editors of TV II are firmly for the latter, Michael Speidel nepos, editor of the Vindonissa stylus tablets, offers support for the place-name referring to the location of the writer, an Absender (Speidel 1996: 35ff.).

are letters to him, draft letters by him and accounts or other documents written under his command. The other three are much less well documented, but are also represented by a variety of tablets. Less certain are three men each registered by one tablet only. Flavianus (Period II), perhaps the same man as a correspondent of Flavius Cerialis, Hostilius Flavianus, and Veranius (Period III) were both evidently prefects at Vindolanda; for Vettius Severus, who received a letter at Vindolanda in Period II, even this is doubtful - not enough survives after his name on the back of the tablet to be sure whether he was a prefect. It should here be stressed that it is far from clear whether only one unit, hence only one commanding officer, was at Vindolanda at any given moment. A few remarks may be offered on the officers as a whole. The tablets register a string of Flavii (Cerialis, Conianus, Genialis, Gentilis, [?Ita]lieus, Proculus, Similis and Vindex) and a Flavianus, some, if not all, of this status. It is worth stressing how surprising it is to find equestrian officers with this name, indicating recent citizenship, early in the reign of Trajan. Very few parallels exist. The fact is, however, that the Flavii known from the Vindolanda tablets probably derived from the Rhineland, an area lacking stone honorific inscriptions or statue-bases that could have attested the existence of Trajanic officers with Flavian citizen status. Decisive proof is still lacking, but (Hostilius) Flavianus, and the Flavii, Cerialis, Genialis and Proculus (and also, whether or not they were officers, Conianus, Similis and Vindex), may be regarded as second-generation citizens, probably men whose fathers (perhaps a maternal grandfather in the case of Flavianus) were granted the franchise for loyalty during the Batavian revolt, and supplied the new commanders for the Batavians, drawn, as before, from the nobilissimi popularium. Perhaps Julius Civilis successfully negotiated the continuation of this vetus institutum and other privileges in his conloquium with Petillius Cerialis on a Nabaliae fluminis pons, of which the details are lost at the end of Tacitus’ Histories (V.26; cf. Strobel 1987: 284f.). This is implied, after all, by Tacitus’ statement in the Germania (29.1), manet honos et antiquae societatis insigne. Another element of the vetus institutum may be the anomalous titles of these officers. The commanders of the two Tungrian cohorts in Britain were still called prefects, not tribunes, even when the units were milliary; and the Ninth Batavians, perhaps already milliary when at Vindolanda, were certainly commanded there by prefects rather than tribunes.19 This makes it tempting to take literally the decurion Masclus’ form of address to Cerialis, regi suó (Inv. 1544, cf. above). Perhaps Cerialis really was, like the commanders of the pre-Flavian Batavian regiments, Julius Civilis and Claudius Paulus, regia stirpe (Tacitus Hist. IV.13.1). As contemporary parallels one might note that C. Julius Antiochus Epiphanes Philopappus

Three particular tablets, already mentioned in connection with the dating of Periods I and IV, are important in showing, respectively, a much reduced garrison (TV II 154: 296 men of coh. I Tungrorum present, 456 absent, of whom 337 were Coris, presumably Corbridge), and a mixed garrison (180, militibus legionaribus (sic) and 181, the equites Vardulli: both Vardullian cavalry and legionary soldiers were clearly there at the same time as the ‘main’ garrison, by then the Tungrians again). The renuntia (127152, 165?) of the Ninth Batavians seem to be a type of document not previously known,17 and the formulaic leave (commeatus) requests to Cerialis and other officers (166177) are also instructive. Could the regular mention of curatores as well as optiones in the former indicate that the curator was the cavalry equivalent of the optio, who was a deputy-centurion, ie. that the curator was a deputydecurion, in charge of equipment? There is reference to desertores in two tablets (226, 320; and cf. 345, expungas, perhaps referring removing the names of deserters from a list, as the editors note). A general feature should be emphasised: the accounts and some of the correspondence give the impression that the major preoccupation of the equestrian officers must have been supply. The tablets have revealed a good score of equestrian officers, almost all from the period c.85-120+, Periods IIV. Some are identifiable as commanders of a regiment in the Vindolanda garrison: Julius Verecundus, prefect of the First Cohort of Tungrians in Period I, Flavius Genialis, probably Period II, and Flavius Cerialis, Period III, both prefects of the Ninth Cohort of Batavians, and Priscinus, prefect of the Tungrians in Period IV.18 Cerialis is by far and away the most copiously attested: well over 100 tablets 16 One may note here that caballus, from the Celtic, discussed by Adams 1995: 124, as a possible vulgarism for horse, is used in a letter to Flavius Cerialis from a brother-officer (Inv. 1246) in a clearly non-pejorative context (ubi caballi belli sunt). Further, Inv. 1575, two draft letters by a man called Florus, are full of spelling and grammatical errors (cf. below) of a kind scarcely found in the published items. 17 Several more renuntia have been found: Inv. 1250 Can[didus optio]; 1311; 1319; 1373; 1418, the only complete example, cf. TV II p. 76; 1482A, the first to mention someone absent; 1552. Cf. also n. 2 above, on the unpublished Inv. 251. 18 Cf. further A.R. Birley 2001b for discussion with full references. TV III, when it eventually appears, will certainly correct details here.

19 See the interesting discussion by Strobel 1987: 287ff., noting, in particular, that there may be significance in the fact that in RIB 1981-2 prefects of coh. II Tungrorum milliaria equitata c.L are called praefectus Tungrorum.

928

A.R. Birley: The Roman army in the Vindolanda Tablets (cos. suff. 109), the grandson of the last king of Commagene, was called ‘king Philopappus’ by Plutarch in his Table Talk (Quaest. Conv. 1.10.1, 628AB; cf. PIR2 J 151), and another man who was an ex-ruler at this time was still called ‘king Alexander’ (ILS 8823).

prefect of an ala, hardly much later than c.97-100, if Genialis is correctly assigned to Period II, which is not quite certain - but he is surely nearly contemporary with Flavius Cerialis (conceivably they were at Vindolanda for a while together, but at some point Genialis wrote to Cerialis from elsewhere, TV II 256). Haterius’ prefecture could have been combined with his function as censitor of the Brittones Anavionenses, ie. the inhabitants of the Annan valley in south-west Scotland. The southern part of this area is easily visible from the line of Hadrian’s Wall a few miles north-west of Vindolanda. There seems to have been a long gap in Haterius’ career between his prefecture of cavalry, with post as censitor, and his first procuratorship, of Greater Armenia, which cannot have begun before 114. Such a gap had already been inferred by H.-G. Pflaum on a priori grounds: ‘en supposant que T. Haterius Nepos eût atteint 50 ans lors de sa nomination à la préfecture d’Égypte, il nous faut en conclure que, sous Trajan, en 114, il n’était, malgré ses 43 ans, que centenaire, ce qui permet de situer sous Domitien le début de son service militaire équestre. Mais il est impossible de préciser si la lenteur relative de son avancement initial doit être imputée à une carrière militaire prolongée ou à une interruption de sa carrière, survenue après la chute du dernier Flavien’ (CP no. 95; cf. PME H 1 + add.).

si me amas frater rogo mittas mihi plagas, ‘If you love me, brother, I ask that you send me some hunting-nets’. Thus begins the draft of a letter (TV II 233) by Flavius Cerialis to his friend Aelius Brocchus, a fellow-officer stationed at a fort not far away. A total of 20 tablets, mostly fragmentary, so far refer to Brocchus. Two are very warm letters from Brocchus’ wife, Claudia Severa, to Cerialis’ wife, Sulpicia Lepidina, the subjects being visits by one to the other. The expensa from Cerialis’ praetorium show Brocchus being a guest for dinner at Vindolanda on four separate occasions in the space of 8 months. Brocchus is only one of a series of correspondents of Cerialis who is called or who calls himself ‘brother’ or ‘colleague’; and their wives call one another ‘sister’. Persons of equal rank were not merely colleagues but ‘brothers’ or ‘sisters’. Brocchus was already known, as it turns out: as prefect of cavalry he dedicated an altar, appropriately to Diana, goddess of hunting, at Arrabona in Pannonia Superior. Here he calls himself by his full name, C. Aelius Brocchus. In the writing-tablets praenomina are virtually never used, except in one or two consular dates.20 A few of the other officers were also previously attested: M. Caecilius September, another friend and correspondent of Cerialis, had been prefect of a cohort in Syria some 15 years before he served in on the northern frontier of Britain. Newly identified is T. Haterius Nepos, later procurator and, after an astonishing series of promotions, prefect of Egypt under Hadrian, from 120-124. He corresponded c.100 with another Vindolanda prefect, Flavius Genialis, perhaps the immediate predecessor of Cerialis. Inv.1379 is a letter from Haterius to Genialis, labelled in the address praef coh, and called domine frater karissime. Haterius hopes that Genialis will indeed ‘come to Coria’ (no doubt Corbridge on Tyne, about 15 Roman miles east of Vindolanda, cf. TV II p. 96), ‘as you have decided’; and adds that he has written in the same terms to Proc̣[ulo], probably Flavius Proculus. An inscription, ILS 1338, from Fulginiae in Umbria, evidently his home, gives Haterius’ career as prae[f. coh]ortis, trib. milit[um], [p]raef. equit., censito[r] Brittonum Anavion[ens(ium)]; then proc. Aug. Armeniae mai[oris] (sc. between 114 and 117), followed by the procuratorships ludi magni, hereditatium et a censibus, the posts a libellis Au[g.], praef. vigilum and praef. Aegy[pti]. The inscription is acephalous, but there is no doubt about the identification, cf. PIR2 H 29. His career between c.114 and 120 was extraordinarily rapid: he probably combined several of the posts held at Rome.

The census taken by Haterius Nepos was surely not to establish what taxes the Anavionenses were to pay: this people will not have had a monetary economy. Theoretically, the intention could have been to establish what they could pay in kind, eg. in cattle. Far likelier, though, it was to discover how many of their men of military age could be conscripted into the Roman army and formed into numeri, to be sent overseas. There seems now to be agreement that the Brittones first epigraphically attested in Upper Germany in the 140s had been there from the beginning of the C2nd. Training of British conscripts is clearly the subject of the remarkable tablet (164, assigned to Period III, with a question-mark) mentioning ‘naked (= without armour?) Britons’ (nụḍị Brittones), very many of them cavalry’, who ‘do not use swords and do not stay mounted to throw javelins’ - with the now famous derogatory and previously unattested epithet Brittunculi. Anavion[enses] appear to be mentioned in an unpublished fragmentary account from the Period III ‘bonfire site’ at Vindolanda (Inv. 1475). It seems legitimate to infer that the Brittunculi were Brittones Anavionenses, and that some of them were put at Vindolanda in the first instance, to be trained by the Batavians (cf. A. Birley 2001c). The Batavians and Tungrians at Vindolanda, so the names of the soldiers, a good 200, suggest, were probably still largely drawn from their regiments’ original place of formation (cf. A. Birley 2001a); and they had a strong sense of common identity. This is nicely implied for the Batavians by the greeting in a letter from Period II to omnes cives et amicos -cives should surely be taken to mean ‘Batavian citizens’ (Inv. 1187). The fact that two or three tablets list men as Trever and Vaṇ[gio?] following their names may mean nothing more than that they were

Haterius was presumably based at Coria, probably as 20 eg. Inv. 1474A, line 20 (Sex. Attio Subur[ano]): Bowman & Thomas 1996: 312. One or two more may be found in the unpublished tablets. Otherwise, there is only 352, Marcus Cocceius Velox (with praenomen written out in full).

929

Limes XVIII

thus distinguished from Batavian homonyms. But it may also underline the almost homogeneous nature of the unit’s make-up. At least one officer of the Batavians regarded the natives with scorn, to judge from Brittunculi. As for the Tungrians, there is a striking echo of this attitude in the draft letter from Period IV, a complaint by a man who had been severely beaten by the centurions, treatment that, he says, should not have been meted out to a hominem trasmarinum et innocentem (TV II 344). By implication, while ‘a man from overseas’ was normally exempt from flogging, native Britons were not.

the writing tablets, inscriptions, brands and graffiti. (Bardon Mill): 18-72. Birley A.R. 1997a Hadrian. The restless emperor. (London). Birley A.R. 1997b Supplying the Batavians at Vindolanda. In W. Groenman-van Waateringe, B. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 273-280. Birley A.R. 1999 The Vindolanda writing-tablets. In P. Bidwell (ed.) Hadrian’s Wall 1989-1999 (Newcastle): 37-47. Birley A.R. 2001a The names of the Batavians and Tungrians in the Tabulae Vindolandenses. In T. Grünewald (ed.) Germania Inferior. (Berlin): 241-260. Birley A.R. 2001b A band of brothers. Equestrian officers in the Vindolanda Tablets. Electrum 5: 11-30. Birley A.R. 2001c The Anavionenses. In N.J. Higham (ed.) The Archaeology of the Roman empire: a tribute to the life and works of Barri Jones. British Archaeological Reports Int. Ser. 940. (Oxford): 15-24. Birley R. 1977 Vindolanda. A Roman frontier post on Hadrian’s Wall. (London). Birley R. 1993 Inscriptions, brands, graffiti. In VRR n.s. II: 73102. Birley R. 1994 Vindolanda Research Reports (VRR), n.s. I. The early wooden forts. The excavations of 1973-76 and 1985-89, with some additional details from the excavations of 1991-93. (Bardon Mill). Birley R. & Birley A. 1994 Four new writing-tablets from Vindolanda. Zeitschrift für Epigraphik und Papyrologie 100: 431-446. Bowman A.K. & Thomas J.D. cf. under TV I and II Bowman A.K. & Thomas J.D. 1996 New writing tablets from Vindolanda. Britannia XXVII: 299-328. Caruana I.D. 1989 Carlisle-Luguvalium. In C.M. Daniels (ed.) The Eleventh Pilgrimage of Hadrian’s Wall. (Newcastle): 24-31. Caruana I.D. 1992 Carlisle; excavation of a section of the annexe ditch of the first Flavian fort, 1990. Britannia XXIII: 45-109. Caruana I.D. forthc. The Roman forts at Carlisle: excavations at Annetwell Street, 1973-1984. Groves C. 1990 Tree-ring Analysis and Dating of Timbers from Annetwell Street, Carlisle, Cumbria, 1981-1984. (Ancient Monuments Laboratories Report 49/90, London). Speidel M.A. 1996 Die römischen Schreibtafeln von Vindonissa. (Brugg). Strobel K. 1987 Anmerkungen zur Geschichte der Bataverkohorten in der hohen Kaiserzeit. Zeitschrift für Päpyrologie und Epigraphik 70: 271-292. Tomlin R.S.O 1998 Roman manuscripts from Carlisle: the inkwritten tablets. Britannia XXIX: 31-84.

Further novelties may be in store from Vindolanda. For the time being it may be registered that the current excavations have discovered the bath-house, balneum, built in Period III. It was on the south side of the fort - and was twice the size of the later bath-house for the cohors quingenaria stationed in the stone fort. One may recall several of the published tablets: TV II 155, from 25 April in one of the years of Period III, lists 343 men employed f̣̣abṛ̣ịcis, including ̣ṣ[tr]ục̣tọṛẹṣ ad balṇeum xviii; [bal]neo may also, perhaps, be read in 322, likewise of Period III; a Period IV account, 181, includes the balniator Vitalis, listed as having paid his three denarii; and 184, a Period V account, mentions charges for several items used in the baths, such as cothurnum (a wooden bath-clog), sudaria (towels) and sebum (tallow). Bibliography21 CP

H.-G. Pflaum Les carrieres procuratoriennes équestres sous le Haut-Empire romain (Paris 1960-1961). ILS H. Dessau Inscriptiones Latinae Selectae. (Berlin 1892-1916). Prosopographia Imperii Romani, 2nd ed., by E. Groag, PIR2 A. Stein et al. (Berlin 1933ff.). PME H. Devijver Prosopographia Militarium Equestrium, 5 vols. (Leuven 1976-1993). TV I A.K. Bowman & J.D. Thomas 1983 Vindolanda: The Latin Writing-Tablets Britannia Monographs 4. (London). TV II A.K. Bowman & J.D. Thomas 1994 The Vindolanda Writing-Tablets (Tabulae Vindolandenses II). (London). VRR I, II Vindolanda Research Reports, cf. below under R. Birley 1993 & 1994 and A.R. Birley 1993. Adams J.N. 1995 The language of the Vindolanda writing tablets: an interim report. Journal of Roman Studies 85: 86-134. Birley A.R. 1993 A review of the tablets, by periods. In E. Birley et al. Vindolanda Research Reports (VRR), n.s. II. The early wooden forts. Reports on the auxiliaries,

21 Anyone able to consult all the present writer’s own contributions listed here (some not widely available) may recognise some recycling in this paper; but also some modifications of interpretation. For neither of which an apology seems needed. Further discussion (with some changed views and corrections of his own mistakes) will appear in A. Birley A band of brothers: Garrison life at Vindolanda (Stroud), delivered to the publisher in September 2001.

930

Food as substance and symbol in the Roman army: a case study from Vindolanda John Pearce Our understanding of the Roman military diet is drawn from textual and archaeological sources which are widely scattered over space and time. However increasing information now exists for exploring its variations according to region, period, unit or rank and for quantifying the relative contribution of different foods. The military diet is of interest not only to the study of the Roman army. It also impinges on wider issues, in particular the influence of the army’s provisioning requirements on local economies and on patterns of long distance exchange and the rôle of the army as an agent of cultural change, in this case of the dietary regimes of recruits and of the societies amongst which soldiers were stationed. References in the Vindolanda tablets, together with archaeological evidence from the same site, offer the opportunity of comparing consumption of food in one military community to the general picture. The presence or absence of particular foodstuffs in these documents has been exploited in connection to this debate, but little use has been made of the quantitative data which these texts supply. The relative frequencies with which individual foodstuffs are mentioned and in particular the quantities of foodstuffs recorded are therefore set out and discussed in this paper. Comparison of the relative importance of different commodities, when quantified, with the current understanding of the Roman military diet suggest significant discrepancies. In particular imported products considered central to the military diet, especially wine and olive oil, are seen to play a minor rôle. This has implications for understanding the general importance of these archaeologically highly visible commodities. Hypotheses to explain the similarities and differences between information from the tablet, the archaeological evidence and the general reconstruction of the military diet are explored. Information in the tablets on diet and dining customs can sometimes be related, for example, to different ranks. However whilst this is a possible source of bias, in general the data from the tablets do not relate to only one section of the military community.

consumption of food in a military setting, without some of the taphonomic and analytical obstacles encountered in the interpretation of archaeological deposits. They also offer the rare opportunity to make comparison with archaeological evidence from the same site. This paper builds on and re-assesses earlier discussions of the tablets as evidence for supply and diet (Birley 1997; Bowman 1994a: 68-71; Bowman & Thomas 1983; 1994; 1996). It is of course dangerous to extrapolate from a small corpus of documents, biased towards the praetorium and pertaining to a brief period in the history of one site on one frontier in its early stages of formation.

Introduction Since Davies’ (1971) fundamental study of the Roman military diet many new data, both textual and archaeological, have accumulated. However less attention has been paid to the framework within which these data should be interpreted and most of this has been directed at the implications of army requirements for local economies and long-distance exchange networks, in the terms of Garnsey’s (1999) discussion, food as ‘substance’. Food as ‘symbol’ however also served to mark status within Roman society and to distinguish Roman from nonRoman. In a military context differences of status and culture were potentially reproduced, but were cross-cut by the imperative to sustain cohesion and morale. ‘Foodways’, both differences in diet and modes of consumption, like other aspects of the daily life of the Roman army (eg. Goldsworthy & Haynes 1999; James 1998; Pollard 1997) will also repay attention as a medium through which relationships in military communities were reproduced and/or transformed (cf. Barrett 1989).

The paper begins with a review of approaches to the Roman military diet and an assessment of the Vindolanda tablets as evidence. The material for military diet in the documents is summarised, paying particular attention to quantitative evidence, and this information is evaluated from two perspectives. First, to what degree does the balance of difference elements compare with the general understanding of the military diet and second, what do their ‘foodways’ reveal about the cultural identity of these military consumers? The garrisons at Vindolanda comprised units of Batavians and Tungrians (E. Birley et al 1993). Roymans (1993; 1996) has identified a marked continuity of Batavian martial traditions into the Roman period, engendered through their service in the Roman army. This ‘trajectory’ distinguishes them from other northern Gallic tribes that fit a more general pattern of ‘civilian’ romanisation in Gaul. The opportunity is briefly taken here to exploit the evidence of the Vindolanda tablets as evidence for the ‘romanisation’ of units recruited from northern Gaul.

To treat the military diet as a monolithic entity masks potential variation. This paper suggests the possibilities for examining such variation, structured around Garnsey’s distinction between food as ‘substance’ and as ‘symbol’. It argues that to establish and interpret what was consumed in different areas and by different elements of the army illuminates both the army’s impact on the societies from which it was recruited and among which it was garrisoned, as well as our understanding of its operation as a community. The paper is not intended as a full-scale reevaluation of the military diet but rather aims to illustrate the potential for different approaches, through a case study from Vindolanda.

Roman military diet

Like other groups of military documents, the Vindolanda tablets offer a fine-grained view of the supply and

Davies’ study (1971) treated the military diet as a single phenomenon, combining archaeological evidence from the

931

Limes XVIII

north-west provinces with papyri and ostraka, mostly from Egypt. A much larger sample of archaeological data has subsequently been excavated from Roman military sites and developments in recovery strategies and analytical procedures have transformed the use to which faunal and botanical data can be put (eg. Dickson 1989; Huntley & Stallibrass 1995; King 1984; 1999a; 1999b). Pollen evidence also bears on the impact of the army’s food requirements (eg. Dumayne-Peaty 1998; Kreuz 1999). Additional collections of documents with relevant information are now published, from Carlisle (Tomlin 1998), Vindonissa (Speidel 1996), Bu Njem (Marichal 1992) and Mons Claudianus (Bingen 1992; 1997) as well as Vindolanda. Further evidence, from instrumentum domesticum is also available (eg. Tomlin 1992). Much of this new information informed Junkelmann’s (1997) recent survey of the military diet, but this gave more attention to amplifying Davies’ outline than to the framework within which the military diet should be studied.

(Breeze 1977; Casey et al 1993: 81, 243; Cool & Philo 1998: 365-66). Yet we need not always anticipate an absolute distinction between the different ranks. For example in the praetorium and the barracks families, slaves, guests and traders passed or lived. The presence of such individuals is well illustrated at Vindolanda (van Driel-Murray 1995). Certain occasions, for example the distribution of food to soldiers by their commanders after sacrifice (eg. Ammianus XXII.12.6) or celebratory feasting (eg. Josephus Bell. Jud. VII .16), may have suspended distinctions. Sharing the diet and dining style of the common soldier in camp is emblematic of the character of an emperor, either of virtue or severity. It is frequently a paired topos with suppression of luxuria among the soldiers (eg. Hadrian (SHA 10.2; Dio LXIX.9), Julian (Ammianus XXV.2.2; 4.17)). Auxiliary soldiers in the Rhineland and in Britain were sometimes represented on their tombstones as symposiasts. There is ‘real’ evidence too for admitting ordinary soldiers to the convivium, for example triclinia in barracks (Richmond 1962: 146) and ownership graffiti on metal and samian vessels (eg. RIB II 2415.63; II 2501-1-14).

There is increasing sensitivity to regional variation. The most thoroughgoing analysis has been King’s examination of the principal meat components of the diet but there has also been attention to other foodstuffs, especially cereals (eg. van der Veen 1992; Huntley & Stallibrass 1995). Inter-regional comparisons across several categories of data have also revealed diversity, for example between north and south Wales (Casey et al 1993; Davies 1997). Attention to such diversity will allow more precise modelling of the impact on local societies of supplying the army (eg. Kreuz 1999; Millett 1990: 56-57). Military requirements have been allowed a great deal of explanatory power for long distance trade routes, in particular the attraction northwards of Mediterranean products in bulk. The fortunes of these routes have been traced through archaeologically visible commodities, especially amphorae and other ceramics, but the significance of these, for example the scale of consumption that they represent and what proportion of diet they comprised, can be difficult to estimate.

The army in general was a privileged community not only perhaps in access to more adequate nutrition (Garnsey 1999: 126), but also perhaps in its access to the ‘products of empire’ otherwise only at the disposal of higher-status groups in the regions in which soldiers were stationed. Through their ‘foodways’ soldiers potentially signified their difference from local populations (eg. Carréras Monfort & Funari 1998: 67; King 1999b: 182-3). Soldiers joining the army, especially from areas where dietary or culinary habits had been little influenced by Roman patterns would potentially have experienced significant changes to styles of food preparation and consumption. They would have encountered not only individual new types of equipment, for example mortaria (eg. Baatz 1977; Haynes 1993), but also a repertoire of specialised vessels for food production and consumption, the availability of which the army consistently ensured (Greene 1979). The continuity of indigenous foodways in army units has also been suggested; Swan (1992) for example has identified close parallels between types of cooking pot from military sites in northern Britain and North Africa. Given the intensive recent study of the pre-Roman and early Roman archaeology of the areas of northern Gaul, there is significant potential for comparing evidence from Vindolanda with the areas from which the garrisons were recruited.

As for variation within the military diet, most attention has been paid hitherto to rank-related differences, most apparent in the architectural spaces in which those of different rank dined. The dining and reception rooms in praetoria have been argued to provide a setting for the construction of social relationships similar to the same rooms in Mediterranean townhouses (Förtsch 1995; Hodgson 1996; Marvell & Owen-John 1997). A small number of studies have identified differences in foodstuffs consumed by, and the dining paraphernalia available to, officers and men. For example the variety and choice of species in faunal and botanical deposits from officers’ accommodation on some British and Dutch sites has been related to the higher rank of their occupants (eg. Kuijper & Turner 1992; Marvell & Owen-John 1997; van Enckevort 1995: 54; Zienkiewicz 1993). Comparison of vessel assemblages has occasionally revealed preferential associations with officers’ accommodation, for example of samian, although so far on the basis of small samples

The tablets as evidence Table 1 summarises the phases identified by the excavations of the 1970s and 1980s (Birley et al 1993: 3). The tablets derive from the earliest, largely pre-Hadrianic wooden phases of the fort. The majority were excavated from levels belonging to Phases 2 and 3, when the site was occupied by the commanding officers’ residence, the praetorium. Contents of the tablets and palaeographic links 932

John Pearce: Food as substance and symbol in the Roman army: a case study from Vindolanda Period

Date (AD)

Context

1

85-c.92

fort ditch

2

92-97

praetorium - western range inc. yard, kitchen & storeroom

3

97-105

ditto (rebuilding)

4

105-c.120

southern end of barrack block, possibly centurion's/ optiones quarters

5

c.120-c.130

?fabrica Table 1: Early phases at Vindolanda

also support the connection of some documents to the praetorium (Bowman & Thomas 1994: 119). However there are also documents unambiguously related to supplies to other elements of the garrison, by the recipients (eg. TV II 180) or in the quantities recorded (TV II 343), even if the relationship to documentation produced in the tabularium is unclear (Birley 1997: 274-75; Bowman 1994b; Bowman & Thomas 1994: 20). The 9th cohort of Batavians is more frequently attested explicitly than the other Batavian and Tungrian garrison units. The correspondence of its prefect, Flavius Cerialis, very probably of Batavian origin (Bowman & Thomas 1994: 25; Birley 2001: 14-17), forms the largest group of letters and several accounts and lists can be connected to him. There are however too many documents without explicit connection to allow a systematic examination of differences in provisioning for different garrisons. Nor is the archaeological context a certain guide to the origin of the document. Deposition of the organic matrix within which the tablets were incorporated, for example during rubbish disposal or levelling, may have moved documents from the context in which they originated, as well as the obvious bonfire deposit (cf. Mons Claudianus; Bingen 1996). During excavation the interface between the layers connected to different periods was often difficult to identify (Birley 1994).

Provisioning of grain on a large scale from outside Vindolanda, presumably for the garrison as a whole, is reported. The tablets also reveal the military taking responsibility for producing some of its own food, for example keeping animals, hunting and possibly beermaking (Birley 1997), although food from such sources is not explicitly referred to as such in any tablet. Most common are small-scale distributions, typically procurement by and disbursement to individuals over several days or months. The amounts of food or other items recorded in some tablets suggest that the individuals to whom they are assigned are collecting them on behalf of a unit or group rather than themselves (eg. TV II 180, 183, 186, 207, 255). The rôle of officers, centurions, optiones and decurions, can be identified in this connection (eg. TV II 182, 183, 93.1495, 93.1544). Amounts issued on individual days, seemingly for consumption in the praetorium, must be for more than one individual (TV II.190; 93.1474). On the smallest scale are individual purchases by soldiers (TV II 181, 184) or small-scale transactions between them (TV II 310, 346) and between slaves (TV II 301), the receipt of gifts (TV II 299, 346) and the record of possibly a single meal (TV II 203). The archaeological context of many tablets suggests a bias to the praetorium, although this connection is rarely explicit (eg. TV II 190, 191).1 TV II 301 is addressed to the slave of the probable prefect Genialis and TV II 302 to the probable slave connected to Verecundus (identified as the prefect of a cohort in TV II 210, 211). A letter referring to food can be linked to the correspondence of Cerialis (TV II 233). 93.1474 clearly relates to Cerialis and his circle.

To what degree then are the forms of supply and consumption documented which we might anticipate in a military context? On current characterisations these are large-scale provision of supplies, from local or long distance sources, as levy, requisition or purchase, and supplies produced or procured by soldiers, placing the soldier:

Texts connected to lower-ranking officers include TV II 193, mentioning a loan to Felicio the centurion, although it is not clear how the food items listed relate to this loan, and TV II 182, although it is not clear for whose consumption the supplies are destined. TV II 299 records the receipt of oysters by a correspondent of the decurion Lucius, although the status of this correspondent is not given. Tablets explicitly recording consumption by individual soldiers are scarce (TV II 180, 184; 93.1544). The context is unclear from which some important documents for diet derive (eg. TV II 203, 204 or 208).

‘at the centre of a web which included his own unit, provincial officers, state officials, friends, veterans, local and distant provincials from both his own province and further afield, merchants, traders, farmers and manufacturers.’ (Breeze 1984: 282)

While the Vindolanda texts often do not explicitly identify sources of supply, nevertheless several strands of this ‘web’ are documented, as foodstuffs (and other commodities) are ordered, as they enter the fort, as they are disbursed to individuals and occasionally as they are consumed (Birley 1997).

1 TV II 191 records food used in p[raetorio], the digital image showing traces not inconsistent with r in line 11

933

Limes XVIII

State of preservation of course means that almost every tablet raises issues of interpretation, but other problems also obstruct use of the texts to reconstruct diet. Foodstuffs are mentioned in many different types of document, inventories, accounts and letters: one text details precise disbursements of grain, another laments the lack of beer. Information on price is rarely recorded with the volume or weight of commodities in the same document. Price information from elsewhere can only be suggestive in reconstructing quantities (eg. Duncan Jones 1982). The number of individuals for whom or number of days for which different amounts of food were intended is rarely known. Several important readings are not certain and food, especially cereal types, is no exception to the problem of matching term to commodity in the Vindolanda texts.2

references to food items, Table 3 those to quantified information on food, whether by price, weight or volume.3 Information from tablets from the 1990s excavations, published or alluded to in publication is listed in Table 4.4 Cereals The quantities of cereals, especially wheat, are greater than those of any other commodity. The 320 and a half modii of frumentum in (TV II 180), totalled from disbursement over several days, would have been sufficient for the daily calorific requirement of over 2000 soldiers (Bowman & Thomas 1994: 123). The individual amounts would have sufficed those to whom they were allotted for several weeks. On a larger scale are individual quantities of bracis (38 modii, possibly further lots of 53 and 63 modii: 91.1108) and 5000 modii of spicae, unthreshed grain (TV II 343).

Some texts refer to foodstuffs consumed elsewhere (eg. TV II 185, 299; 93.1474.55, 98). Some foodstuffs were required for particular occasions or purposes. The Feriale Duranum demonstrates the religious calendar’s requirement for animals for sacrifice (Junkelmann 1997: 156). From Vindolanda there are explicit references to food consumed at festivals (Bowman 1994a: 73; 93.1474,72). The combinations of items in an unpublished tablet (93.1350) may suggest a medicinal use (Bowman 1999). Some texts refer to foodstuffs for animals (eg. TV II 185).

The identification of bracis with emmer seems increasingly unlikely.5 Emmer is a minor component of contemporary botanical assemblages and bracis might be better identified with spelt, occasionally confused with emmer by classical authors (André 1981: 51). In plant assemblages from early Roman military sites, including northern posts (Dickson 1989; Huntley & Stallibrass 1995: 57, Table 5), spelt is predominant, as it is on non-military sites in north-east England (van der Veen 1992) and in Roman Britain in general (Jones 1991). Its predominance also characterises much of the northwest provinces (Kreuz 1999; Küster 1995). Whether derived locally or from further afield, spelt therefore seems a more probable identification for bracis.6

Other corpora suggest what may be absent from the Vindolanda tablets, for example regular supplies for units, as at Carlisle (Tab. Luguval. I; Tomlin 1998), Bu Njem (Marichal 1992: 103) and Pselkis (RMR 80-81) and for individual soldiers, again as at Pselkis (RMR 78). At Wadi Fâwâkhir and Mons Claudianus there is also much greater evidence of individual soldiers and others obtaining their supply through purchase or gift (Bingen 1992; 1997; Davies 1971: 134-5).

Siligo is conventionally bread wheat or soft wheat (triticum vulgare, triticum aestivum). Its supply is not unexpected as it is present, albeit generally in small quantities, on many Romano-British sites (Huntley & Stallibrass 1995: 57, Table 5; Jones 1991) and occasionally in larger volumes (van der Veen in Bidwell & Speak 1994: 250). The editors translate alica as ‘gruel’ (TV II 293). According to Pliny alica was made from zea (NH. XVIII.112), a regional name for emmer. In 93.1495 the term must refer to the cereal rather than its cooked/processed state. As oats are

Vindolanda will offer the opportunity to make detailed comparison of textual and archaeological evidence when faunal, botanical and ceramic reports are fully published (cf. van der Veen 1998). Significant inferences can be drawn from discrepancies between the texts and interim reports (van Driel-Murray 1993) on the archaeological evidence. In general this frontier lacks substantial published archaeological assemblages or works of synthesis on relevant categories of evidence, although there is much potential from current and future postexcavation work, in particular on plant and animal remains (Huntley & Stallibrass 1995).

3

An appendix to Table 3 shows the relationship between Roman measures and modern metric units. 4 Information on foodstuffs was gathered from tablets published in TV II, as well as publications of or allusions to the contents of tablets excavated in the 1990s (Birley & Birley 1994; Bowman & Thomas 1996; Birley 1997; Bowman 1999). The tablets published in the main corpora are referred to by their number in volume II (eg. TV II 190), the 1990s tablets by their inventory number (eg. 93.1474). A current programme (19992001) of digital imaging is producing improved readings of published tablets that are occasionally referred to in this paper. These are my own readings and await evaluation by Alan Bowman and David Thomas. The images will be available online from 2002. 5 The editors initially identified bracis as emmer (Bowman & Thomas 1983: 96; see also Birley 1997: 275) but more recently have preferred to follow André’s (1985: 37) judgement as a cereal of ‘genre inconnu’ (Bowman et al 1990: 49). 6 Both emmer and spelt were recovered in contemporary deposits at Vindolanda but there is no information on their relative frequency (Seaward 1993: 106-7)

Food at Vindolanda The evidence for foodstuffs from Vindolanda (in volumes I and II) is summarised on Tables 2 and 3: Table 2 lists 2

For the more detailed discussion of the identification of individual foodstuffs the reader is referred to the published commentaries on individual tablet, unless a particular argument is developed.

934

John Pearce: Food as substance and symbol in the Roman army: a case study from Vindolanda Table 2: Foodstuffs in the Vindolanda Tablets (Volume II)

Cereals / cereal products Term Amulum

Translation Meal (?)

References II. 204

Avena (?) Bracis

Oats / fodder Wheat - emmer / spelt? Wheat Gruel / wheat emmer ? Barley Bread Ears of grain Twisted loaf (?)

II. 185 II. 191

II. 343

II. 348

II. 160 (2) II. 193

II. 180 (2) II. 233

II. 185

II. 185 II. 180 (2) II. 343 (2) II. 180

II. 190 (7) II. 203

II. 213

Translation Pork fat Roe deer Venison Bacon Bacon-lard Pork cutlet (?) Oysters Eggs Ham Young pig Pork Chicken Suet / tallow Pig's trotters

References II. 182 II. 191 (2) II. 191 II. 182 (2) II. 182 II. 203 II. 299 II. 193 II. 191 II. 191 II. 186 (2) II. 302 II. 184 (4) II. 233

Translation Sour wine (Celtic) beer Wine lees Honeyed wine Wine (Massic?) wine

References II. 190 II. 186 (2) II. 185 (5) II. 302 II. 190 (4) II. 190

Frumentum (H)alica Hordeum Panis Spica Turta

Meat / fish Term Axungia Caprea Cervina Lardum Lardi perna Offella Ostria Ova Perna Porcellus (Caro) porcina Pullus Sebum Ungellae

II. 190 II. 439

II. 302

Wine / beer Term Acetum Cervesa Faex? Mulsum Vinum Vinum (Massicum?)

Pulses / vegetables / fruit Term Faba Fabae frensae Lens Malum Radix Oliva

Translation Beans Bruised beans Lentils Apples Radish / radix Britannica Olives

References II. 192 II. 302 II. 204 II. 302 II. 301 II. 302

935

II. 190 (4)

II. 203

II. 482

II. 191

Limes XVIII

Spices / herbs / condiments Term Alliatum Condimentum Ligusticum Mel Muria Oleum Piper Sal

Translation Garlic paste Spice Lovage Honey Fish sauce Oil Pepper Salt

References II. 208 II. 191 II. 204 II. 192 II. 190 II. 203 II. 184 II. 185

II. 193

II. 202

II. 302

II. 186

II. 191

References to food - possible readings Term Acetum Buturum Callum / alium Cervina Conditum Faba Frumentum (caro) hircina / porcina? Oliva Prunulum Perna Sebum

Translation Sour wine Butter Rind or crackling (pig) / garlic Venison Pickling liquor (?) Beans Wheat Goat meat / pork?

References II. 202 II. 204 II. 233

Olives Small plum? Ham Suet / tallow

II. 208 II. 189 II. 184 II. 319

936

II. 196 II. 208 II. 204 II. 160 II. 186

II. 375

John Pearce: Food as substance and symbol in the Roman army: a case study from Vindolanda Table 3: Quantitative information on foodstuffs (Volume II) Tablet II. 180 II. 182 II. 184 II. 185 II. 186

II. 190

II. 191 II. 192 II. 193 II. 202 II. 203 II. 299 II. 301 II. 302

II. 343

Commodity Frumentum / panis Lardum Lardi perna Axungia Piper Faex Hordeum Sal / avena Sal Cervesa Caro hircina? Caro porcina Hordeum Cervesa Vinum Acetum Muria Axungia Condimenta / caprea / sal / porcellus / perna / frumentum / cervina / conditum / bracis Fabae Mel Condimenta Halica Ovum Acetum? Muria Vinum Ostrea Radices Fabae frensae Pullus Malum Ovum Muria Olivae Spicae Bracis

Price

Number

x x x x x x x x x x

Volume x

Weight x x

x x x x x x x x x x

x (x) (x) x x x

x (x) X x x x X

x x X X X x

x x x x

Volumes and weights For discussion of the problems relating to the correspondence between modern weights and measures see introductions to the volumes of The Roman Inscriptions of Britain. Volume II. Instrumentum Domesticum 1 modius = 16 sextarii 1 metretes = 100 sextarii 1 modius = 8.754 litres 1 sextarius = 0.547 litres 1 metretes = 54.7 litres 1 Roman pound = c. 325.5-327.45grams

937

Limes XVIII

Table 4: Tablets from Vindolanda - 1990s excavations - references to food & food related artefacts Tablet Inv. 91.1022

Commodity bracis + other unknown

Inv. 91.1108

Bracis

Inv. 93.1350

various, including siligo, fabae, uvae, nuculae, ancusa, anesi, senapis, cuminum trullae hunting nets and snares pulli, pulli adempti, anseres, hordeum, cervesa / cervina? halica, siligo contrullium cervesa

Inv. 93.1398 Inv. 93.1462 Inv. 93.1474 Inv. 93.1495 Inv. 93.1503B Inv. 93.1544

Reference Birley & Birley 1994; Bowman & Thomas 1996 A. Birley 1997: 275-76; Birley & Birley 1994 A. Birley 1997: 278; Bowman forthc. Bowman & Thomas 1996.1 A. Birley 1997: 279 Bowman & Thomas 1996 A. Birley 1997: 275-76 A. Birley 1997 Bowman & Thomas 1996

and birds were documented in the faunal remains from Vindolanda, the most frequent being red deer (Seaward, in van Driel-Murray 1993: 109).

attested in the western sector of Hadrian’s Wall in the early Roman period and later in the eastern (Huntley & Stallibrass 1995: 58), it is as plausible that avena (TV II 185) refers to oats as to general fodder.

The consumption of cattle and sheep is not recorded in the tablets, although there are references to products derived from them and to their traction power, as well as to herdsmen (eg. TV II 180, 192, 343). Cows and sheep are however represented in the faunal remains assemblage, the former comprising the largest element (Seaward, in van Driel-Murray 1993: 109). The discrepancy is explored below.8

If fed to horses, the amounts of barley listed, in TV II 180, between 4 and 6+ modii, might have fed 12-18 horses for a day, following Tomlin’s (1998) analysis of barley consumption in Tab. Luguval. I. The barley may also have been destined for human consumption as barley cakes or puls or for brewing. Sewage from Bearsden and other military sites demonstrates some human consumption of barley (Dickson 1989; Huntley & Stallibrass 1995: 52) whether as emergency or punishment rations, as the literary evidence suggests, it is not possible to say.

Wine and beer A range of wine types occurs, acetum, mulsum and possibly faex and vinum massicum. The possible reference to Massic wine is made plausible by epigraphic attestation of ‘vintage’ wines on military sites (Davies 1971: 131; Hassall & Tomlin 1994: n.92). The quantities range from a few sesterces worth of faex (at Pompeian prices enough for only one or two sextarii: Duncan-Jones 1982: 46), to between one and two modii in TV II 190).9 The average capacity of many contemporary amphorae is larger, often three modii or more.10 TV II 190 may therefore record the re-distribution of wine within the praetorium, perhaps in small amounts decanted from amphorae. At the rate of consumption recorded in two C6th AD papyri (P.Oxy. xvi. 1920/2046) (two sextarii per soldier per day), two modii would have supplied the daily wine requirements of 16 men.

Meat Pig, chicken, goose and deer are the only animals certainly recorded in the tablets as being consumed at Vindolanda. Of these pig, usually in preserved form, is predominant. The largest quantities are mentioned in TV II 182. Only in one instance is the price and the volume recorded for the same disbursement (eight denarii two asses for 45lbs of bacon (lardum) and 15.5lbs of bacon-lard (lardi perna)), but the cost of other consignments suggests that they were of a similar scale. At the meat consumption rates documented in P.Oxy. xvi.1920/2046 (Jones 1964: 629) of 2lbs of meat per day, such amounts would have lasted an individual at a month.7 93.1474 raises the profile of chicken (and probably chicks), recording their regular consumption by Cerialis, his household and guests. According to Junkelmann (1997: 152) eggs were not part of a regular military diet but the price recorded in TV II 193 was not prohibitive in comparison to that for other commodities.

The tablets make frequent reference to consumption of beer (and brewers; TV II 182; 93.1474) and preserve information on quantities. TV II 186 twice records the supply of a metretes, TV II 190 two and three modii and three plus modii. At 8 asses the metretes is very cheap

References to caprea and to cervina may record the consumption of both red and roe deer (Bowman & Thomas 1983: 95). Other tablets refer to the hunting of wild animals and birds (TV II 233B; 93.1462) and wild animals

8

Contra Junkelmann 1997: 87, neither TV II 190 nor 191 records the consumption of lamb and beef. Pace Seaward 1993: 112, TV II 191 does not demonstrate that deer meat was issued in everyday rations. 9 Johnson’s 1983: 196 figure, repeated by Junkelmann 1997: 179, of the issue to the troops of 73 modii of wine on 22 June of an unknown year is not supported by the text cited (TV II 190). 10 eg. Dressel 1, 24 litres, Dressel 2-4, 26-34 litres, Gauloise amphorae 26-37 litres: figures from Peacock & Williams 1986: 52.

7

Jones 1964: 629 describes these allocations as ‘positively gargantuan’ but they were perhaps intended to supply dependants as well.

938

John Pearce: Food as substance and symbol in the Roman army: a case study from Vindolanda (Birley 1997: 279); this would have bought only one or two sextarii, of the lowest quality of wine at Pompeian prices (see above). The potential contribution of beer to daily calorific intake should not be underestimated. Rates of consumption from various late mediaeval and early modern contexts offer comparative figures, although difficult to relate to one another or to Vindolanda as in no case do we know the strength of the beer (Clark 1983). At the upper end of attested consumption in C16th England, for example soldiers in the English garrison at Boulogne were issued four and a half gallons of beer (c.20.5 litres) each per week (Clark 1983: 154-5). English navvies building the Paris-Rouen railway consumed five quarts of ale (c.5.7 litres) per day (Coleman 1968 in Garnsey 1999: 44). At the latter rate the daily amounts recorded in TV II 190 would have provisioned 2-4 soldiers, at the rate consumed by the Boulogne garrison 6-9 soldiers.

botanical assemblages (Küster 1995: 12-13). Lead tags provide further evidence for the supply of pepper to the north-west provinces (Schwinden 1985). Coriander and poppy were recovered in the botanical assemblage at Vindolanda. Coriander may be attested in a graffito on a jar at Vindolanda (E. Birley et al 1993: 97). Honey was recorded in an unknown quantity of modii (TV II 192). The lini mellari recorded on an unpublished tablet (93.1350) may have served to sieve honey or were perhaps bandages soaked in honey. The combination of herbs in this tablet may also imply a medicinal use (Bowman 1999). Salt is recorded in large quantities in two tablets (TV II 185, 186). At the rate of 12 plus asses for 85lbs plus of salt (TV II 186), the denarius worth of salt and avena in TV II 185 may also represent a substantial amount, although the proportions of each are unknown. A trade in salt in Iron age and Roman north-east England, possibly based in the Tees estuary may have supplied Vindolanda (Willis pers. comm.), although it may have come from further afield (Anderson 1992: 63).

Pulses, vegetables and fruit If the quantity of beans in TV II 192.3, is correctly read as 55 modii, this represents the second largest of any commodity after cereals. The other recorded quantity is a more modest two modii (TV II 302). The other fruits and vegetables complement the species in the botanical assemblage from Vindolanda (beans, walnuts, hazelnuts and figs; Seaward, in van-Driel Murray 1993: 105-7).

Summary The fundamental rôle of grain in the military diet is well documented, cereals being by far the most significant commodity in terms of volume. Whether the contribution of cereals to diet is as high as Forbes & Foxhall’s (1983) estimate, that cereals account for perhaps three-quarters of dietary needs in the Mediterranean in antiquity, is impossible to assess. Evidence from Vindolanda supports too the importance of pulses in the military diet, suggested by literary evidence and some botanical assemblages (Davies 1971; Junkelmann 1997: 137).

The modius of olives referred to (TV II 302) is a little less than the capacity of contemporary amphorae; for example that of the London 555 was 13 litres (Sealey & Tyers 1989). So far only one reference to olive oil is known. Contra the editors, a digital image of TV II 203 seems to show clearly that olei is not followed by any surviving indication of quantity. The reference occurs in the context of small amounts of food, a (single?) pork cutlet and a sextarius of wine.

The significant contribution of pig meat, especially in cured form, was expected (Garnsey’s 1999: 17 stress on the consumption of fresh pork by the army in Britain seems too emphatic). The high profile of chicken is more surprising. Chicken is only a minor component of faunal assemblages in Roman Britain, the proportion from military sites being second highest after towns (Maltby 1997). The consumption of wild foods seems more significant than faunal evidence would suggest, the proportion on military sites in Roman Britain being generally very low, especially when compared to southern Germany (King 1999a: 146).

Spices, herbs, condiments and others Only one type of fish sauce, muria, possibly of relatively low value, is recorded three times; other types are recorded from amphorae on northern frontier sites, including Vindolanda (eg. Wright & Hassall 1974: 467, no. 44; Curtis 1991). The volumes recorded, one and a half and 8 sextarii (TV II 190, 302) are small in comparison to the capacity of contemporary containers: the two sizes of the Dressel 7-11 ‘salazon’ amphora for example held 14-18 litres and 27-33 litres. Condimenta, fresh or dried herbs and spices, are not necessarily of exotic origin. The price recorded (TV II 193) is relatively low: a half-sextarius cost at most twice as much as the same amount of alica (see above). Aliatum may be a garlic paste, conditum is probably a spiced liquor used in preserves.

The discrepancy is striking between the documentary and faunal evidence from both Vindolanda and other military sites in Roman Britain. Cow accounts for the largest percentages of animal bone assemblages on all military sites, followed by sheep on auxiliary sites (King 1984; 1999a; 1999b). There is also no reference to cheese, expected to be a substantial component of the military diet (Junkelmann 1997: 152-53) and archaeologically visible in ‘cheese-press’ dishes (eg. Dannell & Wild 1987: 69).

The references to cumin and pepper are significant given their archaeological rarity, although cumin is sometimes confused with caraway (Drexhage 1993; Küster 1995: 13). Cumin is attested only once archaeologically in northern Europe and peppercorns in only a small number of 939

Limes XVIII

The documented volumes of wine and beer supply are similar. Neither is likely to have sufficed for a few men for more than a few days. However the presence of cervesarii and the ‘soldier’s complaint’ (93.1544) support the argument that beer was a staple of the soldiers (Bowman 1994a: 76; Dannell & Wild 1987: 70; van der Werff et al. 1997). In literary sources the preference for beer is a topos, illustrating the ‘otherness’ of the barbarian palate (Binsfeld 1972, Ruprechtsberger 1992; RE III. 457-65). Evidence of actual large-scale local production has hitherto only been available from Ptolemaic and Roman Egypt (Drexhage 1997), although suggested also by Diocletian’s Price Edict. In comparison to winemaking, brewing is also relatively invisible archaeologically.11 With the possible reference to a brewer (*bracearius, or possibly a dealer in bracis, see also now the Vindolanda Tablet 92.1110) on a stylus tablet address from Carlisle (Hassall & Tomlin 1994: 496-498, no. 32), the references to brewers are a significant addition to the cervesarii documented in the north-west provinces (Frézouls 1991). Their presence makes local production likely, although some long-distance trade in beer has been proposed (van der Werff et al 1997).

some degree compensate for archaeological biases in preservation, making certain products more visible. For example cured meat products off the bone, bacon and ham, are under-represented in archaeological contexts. Taphonomic and sampling factors also render chicken less visible in faunal assemblages. The documentation is also unlikely to be completely representative of that produced in the fort. Earlier discussion showed that there is an emphasis on the praetorium and on relatively small-scale redistribution of commodities within the fort, an impression strengthened by further analysis of the volumes of individual commodities. Nevertheless most circuits of exchange are represented in the tablets, albeit unevenly. Carréras Monfort and Funari (1998: 50) argue that the absence of references to olive oil relates to the tablets only referring to local purchases, but regional as well as local movements of supplies are documented and other imported commodities are recorded as they move within the fort. The garrisons of forts on the northern frontier were clearly consuming imported wine and especially olive oil, as amphorae assemblages, including those from Vindolanda clearly demonstrate, even if the peak phase of oil amphora representation is not reached until the Antonine period. Examination of the tablets however questions the degree to which either commodity, especially olive oil, formed a staple of military diet in these phases at Vindolanda, especially for ordinary soldiers. Their high archaeological visibility may have exaggerated the importance of their consumption. This evidence also challenges the validity of generalisation from isolated pieces of literary evidence for consumption, especially from ‘oil-rich’ areas, to general supply patterns to the army (eg. Remesal Rodríguez 1997: 49-50). Alternative products may have fulfilled the rôle of oil, for example pig fat, butter and sebum, possibly tallow, supplied in large quantities (TV II 184).

The amounts of wine recorded seem likely to represent decanting from the amphorae or barrels in which the wine arrived. The overall amount of wine recorded over 4-5 days in TV II.190, related to the praetorium supply, represents the equivalent of one to two larger wine amphorae. Over a year this consumption rate would generate 70-140 amphorae, over 50 years 3500-7000. Such figures provide some context for even substantial numbers of wine amphorae in archaeological assemblages. Most surprising is the limited evidence for olive oil. Oil amphorae, in particular the Dressel 20, dominate amphora assemblages from Britain’s northern frontier (eg. Carréras Monfort & Funari 1998). The archaeological representation, as well as papyrological evidence, have led some commentators to argue that oil was a staple of the military diet (eg. Carréras Monfort & Funari 1998; Garnsey 1999: 126).

The recovery of documentation mostly relating to the supply of the praetorium might explain the overrepresentation of some foods. The meat preferences, for pig, chicken and wild foods, are consistent with some evidence from archaeological deposits related to officers’ accommodation, for example from a house in the scamnum tribunorum at Caerleon (Zienkiewicz 1993). We can therefore offer the hypothesis that beef and mutton will be more strongly represented in rubbish deposits produced by ordinary soldiers.

The fruits, vegetables, and condiments, all in relatively small quantities, are consistent with the range attested in botanical deposits from military sites in Germany and Britain (Huntley & Stallibrass 1995; Junkelmann 1997: 139-146; Kreuz 1999; Küster 1995: 18-20). The evidence for cumin and pepper is a more significant addition to the corpus of spices from Roman military sites.

Consumption The most significant discrepancies therefore between the ‘standard’ military diet and that reconstructed from the Vindolanda tablets, a difference also between the tablets and the archaeological evidence from Vindolanda and other auxiliary forts, lie in preferences for meat and the evidence for imported commodities. Preservation biases in archaeological evidence may bear on this. The tablets to

The bias of documentation to the praetorium frustrates extended comparison of the relationship of rank to diet. We can however note exceptions to some generalisations that have been proposed. The differentiation of higher and lower ranks by consumption of wine and beer (eg. van der Werff et al 1997) is not corroborated by the Vindolanda texts. TV II 190 for example shows the regular provision of beer in the praetorium, documented in the same account as wine. According to TV II 184, two denarii worth of pepper were purchased by or owing to Tagarminis; his

11 Ruprechtsberger 1992 discusses beer production in general while Becker & Tegtmeier 1999 and Rieckhoff 1992 identify possible breweries.

940

John Pearce: Food as substance and symbol in the Roman army: a case study from Vindolanda designation as belonging to the century of Ucen(i)us implies him not to have been of high rank. Given the expense of pepper (Schwinden 1985), this is a significant demonstration of the access of lower ranking soldiers to luxuries, further evidence that a ‘convivial culture’ was not merely an afterlife aspiration. This evidence also suggests caution in interpreting the mention of exotic items as evidence for a document’s connection to the praetorium. Higher ranks may have had more frequent but not exclusive access to luxury products, although the evidence of the tablets is not sufficient to confirm this.

both the Tungrian and Batavian areas, as across much of Gaul and Germany (Lauwerier 1983; Lepetz 1993). A local association with sacrifice or feasting connected to ritual may have lent connotations of high status that influenced preferential consumption of these meats in a high-ranking context in the Vindolanda garrison. Other preferences revealed by the tablets can also be read from a local perspective, perhaps relating to the cultural origin of the units, for example in beer drinking in the praetorium and the possible ‘Batavian mos’ (TV II 208), although the latter reference is frustratingly fragmentary.

The insight into praetorium ‘foodways’ is of interest for the culinary ‘romanisation’ of the commanders of auxiliary units and their milieu. Given that the majority of explicit attestations relate to the Batavi, special attention deserves to be paid to their particular form of ‘romanisation’. While their diet was based on common elements in the military diet, grain, bacon, and beans, the inhabitants of the praetorium drew on a wider repertoire of foods. The diversity of wine types suggests some familiarity with wine hierarchies recorded by Pliny (Bowman 1994a: 70). Fish sauce was regularly consumed, although muria is probably the cheapest of the various forms. Herbs and spices from a wide area were consumed, pepper and cumin being the commonest spices in the recipes of ‘Apicius’, as well as garum being among the most common seasonings, required for more than two thirds of these (Curtis 1991; Solomon 1995: 116). Olive oil, also essential for an ‘Apician’ cuisine, was sometimes available, although perhaps not always even in the praetorium. Combinations of items in the same tablet, for example on the left margin of TV II 302 (items sought by the slave of Verecundus) suggest more than the intrusion of occasional exotic items in an unchanged diet. This list could almost be for a proverbial Roman meal, ab ova ad mala. Hunted meat may have alleviated food shortage, but the proportion of wild animals in the faunal remains assemblage is low. The consumption of wild animals was perhaps also connected to elite distinction, through the connotations of virtus attached to hunting (Lane Fox 1996).

The tablets also provide evidence for the broader context of dining, both eating and drinking vessels (TV II 194, 93.139, possibly 93.1503B) and ‘dinner dress’ (TV II 196). Content, archaeological context and palaeographical links suggest that TV II 194 refers to sets of dining equipment, either in ceramic or metal, used in the praetorium. The editors comment that this set of vessels recalls a papyrus inventory of silver dining vessels (Oliver & Shelton 1979). It also recalls the combinations of samian vessels found in early Roman graves in Gaul and Germany, which have been independently linked to the same papyrus (MartinKilcher 1976). The social network of the prefects of Vindolanda extended not only to their peers, some of whom occur in other tablets, but also to the provincial governor as well as to those who sought their patronage and recommendation (Bowman 1994a: 54-7; 74-5). There are therefore glimpses of a convivium culture (cf. d’Arms 1990), linking friends, peers and protégés as well as superiors and subordinates, even if ‘Whether all this betokens the practice of haute cuisine in the praetorium is another matter.’ (Bowman 1994a: 70). 93.1474 demonstrates the variety of guests entertained at Vindolanda, including Cerialis’ peers, the governor and possibly a legionary legate. The fort commanders can be seen as familiar with a range of ‘Roman’ behaviours, not only in the consumption of food. In appearance and hygiene routines (eg. TV II 197), and in the reflections of a literate and literary culture (Bowman 1994b), the tablets also demonstrate familiarity with regimes of Roman taste, the internalisation of which by Gallic elites has been explored by Woolf (1998). The Vindolanda tablets suggest therefore that emphasis only on the continuity of local cultural trajectories does not do justice to the response to Roman power among the Batavi.

In the Apician corpus pig and chicken are the commonest species eaten and young pig and chicks are delicacies within Roman culinary tradition (Lauwerier 1983). Pig has been demonstrated by King to have been preferred in a military milieu, but these meats also form a larger than average component of faunal assemblages from the higher end of the settlement hierarchy, both towns and villas, in Britain (King 1984: 189-90; 1999b: 179-80, Maltby 1997) and northern Gaul (Lauwerier 1988: 126-8; Lepetz 1996: 116-9, 126-7). The tablet evidence does not conform with the preferences from the lower Rhine area in the preRoman Iron age and in the early Roman period, where cattle account for between 50 and 90% of faunal assemblages from the majority of sites (Roymans 1996: 51, 80).

Conclusion In terms of the ‘substance’ of diet, the fundamental rôle of grain, with important contributions from pulses, bacon and salt, plus small quantities of a wider range of foodstuffs and seasonings, are consistent with current expectations of the military diet. Significant differences from expectations have however been established in the relative consumption of different types of meat, especially the preference for chicken and pig, and in the contribution of Mediterranean commodities. The differences in part reflect preservation biases and the derivation of many documents from the

Pig and chicken are however the most common species in funerary deposits in the pre-Roman and Roman period in 941

Limes XVIII

praetorium. However the lack of evidence for bulk supply of wine and especially olive oil challenges the degree to which these were staples of the military diet. It demands further attention to the impact of the provision of grain, meat and pulses to the army, on regional and more distant economies.

André J. 1985 Les noms de plantes dans la Rome antique. (Paris). D’Arms J. 1990 The Roman convivium and the idea of equality. In O. Murray (ed.) Sympotica: a symposium on the symposion (Oxford): 308-320. Baatz D. 1977 Reibschale und Romanisierung. Rei Cretariae Romanae Fautorum Acta 17/18: 147-158. Barrett J.C. 1989 Food, gender and metal: questions of social reproduction. In M.L.S. Sørensen & R. Thomas (edd.) The Bronze age-Iron age transition in Europe. British Archaeological Reports Int. Ser. 483. (Oxford): 304321. Becker W.D. & Tegtmeier U. 1999 Römisches Bier in Xanten. Archäologie im Rheinland 1998: 85-87. Bidwell P. & Speak S. 1994 Excavations at South Shields Roman fort. (Newcastle-upon-Tyne). Bingen J. 1992 Mons Claudianus Ostraca graeca et latina I. (Cairo). Bingen J. 1996 Dumping and the ostraca at Mons Claudianus. In D. Bailey (ed.) Archaeological research in Roman Egypt. Journal of Roman Archaeology Supplementary Series 19: 29-38 Bingen J. 1997 Mons Claudianus Ostraca graeca et latina II. (Cairo). Binsfeld W. 1972 Eine Bierverlegerin aus Trier. Zu CIL XIII 450. Germania 50: 256-259. Birley A.R. 1995 Vindolanda: Das Alltagsleben in einer römischen Grenzbefestigung in Britannien zu Beginn des 2. Jahrhunderts n. Chr.. In W.G. Busse (ed.) Burg und Schloß als Lebensorte in Mittelalter und Renaissance. (Droste Verlag): 9-18. Birley A.R. 1997 Supplying the Batavians at Vindolanda. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995: proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 273-280 Birley A.R. 2001 A band of brothers; equestrian officers in the Vindolanda tablets. Electrum 5: 9-30. Birley A.R. & Birley R.E. 1994 Four new writing-tablets from Vindolanda. Zeitschrift der Papyrologie und Epigraphik 100: 431-446. Birley R. 1994 The early wooden forts. Introduction and analysis of the structures. (Vindolanda Research Reports 1: Bardon Mill). Birley E., Birley R. & Birley A. 1993 The early wooden forts: Reports on the auxiliaries, the writing tablets, inscriptions, brands and graffiti. (Vindolanda Research Reports 3: Bardon Mill). Bowman A.K. 1994a Life and letters on the northern frontier. (London). Bowman A. 1994b The Roman imperial army: letters and literacy on Rome’s northern frontier. In A. Bowman & G. Woolf (edd.) Literacy and power in the ancient world. (Cambridge): 109-125. Bowman A.K. 1999 The Vindolanda writing-tablets. In Atti XI Congresso Internazionale di Epigrafia Greca e Latina (Roma 18-24 Septembre 1997). ( Rome): 545-551. Bowman A.K. & Thomas J.D. 1983 Vindolanda. The Latin writing tablets. Britannia Monographs 4. (London). Bowman A.K. & Thomas J.D. 1994 The Vindolanda writing tablets (Tabulae Vindolandenses II). (London). Bowman A.K. & Thomas J.D. 1996 New writing tablets from Vindolanda. Britannia XXVII: 299-328. Bowman A.K., Thomas J.D. & Adams J.N. 1990 Two letters from Vindolanda. Britannia XXI: 33-52. Breeze D. 1977 The fort at Bearsden and the supply of pottery

As for food as ‘symbol’, association of most documents with the praetorium impeded a systematic study of rankrelated ‘foodways’, although the tablets occasionally demonstrated departures from expectations, for example the consumption of beer in the praetorium and the purchase by an ordinary soldier of pepper. The patterns of consumption and the ‘convivium culture’ identified in the Vindolanda praetorium shed a different light on the mode of ‘Romanisation’ for the Batavian elites than the continuity and amplification of indigenous traditions envisaged by Roymans. Rather the range of foods and the manner of their consumption are analogous to other high status contexts in the north-west provinces, although possible resonance with traditions from the areas from which units were drawn has also been noted. Our understanding of diet and dining even at Vindolanda may be changed by new discoveries (although not so far by the tablets excavated in the 1990s). The problems with the particular data on which this paper has focussed must not be forgotten. Issues concerning individual readings have usually been set aside in favour of examining the bigger picture, but this does not make these difficulties any less acute. However this discussion has, at the very least, provided a frame of reference to assess the significance of new documents and has proposed hypotheses to be tested against new documentary and archaeological evidence. It has aimed to emphasise, from a small-scale case study, the potential variation in military ‘foodways’ and the possible approaches to interpret such diversity. Acknowledgements I would like to thank the British Academy and the General Board of the Faculty of Literae Humaniores, Oxford University, for grants that allowed me to attend the conference. Alan Bowman, Tony Birley and Graham Oliver read earlier drafts, and I am grateful to them and to the participants in the Roman military documents session in Amman for comments. None of them should be incriminated in the shortcomings of the paper, for which I am alone responsible. My thanks are due also to Tony Birley and Alan Bowman for making forthcoming papers available prior to publication. Bibliography Anderson

J.D. 1992 Roman military supply in north-east England: an analysis of an an alternative to the Piercebridge formula. British Archaeological Reports Brit. Ser. 224: (Oxford). André J. 1981 L’alimentation et la cuisine à Rome (2nd ed.). (Paris).

942

John Pearce: Food as substance and symbol in the Roman army: a case study from Vindolanda to the Roman army. In J. Dore & K. Greene (edd.) Roman pottery studies in Britain and beyond. British Archaeological Reports Supplem. Ser. 30. (Oxford): 133-146. Breeze D.J. 1984 Demand and supply on the northern frontier. In R. Miket & C. Burgess (edd.) Between and beyond the walls: essays on the prehistory and history of North Britain in honour of George Jobey. (Edinburgh): 264286 Carréras Monfort C. & Funari P.P.A. 1998 Britannia y el Mediterráneo: estudios sobre el abastecimiento de aceite bético y africano en Britannia. (Barcelona). Casey P.J., Davies J.L. & Evans J. 1993 Excavations at Segontium (Caernarfon) Roman fort, 1975-79. Council for British Archaeology Research Reports 90. (York). Clark P. 1983 The English alehouse: a social history 12001830. (London). Coleman T. 1968 The railway navvies: a history of the men who made railways. (Harmondsworth). Cool H. & Philo C. 1998 Roman Castleford: excavations 1974-85. Vol. 1. the small finds. (West Yorkshire Archaeology Service, Wakefield). Curtis R.I. 1991 Garum and salsamenta, production and commerce in materia medica. (Leiden). Dannell G.B. & Wild J.-P. 1987 Longthorpe II. The military works depot. Britannia Monographs 8. (London). Davies J.L. 1997 Native producers and Roman consumers: the mechanisms of military supply in Wales from Claudius to Theodosius. In W. Groenman-van Waateringe, B.L. van Beek, W.J.H. Willems & S.L. Wynia (edd.) Roman Frontier Studies 1995: proceedings of the XVIth International Congress of Roman Frontier Studies. (Oxford): 267-272 Davies R.W. 1971 The Roman military diet. Britannia II: 122142. Dickson C. 1989 The Roman diet in Britain and Germany. Archäobotanik Dissertationes Botanica 133: 135-154. Drexhage H.-J. 1993 Kuminon und sein Vertrieb im griechischen-römischen Ägypten. Münstersche Beiträge zur antiken Handelsgeschichte 12.2: 27-32 Drexhage H.-J. 1997 Bierproduzenten und Bierhändler in der papyrologischen Überlieferung. Münstersche Beiträge zur antiken Handelsgeschichte 16.2: 32-39. van Driel-Murray C. 1993 The early wooden forts: preliminary reports on the leather, textiles, environmental evidence and dendrochronology. Vindolanda Research Reports III. (Hexham) van Driel-Murray C. 1995 Gender in question. In P. Rush (ed.) Theoretical Roman Archaeology: second conference proceedings. (Aldershot): 3-21 Dumayne-Peaty L. 1998 Forest clearance in northern Britain during Romano-British times: re-addressing the palynological evidence. Britannia XXIX: 315-322. Duncan Jones R. 1982 The economy of the Roman empire: quantitative studies. (2nd ed.: Cambridge). van Enckevort H. 1995 Das Lager auf dem Kops Plateau. In J.S. Kühlborn (ed.) Germaniam pacavi - Germanien habe ich befriedet. Archäologische Stätten augusteischer Okkupation. (Münster): 42-58. Fink R.O. 1971 Roman military records on papyri. (Cleveland). Forbes H.A. L. & Foxhall L. 1983 Sitometreia: the role of grain as a staple food in classical antiquity. Chiron 12: 41-90. Förtsch R. 1995 Villa und Praetorium - zur Luxusarchitektur in frühkaiserzeitlicher Legionsläger. Kölner Jahrbuch 28: 617-634.

Frézouls E. 1991 Les noms de métier dans lépigraphie de la Gaule et de la Germanie romaines. Ktema 116: 33-72. Funari P.P.A. 1996 Dressel 20 Inscriptions from Britain and the consumption of Spanish olive oil. British Archaeological Reports Brit. Ser. 250. (Oxford). Garnsey P. 1999 Food and society in classical antiquity. (Cambridge). Goldsworthy A. & Haynes I. (edd.) 1999 The Roman army as a community. Journal of Roman Archaeology Supplementary Series 34. Greene K. 1979 Invasion and response: pottery and the Roman army. In B.C. Burnham & H.B. Johnson (edd.) Invasion and response. British Archaeological Reports Brit. Ser. 73. (Oxford): 99-108 Hassall M.W.C. & Tomlin R.S.O. 1994 Roman Britain in 1993: II. Inscriptions. Britannia XXV: 293-314. Haynes I. 1993 The romanization of the alae and cohortes of the Roman imperial army from Augustus to Septimius Severus. Unpublished D.Phil. thesis, Oxford University Hodgson N. 1996 A late Roman courtyard house at South Shields and its parallels. In P. Johnson & I. Haynes (edd.) Architecture in Roman Britain. Council for British Archaeology Research Report 94. (York): 135151. Huntley J.P. & Stallibrass S. 1995 Plant and vertebrate remains from archaeological sites in northern England: data reviews and future directions. Research reports of the Architectural & Archaeological Society of Durham & Northumberland 4. James S. 1998 The community of the soldiers: a major identity and centre of power in the Roman empire. In P. Baker (ed.) TRAC 98: proceedings of the eighth annual Theoretical Roman Archaeology Conference, Leicester, April 1998. (Oxford): 14-25. Johnson A. 1983 Roman forts of the 1st and 2nd centuries AD in Britain and the German provinces. (London). Jones A.H.M. 1964 The later Roman empire, 284-602: a social, economic and administrative survey. (Oxford). Jones M. 1991 Food production and consumption – plants. In R.F.J. Jones (ed.) Britain in the Roman period. Recent trends. (Sheffield): 21-28. Junkelmann M. 1997 Panis militaris: die Ernährung des römischen Soldaten oder der Grundstoff der Macht. (Mainz). King A. 1984 Animal bones and the dietary identity of military and civilian groups in Roman Britain. In T.F.C. Blagg & A.C. King (edd.) Military and Civilian in Roman Britain. British Archaeological Reports Brit. Ser.136. (Oxford): 187-217. King A. 1999a Animals and the Roman army: the evidence of animal bones. In A. Goldsworthy & I. Haynes (edd.) The Roman army as a community. Journal of Roman Archaeology Supplementary Series 34: 139-149 King A. 1999b Meat diet in the Roman world - regional inter-site comparisons. Journal of Roman Archaeology 12: 168-202. Kreuz A. 1999 How to become a Roman farmer. A preliminary report on the environmental evidence from the Romanization project. In J. Creighton & R. Wilson (edd.) Roman Germany: studies in cultural interaction. Journal of Roman Archaeology Supplementary Series 32: 71-98 Kuijper W.J. & Turner H. 1992 Diet of a Roman centurion at Alphen aan den Rijn, the Netherlands, in the first century AD. Review of Palaeobotany and Palynology 73: 187-204. Küster H. 1995 Weizen, Pfeffer, Tannenholz; botanische

943

Limes XVIII

Untersuchungen zur Verbreitung von Handelsgütern in römischer Zeit. Münstersche Beiträge zur antiken Handelsgeschichte 14.2: 1-26. Lane Fox R. 1996 Ancient hunting from Homer to Polybios. In G. Shipley & J. Salmon (edd.) Human landscapes in classical antiquity: environment and culture. (London): 119-153. Lauwerier R.C.G.M. 1983 A meal for the dead. Animal bone finds in Roman graves. Palaeohistoria 25: 183-194. Lauwerier R.C.G.M. 1988 Animals in Roman times in the Dutch Eastern River area. (‘s Gravenhage). Lepetz S. 1993 Les restes animaux dans les sépultures galloromaines. In A. Ferdière (ed.) Monde des morts, monde des vivants en Gaule rurale: actes du colloque ARCHÉA/AGER. (Tours): 37-44. Lepetz S. 1996 L’Animal dans la société Gallo-Romaine de la France du nord. Revue archéologique de Picardie 12. (Amiens). Maltby M. 1997 Domestic fowl on Romano-British sites inter-site comparison of abundance. International Journal of Osteoarchaeology 7(4): 402-414. Marichal R. 1992 Les ostraca de Bu Njem. (Département des Antiquités, Tripoli). Martin-Kilcher S. 1976 Das römische Gräberfeld von Courroux im Berner Jura. Basler Beiträge 2. Marvell A. & Owen-John H.S. 1997 Leucarum: excavations at the Roman auxiliary fort at Loughor, West Glamorgan 1982-84 and 1987-88. Britannia Monograph Series 12. (London). Millett M. 1990 The romanization of Britain. (Cambridge). Oliver A. & Shelton J. 1979 Silver on papyrus. Archaeology 32: 22-28. Peacock D.P.S. & Williams D.F. 1986 Amphorae and the Roman economy. (London). Pollard N. 1997 The Roman army as total institution in the Near East? Dura Europos as a case study. In D. Kennedy (ed.) The Roman army in the east. Journal of Roman Archaeology Supplementary Series 18: 211227. Remesal Rodríguez J. 1997 Heeresversorgung und die wirtschaftlichen Beziehungen zwischen der Baetica und die Germanien. (Stuttgart). Richmond I.A. 1962 The Roman siege works of Masada, Israel. Journal of Roman Studies 52: 143-155. Rieckhoff S. 1992 Eine römische Brauerei aus Regensburg. In E. Ruprechtsberger (ed.) Bier in Altertum: ein Überblick. Linzer archäologische Forschungen Sonderheft, 8. (Linz): 27-32. Roymans N. 1990 Tribal societies in northern Gaul. (Amsterdam). Roymans N. 1993 Romanization and the transformation of a martial elite ideology in a frontier province. In P. Brun, S. van der Leeuw & C.R. Whittaker (edd.) Frontières d’empire: nature et signification des frontières romaines. (Nemours): 33-50. Roymans N. 1996 The sword or the plough. Regional dynamics in the romanisation of Belgic Gaul and the Rhineland area. In N. Roymans (ed.) From the sword to the plough. Three studies on the earliest romanisation of northern Gaul. Amsterdam Archaeological Studies 1: (Amsterdam): 9-126. Ruprechtsberger E.M. 1992 Wirtin, füll die Flasche mit Bier Bier im griechisch-römischer Zeit: ein Überblick. In E. Ruprechtsberger (ed.) Bier in Altertum: ein Überblick. Linzer archäologische Forschungen Sonderheft, 8. (Linz): 15-24.

Schwinden L. 1985 Römerzeitliche Bleietiketten aus Trier. Zum Handel mit Pfeffer, Arznei und Kork. Trierer Zeitschrift 48: 121-137. Sealey P. & Tyers P.A. 1989 Olives from Roman Spain: a unique amphora find in British waters. Antiquaries Journal 69: 53-72. Seaward M. 1993 In C. van Driel-Murray The early wooden forts: preliminary reports on the leather, textiles, environmental evidence and dendrochronology. Vindolanda Research Reports III. (Hexham): 91-117. Solomon J. 1995 The Apician sauce: ius apicianum. In J. Wilkins, M.A. Dobson & F.D. Harvey (edd.) Food in antiquity. (Exeter): 115-131. Speidel M.A. 1996 Die römischen Schreibtafeln von Vindonissa. Veröffentlichungen der Gesellschaft pro Vindonissa 12. (Brugg). Swan V.G. 1992 Legio VI and its men: African legionaries in Britain. Journal of Roman Pottery Studies 5: 1-34. Tomlin R.S.O. 1992 The Roman carrot amphora and its Egyptian provenance. Journal of Egyptian Archaeology 78: 307-312. Tomlin R.S.O. 1998 Roman manuscripts from Carlisle: the ink-written tablets. Britannia XXIX: 31-84. van der Veen M. 1992 Crop husbandry regimes: an archaeobotanical study of farming in northern England 1000BC - AD 500. (Sheffield). van der Veen M. 1998 A life of luxury in the desert? The food and fodder supply of Mons Claudianus. Journal of Roman Archaeology 11: 101-116. van der Werff J., Thoen H. & van Dierendonck R. 1997 Amphora production in the lower Scheldt valley (Belgium)? The Valkenburg-Marktveld evidence. Rei Cretariae Romanae Fautorum Acta 35: 63-71. Woolf G. 1998 Becoming Roman: the origins of provincial civilization in Gaul. (Cambridge). Wright R.P. & Hassall M.W. 1974 Roman Britain in 1973. II Inscriptions. Britannia V: 461-470. Zienkiewicz J.D. 1986 The legionary fortress baths at Caerleon. (Cardiff). Zienkiewicz J.D. 1993 Excavations in the scamnum tribunorum at Caerleon: the legionary museum site 1983-1985. Britannia XXIV: 27-140.

944

Vespasianvs Velageno Margaret Roxan a.

the Italian fleets.3 Soon after this Bassus was adlected to the senate and then appointed governor of Judaea. While there he took the fort of Herodium and persuaded the inhabitants of Machaerus to surrender. Until recently epigraphic evidence of his service in Judaea has been lacking but a new study by Professor Werner Eck (1999) remedies this. He died before the siege of Masada, which was finally taken by his successor Flavius Silva.4

71 Apr. 5

The transcript of a complete diploma was shown to me (see Appendix 1 below). The transcript came from a reliable source and there is no reason to doubt the accuracy of its contents. When found the two tablets of the diploma were still fastened together but the wires were cut in order to obtain readings of the inner faces. The inner faces were reported to be still bright and unaffected by corrosion products. Two holes were pierced through each tablet; these were set wide apart, which is common in diplomas of the first three-quarters of the C1st. The capsule covering the seals of witnesses was still present. No measurements were given. It is likely that the tablets are similar in size of those of CIL XVI 12, which were issued on the same day.

A special diploma The recipients of the diploma were still serving, which places the document in the class of ‘special diplomas’. It falls into the same group as CIL XVI 160 of AD110, in which Trajan gave citizenship ante emerita stipendia to members of cohors I Brittonum milliaria Ulpia torquata pia fidelis because of their dutiful and faithful service in the expeditione Dacica. However, in this latter instance the recipients only received citizenship for themselves and presumably had to await discharge before their entitlement to conubium and citizenship for their children became possible. Vespasian showed more generosity in giving the navarchs, trierarchs and remigi the full privileges of veterans, that is citizenship for their children and posterity and the right of legal marriage with their wives. Further, they were given land in Pannonia, perhaps in a newly founded colonia.

The recipients According to Chester G. Starr (1960: 39) navarchs and trierarchs held important ranks in the Roman navy. He considered that the navarchy was “…in many respects comparable to the centurionate”. He further argued that the navarch held a position superior to that of the trierarch (cf. CIL VIII 21025).1 The latter he supposed would command a single ship, whereas a navarch was probably in charge of a squadron. He cited the cursus of P. Petronius Afrodisias, which is preserved in CIL XI 862 as an example of the relative status of the two posts since Afrodisias was promoted from trierach to navarch and finally navarchus princeps. The status of remigi is less clear. Their title implies that they were the oarsmen of a ship. Nevertheless their position in this diploma suggests that they may have been among fighting men of the fleets. If this is the case, the singling out of these three groups to be recipients of a special award emphasises the importance of the support of junior officers in maintaining the loyalty of the sailors to the Flavian cause.

The diploma sheds further light on CIL XVI 17, which was also issued in April 71. H. Nesselhauf restored this fragmentary document with the same formula that appears here: “quod se in expeditione belli fortiter industrieqque gesserant, ex auctorati sunt,” but in this case the recipients were veterans who had served for at least 26 years: “qui sena et [vice] na sti [pendia au] t plura meruissent”. The diplomas issued to veterans of fleets in 70 and 71 all specify 26 or more years. Since the recipients of the present document are still serving it is not possible to estimate how generous Vespasian was being in a reduction of years of service. Nevertheless, the precise reason given for his generosity in both cases underlines his political will to show how well disposed he was to those who served him – a necessary gesture at the beginning of his reign.

The prefect of the Ravenna fleet Sextus Lucilius Bassus is attested as commander of the Ravenna fleet in several diplomas: on 26 February 70, which was issued to beneficiarii, who were serving in this fleet (Roxan 1996), CIL XVI. 14 of 5 April 71 issued to veterans of “classe Ravennate” who were “deducti in Pannoniam”. The defection of Bassus from the cause of Vitellius and his subborning of members of the Ravenna fleet was an important factor in helping to secure Vespasian’s victory (Tacitus Hist. III.12) and this was given recognition in his promotion to command of both of

3

For the Misene fleet, see CIL XVI 12 and 113 of 9 February 71, as well as CIL XVI 15 and 16, both issued on the same day as the present diploma. Recipients of these diplomas were all veterans and were settled in Paestum. It is well attested that legionary veterans were given land or money on retirement but there is little or no other evidence to suggest that veterans of fleets were similarly rewarded. These exceptional awards highlight Vespasian’s need to show his gratitude to his supporters. 4 A fragmentary inscription from Abu Gosh has been re-evaluated by Eck and shown to refer to building work by a cohort whilst Bassus was governor. See also an appendix to this article in which it is tentatively proposed to restore the name of Bassus on a milestone from Jerusalem.

1

CIL VIII 21025 (caesarea M. Caes.) TI. CLAVDIO AVG. LIB. EROT. / TRIERARCHO LIBVRNAE NI/LI EXACTO CLASSIS AVG. / ALEXANDRINAE. L. IVLI/VS C.F. FAB. SATVRNINVS ET M. / ANTONIVS HERACLA TRIER / HEREDES EIVS FECERVNT 2 CIL XI 86 (Ravenna) D.M. / P.PETRONI AFRODISI / EX TRIERARC. NAVARC. ET PRINCIPE CL. PR. RAVEN. / ET GAVI (on)AE GORGONIAE EIVS / AELIVS CARPOPHORVS / PARENTIBVS

945

Limes XVIII

diploma and a list of names of recipients was set up on the Capitol, as was the norm before c.AD90.5 It was fastened to the outer left side of the balcony of the altar to the Julian family. The significance of sites chosen on the Capitol was explored by G. Forni (1986).

A political gesture “Et deducti in Pannoniam”. Tacitus (Hist. III.12.50) observed that the Ravennate fleet of AD69 was largely composed of Dalmatians and Pannonians. Starr (1960: 75, Table 1) produced a table based upon a relatively small sample from epigraphic sources, which supported this statement. He claimed that c.43% of sailors of the Ravenna fleet came from the Balkans compared with 28% for the Misene fleet. In granting land in Pannonia to veterans and ante emerita members of the Ravenna fleet Vespasian once more showed his acuity. These recipients were being returned to their homeland where they could provide a backbone of grateful loyal citizens, who would illustrate to their fellow Planarians just how generous the emperor was to his followers.

The witnesses Five of the seven witnesses to the accuracy of the diploma state that they were principales; as witnesses they should all have been Roman citizens, although not all appear to have tria nomina.6 The list of names of witnesses to the new diploma is new as in most of the diplomas issued before AD74. None of the names has been recorded on any other diploma. In common with other witnesses of the early period they give further information about themselves. T. Flavius Serenus belongs to the Iasi who were located in southern Pannonia; L. Icco Davus of the Breuci came from a similar area; C. Aledo Sammo of the Boii originated in northern Pannonia; C. Obromarus Tossia, whose names suggest an origin in Dalmatia, was also of a sept of the Boii; the names of Breucus Isticanus suggest that he came from the Breuci but he claims to belong to the Andizetes who were situated in the area at the confluence of the Drave with the Danube. C. Valerianus Nigel possesses the most Romanised name and served in the XIIIth Urban cohort, which was stationed at Lyon, where there was an important mint. The 7th and final witness also has a Roman name L. Licinius Aquila. Further information is enigmatic. Possibly CVR indicates curator, but NET does not equate with any known post or people (it is possible that this is a misreading and should be restored as Ne. f.). Diploma texts should have been copied and witnessed in Rome, and the basic assumption is that this group of men – all Roman citizens – was available in Rome to attest to the accuracy of the bronze copy of the original constitution. This is not unreasonable. During the various upheavals and troop movements of 69-70, Rome almost certainly had an influx of soldiers and veterans from units which had fought for one contender or another.

The date of issue of the diploma The first three lines of both faces give the regnal years of the emperor and a date of 71. This date is further ratified by the names of the consuls. The Caesar Domitian, Vespasian’s younger son, was suffect consul with Cn. Pedius Cascus in 71. The day date nonis April corresponds with the day of issue of CIL XVI 14, 15 and 16, whether this had some significance for the Flavian cause is, as yet, unclear. The recipient Velagenus proclaims that he is Eraviscan. His father’s name Covio is listed as Norican, by B. Lörincz (1999) on the basis of one inscription. The Eravisci lived south and west of the stretch of Danube leading up to the Danube ‘knee’ in Hungary. The recipient of CIL XVI 17 merely claimed to be Pannonian, but the fragment of his diploma was found at Grabarje in the Eraviscan domain. It is possible that the new diploma was also found in this area. Velagenus was a centurion. There are other examples of centurions in fleets and Starr suggested that this was the result of a Roman military framework having been superimposed on the original Greek naval organisation. He postulated that the crew of a warship formed a centuria under its centurion, regardless of its numerical strength. The sailors themselves were marines, in the modern sense of the word. They were trained to fight, as the formula of the diploma shows, and took part in warfare as soldiers – expeditione belli foriter industrieqque gesserant. This underlines the fact that the Adiutrix legions were formed with fleet personnel, and explains the background to the petition of veterans of legio X Fretensis, to the governor of Iudaea in 150. They had been formerly in the Misene fleet and asked for verification of their status in order that they could meet the requirements of epikrisis in Egypt (PSI 1929: n.1026).

Various attempts have been made by scholars to argue that witnessing did not necessarily take place in Rome. Saddington suggested that witnesses came from the “…nearest Roman colony to the recipient’s home area” in the case of a diploma issued on 26 February 70 to beneficiarii serving in the Ravenna fleet (Roxan 1996), in this case the colonia was Philippi with an additional witness from Aprensis. The recipient Dernaius gave his home as Dacus, and probably came from a group living south of the Haemus mountains in Thrace. This forms the basis of Saddington’s (1997) contention that witnesses signed in the administrative Roman centre closest to the recipient’s home. He draws a similar conclusion for the 5

The first extant diploma to signal the new site on the wall behind the Temple of the Deified Augustus, near the shrine of Minerva, is CIL XVI 36 of 27 October 90. 6 Epigraphy demonstrates that the use of the tria nomina was not universal until the latter part of the C1st AD

The constitutio The original bronze setting out the privileges of the 946

Margaret Roxan: Vespasianvs Velageno

witnesses of CIL XVI 10, three of whom definitely came from Phillippi and one from Aprensis. The recipient in this case gave his origin as natione Bessus (ie. he was Thracian) so that again the tie between recipient and witnesses is assumed by Saddington, who suggests that the “…signing was done in Philippi”. Other examples are offered, in each case a geographical connection between the recipient and the witnesses to his diploma is sought. Saddington queries whether this may not point to the existence of facilities in certain cities in provinces for the first stages of documentation of recruits and later for the certification of the authenticity of their diplomas. If correct, this theory adds a layer of bureaucracy to the issue of diplomas, which seems unnecessary. The constitutions promulgated in Rome listed all the veterans who were entitled to receive certain privileges according to their type and length of honourable service.7 This applied to men who did not petition (or pay) for bronze copies of a constitutio. Those veterans who requested the copies were probably men who needed proof of their new status for particular purposes eg. registering as citizens in a particular colonia with an ambition of moving up the social scale in that city. It is unfortunate that recipients always gave their original homes rather than the name (and tribus) of the city in which they hope to settle (Speidel 1986).

milliaria Ulpia torquata p. f. c. R. for bravery in the Dacian war. Veterans of the Mauretanian war of Antonius Pius were mentioned as having taken part in the expeditione, in CIL XVI 99 of AD150. The very fact of these specific clauses points to the issue of these particular diplomas as falling into a special class. ‘Normal’ diplomas, of which we know about 500 examples, do not include reasons for their issue other than the qualifying length of service and honourable discharge. It would be strange indeed if reasons for an issue honoris causa were omitted in all these diplomas if Dusanić’s theory is correct. If the general sweep of Roman military history of the late C1st and early C2nd is considered there are a number of warlike phases in which auxiliary troops and sailors would necessarily be involved. It is relatively easy to ‘cherrypick’ dates and events which appear to fit in with the issue of diplomas but this does not constitute proof of a causal connection. It looks like special pleading and undermines the generous concept that surely lies behind the grants. This was to give suitable rewards for long and faithful service to Rome and to create a core of grateful citizens whose loyalty and worth had already been tested.

Another theory has been advanced by Slobodan Dušanić (1982) who believes that diplomata should be regarded as rewards for special merits and that they were not automatically issued to all post-Claudian auxiliaries and sailors who had served the appropriate stipendia. He prefers to assume that the recipients had participated in a qualifying event namely a specific military effort. The witnesses to diplomas, according to his thesis, were soldiers, municipal dignitaries and clerks of freedman status and that they generally came from the place from which the candidates went to Rome. The task of witnesses was to warrant the substance of the document rather than the wording of the text. He examined the six ClaudioNeronian diplomata published at the time of his study and produced suggestions concerning the qualifying events prompting their issue. One such suggestion was participation in Nero’s Parthian War of AD63.

Dušanić S. 1982 The issue of military diplomas under Claudius and Nero. Zeitschrift für Papyrolögie und Epigraphik 47: 149-171. Eck W. 1999 Scripta Israelica Classica 17: 109-118. Forni G. 1986 I diplomi militari dei classiari delle flotte pretorie (inclusi quelli dei classiari-legionari). In W. Eck & H. Wolff (edd.) Heer und Integrationspolitik. Die römischen Militärdiplome als historischer Quelle. (Wien) : 293-321 (republished in M.P. Spediel (ed.) Escercito e Marina di Roma Antica. Raccolta di Contributi da Giovanni Forni. (Stuttgart) 1992: 419450). Lörincz B. 1999 Onomasticon Provinciarum Europae Latinarum Vol. II. (Vienna). PSI 1929 = G. Vitelli, M. Norsa & V Bartoletti (edd.) Papiri greci e latini. Pubbliciazione della Soc. Italiana per la ricerca dei pap. Greci e latini in Egitto. IX. Roxan M. 1996 An emperor rewards his supporters: the earliest extant diploma issued by Vespasian. Journal of Roman Archaeology 9: 247-256. Saddington D. 1997 The witnessing of pre- and early Flavian military diplomas and discharge procedures in the Roman army. Epigraphica LIX: 157-172. Speidel M.P. 1986 The soldiers’ homes. In W. Eck & H. Wolff (edd.) Heer und Integrationspolitik. Die römischen Militärdiplome als historischer Quelle. (Wien): 466481. Starr C. G. 1960 Roman imperial navy 31 BC – AD 324. (2nd. edit, Cambridge).

Bibliography

The main objection to such an idea is the lack of any specific mention of an event leading up to the issue. There are diplomas which actually state why the soldiers named were being given privileges. Such, for example, are the diplomas issued to Velagenus and his conferes, who had clearly fought on the Flavian side during the struggle between Vitellius and Vespasian. They were rewarded ante emerita stipendia and the nature of this award places it in a class of diplomas given under very specific circumstances. Vespasian initiated several such constitutiones.8 Later in 110, Trajan gave a similar award to cohors I Brittonum 7 This is illustrated by the inner face of CIL XVI 153 which has been interpreted as part of a re-used bronze tablet from Rome. 8 See n.6 above and CIL XVI 17, as restored by Nesselhauf, as well as the present issue.

947

Limes XVIII

948