Links to Late Antiquity: Ceramic exchange and contacts on the Atlantic Seaboard in the 5th to 7th centuries AD 1407316397, 9781407316390, 9781407323251

Table of contents :
Cover
Title page
Copyright
Acknowledgements
Contents
List of figures
List of tables
Abbreviated references
Abstract
1 -- Introduction
2 -- History of research
3 -- Late Antique imported pottery on the Atlantic Seaboard: developments in research
4 -- Imported pottery in Britain and Ireland: New evidence and understandings
5 -- Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France
6 -- Ceramic imports to Atlantic Spain and Portugal in Late Antiquity
7 -- Discussion and conclusions: Britain in the Late Antique Atlantic
Bibliography
Appendices

Citation preview

Maria Duggan obtained her PhD in Archaeology in 2016 at the School of History, Classics and Archaeology, Newcastle University. Since then she has taken up a British Academy Postdoctoral Fellowship at Newcastle, with a project entitled ‘Tintagel: Trans-European Connections in the Post-Roman World’. Focusing on this important site in Cornwall, this new project will extend research on the imported pottery found in Britain and advance understandings of connections between the Mediterranean and Atlantic during the 5th to 7th centuries.

BAR  639  2018   DUGGAN   LINKS TO LATE ANTIQUITY

This publication began as an AHRC-funded doctoral thesis, ‘Links to Late Antiquity: Understanding Contacts on the Western Seaboard in the 5th to 7th Centuries’, completed at Newcastle University in 2016. This revised version presents a broad-scale discussion of the evidence for contacts and connections in the Atlantic Seaboard region, based principally on ceramics. It extends knowledge of a category of material with a long history of scholarship in Britain and Ireland: amphorae and fineware vessels of East Mediterranean origin. The presence of this imported pottery at sites in western Britain, such as Tintagel in Cornwall, has frequently been used to suggest direct links between post-Roman Britain and the Byzantine World. This work offers an alternative position – that the wares reflect active and evolving networks of trans-shipment and exchange operating in the Atlantic Seaboard region between the fifth and seventh century. This first examination of parallel French, Spanish and Portuguese publications provides a fresh perspective on this important group of artefacts for understanding early medieval Britain.

Links to Late Antiquity Ceramic exchange and contacts on the Atlantic Seaboard in the 5th to 7th centuries AD

Maria Duggan

‘It is, above all, the analysis of the evidence from the Atlantic seaboard, from France, Portugal and Spain, which makes this work an essential point of reference for its field of research. Future researchers cannot afford to ignore this work.’ Prof. Michael Fulford, University of Reading ‘This is the first major study of Mediterranean material culture in early medieval Britain since Campbell’s 2007 study, and is unique in that it situates British finds within a wider Atlantic coast context. Doing so allows the author to present an original thesis which challenges the received wisdom in this area and presents an alternative model of coastal trade in the period.’ Dr Ben Jervis, Cardiff University

BAR British Series 639 B A R

2018

Links to Late Antiquity Ceramic exchange and contacts on the Atlantic Seaboard in the 5th to 7th centuries AD

Maria Duggan

BAR British Series 639 2018

Published in by BAR Publishing, Oxford BAR British Series Links to Late Antiquity © Maria Duggan The Atlantic Seaboard: view looking north-east from the headland at Tintagel, Cornwall. The Author’s moral rights under the UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reser ved. No par t of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any for m of digital for mat or transmitted in any for m digitally, without the written per mission of the Publisher.

ISBN 9781407316390 paperback ISBN 9781407323251 e-format DOI https://doi.org/10.30861/9781407316390 A catalogue record for this book is available from the British Library

BAR titles are available from: BAR Publishing Banbury Rd, Oxford, [email protected] + ( ) + ( ) www.barpublishing.com

,

Dedicated to: Robert and Veronica Duggan and to Louise Brown

Acknowledgements This work would have been impossible without the help of numerous individuals. I am extremely grateful for the assistance and support provided by John Allan, Scott Ashley, Paul Bidwell, Terry Bruce, Thomas Cadbury (RAMM, Exeter), Aurora Camaño, Barry Chandler (Torquay Museum), Rob Collins, John Davies, Jenny Durrant (RAMM, Exeter), Edgar Fernandes, Sandra Fletcher, Mike Fulford, Susan Greaney, John Hayes, Steve Hill (Plymouth University), Peter Holt, Lilian Ladle, Alan Langmaid (Totnes Museum), John Lund, Malcolm Lyne, Vicky Manolopoulou, Ray McBride, Frances McIntosh, Caron Newman, Colm O’Brien, Naomi Payne, Sophie Moore, Fiona Pitt (Plymouth City Museum), Philippe Planel, Paul Rainbird, Steve Roskams, Alex Turner, Paul Tyers, Joanita Vroom, and Jim Wilson (National Museum of Scotland). I am very grateful for the support given to this project by Devon County Council and the help provided by Bill Horner, County Archaeologist for Devon. Particular thanks must be given to Joachim Le Bomin, Frédéric Berthault, and to David Guitton and Armelle Guériteau (INRAP) who have provided immeasurable assistance with the French material, and to José Carlos Sanchez Pardo, Ana Patrícia Magalhães and José Carlos Quaresma who have been equally helpful with the Spanish and Portuguese finds and sites. Additional thanks must be offered to Adolfo Fernández Fernández for graciously sharing information and expertise, but also for taking valuable time to show me ceramic assemblages during a highly rewarding research trip to Galicia. I am also indebted to Michel Bonifay for his advice and encouragement. My understandings of the Insular finds are founded on the inspirational catalogues, syntheses and discussions published by Charles Thomas, Carl Thorpe, Mike Fulford, Anthea Harris, Bernard Wailes, Ian Doyle, Amanda Kelly and Ewan Campbell, for which I am extremely grateful. Ewan’s generosity in sharing his data online has established a benchmark for future collaborative research in the Atlantic. His assistance with this project has therefore been incalculable and, on a personal level, I am extremely grateful for his encouragement. This work is naturally reliant on the reports, theses and publications of many researchers and ceramic specialists working across the Atlantic and Mediterranean regions, including the inspirational works of John Hayes, Dominique Pieri and Paul Reynolds, and I must express gratitude to everyone cited in the text. I offer particular thanks to those who have shared information and data, and have allowed images to be reproduced. An enormous debt of gratitude is owed to my supervisors Sam Turner, Mark Jackson and James Gerrard – as well as to Kevin Greene – for their unceasing support. I am also grateful for the very helpful suggestions provided by the anonymous reviewers of this text. Finally, I must offer thanks to my family – particularly Veronica, Robert, Damian and Laura Duggan and Teresa Rendall – and to all my friends for always supporting my endeavours.

iv

Contents List of figures ................................................................................................................................................................... viii List of tables....................................................................................................................................................................... xi Abbreviated references .................................................................................................................................................... xii Abstract............................................................................................................................................................................ xiii Chapter 1. Introduction ..................................................................................................................................................... 1 1.1 Summary ................................................................................................................................................................... 1 1.2 Aims and objectives .................................................................................................................................................. 1 1.3 The imported material: an overview ...................................................................................................................... 1 1.4 Historical background and period terminology .................................................................................................... 2 1.5 Outline ....................................................................................................................................................................... 3 1.6 Methodology for data collection: Chapters 4, 5 and 6 .......................................................................................... 3 Chapter 2. History of research.......................................................................................................................................... 7 2.1 Introduction .............................................................................................................................................................. 7 2.2 Research on Late Antique imports to Britain and Ireland ................................................................................... 7 2.2.1 Principal researchers ......................................................................................................................................... 7 2.2.2 State of knowledge ......................................................................................................................................... 10 2.3 Models of transmission: the character of the Insular assemblage and the search for Atlantic parallels ....... 16 2.3.1 Ralegh Radford at Tintagel ............................................................................................................................. 16 2.3.2 Charles Thomas and Bernard Wailes .............................................................................................................. 16 2.3.3 John Hayes’ observations................................................................................................................................ 19 2.3.4 Michael Fulford’s model................................................................................................................................. 21 2.3.5 Ewan Campbell’s research .............................................................................................................................. 22 2.3.6 Additional considerations ............................................................................................................................... 24 2.3.7 The potential for new models ......................................................................................................................... 26 2.4 Conclusions ............................................................................................................................................................. 27 Chapter 3. Late Antique imported pottery on the Atlantic Seaboard: developments in research ........................... 29 3.1 Introduction ............................................................................................................................................................ 29 3.1.1 Issues of classification and terminology ......................................................................................................... 29 3.2 Amphorae ................................................................................................................................................................ 29 3.2.1 History of research in Britain and Ireland ...................................................................................................... 29 3.2.2 East Mediterranean amphorae ......................................................................................................................... 31 3.2.3 British and Irish examples of imported amphorae .......................................................................................... 33 3.2.4 North African amphorae: unravelling the ‘Bv’ classification ......................................................................... 34 3.2.5 Iberian amphorae ............................................................................................................................................ 38 3.2.6 Amphora contents ........................................................................................................................................... 40 3.3 Finewares ................................................................................................................................................................ 40 3.3.1 Introduction..................................................................................................................................................... 40 3.3.2 African Red Slip (ARS) .................................................................................................................................. 41 3.3.3 Late Roman C (LRC)...................................................................................................................................... 45 3.3.4 Late Roman D (LRD) ..................................................................................................................................... 51 3.3.5 Dérivées des Sigillées Paléochrétiennes (DSP) .............................................................................................. 52 3.4 E Ware ..................................................................................................................................................................... 56 3.5 Other Materials ...................................................................................................................................................... 62 3.5.1 Coarsewares and unclassified ceramics .......................................................................................................... 62 3.5.2 Glass ............................................................................................................................................................... 63 3.5.3 Miscellaneous materials.................................................................................................................................. 64 3.6 Summary ................................................................................................................................................................. 64

v

Links to Late Antiquity Chapter 4. Imported pottery in Britain and Ireland: new evidence and understandings ........................................ 65 4.1 Introduction ............................................................................................................................................................ 65 4.2 Tintagel: reassessing the ceramic research .......................................................................................................... 65 4.2.1 Tintagel: history of investigations .................................................................................................................. 65 4.2.2 The Tintagel pottery: research prior to 1990 .................................................................................................. 66 4.2.3 The Tintagel pottery: discussion of the imports reported from the 1990-99 excavations .............................. 67 4.2.4 Discussion and recommendations................................................................................................................... 69 4.3 Early medieval imports on the South Devon coast: a case-study for future directions in research ............... 70 4.3.1 Introduction..................................................................................................................................................... 70 4.3.2 Bantham .......................................................................................................................................................... 70 4.3.3 Mothecombe ................................................................................................................................................... 77 4.3.4 High Peak, Sidmouth ...................................................................................................................................... 81 4.3.5 An amphora recovered from Plymouth Sound ............................................................................................... 85 4.3.6 South Devon case study: discussion ............................................................................................................... 87 4.4 Additional recent British discoveries .................................................................................................................... 88 4.4.1 Pevensey Castle .............................................................................................................................................. 88 4.4.2 Rhynie ............................................................................................................................................................. 89 4.4.3 Druce Farm, Dorset ........................................................................................................................................ 89 4.5 Conclusions and directions in research ................................................................................................................ 89 Chapter 5. Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France .......... 91 5.1 Introduction ............................................................................................................................................................ 91 5.2 Mediterranean imports to western France in Late Antiquity ............................................................................ 91 5.2.1 Imported Mediterranean amphorae................................................................................................................. 91 5.2.2 Imported Mediterranean finewares ................................................................................................................. 94 5.3 Case Study: ceramic imports to Bordeaux........................................................................................................... 95 5.3.1 Introduction..................................................................................................................................................... 95 5.3.2 Saint-Seurin .................................................................................................................................................... 96 5.3.3 Saint-Christoly ................................................................................................................................................ 97 5.3.4 Place Camille-Jullian ...................................................................................................................................... 97 5.3.5 Discussion ....................................................................................................................................................... 99 5.4 E ware: new evidence for production and distribution .................................................................................... 100 5.4.1 Efforts to identify Continental parallels ........................................................................................................ 100 5.4.2 Continental examples of E ware ................................................................................................................... 101 5.4.3 The ‘enigma’ of the origin of E ware ............................................................................................................ 103 5.4.4 E Ware: a product of Bordeaux? ................................................................................................................... 114 5.4.5 E ware in Spain ............................................................................................................................................. 115 5.5 Conclusions ............................................................................................................................................................115 Chapter 6. Ceramic imports to Atlantic Spain and Portugal in Late Antiquity .......................................................116 6.1 Introduction ...........................................................................................................................................................116 6.2 Parallels for the British imports...........................................................................................................................116 6.2.1 Imported Mediterranean finewares in the Atlantic ....................................................................................... 116 6.2.2 East Mediterranean finewares (LRC and LRD)............................................................................................ 117 6.2.3 African Red Slip Ware (ARS)....................................................................................................................... 121 6.2.4 Imported Mediterranean amphorae............................................................................................................... 125 6.3 Regional case studies: sites with imported Mediterranean pottery in Atlantic Spain and Portugal ............ 127 6.3.1 Southern and central Portugal ....................................................................................................................... 127 6.3.2 Northern Portugal and north-west Spain (Gallaecia – Galicia) ................................................................... 134 6.3.3 Northern Spain .............................................................................................................................................. 147 6.4 Conclusions ........................................................................................................................................................... 149 Chapter 7. Discussion and conclusions: Britain in the Late Antique Atlantic ......................................................... 150 7.1 Introduction .......................................................................................................................................................... 150 7.2 Discussion: new models of contact and connection ........................................................................................... 150 7.2.1 Late Roman connections in the Atlantic ....................................................................................................... 150 7.2.2 An Atlantic Network: new understandings of importation and connection in the late-fifth and sixth century ........................................................................................................................................... 151 7.2.3 Vigo: an ‘Atlantic emporium’ ....................................................................................................................... 154 7.2.4 The transformation and end of the Atlantic network .................................................................................... 156 vi

Contents 7.2.5 New models of Atlantic exchange ................................................................................................................ 156 7.3 Future research and recommendations .............................................................................................................. 159 7.4 Conclusions ........................................................................................................................................................... 160 Bibliography ................................................................................................................................................................... 161 Appendices ...................................................................................................................................................................... 173 Appendix A: Concordance of ceramic terminology ................................................................................................ 173 Appendix B: Location maps (individually scaled) .................................................................................................. 176 Appendix C: ARS Chronology .................................................................................................................................. 183 Appendix D: Mothecombe ceramic data .................................................................................................................. 187 Appendix E: High Peak ceramic data ...................................................................................................................... 188 Appendix F: Cawsand amphora ............................................................................................................................... 191 Appendix G: Quantities of amphorae from sites in South Devon, Tintagel and selected sites in south-west Britain ................................................................................................................................................ 192 Appendix H: Ceramic distribution maps. ................................................................................................................ 193 Appendix I: Schematic/speculative maps of supply and exchange ........................................................................ 204 Appendix J: E ware minimum vessel totals after Campbell 2011 and Doyle 2009 .............................................. 206 Appendix K: Atlantic sites with Mediterranean imports and DSPA (E ware totals shown from these sites) and Continental sites with E ware ...................................................................................................................... 208

vii

List of figures Figure 1.1. The political landscape of Western Europe c.525 ............................................................................................. 2 Figure 2.1. Thomas’ distribution map of Mediterranean pottery showing suggested voyages supplying these wares ....... 7 Figure 2.2. Campbell’s distribution maps of all Mediterranean (left) and Continental imports (right) .............................. 9 Figure 2.3. ‘Association of Mediterranean imports with tin and lead production’ in south-west Britain ......................... 13 Figure 2.4. ‘Distribution of imported pottery and Byzantine ‘traders’’ ............................................................................. 15 Figure 2.5. The distribution of fineware comparable to British ‘A ware’ and proposed routes of transmission ............... 18 Figure 2.6. The ‘approximate main distribution areas’ of ARS Forms 103-4 (top) and LRC (bottom) as shown in LRP 19 Figure 2.7. The distribution of LRC Form 3 shown in LRP .............................................................................................. 20 Figure 2.8. Campbell’s map of LRC distribution in the western Mediterranean ............................................................... 22 Figure 2.9. Reynolds’ 2010 distribution of LRC in the western Mediterranean................................................................ 27 Figure 3.1. Riley’s Late Roman Amphora types ................................................................................................................ 30 Figure 3.2. Pieri’s East Mediterranean amphora sub-types (LRA1, 2 and 4 only) ............................................................ 31 Figure 3.3. Imported amphorae from Britain and Ireland by MNV .................................................................................. 34 Figure 3.4. Imported amphorae from Britain and Ireland by percentage of MNV ............................................................ 35 Figure 3.5. Bonifay’s typological development of African amphorae, highlighting selected ‘fifth century’ and later amphorae .................................................................................................................................................................... 37 Figure 3.6. Composite illustration of a ‘Bv’ amphora ....................................................................................................... 39 Figure 3.7. ARS production zones and workshops in the region of Carthage, modern Tunisia ........................................ 42 Figure 3.8. Reconstructed ARS 103B with cross-stamp from Cadbury Congresbury....................................................... 46 Figure 3.9. Radford’s ‘Ai’ fineware subgroup (top) and ‘Aii’ (bottom) from Tintagel ..................................................... 47 Figure 3.10. Hayes’ LRC Form 3 variants C, E and F and Form 10 ................................................................................. 48 Figure 3.11. 1973 Distribution map of DSPA .................................................................................................................... 54 Figure 3.12. Principal forms of DSPA ............................................................................................................................... 55 Figure 3.13. DSPA vessels from Dinas Powys .................................................................................................................. 55 Figure 3.14. DSP sherd with Chi-Rho stamp from Dinas Emrys ...................................................................................... 56 Figure 3.15. Thomas’ E ware form typology ..................................................................................................................... 58 Figure 3.16. Campbell’s E ware forms .............................................................................................................................. 59 Figure 3.17. E ware rim/handle junction showing characteristic ‘thumb-smear’. From Dunadd ..................................... 60 Figure 3.18. E ware base showing characteristic string-marks. From Mote of Mark........................................................ 61 Figure 3.19. Campbell’s distribution of imported glass in Britain and Ireland ................................................................. 63 Figure 4.1. The location of Radford’s investigations on Tintagel Island ........................................................................... 65 Figure 4.2. The location of investigations at Bantham ...................................................................................................... 71 Figure 4.3. Bantham spindle-whorls .................................................................................................................................. 72 Figure 4.4. LRA1 (left) and LRA1 ‘Bantham type’ (right) from Bantham ....................................................................... 74 Figure 4.5. LRA1 (top) and LRA1 ‘Bantham type’ (bottom) from Bantham.................................................................... 74 Figure 4.6. LRC 3 from Bantham: ‘small variety’ (top), 3F (bottom) ............................................................................... 75 viii

List of figures Figure 4.7. Vessel from Bantham described as ‘rich in muscovite’................................................................................... 76 Figure 4.8. The location of investigations at Mothecombe................................................................................................ 78 Figure 4.9. Mothecombe spindle-whorls ........................................................................................................................... 79 Figure 4.10. Imported amphora sherds from Mothecombe: P5a and P6a (left), P7, P8a, b and P9 (right) ....................... 80 Figure 4.11. The location of investigations at High Peak .................................................................................................. 82 Figure 4.12. LRA1 sherds recovered at High Peak in 2012 .............................................................................................. 84 Figure 4.13. Photograph of LRA1 found off Cawsand ...................................................................................................... 86 Figure 4.14. Scaled illustration of LRA1 found off Cawsand ........................................................................................... 87 Figure 5.1. The church of Saint-Seurin, Bordeaux. View facing north across the location of excavations ...................... 96 Figure 5.2. Selected LRA1 and 2 from Place Camille-Jullian ........................................................................................... 98 Figure 5.3. Chronology of DSPA form repertoire at Bordeaux ......................................................................................... 99 Figure 5.4. Wailes’ distribution map of E ware, showing a suggested source ................................................................. 100 Figure 5.5. Possible E ware from Rezé. Lids numbered by Pirault 1-5, bases 10-13 ...................................................... 102 Figure 5.6. Lids from Place Camille-Jullian – equating with E5..................................................................................... 107 Figure 5.7. Selected bowls/‘coupes carenées’ from Place Camille-Jullian – including examples equating with E3 ...... 107 Figure 5.8. Selected cups/ ‘gobelets’ from Place Camille-Jullian – equating with E2 .................................................... 108 Figure 5.9. Cups/ ‘gobelets’ from Place Camille-Jullian with pouring rims ................................................................... 108 Figure 5.10. Selected jars from Place Camille-Jullian – potentially equating with E1 ................................................... 109 Figure 5.11. Trimmed pots/jars from Place Camille-Jullian ............................................................................................ 109 Figure 5.12. Pitcher and Jug from Place Camille-Jullian – equating with E4B and E4 respectively .............................. 109 Figure 5.13. Jugs from Place Gabriel – equating with E4 ................................................................................................111 Figure 5.14. Pot with undulating profile from Îlot Bonnac – potentially equating with E7 ............................................ 112 Figure 5.15. Pierced lid from Puy Paulin showing string-marks – potentially equating with E5 ................................... 113 Figure 6.1. Beltrán’s western Iberian distribution of LRC .............................................................................................. 118 Figure 6.2. Reynolds’ 1995 map of LRC distribution in the western Mediterranean ...................................................... 119 Figure 6.3. The distribution of ‘stamp style A and associated forms’ shown in LRP ...................................................... 122 Figure 6.4. The Iberian distribution of ARS-D in the fourth and first half of the fifth century (left) and second half of fifth and sixth century (right)................................................................................................................................ 124 Figure 6.5. Fabião’s distribution map of Byzantine weights in Portugal......................................................................... 132 Figure 6.6. Examples of LRC 3 ‘small variety’ from Conimbriga .................................................................................. 133 Figure 6.7. LRD from Braga ............................................................................................................................................ 135 Figure 6.8. Patterns of fineware importation at Braga ..................................................................................................... 135 Figure 6.9. The distribution of ‘Terra Sigillata Bracarense Tardía’ in the Atlantic ........................................................ 137 Figure 6.10. Location of excavations at Vigo with late pottery studied by Fernández .................................................... 139 Figure 6.11. LRC forms from Vigo and Britain and Ireland by MNV ............................................................................ 140 Figure 6.12. Example of TSBT Form 8 from Vigo .......................................................................................................... 141 Figure 6.13. DSPA forms at Vigo and in Britain and Ireland by percentage ................................................................... 144 Figure 6.14. Selected E ware from Vigo – including examples of E3 and E4B .............................................................. 146 Figure 6.15. Examples of DSPA with deer stamps from Gijón (top left), Vigo (middle left) and Place Camille-Jullian, Bordeaux (bottom left) ..................................................................................................................................................... 148 Figure 7.1. Small-world systems: one characterised by ‘weak ties’ and the other as a ‘scale-free’ network .................. 157 ix

Links to Late Antiquity Figure 7.2. Circuits of Medieval exchange in the Indian Ocean ..................................................................................... 158 Figure 7.3. The eight circuits of the thirteenth century world system ............................................................................. 159

x

List of tables Table 1. LRC forms and chronology.................................................................................................................................. 49 Table 2. Insular examples of published LRC stamps ......................................................................................................... 50 Table 3. LRD forms and chronology ................................................................................................................................. 52 Table 4. E ware forms and function ................................................................................................................................... 58 Table 5. Relative proportion of amphorae at Tintagel by sherd count, vessel count and percentage ................................ 68 Table 6. Early medieval pottery recovered at Bantham in 2001 ........................................................................................ 73 Table 7. Quantification of all pottery recovered at Mothecombe since 1995 .................................................................... 78 Table 8. Count of amphora sherds from High Peak, Sidmouth ......................................................................................... 82 Table 9. Sherds recovered from High Peak........................................................................................................................ 83 Table 10. Sherd counts from phases of investigation at High Peak ................................................................................... 83 Table 11. Estimated minimum vessel counts from High Peak........................................................................................... 84 Table 12. Vessel counts of late imported amphorae at Bordeaux and Toulouse ................................................................ 96 Table 13. Later Roman and early medieval phases at Place Camille-Jullian, Bordeaux ................................................... 98 Table 14. Pottery from Îlot Bonnac, Bordeaux .................................................................................................................111 Table 15. Fineware totals from Fernández’s research at Vigo ......................................................................................... 140 Table 16. Comparison of the chronology of ceramic importation to Vigo and to Britain and Ireland ............................ 142 Table 17. Imported amphorae from Contexts 19-22 at Vigo ........................................................................................... 145

xi

Abbreviated references A few key publications which are commonly cited have been abbreviated: LRP – Hayes, J. W. 1972. Late Roman Pottery. LRFW1 – Cau, M. Á., Reynolds, P. and Bonifay, M. (eds) 2011a. LRFW1: Late Roman Fine Wares: Solving Problems of Typology and Chronology: a Review of the Evidence, Debate and New Contexts. Provisional List – Thomas, C. 1981. A Provisional List of Imported Pottery in Post-Roman Western Britain and Ireland. Supplement – Hayes, J. W. 1980. A Supplement to Late Roman Pottery

xii

Abstract to refine chronological sequences and to better understand its origins and function. This new review will fully evaluate these developments to determine whether the imported ceramics reflect previously unidentified networks of trade or exchange between communities in western Britain, Ireland, France, Spain and Portugal in Late Antiquity.

In western Britain, particularly the south-west, imported pottery of Mediterranean origin has provided an important means of recognising sites of fifth and sixth century date. These post-Roman imports have typically been seen to represent direct shipments from the East Mediterranean – and therefore to reflect some sort of direct connection between western Britain and the Byzantine world in the later-fifth and sixth century. This model is founded on the apparent scarcity of comparable material on the Western Seaboard and on observations of the unique composition of the British assemblage. Associated finds of pottery of Continental origin have been connected to a separate and subsequent phase of importation in the sixth and seventh century.

The intention of this work is to consider the pottery in its wider Continental and Mediterranean context. Its specific aim is to question the two factors that underpin the model of direct connection with the East Mediterranean. It will assess the character of the British material using published catalogues, syntheses and newly-conducted research on pottery recovered from sites in south-western Britain. The presence of comparable material on the Western Seaboard will also be examined. Comparisons of the composition of the British and Atlantic assemblages will be used to suggest revised models of exchange and contact between sites in Britain, on the Atlantic Seaboard and in the Mediterranean.

Recent archaeological work has increased the quantity and range of the imported pottery identified in Britain and Ireland, while new discoveries on the Atlantic Seaboard have suggested a more complex picture of distribution. The material has been the subject of continuing research

xiii

1 Introduction 1.1 Summary

underpin the traditional model of direct connection with the East Mediterranean: the exceptional character of the Insular assemblage and the apparent lack of parallel material along the Atlantic Seaboard.

Pottery of Mediterranean origin has been recognised at sites of early medieval date in western Britain since Ralegh Radford’s excavations at Tintagel in Cornwall (Radford 1956). Since the 1950s, sherds have been catalogued from sites across western Britain and Ireland and models for the arrival and distribution of these imports have been developed (Thomas 1959, 1981; Fulford 1989; Doyle 2009). Ewan Campbell’s major synthesis of the Insular finds proposed a regular system of importation from the Mediterranean via Atlantic channels operating c. AD 475-550 (Campbell 2007a).

To address the first factor, this work will reconsider the character of the British material, using published catalogues and recent publications as well as primary research and analysis conducted in south-western Britain. To address the second issue, it will assess the presence of comparable material on the Atlantic Seaboard, identifying the types of pottery present and, where possible, relative quantities. Comparisons between the British and Atlantic assemblages will be considered and used to suggest revised models of exchange and contact between sites in Britain, Ireland, France, Spain and Portugal – as well as links with the Mediterranean world – between the fifth and seventh centuries AD.

The arrival of new types and forms of pottery to Britain in the mid to later-fifth century, with a distinct western and coastal distribution, indicated the emergence of a new and discrete post-Roman import system, which has typically been characterised by a model of direct shipment from the East Mediterranean. This model of transportation is founded on two observations: the seemingly unique composition of the British assemblage and the apparent scarcity of comparable material on the Western/Atlantic Seaboard. Pottery of western Gaulish origin has been connected to a separate and subsequent phase of ‘Continental’ importation in the sixth and seventh century.

1.3 The imported material: an overview The imported pottery from the Mediterranean comprises two main categories: amphorae and Red Slip fineware. The amphorae are principally of East Mediterranean and, to a lesser extent, North-African types. Grouped as ‘B wares’ in earlier British publications they are now, more usefully, matched to amphora classifications established in the Mediterranean (Campbell 2007a, 4). Similarly, the ‘A ware’ group coined by Radford was matched to Mediterranean classes ‘African Red Slip Ware’ (ARS) and ‘Phocaean Red Slip Ware’ (PRS) by Thomas’ 1981 Catalogue (Thomas 1981, 3). The latter class has returned to its earlier designation ‘Late Roman C’ (LRC) in some recent publications, and this convention has been followed (Cau et al. 2011, 6).

New information emerging from the Continent has suggested that these interpretations need to be reconsidered. Increasing evidence from Britain has led to new understandings of the character and chronology of the Insular imports, while ceramic data appearing from sites on the Continent have started to fill the ‘gap’ in distribution on the Atlantic Seaboard. Revised patterns of ceramic distribution in the Atlantic suggest that the British ‘import sites’ were integrated into more complex systems of trade or exchange than previously recognised.

The main types of amphorae identified in British contexts are Late Roman 1 (hereafter LRA1) and Late Roman 2 (LRA2), previously classified in Britain as ‘Bii’ and ‘Bi’ respectively (Thomas 1959). Both types were produced between the fourth and seventh century but are not thought to have been imported into Roman Britain. Their identification at sites in western Britain is therefore taken to indicate some connection to a separate import system commencing in the later-fifth century (Campbell 2007a, 19). Other East Mediterranean amphorae (LRA3, LRA4), which occur in lesser numbers on these post-Roman import sites, have been reported from late Roman contexts in Britain (Campbell 2007a, 19-20). Likewise, amphorae of North African origin were imported to Roman Britain, particularly in the third and fourth century (Williams and Carreras 1995, 234) but are also found – though in a smaller proportion to East Mediterranean types – within post-Roman assemblages. The forms of amphora are long-

This new appraisal will examine recent publications on ceramic imports to Britain, particularly those that offer new understandings of the date and character of this import system. It will also assess emerging evidence from the Western Seaboard, particularly from south-western France, north-western Spain and Portugal. This will allow new comparisons to be drawn between patterns of pottery importation and use in Britain and the wider Atlantic region in Late Antiquity. 1.2 Aims and objectives The intention of this work is to consider the ceramic evidence from Britain in its wider Atlantic and Mediterranean context. Its specific aim is to question the two factors that 1

Links to Late Antiquity lasting and cannot usually be closely dated in themselves. Instead, the dates reflect production dates based on typologies established in the Mediterranean, particularly for the Red Slip finewares (Campbell 2007a, 19).

A full concordance of the specific ceramic terminology is provided in Appendix A.

A secondary and subsequent phase of imported pottery from the Continent has also been identified (Thomas 1959). ‘E ware’ is a coarseware of presumed western Gaulish origin, which has a wider distribution in western Britain and Ireland (Campbell 2007a 46-7). The main period for its importation is thought to be the later-sixth and seventh centuries (Campbell 2007a, 46). Present only in very small numbers in Insular contexts a second ware, ‘Dérivées des Sigillées Paléochretiénnes’ (DSP), has been thought to overlap the main phases of Mediterranean and Continental importation (Campbell 2007a, 133).

This work describes patterns in ceramic importation to Britain and the Western Seaboard between the fifth and seventh centuries, although some discussion of preceding and subsequent developments is incorporated. This period witnessed considerable change in the Atlantic region, epitomised by the end of Roman control in the West in the fifth century and the rise of the successor states. It is not feasible to fully outline these complex developments; understandings of the history of Late Antique and early medieval Europe presented in the following chapters have principally been drawn from detailed narratives in Harris 2003, Wickham 2005, Knight 2007 and Collins 2010. The political landscape shifted markedly from the fourth century, when all of the regions considered – apart from Scotland and Ireland – fell under direct Roman control. The map in Figure 1.1 presents the situation c. AD 525, by which point the Visigothic Kingdom was largely restricted to the Iberian Peninsula and Frankish rule was fully established in in the former provinces of northern/western

1.4 Historical background and period terminology

Despite the long history of this research the overall quantities involved remain relatively modest, with c. 322 imported amphorae reported from sites in Britain and Ireland. The amphorae – used to transport commodities such as wine or olive-oil – are more common than tablewares, with c. 126 vessels of Mediterranean and Continental origin reported from western Britain and Ireland.

Figure 1.1. The political landscape of Western Europe c.525. From Jotischky and Hull 2005, 21

2

Introduction assemblage – recently published following excavations at Bantham – is evaluated within a case study of imported pottery in South Devon, and used to assess emerging trends in the British data. Within this case study, original research conducted as part of this project is presented, summarising published reports of pottery recovered at Mothecombe and High Peak in Devon and an amphora found in the sea at Cawsand, Cornwall.

Gaul. Prior to this – between c. 418/9 and 507 – Aquitaine (including Bordeaux) had been subject to Visigothic settlement/authority (Collins 2010, 79-80). Suevic control in north-west Spain expanded from the early-fifth century, although the kingdom was eventually assimilated into the Visigothic State in 585. With the loss of Carthage in 439 all of the Roman holdings in North Africa fell under Vandal rule (Collins 2010, 79). Byzantine control was not re-established until 533. The early-fifth century saw the end of direct Roman control in Britain; the subsequent period is characterised by the division between the surviving indigenous ‘British’ communities of western Britain and the regions of ‘Anglo-Saxon’ settlement and control to the east (Harris 2003, 143). This split has been seen to characterise the Insular distribution of the imported pottery.

The work then moves to discuss the emerging Continental evidence, starting with the Mediterranean imports reported from France, and particularly from Bordeaux (Chapter Five). This chapter also includes a consideration of types of Continental pottery that were imported to sites in western Britain – DSPA and E ware. New evidence for the origin and distribution of these wares is considered, and its importance for wider understandings of Atlantic networks is assessed. The following chapter (Chapter Six) focuses on the considerable evidence for Mediterranean imports in Atlantic Spain and Portugal. This includes a case-study of the site at Vigo, which outlines its unique significance for determining connections with early medieval Britain.

As this project covers a considerable geographical area some complications with the broad chronological terminology were unavoidable. Within the text I have followed Campbell in using ‘early medieval’ to refer to the broad period from the fifth to eleventh century – and, in particular, to refer to the British and Irish sites with imported pottery, thereby distinguishing them from earlier ‘Roman’ patterns (2007a, 3). In relation to Britain, the term ‘Roman’ refers to the period before AD 410 and ‘post-Roman’ to the subsequent period; this division cannot be directly applied across the Atlantic region. The term ‘Late Antique’ has been used to refer to a non-regionally specific period dating from the fourth to the seventh century. All dates mentioned represent AD dates.

Finally, Chapter Seven reviews the overall findings, with specific reference to the patterns of ceramic distribution presented in Appendix H. This new information is used to suggest revised models of contact between Britain, the Atlantic and the Mediterranean, and the wider social and economic importance of these findings is considered. Maps showing the locations of Atlantic sites with this pottery are presented in Appendix B. The respective ‘Site ID numbers’ (linking Appendix B and the tables of ceramic data presented in Appendix K) are shown in square brackets on the first mention of each site within the main text.

Some additional clarification can be offered for the general terminology used to describe the pottery. The amphorae are typically referred to as ‘Late Roman Amphorae’ but this represents a standard term for East Mediterranean vessels of fourth century and later date and does not, necessarily, equate to periods of Imperial control in the West. The terms ‘Mediterranean imports’ or ‘Continental imports’ have been used in reference to the source of the vessels, while ‘Insular imports’ has been used to refer to imports recovered in Britain and Ireland.

1.6 Methodology for data collection: Chapters 4, 5 and 6 The basic aim of the data collection was to assess the relative distribution of selected ceramic types – those identified as imports to post-Roman, western Britain – across the Atlantic region. The plotting of this data within distribution maps would allow geographical and, potentially, temporal patterns in Atlantic supply to be determined. Given the long history of publications on pottery imported to Britain and Ireland, it was not the intention of this project to provide a full reassessment of the Insular evidence. Understandings of the character of the British and Irish import assemblage are, therefore, dependent on catalogues of ceramic finds published by Charles Thomas (1959; 1981) and, principally, on the expanded dataset compiled by Ewan Campbell (2007a; 2011). It was also not viable to view each sherd/assemblage in person; the Insular data has expanded to such a point where such an undertaking was not feasible, and a replication of Campbell’s extensive research was certainly unnecessary. The major assemblage from Bantham was, however, viewed in person, and is considered – together with assemblages from south Devon assessed as part of this project – in Chapter Four.

1.5 Outline After introducing the topic, the material, and the background to this project (Chapter One), the history of research in Britain is reviewed (Chapter Two). This appraisal describes previous attempts to find Atlantic parallels for the British finds, and examines the associated development of models of connection between western Britain and the Byzantine world. A detailed discussion of the major types of pottery and their occurrence in Britain is presented in Chapter Three, updating older understandings of the chronology and typology of these wares with reference to recent Mediterranean research. Chapter Four assesses ceramic research at the major British site of Tintagel and suggests recommendations for future work on its assemblage. The significance of a second large 3

Links to Late Antiquity Data sources and geographical limits

the primary ceramic or excavation reports were located and examined – although many of the accessible sources comprise published syntheses, catalogues, lists of data or annotated maps. The extent and quality of information available for each site, excavation or find-spot was extremely variable, and in many cases little more could be established than the presence of a particular ware or form. As such, the discussion in Chapters 5 and 6 is focused on broad, regional patterns, and on specific ‘case studies’ of larger assemblages and/or sites with dated stratigraphic sequences.

The Insular data presented here is largely founded on the digital archive of Ewan Campbell’s 2007 monograph, Early Medieval Imported Pottery and Glass in the Atlantic Province AD 400-800. The accompanying dataset was lodged with Archaeological Data Services (ADS) and made available online via the University of Glasgow – the latter of which was updated in 2011. This was downloaded, with permission, towards the start of the project in 2012. The dataset was subsequently amended on the University of Glasgow page in October 2012, but this has not been used – any references to Campbell’s dataset therefore refer to the 2011 version. The data for import sites in Ireland is founded on the extensive catalogue published by Ian Doyle and the detailed synthesis by Amanda Kelly (Doyle 2009; Kelly 2010).

Within Chapters 5 and 6 the supply of North African and East Mediterranean pottery to the Atlantic in the fourth century has been considered, but within the database I have attempted to limit the data to types likely to date from the fifth century onward. This was intended to enable closer comparisons with the post-Roman material from Britain – theoretically consisting of imports dating from the later fifth-century onwards. This restriction was not always easy to implement. A break in ceramic importation to Britain has been postulated for the middle decades of the fifth century, but there was not, necessarily, a comparable rupture in supply to Atlantic regions of France, Spain and Portugal. This, unsurprisingly, led to complications with ceramic types and forms that could have been imported in the later-fourth or earlier-fifth century. In these cases, I opted to include them in the dataset, although where possible I have mentioned that they might represent fourth century arrivals. The exception is for Britain, where, following Campbell, vessels considered as potential ‘Roman period’ imports have been discussed but not included in the data-tables or maps (Campbell 2007a, 22). Patterns of supply to the Atlantic in the earlier-fifth century are examined in Sub-Chapter 7.2.1, but it must be acknowledged that this phase remains to be fully understood.

Where possible, the original source publications reporting the discovery of the Mediterranean imports and the DSPA were located and reviewed. However, it must be noted that a significant number of the finds represent unpublished sherds that have been added to the Insular ‘corpus’ since Thomas’ first catalogue. Key references for the British sites with Mediterranean pottery and DSP have, therefore, been provided in Appendix K, but Campbell’s dataset and Doyle’s catalogue should be considered the primary points of reference for the others. Similarly, the primary ceramic reports were reviewed, wherever possible, as part of the data-collection of ceramic types and quantities. However, it was not considered prudent to amend Campbell’s sherd and vessel counts from assemblages which had not been examined in person. It was not the initial intention of this project to consider E ware, but it became clear that important new evidence for the production and distribution of this pottery could not be overlooked – and that the supply of this ware could not be easily separated from the ‘Mediterranean’ phase of importation. Campbell’s dataset was again used as the principal source of evidence, but in this case the majority of the primary sources were not reviewed. The E ware data was again supplemented for Ireland by Doyle’s catalogue (2009). No geographical limit was imposed for data collection or integration from Britain and Ireland, although the distribution of the relevant ceramic wares in Britain was almost exclusively western.

For France, Spain and Portugal, geographical limits were imposed to limit data-collection to sites in Atlantic regions. As a broad principle, sites in coastal regions or with local access to rivers leading to the Atlantic were included. Although it is impossible to prove that such sites were supplied from ships plying Atlantic routes, the ceramic distributions presented in Appendix H suggest that in the majority of cases this would be likely. The majority of sites included are coastal, although a few locations are situated very far inland, thereby reflecting secondary redistribution systems.

Although material from Vigo and Bordeaux was viewed first hand for reference, the Continental figures are essentially reliant on multiple and varied published sources. For France, the principal sources of information were the catalogues of imported amphorae published by Dominique Pieri and by Catherine Amiel and Frédéric Berthault. As discussed in Chapter Six, there has been a long history of research on imported wares to western Spain and Portugal – particularly on LRC – and these numerous sources are specifically detailed. Again, where possible,

For France, the data-collection principally comprised finds from the (current) Regions of Brittany, Pays-de-la-Loire, Poitou-Charentes and Aquitaine (broadly equating with the Late Roman Provinces of Aquitania Secunda/Tertia and Lugdunensis Tertia). A few relevant finds from French sites situated further east along the Channel coast or on rivers reaching the Channel have been included (from the Regions of Lower Normandy, Upper Normandy, Picardy, Nord-Pas-de-Calais, Centre and île-de-France), and it seems increasingly possible that material might have ar4

Introduction Data Collection

rived in northern regions via Atlantic shipments as well as overland/riverine routes. The dataset also includes some information on amphorae found into the Region of MidiPyrénées, but finds from Toulouse itself have not been included as the city seems to have been supplied from the Mediterranean (see Sub-chapter 5.3.1).

The published data for the Atlantic region shows considerable variation in the recording of the imported ceramics, which has hampered the comparative assessment of assemblages. The methodology for data-collection was also driven by previous approaches to quantification. Although Campbell recognised the value of estimated vessel equivalents (EVEs) over sherd counts as a comparative measure, this was not considered to be useful for the characteristically small assemblages of the Atlantic region (2007a, 11-12). Quantification by EVE would also have been hindered by the general scarcity of rim-sherds. Instead, Campbell’s approach – similar to that of Thomas – was to identify sherds to specific vessels and, thereby, to estimate the minimum number of vessels (MNV) by type for each site. Campbell also used this identification of individual vessels to study distributions across individual sites, but this has not been attempted here.

For the Iberian Peninsula, ceramic data has been included from ‘Atlantic regions’ equating to the Late Roman Provinces of Lusitania, Gallaecia and north-western Tarraconensis. This, therefore, comprises all of Portugal and the modern Spanish Autonomous Communities of Galicia, Asturias, Cantabria and the Basque Country. As with France, a few additional outliers were included (from Navarre and Castile and León) which seemed likely to be connected to Atlantic supply, and where it would not make sense to discount them based on modern geographical divisions. The south-eastern Portuguese/Spanish (Lusitanian/Baetican) border represented the furthest limit of data-collection. No data was collected from the Gulf of Cádiz to the south-east of the Guadiana River. Although sites in Andalucía are discussed, data from this region is not included in the appended data-tables (Appendix K). Baelo (Baelo Claudia) was doubtlessly of major relevance to the extension of supply routes from the Straits of Gibraltar into the Atlantic, but the port has been discussed separately within western Mediterranean patterns – which have been comprehensively considered by Paul Reynolds (1995; 2010).

Quantification by EVE is also not commonly used on the Continent, and for effective comparisons the British and wider Atlantic data needed to be as closely aligned as possible. Fortunately, minimum vessel counts were available for many of the Continental sites, and particularly the larger assemblage such as Vigo and Bordeaux. A point of caution is that, in practical terms, this is a somewhat subjective measure – reliant on individual assessments of form and fabric – and can be particularly difficult to assign for assemblages with few diagnostic elements, such as Mothecombe and High Peak. In this case the measure must be considered as an ‘estimated’ rather than ‘absolute’ minimum. As a basic rule the presence of one sherd of a particular type or ware at a site was taken to represent at least one vessel, although it is acknowledged that individual sherds might have been transported between sites.

GIS Mapping The sites with Mediterranean imports and DPSA were plotted in Google Earth, based on information provided in the excavation reports, catalogues or published discussions. Where possible, the findspot/excavation was pinpointed, although in many cases the location could only be assigned to the named town or site. The coordinates (in WGS 1984) were then transferred to a Microsoft Excel spreadsheet where the ceramic data was added – and thereafter transferred to an Access database. Queries were created to compare the relative quantities of pottery at these ‘import sites’; this queried data was exported to Microsoft Excel before being plotted in ARCGIS.10. The basemap used in Appendices B and H was downloaded from Natural Earth (naturalearthdata.com). Where material was recovered from various excavations in one town or city, these have been presented separately in Appendix K, but were summed for the production of the distribution maps. The locations shown in red in Appendix H are scaled proportionally, and this is consistent across all the maps. Those shown in black are not scaled due to the limitations of the data available. The transparency of certain find-spots was increased for clarity, but this is not, otherwise, significant.

In some cases, such as Conimbriga, only sherd counts were available; these were recorded in the primary database and have been frequently noted in the text of Chapters 5 and 6. Where possible, minimum vessel counts were assigned based on the catalogued or illustrated forms, although it is recognised that this might have led to an underestimate of the totals. Where this was not possible or where only the discovery of the particular ware or type had been documented – as with many of the Iberian instances of ARS-D and LRC – the ‘presence’ alone was recorded in the database. To assign some measure of quantity – as presented in Appendices J and/or K – the label ‘P’ was used to signify the presence of a certain ware or type, ‘P+’ for a site considered to have more than ten vessels and ‘P++’ for sites thought to have more than 50 vessels. The use of such labels is admittedly arbitrary, but represents a compromise to allow some presentation – and better understanding – of relative frequencies. ‘P?’ was used to represent a site where the presence of a particular type or ware was uncertain.

The locations of the sites that only had finds of E ware were not located directly from Google Earth; instead the co-ordinates provided within Campbell’s 2011 dataset were used.

Additional points explaining the data-collection for the specific ceramic types are presented in Chapter Three. To 5

Links to Late Antiquity excavated assemblages recovered, ideally, from a ‘sequence’ of closed deposits, and one based on the general presence of wares across sites. Considerations of the latter, he comments, allow the definition of ‘general distribution trends and approximate relative quantities’, although he adds that there will be a ‘degree of loss of temporal definition’ (Reynolds 1995, 2-3). In addition, he notes that ‘published scraps from diverse sites’ might help to ‘confirm an impression of trends in regional imports’ (Reynolds 1995, 3). Although this synthesis incorporates data from excavations of sequenced deposits (Vigo, Place Camille Jullian, Bantham) it is acknowledged that most of the information relates to Reynolds’ second type, and to the presence of ‘scraps’. As such, patterns in the distribution of these wares can only be used to assess broad chronological trends in supply across the Atlantic Seaboard.

summarise: for the finewares all examples of LRC and LRD have been recorded as well as all British examples of ARS considered to represent post-Roman imports. The presence of ARS-D at Continental sites has been recorded, with the acknowledgement that these examples will include fourth century imports (see Sub-chapter 3.3.2). Data for late forms of ARS-C was collected but was not mapped or included in Appendix K. All sites with DSPA have been recorded, although it is unlikely that the French distribution presented is exhaustive. For the amphorae: all of the ‘post-Roman’ British examples have been included (i.e. those presented by Campbell from secure post-Roman contexts). All examples of East Mediterranean ‘Late Roman Amphorae’ from Ireland or other Atlantic sites have been included, with the caveat that these might potentially include late fourth/early fifth century arrivals. All instances of North African amphorae of likely fifth century date that were encountered were also recorded (see discussion in Sub-chapter 3.2.4).

Apparent patterns and ‘gaps’ in the data must be considered with caution. The relative distributions of the ceramic types presented in Appendix H are likely to reflect levels of archaeological research and the availability of published reports as much as the actual spread of these wares. It is also likely that the Insular quantities will be artificially elevated due to the long tradition of research, and to the practice of recording small, undiagnostic sherds in considerable detail. Such fragments are unlikely to be quantified as individual vessels within a large assemblage. Furthermore, as Orton and Hughes note, such maps cannot provide any understanding of the proportional ‘significance’ of the imported ware at the individual site level (2013, 239). Complications with the data from Tintagel are discussed at length in Chapter Four and it has become clear that additional research on the British assemblage is necessary. Finally, it is not expected that the data or distribution maps presented will be comprehensive or final. More information, for example, is available for the spread of East Mediterranean finewares than for other types. Most importantly, the sherds and vessels recorded must be recognised to only represent a surviving fragment of the whole ‘picture’ of trade in the Late Antique Atlantic. Nevertheless, it is hoped that future research from the Atlantic will help to extend the data available, and to develop and clarify these interpretations.

The methodology for the primary ceramic research in Britain is noted in Chapter Four. Limitations of study This project represents an attempt to synthesise and evaluate recent Insular findings and, in particular, new evidence emerging from the Continent. Naturally, such an undertaking requires a consistent approach to data collection and analysis, although it became apparent that the variable availability and quality of the data would impose unavoidable limitations. Most significantly, it became clear that it would not be legitimate to statistically analyse the data, or even to provide relative percentages of ceramic types across the entire dataset or across particular regions. It is only possible to offer broad interpretations based on the relative distributions presented. Where the data permits, more precise comparisons have been attempted between particular assemblages or groups. In his study of patterns of ceramic supply to the West Mediterranean Paul Reynolds described two types of dataset: one based on comparisons of contemporary,

6

2 History of research 2.1 Introduction

Radford mentioned the emergence of evidence for ‘fine red ware’ pottery production in North Africa after the Vandal conquest of Carthage in AD 439, which allowed a fifth or sixth century date for the arrival of the Tintagel material. He was also emphatic that these contacts would not have continued beyond the Arab conquest of North Africa in the seventh and early eighth century. As such, he presented this new ceramic evidence for an ‘Early Christian’ phase of connection as directly aligning with Henri Pirenne’s 1937 thesis – commenting that it was the Arab conquests, and not the Vandal occupation of North Africa, that marked ‘the real break-up of the old classical world based on the unity of the Mediterranean’ (Radford 1956, 69).

This chapter will examine key developments in the long history of research on Late Antique imported pottery found in western Britain and Ireland, and will evaluate how this evidence been understood and presented. It will describe the major significance the imports have had within early medieval British and Irish archaeology – especially as indicators of local and inter-regional trade and diplomacy – and will summarise general understandings of the function and importance of these wares in early medieval Britain. The chapter will centre on a critical appraisal of previous approaches to this study, focusing on the search for Atlantic parallels to the Insular finds, and the use of Continental data in the formation of models to explain the transport of these wares. It will be shown that the limited availability and/or incorporation of information from France, Spain and Portugal has strongly affected interpretations of this pottery – specifically influencing models for contact and exchange in the Atlantic, and between Britain and the Mediterranean. Finally, this chapter will consider the potential for a new approach, situating the Insular finds within a broader, Atlantic perspective.

Charles Thomas After recognising the presence of equivalent imported wares at Gwithian, Cornwall [127] (Thomas 1954), Charles Thomas took over the work begun by Radford, carrying out a full appraisal of the types of pottery and,

2.2 Research on Late Antique imports to Britain and Ireland 2.2.1 Principal researchers C.A. Ralegh Radford Understandings of ceramic imports to post-Roman sites in Britain emerged primarily through Radford’s excavations at Tintagel [110] in the 1930s, where he recognised that much of the pottery recovered – comprising sherds of amphorae and finewares – was of likely Mediterranean origin. Radford’s work at the site (discussed in Sub-chapter 4.2.2) was curtailed by World War Two. His first extended publication on the Tintagel pottery outlined an alphabetical classification for the various wares (see Chapter Three), proposed a fifth/sixth century date-bracket for their arrival and catalogued comparable examples from other British and Irish sites (Radford 1956). Radford saw these varied wares as evidence of trade and contact between the ‘Celtic World’ and the Mediterranean (1956, 68). Although he noted the presence of this pottery at secular sites, as Tintagel had been interpreted as an ‘extensive Celtic monastery’ this supply was also seen to have an ecclesiastical dimension – associated with long-distance pilgrimage, processes of conversion in the West, and the development of the Church in western Britain and Ireland (Radford 1956, 59, 68).

Figure 2.1. Thomas’ distribution map of Mediterranean pottery showing suggested voyages supplying these wares. From Thomas 1990, Figure 3

7

Links to Late Antiquity largely separate to, the phase of Mediterranean contact (1990, 1-11). This final article examined textual sources for contact between Britain, Ireland and Gaul, specifically the phrase ‘Gallici nautae de Galliarum provinciis’ in the Life of Columba, which was taken as evidence of traders from Gaul arriving at Iona [7] (Adomnán, Life of Columba, 1.28).

for the first time, mapping their comparative distributions across Britain and Ireland. His seminal 1959 article provided detailed descriptions of the ‘A-E’ classes and included a first extended discussion of the chronology and origin of the imports – and the possible logistics of their arrival. Confirming the ‘A’ and ‘B’ wares to be wheelturned finewares and amphorae of Mediterranean origin, Thomas revealed their distribution to be concentrated in the south-west of Britain (particularly the region comprising the post-Roman kingdom of Dumnonia), with find-spots extending into southern Wales and Ireland (Thomas 1959, 90-100). This was taken to reflect influence and connection between these regions, not least the ‘sub-Roman’ migration of peoples from Ireland (Thomas 1959, 100).

Ewan Campbell Ewan Campbell’s commenced his research on the imported pottery in the 1980s, leading to the production of his (unpublished) doctoral thesis in 1991. This expanded on the pre-existing interpretations and models of exchange– including the notion of two overlapping, but essentially separate systems – with Thomas’ 1981 catalogue providing the foundation for his dataset of import sites (Campbell 1991, 22, 209-10). However, Campbell’s work was strongly focused on the later Continental wares, particularly in developing understandings of the chronology and characteristics of E ware. His thesis incorporated a detailed spatial and taphonomic study of the Mediterranean and Continental material from the hill-fort of Dinas Powys in the Vale of Glamorgan [96], leading to a reinterpretation of Leslie Alcock’s stratigraphy (Alcock 1963; Campbell 1991, 85-108). This analysis also led Campbell to recognise the wider importance of fragmentary glass recovered at many of the ceramic import sites – as revealing the contemporary importation and use of glass vessels (see Sub-chapter 3.5.2).

The largest Insular group, Class ‘E’, was added by Thomas and recognised to have an alternative, Continental provenance (1959, 96). Although this gritty coarseware was present at some of the same sites, it had a wider distribution than the Mediterranean wares. Thomas recorded its presence at coastal locations across western Britain and Ireland (though not Tintagel) as well as at sites in Scotland – both in the Kingdom of Dál Riata (including Dunadd [10]) as well as within the ‘northern British kingdoms’ (Mote of Mark [19]) (1959, 100). Initially assigned a Rhenish origin, subsequent research by Thomas established E ware as the primary indicator of a later, sixth/ seventh century phase of importation from western France (1959, 98; 1976, 247; 1990).

Over the next twenty years Campbell conducted much of the analysis on imported pottery and glass that continued to appear from excavations at sites in Britain (listed in Campbell 2007a, 2). In 2007 he published his major monograph on the Insular material, which included a full survey of the British and Irish import sites. Most significantly, Campbell presented a revised full database catalogue of the British and Irish finds. This was based on the extensive examination and reassessment of pottery and glass from new and previously published sites– incorporating revisions to Thomas’ identifications where possible (2007a, 2, 11).

The characteristics and research history of E ware are discussed in Sub-chapters 3.4 and 5.4. Thomas also added two more coarseware categories, classes ‘F’ and ‘G’, which were later recognised to largely comprise RomanoBritish wares and were consequently withdrawn (Thomas 1959, 89, Campbell 2006, 4). Over the next three decades Thomas developed his interpretations of the imported pottery, based around the concept of two separate phases of importation, and drawing principally on observations of the Tintagel assemblage, which remained the largest Insular group of Mediterranean wares (1976; 1981; 1988a; 1988b; 1990; 1993). Three publications should be specifically noted. In 1981 he produced a revised full catalogue of the published and unpublished material from Britain and Ireland – described as a ‘Provisional List’. An article published in 1988 offered a ‘new model’ for the transmission of the Mediterranean wares, centred on the identification of specific voyages/ routes supplying luxury commodities to secular powerbases in Britain in return for tin (1988a; see Figure 2.1, an updated version from Thomas 1990). Thereafter, the vessels – or even fragmentary sherds – would have been distributed locally to secondary sites as ‘largesse’ within a ‘network of clientship’ (1988a, 16, 21). Finally, Thomas’ 1990 publication formalised his interpretations of E ware, presenting the pottery as indicative of Merovingian Frankish contact and as attesting a sustained and ‘vigorous trade of Gaulish wine in barrels’ that succeeded, but was

Although Campbell’s catalogue considerably extended the corpus of Insular finds and import sites, it re-affirmed Thomas’ observations on the distribution of the pottery – as essentially restricted to western Britain and Ireland. Although E ware was identified at Pictish sites in eastern Scotland, neither it, the Gaulish ‘DSP’ nor the Mediterranean wares of later-fifth/sixth century date were present at sites in eastern Britain. As such, the system that supplied these ceramics was understood to be different from the cross-channel exchange systems which developed, particularly, in the later-sixth and seventh century (Huggett 1988; Harris 2003; Campbell 2007a, 3). Two distinct systems of long-distance importation to early medieval Britain were recognised: an Atlantic route supplying pottery and associated commodities (principally wine and oil) to western, ‘British’ regions and Ireland, and a 8

History of research 299). Campbell saw this exchange as ‘sustained rather than haphazard’, and not to be fully commercial in motivation (2007a, xiv). He concluded that the contact was driven by demand for metal resources in the East Mediterranean, and the need to access sources of tin, lead and silver in southwestern Britain and, possibly, copper in Ireland (Campbell 2007a, 138; Campbell and Bowles 2009, 305-9). From the major secular power-bases in south-western Britain, the imported goods would have been redistributed to smaller sites (Campbell 2007a, 138). Although products such as wine, oil or tablewares represented everyday commodities in the East Mediterranean, the imports were seen to have attained a new value and political function in their British context:

riverine/overland route supplying luxury items (metalwork, gemstones, amber, ivory, cowrie shells) via the Rhine to ‘Saxon’ communities in the South and East (Harris 2003, 161-88). This model led Campbell to describe an ‘Atlantic Curtain’ for the western ceramic imports, which were not imported to, or exchanged with, Saxon sites – although he noted that this ‘barrier’ was not impermeable to the exchange of glass or metalwork (Campbell 2007a, 140). Similarly, there have been suggestions that some of the Byzantine silver and copper alloy vessels found in Saxon burials might have arrived from western Britain, and ultimately via Atlantic channels (Harris 2003, 179-80). Campbell drew broad conclusions for the early medieval import-system operating via the Atlantic, based on his extended dataset and revised distribution maps (see Figure 2.2), and founded on the model of two identifiable phases of Mediterranean and Continental material (1996; 2007a, 138-9; Campbell and Bowles 2009). The ‘Mediterranean’ phase was seen to represent a wholly new dynamic, distinct from Romano-British import systems that collapsed in the earlier-fifth century (Campbell 2007a, 138). Amphorae and fineware of East Mediterranean and North African origin were imported from the late-fifth to the mid-sixth century, principally to royal or high-status secular sites in the south-west of Britain. As will be discussed, this exchange was not considered to have been ‘mediated’ through sites in the West Mediterranean – except for a ‘late phase’ originating at Carthage – but represented ‘direct contact with Byzantium’ (Campbell and Bowles 2009,

‘They were used by royalty to re-enforce a hierarchical society through gift exchange and control of access to foreign luxuries’ (Campbell and Bowles 2009, 301). The later-sixth and seventh century ‘Continental’ phase of importation was again seen to represent a new trading-dynamic, which emerged after the collapse of the Mediterranean routes in the mid-sixth century. This phase was characterised by a different, broader ceramic distribution, which was not concentrated on the south-west of Britain – although again focused on high-status or royal sites (Campbell 2007a, 138). The imported glass and E ware were seen to represent the surviving vestiges of a diverse but structured exchange system, by which commodities like wine and salt, and luxuries like dyes and spices, were transported from western Gaul to western

Figure 2.2. Campbell’s distribution maps of all Mediterranean (left) and Continental imports (right). From Campbell 2007a, Figure 83

9

Links to Late Antiquity with Ian Doyle (2009) and Amanda Kelly (2010) producing recent syntheses of evidence. Information on the relevant sites has also been collated online (Excavations.ie 2014). Doyle’s revised catalogue of the imports has been used in subsequent chapters of this work as the primary source of Irish data. Like Campbell he observed that the majority of the pottery was found at secular sites, and concluded that the trade was likely to have been controlled by the ‘secular elite’ (Doyle 2009, 32).

Britain and Ireland in exchange for ‘surplus produce’ (Campbell 2007a, 124, 136-8). Again, the impetus for this trade was seen to be external – driven by traders from the Continent with, potentially, some ‘involvement of Aquitanian elites’ (Campbell 2007a, 13, 137). Jonathan Wooding Although Campbell and Thomas had incorporated textual evidence into their interpretations of the imported pottery, Jonathan Wooding’s research on contacts on the Western Seaboard was more directly linked to written sources. His doctoral thesis was founded on a comprehensive review of ancient sources that mentioned long-distance sea-travel, and on their subsequent historiographic treatment (Wooding 1996a, iv). He concluded that the written evidence for continuous and substantial long-distance commerce – and associated cultural ‘diffusion’– along the western sea-lanes between c. 400 and 800 had been overemphasised (Wooding 1996a, iii). He dismissed the pottery as evidence for sustained trade, therefore rejecting notions that a ‘constant backdrop of commerce’ in the Atlantic provided the framework for the ‘cultural development’ of western Britain and Ireland – including the spread of ‘Latin literacy, Christianity and monastic ideas’ (Wooding 1996a, 1). Specifically, Wooding contested Heinrich Zimmer’s model of an extensive fifth-century wine trade operating between south-west Gaul and Ireland, which facilitated the spread of Christian learning (Zimmer 1909; Wooding 1996a, 1-3, 32-3; 1996b; Campbell 2007a, 79). Instead, Wooding favoured a ‘minimalist’ interpretation of long-distance connections in the Atlantic, arguing that the archaeological data revealed ‘smaller scale’ and ‘episodic’ contact and communication (Wooding 1996a, iii).

The growing list of Mediterranean and Continental imports contained in his catalogue suggested that the pottery, and associated commodities, were more widely dispersed in Ireland than had been previously realised. The E ware, in particular, was seen to have a ‘marked Irish distribution’ and, as in Britain, was the most numerous ceramic import (Doyle 2009, 23). However, Doyle concluded that the imports did not provide evidence of ‘large-scale trade’, but were significant for their role in Irish society: ‘…as a means of creating and maintaining relationships though gift-giving and in affirming the social position and identity of those who engaged in this international trade’ (2009, 36).

Wooding mentioned that he had conducted an ‘extensive re-examination’ of the British examples of imported pottery and glass in 1988-9, but he did not include a complete corpus of the material – this was left to Campbell (1996a, iv, 42-3). As the present author’s work is specifically focused on the ceramic evidence, it will not offer a reappraisal of the written sources considered by Wooding. Nevertheless, it will be shown that Wooding’s aim to deconstruct early text-based models led to flawed conclusions regarding the archaeological data. Furthermore, the new Atlantic information introduced in Chapters 5 and 6 will serve to challenge Wooding’s view of the Mediterranean imports and E ware as showing ‘little sign of being part of sustained, long-term importation patterns’ (1996a, 104).

Kelly presented a detailed review of LRA1 amphorae and LRC fineware found on Irish sites, concluding that the former reflected ‘demand for eastern wine by the secular elite’, and that the latter – together with imported glassware – was connected with ‘ostentatious dining displays’ used to ‘consolidate strict social hierarchical frameworks’ (2010, 71-3). Significantly, both Kelly and Doyle situated the Mediterranean imports within a wider model of direct contact between Britain and the Byzantine world, although Kelly observed that trade was likely to have been ‘facilitated by high status secular centres along the Atlantic seaboard’ (2010, 73). Doyle suggested that the small quantities of Mediterranean material found in Ireland indicated that it was a ‘secondary participant’ in this exchange system, but did not discount the possibility of shipments arriving directly from the Mediterranean (2009, 23, 34). Kelly described a connected model of Mediterranean ships travelling via the Atlantic to Britain and Ireland – with much of their cargo being offloaded on the Portuguese coast – and meeting local vessels that transported the wares upriver (particularly along the River Boyne). These imports were assigned a particular value in refuting the impression of Ireland as a ‘wholly insular and marginal contender in the Early Middle Ages’ (Kelly 2010, 72-3).

Research in Ireland

2.2.2 State of knowledge

Both Radford and Thomas’ catalogues incorporated information on parallel finds emerging from early research in Ireland, principally following Ó’Ríordáin’s excavations at the ring-fort of Garranes, Cork [80] (Ó’Ríordáin 1942) which produced a relatively large assemblage of Mediterranean wares – including one vessel of LRC (Ó’Ríordáin 1947). Research in Ireland has continued,

Since Radford’s early research, an increasing number of sites with comparable Mediterranean and Continental imported pottery have been identified in Britain and Ireland and catalogued, principally, by Thomas and Campbell (Thomas 1959; 1981; Campbell 1988; 1991; 2007a). Campbell’s recent monograph listed material from close to 150 sites, which was supplemented by his 2011 dataset 10

History of research control (Barrowman et al. 2007, 335) or, more specifically, a royal site (Turner 2004, 26). Since then, the level of ecclesiastical involvement in both ‘phases’ of importation has been less emphasised (Thomas 1988a; Campbell 2007a, 132, 137; contra Thomas 1976). Thomas, nevertheless, indicated that pilgrims and religious literature might have been carried as a ‘side-line’ to commercially driven trading-routes from the East Mediterranean (1988a, 22). Campbell’s detailed analysis of site-types rejected a ‘direct relationship’ between both groups of imports and ‘early religious settlements’, but noted that Mediterranean and Continental pottery did arrive at religious sites – shown, particularly, by the finds from the monastic site of Whithorn [20], in Dumfries and Galloway (Hill 1997; Campbell 2007a, 117, 122). Whether this reflects direct church involvement in these exchange systems, or the redirecting of imported commodities by secular rulers, remains to be fully established – although the latter seems increasingly likely. It has also been suggested that wine carried in the Mediterranean amphorae was used for communion or, similarly, that sacramental wine was transported from western France (Radford 1956, 68-9; Campbell 2007a, 132)

(2007a, xiv). The characteristics of the British and Irish import sites and the function of the material within early medieval Insular society have been comprehensively discussed by Campbell (2007a; Campbell and Bowles 2009). As such, it is only necessary to summarise general understandings arising from research in Britain. Import sites The strong correlation between the pottery and sites interpreted as secular power-bases has suggested a link between imported commodities and the development of hierarchical power-structures in post-Roman Britain (Campbell and Bowles 2009, 301). The Mediterranean imports in the south-west of Britain have been associated with contemporary changes in society, including ‘increasing social stratification’ and the ‘development of kingship’ (Turner 2004, 30). Local or regional political allegiances may have been fostered and maintained through the control and redistribution of commodities such as wine – and by the structures that amassed mineral resources for exchange. This strong connection between the imports and powerdynamics might even have led to the ultimate demise of these sites. Campbell has suggested that the end of trade with the Mediterranean in the mid-sixth century disrupted ‘social ties’ in the South West, leading to the ‘subsequent collapse of central authority’ (Campbell 1996, 89). For Turner, this helped explain the abandonment of ‘major enclosed settlements in mainland Cornwall’ from the latersixth century (2004, 30). Certainly, E ware is relatively less common in the South West – and absent at Tintagel – indicating that this region was less integrated in the later, ‘Continental phase’ of exchange (Campbell 2007a, 120).

Imported pottery has also been recovered from cemetery sites including Cannington, Somerset [100] (Rahtz et al. 2000), Saint Materiana’s at Tintagel (Nowakowski and Thomas 1990; 1992), as well as the monastic cemetery at Llandough, Glamorgan [95] (Holbrook and Thomas 2005). The presence of imported material at unfortified coastal locations including Mothecombe [124] and Bantham [125] in South Devon established another category of import site, the ‘beachmarket’ or sand-dune site (Turner 2004; Campbell 2007a, 121). These were interpreted as primary import points and trading-stations, characterized by seasonal dwellings and temporary hearths, rather than permanent occupation. However, recent excavations at two of these sites, Bantham and Mothecombe in South Devon (discussed fully in Chapter Four) have suggested the presence of less transient communities.

In Britain, much of the imported Mediterranean pottery has been recovered from enclosed, fortified ‘hill-top’ sites that were occupied, or re-occupied as elite settlements, in the fifth and sixth centuries (Campbell 2007a, 117-24; discussion in Gerrard 2013, 168-77). Significant assemblages of Mediterranean pottery have been published from South Cadbury [106] and Cadbury Congresbury [97] in Somerset (Alcock 1995; Rahtz et al. 1992), High Peak [111] in Devon (Pollard 1966; Rainbird et al. 2013) as well as Tintagel and Dinas Powys. However, sherds of fineware and amphorae have also been recovered from a ‘middle tier’ of settlement sites, including enclosed ‘rounds’ such as Trethurgy in Cornwall [120] (Quinnell 2004, 241). Similarly, Campbell demonstrated a correlation between fortified, high-status or royal sites in Ireland and northern Britain/Scotland and E ware – with the largest assemblage of 25 vessels from Dunadd (Lane and Campbell 2000) – but also noted the distribution of this ware to smaller enclosed sites, particularly in Ireland (Campbell 2007a, 116). Campbell also observed that many of the import sites have produced evidence of craft and metal-working (2007a, 132).

Whether sites represent points of primary importation is often unclear. The majority of the import sites are at coastal locations or close to navigable rivers/estuaries and may have been reached directly by traders from the Atlantic, but it is also likely that many imported sherds represent points of secondary redistribution. The large assemblage from Tintagel certainly suggests it was an initial point of arrival. Although the rocks and tidal differences would have made access into the harbour difficult for a large ship, Campbell concluded that ‘Byzantine vessels’ carrying imported wares would have been able to berth in the Haven (2007a, 120). This might have been facilitated by the use of a ‘local pilot’ (Campbell 2007a, 120; Barrowman et al. 2007, 329). The reinterpretation of Bantham as a ‘port’ establishes the site as a likely point of exchange with the Atlantic, and of redistribution to South Devon – and possibly beyond (Reed et al. 2011). Tintagel, therefore, can no

The original interpretation of Tintagel as a monastic site was eventually discounted (Thomas 1988a, 10-13), and it is now also considered to be centre of high-status, political 11

Links to Late Antiquity tact over a 75-year period both the merchants sailing to Britain and those collating resources for exchange would need to be confident in the other party’s involvement (Campbell 2007a, 130-1). Campbell also indicated that the quantities of LRC imported to Britain were equivalent to those in southern Spain, suggesting that levels of East Mediterranean trade with these regions were ‘comparable’ (2007a, 131; see below). This distribution pattern will be re-evaluated in the subsequent chapters.

longer be viewed as the primary entry-point for material reaching Britain and Ireland (see Barrowman et al. 2007, 329). Elsewhere, Dalkey Island [63], on the south side of Dublin Bay, has been interpreted as a possible coastal trading-station or entrepôt (Thomas 1988a, 22; see discussion in Doyle 1998, 101; 2009, 28). There have also been suggestions the Isles of Scilly might have been used as a temporary stopping-point for traders sailing to Britain, or as a point of exchange (Thomas 1988a, 15; Campbell 2007a, 119). The presence of Mediterranean pottery and E ware at sites in the Scilly Isles and on Dalkey Island would reveal a connection with both ‘phases’ of importation. Campbell concluded, nevertheless, that there were no ‘actual trading settlements’ associated with the later, Continental phase of importation and that none of the island sites with imported wares could be accurately seen as emporia (Campbell 2007a, 119-20, 124).

Tintagel remains, by some margin, the site with the largest quantity of material, with an estimated 150 amphorae and 80 fineware vessels recovered to date (Thorpe 2007, 246; see discussion in Sub-chapter 4.2.3). The recently published report from excavations at Bantham in South Devon has revealed an assemblage with a significant, if smaller, quantity of vessels (Reed et al. 2011). It should be noted, however, that the amount of pottery at Insular sites remains relatively small in comparison to quantities recorded from many Mediterranean assemblages. At Carthage, for example, over 1500 kg of amphora sherds were recovered (Peacock 1984b, 116).

Finally, it should be noted that many of the sherds of imported pottery found in Britain and Ireland are stray, single finds, which have provided little information about the context of their arrival or use.

Campbell criticised a general ‘reluctance’ to see the imported wares as evidence of ‘sustained trade’, adding that western Britain had typically been seen as less ‘economically developed’ (2007a, 138). He commented that Richard Hodges’ analyses of the early medieval economy did not ascribe much importance to the Atlantic routes (Campbell 2007a, 10). Certainly, the phase of Mediterranean importation was only briefly summarised in Hodges’ 1982 monograph on Dark Age Economics, although E ware was considered in more depth as a marker of trade between Aquitaine and the ‘Irish Sea provinces’ (Hodges 1982, 33-9). However, elsewhere Campbell conceded that the Mediterranean imports to Britain were not ‘quantitatively significant’ within the early medieval economy, but important for their use in developing ‘power relationships’ (Campbell and Bowles 2009, 313).

Chronology The established date range for the first ‘phase’ of Mediterranean imports, c.AD 475 to 550, is based on the forms of East Mediterranean and North African fineware found in western Britain – as outlined in Michael Fulford’s influential article on the imported wares (1989, 4). This date-bracket has been re-affirmed since by Campbell (2007a, 19; Campbell and Bowles 2009, 298-9), although there have been suggestions that Mediterranean pottery might have arrived at Tintagel from c. 450 (Thorpe 2007, 245), while a similar date has been suggested for the earliest imported pottery at Bantham (Bidwell et al. 2011, 109). Campbell’s analysis of E ware and glass led him to date the phase of Continental importation from the latersixth century until the late-seventh or early-eighth century (Campbell 2007a, 138-9).

Exchanged commodities Questions of the chronology of the imported pottery, and therefore for the Atlantic trade with Britain, are considered in Chapter Three, with reference to new understandings emerging from research in the Mediterranean and on the Continent.

Olive-oil and wine have typically been offered as the likely contents of the imported amphorae, although other exotic commodities cannot be discounted. This issue is considered further in Sub-chapter 3.2.6 following ongoing research in the Mediterranean. The East Mediterranean finewares have been considered to represent a ‘secondary cargo’ transported as a ‘space filler’ with the amphorae (Campbell 2007a, 16). As such, their presence at British sites can be used as an indirect proxy for commodities that have not survived archaeologically. The Insular distribution of E ware has frequently been seen as an indicator of a trade in wine from western France, although Campbell concluded that this pottery was imported as a container, particularly for dyes (2007a, 51, 80; see Sub-chapter 3.4).

Scale of trade The scale of trade suggested by the imported wares has been a more contentious issue. Thomas initially suggested that a few isolated shipments – or even a single cargo – might account for the material (Thomas 1959, 105; 1988a, 12; 1993, 85). Fulford noted a ‘general recognition’ that the volume of material at post-Roman sites in Britain was ‘altogether not great’, but dismissed the ‘minimalist view’ of the pottery as representing a few shipments (1989, 2, 4). More recently, Campbell has presented a convincing argument for a regular system of sustained trade based on the requirements of exchange: to maintain this con-

Since Radford’s initial publication on the Mediterranean imports, they have been associated with tin, and it is likely that metal was the primary export from Britain (Radford 12

History of research understood. This issue will naturally be affected by new understandings of the routes by which the material was carried, and will be reconsidered in Chapter Seven.

1956, 59; Thomas 1959, 106; Campbell 2007a, 129, 138). Turner commented that it was ‘virtually certain’ that tin production took place in Cornwall in the Roman period, and it is likely that tin-streaming would have continued in the South West into the fifth century and beyond (Penhallurick 1986; Turner 2004, 27-28). The discovery of tin ingots at Praa Sands, Cornwall (Biek 1994) and in Bigbury Bay, Devon (Fox 1995) potentially attest to early medieval production, while ingots have been found at sites with imported pottery, including Trethurgy (Quinnell 2004, 73). As mentioned, Campbell saw increased demand for metal in the East Mediterranean as the driving factor for trade with Britain, pointing to signs of a contemporary ‘scarcity of supply’ in the East (2007a, 129-30). He presented a distribution map of the South West showing a clear association between sites of metal extraction and production – principally of tin, but also lead (and silver) from the Mendips and south Wales – and East Mediterranean imported pottery (Figure 2.3).

The Mediterranean imports to western Britain have frequently been connected to a story in the Life of John the Almsgiver (see Radford 1956, 68; Penhallurick 1986, 237-45; Alcock 1987, 89; Mango 2001; Wilkes 2007). This records a ‘foreign’ sea-captain appealing to John – Patriarch of Alexandria AD 610-619 – to fund a commercial voyage from Alexandria to Britain. John provides a vessel from a church-owned fleet – a dorkon (gazelle) – laden with ‘20,000 modii’ of grain. The ship is blown off course but eventually lands in Britain, where a famine is ‘raging’ and the cargo is profitably exchanged for tin and coinage. On the vessel’s return to Alexandria, the tin is found to have been transformed to silver (Leontius 10). Certainly, the early-seventh century date of the voyage cannot be directly connected with the Mediterranean pottery attested from Britain, and the veracity of this account has been questioned – Knight labelling it as ‘clearly fiction’ (Knight 2007, 157). In addition, none of the pottery from western Britain can be directly connected to the region of Alexandria, but shows a focus on the north-eastern Mediterranean (Antioch region), the Aegean and the Carthage region. Campbell, however, proposed that the story contained some truth, if corrupted, and potentially reflected ‘handed-down sailors’ yarns’ of earlier voyages connected with tin (2007a, 131). Although he noted

It is also possible that other commodities, such as huntingdogs, furs and leather, and potentially slaves, were carried in the returning shipments (Radford 1956, 68; Campbell 2007a, 129). Shipping and traders: textual and epigraphic evidence The identity and origin of the traders conducting this exchange have been considered, but remain to be fully

Figure 2.3. ‘Association of Mediterranean imports with tin and lead production’ in south-west Britain. From Campbell 2007a, Figure 82

13

Links to Late Antiquity the considerable textual evidence for Byzantine merchants in France, Italy and Spain, and concluded that the ‘Syrian’ designation might be credible – reflecting Syrian control over much of this trade and the importance of Antioch as a hub of international exchange (2003, 60-64). Following Dark’s assessment, Harris presented a map of towns in France attested to have ‘Byzantine’ traders, together with locations with imported East Mediterranean pottery (Figure 2.4). These Eastern communities were noted to be positioned at ‘important political and ecclesiastical centres’, principally on the southern coast or on rivers leading from the Mediterranean (Harris 2003, 61). Nevertheless, Bordeaux [203], Poitiers [178] and Tours [169] in the Atlantic region were also included. Notably for Bordeaux, Gregory of Tours recorded the theft of relics from the home of a Syrian merchant named Eufronius, within the context of political upheaval in later-sixth century Aquitaine (The History of the Franks VII. 31). Harris suggested the story betrayed underlying resentment of the wealth or influence of an Eastern community in the town (2003, 61).

suggestions that the destination of the voyage was not Britain, but Brittany or Galicia, he concluded it was ‘certainly Britain’ (Campbell 2007a, 131). Similarly, Wilkes concluded that the account should be viewed ‘an embellishment of a simpler story’, but also argued that it showed that a cargo of tin reaching Alexandria from Britain was ‘credible’ during this period, which strongly hinted at ‘maritime connections of some form’ (Wilkes 2007, 13). Mango used this account to conjecture that ships reaching Britain would have been of relatively large size/capacity. Other details included in the Life were used to show the very mixed nature of cargoes transported from Alexandria – which included dried goods, clothing, possibly ‘silver plate’ and other ‘high value’ goods (Leontius 28). Harris suggested that the ‘dried goods’ listed by Mango might include pepper and other ‘exotic items’ from the Far East (2003, 56), although it should be noted that the phrase ‘Xerophotra himatia’ was translated differently as ‘waterproof garments’ by Dawes (1948). Regardless, the very mixed cargoes suggest that the ceramic finds in western Britain and the Atlantic might represent only the visible remnants of a broader trade in luxuries.

The ceramic data available to Harris was very limited for the Atlantic, comprising a few individual vessels of East Mediterranean amphorae recorded by Bonifay and Villedieu (see Sub-chapter 5.2.1), but the distribution nevertheless showed a correspondence at Bordeaux and Tours. In contrast with France, there was no clear documentary evidence for Eastern traders in Atlantic regions of Iberia (to the north or west of Mérida).

The presence of East Mediterranean traders in the Late Antique West – the region of the former Western Roman Empire – was examined more generally by Ken Dark, based on Louis Bréhier’s early twentieth century review of ancient references (particularly from Sidonius Appolinarus and Gregory of Tours) (Dark 2003, 76-7). The majority of these merchants were referred to by contemporary Western authors as ‘Syrians’, or less commonly, ‘Greeks’ (Dark 2003, 76). However, Dark added that this might have represented a convention, and that the while these traders could be seen as ‘Byzantine’ they did not necessarily originate in Syria. Dark also considered the involvement of Jewish communities in trade in the West, but noted that this was less clear (2003, 76). The textual sources established the activity of Byzantine merchants at ‘Rome, Ravenna, Naples, Mérida, Orléans, Bordeaux, Marseille, Nice, Nîmes, Arles, Autun, Narbonne, Lyon, Vienne, Toulouse, Poitiers, Trier and Paris’ (Dark 2003, 77). These were noted to have all been Roman towns and to be ‘urban centres in the fifth and sixth centuries’ (Dark 2003, 77). No textual sources attest the presence of eastern traders in Britain and, by contrast, there are no confirmed examples of late-fifth or sixth-century East Mediterranean pottery from Romano-British urban sites. Dark noted, however, that from the quantity of eastern pottery it was possible that Tintagel was ‘the British equivalent of the textually-attested sites known from Gaul, Spain and Italy’ (2003, 77). As will be seen, the presence of eastern traders in Britain is not necessary to account for the Insular distribution of the imported ceramics.

Mark Handley’s recent study of epigraphic evidence for travellers in the Late Antique West considered these textual references within a broader examination of ‘Easterners’ recorded across the western Mediterranean and into France and Britain (2011). He reiterated the warning that not all references to ‘Syrians’ or those from the eastern Mediterranean could be directly linked with trade, contrary to the assumptions made by earlier scholars, including Pirenne. Instead, the inscriptions revealed a much more varied list of occupations (Handley 2001, 82-5). However, despite the mass of new information presented, Handley’s distribution maps revealed very little epigraphic evidence for Easterners, traders or otherwise, in Portugal or in Atlantic regions of France and Spain. The exceptions comprise a few (fourth to sixth-century) references to a Syrian at Tavira on the Algarve and to individuals from Asia Minor and ‘the East’ further inland at Mérida and Mértola (Handley 2011, 91-7, 121, Maps 7-9). Some broad conclusions can, nevertheless, be drawn from Handley’s research – principally his observation of a general ‘long slow’ decline in all travel to, and within, the West after the early-fifth century. Inscriptions recording Easterners in the West largely ceased by the early-seventh century, although Handley added that this reflected a decline in ‘Christian epigraphy’ rather than, necessarily, a drop in travel (2011, 103-5).

In her monograph, Byzantium, Britain and the West, Anthea Harris considered the British imports within a wider pattern of Byzantine contact with the West, outlining the transport of a diverse set of exotic commodities and manufactured goods (2003, 41-3). She also examined

Handley’s discussion of the possible associations between the epigraphic evidence for travel and contemporary ceramic distributions are also relevant. He concluded that 14

History of research

Figure 2.4. ‘Distribution of imported pottery and Byzantine ‘traders’’. From Harris 2003, Figure 12, after Bonifay and Villedieu 1989

the two datasets were not easily connected, but revealed two pertinent associations: firstly, that the ‘striking’ number of epigraphic references to individuals from Asia Minor in Spain reflected the high Spanish incidence of East Mediterranean fineware (LRC) – potentially indicating an ‘Asia Minor-Iberia axis of communication’, and secondly, that the early-fifth century peak of epigraphic attestations to Easterners in the West could be equated with a dramatic increase in East Mediterranean pottery in the western Mediterranean (Handley 2011, 110-1).

Questions also remain as to the identity of those involved with the later ‘phase’ of Continental importation to Britain, although Campbell determined that this trade was directed from ‘western Gaul’ and that the ‘traders were Frankish merchants, or merchants based in that area’ (2007a, 136). He added, however, that these might have included traders of Mediterranean, and possibly ‘Syrian’, origin. A model was presented whereby after the breakdown of trade with the Mediterranean in the mid-sixth century, merchants familiar with these routes and markets – having previously acted as ‘intermediaries, pilots or traders’ – took control of the exchange with western Britain and Ireland (Campbell 2007a, 136).

The research conducted by these various authors reveal that although textual and epigraphic evidence demonstrate the presence of ‘Easterners’, Eastern communities and Eastern traders in the West, particularly in the early-fifth century – which can be connected to a general expansion of trade from the East – it cannot be easily used to explain the presence of East Mediterranean pottery in Britain, Ireland or, potentially, elsewhere in the Atlantic.

The scholarly significance of the imported pottery The imported pottery has had particular importance for early medieval archaeology in western Britain, as this period is characterised by a low level of visible material culture (Turner 2004, 26). The Mediterranean wares have, there15

Links to Late Antiquity fore, been valuable in allowing the identification and dating of fifth and sixth century sites in Britain and Ireland. Early in his research Thomas noted how the research on these wares might enable a ‘firm chronology of the archaeology of the so-called ‘Celtic West’, but later criticised the overreliance on these sherds as chronological markers, commenting that they had been ‘too often…picked out by archaeologists as chronological life-rafts in an ocean of uncertainties’ (1976, 245). E ware has also had crucial significance in providing evidence for later-sixth and seventh century activity at sites in western Britain, Scotland and Ireland.

to western Britain and Ireland, interpretations of transportation of these wares have remained fairly static. The pottery has typically been thought to have arrived directly from the East Mediterranean, and therefore to reveal some degree of direct contact with the Byzantine world. As will be shown, this interpretation is founded on two specific observations: the ‘exceptional’ composition of the British assemblage, and the relative scarcity of parallel material from the Atlantic Seaboard.

Beyond this prosaic significance, the imported wares have had a particular importance in understanding the degree to which western regions of Britain were connected to wider political, economic and social transformations in Late Antiquity, not least the emergence of the new kingdoms/ successor-states following the end of Roman control in the West. Dark’s intention to equate Tintagel with Continental urban centres of international trade can certainly be seen in the context of his wider argument – that fifth and sixth century Britain was ‘at the mainstream’ of a European Late Antique culture (2000, 15, 156; 2003, 77). Despite the relatively small numbers of vessels involved, the Mediterranean imports have been used to establish lines of political or cultural contact between western Britain, the East Mediterranean and North Africa from the later-fifth century. Radford’s initial account described the pottery as providing ‘tangible evidence’ of the international contact and trade hinted at in fragmentary historical sources for Britain during this period (Radford 1956, 68). Tintagel, in particular, has accrued an inter-regional importance as a result of these external connections, with Thorpe commenting that:

Radford’s 1956 summary of the Tintagel pottery laid the foundations for these models of transmission. The distribution of the imported wares was said to belong to:

2.3.1 Ralegh Radford at Tintagel

‘…an age when trade was once more flowing along the Atlantic Seaways and avoiding the shores of the narrow sea which had already been blocked by the barbarian conquests’ (Radford 1956, 67). The Mediterranean pottery was therefore linked with the supply of British communities in western regions sometime after the mid-fifth century, and with the (re)opening of Atlantic shipping routes to circumvent ‘blocked’ crosschannel routes. Unlike subsequent authors Radford was fairly optimistic about the prospect of equivalent finds from the Atlantic, and discussed ‘foreign parallels to the fine red ware’ that he had found at Tintagel (Radford 1956, 65-7). He discussed the presence of cross-stamped sherds and sherds with Christian symbols – similar to his ‘Ai’ sub-category – at sites across the Mediterranean, including Constantinople, Athens, Carthage and Marseille. Although he mentioned that these examples were ‘comparatively rare’ in the West, they included sherds found at sites on the Atlantic seaboard – specifically from Bordeaux and Nantes [171]. The finds from the latter location were said to be ‘interesting as indicating the route by which this pottery reached Britain’, by which Radford presumably meant via the Loire (1956, 66). Radford also mentioned the possibility of small quantities of comparable stamped finewares in Spain, and possibly the Algarve, following early Iberian research by Hans Zeiss (Radford 1956, 66). Unfortunately, these early parallels would later be rejected – establishing a divergent trajectory for research on the British finds.

‘The fact that Tintagel was at the end of this complex trade route suggests that the occupants would have wielded far-reaching influence during the fifth to seventh centuries’ (Thorpe 2007, 247). These notions will be examined in the following subchapter. This will focus on the development of models for the transmission of this pottery, and the resulting interpretations of the motivation for exchange in the Late Antique Atlantic. In particular, this discussion will consider how the pottery has been thought to demonstrate the direct shipment of commodities from the East Mediterranean to western Britain. This will be shown to directly result from impressions of the available ceramic data – and specifically the lack/limitations of data from the Atlantic. This examination will focus on understandings of the Mediterranean wares, as the Continental wares have usually been seen to reflect a separate, later trade-dynamic (Thomas 1990, 11). The research history of E ware is discussed in Sub-chapter 5.4.

2.3.2 Charles Thomas and Bernard Wailes Charles Thomas’ early publications reiterated that the imports were transported via Atlantic channels. As much of the fineware had been recognised to have a source in the eastern Mediterranean, the shipments supplying this material to western Britain were taken to originate in the East (Thomas 1959, 104). Thomas’ 1959 article – said to reflect twenty-five years of research on this topic – described the pottery having been ‘brought directly by sea from the Byzantine world through the Straits of Gibraltar’ (1959, 89, 105). He also surmised that the lack of Mediterranean

2.3 Models of transmission: the character of the Insular assemblage and the search for Atlantic parallels In spite of the long period of data collection and extensive publications on the Mediterranean ceramic imports 16

History of research pottery in northern Britain might have resulted from the Mediterranean traders not being prepared to sail further than the Irish Sea (Thomas 1959, 101).

‘…sea-borne all the way via the Straits of Gibraltar, and not taken overland from Marseille to Bordeaux’ (Thomas 1959, 91).

This model of direct shipment from the Byzantine East via Atlantic routes resulted from observations that the Insular assemblage of imported wares was relatively large and essentially isolated. As the research on the imported pottery in Britain developed – principally as a result of cataloguing work by Thomas – a few parallel finds were recognised from sites on the Continent. These were, however, always seen to be a minority in comparison to the British and Irish examples. Most significantly, Thomas’ 1959 article incorporated unpublished research that was being carried out at the time by Bernard Wailes to identify comparable Continental examples of the ‘A’, ‘B’ and ‘E’ wares (ARS/LRC, amphorae and E ware) that had been identified in Britain (Thomas 1959, 90; Wailes 1963). Thomas described how the ‘intermediately-placed’ French and Spanish examples of fineware proposed by Radford (equating with the Insular sub-category ‘Ai’) had been dismissed by Wailes – which consequently made the British finds ‘even more outlying than had been thought’ (Thomas 1959, 90). Specifically, Wailes’ thesis had concluded that Radford’s Continental examples of cross/Christian-stamped wares were actually ‘Visigothic wares’ (later classified as ‘D ware’ or DSP), rather than true parallels for the East Mediterranean finewares found at Tintagel (Wailes 1963 92-100). A concordance of these terms is presented in Appendix A for clarity.

This was a key turning point for British research on the imported pottery: the first possible parallels from the Atlantic Seaboard had been dismissed, and – as a direct consequence – a model of direct shipment from the East Mediterranean via an Atlantic route was proposed. Wailes reasserted this key observation in his thesis: ‘The first piece of evidence to emerge, travelling south, was of a negative character: no parallels to the British Class A and B wares were noted in any of the western French museums.’ (Wailes 1963, 36). ‘Since neither of these wares [ARS and LRC] are to be found in western or south-western France (i.e. on the Atlantic Seaboard) but are confined to the Mediterranean littoral at least in this area, it would seem highly probable that they must have reached the British Isles by a direct sea voyage. This can only have been through the Straits of Gibraltar and northwards along the Atlantic coasts of Europe to the English Channel, the British Channel, and the southern Irish Sea.’ (Wailes 1963, 129). He illustrated this point with the first map to show a likely route for the transmission of the British imports (Figure 2.5). Although the catalogued examples of Mediterranean imports in Britain and Ireland rose significantly over subsequent decades, this model of direct transport from the East remained virtually unchallenged. As Wailes’ research was never published, his conclusions are largely known from the brief summary contained in Thomas’ 1959 article. However, it is clear that his observations – particularly the absence of parallel imported wares in Atlantic France, Spain and Portugal – had a significant impact on Thomas’ interpretations, and on the ensuing research on the British and Irish imports.

Thomas’ 1959 article also reproduced Wailes’ own summary of the various types of late stamped pottery that he had examined in museums in France, Spain, Belgium and the Netherlands in 1958-9 (Thomas 1959; Wailes 1963, 26). These included red finewares (described as ‘terre sigillée claire’ after Lamboglia’s classifications) and ‘Visigothic’ stamped wares of grey and red varieties (equivalent to DSP of reduced and oxidised fabrics). Wailes concluded that the red, fine, ‘A ware’ from Britain did not resemble either of these, but that the British subcategory ‘Aii’ could be equated with a separate group of stamped wares – which he had found to be confined to southern France, Catalonia and the Balearics (Thomas 1959, 91). Thomas used this observation by Wailes to link the British ‘A’ wares to Waagé’s Late Roman B and C fineware categories (established by research at Antioch and equivalent to ARS and LRC) and, simultaneously, to propose that they arrived in Britain directly from the eastern Mediterranean. Wailes would later discuss and refine this concordance in his thesis, although this was never published (Wailes 1963, 36, 129).

Developing Radford’s early speculations, Charles Thomas also proposed an eastern Mediterranean origin for the imported amphorae found in early medieval contexts in Britain. The comparable examples discussed in his 1959 article were largely limited to finds from sites in the eastern and central Mediterranean. Thomas specifically noted that Wailes had failed to identify any ‘Bi’ amphorae (LRA2) in France, but had recognised some examples in Barcelona (Thomas 1959, 92). Comparable finds to Radford’s ‘Class D’ (DSP) were mentioned to have been found in northern and southern ‘Gaulish contexts’, although Thomas – following advice from Leslie Alcock – suggested that this pottery probably originated in Britain, in the region of the Bristol Channel, based on the quantity of vessels recovered from Dinas Powys (Thomas 1959, 95). It was only after the Rigoirs’ work on the distribution of DSP that Thomas retracted this proposal, and conceded that a French origin would be more likely (Thomas 1976, 246). Given this misattribution and Wailes’ lack of success in

Wailes’ assessment of these analogous wares and their restricted distribution in France allowed Thomas to conclude that there was no fineware comparable to the ‘A ware’ from Tintagel at Bordeaux or ‘through the Narbonne gap’. As a consequence, he could infer that the British imports were: 17

Links to Late Antiquity

Figure 2.5. The distribution of fineware comparable to British ‘A ware’ and proposed routes of transmission. Redrawn from Wailes 1963, Figure 170

finding parallels for the pottery in France, Thomas’ first catalogue of the find-spots of the imported wares – contained in the appendix to this 1959 article – did not include Continental examples of the Mediterranean fineware or amphorae, or the DSP.

found outside of Britain and Ireland; Thomas specifically noted that the aim of this catalogue was not to explore the implications of the growing ‘Insular evidence’ (Thomas 1981, 3). This separate treatment certainly propagated the notion of the imports as an essentially Insular phenomenon. In this publication Thomas suggested that Radford’s fineware classes ‘Ai’ and ‘Aii’ should be finally replaced by the Mediterranean categories of Phocaean Red Slip ware (LRC) and African Red Slip ware (ARS), following the publication of Hayes’ typology in LRP and Supplement

Later research by Charles Thomas As in 1959, Thomas’ 1981 catalogue did not include any parallel examples of Late Antique Mediterranean pottery 18

History of research with Charles Thomas, although this did not transpire (Hayes 1972, 7; similarly alluded to in Peacock and Thomas 1967, 39). Following the extended presentation of his fineware typologies, Hayes briefly summarised the distribution of these wares and, pointedly, described the British examples of imported ‘Late Roman’ Red Slip pottery as:

(Thomas 1981, 3; Hayes 1972; 80). Nevertheless, despite this incorporation of Mediterranean research, Thomas did not mention the few examples of fineware from Atlantic sites that were included in Hayes’ corpus (see below). After Wailes’ dismissal of the evidence from France, relatively little information on Continental parallels was incorporated into British publications, and the Insular assemblage continued to be viewed as largely isolated in the Atlantic. As such, Thomas’ subsequent publications repeated the model of direct link between the East Mediterranean and south-western Britain, although a ‘stop-off’ in the Carthage region was added to account for the lesser quantities of ARS and the supposedly North African ‘Bv’ amphorae (1976, 245; 1988a). His 1988 article considered a model of sixth century shipment offered by Paul Arthur – characterised by mixed cargoes carried in smaller ships that ‘tramped the Mediterranean coast, perhaps buying and selling from port to port’ (1988a, 13). Thomas concluded, however, that this model was not applicable for the Atlantic, and that the British finds could only be explained by a ‘planned and intentional’ trip driven by commercial motives.

‘…perhaps the most remarkable phenomenon in the history of exportation of the wares discussed here’ (Hayes 1972, 422). Hayes commented that this exceptional distribution was restricted to the ‘Celtic areas of the west’ – thereby indicating that the pottery arrived directly by sea from the Mediterranean and reflected ‘direct trade-links with the Byzantine world’ (1972, 422). He added that an Atlantic route for their arrival was corroborated by ‘the slight scatter of examples of Late Roman C ware in Spain and Morocco’. Hayes’ maps of the main distribution of ARS Forms 103-4 and LRC Form 3 (Figure 2.6) certainly imply a model of transport to south-western Britain that bypassed Atlantic sites – if not, necessarily, an Eastern port-of-origin. Discussing wider patterns in the distribution of ‘Late Roman’ fineware, Hayes commented that by the mid-fifth

Thomas’ later publications show a gradual change in interpretation. By 1990 he indicated that although the shipping-routes were long-distance, there might have been stopping-points along the Atlantic Seaboard: ‘Classical sources imply that such hazardous journeys were confined to the full summer, and the probability that some part of the voyage involved coast-hopping and occasional beachings raises the problem (valid, wholly relevant, and insufficiently explored) of any intermediacy involving Spain and Portugal’ (Thomas 1990, 11). By the publication of his guidebook to Tintagel Thomas’ opinion had shifted even further; he explicitly suggested that the Mediterranean imports might not have arrived directly, and that it was possible that goods had: ‘…firstly been landed and sold at some European Atlantic entrepôt in Portugal, Spain or Western France (Bordeaux, the ancient Burdigala is such a place.’ (Thomas 1993, 96). It is unclear whether this truly suggests a model of trans-shipment, or merely the existence of Atlantic ‘portsof-call’, but, regardless, the lack of comparable pottery from the Western Seaboard left this idea of intermediate shipment as an unexplored possibility. Most significantly, south-western Britain was never disputed as the ultimate destination of these voyages from the East. 2.3.3 John Hayes’ observations In the introduction to LRP Hayes noted the identification of Mediterranean imports at early medieval sites in Britain and Ireland and mentioned a forthcoming collaborative study

Figure 2.6. The ‘approximate main distribution areas’ of ARS Forms 103-4 (top) and LRC (bottom) as shown in LRP. From Hayes 1972, Maps 29 and 33

19

Links to Late Antiquity

Figure 2.7. The distribution of LRC Form 3 shown in LRP. From Hayes 1972, Map 15

century imports to Spain and Portugal were largely restricted to ARS at sites along the Mediterranean coast, and to some ‘scraps’ of LRC, which provided a ‘tenuous link’ to the British examples (Hayes 1972, 415). These ‘scraps’ comprised LRC from isolated sites in eastern Spain and the Balearics, but, significantly, included Bracara Augusta (Braga) in north-western Portugal [275] (Hayes 1972, 415; and shown in an accompanying distribution map – Figure 2.7). Hayes’ discussion of the French distribution, in contrast, only included finds from southern, Mediterranean France. Although he noted the appearance of later ‘t.s.grise’ products (DSP) in western France, specifically at Bordeaux, Hayes stated that African products did not, as a rule, accompany them (1972, 415-6).

finewares on the Atlantic coast, particularly a telling presence of LRC. However, despite an occasional mention, this never came to be fully incorporated into the British literature. It was at this point that the research on the imports in Britain and research on comparable finds from the Continent started to deviate. Although subsequent British publications referred to developments in Mediterranean research on the pottery (chronology, typology and so forth), limited attention was directed toward the Atlantic Seaboard. Occasional references were made to isolated finds or sites on the Atlantic coast, or to the potential for future discoveries in this region, but these observations were not followed by substantial attempts to scour Continental publications for parallel finds.

Rather more information was included in the appendix to LRP, which listed examples of ARS and LRC vessels – but not the respective forms – from a small number of sites beyond the Straits of Gibraltar. These included Baelo in southern Spain, sites such as Balsa and Marim on the Portuguese Algarve, and, perhaps most significantly, Conimbriga [288] and Tróia [316] in western Portugal. It is not clear how many of these finds represented fifth century or later forms – at least among the ARS – and it seems that a number were subsequently discounted. Nevertheless, this list demonstrates a developing awareness of imported

Hayes’ 1980 Supplement contained additional information on the pottery emerging from Conimbriga in Portugal, where it was described as adding a ‘spectacular new dimension’ to the understanding of the fifth century (Hayes 1980, 482). Specifically, the apparent destruction of the site by the Suevic invasions of 465-8 was considered to provide new dating evidence for the typological sequence of ARS. The presence of forms including ARS 99 and 104 in destruction deposits at Conimbriga was taken to indicate their circulation by this date – which consequently suggested a revised, early date for the equivalent British 20

History of research there is no other way that the integrity of the assemblage which reached Britain could have been maintained unless the ships concerned put together their cargoes in the general area of origin of the pottery that they carried’ (Fulford 1989, 4).

examples. Hayes furthermore suggested that the ‘major importation’ of ARS to Portugal stopped at around the same point (1980, 521). These conclusions were, however, later recognised to be incorrect, and these ARS forms were re-dated to the late-fifth and sixth century following the British excavations at the Avenue Bourguiba site in Carthage (Fulford 1984a, 113-4; Reynolds 1995, 152-3). This reaffirmed that the Insular examples belonged to a new phase of exchange operating in the late-fifth and earlier-sixth century, which was therefore disconnected from Romano-British exchange systems.

The inspiration for contact and exchange with Britain was therefore firmly positioned in the north-east Mediterranean. Trade with Britain – based on the acquisition of tin – was seen to be a ‘deliberate objective’ of some East Mediterranean ships, and connected to a wider expansion of East Mediterranean trade into the West that peaked between c. 475 and 550 (Fulford 1989, 4). Britain, at the ‘periphery’ of this expansion, was therefore brought into contact with Byzantine East. This new connection was taken to account for links ‘hinted at’ in written sources, including allusions to Britain in Procopius’ texts (Fulford 1989, 5). These included a reference to Justinian making payments to barbarians ‘as far as Britain’ (The Secret History, 19). Although Fulford recommended caution in using the passage to reveal ‘political relations between Britain and Byzantium’, he argued that it reflected a genuine awareness of sixth-century Britain (1989, 5).

2.3.4 Michael Fulford’s model The mechanisms for the importation of the Mediterranean pottery to Britain and Ireland were formalised in Fulford’s 1989 article, which reconsidered the Insular finds in the light of research conducted in the western Mediterranean. A model of direct shipment from the East Mediterranean was again proposed to account for most of the pottery, which consequently suggested direct contact between parts of Britain and the Byzantine world between the latefifth and mid-sixth century (Fulford 1989, 4-5). Fulford’s model describes ships sailing from western Asia Minor, the Aegean or Constantinople to targeted sites in Britain and Ireland, and potentially collecting some North African commodities en route (Fulford 1989, 1, 4).

The sites referenced by Fulford – Tintagel, CadburyCongresbury and Dinas Powys – are certainly characterised by a higher proportion of LRC to ARS (Dinas Powys excluded) and of East Mediterranean amphorae to ‘Bv’ (see Appendix K). East Mediterranean wares also dominated the recently published assemblage from Bantham. As such, the unusual prevalence of East Mediterranean pottery in Britain and Ireland cannot be questioned, and clearly reflects some divergence from contemporary trade-patterns in the West Mediterranean (see Chapter 7.2.2).

Rather than being based on the relative isolation of the Insular finds, this model was founded on the exceptional character of the Insular assemblage – the assemblages from the British sites showing a very high proportion of East Mediterranean vessels (LRC and amphorae) compared to North African vessels (ARS and ‘Bv’ amphorae) (Fulford 1989, 2). This was observed to be the reverse of the typical pattern in contemporary assemblages in the West Mediterranean (Fulford 1989, 3). The dominance of East Mediterranean wares was taken to reflect the routes of transport and used to argue against a model of redistribution from a West Mediterranean or Atlantic port:

Nevertheless, the following chapters will show that the East Mediterranean origin of shipments reaching Britain – and the ensuing direct link between Britain and Byzantium – must be reconsidered as a result of new Atlantic data.

‘If the ships which traded with Britain had all originated from ports within the W. Mediterranean or the Atlantic coast of Iberia, we would reasonably expect to find that the majority of the pottery would be of African (Tunisian) origin, confirming with the character of the assemblages in the home ports’ (Fulford 1989, 3).

It must be noted, in relation the other ‘factor’ underpinning the model direct transport, that Fulford did raise the possibility of future, related discoveries from the Atlantic Seaboard. The finds from Conimbriga were mentioned – as the only ‘notable incidence’ of LRC found along the Atlantic to compare with Britain (Fulford 1989, 3). Beyond this exceptional assemblage, the general absence of Atlantic parallels was stated to be ‘puzzling’, but Fulford suggested that it might prove to be a ‘temporary aberration’. An isolated example of a sherd of LRA1 from Brittany was described as a possible indicator of future discoveries (Fulford 1989, 3). This example, from l’Île Lavret [155], off Bréhat – an island off the north coast of Brittany (see Sub-Chapter 5.2.1) – had also been noted by Thomas, who suggested it might reflect the return voyage of a ship that had visited south Devon (1988a, 22).

Similarly, the relatively small quantities of DSP in the Insular assemblages argued against a model of transport focused on western France (Fulford 1989, 3). Fulford noted that these observations did not ‘preclude’ the possibility of some shipments leaving from the western Mediterranean or some ‘traffic’ between western Britain and the Bay of Biscay, but concluded that the analysis of the pottery presented convincing evidence for direct contact between south-western Britain and the Byzantine world: ‘We must look to a predominantly E. Mediterranean origin for the ships which reached the British Isles… 21

Links to Late Antiquity 2.3.5 Ewan Campbell’s research

3C was recorded from Bordeaux. A number of sites in north-eastern and south-eastern Spain were also reported to have examples of LRC, including Baelo and Carteia at the far south of Spain, and, significantly, three sites on the Portuguese coast: Braga (one vessel), Conimbriga (98 sherds) and Tróia (quantity unknown) (Campbell 1991, Illustration 6). These Portuguese sites are examined further in Chapter Six.

Over the course of his publications Campbell developed this model of direct shipment for the Mediterranean wares, which expanded on perceptions of the Insular assemblage as exceptional and isolated in the Atlantic. His doctoral research led him to conclude that the pottery was unlikely to have arrived as a result of ‘sporadic visits by pilgrims and other travellers’ (Campbell 1991, 185). Instead, he proposed a model for the direct shipment of East Mediterranean wares, driven by the ‘centralised market economy’ of the Byzantine Empire and explained by the availability of mineral resources in Britain (Campbell 1991, 186, 206). The coherent and consistent nature of the Aegean ‘package’ of wares typically attested at British sites was taken to show that the trade was direct, and not ‘a product of a chain of cabotage transactions or tramping’ (Campbell 1991, 185). The Insular assemblage was observed to not contain an ‘admixture’ of amphorae and fineware collected at intermediate ports – revealing that Britain was the ‘intended destination’ of these voyages. This conclusion was reinforced by another observation – that only small quantities of LRC (recorded as ‘Phocean Red Slip’) had been recovered from ‘intermediate sites’ in Spain, North Africa and Portugal (Campbell 1991, 185).

A second distribution map and accompanying table listed Continental sites with East Mediterranean amphorae of fifth-sixth century date (Campbell 1991, Illustration 7). Again, these largely comprised sites in Mediterranean regions of France and Spain (principally in Catalonia and the Balearics), but included the aforementioned LRA1 from l’Île Lavret, one possible LRA1 from Conimbriga and one vessel of ‘Bvi’ (LRA4) from St Seurin in Bordeaux [197] (after Keay 1984; see Sub-chapters 6.2.4 and 5.2.1 respectively). Campbell commented that the parallel finds from Portugal indicated an ‘extension of East Mediterranean trade into the Atlantic’, but this did not affect his overall conclusions (1991, 206). Another map suggested possible trade routes from the East Mediterranean and indicated possible ports-of-call in Portugal and France, but this was not discussed further (Campbell 1991, Illustration 8).

Although these few Atlantic parallel finds were not discussed further in the text, they were recorded in an appended list of Continental sites with LRC, and shown on an accompanying map (Campbell 1991, Illustration 6; see Figure 2.8). This data was principally based on LRP, as well as publications by Nieto Prieto (1984) and Mayet and Picon (1986), as discussed in Sub-chapter 6.2.1. As well as three sites in Mediterranean France (including Marseille) which were listed to have LRC, a vessel of LRC

Campbell’s 2007 monograph presented a refinement of the model of direct – though not ‘non-stop’ – transportation, using the pattern of fineware forms (LRC and ARS) as evidence for two non-exclusive phases of importation within a single trading system (Campbell 2007a, 128, 138). These comprised an earlier phase of Aegean imports arriving c. 475-550, followed by a phase of North African importation c. 525-550 (Campbell 2007a, 26, 138). The shift to Carthage as the point of origin for the later shipments was suggested to be related to Justinian’s re-conquest of the city in 533, but although Campbell concluded that there might have been a level of Byzantine state-backing to these long-distance trade routes, he convincingly argued that the British finds could not be explained solely by diplomatic contacts with the West during Justinian’s reign (2007a, 131-2). In this synthesis, and in a subsequent collaborative article, Campbell specifically considered the question of the routes of transmission and the possibility of intermediate locations in the Atlantic (2007a, 126-8; Campbell and Bowles 2009). Following Fulford’s initial comparison of the proportions of East Mediterranean pottery, Campbell outlined five specific characteristics of the British and Irish ceramic assemblage, which distinguished it from contemporary patterns in the West Mediterranean and which necessitated an alternative model of supply. These factors – based principally on Paul Reynolds’ observations of patterns in the West Mediterranean – comprised: • a lack of Gazan or Palestinian amphorae – in contrast to a high proportion of north-east Mediterranean/ Aegean LRA1, 2 and 3; • a disproportionately large quantity of LRA2;

Figure 2.8. Campbell’s map of LRC distribution in the western Mediterranean. From Campbell 2007a, Figure 6 (reproducing Campbell 1991, Illustration 6)

22

History of research of Mediterranean imported pottery from Continental sites. However, these did not strongly affect the proposed models of transportation. His 1996 article again used the western distribution of LRC to demonstrate the exceptional character of the British finds. The fineware was shown to be concentrated in north-eastern and south-eastern Spain and south-western Britain, with a ‘scatter’ of finds on the Atlantic coast of Iberia (Campbell 1996, 86). The quantity of LRC in south-western Britain was stated to be roughly equivalent to that in Mediterranean Spain, rather than it diminishing with distance – which was used to demonstrate that the exchange was directed at Britain and implied deliberate trade rather than ‘casual contact’ (Campbell 1996, 86). It should be noted, nevertheless, that the same article mentioned evidence for tin production in Brittany, and proposed that additional imports from this region might be ‘expected’ (Campbell 1996, 88).

• a low proportion of ARS to LRC (the former typically dominating in the West) and the overall large quantity of LRC in Britain; • an absence of LRD (Late Roman D/ ‘Cypriot Red Slip ware’); • the restricted date-range of the imports – limited to c. 475-550, whereas East Mediterranean amphorae appeared in the West earlier in the fifth century and continued to c. 600 (Reynolds 1995, 135; Campbell 2007a, 127; Campbell and Bowles 2009, 298-9). These particular characteristics demonstrated that there was: ‘something peculiar about the Atlantic trading-system, and that explanations for its existence must differ from those used to account for the western Mediterranean distribution, which can be explained by normal commercial activities involving the transport of staple commodities such as wine, oil and grain’ (Campbell and Bowles 2009, 299).

Campbell defined a specific geographic scope to his 2007 monograph as ‘Britain and Ireland west of the areas of ‘Anglo-Saxon influence’, termed, by preference, the ‘Atlantic West’ (Campbell 2007a, 3). The term ‘Insular’, stated to be ‘in opposition to Continental’ was used to denote sites or finds in Britain and Ireland. However, Campbell frequently refers to ‘Atlantic imports’, ‘Atlantic sites’ and ‘Atlantic trade’ when describing Insular finds and patterns, which might give the (unintentionally) misleading impression that these observations apply to the entire Atlantic Seaboard. The book itself does not contain much information on parallel finds from Spain, Portugal or France, except for a brief discussion on the distribution of LRC and its significance for the British examples (Campbell 2007a, 16). Campbell noted that 26 sites in the West Mediterranean (France, Spain, Portugal and North Africa) had produced finds of LRC, but that most of these only represented a few vessels. Of the 26, only three reached double figures: Benalúa and Baelo in southern Spain and Conimbriga in Portugal (Campbell 2007a, 16). The location of these, on ‘southern coasts’, was considered, together with an apparent gap in the distribution of LRC in the central-east of Spain, to indicate that LRC arrived in Britain directly from the East Mediterranean via southern Spain and Portugal. The figures provided by Campbell were taken from Reynolds’ ceramic study (see Sub-chapter 6.2), the aforementioned article on the western distribution of LRC by Nieto Prieto (1984) and the volume on the ‘sigillata’ from the excavations at Conimbriga (Delgado et al. 1975). Subsequent publications by each of these authors have highlighted additional evidence for Atlantic imports, and will be discussed in Chapter Six.

The assertion that LRA2, rather than LRA1, was particularly common in Britain has certainly marked the British assemblage as distinct from patterns in the West Mediterranean, where this Aegean amphora is generally rare and where LRA1, followed by LRA4, are the most popular types (Reynolds 2010, 106–7). This observation is principally based on the Tintagel amphorae, and is reconsidered in Sub-chapter 4.2.3. Campbell’s five distinguishing ‘factors’ provide a specific framework to assess the British and Irish pottery against wider patterns in the West, including patterns emerging from the Atlantic. As such, they will be addressed in Chapter Seven. These observations had a direct impact on the model of transmission offered for the Mediterranean imports, and re-affirmed that the Insular finds were not mediated through a western Mediterranean port. Their arrival by the Atlantic sea-lanes was not disputed – with Campbell dismissing the rare alternative route offered by Bowman (see below), although the possibility of Atlantic trans-shipment was examined. Campbell noted Barry Cunliffe’s observation that these Byzantine vessels were more likely to have offloaded their cargoes at ports in Portugal, from where local vessels ‘more used to rigours of Atlantic sailing’ would have transported the commodities across the Bay of Biscay to Britain (Cunliffe 2001, 479). However, he considered this unlikely, questioning why these offloaded cargoes would have been trans-shipped to Britain, rather than being distributed in Portugal (Campbell 2007a, 128). The limited amount of comparable pottery in the Atlantic, therefore, revealed that Britain was the ultimate goal of eastern shipments, and that their cargoes were not mediated through an Atlantic port.

Campbell’s earlier map was reproduced to demonstrate the differential distribution of LRC across western Europe – the size of the dots for each find-spot being proportional to the number of vessels from each site (see Figure 2.8). This again presented the finds from western Britain as largely isolated, but also relatively large in quantity – comparable to the western Mediterranean and significantly greater than the quantity on the Atlantic Seaboard. As a

Atlantic parallels in Campbell’s publications Despite Campbell’s research focusing on the Insular evidence, his publications have noted some isolated examples 23

Links to Late Antiquity Ada far exceeded the number of vessels attested at British sites, but concluded that this did not necessarily support a ‘minimalist view’ of the shipments reaching Britain (1996, 100). Nevertheless, she raised a number of factors to refute the model of ‘regular, direct trade’ between Britain and the East Mediterranean. These included the length of time such journeys would take – approximately half of a six-month, restricted, summer season to complete the return-trip – and a general Late Antique trend towards smaller ships and more frequent, but shorter, journeys (Bowman 1996, 101). As an alternative, she proposed that all of the British imports might have arrived through contacts with France – trans-shipped from the Mediterranean coast, and subsequently transported to Britain (Bowman 1996, 102). As mentioned, Campbell discounted this possibility, citing Edward James’ conclusion that the ‘Narbonne-Carcassonne Gap’ was closed by this date (Campbell 2007a, 127).

direct consequence, this reaffirmed the model whereby deliberate contact with sites in south-west Britain was the fundamental force driving Atlantic networks in the fifth to seventh century. Any finds outside of the focus of concentration in south-western Britain could be accounted for by models of redistribution, while sporadic finds on the Atlantic seaboard demonstrated the routes taken, and, perhaps, occasional stopping-points. Campbell noted that ARS was a more common import in the West Mediterranean than in Britain, but his book did not discuss any examples from sites on the Atlantic Seaboard. Likewise, there was no discussion of parallel finds of amphorae outside of Britain, apart from comparisons with the composition of Mediterranean assemblages. Although the text referred to the high percentage (‘28%’) of LRA2 in ‘Atlantic contexts’, compared to low levels at sites in the West Mediterranean, this refers to the pattern of the Insular assemblage (Campbell 2007a, 22, citing an observation from Reynolds 1995, Appendix B.5.). Nevertheless, the appendix to the book and Campbell’s published database did record a few isolated examples of Mediterranean pottery from sites in western France: the single LRA1 sherd from l’Île Lavret and one vessel of LRA2 from the site of Le Yaudet [157] (both in Brittany), one ‘Bv’ amphora and two LRA4 from the site of Saint-Laurent-des-Combes [195] (Gironde, not far from Bordeaux), and three LRA4 from Bordeaux (Campbell 2011). The latter examples represent two further vessels from Saint-Seurin, in addition to the one recorded in Campbell’s thesis.

Anthea Harris Harris similarly proposed that East Mediterranean pottery was imported to southern France and subsequently redistributed inland, noting the presence of amphorae at Tours, Lyon and Bordeaux (Harris 2003, 51). This route was not, however, used to account for the Insular finds. Instead, she used Fulford and Campbell’s analyses to advocate the model of direct shipment via the Atlantic: ‘…a long-distance sea-route around the Iberian Peninsula, with a few coastal stops for provisioning, remains more likely than sequential transhipment’ (Harris 2003, 150).

2.3.6 Additional considerations In general, discussions on the British imports have continued to overlook emerging evidence from the Atlantic seaboard. On the few occasions reference is made to Atlantic parallels, it is generally only the pottery from Conimbriga that is mentioned, demonstrating the continued reliance on Fulford’s 1989 article and Campbell’s research. Nevertheless, there have been some attempts to consider western Britain and Ireland within a broader Atlantic perspective, or to propose alternative methods of transmission for the Insular finds.

Although Harris did not disagree with Britain being the intended destination of East Mediterranean shipments, she was one of the few authors to discuss intermediate stopping-points and the presence of comparable pottery on the Atlantic Seaboard. Harris noted the presence of LRC – said to be more than a ‘mere trace’ – at four sites on the Portuguese coast, which indicated an ‘Atlantic littoral network of distribution’ for this ware (2003, 52). Using Reynolds’ data she described these Portuguese assemblages as representing the highest quantity of LRC in the West apart from south-west Britain, excluding the areas that actually came under Byzantine control (Harris 2003, 52). However, although Reynolds had included four sites from Portugal in his list of western sites with LRC, he only provided specific quantities for one – Conimbriga – and presumably Harris’ observation was based principally on this assemblage (Reynolds 1995, 162, appendix B.2). Harris also reproduced Reynolds’ 1995 map of LRC distribution, which showed the ware as present on the Atlantic Seaboard only at Conimbriga and in south-west Britain (see Figure 6.2).

Leslie Alcock and Anne Bowman Both Alcock and Bowman considered the Insular imports in relation to the wreck site of Yassi Ada, discovered off the coast of Bodrum in Turkey (Bass and van Doorninck 1982). For Alcock, the variation of the LRA1 amphorae (within a cargo of 850-900 amphorae of types LRA1 and 2) carried by the seventh century ship was not reflected at the British import sites. This suggested that comparable cargoes were not reaching Britain (Alcock 1987, 91-2). He added that it was unlikely that these amphorae were ever shipped directly from North Africa or the East Mediterranean, and instead, ‘probable’ that they were trans-shipped between the Mediterranean and the Atlantic (Alcock 1987, 92).

Despite the inclusion of some information from Portugal, the British focus of distribution was not disputed – Harris concluding that the LRC highlighted the ‘unique case

Bowman noted that the quantity of amphorae at Yassi 24

History of research of his subsequent discussions. Wooding recognised that the later forms of imported pottery at the site could be directly equated with the forms found at early medieval sites in Britain, and that this Atlantic distribution was connected to a wider expansion in the ‘traffic’ of East Mediterranean goods (Wooding 1996a, 41-5). The pottery was seen to reflect Byzantine interest in northern Europe, and was described as an ‘intrusive outlier of a new, but short-lived, Byzantine presence in the West’ (Wooding 1996a, 41, 54).

of Britain in relation to eastern Mediterranean imports’ (Harris 2003, 52). In a later chapter she also commented on the dearth of ‘Byzantine imports’ from the west coast of France, and used this to argue that Mediterranean ships traversing the Atlantic routes would not have made frequent stops or stayed close to land (Harris 2003, 149). Citing observations made by Jonathan Wooding (see below), Harris concluded that this lack of French parallels could not be attributed to ‘misidentification’ but reflected a real absence. The presence of ARS and DSP in Britain were, nevertheless, seen to reflect ‘stop-overs’ in North Africa and ‘western Frankia’, respectively (Harris 2003, 150).

Echoing Fulford’s prediction, Wooding suggested that further evidence from western Spain and Portugal might be forthcoming, and that, ultimately, the British finds might end up as ‘isolated speculations arising from a much larger traffic with Atlantic Iberia’ (Wooding 1996a, 42). However, Wooding was much less optimistic about potential Mediterranean imports in central/western France. He described Wailes’ failure to identify parallel pottery in this region, and noted that this ‘negative evidence’ had continued beyond the point where it could be blamed on a lack of identification:

For Harris, the confirmation of long-distance shipments to Britain suggested an underlying political or diplomatic function to this exchange, and, like Fulford, she mentioned the potential evidence from Procopius (Harris 2003, 152). Extending Dark’s conclusions, the possibility of an Eastern ‘mercantile community’ at Tintagel was suggested as another marker of diplomatic links (Harris 2003, 1467). Finally, the prevalence of LRA2 amphorae in Britain provided a potential connection to state-controlled supply. Harris referred to Karagiorgou’s analysis of the distribution of this amphora in the North Balkans and Aegean, which had identified a specific link with the military supply of olive-oil (Harris 2003, 56-7, 146; Karagiorgou 2003, 146-9). However, Karagiorgou had also conceded that these vessels might also have entered the free market as surpluses, and overall it is not clear that her interpretation can be easily transplanted to the Atlantic region (Karagiorgou 2003, 155).

‘Centres such as Bordeaux, Orléans and Tours have all seen extensive exploration and related late-Roman wares are known from these sites. We would not now expect PRS [LRC] or class 43 [LR2] amphorae to exist unrecognised in these centres’ (Wooding 1996a, 43). For Wooding, this gap in the distribution implied that the imports were shipped via Atlantic routes, and not overland through Gaul (1996b, 79). He also described how pottery finds from Spain and Portugal suggested ‘landfalls’ in the region of the Tagus and Mondego Rivers in Portugal. However, despite noting the isolated example of LRA1 from l’Île Lavret, he concluded that there was no evidence that ships transporting late Mediterranean pottery stopped ‘anywhere between Iberia and Cornwall’ (Wooding 1996a, 52).

Jonathan Wooding Notwithstanding Wooding’s minimalist appraisal of the British imports as indicators of Atlantic exchange, his published thesis was optimistic on the subject of Continental parallels. However, relatively little new ceramic data was presented and this issue was not fully addressed. Despite Wooding’s stated intention to consider links between France, Spain and Portugal, he limited the scope of his discussion to the ‘northern-half of the western sealanes, with rather less reference to Spain and Portugal’, and concluded that that there was no ‘natural unity’ between these regions (Wooding 1996a, iii).

Wooding’s conclusions for the later phase of Continental importation envisioned the arrival of E ware pottery to Britain within a ‘tramp-steamer’ model of exchange, characterised by the small-scale, independent transportation of variable cargoes (Wooding 1996a; 1996b). However, he was quite clear that this model could not be applied to the earlier phase of Mediterranean exports to the Atlantic (Wooding 1996a 96; 1996b, 80). Despite his consideration of parallel Mediterranean imports in the Atlantic, the direct model of shipment was not disputed.

Like Campbell, Wooding described how only a ‘scatter’ of parallel Mediterranean imports had been identified from Atlantic Portugal and Spain in comparison to Britain, although he recognised that this reflected the relative quantity of published excavations from these areas (Wooding 1996a, 41). Wooding’s text was rather vague about what comprised this Iberian evidence; it seems that his data is largely based on finewares, and principally on the fifth-sixth century distribution of LRC. As Campbell had done in his thesis, Wooding used Hayes’ research on this fineware and the 1984 article by Nieto Prieto, and he would, therefore, have been aware of the presence of LRC at Braga, Tróia and, above all, at Conimbriga. The finewares from Conimbriga consequently formed the basis

Jeremy Knight The possibility of alternative mechanisms of transportation was considered by Knight in The End of Antiquity (2007, 156-66). Like Wooding, he described the presence of LRC at four sites on the Portuguese coast, and principally at Conimbriga. However, these were not automatically seen to reflect stopping-points on direct shipping-routes to Britain – with Knight suggesting that ‘Syrian’, Gallic, Irish or British sailors might have been operating in the 25

Links to Late Antiquity Atlantic, and that the Mediterranean imports might have been ‘filtered through intermediaries in Mediterranean or Atlantic ports’ (2007, 157). Furthermore, Knight noted the presence of LRC at Bordeaux and the probable production of DSP nearby, which suggested that much of the Insular imports had ‘passed through’ its harbour.

the apparent lack of published, comparative data from sites on the Atlantic Seaboard has left the British finds to be largely examined in isolation. As a direct consequence, this has reinforced notions of special political, diplomatic or commercial links between the newly-emerging British and Irish Kingdoms and Byzantium.

Knight also developed the association between the East Mediterranean pottery and metal resources, but incorporated sources beyond south-western Britain. The large quantity of imports at Tintagel was certainly seen to be associated with tin supplies. However, he also stated that the distribution of LRC in Murcia (south-eastern Spain) might be connected with the presence of lead – and silver – reserves, and suggested a link between finds of DSP at Nantes and early medieval tin extraction in the region (Knight 2007, 155-8).

As a contrast, it is worth considering that the presence of luxury goods from the East Mediterranean in eastern Britain has usually been taken to reflect indirect transport – particularly as resulting from chains of transmission across Merovingian France (Harris 2003, 162). Harris did not discount the possibility of Byzantine traders moving material along these riverine systems, but noted that the Byzantine objects and styles reaching eastern Britain might have arrived with their ‘cultural connotations already transformed by this mediation’ (Harris 2003, 175). Equivalent interpretations have not been suggested for the western British finds of pottery and glass. This differential treatment is doubtlessly due to the relative presence of intermediate finds along the routes of movement – with finds of East Mediterranean metalwork and other exotic finds identified along the Rhone/Rhine corridor (Harris 2003, 175).

2.3.7 The potential for new models This review has not offered an exhaustive assessment of references to the Insular imports but has concentrated on the most significant appraisals of the ceramic evidence. The imported Mediterranean wares have inspired a large volume of text, demonstrating the continued significance of this pottery for understanding economic and political dynamics in western Britain and Ireland from the later-fifth century. The imports are frequently included in monographs summarising the archaeology of the period, where they are typically accompanied by maps of the British and Irish distribution – but without showing comparable information from the Atlantic Seaboard. As such, and despite a few alternative interpretations, the impression that the Insular imports are ‘unique’ has persisted.

However, there have been increasing indications that the apparent ‘gap’ in the distribution of Mediterranean pottery along the Atlantic Seaboard has been overemphasised, and that new data from ‘intermediate’ locations require new interpretations of the Insular finds. As discussed, the Insular imports have typically been characterised by the East Mediterranean LRC. The elevated presence of this fineware in south-west Britain, far from its western Anatolian source, has been thought to indicate an exceptional, long-distance route of transmission. However, new information has emerged to show that LRC was much more widely distributed along the Atlantic Seaboard than previously recognised. Reynolds 2010 study of ceramics and trade in Spain contained new information on its western occurrence. In contrast to his map from 1995, this plot revealed LRC to be well-distributed on the Atlantic coast: present at eight sites in western Portugal and north/northwestern Spain. The data from Vigo [259] in north-west Spain (see Sub-chapter 6.3.2) was not fully available for this publication, but the site was included in Reynolds’ updated distribution map, where LRC was recorded to be ‘common’ (Reynolds 2010, 170, Map 12; see Figure 2.9).

This impression has directly resulted from two key observations of the character of the British import assemblage: firstly, the relative isolation of the Insular material – arising from Wailes’ research on the Continent, but never countered by subsequent appraisals; and secondly, the exceptional character of the Insular assemblage – specifically the unusual prevalence of East Mediterranean material, as presented by Fulford, and the broader list of atypical attributes summarised by Campbell. These observations have supported assumptions that deliberate contact with sites in south-west Britain was the fundamental force driving Atlantic exchange systems between the fifth and sixth century. Any finds outside of the focus of concentration in south-western Britain have been accounted for by models of redistribution, while sporadic finds on the Atlantic seaboard are seen to demonstrate the routes taken, and, perhaps, to indicate intermediate stopping-points. As the research progressed, a few publications mentioned occasional parallel finds at Continental sites – principally the LRC fineware from Conimbriga – but these had little impact on the broad conclusions, and were typically seen as outliers to the main British distribution. Dark, for example, suggested that the few finds of LRC in western France and Spain might be explained as a ‘by-product of directional trade aimed at western Britain’ (2000, 127). Overall,

Cunliffe offered a new impression of the Late Antique Western Seaboard in his 2008 monograph. He stated that continuity in long-distance Atlantic trade was revealed by the presence of Mediterranean pottery – East Mediterranean and North African – along the Atlantic coast of Iberia, and noted that ‘considerable quantities’ had been found in ‘the estuaries of the Tagus, Mondego and Miño Rivers’ and as far north as A Coruña [222] (Cunliffe 2008, 432). The specific evidence was not expounded, but significantly, these finds were not presented to signify the occasional stopping-points of Byzantine ships. Instead Cunliffe saw the pottery as ‘speaking’ of ‘trading ships 26

History of research

Figure 2.9. Reynolds’ 2010 distribution of LRC in the western Mediterranean. From Reynolds 2010, Map 12.

sailing north from Iberia to serve the more northerly of the Atlantic communities’ (Cunliffe 2008, 433). These, he added, were more likely to have been ‘Atlantic-based vessels’, although he conceded that it was ‘not impossible’ that Mediterranean vessels might occasionally have ‘braved the Bay of Biscay’.

est within British and Irish archaeology. The pottery has provided crucial dating evidence for early medieval sites in western Britain and Ireland, based principally on the typo-chronological sequences of the fineware developed in the Mediterranean. New finds have continued to appear in Britain and Ireland as a result of rescue and research excavations, while the re-analysis of older site assemblages has revealed a greater variation within the British assemblage as a whole. Nevertheless, interpretations of these imports have never fully addressed the British assemblage within the wider Atlantic. Models of contact have been repeated, but rarely questioned.

Finally, in a recent article, Portuguese archaeologist Carlos Fabião has criticised previous interpretations of Late Antique trade that overlooked evidence from the western Iberian coast. He commented that the Atlantic represented a ‘forgotten paradigm’ and added that ‘Lusitania’ had typically been viewed as a ‘remote periphery’ (Fabião 2009, 43). In particular, Fabião took exception to the continued production of maps showing the transmission of Mediterranean imports to Britain that ignored evidence from the Iberian Peninsula or overlooked the importance of the Atlantic routes. In contrast, Fabião’s research indicated that the Atlantic routes had acquired a new relevance in Late Antiquity, reflecting contemporary ‘difficulties’ in road and riverine transport (Fabião 2009, 43-4).

Recent publications from the Continent have started to offer new information on the supply of ‘Late Roman’ Mediterranean pottery to the Atlantic region, and the subsequent circulation of early medieval wares. This emerging data provides a new opportunity to examine patterns of trade or exchange to Britain between the fifth and seventh centuries. In particular, it has become increasingly evident that established models of transmission are founded on a restricted Atlantic dataset.

These recent publications reflect an increasing awareness of the complexity of exchange along the Atlantic Seaboard from the fifth century. Specifically, a picture of Mediterranean pottery being present at sites up the Atlantic coast has emerged – indicating that the British imports were not isolated.

The rest of this new appraisal will directly address the two factors that have underpinned the model of direct shipment to Britain. The composition of the British assemblage will be reviewed in Chapter Three as part of a wider examination of the categories of material found in western Britain. Campbell’s research focused largely on E ware, and on what he saw as the ‘Continental trading system’ and although his 2007 monograph provided a description of the Mediterranean imports, there is a need to provide

2.4 Conclusions The ceramic imports have remained a topic of major inter27

Links to Late Antiquity address the apparent dearth of comparable material from the rest of the Atlantic. The ‘sparse’ distribution of East Mediterranean wares on the Iberian coast will be re-evaluated, as well the supposed ‘lack’ of material in Atlantic regions of France. At the same time, the Insular evidence will be reconsidered in relation to the new Atlantic data. These conclusions will be brought together in Chapter Seven to consider the British and Irish imports within their wider, Atlantic context.

a more ‘up to date’ summary (2007a, 3, 14-26, Appendix 1). Ongoing research by scholars in the Mediterranean has revealed new information on the chronology and typology of the Mediterranean wares, and on the contents of the amphorae. These new understandings will be developed in Chapter Four within a discussion of new evidence from Britain, and how this affects the ‘exceptional’ character of the Insular assemblage. Chapters Five and Six will

28

3 Late Antique imported pottery on the Atlantic Seaboard: developments in research 3.1 Introduction

This situation has started to change, particularly through Ewan Campbell’s research, which has increasingly promoted the usage of standardised terminology. It should be noted, however, that the continuing use of the British classifications reflects the condition of the pottery as commonly encountered on British sites. The sherds are often very small and abraded, and frequently comprise undiagnostic body-sherds. Complete or semi-complete examples are very rare. This, consequently, limits the extent to which finds from Britain can be matched to the increasingly precise typologies employed in Mediterranean research. A few British sites have provided semi-complete vessels or vessels that could be reconstructed to reveal the full profile. In particular, the most recent work at Bantham produced a number of examples of amphorae in good condition that could be reconstructed for illustration, and this has provided an opportunity to identify types using the more closely sub-divided classifications (see Subchapter 4.3.2).

This chapter will present a review of the types of pottery imported from the Mediterranean to the Atlantic Seaboard between the fifth and seventh centuries AD. Pottery of Continental origin (DSP and E ware) that has been identified as imports at sites in western Britain and Ireland will also be described. The development of the classifications used in Britain and in the Mediterranean will also be discussed. In particular, this chapter will review current understandings of the origin, production history, typological development and chronology of these ‘import wares’. As this summary is based principally on the major categories of pottery imported to Britain and Ireland, late sigillata or finewares produced in France or Spain will not be reviewed – including late Spanish sigillata (elsewhere TSHT) or Continental imitations of Mediterranean products. Similarly, coarse and cooking wares will not be described at length. These types will, however, feature in the discussions presented in Chapters 5, 6 and 7 as indicators of wider ceramic production, consumption and exchange. This chapter will also consider current understandings of the function of the imported wares, focusing principally on the amphora contents. The chapter will conclude with a brief review of other materials and commodities, including glass, that have been considered to be features of this same system of exchange. Nevertheless, the focus of this project is the major classes of pottery imported to western Britain and Ireland, and this chapter reflects this primary concern.

Issues of terminology have created additional difficulties as part of this project. Each of the ceramic classes has been recorded under a variety of names within reports from Britain, France, Spain and Portugal. The individual types, forms and variants have also been identified using various typological schemes. Appendix A, therefore, presents a concordance of the terminology used to classify these wares, as well as any abbreviations used in the following chapters. This table also notes the presence or absence of these wares (where known/specified) at Roman period sites in Britain (up to c.AD 410) and in early medieval Atlantic contexts.

This chapter also presents an opportunity to summarise the British examples of these categories of material – drawing principally on the catalogues produced by Charles Thomas and Ewan Campbell, but also including some recent discoveries that will be discussed further in Chapter Four.

3.2 Amphorae The vast majority of the imported sherds recovered from early medieval sites in Britain and Ireland are from amphorae: ceramic vessels designed for transportation and storage – particularly of liquids, and most commonly olive-oil, wine and fish-sauces.

3.1.1 Issues of classification and terminology As summarised in Chapter Two, there has been a long history of research on the late Roman and early medieval pottery imported to Britain. This has resulted in a terminological complexity and, as a result, a degree of disparity between research in the Mediterranean and Britain. Although research from the Mediterranean has been incorporated into studies on the British imports, the categories developed at Tintagel and by Thomas’ work at sites such as Gwithian have lingered in British publications, particularly for the imported amphorae. This has hampered the integration of British data with findings emerging from the Atlantic region.

3.2.1 History of research in Britain and Ireland Sherds of ridged pottery that were later recognised as amphorae have been collected and described from sites in south-western Britain since the nineteenth century (see Sub-chapter 4.3.4). It was, however, Radford’s excavations at Tintagel in the 1930s that led to the first identification of sherds as fragments of amphorae, and to the first attribution of a Mediterranean source. His 1956 article briefly describes a collection of buff to pink sherds with ‘deep regular rilling’ formed by combing, that were clearly not of Romano-British date (1956, 60). Radford 29

Links to Late Antiquity

Figure 3.1. Riley’s Late Roman Amphora types. Redrawn from Riley 1981, 117, Figure 10

adds that the ‘shape’ could not be reconstructed – the accompanying photograph showing fragments rather than identifiable vessel profiles (1956, 61, Plate VI). Most significantly, Radford also mentions comparable sherds from Garranes in Ireland, demonstrating that equivalent vessels were imported to other sites across the Atlantic region (1956, 60). As part of his alphabetical classification system these amphorae fragments were grouped as ‘Class B’. Although Radford indicated that this was not a homogenous group, which suggested a range of origins, it is clear that his descriptions principally relate to Late Roman 2 (LRA2) (Campbell 2007a, 19). This type was subsequently recognised as particularly characteristic of the Tintagel assemblage.

Radford’s suspicions of an East Mediterranean source were reiterated in Thomas’ 1959 article. Here, Thomas extended Radford’s classification, subdividing Class B into sub-groups ‘Bi, Bii’, ‘Biii’ and ‘Biv’ on the basis of fabric, surface-treatment and vessel shape. Bi and Bii were noted to be found in association at a number of British sites, as well as in the Mediterranean (Thomas 1959, 92). ‘Biii’ represented a ‘catch-all’ category for un-combed or ridge-less sherds which could not be otherwise classified; the term would be later be replaced by ‘Bmisc’ and later again, ‘untyped’ (miscellaneous) to cover any unattributed sherds (Thomas 1981, 17; 1988a, 12). By his 1981 catalogue Thomas had also added sub-classes ‘Bv’ to refer to amphorae purportedly of Byzacenan (Tunisian) origin and 30

Late Antique imported pottery on the Atlantic Seaboard: developments in research noted, however, that the evidence for LRA4 is less clear. The main characteristics of these types are summarised below, but more detailed descriptions can be found in Peacock 1984b, Tomber and Williams 1986, Peacock and Williams 1986, Pieri 2005 and online at the University of Southampton Digital Resource (2005)

‘Bvi’ for vessels originating in ‘Gaza’ (1981, 15-6). Unlike the fineware, these amphora classifications were not quickly made obsolete by terminology emerging from Mediterranean research. The ceramic reports and discussions within publications on British and Irish sites with imported wares typically referred to Thomas’ sequence (for example Rahtz et al. 1992; Alcock 1995; a notable exception would be Quinnell 2004). A major article on the types present in Britain was published in the first volume of the Journal of Roman Pottery Studies (Tomber and Williams 1986). This noted alternative classifications, but was principally based on the British ‘B-ware’ sequence. Wooding’s published thesis, by contrast, employed the numerical classification outlined in Peacock and Williams’ guide to amphora types (Peacock and Williams 1986; Wooding 1996a).

Research on East Mediterranean amphorae Subsequent research in the Mediterranean has developed understandings of these amphorae, revealing that these ‘standard’ types were produced at a multiplicity of workshops widely distributed across the eastern Mediterranean (Bonifay 2007, 97-8). Furthermore, Riley’s typological sequence has itself been considerably refined. In their 1995 article on the amphorae from Marseille, Bonifay and Pieri highlighted the need to develop Riley’s classifications, stating that the types no longer reflected the complexity of ‘typo-chronological variation’ (1995, 117). Pieri’s ensuing publications, based principally on finds from south-eastern France, have outlined detailed sub-divisions of Riley’s types (1998; 2005; 2007; see Figure 3.2).

More recently, the British finds have been (largely) correlated with the eastern Mediterranean ‘Late Roman Amphora’ sequence developed by Riley at Benghazi and Carthage – based on the most common fourth to seventh century types (1979; 1981) (see Figure 3.1). For East Mediterranean amphorae this sequence superseded an earlier typology developed by Keay, developed through research in Spain (1984). Keay’s typology, nevertheless, remains in standard use for amphorae of North African and Iberian origin. Ewan Campbell opted to use Riley’s types in his 2007 monograph, stating that these had become ‘standard nomenclature in Mediterranean publications’ (2007a, 19). As such, the British Bi amphora can be identified with Riley’s Late Roman amphora 2 (LRA2), Bii with LRA1, Biv with LRA3 and Bvi as LRA4. The last remnant of the B-ware sequence retained by Campbell was ‘Bv’ for amphorae of supposed North African origin – as will be discussed below.

The standard eastern amphora types of Riley’s sequence (LRA1-7) were in widespread production for a number of centuries and are, therefore, not in themselves closely dateable. The Insular imports have consequently been dated on the basis of associated fineware; the established date for the arrival of imports into post-Roman Britain, c. 475-550, is based on dated typologies of LRC and ARS, not amphorae (Campbell 2007a, 19). Pieri’s sub-divisions have consequently allowed the Late Roman Amphorae to be dated much more closely, which is particularly useful for sites which have amphorae but not fineware. It should be noted, however that given the typically small fragments

Despite the recognition that the British finds could be directly equated with well-documented types of amphorae distributed across the Mediterranean, Campbell noted that Thomas’ B-ware classifications has become ‘embedded’ in the British literature. The term B-ware and the Bi-Bv sub-types are still regularly used in British reports, notably in the most recent publication on the Tintagel pottery (Barrowman et al. 2007; see Sub-chapter). The East Mediterranean amphorae discussed in subsequent chapters have been recorded using Riley’s types (and subsequent refinements), although other classifications may have been used in the original reports. 3.2.2 East Mediterranean amphorae The main types of amphorae identified at early medieval Insular sites are Riley’s LRA1 and LRA2. Both types were produced between the fourth century and seventh century and were preceded by earlier ‘antecedents’ (University of Southampton 2005), but they are not thought to have been imported to Britain during the period of Roman control (prior to AD 410). LRA3 and LRA4 are reported to be present in late Roman urban contexts in Britain (Campbell 2007a, 19-20, 125-6; see Sub-Chapter 7.2.1). It should be

Figure 3.2. Pieri’s East Mediterranean amphora sub-types (LRA1, 2 and 4 only). After Pieri 2005, Figures 25, 45, 66

31

Links to Late Antiquity 70-4; 2007, 150). LRA1B is broadly dated from the sixth to the mid-seventh century (Pieri 2005, 75-7).

found at Insular sites – and the frequent lack of diagnostic sherds such as handles – attribution within such close subdivisions is not usually feasible. Nevertheless, these refinements are increasingly being used in Continental and Mediterranean publications – as at Vigo (Fernández 2014) – and have the potential to reveal more subtle shifts within patterns of Late Antique trade. The closer identification of the British examples should be a specific goal of future research, especially where larger diagnostic sherds are recovered.

LRA1 ‘Bantham-type’ amphora Excavations at Bantham in south Devon produced vessels recorded as a variant of LRA1 (Bidwell et al. 2011, 97-9). This is described and discussed in Sub-chapter 4.3.2. Late Roman 2 amphora (LRA2) – Bi (Thomas 1959), Class 43 (Peacock and Williams 1986)

A greater range of amphorae have been found at Atlantic sites than in Britain. Only the major types found in Britain and Ireland are described below, but Atlantic examples of other types are noted in Chapters 5 and 6.

This globular amphora is distinguished by ‘a short conical neck with a high everted rim and bowed handles’ with deep, horizontal grooving/rilling on the upper part of the body (Peacock and Williams 1986, 182). The fabric is generally hard, fine-grained and smooth, with scattered limestone (Peacock and Williams 1986, 184). LRA2 is the standard late Roman Aegean amphora, with production sites identified on Chios and Cnidos and the Argolid region of Greece, although a more widespread origin is possible (University of Southampton 2005). As with LRA1, Pieri was able to identify chronological developments in the shape and surface treatment, allowing him to divide LRA2 into three sub-types, LRA2A-C. The earliest of these, LRA2A, is characterised by a spherical body and a funnel-shaped rim; this was suggested to be typical of the fifth and earlier-sixth century (Pieri 2005, 88). From around the mid-sixth century this shape was superseded by LRA2B with an elongated collar (Pieri 2005, 88). By this point the surface-rilling was typically wavy rather than straight (University of Southampton 2005). A second late variant, LRA2C, was identified with a vertical, rather than flaring neck; this can be seen to precede the ‘globular amphorae’ that characterise eighth and ninth century amphora-exchange in the eastern Mediterranean (Pieri 2005, 90). LRA2C has elsewhere been classified as LRA13, by which term it was recorded at Vigo (Riley 1979, 231; Fernández 2014, 344)

Late Roman 1 amphora (LRA1) – Bii (Thomas 1959), Class 44 (Peacock and Williams 1986) The characteristic Late Roman 1 amphora (LRA1) has an ovoid body with a rounded base, relatively thin walls, a thickened rim and stumpy, grooved handles (Peacock and Williams 1986, 187; Williams 2005, 159; University of Southampton 2005). There is typical ‘clapboard’ ridging across the centre of the body, narrowly-spaced ribbing at the top and bottom as well as distinctive grooves on the interior surface. However, there is considerable variation in these features as the type evolved over time. Following pioneering investigations by Empereur and Picon, production has been identified at a number of sites in Southern Turkey (particularly Cilicia), Rhodes and Cyprus (including Paphos), with a concentration in the Gulf of Iskenderon (Empereur and Picon 1989; Williams 2005, 160-1). The form was even imitated by North African producers (Bonifay 2005, 457). The fabric is typically hard and sandy with limestone inclusions, although both fabric and colour and can vary considerably, reflecting the range of production sources. Petrological analysis by David Williams indicated that it was possible to distinguish sherds of Cilician/Syrian and Cypriot origin, while Pieri described five fabric groups and connected these to individual or multiple sources (Pieri 2005, 84; Williams 2005, 166). To date, however, the British sherds have not been matched to specific origins. LRA1 was one of the most important and widely distributed transport vessels of this period and was the most common East Mediterranean amphora imported to sites in the West Mediterranean, exemplifying a general expansion of trade from the east by the mid-fifth century (see Sub-chapter 6.2.4)

Late Roman 3 Amphora (LRA3) – Biv (Thomas 1959), Class 45 (Peacock and Williams 1986) This amphora is characterised by a long, slender neck, high shoulders and a tapering foot, which becomes more pointed and closed by the sixth century (University of Southampton 2005). LRA3 are typically thin-walled, dark red-brown in colour with a characteristic highly micaceous fabric (Peacock and Williams 1986, 189). Several origins for this type have been proposed, principally around western Asia Minor, with Pieri highlighting Aphrodisias and Ephesus as likely production centres (University of Southampton 2005; Pieri 2005, 100). Vessels of this type have a long history, with hollow-footed, single-handled versions being produced from the first century. Although earlier versions have been identified in Romano-British contexts (Tyers 1996, 102-3), the presence of two-handled vessels – which appear from the fifth century – has been taken to indicate post-Roman importation (Campbell 2007a, 19). It is not a common import to the western

Pieri’s refined sequence contained two sub-types (‘morphological generations’) of LRA1 – the early ovoid LRA1A and the later, larger-mouthed, cylindrical LRA1B (see Figure 3.2). The later sub-type was further divided into LRA1B1 and LRA1B2 – the latter having a funnelshaped mouth (Pieri 2005, 70-7). Pieri also identified small versions (‘sous-modules’) of each sub-type. The first examples of LRA1A were dated by Pieri to the end of the fourth century, with the transition between LRA1A and 1B assigned to the late-fifth or start of the sixth century (2005, 32

Late Antique imported pottery on the Atlantic Seaboard: developments in research century to the eighth century (University of Southampton 2005). Pieri (2005, 116-122) outlines a detailed typological sequence for the later examples, including standard and small versions, but this is currently of less relevance for the Atlantic.

Mediterranean after the sixth century and therefore does not feature in the quantification of amphorae from the late contexts at Vigo (Pieri 2005, 95; see Sub-chapter 6.3.2). Pieri described two sub-types of this amphora, of which LRA3A was the standard late version distributed around the Mediterranean; this was further sub-divided on the basis of rim shape, although Pieri noted that as this was often missing, such attributions were difficult to apply (2005, 96). A parallel sequence of small versions (‘sousmodules’) were categorised as LRA3B; Pieri noted that these did not reach Mediterranean France beyond the fifth century (2005, 98).

Late Roman Amphora 7 (LRA7) – Class 52 (Peacock and Williams 1986) The final type from Riley’s original sequence has also not been confirmed as an import to early medieval Britain or Ireland and has no ‘B-ware’ equivalent. However, an example has been identified at Vigo. Characterised by its tubular neck, tapering body and solid spike, as well as its dull brown colour and rough sandy fabric, production of LRA7 was centred on the Nile Valley (University of Southampton 2005; Fernández 2014, 346). Pieri has identified a detailed sequence for LRA7 across its long production history – from the fourth to the eighth century (2005, 128-32).

Late Roman 4 Amphora (LRA4) – Bvi (Thomas 1981), Classes 48 and 49 (Peacock and Williams 1986) Amphorae of this broad tradition have frequently been referred to as ‘Gazan’ although kilns have been identified in southern Palestine and Israel (University of Southampton 2005). This ‘cigar-shaped’ amphora has ring-shaped handles situated on the body, with ridging at the shoulders and base and characteristic ‘clay accretions’ on the shoulder and rim (University of Southampton 2005). Typically ‘drab brown’ in colour the fabric is hard, grainy and sandy with scattered limestone (Peacock and Williams 1986, 196-9).

In addition, a wider range of East Mediterranean types have been identified at Vigo, although in smaller numbers, revealing a more complex picture of exchange between the East Mediterranean and the Atlantic than has been shown in earlier publications (Fernández 2014, 346-8). 3.2.3 British and Irish examples of imported amphorae

LRA4 also shows a long and complex typological development. Specifically, over time the vessels became longer and narrower. Early (second to third century) vessels with thick walls and wide bodies were categorised by Peacock and Williams as Class 48. This was noted to be the precursor of the narrower, elongated Class 49, elsewhere classified as Almagro 54 (Peacock and Williams 1986, 196-9). Pieri has more recently outlined a detailed sequence of shapes for this type: LRA4A1 dating from the fourth/early-fifth century; LRA4A2 dating, roughly, from the second-quarter to the end of the fifth century; LRA4B1 dating across the later-fifth and first half of the sixth century, and LRA4B2 dating up to the later-sixth and seventh century (Pieri 2005, 101-7; see Figure 3.2). His LRA4B3 type represents its latest, late-seventh/early-eighth century incarnation. Notably, a chronologically extended range of LRA4 has been identified in Bordeaux (see Sub-chapter 5.3); Berthault’s use of ‘LR 4c’ for sixth/seventh century examples can be equated with sixth or early-seventh century versions of Pieri’s LRA4B sub-type (Berthault 2012, 316).

Ewan Campbell’s 2007 monograph presented a minimum total of 250 post-Roman imported amphorae (all types by MNV) from Britain and Ireland (2007a, 22). By his 2011 dataset this total had risen to 284 vessels. Additional Insular examples have also been reported in recent years, including new finds from Ireland (Doyle 2009) and following the publication of the Bantham excavations (Reed et al. 2011). New amphora sherds have also been recorded from Mothecombe and High Peak in South Devon (see Sub-chapter 4.3). Additional recent finds from Britain include 21 sherds of LRA1 (representing at least two vessels) from Glastonbury Abbey [101], which were identified during a review of material excavated by Radford in the 1950s (Allan et al. 2015). This discovery has extended the history of occupation at the site into the later-fifth or sixth century; previously finds of post-Roman imported amphorae had only been reported from Glastonbury Mound [102] and Tor [103]. These new finds bring the suggested minimum total to 322 amphorae of all types. Of these, 225 have been directly identified with East Mediterranean types (see Figure 3.3). The remainder comprise Late Roman Amphorae of unidentified types, ‘North African’ amphorae of potential post-Roman date, and amphorae of the provisional ‘Thick walled’ and ‘Bv’ classes which are discussed below. Campbell’s synthesis used the category ‘LRA’ (Late Roman Amphora) to denote amphorae of uncertain origin, which he notes as superseding Thomas’ ‘Biii’ and ‘Bmisc’ categories (Campbell 2007a, 19). This classification has been followed.

Bag-shaped amphorae/Late Roman Amphora 5-6 – Class 46 (Peacock and Williams 1986) ‘Bag-shaped’ amphorae of this broad, Palestinian tradition have not (to date) been identified within the post-Roman Insular assemblage – and were therefore not included in the ‘B-ware’ classification – although examples have been identified from Bordeaux. Characterised by its bag-shaped body with ‘corrugated’ surfaces and ring-handles vessels of this general type were produced form the first/second 33

Links to Late Antiquity

Figure 3.3. Imported amphorae from Britain and Ireland by MNV. Based on Campbell 2011; Doyle 2009; and additional data

moved, the relative proportion of LRA2 to LRA1 drops from 25.2% to 21.1% (see Figure 3.4). It may be that this Aegean amphora is particularly well represented/identified at Tintagel. These observations are, however, complicated by the restricted data from Tintagel. If the overall assemblage is considerably larger, these relative proportions must be reassessed.

The assemblage from Tintagel, seemingly containing the largest quantity of amphorae in Britain or Ireland, has presented particular difficulties. Specifically, issues of quantification and the absence of a full, published catalogue have obscured the quantity of vessels recovered from the site. The recent report of excavations at Tintagel suggested a total of 150 amphorae and 80 fineware vessels, but these were not sub-divided by type (Thorpe 2007, 246; Barrowman et al. 2007, 317, 329). This issue is discussed in detail in Sub-chapter 4.2.3. By necessity, Campbell’s lower minimum figure for the site (99 amphorae) has been used within this project to consider the relative proportions of amphora types, but it seems likely that this considerably underestimates the quantity of vessels from the site. If the suggested amphora total from Tintagel is changed from 99 to 150, the overall Insular amphora total would rise to 373.

A final observation relates to issues of identification. It is increasingly clear that there is a considerable quantity of vessels (roughly 20%) that have not been assigned to a type or origin – either due to their fragmentary state/lack of diagnostic elements, but also due to problems in identifying fabrics. As additional information becomes available on the range of amphora distributed in the Mediterranean – and particularly in the Atlantic – this situation should improve. However, these refinements will ultimately depend on collaborations between British researchers and those working in regions with more complete vessels and larger assemblages.

Despite these problems, certain broad conclusions about the total Insular amphora assemblage can be made. Firstly, that it is dominated by amphorae of East Mediterranean origin. Vessels of likely North African origin are rare, especially if much of the ‘Bv’ category cannot be assigned an African origin (see below, 3.2.4). This would appear to reinforce the observations outlined in Chapter Two – that the Insular assemblage is markedly different to patterns in the West Mediterranean. Of the East Mediterranean types, LRA1 is clearly the most common. Its increasing representation in the Insular group is certainly due to the new data from Bantham, but does reflect a wider trend emerging from British sites. The quantity of LRA2 is definitely high and does appear to be a characteristic of Britain and Ireland (see Appendix H.6). However, if the figures from Tintagel – based on Campbell’s lower estimate – are re-

3.2.4 North African amphorae: unravelling the ‘Bv’ classification Since Thomas’ addition of the ‘Bv’ Class there has been a general understanding that North African (Zeugitanan/ Byzacenan) amphorae, as well as ARS, were imported to post-Roman Britain. However, the evidence for this is, currently, much less clear than for the other classes of imported pottery. Thomas described how the 1981 ‘sorting’ at Tintagel led to the classification of about 200 fragments as ‘Bv’, representing three or four individual vessels (Thomas 1981, 15). Many of these comprised sherds col34

Late Antique imported pottery on the Atlantic Seaboard: developments in research

Figure 3.4. Imported amphorae from Britain and Ireland by percentage of MNV. Based on Campbell’s lower estimated vessel count – shown including and excluding the Tintagel amphorae

lected during Radford’s excavations which had originally been thought to have been medieval in date (Thomas 1988c, 425). The revised identification of these sherds followed research by David Peacock on the presence of North African amphorae at late Romano-British sites (1977a, 270-2). This work had identified fragments of amphorae from Gloucester in a ‘distinctive hard red fabric…with an off-white slip finished with vertical smoothing’ as being similar to sherds recovered at Tintagel – contained within the un-typed ‘Biii’ group (Peacock 1977a, 270). The North African source was identified on the basis of a comparable ‘complete vessel’ from Holborough in Kent, which Peacock matched to a sub-type of amphorae known in the Mediterranean and classified by Clementina Panella as the fourth century ‘Africana IID’ (Panella 1972, 102; Peacock 1977a, 271). Panella’s ‘Africana’ series was in itself a refinement of the broad ‘Africana Grande’ categorisation created by Zevi and Tchernia to describe large, cylindrical African amphorae of the third and fourth century – produced in parallel with the small ‘Africana Piccolo’ (Zevi and Tchernia 1969).

sherds they could not usually be matched to the developing Mediterranean typologies – specifically the numerical sequence devised by Keay which had entered standard use (1984). These finds were nevertheless taken to indicate that vessels of a ‘general North African cylindrical type’ were imported to Britain from the third century, and particularly in the later-fourth and early-fifth century (Peacock 1977a, 271-2). The Tintagel sherds suggested a later, postRoman phase to this importation, but the apparent absence of equivalent sherds from other known post-Roman sites in the south-west indicated a focus on the early-fifth century, rather than later. Significantly, Peacock mentions equivalent North African imports from a site in Bordeaux; although he does not name this, the citation given relates to the investigations at the Church of Saint-Seurin – which is discussed in Sub-chapter 5.3.2 (Coupry 1969, cited in Peacock 1977a, 272). These late ‘Romano-British’ imports from North Africa were the subject of a 1995 ‘reappraisal’, by which point a range of types and sources (including examples of Keay 25/Africana III amphorae) had been recognised (Williams and Carreras 1995). Here, the amphorae are directly interpreted as transport vessels for olive-oil as part of an importation system which superseded, but never equalled, the earlier Imperial arrivals of oil in the Baetican Dressel 20 amphora (Wiliams and Carreras 1995, 234-5). The

Subsequently, more examples of North African amphorae were recognised from a number of British sites in late Romano-British contexts – the majority from urban sites in southern and western Britain, including London, Southampton, Dorchester and Exeter. As most were body35

Links to Late Antiquity presence of North African sherds in late Roman contexts at Exeter is explained by its position as a ‘convenient portof call’ for ships plying the Atlantic route, or arriving from Bordeaux (Williams and Carreras 1995, 238). The latest of these identified was a Keay 25B dated to the late-fourth/ early-fifth century (Holbrook and Bidwell 1991, 218; Williams and Carreras 1995, 250). Williams and Carreras discuss the possible arrival of North African amphorae after the end of Roman control in AD 410; a few fragmentary body-sherds from Bush Lane, London, Mucking and Market Hall, Gloucester are offered as potential fifth century imports, but this could not be confirmed (1995, 240). It is only the sherds from Tintagel and an isolated example from Dinas Powys that are taken to represent definite postRoman arrivals. These are, however, interpreted within the model of direct shipment from the East Mediterranean – as representing a ‘subsidiary cargo’ collected in North Africa/ Carthage (Williams and Carreras 1995, 241). Significantly, the authors describe the difficulty in obtaining a ‘reliable figure’ for the post-Roman North African imports, but note that apart from Tintagel, they would appear to be ‘thin on the ground’ (1995, 240-1).

belled ‘spatheia’ (Keay 26), were in general production from the fifth to the seventh century; within this timeframe he outlined three chronological ‘generations’ (Spatheion 1-3) (Bonifay 2004, 125-9). The only spatheion specifically attested from Britain is from Shadwell [*A], London (of Type 1). Dating to the late-fourth or fifth century, this was found in a probable fifth century context, and represents a likely import of the first half of the fifth century (University of Southampton 2005; Douglas et al. 2011, 68, 172; Williams 2011b, 80). This has been shown in Appendix H.8. Additional examples of spatheia have been identified in Atlantic contexts, as noted in Chapters 5 and 6. Bonifay’s final late group, ‘Africaines cylindriques de grandes dimensions’ represents a heterogenous ‘family’ of vessels produced between the fifth and seventh century, within which Bonifay has again identified two broad ‘generations’. The first ‘generation’, which includes Keay 59, Keay 35A and 35B and Keay 55-57 are broadly dated between the late-fourth and mid-sixth century (Bonifay 2005, 453-4). The second ‘generation’ of even larger vessels includes the major export type Keay 62 (including its many variants; the type is broadly dated between the sixth and early-seventh centuries) and Keay 61, characteristic of the late-sixth and, particularly, seventh century (Bonifay 2004, 137-41; 2005, 455-6). It is, therefore, these types, rather than the earlier ‘Africana’ series, that should comprise any later-fifth or sixth century North African imports to western Britain and Ireland. Vessels of both ‘generations’ of late cylindrical types have been identified from sites on the Atlantic Seaboard, and have been plotted in the map in Appendix H.8. To attempt some parity with the British post-Roman sites, and to restrict the data to likely fifth century imports, only vessels of Bonifay’s latter two chronological groups (spatheia and the ‘Africaines cylindriques de grandes dimensions’ have been recorded within this project. However, it is acknowledged that this creates difficulties in understanding the shifting patterns of later-fourth and earlier-fifth century importation to Britain and the wider Atlantic. Future research is necessary to refine understandings of amphorae such as Africana III/ Keay 25 – especially the latest versions – and the relationship between their British and Continental Atlantic distribution.

Although Williams and Carreras catalogued these Romano-British examples using standard Mediterranean terminology, the ‘Bv’ classification has persisted for the Tintagel sherds, as well as the association with the broad ‘Africana Grande’ type. Following Thomas, Wooding directly equated ‘British Bv’ with ‘Africana II’ and Peacock and Williams’ Class 34 (1996a, 42). Similarly, Campbell described ‘Bv’ as a ‘general term for late North African cylindrical amphorae of the ‘Africana Grande’ type’, which included Keay type 7 (equivalent to Africana IID) (2007a, 19). It should be noted, however, that Campbell had previously assigned a tentative identification of Keay 55B for the aforementioned vessel from Dinas Powys, although this was catalogued as ‘LRA’ (unidentified) in his most recent dataset (1998, 128, 134, Figure 28.25; 2011). Since the 1980s research on North African amphorae has developed considerably in the Mediterranean, particularly as a result of Michel Bonifay’s work in Africa, and his parallel publications on imports to south-eastern France/ Marseille (2004; developments summarised in Bonifay 2007, 98-100). Most significantly, Bonifay has refined and described the complex typological trajectory of North African amphorae beyond the classic third and fourth century ‘Africana’ types, whilst retaining Keay’s terminology (see Figure 3.5). Bonifay describes three broad categories that characterise amphora production from the fifth century. Firstly, the latest examples of the ‘amphores cylindriques de moyennes dimensions’ (Keay 25.2/Africana IIIC) produced up to the mid-fifth century (Bonifay 2004, 119). The ‘medium-sized’ Keay 25/Africana III series have been said to fall ‘typologically between the ‘Africana Grande’ series and later, larger, cylindrical amphorae (University of Southampton 2005). Following this series, African production is characterised by a ‘double tendency’ towards increasingly smaller and larger types. Bonifay’s second late category, small tapered vessels la-

Insular examples of ‘Bv’ and North African amphorae Campbell’s 2011 dataset included 34 examples of Bv amphorae from eight Insular sites, as well as one from Saint-Laurent-des Combes in France (see Sub-chapter 5.2.1). These had been identified from various excavations, including Dalkey Island (see Doyle 2009, 46) on the basis of fabric and typical vertically-smoothed white surfaces. Campbell nevertheless noted that it was rarely possible to identify specific types as the Insular examples were ‘so fragmentary’ (2007a, 19). The vast majority of these Insular examples recorded as ‘Bv’ come from Tintagel (see Figure 3.4). By 1988 Thomas had suggested the Tintagel ‘Bv’ numbered ‘probably not less than 27’ (1988a, 12). 36

Late Antique imported pottery on the Atlantic Seaboard: developments in research

Figure 3.5. Bonifay’s typological development of African amphorae, highlighting selected ‘fifth century’ and later amphorae. After Bonifay 2004, Figure 46

Campbell subsequently included a minimum total of 23 vessels from the site (2011). As the Tintagel assemblage has never been fully published, relatively little information is available on these vessels and they have never been further identified (see Sub-chapter 4.2.3). The recent identification of sherds from Bantham and Mothecombe as specifically ‘North African’ would suggest that vessels of Bonifay’s later types were imported to post-Roman sites in Britain, or otherwise would represent residual sherds of fourth century vessels. Unfortunately, neither site has

produced diagnostic sherds which might clarify this question. The sherds from Bantham were noted to resemble the aforementioned late Roman examples from sites such as Exeter, but their association with the East Mediterranean imports suggests they might equate with the later-fifth and sixth century African types (Keay 35, 62, spatheia?) recorded elsewhere in the Atlantic. Although African amphorae would appear to represent a small proportion of the post-Roman imports, the general 37

Links to Late Antiquity ity – based on apparent similarities with vessels produced at Bueu in Galicia [256], approximately fifteen kilometres to the north of Vigo (Fernández Fernández, pers. comm). This research is at an early stage but has the potential to revise understandings of ceramic supply in the Atlantic.

origin of the vessels classified as ‘Bv’ remains unresolved. Thorpe’s 1997 report from Tintagel stated that despite the absence of any rim-sherds of this type, the ‘Bv’ vessels resembled ‘Africana Grande’ amphorae, which were typically connected to the transportation of olive-oil. However, it also stated that fabric analysis had suggested an East Mediterranean origin (Thorpe 1997, 78). The ceramic report of the 1990s excavations (Thorpe 2007) again noted the type as unprovenanced and again noted a resemblance to ‘Africana Grande’. However, the ‘Bv’ fabric is described as ‘very sandy’ and pale buff-brown in colour, rather than the typical ‘brick red’ previously given for the African imports (Thorpe 2007, 233). By this point the first diagnostic sherds of ‘Bv’ had also been recovered, allowing a composite profile illustration to be produced (see Figure 3.6). The twisted, squared handles and sloping body of the illustrated vessel cannot be matched to any of the main African types, while the description of the amphora walls as ‘thick’ and ‘ridged’ would again argue against a North African origin (Thorpe 2007, 233). The shape itself is rather more suggestive of late Lusitanian productions illustrated in Fabião 2008. Chemical analysis of selected amphora sherds was also carried out within this recent phase of work at Tintagel (Jones 2007). This indicated that sherds classed as ‘Bv’ – many of which were ‘not uniform in hand-specimen’ – represented a fairly chemically-coherent group. However, although the source might be North Africa, it was not possible to discount southern Spain or even the eastern Aegean (Jones 2007, 254).

It was not the aim of this project to re-classify the Insular finds, and this has not been attempted within the given timescale. However, it has become apparent that in order to fully appreciate the parallels between Britain, Ireland and the Atlantic – and specifically links between Galicia and south-west Britain – future research on this pottery is essential. The ‘Bv’ category represents a specific case where the British research has strayed from parallel Mediterranean research, and where a significant group of vessels remains effectively unidentified. Bonifay has specifically criticised the ongoing reliance on categories like ‘Bv’, which do not represent homogenous groups or reflect current, Mediterranean classifications (Bonifay 2008, 1124). Therefore, a recommended target for future research must be a fully revised, published catalogue of the British finds – especially the Tintagel assemblage – including detailed fabric descriptions and illustrations. As the interpretation of the Insular imports has been frequently been based on the relative proportion of types by source, the resolution of such issues has major implications. As the ‘Bv’ amphorae from Britain and Ireland recorded by Campbell have not been viewed in person they have not been re-classified as part of this project, although it is clear that the category can no longer be seen as an appropriate or useful. The total can, however, be reduced from 33 to 27 vessels on the basis of sherds from Bantham and Mothecombe being split into provisional ‘Thick Walled’ and ‘North African’ categories to reflect differences in fabric, surface and wall-thickness (see Sub-chapter 4).

Reynolds has commented on a ‘generally-held belief’ that Tunisian amphorae are common among the British postRoman imports, based on the supposed North African attribution of the ‘Bv’ class (2010, 109). He subsequently noted that a ‘major component’ of the Tintagel assemblage was ‘thick-walled, buff-coloured amphorae’ which, he suggests, could be best matched to amphorae from the Bay of Cádiz or the Algarve coast (Reynolds 2010, 292, note 392). The only rim – presumably the rim-sherd illustrated in Figure 3.6 – was noted to resemble the Iberian Keay 16 amphora (Reynolds 2010, 293). These identifications were based on a viewing of a sample of the pottery from Tintagel and Bantham conducted in 2002. The ceramic assemblage from Bantham also included body-sherds of ‘thick-walled’ cylindrical amphorae, which were stated to closely resemble ‘Bv’ vessels from Tintagel (Bidwell et al. 2011, 102; see Sub-chapter 4.3.2). A sherd from Mothecombe was tentatively linked to this ‘thick-walled’ classification (see Sub-chapter 4.3.3).

3.2.5 Iberian amphorae As discussed, it seems increasingly likely that the Insular assemblage contains a significant proportion of Iberian amphorae, although this attribution clearly requires additional research. Beyond questions of the Tintagel ceramics, amphorae of southern and western Iberian origin are being increasingly suggested as a factor of Late Antique exchange in the Atlantic. This project has focused on imported pottery of Mediterranean origin, and has not directly considered the movement of late Baetican or Lusitanian amphorae. The exportation of these types appears to have been focused on the West Mediterranean. However, some late Iberian amphorae may have been transported northward along the Atlantic Seaboard, accounting for the small scatter of vessels mentioned in Chapter Five. These relate to fragments of Almagro 50 and 51B/51C amphorae recorded from sites in western Gaul, although it should be noted that some of these older identifications have been questioned (Bonifay; Fernández Fernández; Le Bomin, pers. comm.). It is hoped that ongoing research by Le Bomin and Guitton will resolve the question of Iberian exports to Atlantic Gaul in the fifth century.

Clearly there remains considerable uncertainty about the ‘Bv’ class, but it seems that the majority of sherds classified as such are unlikely to have a Tunisian origin. Although this will need to be verified by typological and petrological analysis, it seems increasingly likely that many of these sherds represent a group of vessels of Iberian (and potentially Atlantic) origin, hitherto unrecognised within the British post-Roman suite of imports. Furthermore, a north-western Iberian origin is a possibil38

Late Antique imported pottery on the Atlantic Seaboard: developments in research

Figure 3.6. Composite illustration of a ‘Bv’ amphora. From Thorpe 2007, Figure 130

The long, cylindrical Almagro 50 has been dated to between the third and fifth century (University of Southampton 2005). The ‘piriform’ Lusitanian type Almagro 51A-B is broadly dated between the third and sixth century (University of Southampton 2005). Almagro 51C has been typically assigned a third to fifth century date, although this type might also show some sixth century continuation. The complex sequence of Lusitanian amphorae has been principally disclosed by Fabião, who has suggested that most, if not all, of the vessel-types could be linked to the transportation of fish-sauce/salted-fish products (2008, 743). Describing the broad chronological development of Lusitanian amphorae he has noted a late ‘apogee’ of production in the fourth and fifth century, typified by the Almagro 51A-B amphora and

concentrated on, though not limited to, the Algarve coast (Fabião 2008, 740). Beyond the late-fifth century, Lusitanian amphora production is less well understood and defined, but Fabião has suggested the continuation of production for export into the sixth century, characterised by smaller vessels (2008, 743). This issue is discussed further in Chapter 6.3. The identification of Iberian amphorae of both Lusitanian and southern-Spanish/Baetican origin at Vigo suggests a more complex pattern of amphorae exchange in the Atlantic than has previously been understood (Sub-chapter 6.3.2). No specific examples of late Iberian amphorae have been published from Britain or Ireland, but it seems likely that this will change. 39

Links to Late Antiquity 3.2.6 Amphora contents

Oil has typically been considered the likely cargo of the North African imported vessels (Williams and Carreras 1995, 240), although this is based on a straightforward association between later Roman imports to Britain and the ‘Bv’ category. Indeed if much of the ‘Bv’ amphorae can be equated with Iberian/Lusitanian amphorae then fish-products must be considered as a possibility (Fabião 2008, 743). Furthermore, Bonifay has also questioned whether too much emphasis has been placed on oil as the likely content of North African amphorae. He records that many of the types distributed in Late Antiquity (including Keay 35B and, in part, Keay 62) have been found with traces of pitch, and would not, therefore, have carried oil (2004, 463-75; 2007, 100). Fish-sauces or wine would be the probable alternatives. The spatheia are also unlikely to have carried oil (see discussion in Reynolds 2010, 284, n.331).

The question of which commodities were stored and transported within the imported amphorae has been frequently debated in publications on the Insular imports, but this has not been fully resolved. Wine and olive-oil would seem to be the most likely products imported to Britain, but other exotic commodities such as fish-sauce (garum, liquamen), olives or dried fruit remain possibilities. Thomas initially suggested that LRA2 might be associated with wine and LRA1 with olive-oil and later affirmed that oil was likely to be the major commodity imported to Britain in amphorae (1959, 92; 1990, 16). However, individual amphora types have not proved easily matched to specific products. Standard shapes were imitated, and vessels could be frequently reused, as indicated by the slate amphora stoppers found at Tintagel (Barrowman et al. 2007, 317-8; Campbell 2007a, 24). To date, scientific analysis of resins or residues has rarely produced unambiguous results. Following the recent excavations at Tintagel, residue analysis by gas chromatography was conducted on a number of sherds – fragments identified as LRA2 and ‘Bv’ but not LRA1 – but the results were inconclusive (Barrowman et al. 2007, 256-70).

Nevertheless, it must be remembered that amphorae may only represent the surviving component of a more diverse exchange. Bonifay has highlighted the probable transport of wine or even fish-sauces in barrels, oil in skins, as well the crucial transport of grain in sacks (Bonifay 2007, 96). Thomas had previously proposed that the importation of wine in barrels was a major feature of trade between western Gaul and Britain/Ireland, and accounted for the distribution of E ware, while Alcock had associated DSP with the transport of wine from Bordeaux in barrels – commenting on their superior efficiency as containers (Alcock 1987, 90; Thomas 1990). Unfortunately, there is little evidence to confirm this hypothesis.

Nevertheless, wine has been increasingly highlighted as a major exported commodity in Late Antiquity, and is considered the principal content of the East Mediterranean amphorae imported to the West (Pieri 2007, 149). LRA1, the most common of the Insular imports, has been increasingly associated with wine (Pieri 2005, 84-85), although Bonifay has commented that oil cannot be completely discounted (2007, 100). A recent study of late amphorae imported to Italy concluded that of seven LRA1 analysed, all had resin coatings and five produced traces that could be related to wine, although indications of subsequent reuse were also identified (Pecci et al. 2010). Resin or pitch coatings, intended to prevent wine seeping through the vessel walls, have been observed on LRA1 elsewhere in the Mediterranean (Williams 2005, 161). Analysis by photoacoustic spectroscopy was conducted on an amphora found in Plymouth Sound (see Sub-chapter 4.3.5). This also hinted at wine as a likely cargo, although this technique is not in standard use.

3.3 Finewares 3.3.1 Introduction Although the amphorae predominate within the Insular assemblage, the finewares have been particularly significant in providing the core dating structure for this early medieval period of importation. The characteristic imports distributed in the Atlantic are African Red Slip ware and Late Roman C ware (Phocaean Red Slip ware) of the Mediterranean red-slipped fineware tradition. Within the Mediterranean these were essentially tablewares, used for dining as cheaper alternatives to precious metals; the shape and decorative features frequently replicate metal ‘prototypes’ (Hayes 1972, 282-3; 1997, 17-8). It seems likely that the vessels found in Atlantic regions of Spain, Portugal (and probably those in Bordeaux) were used in much the same way. The continuing demand for fine tableware across the Atlantic region– as discussed in Chapter Six – indicates continuity in dining practices beyond the political transformations of the fifth century. A dining function can also be considered for the British and Irish examples. However, given their rarity in Insular regions, the fact that much of western Britain and certainly Ireland is considered to be ‘aceramic’ during this period, and their association with long-distance exchange networks, it is possible that the vessels were used and/

Pieri discussed the difficulties in assigning a likely content for LRA2, due to issues of reuse. However, the presence of pitch on various examples, including vessels from the seventh century wreck of Yassi Ada, again suggested wine as a primary content (Pieri 2005, 93). LRA4 has been directly associated with the famed, fine wines of Gaza – as lauded by Gregory of Tours (The History of the Franks, VII.29) – and, consequently, as connected to liturgical use and ecclesiastical supply (Pieri 2007, 152). This type is rare in the British post-Roman imports, but is increasingly being recognised as a factor of late Atlantic exchange (Chapters 5 and 6). Pieri’s research also favoured wine as the likely cargo of LRA3 amphorae (2005, 101). The identification of the latter type as a ‘water jar’ has been discounted (Campbell 2007a, 24). 40

Late Antique imported pottery on the Atlantic Seaboard: developments in research Tunisia. This red-slipped tableware represents a continuation of the earlier Roman sigillata tradition, although the slipped surfaces were less glossy as a result of simpler firing methods used in production (Hayes 1997, 59). Until regional imitations developed in the later fourth and fifth centuries ARS monopolised fineware distribution across the Mediterranean (Hayes 1997, 59), and occasionally to more distant regions such as Ethiopia, the Black Sea and the Danube Frontier (Hayes 1972, 422; 2008, 71). A few vessels have been recorded from Roman contexts in Britain (Bird 1977), although the ware does not seem to have been a regular import. The overall distribution of ARS is widespread but concentrated on coastal areas, indicating that it was largely a seaborne product (Hayes 1972, 414).

or valued differently at British sites than at locations further south in the Atlantic. The lack of comparable local material suggested to Campbell that these vessels might have ‘acquired a cachet beyond their normal importance’ (Campbell 2007a, 16). Beyond their primary use, broken sherds might also have acquired some secondary significance. The information on the Insular examples of imported fineware discussed in this chapter is largely reliant on Campbell’s assessment (2011), as updated from Thomas’ 1981 Provisional List. However, as with the amphorae this might underestimate the total quantities. As mentioned, Thorpe estimated the sherds recovered from Tintagel to represent at least 80 red-slipped vessels, but these were not sub-divided by ware or form (Thorpe 2007, 246). Revisions to Campbell’s data are discussed for each ware, but it is acknowledged that future research might affect the ensuing interpretations.

History of Research As with the amphorae, the long history of classification of late Roman finewares has resulted in a complex terminology, particularly for the North African finewares. This has created difficulties in equating the finds from Britain to those from Continental and Mediterranean sites. Waagé’s initial classification of the late red-slipped finewares found in the Athenian Agora (Waagé 1933) identified ‘Late A’, ‘Late B’ and ‘Late C’ categories. Although ‘Late A’ and ‘B’ had different fabrics they were later recognised to both be of North African origin and were grouped together by Hayes as ‘African Red Slip Ware’, following the term coined by Kathleen Kenyon at Sabratha (Hayes 1972, 2008, 67-8). Waagé’s ‘Late C’ ware, later termed ‘Phocaean Red Slip’ or ‘Late Roman C’ ware, is summarised below.

Since the publication of Hayes’ Late Roman Pottery (hereafter LRP) the British finewares have typically been classified using standard Mediterranean terminology. Full descriptions of the fabric and forms are provided in LRP and subsequent revisions (Hayes 1972; 1980; 2008). The main characteristics and features have been summarised below, focusing on recent understandings of these wares arising from research in the Mediterranean. Pottery from the Athenian Agora was the primary source for Hayes’ original chronology, supplemented by Waagé’s publications from Antioch for the later finds (Waagé 1948; Hayes 1972, 1). Since the publication of LRP, evidence from numerous excavations and surveys across the Mediterranean has allowed considerable amendments to Hayes’ dates for individual forms and variants (see Carandini et al. 1981; Fulford and Peacock 1984; Bonifay 2004). The urgent need to standardise descriptions and establish consensus on the typologies and chronology of these wares was highlighted at a specialist workshop held in 2008, leading to the creation of a new publication series – Late Roman Fine Wares in the Mediterranean (Cau et al. 2011b). General observations on the finewares (production, typology and dating), as summarised in the first volume (LRFW1), are noted throughout this chapter (Cau et al 2011c).

3.3.2 African Red Slip (ARS)

A separate classification system was developed by Lamboglia based on finds from Italy, who labelled these same North African products as ‘terra sigillata chiara’ (Lamboglia 1963). This class was divided by fabric and surface appearance (including the quality of the slip) into four major series, A, B, C and D, although the ‘B’ group was later recognised as a Gaulish product (Hayes 2008, 67). Lamboglia’s categories of T. S. Chiara A, C and D have persisted (often substituting the French ‘claire’ for Chiara) and were utilised, with minor re-labelling, in Atlante I (an encyclopaedia of Late Roman Mediterranean pottery) as the major North African series T.S. Africana A, C and D (Carandini et al. 1981, 12). A fifth century ‘E’ series was added, as well as smaller transitional series A/D and C/E (Carandini et al. 1981, 12). Atlante I incorporated Hayes’ sequence of African Red Slip Ware forms into a larger, expanded catalogue of fineware forms; these were classified into sub-groups of the major series based on fabric: A1-2, A/D, C1-5, D1-2 and E. Consequently, Hayes’ preferred term ‘African Red Slip Ware’, has typically been divided into these major series, here abbreviated as ARS-A, ARS-C, ARS-D etc. (Hayes 2008, 67-8).

The major late Roman fineware was produced in North Africa between the first and seventh centuries AD, initially in the region around Carthage, although production later spread south and inland across the region of modern

From the publication of Atlante I considerable efforts have been made to match these groups and sub-groups of ARS to specific regions of production in North Africa. The earliest series, ARS-A, dating to between the later first and

Further revisions arising from this ongoing research might impact on the broader chronology of the British early medieval imports. Nevertheless, it should be noted that there might be some lag between the assigned date-bracket for a form or variant close to its source, and at the furthest reaches of its distribution. Furthermore, these dates do not account for the use-life of the vessel, which might be extended in regions with limited or intermittent supply.

41

Links to Late Antiquity third centuries is generally considered to have been produced in North Tunisia in the region of Carthage, although a central Tunisian origin has recently been suggested (Cau et al. 2011c, 4). ARS-D is seen as the successor of this industry, representing Waagé’s ‘Late B’ category, and was produced in the Carthage region between the late third and seventh centuries (Reynolds 1995, 6). This was divided in Atlante into two fabric sub-groups: ‘D1’and ‘D2’ (Carandini et al. 1981, 78). Although no northern production sites have been identified for the earliest ARS products, sites producing the later Roman ARS-D series have been identified in North Tunisia, including a major centre at El Mahrine, which started production around the 320s and which has been associated with the ARS-D1 fabric (Mackensen 1993; 1998, 33, Cau et al. 2011c, 4, 5). The ARS-D2 fabric has been linked to the production centre discovered at Oudhna (the only ARS production site where kilns have been excavated) as well as the so called ‘Atelier X’ workshop (Hayes 2008, 68; Cau et al. 2011c, 4). An intermediate category C/D was added by Bonifay to classify the fineware from the production site at Sidi Khalifa (Bonifay 2004, 49). Production spread south from the Carthage region (the area of the later Roman province of Zeugitana), to Central Tunisia (Byzacena) with the start of the A/C series around AD 200 (Reynolds 1995, 6). The major central Tunisian series – ARS-C – was common from the third until the fifth century, although the end date has been extended toward the mid-sixth century (Mackensen 1998, 27; Hayes 2008, 69; Cau et al. 2011c, 5). ARS-C was divided in Atlante I into five sub-groups, ‘C1’ to ‘C5’ – the latest of which is characteristic of the second half of the fifth century but might date into the middle of the following century (Cau et al. 2011c, 5). LRFW1 noted, however, that these subgroups represent the chronology of the central Tunisian industry rather than always clearly distinguishable fabric groups, and recommended against their use where attributions were uncertain (Cau et al. 2011c, 5). Production of ARS-C has been identified in central Tunisia, most notably at Sidi Marzouk Tounsi and possibly El Djem (Mackensen 1998, 26-30; Cau et al. 2011c, 4).

Figure 3.7. ARS production zones and workshops in the region of Carthage, modern Tunisia. From Bonifay 2004, Figure 22

various subdivisions of ARS has increasingly been questioned. Reynolds described the ‘impossibility’ of matching forms to production areas based on the fabric or surface condition, and commented that plotting trends in the distribution of ARS using the ARS-A to E subdivisions was impractical (Reynolds 1995, 6, 12). At the same time production centres have been identified that made vessels which cannot be easily classified as ARS-C or D (Cau et al. 2011c, 4).

The source of the ‘E’ category identified in Atlante I has not been identified, but would seem to be in the coastal areas of ‘south-central’ Tunisia (Hayes 2008, 69). Production of ARS has also been identified further inland in southwestern Tunisia (Bonifay 2004, 51) but the distribution of these products is not directed towards the Mediterranean but inland to sites in the African interior (Bonifay 2004, 51). Figure 3.7 shows Bonifay’s map of the principal ARS production zones and known workshops.

Although the limitations of this system were highlighted by LRFW1, its general acceptance by Mediterranean ceramic specialists was noted, and it was recommended that until more is known about the North African production sites, it should be retained (Cau et al. 2011c,4). As many of the Continental finds of African Red Slip have been classified according to this structure, reference has been made to these sub-divisions in the discussions in subsequent chapters. However, to reflect these problems this project will focus on vessel form as the basic unit of comparison for ARS (as with the other finewares). The various regional series of ARS have, however, been used as a guide in establishing limits for data collection. While the presence of Late Roman C (Phocaean Red Slip) or Late Roman D (Cypriot Red Slip) in Atlantic contexts can

Despite these advances, understandings of the development and history of these production centres and their products remain incomplete. As a result, it is possible to suggest broad regional origins for the examples of ARS from Atlantic sites considered here, but it is not possible to categorically match the majority of them to individual production centres. Furthermore, the usefulness of these 42

Late Antique imported pottery on the Atlantic Seaboard: developments in research for the post-Roman examples in Britain fall within the ARS-D series and, according to current understandings, would have a general origin in the Carthage region of North Tunisia. At present, none of the British examples has been matched to specific production sites, or defined fabric groups, although the range of forms (including those attested to El Mahrine, Oudhna, Atelier X?) would suggest imports were not limited to a single production source.

be taken as evidence of fifth century or later importation, ARS was produced and exported in the preceding centuries. As a result, it was decided within this project to focus data-collection on ARS-D. Published examples of ARS that were not matched to dateable forms or to any of the regional series have not been used and are not included in Appendix K. Examples published only as ‘ARS-C’ have not been recorded, although the presence of ARS-C forms of potential fifth century date at certain Atlantic sites has been mentioned in the text. All examples of ‘ARS-D’ encountered have been recorded, with the caveat that these might include fourth century imports. However, published examples of identified later forms and variants of ARS-D have been presented within distribution maps Appendix H.1.2-3.

Forms The earliest ARS products, made from the first century AD, were closely modelled on forms of the Gaulish sigillata (Samian) industries, but by the second century the North African industry was producing a range of simpler forms, particularly broad dishes with low feet, that could be easily stacked for shipment (Hayes 1997, 59). Although the ARS repertoire of forms is more extensive than other Late Red Slip wares, the most common forms are dishes and bowls. The full sequence of development is outlined in LRP (Hayes 1972, 15-8). General trends include: an increase in the diameter of dishes through the fourth and fifth century, a reappearance of ‘ring-footed’ dishes and plates from the fifth century (including Forms ARS 88 and 103-104), and the production of small bowls with ‘a high flaring foot and sagging floor’ (including ARS 99C) and flanged bowls (ARS 91) in the later-fifth and sixth century (Hayes 1972, 16). Hayes’ classification has remained the primary system for identifying dated vessel-types and variants, although additional forms and sub-forms were subsequently presented in Atlante I and in the publication of the British excavations at Carthage (Fulford 1984a).

Post-Roman imported examples of ARS were first identified in Britain following excavations at Tintagel, Cornwall and were grouped together with examples of Late Roman C ware under Radford’s ‘Class A’ (Radford 1956, 61-2). An account of the history of the identification of these Red Slip finewares in Britain and their concordance with Mediterranean classifications is provided in Subchapter 3.3.3. Description In general, ARS is coarser, paler and less glossy than earlier sigillata wares (Hayes 1972, 13). The fabric is typically granular, with common quartz and lime inclusions, but no mica (Hayes 1972, 14; 2008, 68). The colour is typically orange-red to brick-red; vessels are covered wholly or partly by a similar-coloured slip (Hayes 2008, 68).

Examples of ARS published from Mediterranean and Continental sites are typically matched to Hayes form numbers (open red-slipped vessels comprise the sequence ARS 1 to 111), although other classifications are also used if there is no suitable Hayes form. A number of vessels at Vigo, for example, were identified using Fulford’s sequence from Carthage (1984a) which included forms and variants not covered by Hayes’ typology.

ARS-C has a fine fabric, with a thin-slipped, smooth surface and was commonly made in moulds (Hayes 2008, 69). This series frequently features relief decoration. ARS-D is described as rather more robust with a granular clay (Hayes 2008, 68) and is typically slipped only on the inside of the vessel and onto the rim (Hayes 1997, 59). ARS-D1 is typified in LRFW1 as ‘clean breaking’ with few inclusions and a matt slip, while ARS-D2 has a coarser-grained fabric and a glossier slip (Cau et al. 2011c, 4). Bonifay has proposed additional ARS-D fabrics, D3-D5, based on vessels from Marseille and Reynolds’ observations at Alicante (Bonifay 2004, 48). Hayes commented, however, that not all examples fall easily into these fabric categories, reflecting the regional complexity and dispersed nature of ARS production (Hayes 2008, 68). Similarly, Bonifay has questioned whether these traditional categories remain appropriate for current research on ARS production and distribution (Bonifay 2004, 48).

Decoration The earliest examples of ARS are not decorated apart from occasional rouletting. Between the third and fourth centuries, ARS vessels categorised as ARC-C commonly bore applied decoration, including figural motifs (Hayes 2008, 70). From the fourth century stamped decoration appears on the interior base of dishes and large bowls (Hayes 2008, 70). Hayes has defined a chronological sequence for these stamps, commencing with repeated ‘geometric and floral’ motifs in the period c. 325-440 (his styles A-C). From the mid-fifth century these are replaced by Christian motifs of ‘cross-monograms’ and animals which are confined to a narrow band on the base (style D). By the sixth century, larger Christian motifs are used, including human figures and large crosses (Style E) (Hayes 1972, 212-22; 2008, 70). The detailed chronology of ARS stamp-decoration

Campbell noted that the ARS sherds recovered in Britain often have a degraded fabric and the slipped surface may not survive (Campbell 2007a, 18). As only very limited observations could be made about the surface appearance, it would be extremely difficult to match the few British examples to the regional series and fabric divisions distinguished in North Africa. Nevertheless, the forms identified 43

Links to Late Antiquity first presented in LRP has, in association with the dated forms, permitted a tight dating structure for these wares, which has consequently assigned this ware a key role in dating deposits at consumption sites.

It is not possible to discuss the dating issues of each form present on the Atlantic Seaboard, but the various datebrackets indicated for the more common forms given in Appendix C attest to the complexity of this issue.

Roulette decoration also appears on some vessels, including the distinctive ‘feather rouletting’ that is a feature of ARS from the late-fourth to sixth century, and particularly on the flanged bowl form ARS 91. Although Hayes suggested feather rouletting represented a ‘conscious effort to reproduce silverware decoration’ (1972, 282), he has also repeated the suggestion that as this form bears strong similarities to earlier Roman mortaria, the decoration might have had a use during dining – possibly to grate condiments or garlic (Hayes 2008, 71). Variants of Form ARS 91 have been identified in post-Roman assemblages in Britain, although it is unclear whether these, or any of the imported finewares, would have been used for dining.

British examples of ARS Small quantities of ARS have been identified from Roman-period contexts in Britain, although this was outside of its standard Mediterranean distribution. These examples were described by Bird (1977) and catalogued using Hayes form numbers. She notes that the distribution is widespread but small, and dates from the later first century to the late-fourth or early-fifth century (Bird 1977, 272). The latest of these are: two sherds from a bowl with applied motifs – Form ARS 52B (of the central Tunisian ARS-C series) – from Old Ford, London, which is dated in LRP from c. 280/300 to the late-fourth century but with late variants into the early-fifth century (Hayes 1972, 78), and one rim-sherd of ARS 67 from Southwark [*B], London (Bird 1977, 275). The large bowl ARS 67 (of the ARS-D regional series) is dated in LRP to c. 360-470. The Southwark example is stated to be from a late-fourth century context and does seem to fit early in the series of variants established by Bonifay, suggesting it was a probable ‘Roman period’ arrival (Bird 1977, 275; Bonifay 2004, 171-3). Bird also suggests a heavily rouletted base-sherd from Cirencester is of a similar late African fabric (Bird 1977, 275). A North African lamp found at Silchester is also taken to represent a comparable late Roman arrival (Bird 1977, 276; Fulford 1977, 56).

Chronology ARS was produced from the first century and became increasingly well distributed in the western Mediterranean from the second century– where it reached its zenith between the fourth and fifth centuries. In the East, Late Roman C ware (LRC) and Late Roman D ware (LRD) took over the market from ARS in the fifth century. Broader patterns in the distribution of ARS throughout the Mediterranean and specifically into the Atlantic are discussed in Sub-chapter 6.2.3. The production of ARS seems to have lasted into the last decades of the seventh century (Bonifay 2004, 207), with very late exports reaching the Crypta Balbi in Rome up to c. AD 700 (Saguì, 1998, 309; Hayes 2008, 82). Hayes has suggested that localised production continued into the eighth century (Hayes 2008, 82). Although some of the most common ARS forms could be relatively long-lasting, only a few forms remained in production for more than a century without ‘drastic modification (Hayes 1972, 14). As a result, these forms – and more significantly the individual variants – have been used to provide a relatively tight and consistent dating structure for assemblages across the Mediterranean and beyond. The dates suggested by Hayes for the production of individual forms have been continually questioned and refined since 1972, for example at Carthage (Fulford 1984a), in the West Mediterranean (Reynolds 1995) and following Bonifay’s research in North Africa (Bonifay 2004). Nevertheless, LRP has been heavily relied upon for dating finds from consumption sites, which was criticised by Hayes himself (1998, 9). LRFW1 commented that the typo-chronology in LRP required ‘serious revision’ and offered a summary of the suggested corrections for dates of common forms (Cau et al. 2011c, 5). These dates have been preferred for the dating of Atlantic examples discussed in subsequent chapters. This ongoing process of revision in the Mediterranean will doubtless have further impact on the chronology of the imported finewares in British and other Atlantic contexts, and therefore on the wider chronology of the Atlantic exchange system.

These earlier examples of ARS in Britain have been interpreted as possessions that might have arrived in Britain with their owners, rather than as traded items, and as revealing individual mobility rather than long-distance exchange systems (Bird 1977, 272). As discussed, the publication of the imported fineware from Tintagel marked the first recognition that North African pottery was imported to post-Roman Britain. These later finds have, by contrast, typically been considered to represent traded commodities. Therefore, the arrival of ARS vessels in Britain is traditionally split into two distinct phases of Roman and post-Roman imports (both comprising small quantities of vessels), with a definite gap in the chronology between the late-fourth/early-fifth century and the later-fifth/sixth century. Within the scope of this project it has not been possible to re-examine individual British examples of ARS and the identifications for the post-Roman examples provided in Campbell’s synthesis have largely been followed (Campbell 2007a, 2011). Where illustrations have been published some variants have been suggested (see Appendix C), but it should be noted these were not based on examination of the sherds. Campbell’s ARS data comprised revisions and additions to identifications originally made by Charles Thomas (1959; 1981). Although Thomas’ 1981 catalogue of imported pottery matched the British examples of ARS to the typology in Hayes’ LRP, Campbell utilised the number sequence from the British 44

Late Antique imported pottery on the Atlantic Seaboard: developments in research for ‘post-Roman’ examples of ARS in Britain is 35 vessels from ten sites (36 vessels including the ARS 67 from Southwark).

excavations at Carthage (Fulford 1984a). He applied the dates suggested for the ARS forms at Carthage to suggest a chronology for the imports to Britain, concluding that all of the British examples could be dated to the sixth century (Campbell 2007a, 18). Although the forms themselves have longer date ranges, the overall composition of the British assemblage was seen to match the specific pattern of ARS at Carthage in the period c. 525-550 (Campbell 2007a, 18). The absence of later-seventh century forms such as Fulford 67 and 68 (Hayes 106 and a possible associated form) was seen to corroborate this general view (Fulford 1984a, 74; Campbell 2007a, 18).

To enable clearer comparisons between the British finds and the Mediterranean and Atlantic regions where Hayes’ sequence is in most common use, it was appropriate to reclassify the British examples using Hayes form numbers. Forms that are not covered by Hayes’ sequence have retained alternative classifications. The forms identified by Campbell are: ARS 88, 91C, 96, 99, 103, 103B, 104, 104B as well as ‘Fulford forms’ 42, 50 and 52 which cannot be directly matched to Hayes forms. Some of those classified as 50 and 52 are likely to equate with forms E.M. 18 or E.M. 24 from El Mahrine (Mackensen 1993, Tafel 62, 65), but it is recommended that this group – which includes sherds from Tintagel for which illustrations have never been published – should be subject to a full reassessment. In this, and subsequent chapters any form numbers for ARS (as well as LRC and LRD) refer to Hayes’ typologies, unless specified otherwise.

Although Campbell noted that no individual vessel need be confined to this narrow date range, the ‘overall pattern’ suggested the importation of ARS within this approximate 25-year window (Campbell 2007a, 18). Significantly, Campbell concluded that the ARS imports were therefore slightly later to arrive than the ‘Late Roman C’ vessels. This was developed further and used to date a so-called ‘African package’ of imports, comprising ARS and North African amphorae, said to have succeeded the phase of imports from the East Mediterranean arriving between c. 475-525 (Campbell 2007, 18). Bonifay has, however, suggested that all of the imported fineware forms could easily be contemporary arrivals in the first half of the sixth century (Bonifay 2008, 1123); this would eliminate the need for a model of importation with successive ‘Aegean’ and ‘African’ phases.

As noted by Campbell (2007, 18) the only sherd of ARS from Britain with an identifiable stamp is from CadburyCongresbury. This stamp of a ‘jewelled cross’ on the base of a Form ARS 103B (a large shallow dish) was classified in the site pottery report as Hayes’ style Eii, which is dated in LRP to c. 530 to 600 (Hayes 1972, 222; see Figure 3.8). For Campbell, this supported his assessment of the British ARS as dating to the second quarter of the sixth century (Campbell 2007a, 18).

It does seem that Britain was not receiving ARS in the middle decades of the fifth century (the possible exception would be the vessel from Pevensey discussed in Sub-chapter 4.4.1). Instead, the forms present at postRoman sites point to their arrival (rather than, necessarily, their deposition) in the later-fifth and, particularly, sixth century. It is also apparent that very late-sixth or seventh century forms of ARS are not present among the British or Irish examples. One possible exception is a sherd from Iona identified as Hayes 104B (see Sub-chapter 6.3.2). Although Campbell did not specify the variants for the other examples of 104, most would appear to match 104A – thereby confirming the generally early 6th century chronology. The only other potentially late Insular ARS is the example from Pevensey identified as ARS 99C.

3.3.3 Late Roman C (LRC) History of Research By the later fourth century AD, competing red-slipped fineware industries were established in the East Mediterranean, which, over the course of the fifth century, increased their distribution at the expense of ARS. This is linked to contemporary shifts in Mediterranean trade networks and, specifically, the growth of Constantinople as a market (Hayes 1997, 62). The two major competitors for ARS in the East Mediterranean have been classified as Late Roman C Ware (LRC), which is also known as Phocaean Red Slip Ware, and Late Roman D Ware (LRD), which has previously been termed Cypriot Red Slip Ware. Between the fifth and seventh centuries LRC was widely distributed throughout the eastern Mediterranean (Hayes 1972, 323), but by the later-fifth century was increasingly distributed in the West Mediterranean as part of a widespread expansion of eastern goods (Reynolds 2010, 105-6).

Thomas’ 1981 catalogue recorded sixteen examples of ARS from Britain and Ireland. Campbell’s 2007 synthesis recorded a minimum of 32 vessels of ARS from ten sites in Britain and Ireland, although approximately half of these were from Tintagel (Campbell 2007a, 18). His 2011 dataset included two further vessels, representing the discovery of possible sherds of ARS at Lellizick, Cornwall [113] and a possible example from Kilree in Ireland (Wessex Archaeology 2008; Campbell 2011). The latter example was, however, subsequently discounted as ARS and awaits a revised identification (Kelly pers. comm.). To this total can be added three vessels recently identified at Pevensey Castle [108], which will be discussed in full in Chapter Four. Therefore, the current suggested total

LRC was originally categorised by Waagé at Athens based, principally, by the distinctive shape of vessels that would later be classed as Hayes form 3 (Waagé 1933, 298-9). The origin of these vessels had not been identified by the 1972 publication of LRP prompting Hayes to retain Waagé’s label (Hayes 1972, 323). Hayes, nevertheless, identified a probable origin in ‘Asia Minor’, suggesting that this 45

Links to Late Antiquity

Figure 3.8. Reconstructed ARS 103B with cross-stamp from Cadbury Congresbury. From Rahtz et al. 1992, Figure 116

industry was a continuation of second and third century ‘Eastern Sigillata’ production in the region of Candarlı close to Pergamon (Hayes 1972, 316, 323).

wasters, ‘moutons’ of fused vessels and kilns (outlined in Mayet and Picon 1986, 129). Mayet and Picon’s research confirmed Phocaea as the major source of this ware by chemical analysis, comparing samples taken from vessels identified as LRC across the Mediterranean with wasters collected from the production site. Significantly, they used the identification of the production centre as an opportunity to confirm the western distribution of this ware, recording find-spots in Italy, southern France and Spain and on the Atlantic Seaboard, including Britain, although they did not specifically analyse any samples from Atlantic sites (Mayet and Picon 1986, 130-2). This development is discussed in more detail in Chapter Six.

By the time Hayes’ Supplement was published in 1980 production sites in western Turkey had been identified at Phocaea (Foça) in north-western Turkey, leading him to re-label the ware as ‘Phocaean Red Slip’, thereby abandoning the last of Waagé’s alphabetical classifications (Hayes 1980, lix) Production wasters had first been reported from Phocaea in 1969 (Langlotz 1969, 381, Figs 4-6) but were not immediately identified as Late Roman C. Subsequent fieldwork identified a production centre with numerous 46

Late Antique imported pottery on the Atlantic Seaboard: developments in research Subsequently, a secondary production site was identified at Grynion, north of Phocaea/Foça, which produced sherds of the earlier ‘Çandarlı ware’ as well as LRC, suggesting some degree of continuity of production (Empereur and Picon 1986, 143; Hayes 2008, 84). It was initially suggested that products from Grynion were not represented among the Mediterranean samples studied by Mayet and Picon, and may therefore have only had a local distribution (Empereur and Picon 1986, 144). This localised distribution was reiterated by Vaag’s study of LRC ‘main and secondary production’ which also discussed the presence of ‘imitation’ LRC vessels at a range of sites including Sardis, Emporio and Anemurium and possible LRC production at Pergamon (Vaag 2005, 133-6). Hayes, however, has suggested that a few of the LRC vessels from the Athenian Agora might be Grynion products (Hayes 2008, 84). The recent LRFW1 listed the sources of ‘a large percentage of long-distance exports’ as Phocaea/Foça, Gryneon (Grynion), Çandarlı and another ‘unidentified product’, and also noted ‘satellite’ centres with localised distribution at Ephesus and Pergamon (Pergamum) (Cau et al. 2011c, 6). Further research is necessary to fully understand the production of vessels in this region of north-western Turkey and to ascertain the relationship between individual production sites and fabrics and forms that match the established descriptions of LRC, and as a result, it remains to be fully established whether the British and Irish finds of LRC represent a single production source. As a reflection of this uncertainty specialists have increasingly reverted to using the original designation Late Roman C (LRC) (as in Cau et al. 2011c, 6), and this example has been followed here. History of Identification in Britain

Figure 3.9. Radford’s ‘Ai’ fineware subgroup (top) and ‘Aii’ (bottom) from Tintagel. From Radford 1956, Figures 13 and 14, not to scale

In Britain, examples of LRC and (to a lesser extent) ARS formed the original collection of finewares from Tintagel that were described by Radford (1956) and published as his ‘Class A’. This category was sub-divided by Radford into ‘two varieties of fine red ware’, ‘Ai’ and ‘Aii’, based on fabric distinctions, particularly hardness, and on the presence of stamped crosses on vessels within the Ai group (Radford 1956, 60-2). In LRP Hayes linked Radford’s Aii with Waagé’s Late B and C (therefore with both ARS and LRC) and incorrectly described Ai as a ‘derivative ware’ not identified elsewhere (Hayes 1972, 7). This seems to be an attempt by Hayes to reconcile Radford’s illustrations with Charles Thomas’ initial concordance, the latter of which equated Aii with either Late Roman B or C and described Ai as connected to it in some ‘unspecifiable way’ (Thomas 1959, 90).

the British finewares to Mediterranean classifications, as conducted by Bernard Wailes, was never published (Wailes 1963). Wailes realised that Radford’s categorisation of the finewares on the basis of hardness was not reliable, and – following a re-examination of the sherds – re-classified a number of the Aii sherds as Ai. This allowed him to directly equate Ai with LRC and Aii with ‘Late Roman B’ (ARS) (Wailes 1963, 101-6). Nevertheless, it seems that Wailes’ reassessment was followed by Thomas in his later publications (i.e. Thomas 1993, 62). Appendix A shows the concordance of these terms. Radford was uncertain as to the origin of Ai, suggesting Egypt as a possibility, while, intriguingly, noting the similarity of the Tintagel sherds to stamped pottery recovered close to Oudna (Oudhna) near Tunis and at Carthage in North Africa. In contrast, the Aii group, which did in fact contain vessels of North African origin, was assigned a probable Gaulish or Spanish origin (Radford 1956, 65-6). An East Mediterranean origin for both groups of fineware was initially proposed by Charles Thomas, who nevertheless retained Radford’s ‘British’ terminology (Thomas

Radford’s illustrations from this first published classification, however, show clear examples of Hayes form 3 (LRC 3) within both sub-categories (see Figure 3.9). Campbell recently resolved this confusion, noting that although Radford’s Ai and Aii do not equate entirely with ARS and LRC, Radford’s Ai group generally comprises LRC, while Aii ‘corresponds partly’ with ARS (Campbell 2007a, 1416). It is regrettable that the first major attempt to equate 47

Links to Late Antiquity 1959, 91). However, it was the publication of LRP in 1972 that finally confirmed the respective North African and East Mediterranean origin of ARS and LRC. After the publication of Thomas’ Provisional List in 1981 British researchers have generally followed Hayes’ terminology of African Red Slip and Phocaean Red Slip wares, although the concept of ‘A wares’ as an inclusive term for both post-Roman imported finewares has lingered in British publications (e.g. Hollinrake 2007, 337).

Chronology Hayes provided no major revisions to his date sequence of LRC in his 1980 Supplement, but subsequent research has had implications for the dating some of these established forms and variants. Although the production of LRC is thought to have developed from an earlier tradition of sigillata production in north-west Turkey, the fourth century transition between these industries remains unclear (Hayes 2008, 84). In LRP Hayes dated the emergence of the first LRC product, LRC1, to the later fourth century (Hayes 1972, 327), although more recently assessed assemblages, including groups from the Athenian Agora, have suggested a later, fifth century starting date (Cau et al. 2011c, 6). Hayes notes that imports arrive at Athens much more regularly from c. 390-400, coinciding with a drop in ARS imports, before the major period of importation during the ‘middle phases of the ware’ between c. 450-550 (Hayes 2008, 85). This period corresponds broadly with the maximum distribution of this ware. The latest phase of LRC production is also unclear; the ware is certainly produced and exported until the mid-seventh century, but the possibility of later-seventh century production remains an area of debate (Hayes 2008, 86; Cau et al. 2011c, 7).

Description The description of LRC provided in LRP remains basically unchanged, although the reported uniformity and consistency of the fabric in the vessels assessed by Hayes might obscure greater variation closer to the production region. Extensive descriptions are provided in Hayes 1972, 1997 and 2008. The fabric of LRC is hard and fine-grained, comparing well to ARS-C (Hayes 2008, 84). The clay is usually ‘brick-red to maroon or purple-red’, although Hayes notes that occasional poorly fired vessels can be orange (Hayes 2008, 84) and Campbell has described burnt grey sherds within the British examples (Campbell 2007a, 14). A thin, often dull slip, usually slightly darker than the clay, covers the entire body. Harder fired examples can have a metallic appearance, or be speckled with particles of lime (Hayes 2008, 84). The walls are usually thin but become increasingly thinned towards the middle of the vessel, while the rims are commonly discoloured or blackened, indicating that they were stacked during firing (Hayes 1997, 64). Hayes also notes the presence of ‘spatula marks’ on the smoothed interior of the vessels and scratch marks on the exterior which might indicate the use of moulds during manufacture (Hayes 2008, 323-4).

The specific typological sequence of Hayes form 3 (comprising Forms 3A to 3H) remains an issue of major current interest, as discussed in the recent LRFW1 (Cau et al. 2011c). This review highlighted the difficulties in matching individual sherds to the published variants and their assigned dates and, significantly, questioned the validity of this detailed sequence. Although the form clearly develops across its span of production, it was suggested that some of the variants might represent parallel production at ‘several contemporary workshops’ (Cau et al. 2011c, 6), which reflects the new understandings of the complexity of the LRC industry. This difficulty in identifying Form 3 variants is exacerbated in British and Irish contexts because of the typically abraded and fragmentary condition

Forms The range of forms classified by Hayes is small (LRC1 to 10), and the overall repertoire is restricted in comparison to ARS and limited to dishes and open bowls (Hayes 1972, 325-46; 2008, 84). However, some of the forms, particularly LRC 3, show high levels of stylistic development across extended lifespans and, as a result, a range of dated variants with distinctive rim shapes have been identified (see Figure 3.10). Hayes’ classification of these variants (following Waagé) has consequently had great significance in establishing chronological markers for sites with LRC across the Mediterranean and into the Atlantic. The archetypical shape for LRC is Hayes form 3 (LRC 3), a shallow dish/bowl with a vertical projecting rim (Hayes 2008, 84). Its typological development of this form is characterised by Hayes as ‘a progressive shortening and thickening of the rim with a corresponding increase in the overhang at the bottom’ (Hayes 1972, 329). LRC 3 superseded LRC 1 and 2 in the second half of the fifth century and evolves into LRC 10 by the mid to later-sixth century. The latest form produced in LRC, LRC 10 – a dish/bowl with a more flattened/knobbed rim – typified production until the end of this industry (Hayes 1972, 343; see Figure 3.10).

Figure 3.10. Hayes’ LRC Form 3 variants C, E and F and Form 10. After Hayes 1972, Figures 67-9, not to scale

48

Late Antique imported pottery on the Atlantic Seaboard: developments in research Table 1. LRC forms and chronology Hayes form

Variant

LRP date

Revised Date

All LRC forms

Late-4th century to mid-7th century.

Early-5th century to later-7th century

1

Late-4th century to early-5th century

Early-5th century+

2

c. 370 to 450 (not after mid-5th century)

c. 400?+

3

c. 450 to 550

c. 400?+ Comprehensive publication of LRC 3 series stated to be a priority in LRFW1

3B

c. 460 to 475

3C

c. 460 to 475

3D

Late-5th century

3E

Around 500

3F

6th century c. 525 to 550?

c. 520 to 550

3G

6th century c. 525 to 550?

Considered fully in Fernández 2014, 353 and dated to second half of 6th century at Vigo

3H

6th century

5

c. 460 to 550 5A

c. 460 to 500

5B

c. 500 to 550

6

Early-6th century

8

Second half of 5th century

10

c. 570 to 660+ A-B

Late-6th to early-7th century

C

Early to mid-7th century

c. 550 to later-7th century?

of the sherds (Campbell 2007a, 14). The ‘Atlantic’ examples that will be discussed are typically published within a hierarchy of identification – many solely as LRC, some of these to individual forms (typically Form 3) and thereafter, where possible, to specific variants. As a result of these various issues the identified variants should be seen to provide guides for the chronology of the imports rather than as representing fixed datable markers. It is therefore possible to assess broad trends in the distribution of LRC across the Atlantic region, but not to map the precise distribution of specific, closely-dated variants from sites on the Western Seaboard.

typically considered to have been imported beyond this date (Cau et al. 2011c, 6). Table 1 shows the main exported forms and more common variants of LRC as classified in LRP with suggested dates of manufacture (from LRP, with suggested revisions following Hayes 2008; Cau et al. 2011c; Quaresma and Morais 2012 and Fernández 2014). Decoration LRC vessels are commonly decorated with rouletting, particularly on the outside of the rim. Hayes notes that this rouletting is distinctively varied and can comprise a varying number of rows of impressed marks made simultaneously (Hayes 1997, 64). The larger LRC vessels are also typically decorated with stamps on the interior of the base, although the earliest product, LRC 1, is undecorated and in general stamps become less frequent from the mid-sixth century. The later versions of LRC 10 are also usually undecorated (Hayes 1972, 346). Hayes describes the stamps as having a ‘simple impressed outline’ and a ‘childish quality’ (Hayes 1997, 64) and has identified three main phases of decorative styles (described in Hayes 1972, 346; 2008, 84). The earliest phase, Hayes’ Group 1, replicates

Variant G of Form 3 was not illustrated in LRP but has been subsequently considered by Fernández Fernández based on LRC found at Vigo in north-west Spain (see Sub-chapter 6.3.2). He describes LRC 3G as having a low, characteristically convex and chunky rim which he considered to represent the latest in the Form 3 sequence, and the transitional shape between Form 3 and Form 10 – although he also identified a transitional category of LRC 3/10 (Fernández 2014, 245). Both were relatively common at Vigo. LRFW1 highlighted the importance of variant 3G for identifying mid-sixth century deposits; this might have future significance for Britain as finewares have not been 49

Links to Late Antiquity the ‘floral schemes’ of ARS produced between c. 350430. Group 2, focused on the fifth century, parallels the repeated animal motifs and crosses of ARS style D (dated rather earlier at c. 440-500+ in LRP), which are usually within bands of rouletting or grooves. The third, typically sixth century style, features individual, central stamps of animals or crosses. The few stamped sherds from Britain were originally tied to Hayes’ stamp typology by Thomas (1981, 7-8) and subsequently corroborated and/or revised by Campbell (Campbell 2007a, 14). Campbell used the identifiable stamps, in conjunction with the forms, to argue for a restricted date-bracket for the importation of LRC to Britain. The small number of British stamped sherds have been assigned to Hayes’ groups 2 and 3 (as shown in Table 2), therefore fitting broadly within a later-fifth to midsixth century window – or more specifically, according to Campbell, to the period 500 ± 25 (Campbell 2007a, 14).

British and Irish examples of LRC It has not been possible within the scope of this project to re-examine all previously identified British and Irish examples of LRC, and therefore, as with the ARS, Campbell’s classifications have largely been followed (Campbell 2011). The exception is for the Irish imported finewares where Doyle’s totals have been followed (Doyle 2009, 47). According to Campbell’s 2007 synthesis, sherds representing a total of 62 vessels of LRC have been recovered from Britain and Ireland. Of those that could be identified, all were matched to LRC 3 and some to specific variants within the range of 3C-3F (Campbell 2007a, 14-6). Campbell’s 2011 dataset records 66 vessels, following the inclusion of unpublished, potential examples of LRC from Hay Close, St. Newlyn East, Cornwall [119] (Campbell 2011), Woolston, Somerset [105] (Campbell 2011; Costen 2011, 56; site described in Tabor 2007) and by the addition of a recently identified sherd from Collierstown, Meath [51], in Ireland (Kelly 2010). However, the total figure for LRC can be slightly reduced. Campbell had recorded two vessels from Garranes and three from Bantham, but Doyle’s 2009 catalogue of Irish examples indicated only one certain LRC vessel from Garranes (Doyle 2009, 47), while the recent Bantham publication recorded only two definite LRC vessels (Bidwell et al. 2011, 106). As a result of these minor revisions an approximate total figure of 62-64 vessels is suggested for Britain and Ireland (see Appendix K), although there are a few further, tentative identifications. However, this has no major implications for the general assessment of LRC as a category amongst these imports. British and Irish finds of LRC represent a relatively small number of vessels that have had major significance for the overall interpretations of this post-Roman exchange system.

Although further fragments of stamps have been recorded from Tintagel, including another possible cross, part of a human head and part of a peacock tail, these have not been identified with certainty or matched effectively to datable decorative groups (Thomas 1981, 7-8). However, the dating structure of the stamps is dependent on their association with identified LRC variants and, as discussed, there have been revisions to this typo-chronological sequence since the publication of LRP – based, in part, on newly identified key assemblages in the East Mediterranean and the Black Sea (Cau et al. 2011c, 6). These revisions might impact on the dating of the British stamped sherds. As Hayes has suggested that LRC 3 might have originated closer to the turn of the fifth century (Hayes 2008, 85) it might be that the Group 2 stamps associated with early variants (A-C) of LRC 3 consequently belong earlier in the fifth century. The stamped vessel from Dinas Powys, identified as LRC 3C by Thomas and 3C/E by Campbell (Thomas 1981, 7; Campbell 2007a, 14), might, therefore, have arrived slightly earlier in the fifth century than previously considered. Such revisions will only be possible following the reassessment of the Form 3 typological sequence, as recommended in LRFW1 (Cau et al. 2011c, 6). Although sherds of LRC with roulettedecoration have been recovered in Ireland, no stamped sherds have been identified (Kelly 2010).

Campbell, citing a distribution map in LRP (Hayes 1972, 464), states that apart from some isolated examples of later forms at Mediterranean sites such as Marseille, LRC was only exported outside of the East Mediterranean during the span of production of LRC 3 (Campbell 2007a, 14). The date range of LRC 3 would therefore define the date for the exportation of this ware to Britain. Reynolds has stated that LRC started to appear at sites in the western Mediterranean from c. 450, but that LRC 2 and early vari-

Table 2. Insular examples of published LRC stamps. After Campbell 2007a, 14; Thomas 1981, 7-8 Site

Figure no.

Form

Stamp

Hayes stamp number/group

Date

Dinas Powys

Alcock 1963, 127

LRC 3F? (Thomas has 3C, Campbell has 3C/F)

Repeated motif of five+ running or leaping animals. These were interpreted as ‘felines, possibly leopards’ by Alcock (1963, 128) but were described as hares in Campbell 2007a, 14

35, Group 2

Before 500

South Cadbury

?

Cross

79 or 68?

Late-5th/mid6th century

Tintagel

?

Cross Stamps from two Vessels

71/72, Group 3

Late-5th/early6th century

?

Fragment of Peacock tail?

Group 3

50

Late Antique imported pottery on the Atlantic Seaboard: developments in research Hayes that it must have been produced on the island and should therefore be re-labelled ‘Cypriot Red Slip Ware’. Although a production site has never been identified on Cyprus, Hayes’ designation has remained in standard use. More recent research has, however, questioned the assumed Cypriot origin of this ware. The identification of seven production centres with associated kilns in the region of Gebiz, near Antalya in Southern Turkey (close to the ancient city of Pednelissos) has suggested that LRD was partly or entirely produced in this region of southern Anatolia (Jackson 2012).

ants of LRC 3 are largely absent (Reynolds 1995, 35). Later variants of LRC 3 (C-H) dominate the subsequent distribution of this ware in the later-fifth and early-sixth century. Campbell, however, comments that the later variants of 3G and 3H, dating to the mid to late-sixth century, are ‘notably’ absent from the British assemblages (Campbell 2007a, 14). Reynolds also states that the latest forms of LRC have a very restricted distribution in the West, with LRC 10 present only at Marseille and sites in southern-eastern Spain and, at the very limits of this distribution, in Britain (Reynolds 1995, 35). The identification of the latest variant of the final LRC form at Tintagel (which is dated well into the seventh century) would therefore have major implications for the chronology of the British imports. Thomas identified three vessels of LRC 10 at Tintagel (Thomas 1981, 6), although by 1988 he had concluded that the LRC (‘PRSW’) was all of Form 3 (Thomas 1988a, 12). Campbell later reclassified these vessels as LRC 3E/F, commenting that Thomas’ identification had ‘entered the literature’ on the British imports and obscured the ‘tight’ chronological range of the British imports (Campbell 2007a, 14). The presence of LRC 10 amongst the material from Tintagel has indeed persisted as a feature of very recent summaries of LRC distribution (as in Reynolds 2010, 108). Although the report from the 1990-99 excavations at Tintagel confirmed that LRC 3 was the sole form identified from the site (Thorpe 2007, 234), the absence of a full illustrated catalogue of the Tintagel fineware showing forms and variants, has allowed the phantom presence of LRC 10 to persist (see Sub-chapter 4.2.3). Campbell’s rejection of LRC 10 within the British assemblage has been respected throughout this study and, as a result, the latest forms of LRC in Britain and Ireland are not currently considered to date past the mid-sixth century. Nevertheless, recent evidence from the Atlantic seaboard has revealed the tantalising presence of later LRC forms and variants, produced in the later-sixth and into the seventh century. This new evidence and the implications for the chronology of the Atlantic trade are discussed in Chapter Six.

This recent survey in south Turkey has revealed a complex picture of scattered fineware production sites producing vessels within a family of forms fitting the classification of LRD. Large quantities of production wasters have been collected as well as fused ‘moutons’ comprising multiple forms fired simultaneously (Jackson 2012, 109). Naturally, such information on forms produced contemporaneously, could affect our understanding of the chronological development of LRD. Significantly, each of the forms classified by Hayes has been identified at these production sites, as well as supplementary forms and variants (Jackson 2012, 96-109). In recognition of the continuing uncertain origin of vessels classified as ‘Cypriot Red Slip’, it has been suggested that the ware should revert to its previous name of ‘Late Roman D Ware’ (Jackson 2012, 90). This change in terminology was also employed in LRFW1 and has therefore been followed in this publication. Description Hayes’ descriptions of the fabric, forms and variants of LRD has remained in use throughout the Mediterranean and formed the basis for identifying this ware. The fabric is described as fine-grained and smooth, with occasional lime that can rupture the surface (Hayes 2008, 89). Hayes describes the fabric as identical to the earlier Roman ‘Cypriot Sigillata’, although the slip is ‘thinner and less glossy’ (Hayes 2008, 89). No direct link has, however, been established between these two industries. As with LRC a slip was usually applied to the whole surface, apart from on some of the larger ‘basin’ forms (Hayes 1972, 371). The colour of the body and slip are typically dark-red to maroon, although of all the major Red Slip finewares LRD is distinctive for its wide range of colours (Hayes 2008, 89). Jackson has connected the variability of fabric and colour in LRD to the relatively simple kiln technology utilised in its production (2012, 110). As with LRC the vessels are frequently discoloured on the rims suggesting stacking during firing (Hayes 1972, 371).

3.3.4 Late Roman D (LRD) The last of the three major Late Roman Mediterranean Red Slip finewares is Late Roman D Ware, which has previously been termed Cypriot Red Slip Ware. Like LRC this ware was transported in volume across the East Mediterranean but, more specifically, has been identified in large quantities at sites on Cyprus, in Asia Minor and the Levant and is generally rarer in the Aegean and the West Mediterranean (Hayes 2008, 89).

Forms History of Research As with LRC, the LRD repertoire is rather restricted, and is characterised by distinctively low or flat-based dishes produced by hollowing out the base rather than adding a true ‘foot-ring’ (Hayes 1972, 372; 1997, 64). The forms are noted to be less standardised than ARS, and are also less discrete and tend to merge into each other (Hayes 1972,

Waagé’s original classification of late imported finewares at Athens comprised Late A, B and C, but by 1948 he had identified a ‘Late D’ category (Hayes 1972, 5; Waage 1948, 52). By the publication of LRP in 1972 the large quantities of ‘Late D’ recovered from Cyprus had convinced 51

Links to Late Antiquity 372). This was reflected at the production sites in southern Turkey where ‘transitional forms’ were frequently recognised (Jackson 2012, 96). The majority of LRD forms fit within three broad groups of small bowls, large basins, and, most commonly, medium-sized dishes (Hayes 1972, 372). The most common forms are the dishes LRD 2 and its successor LRD 9; the transition between these forms was placed close to the mid-sixth century by LRP (Hayes, 376). The typological range of LRD was extended by Meyza, who attempted a revision of Hayes’ classification (Meyza 2000). Nevertheless, the majority of sherds from the southern Anatolian production sites could be matched to forms described in LRP (Jackson 2012, 96), reinforcing the continuing validity of Hayes’ sequence.

Table 3. LRD forms and chronology Hayes form

Variant

LRP date

Possible revisions

All forms

End of the 4th century to c. 700

End of production in 8th century?

1

Late-4th century to third quarter of the 5th century

2

Late-5th to early-6th century (transition to Form 9 in mid-6th century)

3

Third quarter of the 5th century to second quarter of the 6th century

4

5th century?

5

Mid to late-6th century

6

6th century?

7

Second half of the 6th to early-7th century

8

‘No direct dating evidence’

9

c. 550 to 700

Chronology The twelve forms classified in LRP have remained central to the identification of this ware at consumption sites in Mediterranean and beyond. LRFW1 noted that there had been few revisions to the dates assigned in LRP (Cau et al. 2011c, 7; see Table 3). Jackson acknowledged that the recent discoveries in Turkey have limited impact on the dating of these forms, as the finds were not from datable, stratigraphic contexts (2012 109). Once questions of the origin of this ware are resolved, however, further revisions to the dating structure of LRD might be possible using comparable finds from consumption sites. Of significance to the chronology of LRD are recent suggestions that the ware might have been produced into the eighth century. Pamela Armstrong has suggested a very late end date for LRD 9 (Armstrong 2009, 159-61), although the distribution of very late products seems unlikely to have extended beyond the East Mediterranean. Decoration LRD vessels are usually decorated with ‘coarse-rouletting’ in a large single band or multiple narrow bands, probably in imitation of ARS decoration (Hayes 1972, 372). The rims are often grooved and sometimes bear a distinctive incised wavy line. As with ARS-D and LRC stampeddecoration is also common, particularly on forms LRD 2 and 9. Typically there is a single stamped cross on the interior at the centre of the base, although other stamps of leaves or rosettes were also employed (Hayes 1972, 372; 2008, 89).

A

c. 550 to 600

B and C

c. 580/600 to end of 7th century

10

Mid-7th century?

11

c. 550 to 650+

12

Mid-7th century

Produced into the 8th century – or beyond?

to be clarified about the interactions that facilitated the distribution of this fineware in the eastern Mediterranean before the mechanisms by which it reached the Atlantic can be revealed. Nevertheless, it seems likely that further examples of LRD will be identified on the Atlantic Seaboard. Notwithstanding its general rarity outside the eastern Mediterranean, there seems no particular reason why this ware might not be similarly identified in Britain as a minor component of early medieval assemblages.

LRD in the Atlantic Although ARS and LRC have been recovered from postRoman contexts in Britain, to-date no sherds of LRD have been identified. However, a small number of examples of LRD have been identified at sites in western Iberia, marking the westerly extent of its distribution. These will be discussed in full in Chapter Six. Such finds would previously have been linked to trade with Cyprus or to the redistribution of Cypriot products, but until the production history of LRD is more fully understood, the precise origin of these Atlantic examples must remain in question. Much remains

3.3.5 Dérivées des Sigillées Paléochrétiennes (DSP) The final category of imported fineware identified at early medieval sites in Britain and Ireland is of Continental, rather than Mediterranean origin. Present in smaller quantities in Insular contexts than LRC and ARS, it was described 52

Late Antique imported pottery on the Atlantic Seaboard: developments in research been used in this project. The ware has elsewhere been termed ‘céramique estampée’, including in publications detailing finds from Bordeaux (Soulas 1996; 2000; 2012). Hayes used the term ‘Gaulish T.S. Grise’ in LRP (1972, 402). A number of publications describing finds from northern Spain (referred to in Chapter Six) have also used the term ‘Terra Sigillata Gálica Tardía’ (Fernández Ochoa et al. 1992; Uscatescu et al. 1994; Filloy Nieva 1997; see discussion in Fernández 2014, 271).

by Thomas as a ‘kind of sideline’ to the ‘main tale’ (1993, 62). Nevertheless, it has also been seen as significant to the broader chronology of the Atlantic systems – as characteristic of a transitional period of exchange, which overlapped distinct phases of Mediterranean and Continental (E ware) importation to Britain (Campbell 2007a, 32). History of Research Radford’s initial, alphabetic classification of the pottery from Tintagel (which did not include ‘native coarsewares’) comprised groups labelled ‘A-D’. The ‘A’ and ‘B’ classes represented amphorae and red-slipped fineware respectively. His ‘Class C’ – a coarse grey ware – was later recognised to be of medieval date and disregarded (Campbell 2007a, 4). The final group, ‘Class D’, was described as a ‘fine, soft, blue-grey ware with a black wash’ which was represented by a few fragments of bowls and mortaria (Radford 1956, 60). Despite identifying a Mediterranean origin for classes A and B, Radford was unable to specify a source for this grey/black fineware. By the end of the decade, Thomas was able to discuss the presence of comparable pottery at Dinas Powys, as part of his first major publication on the imported wares (Thomas 1959, 94; subsequently published in Alcock 1963, 135-7). Although Thomas noted possible equivalent vessels from France, he concluded that the origin proposed by Alcock – the Bristol Channel area – could be justified (Thomas 1959, 95).

Later research on DSP has continued to reveal further details of the ware, although the initial classification and form-series established by the Rigoirs has not been superseded. Significantly, there have not been any subsequent major synthesising publications on the pottery and its circulation. The distribution map of DSPA presented in 1973 is, therefore, the foundation of the map presented in Appendix H.4 (Rigoir et al. 1973, Figure1, see Figure 3.11). Unlike DSPP and DSPL which are focused on the Mediterranean, DSPA is shown to be concentrated on the area between the Loire and Garonne. A number of additional find-spots have since been added, although Appendix map H.4 is not expected to fully represent the current French distribution of the ware. Although finds of DSP were identified on the Mediterranean coast of Spain, the Atlantic group was not initially considered to have been distributed south of the Pyrenees (Rigoir et al. 1973, 207; Soulas 1996, 252). This has been contradicted in recent publications from northern Spain (see Sub-chapter 6.3.3). A recently published review of DSP in northern France has also extended the known reach of the Atlantic production group. This demonstrated that DSPA was distributed in the Seine basin and along the south Channel coast, including small quantities of vessels at Cherbourg [142] and Bayeux [147], Rouen [143] and as far north as Wailly-les Arras [141] in the Nord-Pas-deCalais Region (Chaidron 2008, 247-9).

It was not until the publication of research in France conducted by Jacqueline and Yves Rigoir that that the Gaulish origin and post-Roman date of the pottery was confirmed. The British examples of ‘Class D’/’D ware’ could be directly matched to a regional sub-class of fine, wheel-turned pottery produced in southern France from the fourth century, and that had previously been termed ‘Visigothic ware’. Three regional groups were outlined in J. Rigoir’s 1968 publication, with notable concentrations at Marseille, Narbonne and Bordeaux (Rigoir 1968, 185). These were labelled the ‘Provençal’, ‘Languedoc’ and ‘Atlantic’ groups respectively. Analysis of stamped vessel suggested that products of the latter group were distributed from the Bordeaux area to sites in western France by maritime routes (Rigoir 1968, 184, Figure 5). A subsequent publication specifically considered the characteristics, chronology and distribution of the Atlantic production group, and again highlighted the significant quantities recovered from various investigations at Bordeaux (Rigoir et al. 1973, 208-9). To date, however, no production sites have been discovered for any of the three regional groups.

Excavations in Bordeaux have continued to reveal very large quantities of the ware, confirming a likely, major source for the Atlantic group close to the city – although it seems that smaller-scale production-centres might also have supplied towns in the south-west (Soulas 1996, 252). Soulas observed that although thousands of sherds had been found in Bordeaux, most of the Aquitanian sites with DSPA produced less than ten sherds (Soulas 1996, 252). Although the ware seems to have had a wide distribution, which includes Britain and Ireland, it was certainly primarily intended for the Bordeaux market. The investigations at Place-Camille-Jullian [201], for example, revealed a homogenous group of 4680 sherds of DSPA, all thought to represent the same, local production centre (Soulas 2000, 145; 2012, 247). Comparisons between this assemblage and the British examples are considered in Sub-chapter 5.3.4.

The terminology for the pottery was modified in subsequent publications, and overall, terminological complexity has created difficulties in equating finds of this ware. J. Rigoir’s initial publication referred to ‘sigillées paléochrétiennes grises et orangées’, but by the 1973 article the pottery was termed ‘dérivées des sigillées paléochrétiennes’. This designation seems to be increasingly favoured, in conjunction with the abbreviation ‘DSP’ (therefore DSPA, DSPL, DSPP for the regional sub-groups), and has consequently

Description DSP has been described as the last in a ‘vast family’ of wares deriving from the earlier Imperial sigillata industries, with forms and decoration styles that mirror ARS 53

Links to Late Antiquity

Figure 3.11. 1973 Distribution map of DSPA. From Rigoir et al. 1973, Figure 1

(Bonifay and Rigoir 1986, 40). The form series principally comprises tablewares, including plates, dishes and cups, although other, ‘utilitarian’ forms, including mortaria are included in the repertoire.

several ‘poinçons’/stampers arranged around a central stamped ‘medallion’. The Rigoirs’ research established the distinctive decorative style for DSPA and formalised a numerical stamp-series for the group. Rosette stamps and, particularly, palm/palmette stamps feature heavily, while DSPA is further distinguished by the common occurrence of human and animal motifs (Rigoir et al. 1973, 246-56). Of the animals, stag/deer motifs are particularly characteristic of the Atlantic series (see Sub-Chapter 6.3.3).

The two Mediterranean production groups were effectively separated by the Rhone. To the west, vessels of DSPL are typically oxidised, whereas the Provençal DSPP, including the large quantity from Marseille (and seemingly produced in its locality), are usually grey (Bonifay and Rigoir 1986, 40-1). DSPA is considered to be the most homogenous group, being almost entirely grey – as a result of ‘reducing’ firing conditions – with a typically fine fabric showing few visible inclusions to the naked eye (Bonifay and Rigoir 1986, 42; Soulas 1996, 237). The grey-black slipped surfaces are usually smooth and polished. DSP is characterised by stamped decoration, although rouletting is also a common feature. The stamped designs are usually situated on the inner base of the vessels, and are constructed from

Forms As mentioned, the form series for DSP established by the Rigoirs has remained in use. Notably, the form repertoire of the Atlantic group is more restricted than DSPL, and much more so than DSPP. Of the original series of 32 DSP forms, only eleven were originally identified in DSPA. Of these, the most common were plates Rigoir 1 and 4, bowls Rigoir 6, 16 and 18, and to a lesser extent bowl 30 54

Late Antique imported pottery on the Atlantic Seaboard: developments in research and the mortaria form 29 (Rigoir et al. 1973, 222, Figure7; see Figure 3.12). Within the Provençal group a trend over time towards more utilitarian forms was observed, including the mortaria form 29 (Bonifay and Rigoir 1986, 42). Soulas later observed that the Atlantic version of this form, though present at Bordeaux from the fifth-century, became increasingly well-represented during the sixth century (Soulas 2000, 149). An originally unidentified ‘pitcher’ form with a tubular spout was later added to the DSP form series – Rigoir 36 – on the basis of finds from Marseille, and was subsequently found at Bordeaux (Soulas 2000, 149-50). This shape was also produced in contemporary coarse grey-wares in the south-east (Bonifay and Rigoir 1986, 42). It also seems to have been the inspiration for the E ware form E4B. Soulas later documented other rarer forms from Bordeaux, including flagons, jars and a chalice, none of which have been identified in Britain or Ireland (Soulas 1996, 247-9).

Figure 3.12. Principal forms of DSPA. After Rigoir 1973, Plates 1, 9, 14 and 15 and Soulas 1996, Figures 9 and 10, not to scale

its later history was clouded by issues of residuality (2000, 146-7). Over this period, however, the repertoire in production evolved and the relative proportions of particular forms shifted. In addition, it is notable that examples of DSPL were identified in the earlier, fifth-century phases at Place Camille-Jullian, suggesting an early demand for imported tablewares – presumably distributed overland from the south-east – that was subsequently met by local, DSPA, production (Soulas 2000, 146).

Chronology The emergence of DSP can be dated to the mid-fourth century, with the Provencal group (most typical of the fifth century) developing a little later than the Languedoc production. The Atlantic group is certainly later, and has been considered as particularly characteristic of the sixth century (Bonifay and Rigoir 1986, 40-2). Campbell repeated the general consensus that production of DSPA centred on the sixth century (2007a, 28). As he noted, this was largely corroborated, if slightly extended, by the emerging evidence from Bordeaux. The excavations at Place Camille-Jullian provided the first major stratified assemblage from which the wider chronology of DSPA could be examined. Soulas was able to broadly demonstrate that the Atlantic group was supplied to Bordeaux from the fifth century, throughout the sixth century and into the seventh century, although

British and Irish examples of DSP/DSPA The Rigoirs’ first publication on DSPA identified the vessels from Dinas Powys as belonging to the same production group (1973, 207). The group from this site remains the largest collection from any Insular site, representing nine vessels (Figure 3.13). Overall, the quantity

Figure 3.13. DSPA vessels from Dinas Powys. From Alcock 1963, Figure 28

55

Links to Late Antiquity (2007a, 31, 138). Campbell observed that sherds of DSPA had been found in association with Mediterranean pottery and E ware at a number of British sites. Its presence in association with ARS at Tintagel (which had no E ware) indicated that it was being imported in the first half of the sixth century, while its occurrence at northern sites such as the Mote of Mark and Dunadd (dominated by E ware), suggested it was still being imported later into the latersixth century, after the Mediterranean wares had ceased to arrive (Campbell 2007a, 31). The identification of DSPA at Vigo allows these models to be questioned, in relation to the wider Atlantic chronology of this ware (Sub-chapter 6.3.2).

of DSP from Britain and Ireland is relatively small, currently comprising a total of 29 vessels from fifteen sites (Campbell 2011). Campbell observed that the fine fabric and some of the forms of the Dinas Powys vessels could be directly equated with the major group of DSPA from Bordeaux (2007a, 29-30). However, he also noted that some of the other British examples had more varied fabrics, decorations and forms, and could potentially be the products of other, smaller French manufactories (2007a, 28). It should be added that the identification of centres producing imitations of DSPA in north-western Spain means that a possible Iberian origin for some of the Insular finds cannot be discounted (see Sub-chapter 6.3.2). Further research is clearly necessary to fully establish the source of the British examples. Campbell also commented that even if Bordeaux was the ultimate source, this did not imply a direct connection to the port. Instead, he proposed the Loire as a more likely point of exportation (2007a, 31). As will be discussed, however, emerging evidence of other classes of material – principally E ware – indicates that DSPA can be taken as evidence of links between Bordeaux/Aquitaine and western Britain.

3.4 E Ware The last major group of early medieval pottery imported to western Britain and Ireland has been classified as ‘E ware’. This is the most commonly identified group at Insular sites, and comprises coarseware vessels of varied form in a hard, gritty fabric. Although E ware is now recognised as a Continental product, its precise source has defied identification. It should therefore be stressed that the ‘ware’ was described and classified based on finds from Britain and Ireland and does not necessarily reflect the identified ceramic output of a known, regional production centre.

A limited range of forms of DSPA have been identified in Britain and Ireland – comprising Rigoir 4, 16 and 29 – that does not fully reflect the series produced in Bordeaux. The mortaria form Rigoir 29, present at seven sites, has been noted to be particularly common in Britain in comparison to Bordeaux (Campbell 2007a, 27).

Campbell’s published research on E ware remains the most comprehensive account of the fabric, typology and chronology of this ware (Campbell 2007a, 32-53; Appendix 1 and 2). Although this sub-chapter will present a concise description of E ware, it is unnecessary to repeat Campbell’s extensive analysis. Instead, the aim within this project was to consider E ware in its Atlantic context. This

Although DSPA clearly had a table/culinary function in its place of origin, the small number of Insular finds suggest that it was not specifically and regularly imported for this purpose. Instead it is likely that it arrived as a ‘secondary cargo’ with other commodities. The recent publication from Tintagel proposed the vessels were distributed as a minor accompaniment to a trade in Gaulish wine in barrels (Thorpe 2007, 241). As with the other classes, the sherds themselves may have taken on a secondary significance. One of the more unusual British finds was recovered from Dinas Emrys [85]; this was a disc cut from a sherd bearing part of a Christian symbol – a Chi-Rho with a sun and omega inside a ‘toothed border’ and part of a palmette symbol (Thomas 1981, 8; Campbell 2007a, 27, 29, Figure 19). Thomas originally assigned this to his (subsequently rejected) ‘Group G – Late Colour-Coated Type’ (Thomas 1959, 110), and later identified it as part of a Mediterranean lamp (Thomas 1981, 8). More recently Campbell re-identified it as oxidised DSP, either a rare oxidised example of DSPA or as belonging to another of the regional producers (Campbell 2007a, 27). Regardless, it seems reasonable to concur with the initial interpretation that it was brought to the site as some sort of token or souvenir (Savory 1960, 61-2; see Figure 3.14). As mentioned, Campbell’s analysis concluded that the arrival of DSPA in Britain overlapped the phases of Mediterranean and Continental importation, and possibly represented a ‘precursor’ to E ware imports; the shift between these systems was dated to the mid-sixth century

Figure 3.14. DSP sherd with Chi-Rho stamp from Dinas Emrys. From Savory 1960, Plate 8b

56

Late Antique imported pottery on the Atlantic Seaboard: developments in research By the publication of his final article in 1990, fragments of nearly 200 E ware vessels had been recovered from approximately 60 sites in Britain and Ireland (Thomas 1990, 6-7). Around this time, Ewan Campbell was also developing his own research on this ware, which, together with imported glass, would form the focus of his thesis and subsequent research (Campbell 2007a, 3)

research has taken a rather unexpected turn; information has been provided that reveals an unexpected extension to the distribution of this pottery, while crucial information on its probable source has also come to light. History of Research Thomas’ addition of an ‘E’ category to Radford’s ‘A-D’ classes of early medieval imported pottery stemmed from analysis of pottery recovered during his excavations at Gwithian (Thomas 1959, 89). However, vessels of this ware had previously been found at sites in Britain and Ireland. The earliest examples (later classified as an ‘E2’ beaker and an ‘E4B’ jug) were recovered at Buiston Crannog [14] in Scotland in 1882 (Munro 1882, 233; Campbell 2007a, 4). Craw described a collection of sherds from Dunadd in 1930, to which he assigned a possible sixth to ninth century date, and noted the presence of comparable finds at the Mote of Mark (Craw 1930, 123-6). Echoing the early history of research on the Mediterranean wares, vessels that would later be classed as E ware were recovered at Garranes in Ireland (Ó’Ríordáin 1942, 125-7). Examples were also recovered at Lagore Crannog [56] in Ireland, which were recognised by Hencken to be post-Roman in date and Continental in origin, although a petrographic examination that was carried out could not identify a more specific source (Hencken 1950, 8-9, 124-5). Ó’Ríordáin had compiled this growing Irish evidence in his 1947 article on ‘Roman’ material in Ireland, which listed parallel examples from Ballinderry [61], Ballycatteen [82] and Garryduff [77] (Ó’Ríordáin 1947, 65-73). Despite the title of the article, he recognised the pottery to be early medieval in date. However, although Ó’Ríordáin mentioned Hencken’s suggested provenance, and noted a corroborative appraisal made by Gerhard Bersu, he expressed doubt that the pottery represented an imported ware (Ó’Ríordáin 1947, 72-3).

Description As Campbell provided an extensive description of E ware in his 1991 thesis and 2007 monograph and appendices, it is only necessary to summarise the main features here. E ware is essentially a coarseware pottery, although its exact function has been disputed. This is in contrast to the amphorae which were produced and (initially) used as storage/transport vessels and the finewares which were, at least in the Mediterranean, intended for dining. The pottery comprises a range of vessel forms, principally jars, but also beakers, bowls, jugs and lids. Decoration is very rare, and limited to single, horizontal, incised lines. The vessels are usually hard, suggesting the pottery was fired in kilns with generally oxidising conditions, at temperatures of around 1000° (Campbell 2007a, 32; Appendix 2, 7). The fabric is characteristically gritty, due to its quartzsand rich fabric, and vessels typically have a ‘pimply’ surface (Campbell 2007a, 32). Campbell notes that the clay came from an iron-poor deposit, to which quartz was added as a temper, but comments that the mineralogy is not particularly distinctive, hampering the identification of its source. White quartz is the principal inclusion, although red iron-ore is also present in variable amounts. Campbell described two fabrics, E1 and E2, the latter of which represents only a few vessels and is distinguished by having rounded rather than sub-angular quartz inclusions (Campbell 2007a, 32, Appendix 2, 2).

Charles Thomas drew this disparate evidence together and added the category (as Class E/ E ware) to the alphabetical system established by Radford at Tintagel, despite the pottery not actually occurring at Tintagel. He intitially added the class in his interim report of the Gwithian excavations of 1953-4, noting seventeen sherds of a similar type to that recognised at Garranes and other Irish sites (Thomas 1954, 68). Although he never fully published these excavations, Thomas continued to develop his research on E ware. His 1959 article introduced the class in more detail, and established the numerical typology based on broad vessel shapes – classified as ‘Ei’ to ‘Ev’ (Thomas 1959, 96-9).

Although the pottery is essentially a white-firing ware (Campbell describes the ‘fresh surface’ as ‘off white, cream or beige’) the actual colour of sherds can be extremely variable, and is affected both by firing and subsequent depositional conditions. Campbell notes that variations in the firing can produce ‘grey patches or surfaces’ or orange or yellow tinged surfaces, that soil conditions can stain the sherds brown or black, and that varying amounts of iron-ore inclusions can result in vessels with a pink or red appearance (2007a, Appendix 2, 2). Unsurprisingly, it seems that this significant variability in colour has hindered the identification of this ware outside of Britain and Ireland. In general, it seems that colour is not a useful factor for identifying this category, and other characteristics – particularly form-type and production features – are more reliable indicators. Given the absence of a known production source, and the importance of E ware as a ‘widespread and critical element of the material culture of Insular Atlantic sites’ Campbell described these characteristic features with considerable care (2007a, 32).

By 1959 Thomas had recognised that E ware represented a Continental import – commenting that despite its overlapping distribution, it was only connected to the Mediterranean wares by a ‘common market’. However, this 1959 publication only marks the start of his efforts to identify a precise origin (Thomas 1959, 96). From the 1950s onward Thomas continued to expand his research on E ware, developing interpretations and extending the known distribution (Thomas 1976; 1988; 1981). 57

Links to Late Antiquity Forms Charles Thomas developed the first form typology of E ware, establishing a numerical sequence, with Ei representing jars, Eii beakers, Eiii bowls, Eiv jugs and Ev lids (Thomas 1959, 97-8, see Figure 3.15). A refinement of this typology with new illustrations – drawn by Mary O’Donnell – was presented in his 1990 publication (Figure 3.15). Thomas’ typology was again extended and refined by Campbell, who sub-divided a number of the forms as detailed in Table 4. This typology has been followed during this project. E1 jars are by far the most common form among the British and Irish finds, followed by the small carinated beakers classed as E2 and then by the E3 bowls. The standard E4 jugs have pinched trefoil spouts, although Campbell notes that a complete profile of this form has not been identified (2007a, 39). A rare but significant addition by Campbell is the E4B jug with a tubular spout (Figure 3.16), which he notes is adapted from the E1B jar, although the form itself seems to be based on unusual form of DSPA, the three-handled Rigoir Form 36 (see Figure 3.12) (Campbell 2007a, 36-9). ‘E5’ represents lids which were probably used with the E1 jars. These are generally rare, although a number of examples have been identified in Ireland, including from Dalkey Island (Doyle 2009, 46). Campbell notes that one of the British examples of an E5 lid (from Bar Point, St Mary’s [139]) is perforated, while a number

Figure 3.15. Thomas’ E ware form typology. From Thomas 1990, Figure 1, drawing of Irish finds by Mary O’Donnell

of the Irish examples (including a lid from Caherlihillan) share this feature (Doyle 2009, 23). The last two forms, E6 and E7 were classified by Campbell and, to date, are represented only by single, fragmentary, examples. Production Characteristics The vessels are wheel-thrown, demonstrating the continuation of wheel-throwing techniques on the Continent. Any handles are also wheel-thrown, which is typical of French medieval wares and is a technique not used in Britain (Campbell 2007a, 34). Campbell notes that another characteristic feature is that the handles and spouts are luted to the body with a ‘smeared finger-mark’ or ‘thumb-smear from the upper surface of the rim across the handle’ (Campbell 2007a, 34, 39, Plate 13; see Figure 3.17). Despite examining other contemporary Merovingian pottery, in an effort to extend the distribution of this ware, Campbell did not encounter this feature elsewhere, and notes that it might be specific to the production zone (2007a, 39). The vessels have smoothly-curved outer surfaces, while the inner surfaces show ‘finger-rilling’, which, Campbell notes, may indicate that a ‘jig’ (a ‘curved strip of bone or wood’) was held against the outer surface while the vessel was being turned (2007a, 4). The bases are also distinctive, usually bearing string-marks showing how the vessel was cut-off and removed from the wheel while it was still rotating (Campbell 2007a, Appendices 2, 4) see Figure 3.18.

Table 4. E ware forms and function. After Thomas 1959; 1990 and Campbell 1991; 2007 E ware Forms (Thomas 1959)

Thomas’ Descriptions

Campbell’s sub-types (Campbell 1991; 2007)

Campbell’s Descriptions (2007a)

Ei

Jars

E1

Multi-purpose storage jars

E1B

larger variant

E1C

Smaller variant

Eii

Beakers

E2

Beakers

Eiii

Bowls

E3

Bowls

Eiv

Pitchers

E4

Jugs with pinched trefoil spouts

E4B

Same form as E1B but turned into pitcher by addition of spout and handle

Ev

Lids

E5

Lids used on E1 jars

-

E6

‘Bottle’? from Dumbarton Rock [12]

-

E7

‘Double-carinated biconical vessel’ from Samson [140]

Function The function of the E4 and E4B jugs/pitchers is clearly connected to pouring liquids – possibly wine – while the function of the other forms is less clear. Thomas described the pottery as a ‘kitchen-ware’ and the forms as ‘whollydomestic’ (Thomas 1990, 1). He labelled the E1 jar as a ‘cooking pot’, but Campbell has argued against this attribution. His analysis of the E ware fabric indicated that it 58

Late Antique imported pottery on the Atlantic Seaboard: developments in research

Figure 3.16. Campbell’s E ware forms. From Campbell 2007a, Figure 21

might be ill-suited to cooking, and he suggested that the vessels might have had a multi-purpose function, including table-use and storage (Thomas 1959, 96; Campbell 2007a, Appendix 2, 25). Nevertheless, the pierced holes on some E5 lids may have been to let steam escape, indicating some cooking function for these and the associated

E1 jars (Campbell 2007a, 48; Doyle 2009, 23). Campbell stated that it was often difficult to distinguish blackened stains from the original firing or from depositional conditions on E ware vessels from sooting caused by cooking (2007a, 49). Even where soot-deposits can be identified this does not, necessarily, indicate the primary function of 59

Links to Late Antiquity

Figure 3.17. E ware rim/handle junction showing characteristic ‘thumb-smear’. From Dunadd. By kind permission of National Museums Scotland

the vessel. A number of the British examples of E ware seem to have been reused for ‘medicinal’ or ‘industrial’ purposes and Campbell describes vessels from Loch Glashan [11] and Port y Candas [22] that seem to have been trimmed down for reuse (2007a, 49). In addition, sherds from the Mote of Mark were noted to have ‘glazelike deposits’ on the surface, indicating some industrial association, although these were not identified with certainty as E ware (Campbell 2007a, 49-50).

consumption practices associated with the vessels or their exotic contents, or the long-distance connections that they represent, were significant to the local power-dynamics of these high-status sites. Campbell has also suggested that a number of the Insular examples of E ware might represent the personal property of traders themselves (2007a, 51) Chronology The chronology of E ware has been debated since the ware was classified but remains a matter of continuing uncertainty. This is, at least in part, due to the lack of an identified production source. However, as the ware is frequently discovered at sites with little other datable material, the pottery has been ascribed considerable importance for dating early medieval sites in Britain and Ireland. Campbell notes that ‘E ware is generally the only material found on Atlantic sites of this period which is theoretically capable of independent external dating’ (Campbell 2007a, 44).

However, despite this evidence for reuse, Campbell concluded that the primary function of the E1 jars, the most common of the forms, was as a storage container, and that these vessels were imported for their contents rather than any intrinsic value or function (Campbell 2007a, 51). Traces of madder, used as a purple dye, have been found on a number of vessels, and dyestuffs were therefore suggested by Campbell one of the potential, exotic commodities that might have been transported in E ware (2007a, 49, 80). The function of the E2 ‘beakers’ is uncertain, and it is also possible that they were used as containers. However, it is also possible that they were used as drinking vessels, and it might be that these, the E4 jugs and pitchers, and possibly the E3 bowls were connected to the consumption of wine (Campbell 2007a, 52). Although E ware has been recovered at a number of high-status sites in Britain and Ireland, Campbell concluded that the vessels themselves are unlikely to have been seen as a marker of ‘prestige’ (2007a, 51). It might be, however, that the

The first classified examples of E ware were included by Thomas, together with the imported Mediterranean finewares and amphorae, in his ‘Group 1’ pottery from Gwithian. This group was found to stratigraphically predate the ‘bar-lug’, ‘grass-marked’ pottery categorised as ‘Group 2’ (Thomas 1954, 66-70). This, as part of a wider consideration of sites with E ware, allowed Thomas to suggest a date for its importation from the fifth century 60

Late Antique imported pottery on the Atlantic Seaboard: developments in research

Figure 3.18. E ware base showing characteristic string-marks. From Mote of Mark. By kind permission of National Museums Scotland

(though largely from the sixth century) up to the early eighth century (Thomas 1959, 104). Over time, however, this early date for the first imports was questioned and eventually retracted (Campbell 2007a, Appendix 2, 16). By his latest major publication on the ware, Thomas had divided the systems supplying early medieval western Britain into two separate phases of Mediterranean and Continental imports. E ware was considered to be the primary marker of this secondary phase, which was described by him as ‘best seen’ in the late-sixth and seventh century (Thomas 1990, 1).

2, 16). Of particular importance to this is the analysis of the stratigraphic sequence at Clogher, Co. Tyrone [28], where deposits producing E ware were separated from a deposit producing Mediterranean amphorae by a sterile ‘yellow’ layer of boulder-clay and stones (Warner 1979, 37-8). This was interpreted as a deposit connected to a phase of re-building at the ring-fort, which was dated, by a consideration of radiocarbon dates and historical events, to the second half of the sixth century (Warner 1979, 38). Warner’s conclusions have been influential in providing a chronological framework for these wares at other sites where the stratigraphy is less clear. However, Doyle has commented that more recently studied sites in Ireland reveal some overlap between E ware and the imported amphorae in the mid to late-sixth century (Doyle 2009, 25).

Campbell provided a full discussion of the issues surrounding the chronology of E ware in his 2007 synthesis (Campbell 2007a, 44-6; Appendix 2, 13-22). He suggested that the importation of this ware took place from the midsixth to the early-eighth century, particularly between the later-sixth and later-seventh century, with a floruit in the earlier-seventh century. However, he noted that the current accepted chronology ‘rests on a variety of evidence, none of which is conclusive, but which taken together points to a restricted period of time’ (Campbell 2007a, 46; Appendix

Campbell described how he reached a similar conclusion to Warner about the separation of the two phases of importation following his own research. This was based on the analysis of E ware from a number of various sites studied in combination, including Dalkey Island, and specifically on Hill’s detailed phase-sequence from Whithorn (Hill 61

Links to Late Antiquity ent. Within the scope of this project it was not possible to re-examine the Irish material and verify these totals. Therefore, Campbell’s 2011 dataset has been used as the principal source for the quantities of E ware in Ireland, but any new sites identified by Doyle have been added to the database. The vessel totals quoted by Doyle are noticeably higher than Campbell’s, but this might reflect differences in methodology and quantification, as well as recent reassessments of older assemblages. In general, more research on this ware would be useful to formalise identifications and standardise descriptions, which would allow a more precise understanding of its distribution.

1997, 322-6). Although Campbell questioned the merits of an overly simplified link between deposition and supply, the pottery and glass from Whithorn demonstrated a ‘clear stratigraphic progression from Mediterranean pottery to Continental’ with a confirmed floruit for E ware in the seventh century (Campbell 2007a, 108, Appendix 2, 20). The evidence from Whithorn was also used to indicate that the importation of E ware ceased in the early eighth century. Like Doyle, Campbell conceded that some sites show Mediterranean pottery and E ware to be stratigraphically associated. However, he was critical of the conclusions reached by Charles Thomas at the two sites with the ‘strongest’ arguments for this contemporaneity, Tean [134] and Gwithian – neither of which were fully published (Campbell 2007a, Appendix 2, 17). Campbell concluded that there was strong evidence that the importation of E ware commences after the cessation of Mediterranean imports in around AD 550.

3.5 Other Materials This appraisal has intentionally concentrated on the major categories of pottery imported to Britain and Ireland. The parallel identification of these wares at sites on the Western Seaboard confirms them as key markers of Late Antique/ early medieval exchange in the Atlantic. However, other ceramic wares, as well other categories of material, were also transported, traded or exchanged across this region. Limitations of time prevent a full review of this material, and it must be noted that the Insular finds were comprehensively considered by Campbell (2007a, 54-82).

E ware was produced in a limited range of forms and lacks the sequence of identified variants seen with the finewares. It seems probable that the entire repertoire was in contemporary production, although it might be that further analysis of the typological development will allow this dating to be refined.

3.5.1 Coarsewares and unclassified ceramics Origin Not every sherd of pottery found at early medieval sites in Britain and Ireland has fallen neatly into the categories discussed above. This is doubtlessly exacerbated by their typical small size and abraded condition, but there have been suggestions that the Insular assemblage contains a wider variety of imported wares than previously recognised. Sherds from Tintagel have been identified as imported Mediterranean coarsewares, but although Ewan Campbell considered that this was not inherently implausible, he was not able to corroborate these findings (Campbell 2007a, 24-5).

Campbell’s 2007 monograph describes E ware as coming from a ‘single, coherent source…within the area of Merovingian France’ (Campbell 2007a, 48). However, despite considerable speculation and effort since the ware was first classified, the specific origin of the pottery has never been identified. This question is fully discussed in Sub-chapter 5.4, where it is suggested that the issue is approaching resolution – with Bordeaux proposed as a – and possibly the – likely source. British and Irish Examples of E ware

This issue is discussed in greater detail in Sub-chapter 4.2.3. Imported coarsewares of Mediterranean origin have, however, been recorded from sites in western Iberia, most notably at Vigo, and have been remarked on in Chapter Six. These finds augment Reynolds’ detailed study of coarseware distribution in the West Mediterranean – itself reliant on formative work on late coarsewares conducted at Carthage (Fulford and Peacock 1984, 155-231; Reynolds 1995, 86-105).

In 1959 and 1981 Charles Thomas presented complete catalogues of E ware sites in Britain and Ireland, which was expanded by Campbell’s 2007 monograph and updated dataset (Campbell 2011). In Ireland, recent research conducted by Ian Doyle has increased the number of sites known with E ware; his 2009 catalogue of Mediterranean imports was appended by a table of E ware from Irish sites (Doyle 2009a, 60-1). This included new finds from sites excavated as part of the M3 road project: Collierstown, Garretstown [53] and Roestown [55]. Doyle’s E ware table is also based on previous catalogues published by Ewan Campbell, Charles Thomas and Mary O’Donnell, the latter of whom completed an unpublished MA thesis in 1984 (it has not proved possible to view this work).

In addition, Campbell also described ‘unclassified wheelthrown coarsewares’ and ‘unclassified pottery’ which represented possible imports of Mediterranean or Continental origin (some of which may have been related to E ware) but with different forms or fabrics (2007a, 52-3, 76-8). Further international collaboration will hopefully allow much of this unidentified material to be identified. In particular, current research on early medieval coarsewares in south-western France should lead to the characterisation of distinct regional productions, which can thereafter be compared to the scattered Insular examples of E ware.

Campbell’s 2011 published dataset noted the inclusion of data from Doyle, but there are some discrepancies between Campbell’s catalogue and Doyle’s 2009 list, specifically the number of sites, number of vessels and forms pres62

Late Antique imported pottery on the Atlantic Seaboard: developments in research 3.5.2 Glass

than cullet (2007a, 54, 96). The latter observation was based on a detailed consideration of the Dinas Powys material, as well as the presence of complete vessels at sites in Ireland (Campbell 2007a, 92-6). Campbell’s recent monograph presented an estimate of over 300 vessels from over 45 sites – mostly high-status settlement sites – in western Britain and Ireland, with additional data presented online by 2011 (2007a, 54; see Figure 3.19).

A short appendix to Radford’s first publication on the Tintagel finds records glass from the same contexts as the imported pottery, most of which was assigned a probable eastern/Egyptian origin (Harden 1956a, 60). The same volume includes Harden’s major study of early medieval glass in Britain (1956b). This concentrated on evidence from eastern Britain, dating particularly between the fifth and seventh century, during which period glass vessels were deposited as grave-goods in Anglo-Saxon inhumations. However, Harden also noted the contemporary presence of fragments of glass at settlement sites in the ‘Celtic areas’ of western Britain, Scotland and in Ireland (1956b, 149-51). Harden offered a northern Gaulish/ Rhineland origin for much of the glass in eastern Britain, apart from the vessels with opaque white trailed decoration which he considered to have originated ‘further west’ (1956b, 147-8). Most significantly, he suggested that sherds of imported glass were transported to western regions as ‘scrap’ (cullet) for reuse in jewellery and metal objects (1956b, 151-2). As more finds of glass were recovered at western sites, the association with the imported pottery – Mediterranean imports as well as Continental E ware – was confirmed.

Campbell divided the glass imports into five broad ‘traditions’, although he noted that these did not indicate common or confirmed production sources (2007a, 54-5). These comprised: ‘Group A’ – vessels of ‘Late Roman/ Mediterranean tradition’, which were typically colourless with ‘wheel-cut’ decoration; ‘Group B’ – vessels of ‘Anglo-Saxon or Germanic tradition’, which were typically darker in colour; and ‘Group C’ of ‘Atlantic tradition’, characterised by thin, yellowish glass frequently decorated with opaque white ‘marvered’ decoration. The final two groups represented undecorated vessels of the ‘Atlantic tradition’ – ‘Group D’ – and a group unique to Whithorn – ‘Group E’ (Campbell 2007a, 54). The majority of the glass was identified as Group C or D, while the pale yellowish colour and white-trailed, marvered decoration were observed to be particularly representative of the ‘Atlantic West’ in general, potentially indicating a ‘restricted source’ (Campbell 2007a, 55, 64).

Ewan Campbell’s research has considerably expanded understandings of glass in western Britain and Ireland, and led to revisions of Harden’s initial suppositions. Specifically, Campbell recognised the importance of considering the glass and pottery together, as components of the same exchange system (2007a, 54). He concluded that glass was a ‘major import’ to western regions, which, crucially, arrived at settlement sites as intact vessels rather

Differences in the characteristic forms were observed between Groups B and C – the former group more varied but typified by claw beakers and palm cups, and the latter dominated by cone beakers (Campbell 2007a, 54). The low proportion of bowls in both, contrasting with Group A, suggests a broad early medieval focus on drinking vessels (Campbell 2007a, 73). The prevalence of glass drinking vessels at high-status Insular sites can, potentially, be associated with patterns in the ceramic evidence – as discussed in Chapter Seven – which suggest that demand for wine characterised the northern reaches of the Atlantic supply network. Glass finds of the ‘Germanic’ and ‘Atlantic’ traditions were also suggested to represent very different import channels, with vessels of Group B being transported/exchanged overland from Saxon England and the majority of the glass arriving via the same Atlantic routes as the imported pottery (Campbell 2007a, 60). Group A glass was observed to be concentrated on south-western sites with imported Mediterranean pottery, and was therefore connected with this ‘trading system’ (Campbell 2007, 136). As the distribution of Groups C and D echoed that of DSPA, they were associated by Campbell with the Continental phase of imports, although elsewhere he suggested Group C spanned ‘both import systems’ across the sixth and seventh century (Campbell 2007a, 68, 135). Following the surge of data on imported pottery, new information on early medieval glass has started to emerge from sites on the Atlantic Seaboard. Campbell has noted Iberian and western Gaulish parallels for vessels of

Figure 3.19. Campbell’s distribution of imported glass in Britain and Ireland. From Campbell 2007a, Figure 39

63

Links to Late Antiquity evidence for blowing/vessel production in western Britain or Ireland, except potentially at Whithorn (Campbell 1991, 64, Table 11; 2007a, 72, 92). This two-stage system of glass importation mirrors the proposed model of Atlantic trans-shipment for the Insular ceramic imports (see Chapter Seven). The crucibles associated with glass production, assigned to Phase 14, comprise vessels of wheel-turned domestic pottery, seemingly equivalent to Insular E3 bowls (see Sub-chapter 5.4.2).

Group A, including some found at Tintagel (2007a, 58). Most significantly, he has also discussed information on early medieval glass and glass-working emerging from Bordeaux, and highlighted parallels between his ‘Group C’ and the Aquitanian material described by Foy and Hochuli-Gysel (Foy and Hochuli-Gysel 1995; Campbell 2007a, 65). This publication described glass from several excavations in Bordeaux, including Saint-Christoly and Saint-Seurin, but most notably from Place CamilleJullian. In addition, finds were recorded from several villa sites in Aquitaine, revealing both occupation and the use of glass in rural areas into, at least, the sixth century (Foy and Hochuli-Gysel 1995, 164).

It seems clear that future research is needed to fully determine connections between the Aquitanian and Insular glass– as indicated by Campbell (1995, 37). A final issue complicates the seemingly neat association with Bordeaux. Large quantities of glass – as yet unpublished – have been recovered at Vigo, suggesting it was also a point of secondary manufacture (Fernández Fernández pers. comm.). The material – in colour and decoration – appears to show parallels to the Insular Group C. As Vigo would appear to be central to ceramic exchange in the northern Atlantic, a link with early medieval glass in western Britain and Ireland would not be surprising. Again, this must be a focus for future investigations.

Like the Insular category, the Aquitanian material was observed to be dominated by cone beakers and bowls, which frequently bore opaque white-trailed decoration – although the typical colour was given as green/olive-green (Foy and Hochuli-Gysel 1995, 155). Given these similarities, Campbell proposed that Groups C and D might originate close to the source of E ware and DSPA and ‘in the Bordeaux region’ (2007a, 73, 138). In the recent publication on Tintagel Campbell confirmed Bordeaux as the source of glass vessels of Groups C and D (2007c, 225).

3.5.3 Miscellaneous materials However, the full report of glass from the Place CamilleJullian excavations was not published until after Campbell’s 2007 monograph. Foy’s report recorded 715 sherds of glass dating between the fifth and eighth century, representing an estimated minimum of 174 vessels – approximately half of Campbell’s total for western Britain and Ireland (Foy 2012, 335). This equates with the large quantity of ceramics discussed in Chapter Five, and confirms the port-city as a primary hub for Atlantic exchange. Although the analysis was complicated by issues of residuality, a definite peak in glass-use was recognised for the sixth century – paralleling the floruit of East Mediterranean ceramic importation (Foy 2012, 335). Fragments assigned to Phase 14 (see Sub-chapter 5.3.4) indicate continuing use into the seventh century; this phase appears to have witnessed changes in the typical colour and decoration, including the disappearance of the white-trailed decoration (Foy 2012, 344).

Beyond pottery and glass, additional items have been identified or suggested to represent early medieval imports to western Britain and Ireland; these again would benefit from a wider Atlantic appraisal. Campbell’s recent discussion of these finds considered coins and Menas flasks/ampullae as well as other rare finds, including metalwork, mosaic tesserae and the possible Byzantine intaglio from Cefn Cwmwd in Anglesey [84] (Campbell 2007a 74-82; Henig 2012). Many of these items, however, were stray finds or were of questionable provenance, and overall they did not impact on his broader conclusions (2007a, 78, 82). Harris also considered these non-ceramic items, but also noted their rarity in comparison to the Mediterranean pottery (2003, 144). Sub-chapter 5.3.4 contains a brief discussion of coins, relating to recently published finds from Bordeaux. 3.6 Summary This summary of the major types of pottery imported to the Atlantic has revealed the narrative of contact between early medieval western Britain and Ireland, the Mediterranean and the Continent to be based on the detailed analysis of ceramic assemblages. These understandings are reliant on classification sequences and typologies developed in Britain and the Mediterranean – which have been subject to continual revision. The next chapter will discuss the distribution of these wares in Britain and Ireland, expanding on previous cataloguing and interpretive work – particularly by Charles Thomas and Ewan Campbell.

Foy also provided additional details of the evidence for glass-working at Bordeaux. Although workshop structures have not been identified, waste-fragments of raw material and deformed vessels as well as ceramic crucibles demonstrate the ‘secondary’ manufacture of glass vessels, particularly associated with Phase 14 (Foy 2012, 335, 344, 352-71). Slabs of raw material, likely to have been produced in primary manufactories in Egypt/Syria-Palestine, would have been imported to western regions for re-firing (Foy 2012, 353). Although glass-melting seems to have been a regular practice at Insular import sites, there is no

64

4 Imported pottery in Britain and Ireland: new evidence and understandings 4.1 Introduction

pottery imported to post-Roman Britain (Radford 1956). The site has also produced, by a considerable margin, the largest assemblage of imported Mediterranean pottery in Britain or Ireland. It is, therefore, unsurprising that Tintagel came to dominate the research on these imports, and that impressions of the site and its assemblage have steered the overall interpretations of connections to post-Roman western Britain. However, difficulties in establishing the quantities of material recovered from the site and the range of material represented (types, forms and variants) have simultaneously distorted the character of the British import assemblage and limited the usefulness of the Tintagel material for wider Atlantic or Mediterranean comparisons. After over eighty years of research on the Tintagel pottery it remains difficult to understand the actual number of vessels represented by the numerous sherds that have been recovered. A fundamental question can be posed: how much pottery has been recovered in total from the site and of which types?

The principal intention of this project is to evaluate the imported pottery from early medieval sites in Britain within a wider Atlantic perspective. This necessitated a review of the published British material and a reappraisal of the characteristics of the Insular import assemblage. As discussed in Chapter Two, the imports have a long history of research in Britain, commencing in the 1930s. However, new identifications of Mediterranean and Continental ceramic imports continue to be made at early medieval sites in Britain and Ireland. At the same time, recently published discoveries of imported pottery are starting to suggest a greater degree of complexity within the British material and that established models for the arrival and subsequent distribution of these wares require revision. This chapter will present a discussion of some of this recent evidence, commencing with a review of the recently published evidence from Tintagel. The history of ceramic recovery and classification at the site is summarised, and recommendations for future directions in research on the assemblage are suggested. The core of this chapter comprises a presentation of new evidence from sites in south Devon, following the publication of a large assemblage of imported pottery from Bantham (Reed et al. 2011). The discovery of significant quantities of material at this site and its subsequent, detailed publication, provides an opportunity to examine the nature of imports reaching the south Devon coast – and to assess these findings for the British imports as a whole. The material from Bantham is considered as part of a case-study of sites on the South Devon coast. This incorporates original research carried out on the imported pottery recovered at Mothecombe and High Peak in Devon and an isolated marine-find from Cawsand in Cornwall [123]. Finally, some additional recent discoveries from Britain will be briefly summarised. The identification of East Mediterranean pottery at Rhynie in eastern Scotland [3] and ARS fineware at Pevensey Castle in East Sussex [108] have significantly extended the established distribution of Mediterranean pottery in early medieval Britain.

4.2.1 Tintagel: history of investigations Radford’s major excavations on Tintagel Island (principally at the sites labelled A-G – see Figure 4.1) took place between 1930 and 1939, but were halted by the outbreak of the war. Minor excavations took place in 1955, including work in the Great Ditch, although the precise location and extent of this work is unclear (Barrowman et al. 2007, 3). Although Radford published an interim report of his 1930s work and various articles on the site, no final report was ever published and many of the original site records were destroyed by bombing in Exeter (Barrowman et al.

This new information presents an opportunity to review the nature of the pottery reaching Britain using current understandings of the material established from research in the Mediterranean (discussed in Chapter Three) and in light of emerging evidence from the Atlantic Seaboard (Chapters Five and Six). 4.2 Tintagel: reassessing the ceramic research Figure 4.1. The location of Radford’s investigations on Tintagel Island. From Barrowman et al. 2007, Figure 3. By kind permission of the Society of Antiquaries of London

Excavations conducted in the 1930s at Tintagel [110] by Radford led to the first classifications of Mediterranean 65

Links to Late Antiquity north-eastern part of the island), including the newly identified ‘Lower Terrace’, and at Site T (the great Ditch on the mainland) (Barrowman et al. 2007, xiv, 21, 33-6). The final report of these recent excavations sites was published in 2007, coinciding with the publication of Campbell’s major synthesis. The pottery report that was produced by Carl Thorpe is discussed in Sub-chapter 4.2.3.

2007, 5). As a result, the majority of the finds from his excavations cannot be tied to specific parts of the site. Various smaller excavations have been carried out at the site following Radford’s work, largely related to the maintenance of the site, which have typically produced (redeposited) Mediterranean pottery. The reporting of these findings has typically consisted of sherd counts for the types of fineware (ARS, LRC and ‘D ware’/DSP) and the ‘B ware’ amphora classes. Minor excavations were carried out by Cornwall Archaeological Unit in the Lower Ward on the mainland in 1986, which produced 74 sherds of Mediterranean imports (Thorpe 1988, 22). In 1988 a section was dug to the area north of the Inner Ward, which produced at least 50 sherds of various Mediterranean classes, including amphorae and fineware (Thomas 1988b, 58-60). Although illustrations of a selection of the sherds were provided in the former report, and sherd counts by type were presented for the latter, neither report related the finds to an overall, running total vessel count for the site.

4.2.2 The Tintagel pottery: research prior to 1990 The pottery from Tintagel has been the subject of various phases of research and revision since Radford’s first classifications were published (Radford 1956). Radford’s article does not provide sherd counts for the imported pottery, although a ‘considerable amount’ was noted to have been recovered (Radford 1956, 59). Vessel numbers were, however, suggested for the fineware sub-groups: more than twenty vessels of ‘Ai’ and over twelve of ‘Aii’ (Radford 1956, 61-2). Charles Thomas commenced his work on the imported pottery in the 1950s. His various articles on the British imports used the material from Tintagel as the principal site for developing models of connection between Britain and the Mediterranean. Thomas’ 1981 Provisional List of imports in Britain and Ireland contains a discussion of the site and its status; here Thomas retracted his earlier monastic interpretation and framed the site as the most important of the primary landing sites for Mediterranean shipments (Thomas 1981). By 1988 he had suggested that all the material at Tintagel might represent one specific commercial voyage from the East Mediterranean, directed towards the site to trade for locally-streamed tin (Thomas 1988a, 13).

In 1989 and 1990 excavations were carried out in the ‘Steps’ area between the Inner Ward and Site F, as part of reconsolidation work (Batey et al. 1993). Thirteen sherds of LRC were identified by Carl Thorpe including two abraded rims identified as Hayes form 3 and five sherds of ARS (Batey et al. 1993, 55). Although illustrations were provided for some of these, it is not clear whether these all represent ‘new’ vessels or might be matched to earlier finds. Over two hundred sherds of amphorae were also collected, including LRA1 (88 sherds), LRA2 (103 sherds), LRA3 (4 sherds) as well as 43 sherds of the ‘Bv’ category (Batey et al. 1993, 55-6). Thorpe’s identifications include three sherds in an ‘Almagro 51A-B Fabric’ which is compared to the equivalent Keay 19 amphorae (Batey et al. 1993, 56). A southern Spanish or North African origin is suggested for these, although this type was produced in western Iberia (University of Southampton 2005). This is the only specific mention of a possible Almagro 51 amphora from Tintagel and seems to be the earliest mention of a possible Iberian import, and consequently deserves further inquiry.

Thomas’ 1981 catalogue counted 39 fineware vessels from Tintagel: approximately 26 LRC, 10 ARS and 3 DSP vessels, as well as at least 50 amphorae (Thomas 1981, 6-19). In the catalogue the finewares are matched to Hayes’ terminology and listed by Hayes form and variant. From 1983 Thomas supervised the assembly of the Tintagel collection at the Royal Cornwall Museum in Truro (where it is remains held) and commenced a review of the pottery (Thomas 1993, 67-8). He notes that an ‘early aim’ of this work was to establish the minimum vessel quantities for the amphora sherds (Thomas 1993, 67-8).

Excavations producing imported pottery have also been carried out at the church of St Materiana on the mainland, which seem to have produced significant amounts of amphora sherds, including a sherd of ‘Bv’ that was clipped into a disc (Nowakowski and Thomas 1990, 14-5; 1992, 10-11). Full totals were not published, although, unusually for Tintagel, LRA1 was said to be the most common (Nowakowski and Thomas 1992, 10).

In 1988 Thomas, together with Thorpe, produced a catalogue of the Tintagel pottery, based on this recent sorting, although this was never published. They later commented that only 10 to 15% of this pottery had any provenance and that most of the material had remained unsorted since the 1930s (Thomas and Thorpe 1993, 59). It has not been possible to view this catalogue, although, in a subsequent article from 1988, Thomas provided overall vessel totals for the pottery collected from the site based on this cataloguing: 30 LRC, 18 ARS and at least 141 amphorae (see Table 5) (Thomas 1988a, 12). By 1993 Thomas described the site as producing three to four thousand amphora sherds (Thomas 1993, 62) and hypothesised that as only 5% of the site had been excavated, the original totals arriving

A more recent major phase of research at the site was carried out between 1990 and 1999 for English Heritage (led by Christopher Morris of Glasgow University), following a fire on the island in the 1980s (Barrowman et al. 2007, xiv). As part of this work, the surviving part of Radford’s site archive was fully re-appraised. The general location of Radford’s trenches was identified and excavations/reexcavations were carried out in Site C (terraces on the 66

Imported pottery in Britain and Ireland: new evidence and understandings reference to form types and variants for the fineware (Campbell 2011). Campbell was given access to Thomas and Thorpe’s unpublished catalogue and based his totals on this and his own observations of the assemblage (1991, 139). Campbell’s most recent totals suggested a minimum total of 99 amphorae for the site and 44 fineware vessels (Campbell 2011). Originally, he considered Thomas’ counts for ‘Bv’ overly high and instead suggested a minimum of ten for the site (Campbell 1991, 139). By his 2011 dataset these totals had risen significantly to 23 (Campbell 2011). If some, or indeed most, of these vessels can be proven to represent Iberian amphorae, this must considerably impact on understandings of Tintagel within systems of Atlantic exchange.

at the site might have represented one or more ‘complete shiploads’ of six or seven hundred amphorae (Thomas 1993, 71). Thomas later commented that the quantity from Tintagel was ‘not only dramatically greater than that from any other single site dated to about AD 450-600 in either Britain or Ireland, but also larger than the combined total of all such pottery from all known sites’ (Thomas 1993, 71). Campbell’s assessment of the pottery – conducted as part of his doctoral research – counted ‘almost 1500’ sherds of amphorae, representing about 80 vessels (1991, 140). He added that the number of storage boxes gave the impression of a much larger quantity.

Figures were ‘conservatively estimated’ by Thorpe in the 2007 excavation report for the overall quantity of imported vessels at Tintagel, represented by surviving fragments: 150 amphorae ‘of all types’ and 80 fine tablewares (Thorpe 2007, 246). However, these suggested vessel counts were not subdivided into amphora or fineware types (e.g. LRA1 vs LRA2 or ARS vs LRC vs DSP). The higher quantity of vessels suggested for the site was, however, considered to indicate the arrival of more than one voyage between c. 500-600, contradicting Thomas’ 1988 model for the arrival of the Tintagel pottery (Thomas 1988a; Thorpe 2007, 246). It seems therefore, that Campbell’s minimum figures for Tintagel may be an underestimate. Nevertheless, as his database represents the only published summary of the pottery by type, form and variant, his totals have been used in this project – although a note of caution should be added that any conclusions on the overall character of the British material must be considered to be provisional.

4.2.3 The Tintagel pottery: discussion of the imports reported from the 1990-99 excavations The excavations conducted from 1990-99, which were published in 2007, produced a further 1821 sherds of pottery (Thorpe 2007, 232). These quantities alone – discounting Radford’s finds – were stated to ‘dwarf’ those from other British import sites, including Dinas Powys, South Cadbury and Whithorn (Thorpe 2007, 246). The pottery from the Lower Terrace was also published separately in 1997 (Thorpe 1997). It should be noted that some of the pottery recovered in the 1990s excavations might represent sherds discarded by Radford. A number of sherds were found in the backfill of Radford’s ‘Site C’ trenches that had ‘clean breaks’, suggesting ‘a lack of familiarity’ with these wares by Radford’s 1930s workforce (Barrowman 2007, 79-80). Campbell has furthermore suggested that the fineware collection, largely comprising rim and base fragments, indicated that Radford’s workers had discarded undiagnostic or undecorated sherds (Campbell 1991, 139; 2009, 301).

Finewares

The 2007 report describes the types encountered in each of the excavated areas and provides relative quantities by type, but these are only given as percentages of the total sherd count. The focus of the report is the relative distribution of pottery by volume and type – for example the absence of red-slipped wares from the Upper Terrace, and how this might relate to varying accessibility to imported wares for the occupants, or differences in function across the site (Thorpe 2007, 236, 244). The report does not provide a detailed summary of the material recovered or discuss how these ‘new’ sherds relate to the complete assemblage from Tintagel or to the British imports as a whole. Bonifay has stated that it was regrettable that the material was not more extensively presented with a complete catalogue and more illustrations (2008, 1124).

Seventeen ‘new’ sherds of ARS were reported from the various areas investigated across the site in the recent phase of work (Thorpe 2007, 235), but although small diagnostic rims were mentioned, these were not illustrated, and it is not clear how these related to the ARS vessels catalogued by Thomas (1981, 8-9). Two sherds of ARS, including a base-sherd with part of a foot ring were recovered from Site C during the recent excavations and recorded as a ‘Hayes form 50 bowl dating from c.525-33’ (Thorpe 2007, 240). This rather refers to Fulford form 50, which does not equate to a specific Hayes form – Fulford noted typological resemblances to Hayes forms ARS 93A, 93B, 94, 98 and 108 – but has previously been identified among the Tintagel ARS by Campbell (2011). Fulford had dated his Form 50 to c. 500-575, but noted that its popularity at Carthage had passed by c. 575, or possibly earlier (Fulford 1984a, 68-9).

The ongoing confusion about the total quantities was highlighted in the 2007 report, which mentions Leslie Alcock’s significantly higher quoted estimates (Barrowman et al. 2007, 329). Nevertheless, it is only Campbell’s work that has provided a published, detailed breakdown of the Tintagel material by minimum vessel count with

Attention is drawn to an ARS sherd from a deposit in the Lower Ward with a radiocarbon date of AD 340-530 at 95% confidence and AD 390-430 at 68%, as ARS is reported to usually date to the sixth century (Thorpe 2007, 245), and these dates might suggest an earlier arrival. However, as discussed in Chapter Three, the sixth century focus for 67

Links to Late Antiquity count, rather than weight or minimum vessel counts, and it is unclear how the figures relate to proportions of vessels. A further 228 sherds were recorded as ‘unknown’, although the quantification tables for Site T record them as ‘unknown B ware’ (Thorpe 2007, 245-6). These ‘unknown’ sherds have therefore been counted as amphorae of unknown type in the creation of Table 5. This shows the relative proportion of the major amphora types from Tintagel based on the sherd counts published for the 199099 excavations, the provisional vessel counts provided for the pottery found at the site up to 1988 (Thomas 1988a, 12) and minimum vessel counts from Campbell’s 2011 dataset, which included the pottery from the recent excavations. It is noticeable that although the sherd counts from the recent excavations would indicate that LRA2 massively dominates, the estimated/minimum vessel counts reflect a much lower bias toward LRA2. As the high proportion of LRA2 to LRA1 is noted to be a unique feature of the British material as a whole, it is interesting to consider whether this might be, at least to some degree, a construct of the quantification strategies. It does seem, nevertheless, that LRA2 was a major feature of the imported pottery reaching Tintagel.

ARS imports to post-Roman Britain reflects a general impression of the identified forms, rather than hard limits. Although a sherd belonging to the earlier range (AD 390430) would certainly be unusual in Britain, an ARS vessel dating to the later-fifth century would not be surprising. 28 sherds of LRC were recovered during the recent excavations. Again, these are reported to include diagnostic rim-sherds, of which one is noted and illustrated – ‘RF3434’ (Thorpe 2007, 232-3 Figure 127). This is a rim fragment of Form LRC 3, possibly LRC 3F, although this is not clear from the illustration. The one sherd of DSP recovered was published as the rim of ‘either a mortaria or a large bowl’ (form Rigoir 1 or 2) (Thorpe 2009, 235). Campbell’s database identified this sherd as from a Rigoir Form 29 mortaria and noted that it might belong to a vessel of this form previously identified from the site (Campbell 2011). Amphorae The highest proportion of the sherds recovered during the recent investigations was from amphorae. Although the Mediterranean terminology (LRA1 etc.) is noted, these are principally listed using Radford classes of Bi, Bii etc, and as ‘B-ware’ within the general discussion. Of the 1821 sherds of post-Roman imported pottery, 914 were identified as LRA2, 375 as LRA1 and 19 as LRA3, while 118 were assigned to the ‘Bv’ category (Thorpe 2007, 2323). A number of ‘amphora stoppers’ were also recovered (Thorpe 2007, 236) as well as evidence of amphorae being modified into spindle-whorls (Barrowman et al. 2007, 87).

Of particular significance is the discovery of a handle classed as ‘Bv’ from the Middle Terrace which is reported to be the first from the site, and which allowed the first reconstruction of a ‘Bv’ vessel (Barrowman et al. 2007, 88; Thorpe 2007, 238-40, Figure 130; see Figure 3.6). However, this is not equated to any specific Mediterranean amphora type and no details of the fabric, surface, or dimensions are provided beyond the illustration and the general description of the ‘Bv’ group – discussed in more detail in Sub-chapter 3.2.4. A definite identification cannot be offered here, but it seems likely that this represents an Iberian rather than African amphora.

Although the quantity of LRA1 is therefore clearly lower than LRA2, the quantification is only given by sherd

Table 5. Relative proportion of amphorae at Tintagel by sherd count, vessel count and percentage. *includes ‘untyped and coarseware vessels’ of which Thomas assigns four a North African and 30+ an East Mediterranean origin. Sherd count from 1990-99 excavations (Thorpe 2007, 232-3)

‘Provisional site totals’ (Thomas 1988a)

Minimum vessel counts (Campbell 2011)

Sherd Count

%

Vessel Count

%

Vessel Count

%

LRA1

375

21.9

35

24.8

29

29.3

LRA2

914

53.3

40

28.4

34

34.3

LRA3

19

1.1

5?

3.5

3

3.0

‘Bv’

118

6.9

27+

19.1

23

23.2

‘Unknown’

288?

16.8

34*

24.1

N/A

-

‘LRA’

-

-

-

-

10

10.1

Total

1714

100

141

100

99

100

68

Imported pottery in Britain and Ireland: new evidence and understandings specialists familiar with contemporary Mediterranean coarse and cooking wares.

Coarsewares As well as the relatively large quantity of sherds recovered from Tintagel, the range of pottery has also added to the impression of Tintagel as of key importance within Atlantic systems and as exceptional to the other British sites. Following the sorting of the Tintagel material in the 1980s, a number of sherds that could not be matched to the typical fineware or amphora categories were identified as imported Mediterranean coarsewares; Thorpe noted that these might represent amphorae or bowls and casseroles (Thorpe 1988, 24). An extensive fabric series was established to classify these (Thorpe 1988, 25; 1993, 56; 2007, 233). Thorpe’s report of the pottery from the 1986 excavations in the Lower Ward includes examples of three of these groups, which are of various colours and said to resemble North African coarsewares, but as they contained mica an East Mediterranean origin was suggested (Thorpe 1988, 25). A wider variety of ‘coarseware’ fabric groups, including those labelled ‘Eastern Mediterranean Red Ware’, ‘Eastern Mediterranean White Surface Ware’ and ‘Eastern Mediterranean Sandy Cream Ware’ were identified from the excavations in the ‘Steps area’ in 1989-90 (Batey et al. 1993, 56). The significance of these coarsewares was first discussed at this time; they are noted to comprise ‘pans’ and ‘skillets’ and said to be ‘the sort of thing one would perhaps expect to find in a ship’s galley amongst other places’ (Thomas and Thorpe 1993, 59). Recommendations were made for future collaboration to identify these wares, although an East Mediterranean origin has been repeatedly suggested (Batey et al. 1993, 57). References have been made to a forthcoming report of the Tintagel fabric series that would describe these coarsewares (Batey et al. 1993, 56; Thorpe 1997, 78; 2007, 233), but this has not yet been published.

In 1988 Thomas suggested that a total of 30 ‘untyped and coarseware vessels’ of eastern Mediterranean origin and a further four of North African origin had been recovered, although it is not clear how many of these relate to unclassified amphorae (Thomas 1988a, 12). Thorpe’s report of the 1990-99 excavations recorded sherds from two of the ‘imported coarseware’ classes, as well as further ‘miscellaneous items’ that might represent unidentified amphorae or other imported wares. A single sherd from an unidentified thick-walled vessel is suggested to possibly represent an amphora not previously recognised at the site (Thorpe 2007, 233). Additionally, the report describes 287 sherds of ‘Unknown fabric’, two of which bear ‘cruciform decoration’. No further details of the colour, fabric or any indication of shape was provided, although a fragment of a cross is illustrated (Thorpe 2007). Clearly, this is a significant quantity of material that has potential parallels in the British or wider Atlantic assemblage and which deserves further research. Thorpe had previously referred to sherds of ‘miscellaneous’ types from Tintagel, including ‘squared corrugation ware’ and ‘closely grooved ware’ (Thorpe 1988, 25). It might be that parallels to the ‘LRA1 type’ amphora identified at Bantham (Sub-chapter 4.3.2) could be found within these unidentified groups. Chronology Thorpe’s recent report concluded that imported pottery was arriving at Tintagel ‘in some quantity from AD 450’ onwards with the bulk between c. 450-550, but also that imported Mediterranean pottery was also recovered in association with deposits on the lower terrace dated by radiocarbon to cal. AD 560-670 (95% confidence) (Thorpe 2007, 245). A potential continuation of Mediterranean imports into the later-sixth century at Tintagel would challenge the typical models presented for Britain, where Mediterranean imports are not thought to continue past c. 550. However, it must be noted that, currently, there is no evidence of this late date from the pottery. As noted, Campbell dismissed the presence of LRC 10 at Tintagel and, overall, the fineware imports cannot be dated past the mid-sixth century. The ‘function of the site’ is suggested to have changed with this break – and although the radiocarbon dates might hint at later importation, the absence of E ware indicated Tintagel was cut off from the ‘Continental import system’ (Campbell 2007a, 120).

Crucially, Campbell disagreed with this identification, stating that he could find no evidence of East Mediterranean coarsewares in the assemblage (Campbell 1991, 139). In 2007 he reiterated that only a very limited quantity of the sherds identified at Tintagel might actually be imported coarsewares, and stated that instead, the majority might be from amphorae – possibly from types that were previously unrecognised amongst the British assemblages (Campbell 2007a, 24). These contradictory opinions have two possible solutions – both of which have implications for the British imports as a whole. As Mediterranean coarsewares and cooking wares have now been identified on the Atlantic Seaboard (specifically at Vigo, Sub-chapter 6.3.2), it seems more plausible that Mediterranean coarsewares might have been recovered at Tintagel. Perhaps these formed a minor component of the imported wares, or alternatively might indicate the possessions of traders themselves at Tintagel? However, Campbell’s familiarity with the British material might suggest his conclusion was correct, and therefore that the British imported amphorae are more varied than has previously been accounted for in the literature on these wares. It seems that only a full review of these sherds will resolve this issue, and that this can only be carried out in consultation with ceramic

4.2.4 Discussion and recommendations It is clear that there remains considerable uncertainty about how much pottery has been recovered from the site and of which types. Specifically, what is lacking are published totals for the entire pottery assemblage from Tintagel, divided by type, form and variant (where identifiable), preferably with a range of quantification measures and full illustrations of diagnostic fragments. This would doubtless prove a significant undertaking, and of course the totals 69

Links to Late Antiquity sites with only a few sherds – or even one fragment – taken to represent the presence of a single imported vessel.

would only ever be a sample of the material originally imported. The amount of pottery discarded by Radford is unclear, as well as the amount remaining unexcavated. As mentioned, Thomas estimated that only approximately 5-10% of the Period II deposits had been excavated (Thomas 1988a, 12). It must also be remembered that the pottery is likely to signify a much greater quantity of archaeologically invisible commodities that were also imported. Although Campbell suggested that sherd counts and weights were misleading in assessing the relative significance of British import sites due to the ‘vagaries of the taphonomic processes at work’ (Campbell 2007a, 11), it might well be useful to compare sherd counts and weights between certain larger sites, specifically Tintagel and Bantham. A full catalogue for Tintagel using a variety of quantification measures would allow comparison of the sherd to vessel ratios from these two sites, which might reveal differences in deposition or consumption.

In 2001 a large quantity of pottery was excavated from the site at Bantham in South Devon (Reed et al. 2011). Although the assemblage is smaller than that from Tintagel the Bantham collection represents the first from a British site to be on a comparable scale. Unlike Tintagel, the ceramic report from the site was published with a breakdown of types and forms using a range of quantification measures (Bidwell et al. 2011). The considerable size of the assemblage has helped to rectify the prior dominance of Tintagel, while the specific details of the material provides an opportunity to reassess the nature of the British imports as a whole. It was clear from the publications that the Bantham material differed in many ways from the Tintagel assemblage, both in the condition of the sherds and the range of types. This suggested that a different pattern of supply to this region may have been in operation, or that established understandings of the range of ceramic material imported to Britain might need to be reconsidered. In this sub-chapter the range and significance of the pottery from Bantham have been considered within a wider case-study of imports to this region of the South Devon coast.

Finally, Tintagel also remains the last bastion of the original Radford/Thomas nomenclature, which has limited comparison between British and Continental scholars. A future recommendation would be that the pottery should be quantified using current Mediterranean classifications, referring to recognised variants, sub-types and fabric groups/production centres where possible. This is critical in order to allow effective comparisons between sites in Britain, Ireland and the Atlantic Seaboard.

As part of this case-study other sites in the region that had imported ceramics will be discussed (see Appendix B.4 for site locations). The relatively small ceramic assemblage recovered from the site at Mothecombe was assessed in 2012, following an examination of the published Bantham material. This allowed parallels to be made to the fabric groups and types identified at Bantham. In 2012 excavations carried out at High Peak in East Devon by a commercial archaeology unit produced eleven sherds of pottery, which were suspected to be of post-Roman date (Rainbird et al. 2013). As part of the analysis of these sherds, a review of the entire assemblage was carried out for Devon County Council – the results of which have been summarised and considered in relation to material from Mothecombe and Bantham. Finally, as part of this casestudy, the upper portion of an amphora that was recovered from a marine context near Cawsand in Cornwall will be discussed. Although this was discovered in the 1970s, it was not published at the time and was not included into catalogues of early medieval imports. Although the amphora was recovered just into Cornwall, the closest import sites to the find-spot are Mothecombe and Bantham, and the publication has therefore been summarised as part of this examination.

As discussed, the ‘rough total’ figures for the site suggested by Thorpe (150 amphorae, 80 fineware vessels) has been used wherever possible to indicate the approximate number of vessels recovered to date at Tintagel. However, as these are not subdivided in a way that would allow comparison, Campbell’s published totals have been used to frame discussions based on proportions of types or the presence/absence of specific variants. 4.3 Early medieval imports on the South Devon coast: a case-study for future directions in research 4.3.1 Introduction Although the precise character of the Tintagel assemblage remains to be clarified, it is clear that the site produced much greater amounts of Mediterranean sherds than any of the other British or Irish import sites. Consequently, as the research on the imports progressed, specific features of the Tintagel material came to define the British assemblage as a whole – particularly the focus towards LRA2. Where proportions of amphora types have been published (see Campbell 2007a, 23) the data from Tintagel has, unsurprisingly, skewed the impression of the British material. This difficulty reflects the long-standing nature of the British post-Roman Mediterranean import assemblage: dominated by one site with a very large quantity of pottery (relatively for Britain), followed by a ‘second’ tier of sites such as South Cadbury, Cadbury Congresbury and Dinas Powys with estimated minimum vessel counts of between, roughly, ten and twenty. Below this is a larger number of

4.3.2 Bantham As the research on the import sites developed, many came to be interpreted as centres of local political power – with the imported commodities associated with elite consumption, or seen to be somehow connected to the formation or maintenance of local power. The site at Bantham [125] is rather different, and instead came to be interpreted as 70

Imported pottery in Britain and Ireland: new evidence and understandings pottery, spindle-whorls, bone combs and implements, flint arrow-heads as well as metal items including fish-hooks (Elliot 1901, 477). In the mid-twentieth century Aileen Fox examined the pottery collected in the early nineteenth century, together with pottery that had been recovered in the early 1950s from ‘exposed occupation horizons’ (Bidwell et al. 2011, 85). She recognised these sherds as comparable to pottery that had recently been identified at Tintagel and Garranes (Fox 1955, 55). The site was therefore recorded in the earliest of Charles Thomas’ catalogues of British sites with post-Roman imports: Thomas 1959 (108-9) records LRA1 and other possible amphorae as well as E ware from Bantham; Thomas 1981 (12-21) lists two LRA1, two Miscellaneous amphorae (LRA) and one possible E ware vessel.

a ‘beach-market’ site, where people would meet seasonally to exchange goods. The recent excavations at the site, however, suggested that revisions to this interpretation might be necessary. A large and very varied ceramic assemblage was recovered, which not only reveals the importance of the site within Atlantic systems of exchange, but challenges conventional views of the character of the British imports and established models for their arrival. Bantham: history of investigations The site at Bantham lies on the south coast of Devon, where the River Avon flows into Bigbury Bay (Figure 4.2 shows the location of the various investigations conducted in the area, including the most recent 2001 excavation).

Excavations were carried out at two exposed areas of shell and bone in 1978 by Bob Silvester, at the south of the ‘camp’ area (Silvester 1981, 91). These were found to overly a series of temporary hearths and hollow features, which were linked to ‘wind breaks’ and ‘camp-fires’ associated with seasonal occupation. After their abandonment, rubbish deposits had been dumped on these features; these dumps produced the majority of the finds. These included bone, bone-combs and objects, iron and other scraps of metal including two penannular brooches (Silvester 1981, 96-106). Silvester’s report describes the twenty sherds of pottery from the site recovered up to that date, including those from the early twentieth century. These are typically small and worn, but notably include Roman and post-Roman sherds, a number of which have been turned into spindle-whorls (see Figure 4.3). A fragment of samian ware and of early medieval imported amphora both showed ‘abortive attempts at perforation’ (Silvester 1981, 105). This suggested that sherds might have been deliberately taken to this area of the site to be made into spindle-whorls (Silvester 1981, 114). The reuse of imported amphorae for the manufacture of spindle-whorls has been attested at other early medieval sites, including Gwithian and Tintagel (Thomas 1993, 68; Nowakowski et al. 2007, 45). The reused sherds from Bantham were LRA1 as well as earlier Roman sigillata/samian thought to have been collected from a nearby deserted Roman site (Silvester 1981, 105, 113). Whorls of bone and slate were also recovered.

The earliest mentions of the site date to the mid nineteenth century, when ‘cartloads’ of bones were recorded to have been removed from the site, probably from a marshy area to the west of the 2001 excavation (Silvester 1981, 89; Bidwell et al. 2011, 83). Around the turn of the nineteenth century a feature to the north was identified as a ‘camp’, although it was later suggested that this represented a natural dune formation (Elliot 1901; Jenkins 1902, Griffith 1997, 124). At the time two very extensive ‘kitchen middens’ were described to be on the Ham. The closest to the ‘camp’ was said to be frequently eroded by the sea – each high tide exposing a new section of midden and revealing numerous artefacts, including shells, split bones,

Overall, the site was seen to be different to post-Roman coastal settlements such as Gwithian, Cornwall and Tean in the Scillies – characterised, instead, by seasonal occupation: ‘not for decades but for short periods over several years’ (Silvester 1981, 114; Thomas 1985, 195). Silvester, following an original proposal by Fox, suggested that the site might represent a seasonal trading-station or ‘mart’ – where local people would converge to exchange commodities with merchants. Although he did not speculate on the mechanisms by which imported commodities reached the South West he proposed that the site was a primary entry-point for imported goods, that this demand indicated the region had products for exchange, and that this trade was probably controlled by a ‘well-established secular aristocracy’ (Silvester 1981, 112-4).

Figure 4.2. The location of investigations at Bantham. From Reed et al. 2011, Figure 1. Drawing by Tony Ives, Exeter Archaeology

71

Links to Late Antiquity

Figure 4.3. Bantham spindle-whorls. Courtesy of Torquay Museum. Identifications by David Peacock in Silvester 1981, 105-6: number 1 described as ‘similar’ to LRA1; 3 identified as LRA1; 2, 4 and 5 as unidentified pottery; 6 as samian ware; 7 and 8 as bone; 9 as slate

from context (501) – a thin, but extensive, deposit of charcoal rich material, with frequent bone and shell. In total 88% (471 sherds) of the stratified pottery came from, or could be joined to, sherds from this deposit (Reed et al. 2011, 88). The authors suggest this represents a layer of material eroded from underlying charcoal rich midden deposits and hearth-fills. The good condition of the pottery suggested that it had only been exposed for a short time before it was covered by wind-blown sand (Reed et al. 2011, 89). A similarly formed ‘occupation horizon’ (521) was identified within a later phase (Phase 6) which produced ironworking slag. As well as Mediterranean pottery this layer produced three sherds of E ware (Reed et al. 2011, 92). Smaller quantities of pottery were recovered from other deposits, some of which were considered to be residual.

Further small investigations were carried out at the site in 1982 and 1997. The latter did not produce any postRoman features or artefacts, but revealed a dry-stone enclosure of ‘probable prehistoric or Roman date’ and 38 sherds of second to fourth century pottery (Griffith 1998, 122-5). However, Griffith’s salvage recordings of drainage works at the mouth of Buckland Stream in 1982 produced further deposits containing bones and shell and a hearth (Griffith 1986, 39-42). Radiocarbon dates were taken from bone and charcoal from within the hearth, which dated to cal. AD 350 ± 95 and 605 ± 90 (Griffith 1986, 43-4). A sherd of LRA1 was found immediately above this hearth and reported to be in situ (Griffith 1986, 43). The most recent excavation in 2001 took place at the eastern, seaward side of the peninsula, close to the location of Griffith’s 1982 recording, and was carried out as part of the rebuilding of a surf clubhouse (Reed et al. 2011). The demolition of the original structure left an almost vertical section revealing a sequence of charcoal-rich deposits and wind-blown sand underneath 1.5m of dune sand (Reed et al. 2011, 85-6). Various layers of occupation deposits were identified which were interspersed by layers of windblown sand. These were cut by phases of hearths, pits and stake-holes suggesting phases of temporary use (Reed et al. 2011, 87-92). A fill from a hearth in the earliest phase of occupation (Phase 1) produced a radiocarbon dates of cal. AD 430-600 (at 95% confidence); a sample from a stratigraphically later hearth (Phase 5) produced a date of cal. AD 535-640 (at 95% confidence) (Reed et al. 2011, 88-92).

Within the report attention is drawn to the differences between the artefacts recovered in 1978 and 2001 (Reed et al. 2011, 128). Although the 2001 excavation produced over 2000 fragments of bone, nearly double the quantity was recovered in 1978. The earlier excavation, however, produced very little pottery, apart from the reworked sherds and spindle-whorls – but did produce a greater range of metal and bone artefacts. The authors suggested that this indicates different functional zones at the site. Silvester’s 1978 work revealed an area of ‘manufacture and occupation’, while the 2001 excavations revealed an area without buildings where the contents of the amphorae were consumed (Reed et al. 2011, 128). It might also be possible that the contents were decanted here for further redistribution. The large quantities of imported pottery recovered in the recent excavation, together with features suggesting per-

Large quantities of imported pottery were recovered from the excavation of these layers and features, principally 72

Imported pottery in Britain and Ireland: new evidence and understandings manent occupation and craft-working led to a revised interpretation of the site. Rather than a ‘beach-market’ site of temporary seasonal gathering and exchange, the authors advised that Bantham could be seen as a place of permanent occupation, craft-working and production, and as a primary import site for international trade in the post-Roman period (Reed et al. 2011, 129-31). They proposed that Bantham should now considered as a ‘port’, connected to Atlantic trading channels, with the middens of bone and amphorae indicating the communal feasting practices of its residents (Reed et al. 2011, 132).

report (Bidwell et al. 2011, 94, 110). Although the earlier finds from the site were examined during the assessment, the authors did not match any of these to the vessels identified from 2001. This is probably due to the small, worn or reworked condition of the earlier finds. Unlike the imported pottery at many of the British sites, particularly Tintagel, the pottery from the 2001 excavation is in very good condition and is characterised by relatively large and unabraded sherds. This allowed a level of identification unusual for sites in Britain with post-Roman imported pottery.

Early medieval pottery from Bantham

The ceramic assemblage was consequently published in considerable detail, using terminology current in the Mediterranean and with detailed quantification measures, including sherd count and weight (both only presented as a percentage of the total) and MNV (Bidwell et al. 2011, 93, Table 1). The report also indicates that the vessels were quantified by estimated vessel equivalents (EVE) based on rim circumferences.

A total of 719 sherds were recovered in 2001, with a weight of 20.45 kilograms and representing an estimated 63 vessels. 692 of these sherds were identified as amphorae, representing 52 vessels and therefore the vast majority of the assemblage. Table 6 shows a summary of the 2001 excavation assemblage, based on tables in the published

However, a number of vessels that are only represented by base or body-sherds are assigned an EVE value, and therefore it must be presumed that the headings in the report’s Table 1 are in the wrong order, and that the figures provided are minimum vessel counts. This assumption has been followed in the production of Table 6 and in the following discussion. It is worth commenting that the complete ceramic data is not available online, as indicated in the report, and that is not possible to establish sherd counts for many of the identified vessels.

Table 6. Early medieval pottery recovered at Bantham in 2001. After Bidwell et al. 2011, Table 1

Class

Type

Vessel Count

Finewares

LRC

2

Oxfordshire Parchment Ware

1

Unidentified

2

‘Byzantine Redware’

1

E ware

3

South Devon ware

1

‘Vessel rich in muscovite’ *

1

Coarsewares

Amphora

As part of this doctoral project the pottery excavated at Bantham in 2001 and held at Plymouth City Museum and Art Gallery was examined. The pottery represented comparative material for the analysis of the Mothecombe and High Peak ceramic finds and also provided a ‘test-case’ to evaluate a large, complex and recently studied British assemblage. During this examination, the previously sorted bags of pottery were matched to the identified vessels from the 2011 report. It must be noted that many of the illustrations of amphorae provided in the 2011 report (some of which are reproduced below) represent composite reconstructions and do not wholly reflect the assemblage as can be viewed in the museum (Bidwell et al. 2011, 93). The finds from earlier investigations at the site, held at Torquay Museum, were also inspected.

% LRA1

20

38.46

LRA1 ‘Bantham-type’

6

11.54

LRA2

1

1.92

East Mediterranean

5

9.62

LRA4

2

3.85

‘Thick-walled’ /Bv

5

9.62

North African

2

3.85

Unidentified

11

21.15

Total amphorae

52

100

Amphorae The 2011 report identified fifty-two imported amphorae of diverse types. The majority of these (nineteen vessels) were LRA1, which were classified into three broad fabric groups based on colour and inclusions. These were not matched to specific regions of production in the report, but the variation in colour and fabric suggests that LRA1 from various locations in the East Mediterranean might be represented. Although the LRA1 from Bantham were not considered by the authors to be inherently datable, the illustrated shapes match most closely to Pieri’s ‘transitional

*(not included within the original quantification) 73

Links to Late Antiquity

Figure 4.4. LRA1 (left) and LRA1 ‘Bantham type’ (right) from Bantham. After Reed et al. 2011, Figures 14.1 and 17.21

LRA1. In stark contrast there was only a single, small body-sherd of LRA2. One possible sherd of LRA2 was identified by Thomas from the earlier finds at the site (Silvester 1981, 105), although this could not be verified by the present author from the material held at Torquay

LRA1A’ or LRA1B1 type, and therefore would best fit a date in the later-fifth and earlier-sixth centuries (Pieri 2005, 70-5; see Figure 3.2). One of the vessels had a painted dipinto on the shoulder; this was not translated with certainty but may have noted the capacity of the vessel (Bidwell et al. 2011, 103). Two more had graffiti that were incised after firing. The presence of an additional vessel was identified from a disc cut from an amphora for use as a stopper (Bidwell et al. 2011, 97). There were also sherds from six other vessels that had the general form of LRA1, but with certain differences, specifically a soft, powdery, pale cream-coloured fabric, distinctive arched handles, thin necks and a sharp angle between the neck and body. Perhaps most distinctively, they bear grooving at the neck of the vessels which gives the effect of ribbing. These are described to have been made with a ‘flat-ended implement’ (Bidwell et al. 2011, 99). Although these were recognised as a distinctive ‘sub-type’ of LRA1, the origin of these vessels was not ascertained. It was noted that there were no similar vessels within the Tintagel assemblage, but that similar sherds had been recovered at High Peak (Sub-chapter 4.3.4) and possibly at Wroxeter (Bidwell et al. 2011, 99). Figure 4.4 and Figure 4.5 show Bantham ‘vessel number 1’, a ‘typical’ example of LRA1, and ‘vessel 21’, one of this variant ‘LRA1-type’. Parallels have been observed between this ‘Bantham-type’ and some of the LRA1 recovered at Vigo in north-west Spain (Fernández Fernández, pers. comm), although this remains to be fully established. It is likely that these represent an alternative source than the ‘standard’ LRA1 found elsewhere on the Western Seaboard. Significantly, the 2001 assemblage was dominated by LRA1 (see Table 6). Including the variant LRA1 ‘Bantham-type’, 50% of the vessels were identified as

Figure 4.5. LRA1 (top) and LRA1 ‘Bantham type’ (bottom) from Bantham. © Plymouth City Council (Museums Galleries Archives)

74

Imported pottery in Britain and Ireland: new evidence and understandings Museum. Although Campbell’s 2007 synthesis predated the final publication of the Bantham material, he had been informed of the findings and noted that Bantham was unusual in having an assemblage dominated by LRA1 with few LRA2 (Campbell 2007a, 26). He proposed that there might be a chronological explanation for this (imports veering between LRA1 and LRA2 at slightly different times), that it might represent differential distribution at the site, or might possibly represent a ‘specific supply mechanism’ (Campbell 2007a, 26, 103). The authors of the 2001 report also noted that the high proportion of LRA1 might indicate a generally early date for the assemblage (Bidwell et al. 2011, 94; 112). Reynolds, however, suggested that the absence of LRA2 at Bantham simply demonstrated that cargoes of LRA1 and LRC could arrive without LRA2 (Reynolds 2010, 110).

to North African amphorae. The authors propose that a handle found by Jenkins in the early twentieth century might be related to this type, which they suggest is similar to handles on the North African Keay 62. However, the authors also indicate that the fabric of these vessels is similar to those classed as ‘Bv’ at Tintagel, which are said to have an ‘unknown eastern Mediterranean source’ (Bidwell et al. 2001, 102). As it is increasingly likely that the ‘Bv’ at Tintagel have an Atlantic source, it seems plausible that these ‘thick-walled’ vessels from Bantham might also represent Iberian imports – and would therefore reflect connections between this region and sites in Devon in the fifth or sixth century. Future research is necessary to compare the Bantham ‘thick-walled’ type to the Tintagel ‘Bv’, and then to equate these with contemporary vessels produced in Spain or Portugal.

As discussed in Chapter Two, the assertion that LRA2 was disproportionately high among the imports identified in Britain has marked the British assemblage as distinctive from patterns in the West Mediterranean – and has reinforced the model of direct supply. Although LRA1 is the most common amphora found in early medieval contexts in Britain, it is usually accompanied by LRA2. The proportions attested at Bantham would, therefore, be closer in-line with contemporary patterns in the West. It seems possible, furthermore, that the Bantham assemblage attests to a different system of supply focused on South Devon, to that which supplied Tintagel.

Five amphorae were assigned a probable East Mediterranean origin but could not be matched to published types. Eleven further amphorae were grouped that could not be tied to specific types or origins and were recorded as ‘unidentified’ amphorae. These two categories clearly represent a significant proportion of the overall assemblage (nearly a third of the vessels). It seems possible that future collaboration between researchers at Atlantic sites will allow new identifications to be made for these and other unsourced amphorae from Insular sites. The established model of direct contact with the East Mediterranean is, naturally, based on those vessels that could actually be identified. As the range of identified types increase, a different and more complex picture of this exchange seems likely to emerge.

Beyond the proportion of LRA1 and LRA2, other features of the Bantham assemblage were unusual for an early medieval site in western Britain. Firstly, the amphorae included two examples of LRA4, although the quantity of mica in these suggested they might be imitations of Gazan vessels (Bidwell et al. 2011, 100). LRA4 has not been seen as a typical feature of the British imports. Campbell indicates that it has only been recovered at Bantham, as well as a possible, abraded rim from Dinas Powys (Campbell 2007a, 22).

Finewares Two vessels of LRC were identified from Bantham, both of the typical Hayes form 3 (see Figure 4.6). One vessel was identified as Form 3 ‘small variety’ which the authors suggest might have a rather early date (Bidwell et al. 2011, 106, 111). The closest example in Hayes’ LRP (Hayes 1972, 334.32) is said to be related to LRC 4, which is assigned a possible date of production in the second quarter of the fifth century (Hayes 1972, 338). However, the illustration shows the outer face of the rim to be convex, rather than concave, which might suggest a later date to be more

Secondly, at least two amphorae of North African origin were identified (Bidwell et al. 2011, 94). As discussed, the British ‘Bv’ group has traditionally been assigned a North African origin, although it might be that some, or even most, of these were actually produced in Spain or Portugal (Sub-chapter 3.2.4). However, a couple of sites have produced sherds that are more likely imports from North Africa. As well as the six sherds from Bantham, which were not illustrated in the 2011 report and could not be matched to a specific type of amphora, a sherd from a North African amphora was identified at Mothecombe. Finally, a particularly interesting amphora group from Bantham is the so-called ‘thick walled’ amphorae, of which five vessels were identified from the 2001 assemblage. These had walls up to 250mm thick with ‘buff-pinkish, sandy fabrics’ and broad, shallow rilling on the exterior (Bidwell et al. 2011, 100). The reconstruction suggested that the vessels were cylindrical, and similar

Figure 4.6. LRC 3 from Bantham: ‘small variety’ (top), 3F (bottom). From Reed et al. 2011, Figure 23.52-3

75

Links to Late Antiquity fabrics. Examples of this local, micaceous, hand-made ware were illustrated in the recent monograph from Vigo (Férnandez 2014, 525, Appendix 3.2, 7-8). Férnandez Férnandez has recently viewed the sherd from Bantham and agreed that Vigo would be the likely origin, although further analysis is necessary to verify this attribution. If it can be established that that this sherd was produced in Vigo, this would represent the first pottery produced in north-west Spain to be identified from an early medieval site in Britain or Ireland – and would confirm some sort of link between Galicia and south Devon. This adds considerable weight to the argument for north-west Spain being an important ‘stopping-off point’ for shipments heading toward Britain, or – as is more likely – that Vigo acted as a redistribution point within the Atlantic exchange systems. As only a single vessel was identified it seems quite plausible that it represents the personal property of traders arriving from Vigo, or was exchanged for use at Bantham itself. It is not clear how many sherds of this vessel were recovered at Bantham, although they are reported to have been found in the large deposit (501), apart from two sherds recovered from hearth (541).

likely. The other vessel was identified as LRC 3F, which has previously been identified in Britain and is dated by Hayes to the second quarter of the sixth century (Hayes 1972, 338; Bidwell et al. 2011, 106). The publication does not record any ARS from the site, although a body-sherd that represents one of the two ‘unidentified’ fineware vessels bears similarities to ARS-D fabric and the original label on the finds-bag notes ARS as a possibility (Bidwell et al. 2011, 106.56). Coarsewares One vessel was identified as ‘Eastern Mediterranean Red Ware’ and compared to the imported Coarse Ware Fabric 1 identified at Tintagel (Bidwell et al. 2011, 106). As discussed, these fabric categories have not been fully published. Seven sherds of E ware, representing a minimum of three vessels, were found during the 2001 excavations. The forms were not identified, but were thought to be from closed vessels (Reed et al. 2011, 111). Two of the vessels were from the ‘upper sequence of dune deposits’, which fits the interpretation of E ware as a later feature of the imports to Britain. Three sherds, however, were recovered from the stratigraphically earlier deposit (501), although these might have been intrusive (Reed et al. 2011, 111). However, the authors noted that E ware might still have arrived at the site at an earlier date than in other parts of Britain, and might, therefore point to long-standing connections between Aquitaine and Devon (Reed et al. 2011, 111). Nevertheless, given the large quantity of amphorae, there was relatively little E ware. It might be that, as at Tintagel, Bantham was not well integrated into the later phase of Continental imports. The complete absence of DSP from Bantham points to a longer-term focus towards the Mediterranean wares and the associated supply-routes.

It should be noted that the 2001 excavation also produced seven fragments of imported glass, mostly from the deposit (501) that produced the majority of the pottery. Cool identified these as representing two imported vessels, rather than cullet, and as probably linked with the consumption of imported wine at Bantham as part of ‘elite drinking rituals’ (Cool 2011, 117-9). Six fragments came from a probable beaker of pale-green glass fitting Campbell’s Atlantic ‘C/D’ group of likely Continental origin (Cool 2011, 118-9). Campbell suggests an origin for this group near the source of DSPA and E Ware, but indicates that these glass imports might have spanned both of his phases of Mediterranean and Continental ceramic importation (Campbell 2007a, 68, 73). The other find was a body fragment of blue-green glass that could not be securely identified. The significance of the Bantham assemblage

A vessel of local ‘South Devon Ware’ was identified as well as two other coarseware vessels that were not included in the original quantification (Bidwell et al. 2011, 107). One was a Gabbroic ware of probable Cornish, postRoman manufacture. The origin of the final vessel was not ascertained, but might have particular significance for understandings of the site and wider mechanisms of supply in the Atlantic region. At least three sherds were found of a micaceous handmade vessel, including a complete profile sherd (Bidwell et al. 2011, Figure 24.63). This was described as a ‘vessel rich in muscovite’ on the basis of its principal inclusion (Bidwell et al. 2011, 108; see Figure 4.7). No parallels to the vessel were apparent from sites in the South West, including Tintagel, and although connections were made to broadly similar fabrics in Brittany, the authors concluded that the origin remained an ‘unresolved problem’ (Bidwell et al. 2011, 108).

Although the sites seem to have functioned as very different settlements, the excavations at Bantham in 2001 demonstrated that Tintagel was not the only site in Britain

During the course of this project, it was possible to inspect vessels of local manufacture recovered from excavations at Vigo in north-west Spain, which had noticeably similar

Figure 4.7. Vessel from Bantham described as ‘rich in muscovite’. After Bidwell et al. 2011, Figure 24.63

76

Imported pottery in Britain and Ireland: new evidence and understandings The first survey at Mothecombe was carried out after the discovery of sherds of pottery in dark deposits eroding from the back of Meadowsfoot beach in 1959, which were recognised as imported amphorae of post-Roman date (Fox 1961). The features were interpreted as temporary hearths and ‘shanties’, constructed by traders on at least two occasions (Fox 1961, 80). Immediate parallels were drawn to Bantham, and the site was likewise interpreted as a ‘beach-market’. The finds were included in Thomas’ later catalogue of early Mediterranean imported pottery (Thomas 1981).

to receive a large number of amphorae. The quantity at Bantham is much greater than other sites in the South West, such as Dinas Powys or South Cadbury (see Appendix G). Nevertheless, Thorpe’s estimate of a total of 150 amphorae at Tintagel is still three times as large as the 52 vessels identified at Bantham. However, the sherds from the recent Bantham excavations were typically larger (Bidwell et al. 2011, 93). This certainly suggests differences in the deposition of amphora sherds at the two sites, but might also indicate that the Bantham assemblage would compare more favourably to that from Tintagel if figures for total weight by type were available for both sites. Similarly, a comparison of sherd counts per square metre excavated could be informative. Such comparisons would require more detailed information from Tintagel, but it might be that, using alternative quantification measures, the dominance of Tintagel within the British assemblage would be much reduced. The 1978 and 2001 excavations at Bantham only examined a small area, and the authors indicated it was likely that much more could be found at the site (Bidwell et al. 2011, 109). It was suggested that if a larger area had been excavated at Bantham the assemblage would equal that from Tintagel. However, as previously noted, it was estimated that only 5-10% of the post-Roman levels at Tintagel have been excavated. Regardless, the range of material identified at Bantham, beyond the ‘typical’ range of material from British sites, demonstrates the considerable importance of the assemblage.

In the 1990s further finds of animal bone and charcoal were observed eroding from the beach, and the site was surveyed in 1992 (Agate et al. 2012, 351). Various phases of investigation and recording were carried out thereafter to characterise the archaeology at the site and to assess the extent of erosion (Agate et al.2012, 351). Over this period a small number of sherds were collected from the beach area. Excavations were carried out at the site between 2004 and 2011 by Newcastle University and the University of York. A number of successive hearth-features were revealed from the east side of the beach (Area 1; see Figure 4.8) which again, indicated temporary occupation at the site (Agate et al. 354-7). At the west of the beach (Area 2) the earliest phase of activity comprised hearth features and a robbed structure that suggested possible craft-working. However, most significantly, two later phases revealed truncated, horseshoe-shaped cuts that were interpreted as the foundations for two successive, large timber structures (Agate et al. 2012, 358-64). Sherds of early medieval pottery and reused sherds of Roman pottery were recovered in association with and covering the hearths in Area 1, and in deposits associated with, and overlying, the building foundations. The discovery of these large timber structures at Mothecombe provided the first indications that the site was permanently occupied and that, like Bantham, it was a consumption site for imported commodities such as wine (Agate et al. 2012, 382). A single sherd of pale green glass of early medieval date was also recovered, although this was identified as belonging to Campbell’s ‘Germanic’ Group B, rather than with the Atlantic group of imports (Campbell 2012, 379).

The identification of the LRA1 ‘Bantham-type’ represents the first example of this variant amphora type from Britain, of which there might be equivalent examples in Spain. An Iberian connection might also be also demonstrated by the ‘thick-walled’ amphorae, although this remains to be established. Perhaps most significantly, the possible identification of a coarseware vessel from Vigo hints at some sort of connection between the ‘beachmarket’ or ‘port’ at Bantham and north-west Spain. 4.3.3 Mothecombe This sub-chapter summarises the ceramic report and discussion published following the 2004-11 excavations at Mothecombe (Agate et al. 2012; Duggan 2012). The ceramic assemblage from Mothecombe [124] is much smaller than that from Bantham but reveals a number of similarities in terms of the types of pottery recovered. Specifically, the site also produced a higher proportion of LRA1 to LRA2. As discussed, this emerging pattern might represent a regionally specific supply or variations in the chronology of importation but might also suggest the British assemblage to be less ‘unique’ than previously considered.

Pottery recovered from Mothecombe prior to 1995 The pottery found at Mothecombe prior to 1995 is held in Totnes museum and was re-examined in 2011. A number of the identifications were changed from those presented in the original publication from the site (Fox 1961) and Thomas’ 1981 catalogue. The collection at Totnes comprises the sherds that were collected from the site in 1959 (Fox 1961), as well as three fragments of amphora handles found at the north end of Mothecombe beach. These had been previously identified as LRA2 handles (Turner and Gerrard 2004, 173).

Mothecombe: history of investigations Two granite-tempered sherds were collected from the back of the beach in 1960 and identified as locally produced ware, highlighting the question of the local ceramic

The site at Mothecombe is also located in South Devon, to the west of Bantham, at the mouth of the River Erme. 77

Links to Late Antiquity

Figure 4.8. The location of investigations at Mothecombe. From Agate et al. 2012, Figure 4

context in the fifth and sixth centuries (Fox 1961; Turner and Gerrard 2004). Although undated, they were found in general association with LRA1 sherds and might represent contemporary vessels. Gabbroic pottery has demonstrated continuity in early medieval pottery production in Cornwall, but there is no evidence for contemporary ceramic production in Devon (Thorpe and Wood 2011, 276-80). A gabbroic pot of post-Roman date was recovered at Bantham, which was thought to have arrived with Mediterranean imports shipped eastward from Cornwall (Reed et al. 2011, 116). This suggests that despite the lack of ceramic production in Devon, the sites at Mothecombe and Bantham had some limited access to British products, as well as imported, wares.

The Roman Pottery Although no features of Roman date were identified, the excavations produced six small and abraded sherds of Roman pottery, widely ranging in date, and representing regional imports and vessels imported from Gaul. Two fragmentary and one complete ceramic spindle-whorls were also recovered (Figure 4.9), which represent reused Roman vessels. The discovery of these whorls in association with early medieval imports, and without obvious Roman activity at the site, suggests that interesting sherds may have been deliberately collected and brought to the beach from elsewhere. Roman pottery of earlier, date has been found in association with later imported Mediterranean wares at other early medieval import sites. At Trethurgy some sherds of samian were deposited long after their period of manufacture, including one sherd in association with early medieval imports, indicating either a very long sur-

The post-1995 assemblage (including the 2004–2011 excavations) This combined post-1995 assemblage (Table 7) comprises a total of sixty-two sherds of pottery with a total weight of 443.2 g (Table 5). The sherds were generally small, and many were abraded. The average weight of the Roman sherds (including the spindle-whorls) was 3.7 g and of the imported amphora sherds 8.7 g. The detailed identification of types was limited by the size and condition of the pottery, and by a lack of diagnostic sherds. This also prevented quantification by estimated vessel equivalent (EVE). Instead, the assemblage was quantified by count, weight and minimum vessel number. The latter was inferred by joining sherds and sherds of the same fabric presumed to derive from the same vessel. Fabric descriptions, based on visual study at ×20 magnification for the Roman and early medieval sherds, are provided in Appendix D. Colours were described by Munsell number and colour name.

Table 7. Quantification of all pottery recovered at Mothecombe since 1995

78

Period

Sherd Count

Total Weight (g)

MNV

Roman (includes 3 spindle-whorls)

9

33.7

7 (includes spindlewhorls)

Imported amphorae

18

156.1

5 early medieval, 1 uncertain

Medieval and post-Medieval

26

238.5

19

Unidentified

9

14.9

1

Total

62

443.2

33

Imported pottery in Britain and Ireland: new evidence and understandings

Figure 4.9. Mothecombe spindle-whorls. From Agate et al. 2012, Figure 14

vival in use for certain vessels or the curation of selected sherds (Quinnell 2004, 99). The collection and reuse of Samian was identified by Cool as a typical feature of late Roman sites; she suggested that the red colour might have had some social or religious significance (2000, 53-4). The source of the Roman sherds at Mothecombe is not certain, although Roman material has been recovered in the general vicinity. A scrap of possible samian was found c. 2.5 km away at Oldaport, supporting a Romano-British date for the first of two phases of defensive earthworks there, although, since the sherd was highly abraded, it might also be residual (Farley and Little 1968; Rainbird 1998, 158). Excavations at Bantham in 1997 revealed a possible Romano-British enclosure as well as 38 sherds of second to fourth-century pottery, including samian and blackburnished ware (Griffith and Reed 1998). Roman activity has also been identified at Mount Folly, located between Mothecombe and Bantham (Reed et al. 2011, 130).

this purpose (Campbell 2007a, 87). Although spinning is only the assumed function of these items, their presence at Mothecombe hints at the processing of wool, possibly with associated cloth production, as part of the activities of daily life (Croom 2011, 58). These ceramic spindle-whorls also reveal that sherds of Roman and early medieval pottery – possibly having some specific significance (considered exotic or attractive?) – could remain in circulation with a functional importance long after their original production date, or in the case of the amphorae, the consumption of the contents. This, naturally, has implications for the use of these sherds for dating, but is also suggestive of a continuing social relevance for certain ceramic items when pottery might otherwise be a rare material. Campbell has proposed the possible structured deposition of pot-sherds at Cadbury Congresbury and Trethurgy, suggesting the imported pottery may have acquired some ‘talismanic significance’ (Campbell 2007a, 97). However, he also suggested that other samian sherds found at Dinas Powys might have been intended for reuse in industrial processes, specifically as a red colourant (Campbell 2007a, 88).

As discussed in Sub-chapter 4.3.2, Silvester’s excavations at Bantham revealed evidence of the on-site manufacture of spindle-whorls (Silvester 1981, 97–106). It may be that the spindle-whorls found at Mothecombe were made locally, perhaps even at Bantham, or produced on site from pottery recovered in the vicinity. Fragments of whorls made from sherds of samian were also found at Dinas Powys; the sherds had presumably been brought to the site for

Early medieval pottery recovered at Mothecombe The assemblage collected after 1995 contains eighteen sherds identified as amphorae of which sixteen were identified as early medieval imports – representing a minimum 79

Links to Late Antiquity

Figure 4.10. Imported amphora sherds from Mothecombe: P5a and P6a (left), P7, P8a, b and P9 (right)

of five vessels (see Figure 4.10; descriptions in Appendix D). None had traces of dipinti or graffiti, or showed evidence of reuse. Although the previously published amphora sherds from Mothecombe were viewed, no joins or obvious similarities in fabric could be identified. As such, a revised total minimum vessel number for imported amphorae at Mothecombe is suggested in Appendix G. There is considerable variation in the fabrics of the four LRA1 vessels identified. This is similar to Bantham, where the LRA1 amphorae exhibited a variety of fabric types, and is likely to indicate a range of production sources (Campbell 2007a, 19). The lack of diagnostic sherds prevented close identification on the basis of developments in the shape of LRA1, as discussed in Sub-chapter 3.2.

is relatively large (possibly ten vessels), and compares favourably to the quantities identified at High Peak, Trethurgy, Dinas Powys and Cadbury Congresbury (see Appendix G). Campbell has proposed that the vessel-tosherd ratio represents a much more informative measure than sherd counts for comparing patterns of consumption and disposal at early medieval import sites (Campbell 2007a, 86-92). The sherd-to-vessel ratio for the amphorae at Mothecombe is particularly low at 1:2.4, in comparison to 1:13.3 from the 2001 Bantham excavations (Bidwell et al. 2011). These ratios suggest that at Bantham the vessels were discarded largely whole, although some sherds may have been reused. In contrast, the Mothecombe excavations seem to have produced isolated fragments that represent a range of different vessels.

The significance of the Mothecombe assemblage The Mothecombe assemblage would seem, in general, to comprise a comparatively small number of worn bodysherds from a relatively large quantity of vessels. The assemblage appears to attest to a relatively high number of imported wine-amphorae which arrived at the site from a range of export centres. The imported commodities may have been consumed by the residents of the large timber buildings, or else redistributed as part of local exchange systems. The presence of imported glass and wine implies that the inhabitants of the large structures might have displayed status or exhibited long-distance connections through drinking practices.

Apart from the handles identified in 2004, no further sherds of LRA2 were recovered from the excavations at Mothecombe; as at Bantham, LRA1 seems to have been the main import. Overall, the range of material observed at Mothecombe is similar to that from Bantham. There were no imported fineware vessels and no E ware; the quantity of both was low at Bantham. Neither assemblage contained DSP. Despite the very small sherd count at Mothecombe, the estimated quantity of amphorae suggested for the site 80

Imported pottery in Britain and Ireland: new evidence and understandings vestigations are recorded in the accounts of Sidmouth antiquarian and diarist P.O. Hutchinson (Hutchinson 1871-2). These diaries have recently been digitised by the Devon Record Office and the ‘In the Footsteps of Peter Orlando Hutchinson Project’ and are available online. The diaries record the recovery of pottery and bone from the site by Hutchinson in the mid to late nineteenth century, and by a contemporary antiquarian Rev. R. Kirwan, who carried out limited excavations at the site.

The mechanisms for the arrival of goods at Mothecombe are not clear, but it seems likely that the supply was associated with the shipments reaching Bantham via Atlantic channels. As discussed, the Bantham assemblage might reveal Iberian connections, specifically to north-west Spain, and it might be that the vessels at Mothecombe arrived from the East Mediterranean in shipments that stopped at an Iberian port – or else were redistributed from Vigo and arrived with traders from Galicia. At the same time, it is entirely possible that the residents of the buildings at Mothecombe had no direct contact with traders operating in the Mediterranean or Atlantic, but were connected to these systems through local interactions.

As part of this review, 28 sherds within the RAMM collections in Exeter were directly tied to these antiquarian investigations. These are all small body-sherds and almost entirely comprise sherds of Late Roman Amphorae. Their condition ranges from abraded to good. Fifteen sherds had paper stamps bearing ‘High Peak’ and the initials ‘RK’ and it is likely that these are some of the fragments found by Kirwan around 1871. However, as Hutchinson’s diary records Kirwan exhibiting fourteen ‘patterned’ sherds as well as plain sherds, it might be that further sherds were also found by Kirwan. At some point, Hutchinson made a detailed illustration of nine sherds from High Peak, although some of these also bear ‘RK’ stamps. Hutchinson’s diaries also indicate that a number of sherds had been recovered at the site by him from 1859 onwards.

4.3.4 High Peak, Sidmouth A reassessment of the early medieval ceramic assemblage from High Peak, Sidmouth [111] held at RAMM (Royal Albert Memorial Museum), Exeter was conducted in December 2012 for Devon County Council (Duggan 2013a). At the same time, an assessment was made of a number of suspected early medieval imported sherds, which had been recovered from the site during excavations in the summer of 2012 by AC Archaeology (Duggan 2013b; Rainbird et al. 2013). The results of this reappraisal have been summarised; the primary ceramic analysis is presented in Appendix E.

There are also two sherds in the collection that are catalogued as ‘excavated by D’Urban’.

High Peak: history of investigations G.E.L. Carter 1929 The site at High Peak, although now situated on the coast, is very different in character to Bantham and Mothecombe, representing a Neolithic site that was reoccupied in the post-Roman period. The site is situated on a cliff-summit to the south-west of Sidmouth in east Devon (see Appendix B.4). Although most of the site has been lost to coastal erosion, the surviving ramparts and finds indicate that High Peak may have been a major, high-status hillfort site in the fifth or sixth centuries (Campbell 2007a, 122). The reuse of prehistoric hilltop sites as fortified settlements in the early medieval period has been previously recognised in south-west Britain, most notably at South Cadbury and Cadbury Congresbury in Somerset (Rainbird et al. 2013, 48).

Excavations at the site were carried out by G.E.L. Carter in 1929 (Carter 1930). His report mentions the discovery of ‘Iron Age pottery’ at the ‘foot of the vallum under an accumulation of debris’ and ‘Romano-British ware’ (Carter 1930, 119). However, he added that ‘the remains of the Romano-British period are very slight and do not include any of the finer wares’. In Pollard’s 1966 report, Charles Thomas assigned twelve sherds of early medieval date to Carter’s 1929 excavation (1966, 53). However, the two sherds assigned to Carter that were illustrated in the report (Pollard 1966, Figures 12.3 and 12.10) both have original ‘RK’ stamps and were presumably deposited at RAMM in the nineteenth century. At the time of Thomas’ assessment, the sherds illustrated by Hutchinson (Pollard describes them as ‘Hutchinson’s finds’) could not be located in Exeter Museum; instead these were described and quantified from the illustration alone (Pollard 1966, 35). The quantity assigned by Thomas to Hutchinson is nine, which is the number of sherds in Hutchinson’s nineteenth century illustration.

Artefacts from High Peak have been recovered from the site in a number of intermittent phases, although it was not until excavations in the 1960s that sherds from the site were recognised as belonging to imported amphorae of post-Roman date (Pollard 1966; see Figure 4.11). This material was assessed by Charles Thomas, and the site was added to subsequent catalogues of Mediterranean imports (Thomas1981).

It seems that Thomas was led to believe that the sherds which could be located in the museum belonged to Carter’s excavation, rather than representing additional ‘antiquarian’ finds. This suggests that Thomas did not, in fact, assess any sherds from Carter’s excavations. The location, type and date of any sherds recovered by Carter are therefore uncertain.

‘Antiquarian’ investigations at High Peak The material from High Peak that was found prior to Pollard’s excavations (1961-4) was collected during antiquarian investigations of the site in the mid to later-nineteenth century. Many of the details of these in81

Links to Late Antiquity

Figure 4.11. The location of investigations at High Peak. From Rainbird et al. 2013, Figure 1

counts by type that is presented in Pollard’s report cannot be taken as correct.

Pollard 1961-4 Further excavations were carried out at High Peak in 1961 and 1964 by Sheila Pollard, which established two phases of Neolithic and ‘Dark Age’ activity at the site (Pollard 1966). Early medieval finds were identified in association with two surviving ramparts (labelled as the inner and outer rampart) and in the intervening ditch, as well as in occupation deposits, interpreted by Pollard as ‘cooking areas’ on a terrace on the landward site (Pollard 1966, 3544). These discoveries led Pollard to conclude that the site had not been occupied in the Iron Age or Roman period, but had been refortified during the fifth and sixth centuries AD (Pollard 1966, 56; Rainbird et al. 2013, 25).

Pollard’s report of the excavations conducted between 1961-4 records that amphora sherds were recovered from four features: the ditch, small outer rampart and two separate ‘occupation deposits’ identified across trenches labelled A, B and G (Pollard 1966; see Figure 4.11).

Table 8. Count of amphora sherds from High Peak, Sidmouth. After Pollard 1966, 53, analysis by Charles Thomas Amphora Type

The ceramic assessment on the sherds from this excavation and the earlier finds was carried out by Charles Thomas. He identified the sherds as coming from Late Roman amphorae, specifically as ‘Bi’, ‘Bii’ and ‘Bmisc’ (LRA2, LRA1 and LRA respectively) and provided sherd counts for each of these types (Table 8) (Pollard 1966, 53-4). However, as it must now be assumed that Thomas did not see any pottery recovered by Carter, the quantification of the sherd 82

Excavation

Bi

Bii

Biii

1871

2

7

-

1929

2

9

1

1961-4

3

60

-

Imported pottery in Britain and Ireland: new evidence and understandings described by Munsell number and name, although due to depositional conditions the colour of the sherds from the same vessel differed significantly. Few obvious new joins could be identified within the collection, especially as the surfaces of many of the sherds were rather worn.

Examination and quantification in 2012 Although the sherds recovered from Pollard’s excavation are generally marked with the trench letter, the feature within the trench is not specified and the context of discovery is not noted in the ceramic report. In addition, only 54 of the sherds are accounted for in the Pollard’s section covering the ‘Dark Age’ features, out of an original published site total of 63 (Pollard 1966, 42-4, 53). As a result, many of the sherds within the High Peak assemblage cannot be assigned with certainty to a specific feature. The sherds recovered from Pollard’s excavations are all body-sherds; again, there are no diagnostic sherds, such as handles or rims. However, in general they are more abraded than the antiquarian finds. A number of the sherds do not have ridging or ribbing but considered overall the entire assemblage seems to be sherds of Late Roman Amphorae. Many of the sherds are very small. A number have been glued together, and some of these were illustrated in Pollard’s report as a group (Pollard 1966, Figure 12.5, for example, is a composite illustration of glued, joining sherds). For this revised quantification every scrap was counted even if it had been glued together – which has raised the overall sherd count to 97 (339g) from the original 63 (Pollard 1966, 53).

As noted, the most recent assessment of the High Peak pottery (Pollard 1966, 52-5) provided a breakdown of sherd quantities by type for each phase of investigation, but did not propose an estimated total number of vessels. Although such an estimate is important for interpreting the assemblage and for allowing comparisons to other sites, it must be stressed that the nature of the assemblage caused difficulties in arriving at a figure. As there were no diagnostic sherds, this figure was based on general similarities of fabric and surface treatment, and should be considered as an approximate minimum rather than a fixed, precise total. Similar sherds were assigned to fabric groups irrespective of the date or location of their recovery (see Appendix E). These fabric groups may indeed represent individual vessels, but it is also possible that they represent sherds from numerous vessels with similar fabric. These groups suggest that sherds from the same vessel may have been collected at various dates or even from different points across the site. Early medieval pottery from High Peak

The pottery was quantified by sherd count and by weight. The absence of any diagnostic elements prevented quantification by EVE, but an estimate of the minimum number of vessels represented by the sherds was attempted (Table 10). Fabric descriptions were based on visual study at x20 magnification, but unfortunately, it was not always possible to form these from fresh breaks. Colours were

The pre-2012 assemblage from High Peak comprises 128 sherds with a combined weight of 793g (see Table 9). It was possible to match four of the sherds in the collection directly to Hutchinson’s illustrations. A further three

Table 9. Sherds recovered from High Peak. The fabric groupings and identifications are presented in Appendix E Sherds with ‘RK’ stamps

15

Sherds illustrated by Hutchinson (nine shown in the illustration)

4 (Four can be clearly identified, a further three are less certain and carry ‘RK’ stamps, two could not be located)

Sherds catalogued as ‘excavated by D’Urban’

2

Unassigned ‘antiquarian’ sherds

7 (includes 1 handwritten stamp)

Unassigned sherds

3

Sherds excavated by Pollard

97

Table 10. Sherd counts from phases of investigation at High Peak. Figures in parentheses are weight in grams Collection Phase

LRA1

LRA1 ‘Bantham- type’

LRA2

UnID LRA

UnID

Totals

Sherds from ‘pre-Pollard’ investigations

17 (260g)

0

2 (35g)

8 (137g)

1 (15g)

28 (447g)

Pollard

80 (293g)

3 (14g)

0

14 (32g)

0

97 (339g)

Unassigned

3 (7g)

0

0

0

0

3 (7g)

AC Archaeology 2012

11 (43g)

0

0

0

0

11 (43g)

Totals

111 (603g)

3 (14g)

2 (35g)

22 (169g)

1 (15g)

139 (836g)

83

Links to Late Antiquity seem to be possible matches but also bear ‘RK’ stamps suggesting they were found by Kirwan. The sherds found prior to 1966 cannot, therefore, be definitively matched to individual collectors or locations. It is also clear that Hutchinson’s illustrations were not exactly to scale, and over-emphasised certain features such as inclusions or surface treatment. It also seems likely that a number of the illustrated sherds have been worn or broken since they were originally illustrated (including some of the more recent finds, such as Pollard 1966, Figure 12.2).

The significance of the High Peak pottery The sherds recovered by Pollard come from approximately five vessels, all of which have been classified as LRA1, LRA1 ‘Bantham-type’ or unidentified LRA. It is possible, however, that a few sherds of all but one of these vessels (the ‘Bantham-type’/ ‘Group 6’) were also collected in the nineteenth century. In contrast, the lower number of nineteenth century, ‘antiquarian’ finds represents a greater variety of vessels (approximately seven) – again, mostly LRA1 or unidentified LRA, but also possibly LRA2. The sherds recovered in 2012 seem to represent one Late Roman 1 amphora that is ‘new’ for the site and two sherds from one of the vessels partially recovered by Pollard. Overall, there is a high degree of variation in the fabrics of the LRA1 vessels, which is similar to Bantham and Mothecombe – suggesting a range of production sources for the vessels arriving at these South Devon sites.

Excavations by AC Archaeology in 2012 Excavations at High Peak were carried out by AC Archaeology in the summer of 2012 as part of work on the South West Coastal Path. These revealed an ‘outer ditch’ at the site that may have been partially excavated but not fully recognised by Pollard, and which adds an element to the known post-Roman fortifications. The recognition of a bivallate enclosure at High Peak implies that the site represented an ‘elite settlement’ with ‘substantial defences’ (Rainbird et al. 2013, 49-50).

Apart from one unidentified sherd, the entire early medieval ceramic assemblage is comprised of amphora sherds. However, unlike Bantham there are no sherds of Gazan amphorae, North-African amphorae and no obvious examples of the so-called ‘thick-walled amphorae’ (Bidwell et al. 2011, 100-102). Nevertheless, there remain a significant number of sherds that could not be classified, due to a lack of diagnostic elements and non-distinctive fabrics. The absence of rims or handles prevented the identification of sub-types of the amphorae, such as those discussed at Bantham, limiting the chronological information available.

The excavations produced eleven body-sherds of amphora with a combined weight of 43g (see Figure 4.12). Including these eleven sherds, the overall total sherd count for the site is 139 with a total weight of 836g. A full quantification of the sherds by type and investigation phase is provided in Table 10 and a minimum estimate of the number of vessels represented by the body-sherds in Table 11.

As discussed in Sub-chapter 4.3.2, the Bantham ceramic report noted the presence of two body-sherds within the High Peak assemblage in the same fabric as the LRA1 ‘Bantham-type’, but reported a lack of obvious parallels from other British sites (Bidwell et al. 2011, 99). During the examination of the High Peak material three sherds were recognised that had similar fabric and colour and bore the same, distinctive, square-grooved surfaces. However, as there are no diagnostic sherds within the High Peak collection, it is not certain that the complete vessels were of the same shape as those at Bantham. Nevertheless, it seems likely that amphorae of this distinctive ‘Banthamtype’ were imported to High Peak, possibly arriving via shipments reaching Bantham itself. It is possible that further examples of this type will start to be recognised among the British imports, but it is also possible that these finds represent a specific cargo containing an unusual amphora type (of possible Iberian origin) that only reached this region. It may prove significant that this sherd was identified by Charles Thomas as LRA2, as it leaves the possibility that some of the other British sherds classified as ‘Bi’ might, in fact, belong to this or other less common forms of LRA1.

Figure 4.12. LRA1 sherds recovered at High Peak in 2012. From Rainbird et al. 2013, Figure 8

Table 11. Estimated minimum vessel counts from High Peak. Based on distinct fabric groups (see Appendix E) LRA1

LRA1 ‘Bantham Type’

LRA2

UnID LRA

UnID

Total

4/5+

1

2?

2+?

1

10/11+

Unlike Bantham there were no sherds of Red Slip fineware, no E ware and no obvious sherds of early medieval pottery produced in Britain (Bidwell et al. 2011, 106-8). Overall, the range of pottery represented by the High Peak sherds 84

Imported pottery in Britain and Ireland: new evidence and understandings tention of Mr Peter Holt as part of the SHIPS Project, which studies the maritime history of Plymouth, and was displayed on the project website (SHIPS Project 2012, Amphora 12A10).

is rather limited. The origin of the ‘Bantham-type’ is uncertain, but most of the material can be linked to the East Mediterranean. The assemblage provides evidence of the transport vessels in which imported, exotic commodities were brought to the site, but contains no fineware. However, as much of the site has been destroyed by erosion, it is probable that some material has been lost.

The amphora The surviving portion of the amphora (Figure 4.13 and Figure 4.14) comprises the rim, handles and upper body. There is some marine encrustation on the upper part of the neck and the underside of one handle, but overall the condition of the amphora is very good. More detailed macroscopic examination of the fabric from a fresh break would enable a better identification of inclusions and might suggest a potential origin for the vessel. Nevertheless, the amphora can be clearly identified as the upper part of a Late Roman 1 amphora (LRA1). There are no painted markings (dipinti) on the outside of the amphora and no graffiti. However, there are patches of a reddish-brown residue (Munsell 2.5YR 6/3) concentrated on the lower portion of the interior walls on one side of the vessel. Chemical analysis of this residue was carried out at Plymouth University as discussed below.

A minimum count of ten or eleven vessels was estimated for the site, which is much less than estimates for Bantham and Tintagel, but is broadly comparable with the estimates for other, significant early medieval sites in the South West with imported amphorae (see Table 10 and Appendix G). However, this might also reflect the various methodologies used and the varying condition and nature of the assemblages. The prevalence of small, worn, undiagnostic body-sherds within the High Peak collection presents similarities with the pottery from Mothecombe but contrasts markedly with the assemblage from Bantham. This might indicate that the sherds from High Peak were redeposited, or exposed to adverse conditions after the original vessels were broken, or reflect differences in the consumption and disposal of the vessels between the sites.

The Cawsand amphora presents a relatively rare opportunity among the British imports to attempt a more specific typological identification. Unfortunately, the bottom of the vessel was not recovered and as such it is not certain whether the vessel was pear shaped, as with LRA1A, or more cylindrical, as with later versions. Nevertheless, the combination of twisted, grooved handles and a fairly wide mouth points to a late-fifth or early-sixth century date (Pieri 2005; University of Southampton 2005). Pieri notes that the external diameter of the mouth of LRA1A is more constricted at 5-7cm while LRA1B is wider at 10-12.5cm (Pieri 2005, 70-5). Overall, the amphora would seem to match more closely to Pieri’s sub-type LRA1B1 (see Subchapter 3.2.2). A late-fifth century or earlier-sixth century date for this amphora would be a good fit for this amphora, which matches very well with the established date for imports to post-Roman Britain. The form also compares closely with an illustrated LRA1 from Bantham (Bidwell et al 2011, 95, Figure 14.1).

4.3.5 An amphora recovered from Plymouth Sound This sub-chapter summarises the report on an amphora that was discovered by a diver off Cawsand [123] in Plymouth Sound in the early 1970s (Duggan 2013c). Although the find was brought to the attention of Plymouth City Museum in 1973 and recognised as a possible early medieval import, it was not included in subsequent catalogues of imported Mediterranean pottery. A full description of the amphora and its fabric is presented in Appendix F. Background to discovery This upper part of a ceramic vessel was recovered by Mr Terry Bruce, a former police officer and sports diver of Plymouth, Devon in the early 1970s while diving off Cawsand in Plymouth Sound. The precise location has not been disclosed to the author. It was found buried upright with the rim of the neck showing above the seabed. After brushing the surrounding sand away Mr Bruce lifted the vessel by its handles. It was easily recovered and without any obvious resistance, but as Mr Bruce notes that as he did not dig any further into the sand it is not certain that the lower portion of the vessel was not left beneath the seabed. The artefact appeared to be an isolated find, although Mr Bruce did not investigate the location beyond a brief examination of the surrounding sea-floor.

Potential contents Further to the discussion of amphora contents in Subchapter 3.2.6, residue analysis of this amphora was carried out by Professor Steve Hill of Plymouth University in 2013 (see Appendix F). The identification of wine as the likely product contained in this vessel matches well with our understanding of this type of amphora, although this may not have been the original contents.

At the time of discovery, the vessel was shown to the curator of the Plymouth City Museum and Art Gallery, Mr James Barber, who examined, drew and photographed the object. At this point it was tentatively identified as being an amphora of likely Mediterranean origin and potentially dating from the 5th to the 7th century AD. However, the vessel was not published, and the amphora was returned to the finder. In 2012 the amphora was brought to the at-

Significance No sites with imported early medieval pottery have been identified to date on the Rame Peninsula, but it is tempting to see this amphora as signalling imported material arriving in this area, or otherwise the shipping of goods west 85

Links to Late Antiquity

Figure 4.13. Photograph of LRA1 found off Cawsand

across Plymouth Sound. The closest sites with early medieval imported pottery are Looe Island, Cornwall [122], to the west, and to the east – across Plymouth Sound and into Devon – the coastal sites of Mothecombe and Bantham (see Appendix B.4). The single sherd of amphora recovered as a surface find at Looe Island was identified as LRA2 (Thomas 1981, Todd 1983). Campbell (after Olson 1989) notes that the imported pottery at Looe Island might be linked to the presence of an early monastery, but overall argues against a direct link between imported pottery and early medieval religious settlements (Campbell 2007a, 122). Further work at Looe Island by Time Team in 2008 failed to produce any further early medieval finds (Wessex Archaeology 2009).

a firm archaeological context, it fits extremely well with the chronology and pattern of material recovered from local sites with early medieval imported pottery. As such, it seems reasonable to consider the vessel as a potentially genuine post-Roman import, rather than an ‘antiquarian’ object that ended up in a marine context. This cannot be proven, and the specific circumstances of its deposition remain unclear, but it does seem reasonable to consider it as potentially another example of East Mediterranean pottery imported to early medieval Britain. In 2014 a second amphora was recorded as part of the SHIPS project. It has not been possible to identify this example in person, but (from photographs) it appears to be the upper part of a Late Roman 1 amphora: the neck, rim and ‘stubs’ of the top of both handles are present. It was recorded as having been found by a diver in the Tamar, but there is no further information on the location or circumstances of its discovery (Holt, pers. comm.). As such, although there are similarities with the shape of the Cawsand amphora and the circumstances of its discovery, there is greater doubt that this new example represents a genuine early medieval import.

As discussed, LRA1 is the most common overall for sites in Britain. This type was the most common amphora recovered at High Peak and Mothecombe and provided by far the largest proportion at Bantham (see Appendix G). As such, the discovery of a LRA1 of this date in waters close to the south Devon coast would certainly be exceptional in its discovery, but not unusual in its type. Although the Cawsand amphora was not recovered from 86

Imported pottery in Britain and Ireland: new evidence and understandings

Figure 4.14. Scaled illustration of LRA1 found off Cawsand

4.3.6 South Devon case study: discussion

a very small proportion of the assemblage. Although finewares are generally rare in Britain, this might indicate that the sites in South Devon were specifically focused towards the supply of amphorae, or were at the periphery of a separate system that supplied finewares (perhaps in association with ‘archaeologically invisible’ commodities). The comparative absence of E ware is also a defining feature of these sites in South Devon. Campbell noted the relatively small quantity of E ware in the South West and concluded that no new sites were established in this region during the later phase of exchange characterised by this pottery. He suggested that following the sixth century ‘collapse’ of the ‘major centres’ that received Mediterranean wares, only limited connection was retained with the smaller, coastal sites (Campbell 2007a, 121). The relative scarcity of E ware in this region, therefore, suggests that importation to South Devon declined by the mid-sixth century, and that there was, therafter, limited contact with systems focused on south-western France.

The presence of imported Mediterranean pottery in south Devon has typically been explained by a model of the exchange of locally produced tin for exotic commodities, particularly wine (Reed et al 2011, 129-30). Geoarchaeological analysis has revealed evidence of tinstreaming to the north of Bantham and Mothecombe on Dartmoor (Thorndycraft et al. 2004). The recovery of forty tin ingots of possible early medieval date from the sea in Bigbury Bay, off Mothecombe can be connected to these exchange systems operating along the south coast (Fox 1995, 21-22). The amphora recovered off Cawsand might also attest to these shipments. Although the sites/find-spots are rather different in character, these sites on the south Devon coast show a similar pattern of imported pottery. The focus on Late Roman 1 amphora at High Peak is similar to Bantham and Mothecombe, in contrast to Tintagel, which is dominated by LRA2. No imported finewares were recovered from Mothecombe or High Peak, while only three imported fineware vessels were identified at Bantham, representing

Perhaps the most revealing aspect of this case-study is the potential evidence for contacts between the region and Spain or Portugal, although the sources of the LRA1 87

Links to Late Antiquity in Britain at Tintagel and Dinas Powys (Campbell 1988, 128 – recorded as Fulford 58; Timby 2011a, 49; b, 145). Thomas catalogued variants 99A and 99B of this form at Insular sites, including examples from Tintagel (Thomas 1981, 8-9). However, variant 99C, as specified here, is potentially very late – indeed it would be later than any other British or Irish ARS. Within the Pevensey report a late-sixth to seventh century date is given for 99C (Timby 2011b, 145 citing Bonifay 2004, 179), although LRFW1 suggested production up to the later-seventh century (Cau et al. 2011c, 5). Unfortunately, these two sherds are not illustrated, preventing comparison with other British finds. The final two sherds, from the same vessel, were were found within a sequence of ‘dark earth’ deposits inside the Roman fort wall (Timby 2011a, 43). One of these – a rim-fragment – is illustrated in the report (Timby 2011a, Figure 3.3, 48). It is described to have roulette-decoration on the upper rim and as being closest in form to Hayes form 75 (Timby 2011b, 145). This form is noted to be uncommon in LRP and has not previously been identified at any British site, or at any Atlantic site to date. The early to mid-fifth century date given by Timby for this sherd is noted to be ‘late amongst the British finds’ of ARS (2011, 145). The date given in LRP is c. 420-450 (Hayes 1972, 124).

‘Bantham-type’, and ‘thick-walled’ amphorae need to be established. Similarly, further work is required to determine the origin of the coarseware sherd from Bantham. The possibility that it originated in Vigo suggests there might have been specific links between these regions, and might even indicate the presence of traders from Galicia arriving at sites in south Devon. 4.4 Additional recent British discoveries As well as the new evidence from the sites in South Devon case there have been other recent finds of early medieval imported pottery from sites in Britain and Ireland that have the potential to revise understandings of the chronology, range and volume of this import-system. These have been noted in Chapter Three, as part of the revised totals for the British and Irish assemblage. However, three particular discoveries have been highlighted here, due to their significance in expanding the established pattern of the distribution of the Mediterranean imports. 4.4.1 Pevensey Castle A recent British discovery with potential implications for the chronology and distribution of the Mediterranean imports is the identification of ARS at Pevensey Castle in East Sussex [108], following excavations (conducted in advance of construction work) at the site in 1993-95 (Fulford and Rippon 2011). Although African Red Slip ware is an occasional find in Romano-British contexts up to the fourth/early-fifth century (see Sub-chapter 3.3.2), its late-fifth/sixth century distribution has been seen as completely separate and associated with a new system supplying imported goods to western regions of Britain. Pottery arriving via these late Atlantic channels has not been thought to penetrate into eastern/Anglo-Saxon territories. This discovery at Pevensey therefore represents the site furthest east along the south coast with potential fifth/ sixth century forms of this pottery, and lies well beyond the established post-Roman distribution of Mediterranean wares.

Although small numbers of ARS of later-fourth or potentially early-fifth century date have been recognised as ‘late Roman’ imports to Britain, it is clear that the group from Pevensey represents a very unusual collection. Timby repeats Bird’s interpretation for the ARS vessels found at Romano-British sites – that they were unlikely to represent ‘traded cargoes’, but instead represented personal belongings (Bird 1977, 272; Timby 2011b, 145). However, it is likely that if the Pevensey group was found in western Britain, it would automatically be tied to fifth century long-distance import systems. This alternative interpretation is offered in the discussion section of the excavation report, where the finds are stated to be the first examples of ARS from south-eastern England with a ‘date-range starting after c. 400’, and are said to alter the view that post-Roman Mediterranean imports are ‘only to be found in western Britain’ (Fulford and Rippon 2011, 125).

One sherd of ARS, found in ‘slumping deposit’ outside of the Castle was tentatively identified as the flange from a bowl of Hayes form 91 (Timby 2011a, 45; b, 145). Although the variant was not specified, the sherd was connected to Bonifay type 49-50, and therefore Hayes 91A/B, which has not previously been identified in Britain (Timby 2011a, 45; b, 145). Variant C of this form has been previously recognised among western British imports at Tintagel and Dinas Powys (Thomas 1981, 8; Campbell 2007a, 17), as well at sites on the Atlantic seaboard including Bordeaux (Bonifay 2012, 253). In the ceramic report Timby suggests an early to mid-fifth century date for the Pevensey sherd, in line with the earlier variants A and B (Timby 2011b, 145), whereas in western British contexts ARS 91C is usually seen as a sixth century find (see Appendix C). A second sherd of ARS – redeposited in the trench of a robbed-out tower foundation – was matched to Hayes form 99, which has been identified previously

The excavations at Pevensey suggested that there was not any ‘significant break of occupation’ at the fort after the late Roman period, and early Saxon pottery was recovered, as well as fragments of sixth or seventh century glass (Fulford and Rippon 2011, 124-5). The authors suggest that these, together with the imported ARS, point to a ‘high-status element’ at the site in the early medieval period that had trading contacts ‘looking both eastwards up and across the Channel to the Rhine, and westwards to the Atlantic and the Mediterranean beyond’ (Fulford and Rippon 2011, 125). Whether these particular sherds of ARS arrived via Atlantic or Rhenish channels remains unclear. The authors note sherds of ARS recovered in the Netherlands (of late-fourth/early-fifth century date?) and at Cologne that potentially attest to these two alternative routes of supply (Fulford and Rippon 2011, 125). 88

Imported pottery in Britain and Ireland: new evidence and understandings it seems likely that the amphorae would have reached the site overland from western Scotland (Noble et al. 2013, 1142). The fragments of pottery and glass might only represent the surviving elements of a wide variety of imports. Nevertheless, the distance that these vessels travelled makes it clear that the imported commodities – which presumably included wine – were highly prized. The connections these products signified may have been equalled valued. The fifth and sixth centuries in Scotland witnessed the development of the Pictish kingdoms through the consolidation of local power, as evidenced by the construction of forts and enclosures (Noble et al. 2013, 1143). It might be that – as in south-western Britain – the acquisition, possession and redistribution of exotic commodities were instrumental in the formation and maintenance of local and regional power-structures. However, the specific origin of these products might not have been important, or even apparent, to the site’s inhabitants.

A body-sherd from a possible East Mediterranean amphora at Pevensey might also be associated with these fineware imports (Fulford and Rippon 2011, 125). The sherd is described to have a micaceous fabric and to potentially be from a ‘wide-bodied vessel’, but could not be matched to specific type (Williams 2011a, 146). Reference is also made to the possible discovery of DSP at the site during earlier excavations (Fulford and Rippon 2011, 125). Lyne’s analysis of these earlier finds identified DSP, specifically a red bowl of Rigoir 15A – provisionally classed as a Languedoc product – and a grey bowl of Rigoir 17 (Lyne 2009, 101). Lyne suggested a stamp from the Atlantic group for the oxidised bowl, but the overall decoration appears to resemble Argonne ware rather than DSP (Lyne 2009, 118, Figure 32.27). Rigoir et al. 1973 did not include Form 17 in the Atlantic repertoire and it has not been identified at sites in western Britain or on the Atlantic Seaboard. It is possible that these identifications could be refined by further comparisons with Continental examples.

4.4.3 Druce Farm, Dorset The group of ARS from Pevensey raises the possibility of future identifications of Mediterranean pottery in early medieval Britain beyond its traditional, western distribution. The sherds of ARS 91 and 75 could be linked to the earlier phase of expansion of ARS into the Atlantic. It is more difficult to account for the sherd of ARS 99C. Its very late date suggests it cannot be easily associated with the vessels reaching western Britain by Atlantic channels. It may be that further research will reveal more complex patterns of Mediterranean imports in eastern Britain in the fifth and sixth century – and potentially even beyond the mid-sixth century.

At a late stage in this project the author was advised of the discovery of two amphora sherds at Druce Farm Roman Villa in Dorset [109], close to the town of Dorchester; these had been identified as LRA1 by Malcolm Lyne (Ladle, pers. comm.). Like the sherds from Pevensey, these finds extend the known distribution of the imported wares, representing the most easterly example of an identified, East Mediterranean amphora – from a secure post-Roman context – found along the south coast of Britain. Excavations are continuing at the site. 4.5 Conclusions and directions in research

4.4.2 Rhynie These recent publications have revealed a greater degree of complexity within the British assemblage, showing that the pottery imported to post-Roman Britain is both more varied and more widely distributed than previously recognised. New finds from the south coast of Britain demonstrate the possible extension of Atlantic supply networks into the Channel, while the finds from Rhynie reveal the distribution of Mediterranean pottery into northeastern Scotland via overland channels.

Recent discoveries have also extended the northerly distribution of imported Mediterranean pottery. Excavations at Rhynie in north-east Scotland [3] revealed new features at this Pictish power centre, famed for its carved stone monuments, and produced a remarkable collection of amphora sherds (Noble et al. 2013). These represent the northernmost finds of eastern Mediterranean pottery in Britain or Ireland (Noble et al. 2013, 1145). Excavations in 2011 and 2012 revealed a series of enclosures with a possible timber wall, and the presence of timber buildings, including a possible timber hall. These features were dated by radiocarbon to the later-fifth and sixth century (Noble et al. 2013, 1142). The finds included a small collection of sherds of LRA1 and LRA2 – identified by Ewan Campbell – as well as fragments of imported glass, reported to originate in western France (Noble et al. 2013, 1142). The discovery of moulds and crucibles also indicated metalworking activities at the site.

The assemblage from Bantham is of key significance in challenging understandings of the British imports, some of which – as seen in Chapter Two – underpin the models of direct contact with the East Mediterranean. Although Tintagel remains the largest assemblage, individual features of its assemblage can no longer be taken to characterise the entire Insular distribution. Instead, the finds from Britain – and specifically those from south Devon – can be more closely aligned with patterns in the West Mediterranean, where LRA1 dominates as an East Mediterranean import. However, the more general Insular focus on East Mediterranean types over North African pottery remains to be considered.

These discoveries, far from the recognised concentration of sites with Mediterranean pottery imports in the southwest of Britain, reveal very long-distance links to the site. Somehow, possibly indirectly, the inhabitants of Rhynie were able to connect with channels supplying wine via the Atlantic. As the site is unusually far from the coast,

The identification of the LRA1 ‘Bantham-type’ demonstrates that closer attention to variant amphora types 89

Links to Late Antiquity vessels and groups of pottery might extend the range of material known to be distributed in the Atlantic in the fifth and sixth century – as suggested by the ‘Muscovite-rich’ vessel from Bantham. This is particularly important for Tintagel. The publication of parallel assemblages from the Atlantic Seaboard – as will be discussed in Chapters Five and Six – should allow the British material to be identified with more precision. This has considerable potential to produce new narratives of contact and exchange in the Late Antique Atlantic.

– beyond Riley’s standard series – has the potential to reveal connections between Insular sites and with sites on the Continent. Furthermore, these recent discoveries hint at the existence of specific connections between communities in south-western Britain and the Iberian Peninsula (possibly mediated by Vigo). Within the timescale of this project it has not been possible to verify the source of the ‘Bantham-type’, ‘Bv’ and ‘thick-walled’ amphora but this must be a priority for future research. It is also possible that additional examinations of ‘unidentified’

90

5 Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France 5.1 Introduction

for understanding the British and Irish finds – and a casestudy of the Mediterranean imports found at Bordeaux will be presented. Sub-chapter 5.4 will examine research on E ware – the class of pottery thought to characterises a phase of ‘Continental’ importation to Britain. This will, for the first time, discuss large numbers of comparable pottery published from France, in order to propose a likely origin for this ware, extend its distribution and reveal details of its development and chronology.

The finds of imported Mediterranean pottery from Britain and Ireland have been considered to be largely isolated on the Atlantic coast, apart from a small scatter of equivalent material – principally LRC – on the western Iberian coast. In particular, the Atlantic coast of France has been portrayed in British publications as almost completely lacking in comparable material. This is exemplified by Wooding’s observation – that the dearth of parallels for the Mediterranean imports found in Britain and Ireland in central/western France went beyond the ‘failure of French archaeology to discover examples’ but reflected a genuine absence (Wooding 1996, 42). However, this widely accepted interpretation was principally based on Wailes’ research from the late 1950’s, which was never superseded. Campbell’s 2007 catalogue included a handful of sites on the French Atlantic coast that had produced Mediterranean imports, specifically amphora sherds from l’Ile Lavret and Le Yaudet, in Brittany, St-Laurent-desCombes in Gironde, Aquitaine, and three vessels from Saint-Seurin, Bordeaux (Campbell 2007a, xvii-xix; 2011). However, these did not affect his overall interpretation – that the Insular distribution of this Mediterranean pottery represented direct contact and exchange with the tinproducing regions of south-west Britain, and subsequent redistribution.

5.2 Mediterranean imports to western France in Late Antiquity 5.2.1 Imported Mediterranean amphorae Late Roman amphorae and fineware of fifth century and later date have been found in substantial quantities in southeastern France, most notably at the Mediterranean port of Marseille. In particular, the excavations conducted at the harbour of ‘La Bourse’ between 1967 and 1984 produced significant quantities of East Mediterranean and African amphorae and African Red Slip fineware (Bonifay et al. 1998, 55-196, 380-401). This revealed commercial vitality at the port in Late Antiquity, and provided evidence of ongoing, if fluctuating, connections to the Byzantine East Mediterranean and North Africa between the fifth and seventh centuries (Bonifay et al. 1998, 377).

As reviewed in Chapter Two, research on the British and Irish imports diverged from research on the Continent and emerging evidence from France, Spain and Portugal was never fully incorporated into discussions of the Insular imports. As a consequence, the resulting interpretations – based on the types present, relative proportions and overall distribution – are distorted. Recent research on the Continent has started to fill this ‘gap’ on the Atlantic Seaboard; although the total quantity of Mediterranean imported pottery published from the Atlantic regions of France is not great in relation to some Mediterranean sites, it is much more significant than has been implied in earlier publications. Mediterranean pottery of fifth and sixth century date is certainly present at sites in western France, and this data is likely to increase. In particular – and contra Wooding’s observation – increasing evidence has emerged from Bordeaux to reveal that the city did receive significant quantities of Mediterranean pottery in this period, and was a key site within early medieval Atlantic exchange systems.

Bonifay’s 1986 article on the pottery from La Bourse summarised the recovered African, Spanish and East Mediterranean amphorae (which included all of the amphora types found at Insular sites), but also mentioned the occurrence of these types in other parts of France. Although nearly all of these examples reflected a Mediterranean distribution, he noted the presence of LRA4 amphorae at Saint-Seurin in Bordeaux (Bonifay 1986, 281). A subsequent article reaffirmed the distribution bias toward the Mediterranean coast, but indicated the potential for future discoveries in the Atlantic region, particularly in Aquitaine (Bonifay and Villedieu 1989, 42-3). The presence of a LRA4 at Tours and a sherd of LRA1 in Britanny (Lavret) were specifically recorded (Bonifay and Villedieu 1989, 41-2, Figure 17). Pieri later recognised that further synthetic research on Late Antique ceramic imports to France was necessary, commenting that the distribution of this pottery was little understood beyond the Mediterranean littoral (Pieri 2005, 1). His 2005 study of the Byzantine wine trade compiled published and unpublished data on imported East Mediterranean amphorae in France of fifth to seventh century date. Although his synthesis covered the whole of modern France, Pieri remarked that his conclusions nec-

This chapter will review this mounting information, and evaluate its impact for understanding contact and exchange on the Atlantic Seaboard. Finds of imported Mediterranean from a scatter of sites in western France will be discussed – highlighting the relevance of this data 91

Links to Late Antiquity comparison to those from British sites such as Dinas Powys and South Cadbury, much less to Tintagel or Bantham. As such, it is not evident whether these finds from Brittany indicate primary import sites or reflect the redistribution of commodities from the south of Britain. Significantly, the presence of sites with Mediterranean imports on the Atlantic coast of Iberia is discussed in the report from Le Yaudet, as well as the possibility that these locations acted as ‘trans-shipment points’ for Atlantic exchange system. The authors, consequently, speculated that further finds of Mediterranean pottery from north-western France were likely to follow (Cunliffe and Galliou 2007, 194). Cunliffe and Galliou also discussed whether the evidence from Le Yaudet – in demonstrating the continuation of post-Roman maritime connections – might be associated with the ‘supposed’ emigration of people from the south of Britain to ‘Armorica’ (Brittany) in the fifth century. However, they suggested that this migration might simply represent the ‘intensification of existing relationships’ between these Atlantic communities, symbolised by shared access to exotic, Mediterranean commodities (Cunliffe and Galliou 2007, 194-6). Although the specific connections of these sites to Atlantic exchange systems remains to be clarified, their identification starts to reveal a narrative of contact along the Western Seaboard.

essarily focused on sites in the south, given the rarity of examples and generally poorer data in northern regions (Pieri 2005, 2). The number of vessels recorded from sites in northern France was certainly very small in comparison to the large quantities of imported amphorae recorded in the south-east, especially from Marseille (Pieri 2005, 7). Nevertheless, his catalogue presented a scatter of finds of Late Roman amphorae at a number of sites in western France. All of these examples were, however, recorded by Pieri’s as ‘additional sites’ – typically isolated examples from poorly interpreted or badly stratified sites. Most of these have never been published and often there is limited information available on the location or context of discovery, consequently limiting any interpretations. An example of LRA1 was recorded from the early medieval – possibly monastic – site of l’Île Lavret, Bréhat in Brittany; this had been previously cited by Fulford as demonstrating the potential for future Atlantic finds (Fulford 1989, 3; Pieri 2005, 50). Giot and Querré’s 1985 report on the pottery from l’Île Lavret made it clear that they had hopes of identifying parallels for the British post-Roman imports from the excavation. Although no ‘A’ wares (ARS/LRC) were found, they suggested the possibility that small, undecorated sherds might have been missed amongst the earlier finewares. However, a body-sherd with characteristic ribbing was recovered from a midden deposit overlying levels containing Gallo-Roman pottery, and identified as ‘Bii’ (Giot and Querré 1985, 96). Despite its small size, Pieri was able to refine the original classification to LRA1B. A group of 130 coarseware sherds scattered across the site were reported to show similarities with E ware, but this association was later rejected (Giot and Querré 1985, 97-9).

To date, there is limited evidence for Mediterranean wares being distributed further east along the Channel. A bodysherd of LRA1B of sixth century date was found in an excavation at the Collège de France in Paris [154], although it seems possible that this arrived via overland/ riverine routes (Pieri 2005, 54). An isolated example of LRA1A from Metz in north-west France – presumably arriving via the Rhine – also demonstrates the sporadic presence of East Mediterranean amphorae at inland sites (Pieri 2005, 53).

Also on the north coast of Brittany, but published after Pieri’s volume, excavations at the promontory settlement of Le Yaudet produced four body-sherds, probably from a single LRA2 amphora, which were identified by David Williams (Williams 2007, 38). The sherds were found in a midden deposit close to a rectangular stone building with an oven, associated with the fourth/fifth century re-occupation of the site. The midden also produced some late Roman (fourth and early-fifth century) pottery and metalwork, as well as the neck of a glass vessel of mid-fourth to mid-sixth century date (Cunliffe and Galliou 2007, 42, 193-4). It is not clear whether the site was fortified at this point, but it seems to have been continually occupied from the late Roman period into the fifth and sixth centuries, during which time it was connected to networks that provided access to exotic goods, including the contents of East Mediterranean amphorae (Cunliffe and Galliou 2007, 194). These features present clear comparisons to the highstatus settlements in the south-west of Britain that received Mediterranean imports. Whether there was a comparable hiatus in these long-distance connections in the mid-fifth century – as hypothesised for western Britain – is unclear.

To the south, within the Pays-de-la-Loire, Pieri’s catalogue records the unpublished discovery of a sherd of LRA1 from the surroundings of Vaas, close to the sanctuary at Aubigné-Racan [167] (Pieri 2005, 49). Approximately 45km to the south-west, at Tours, a LRA4 amphora is reported to have been found in an ‘old’ excavation (Pieri 2005, 55). No further information is available for this example; Pieri’s source is an observation by David Peacock reported in Riley’s original discussion of this amphora type (Riley 1979, 220). As such, the likely date of this vessel cannot be ascertained. Further east and down the Loire, a rim/neck of LRA2 is recorded from Rezé [172] (ancient Ratitatum) (Pieri 2005, 53). Situated on the south bank of the Loire from Nantes, excavations have been carried out within the town from the nineteenth century, with more recent investigations revealing evidence of late Roman/Late Antique activity (Pirault 1997, 299-300). Excavations at the south of the ancient settlement revealed a residential area dating to the end of the first century AD that seems to have been reoccupied between the fifth and seventh century (Pirault 2007; Pieri 2005, 53). Pottery of this date – mostly DSPA

The quantities of fifth and sixth century imports currently identified in north-western France are certainly small in 92

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France To the west of the Midi-Pyrénées Region, six North African spatheia of fifth century date were recorded from the cemetery of Saint-Gény at Lectoure [210] (Amiel and Berthault 1996, 257). These had been reused for infant inhumations, a practice that has not been identified in postRoman Britain, but has been recorded at a number of sites in south-western France – most-notably at Saint-Seurin, Bordeaux (Piot 2001, 104-7; Sub-chapter 5.3.2). To the west of Lectoure, imported amphorae were identified from a villa (the ‘Villa de Séviac’) at Montréal du Gers [209]. This group comprised one Dressel 30 amphora (a third century type), another amphora recorded as Almagro 51C (assigned a fourth century date although it seems possible that this is later), fragments of undated spatheia and LRA4 and also one North African Keay 62 amphora of sixth century date (Amiel and Berthault 1996, 258). This was the only villa site in the south-west that produced amphorae of sixth century date, although the authors comment that this does not necessarily indicate the abandonment of rural, agricultural properties, but rather changes in consumption connected to new patterns of land-management.

and associated coarsewares – was recovered from the latest occupation levels and, in particular, from abandonment deposits associated with the re-occupation of one of two ‘domus’ buildings (Pirault 1997, 300). The principal forms of DSPA recovered were Rigoir Forms 4 plates and Form 6 bowls, but Form 30 lids and 29 mortaria were also represented, as well as a possible example of DSPP from Mediterranean France (Pirault 1997, 300). In addition, a few fragments of amphorae were recovered, the latest of which is likely to be the LRA2. The other examples were recorded as Iberian amphorae of later Roman date: one as a rim of a western Iberian Almagro 50 and another as Almagro 51C (although these identifications are open to question), as well as two examples of Baetican Dressel 23 amphorae (Pirault 1997, 300; University of Southampton 2005). No published examples of fifth or sixth century Mediterranean pottery could be found from PoitouCharentes, but it is likely that data will emerge from this region. Further south, particularly in areas surrounding the Garonne and Dordogne, there is much more evidence for Mediterranean amphorae. Pieri’s catalogue includes data discussed in an earlier French article by Amiel and Berthault on late Roman and Late Antique imported amphorae in south-western France (Amiel and Berthault 1996). This study reviewed the types and relative proportions of North African, East Mediterranean and Spanish amphorae found in the region between the third and sixth century, drawing specific distinctions between the pottery supplied to the main urban centres at Bordeaux in Aquitaine and Toulouse in the Midi-Pyrénées. Small quantities of late amphorae were documented at a number of small urban sites, villa sites and rural sites in the south-west – principally North African amphorae and LRA4 – but the authors comment that beyond the fifth century, Mediterranean imports were, on the whole, only recovered at the larger urban sites (Amiel and Berthault 1996, 257). This observation is seen to reflect excavation bias towards the larger urban centres, but also to possibly reveal socio-economic change – that by the fifth century these smaller towns were no longer occupied by an ‘elite’ capable of accessing imported luxuries (Amiel and Berthault 1996, 257).

A group of sites with imported amphorae are also recorded along the Dordogne. Excavations at the town of Périgueux [190] (on the River Isle, a tributary of the Dordogne) produced two examples of LRA4 amphorae that were assigned probable dates, as well as an African amphora of fourth century date. One of the LRA4 was identified as a fifth century type (stated to be equivalent to the Zemer 53 amphora, although this is usually seen as earlier in date) and the other as a sixth century type (Almagro 54) (Amiel and Berthault 1996, 257). LRA4 was also identified close to the villa at Montcaret [196], thought to be situated on an ancient route between Bordeaux and Périgueux and excavated from the nineteenth century (Amiel and Berthaut 1996, 258; Berthault 1997, 159). The shoulder of LRA4 was recovered from the fill of tanks associated with viticulture, close to the villa, and was identified as a probable sixth century example (Berthault 1997, 159-65). This sherd, together with the DSP found at the site, is testament to the fifth and sixth century occupation of the site and associated late modifications at the villa complex. To the west of Montcaret, and also close to the Dordogne River, excavations in 1987 at another villa at Gueyrot, Saint-Laurent-des-Combes, produced a body-sherd of an African amphora and the upper parts of two LRA4 (Hochuli-Gysel et al. 1995; Amiel and Berthault 1996, 258). These were found in a rectangular pit, cut into late occupation deposits, that seems to have functioned as a waste-pit. As well as the amphorae, the pit produced DSP and fragments of over fourteen glass vessels. Intriguingly, the DSP was not of the Atlantic group, but possibly of the Languedoc group; its arrival at the site may have pre-dated the earliest DSPA production at Bordeaux in the fifth century (Hochuli-Gysel et al. 1995, 181). Pieri records the East Mediterranean amphorae as LRA4A2 and as dating to the first half of the fifth century, although Berthault suggests an early-sixth century date might be possible for one of the rims (Hochuli-Gysel et al. 1995, 182; Pieri 2005,

A number of smaller urban sites, including Agen (Aquitaine) and Eauze (Midi-Pyrénées) were recorded to have produced fourth century pottery, but only a small number had amphorae that could be clearly dated to the fifth century or later. In particular, a number of examples of LRA4 were recorded from sites in the south-west, but the longevity of this type means that the likely date is not always clear. This exemplifies the difficulty in equating patterns in the imported pottery between Britain and the Continent. As noted in Sub-chapter 3.2.2, this type of amphora is not automatically assigned a post-Roman date in Britain. Some of these Continental examples have been specifically identified as the early sub-type LRA4A, and could, therefore, equate to the late-fourth/early-fifth century examples found at Romano-British urban sites (Campbell 2007, 19). 93

Links to Late Antiquity sixth century imported finewares are less common in Atlantic France than in south-western Britain, with no published assemblages on the same scale as Tintagel, or even Cadbury Congresbury.

33). The excavations identified a phase of construction at the villa in the fourth century, and dated its abandonment to the fifth or sixth century (Hochuli-Gysel et al. 1995, 177). Three spatheia are recorded from Libourne [194], where the River Isle meets the Dordogne, but these are not described further, and the specific date is unclear (Amiel and Berthault 1996, 257). Similarly, another undated spatheion is recorded from the village of Saint-Georges de Montagne [193] (Amiel and Berthault 1996, 258).

Tyers’ 1996 distribution map of ARS in Roman Pottery in Britain does not show any parallel examples in northern and western France, although it should be noted that this data is not, necessarily, up to date (1996, 153). The Dictionary of Roman Pottery in Northern Gaul, published in 2010, records the presence of ARS-D at Cherbourg, Lower Normandy, apparently as part of a minor, northeastern, inland distribution of this ware to larger towns, including Paris, Trier, Cologne and Orléans (Brulet et al. 2010, 206). At present, western France lacks a similar synthesis. The possible identification of ARS from Cherbourg is also mentioned by Giot and Fichet de Clairfontaine, in their discussion of fifth to fifteenth century pottery in Brittany, as well as possible sherds of ARS-D at Granville – also in the Department of Manche (Giot and Fichet de Clairfontaine 1996, 56). The authors also note a possible sherd of ARS from Corseul in Brittany, but no form or production group is given. The recorded discussion appending Pirault’s article on the Rezé assemblage mentions possible examples of ARS-D from the site, but it has not been possible to verify this (Pirault 1997, 306).

It may be significant that this group of sites in eastern Aquitaine, close to the Dordogne River, with fourth and fifth century imports are not far from Vayres (Gironde), where production of céramique à l’eponge has been identified (see Sub-chapter 7.2.1). This ware was exported to sites in Britain, including Exeter, in the fourth/early-fifth century. The identification of fifth or even sixth century Mediterranean amphorae in the area potentially indicates enduring connections between this region and the Atlantic Seaboard. An example of LRA4 was also identified at Dax [211], to the far south-west of Aquitaine, although the date of this vessel is unclear (Amiel and Berthault 1996, 257). A doctoral research project is currently being conducted by Joachim Le Bomin at the University of Paris 1 PanthéonSorbonne to compile evidence of late Roman and early medieval Mediterranean imports to western France, which should reveal additional examples of fifth and sixth century imported amphorae. These published examples, nevertheless, reveal that Mediterranean amphorae of fifth century or later date are not exclusive to sites in western Britain, but are present, albeit in small quantities, at a scatter of sites in north-western and south-western France. Nevertheless, what has become clear – through Pieri’s catalogue, Amiel and Berthault’s discussion, and additional publications – is that the site in Atlantic France with the largest quantity of late Mediterranean imported amphorae is Bordeaux (see Sub-chapter 5.3). In particular, the recent publication of the amphora group from Place CamilleJullian has revealed a large group of late types that shows similarities, and important differences, to the British finds.

Once again, the evidence from Bordeaux stands out. The presence of imported Mediterranean finewares in southwestern France was confirmed by an example of LRC 3 from Saint-Christoly [202] (Mayet and Picon 1986, 130). Although Wooding had discounted the presence of LRC in western France, this single French example was included by Campbell in his map of LRC distribution (see Subchapter 2.3.5). Since then, however, the information from Bordeaux has been significantly supplemented with the publication of finewares – ARS and LRC – from excavations at Place Camille-Jullian, as discussed below. Despite the scatter of imported amphora sherds along the Atlantic seaboard and up the Loire, to the north of Bordeaux there have been no other published identifications of LRC, although it seems likely that this will change (Guitton, pers. comm.). This relative scarcity of information is doubtlessly connected to the long history of research on these finewares in Britain, and, as with the amphorae, current research by Le Bomin should provide new understandings of late fineware imports to western France. Furthermore, the very patchy information from central-western France – equating to the late Roman Province of Aquitania Secunda – will be expanded by ongoing research by David Guitton (INRAP), following the analysis of pottery from recent excavations and the reassessment of older assemblages (such as from Poitiers). This will consider wider trends in Gaulish productions (DSP, Argonne ware and coarsewares) as well as Mediterranean imports. Most notably, recent investigations at Saintes [184] have revealed an important assemblage of imports, including North African amphorae of late-fourth/ early-fifth century date (equivalent to the late Romano-

5.2.2 Imported Mediterranean finewares This review of imports in Atlantic regions of France has, by necessity, concentrated on amphorae. Published parallels for the imported Mediterranean finewares recorded in western Britain remain relatively scarce. As discussed in Sub-chapter 3.3.2, it was decided within this project to only record examples of ARS from Continental sites identified as ARS-D (unless specific late forms of ARS-C were stated), in an attempt to maintain some parity with the post-Roman imports recovered in Britain. Consequently, any potential examples of early forms of ARS or ARS-C from western France have not been considered. The examples discussed here are not expected to represent an exhaustive list, but to reflect the currently available evidence. At present, this would indicate that late-fifth and 94

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France The comparative examination of these urban assemblages by Amiel and Berthault revealed relatively small amounts of third century imported amphorae at both Bordeaux and Toulouse. However, there seems to have been a considerable increase, from the fourth century, of Spanish and North African imports – marked by a higher proportion of Spanish vessels at Toulouse, and North African amphorae at Bordeaux (Amiel and Berthault 1996, 256). This would suggest a growth in long-distance contact at both towns in the fourth century. East Mediterranean amphorae appear at both towns from the fifth century (Amiel and Berthault 1996, 256), linked to a widespread expansion of East Mediterranean wares (Reynolds 2010, 105) and paralleling their later-fifth century distribution in western Britain (see Table 12).

British imports considered in Sub-chapter 3.2.4), as well as ARS-D of equivalent date (including Hayes 61 and 63), but also ARS-D dating to the late-fifth and early-sixth century (including Hayes 87) (Guitton forthcoming). These finds remain to be published, but it has become apparent that Saintes, like Bordeaux, will increasingly feature in discussions of Late Antique and early medieval exchange in the Atlantic. For now, understandings of imported finewares in Atlantic France must remain provisional. It does seem, nevertheless, that western France does not have the same quantities of late-fifth and sixth century fineware as seen in Atlantic Spain and Portugal. As the quantity of imported amphorae in Britain and Ireland far exceeds the number of fineware vessels, the scarcity of imported finewares in western France could be seen to reflect a similar bias. The quantity of imported fineware vessels might therefore rise as the number of amphorae increases, but at a proportional rate. These relative differences may also reflect underlying patterns in Atlantic trade. As will be discussed, fineware vessels and amphorae may have been shipped separately (finewares being carried with grain). As such, these differences in distribution might hint at distinct routes in operation and/or diverse cargoes being carried along the Atlantic seaboard. Such patterns cannot, however, be easily used to reconstruct the primary distribution of specific commodities. Instead, complex systems of secondary redistribution, connected to demand, consumption and exchange would have contributed to this differential dispersal.

The relatively large quantity of imported amphorae found at Bordeaux demonstrates the significance of the city within Atlantic exchange systems. Amiel and Berthault recorded 79 imported amphorae of fifth century or later date from Bordeaux, but this figure is now approximately 100 (see Appendix K). This is equivalent to the conservative estimate of the total number of amphorae recovered from Tintagel but is less than the 150 vessels indicated by Thorpe (see Sub-chapter 4.2.3). The figure from Bordeaux is also more than the number of amphorae identified from Bantham (Reed et al. 2011). Although potential differences in ceramic methodology and excavation history prevent a straightforward comparison, it is clear that the growing data from Bordeaux presents an Atlantic location with an assemblage on a scale with the largest British groups. However, the specific forms of Mediterranean pottery imported to Bordeaux must be considered against the British examples before a closer connection can be established. Pieri’s catalogue notes three sites in Bordeaux with late Roman Mediterranean imports: Saint-Seurin, SaintChristoly and Place Camille-Jullian. These same three sites provided the data for Bordeaux in Amiel and Berthault’s analysis, although individual totals were not presented for each site. Additional information on Mediterranean imports has been provided by Armelle Guériteau, and is noted in Sub-chapter 5.4.3.

5.3 Case Study: ceramic imports to Bordeaux 5.3.1 Introduction The evidence compiled by Amiel and Berthault from southwestern France revealed a scatter of sites with late Roman imported amphorae – characterised by a bias toward LRA4, then smaller quantities of other East Mediterranean and North African types (Amiel and Berthault 1996). The amphorae were largely of fourth or early-fifth century date; vessels of later-fifth or sixth century date (more directly comparable with the British post-Roman imports) proved to be rarer. However, their study demonstrated that the largest quantities of vessels, and with the longest chronological range, had been recovered from excavations in Bordeaux (in the modern Region of Aquitaine) and Toulouse (in the Midi-Pyrénées). These urban centres were within the region settled by the Visigoths after AD 418 (Wickham 2005, 44). However, despite Bordeaux being the capital of the late Roman province of Aquitania Secunda in the fourth century, it was Toulouse that became the capital of the Visigothic kingdom in the fifth century. The court, nevertheless, often resided at Bordeaux, demonstrating its continuing importance (Barraud and Maurin 1996, 38). After the Frankish conquest of Aquitaine in the early-sixth century Bordeaux changed hands frequently, but there is limited textual information on the evolution of the town from the early-fifth to the sixth century.

The archaeological investigations conducted at Bordeaux [197-204], particularly from the 1970s, revealed the ancient town to have been at its maximum extent by the start of third century. However, by the end of third century large, quadrangular rampart walls were constructed that enclosed a more restricted area (Barraud and Maurin 1996, 42). The fortification and contraction of the town was also accompanied by a major change in the focus of the settlement. The later Roman settlement was centred on the Devèze, a stream running east-west into the Garonne (see Appendix B.7). The port at the mouth of the Devèze was integrated into this fortified area, demonstrating the importance of protecting and securing external connections and trade. By contrast, areas to the north, the location of large municipal structures and the amphitheatre, appear to have been abandoned by this date (Barraud and Maurin 1996, 41-9). The 95

Links to Late Antiquity Table 12. Vessel counts of late imported amphorae at Bordeaux and Toulouse. After Amiel and Berthault 1996, 256, Tables 1 and 2

Toulouse

Bordeaux

Spanish

African

East Med

Indeterminate

Total

3rd century

3

6

-

-

9

4th century

61

38

-

-

99

5th century

70

16

12

-

98

6th century

-

-

3

-

3

Uncertain Date

4

5

-

7

16

Total

138

65

15

7

225

3rd century

-

6

-

-

6

4th century

10

81

-

-

91

5th century

7

35

9

-

51

6th century

-

4

24

-

28

Uncertain Date

-

2

3

3

8

Total

17

128

36

3

184

the mid-fourth century the site had developed into a large early-Christian necropolis which continued in use into the seventh century (Barraud and Maurin 1996, 52). The excavations at Saint-Seurin produced evidence of the earliest Christian burials in Bordeaux. The lid of a fourth-century stone sarcophagus decorated with Christian imagery was found during the first excavations at the site in 1909-10 (Barraud and Maurin 1996, 37). These early investigations and subsequent work in the 1960s revealed a dense cemetery characterised by a wide variety of burial-types. More marble sarcophagi were discovered, as well as tile-burials, burials with stone and wooden coffins and burials without any fittings. Furthermore, a large number of amphorae were uncovered that had been reused for infant inhumations (Watier 1973).

excavations at Saint-Christoly and Place Camille-Jullian, both inside the fortifications, revealed the areas around the port to be densely occupied and economically active from the fourth century onward, and demonstrated the continuing operation of long-distance trading networks. The investigations at Saint-Seurin also revealed evidence of these long-distance connections, but moreover provided indications of contemporary social transformations – specifically the emergence of Christianity. 5.3.2 Saint-Seurin The earliest examples of imported amphorae from Bordeaux were recovered from excavations beneath the basilica church of Saint-Seurin, to the north-west of the Roman town and outside of the Late Antique fortifications (Barraud and Maurin 1996, 51; See Appendix B.7, Figure 5.1). By the start of the fourth century a number of mausolea (many decorated with frescoes) had appeared in this area, centred on a building of unknown function. By

A total of 71 ceramic vessels were identified that had been used for burial, although Brigitte Watier, who published the assemblage in 1973, estimated the presence of another 60 vessels based on unidentifiable fragments. 69 of the identified vessels were amphorae, while two jugs were found that had also been reused (Watier 1973, 115). Watier’s report does not include a full quantification by type, but the majority of the imported amphorae would appear to be of African origin and to date to the fourth and fifth century. The presence of vessels equating to spatheia is recorded from the later levels, although it is not clear how many of these were present. Three larger cylindrical amphorae are also described, which would appear to include the two vessels of North African Keay 36 given by Berthault as possible sixth century imports (Watier 1973, 27; Berthault 2012, 317). In addition, six examples of East Mediterranean LRA4 were recovered, which had also been used for infant burials. Four of these were later identified as LRA4A of fourth or fifth century date, while one was an LRA4B of sixth or seventh century date (Pieri 2005, 50; Berthault 2012, 317).

Figure 5.1. The church of Saint-Seurin, Bordeaux. View facing north across the location of excavations

96

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France 5.3.3 Saint-Christoly

British post-Roman imports, and presented contrasting evidence for connections between the town and the Atlantic Seaboard. The particular importance of this excavation is that it revealed a continuous stratigraphic sequence between the first and fifteenth century and provides information on the very latest Mediterranean imports to Bordeaux (Amiel and Berthault 1996, 255).

A major excavation was conducted at Saint-Christoly between 1981 and 1982, within the heart of the later Roman fortified town. The excavations revealed a densely occupied urban area along the Devèze that seems to have functioned in the fourth and fifth centuries as a ‘rear-port’ for the frequent transfer/transhipment of goods arriving from the Garonne (Gauthier 1983, 453; Barraud and Maurin 1996, 49). Numerous buildings and structures were uncovered, including large private dwellings (some with elaborate mosaics) baths, workshops, as well as the warehouses, granaries and wooden quays associated with the port itself (Barraud 2012, 10).

The excavation was conducted between 1989 and 1990, prior to the construction of an underground car-park, although various delays meant that the disparate reports were not collated into the final publication until 2012 (Riuné-Lacabe 2012a, 22). The site itself was close to the Roman port, and situated to the south of the Devèze (see Appendix B.7). Although at the periphery of the earlier Roman town, this area became central to the late Roman settlement, following the construction of the town walls. The excavations produced evidence of occupation and activity in the area from the first century onward, and in particular revealed the presence of warehouses connected to the port (Barraud and Maurin 1996, 49). In the earlyfifth century a large domestic structure with an extensive heating system was built at the site – presumably linked to escalating demand for residential space. This building was abandoned in the late-fifth or early-sixth century, at which point three more modest buildings were installed in the ruins (Riuné-Lacabe 2012b, 97-125). These buildings were themselves destroyed/abandoned in the seventh century. This marked the latest evidence of habitation at the site until the eleventh century, although multiple pits of seventh or eighth century date were identified and excavated (Riuné-Lacabe 2012b, 119-22).

Although the finds from the site have been exhibited, the excavation has never been fully published. Nonetheless, Frédéric Berthault, who inspected the amphorae, has graciously shared his vessel counts by broad amphora type, and this has been integrated into Appendix K. However, as this does not comprise a detailed breakdown of types by date, the finds from Saint-Christoly cannot be directly equated with the totals given in Amiel and Berthault 1996. The count from Saint-Christoly also includes a number of North African and Spanish amphorae dating to the fourth century or earlier, which have not been included in Appendix K. Two additional Iberian amphorae were also identified: one Almagro 50 and one Almagro 51C (Berthault, pers. comm.). Both types are typically dated from the third to the fifth century (University of Southampton 2005). Of the later types, Berthault records 23 spatheia and five North African amphorae of ‘grande dimension’ (Berthault, pers. comm.). Although the exact dates for these are not clear, they potentially represent fifth century vessels. Berthault has elsewhere commented on the low quantity of sixth century pottery from the excavation, as the latest levels had typically been destroyed by the construction of modern cellars (Berthault 1999, 252). Finally, three examples of LRA4 were identified. These were recorded by Pieri as LRA4A, and therefore of fifth century or earlier date (Pieri 2005, 50).

These late features appear to be covered by a ‘dark-earth’ deposit, sealing the early medieval features. A phase sequence was developed for the site, based on the finds (principally the pottery) and stratigraphy, although not on scientific dating. This is shown in Table 13. The excavations produced significant quantities of Late Antique imported pottery, specifically ARS and LRC (Bonifay 2012) and amphorae of Spanish, North African and East Mediterranean origin (Berthault 1999; updated version 2012). The pottery shows many comparable features with the British imports, but also notable differences.

The presence of LRC from the excavations at SaintChristoly was noted by Mayet and Picon (1986, 130), but it is not clear if this was the only example of Mediterranean fineware found at the site. However, it is certain that the excavation produced very large quantities of DSPA, confirming Bordeaux as a production centre (Gauthier 1983, 450). Soulas notes that Saint-Christoly and Place CamilleJullian each produced many thousands of sherds of DSPA (Soulas 1996, 237). In addition, over 800 coins were recovered from the excavations at Saint-Christoly. These dated from the first century onward, but notably included a Frankish imitation of a golden tremissis of Justinian I. (Gauthier 1983, 450).

Amphorae The amphora assemblage comprised 57 amphorae, of which 49 could be dated to the fifth century or later (Berthault 2012; see Appendix K). The group demonstrates parallels to the post-Roman imports in Britain, comprising largely North African amphorae and East Mediterranean amphorae, although there was a higher proportion of the former in the fifth century and the latter in the sixth (Berthault 2012, 314). The fifth century examples also include four amphorae of possible Lusitanian origin – one recorded as Almagro 51C and three as Almagro 51B – although these identifications are subject to debate (Berthault 2012, 315; see Sub-chapter 3.2.5). Iberian amphorae have not

5.3.4 Place Camille-Jullian The publication of the excavations at Place Camille-Jullian (Maurin 2012) revealed new data of major relevance to the 97

Links to Late Antiquity Table 13. Later Roman and early medieval phases at Place Camille-Jullian, Bordeaux. Based on Soulas 2000, 145 and Maurin 2012, 84-125, 412 Phase

Suggested Date

Activity

8

c. AD 130 to 400

Occupation/use of warehouse structures

9

c. 400-450

Destruction of Phase 8 structures

10

c. 425-480

Construction of large residence with extensive heating system

11

c. 480-520

Destruction of large Phase 10 building

12

c. 500-600

Construction and occupation of three smaller domestic buildings within ruins of Phase 10 building (‘Buildings 1-3’)

13

c. 580-620

Abandonment/destruction of ‘Buildings 1 and 2’, continuing occupation and remodelling of ‘Building 3’

14

7th/8th century

Continuing occupation/partial retraction of ‘Building 3’. A large number of pits dug across the site – of uncertain but probable artisanal function

Figure 5.2. Selected LRA1 and 2 from Place Camille-Jullian. From Berthault 2012, Figure 8

typically been seen as part of the ‘package’ reaching postRoman Britain, although amphorae of southern-Spanish and possibly Portuguese origin have potentially been identified at Tintagel (see Sub-chapter 3.2.4). An Almagro 51 (showing similarities to Almagro 51C) had been previously identified from Bordeaux, within an assemblage of nearly 500 amphorae found during a rescue excavation at the Allées du Tourny (Laubenheimer and Watier 1991, 14). This example was not dated, although the latest activity at this site was assigned to the late-third century (Laubenheimer and Watier 1991, 8).

Fineware Bonifay’s report on the fineware from Place CamilleJullian also discusses possible links with supply to Britain, but he suggests that the forms do not necessarily reveal a straightforward parallel. The identified LRC from Place Camille-Jullian (five sherds from two vessels) is all of the same form LRC 3 that characterises its British distribution (Bonifay 2012, 257-8).

A significant proportion of the total assemblage consisted of North African amphorae, particularly from the fourth century. The earlier arrivals comprise examples of Africana IIC and Africana III (Keay 25) (Berthault 2012, 315). The later African vessels comprised eleven spatheia and two Keay 57 amphorae of fifth century date, four Keay 62 amphorae of sixth century date, and one vessel identified as a Keay 61 amphora of seventh century date (Berthault 2012, 314-6). Although the presence of African amphorae has been observed within the British and Irish imports, these are much less common, and have not, to date, been identified to specific types (see discussion in Sub-chapter 3.2.4).

The ARS from Place Camille-Jullian (a minimum of 14 vessels) also shows some similarities to the vessels recovered in post-Roman contexts in western Britain, principally Hayes forms 91C, 99A, 103 and 104 (Bonifay 2012, 256). However, other later forms were recognised which have not been observed within British assemblages – specifically ARS 90, 105 and 109A – which again demonstrate the continuing importation of Mediterranean pottery to Bordeaux in the first half of the seventh century (Bonifay 2012, 256). These suggested a stronger association with the fineware assemblage from Vigo, although certain discrepancies were also observed (Bonifay 2012, 256; Sub-chapter 6.3.2).

Most significantly, whereas the ceramic data from SaintChristoly and Saint-Seurin would suggest exchange with the Mediterranean ceased by the start of the sixth century, the latest East Mediterranean and North African amphorae from Place Camille-Jullian indicate importation into the early-seventh century (Berthault 2012, 317), well beyond the mid-sixth century date generally given for the end of Mediterranean pottery imports to Britain. If Bordeaux was indeed supplied as part of the same exchange systems, this signals a foreshortening of the northern extent of the Atlantic routes by the later-sixth century.

The Place Camille-Jullian excavations have additional relevance for dating finds of DSPA in Britain and Ireland. The site produced 4680 sherds of this ware, although a large proportion were thought to be residual in later contexts; a minimum total of 1075 vessels was identified from levels dating between the fifth and seventh century (Soulas 2012, 247). Campbell has suggested a largely sixth century date for the Insular examples of DSPA (up to a minimum of 29 vessels from 16 sites), as the forms encountered seem to belong later in the DSP repertoire (Campbell 2007a, 27-8). Soulas’ chronological table of DSPA at Place Camille98

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France Jullian records the presence of Rigoir Form 29 mortaria (as found at Tintagel and Dinas Powys) from the later-fifth century and throughout the sixth, and Form 16 intermittently from the early-fifth century, though more frequently from the mid-fifth century onward (Soulas 2012, 247) (see Figure 5.3).

into the sixth century; this form peaks at Place CamilleJullian in Phase 14, although it is present from Phase 12 onward (Soulas 2000, 14-50). The pitcher Rigoir 36 is replicated in the contemporary coarseware assemblage found at Bordeaux. This is almost certainly the E ware form E4B identified in British contexts (see Sub-chapter 5.4.3).

Campbell’s synthesis revealed the British and Irish assemblage to be characterised by Rigoir Form 4 plates, Form 14/16 bowls (the absence of Rigoir 14 at Bordeaux might suggest Rigoir 16 is more likely) and Form 29 mortaria (Campbell 2007a, 27). This pattern is said to contrast with Bordeaux, where Rigoir 1, 4 and 6 are the most common and Rigoir 29 is generally uncommon. Campbell suggested that this difference might represent the deliberate selection of forms for export, or else reflect chronological factors (Campbell 2007a, 27). Examination of the data from Place Camille-Jullian confirms that the British collection does largely reflect the typo-chronology of DSPA as evidenced at Bordeaux. The quantity of DSPA mortaria (Rigoir 29) found in Britain is certainly disproportionally high, and it might be that this form was, as Campbell suggested, specifically desirable at Insular sites. However, the British collection as a whole equates well with the typical pattern of DSPA from phases 11 to 14 at Place-Camille Jullian (late-fifth century onward), by which point Rigoir 1 had declined and Rigoir 29 had appeared (Soulas 2012, 247). This date-bracket matches well with the established date for the appearance of the Mediterranean wares in Britain in the later-fifth century, and particularly, the first half of the sixth.

It may be that future considerations of the stratigraphic relationship between DSP forms and the Mediterranean imports from Bordeaux will have additional implications for the chronology of Mediterranean and Continental imports in Britain, especially as the published data from France increases. Coins Finally, a major discovery at Place Camille-Jullian was the first Byzantine coins of sixth/seventh century date from excavations at Bordeaux. Of the six bronze coins recovered, one was assigned to Justin I (AD 518-527), one to Justinian I (this was dated after 538/539-598), two to Maurice (582-602), and two to Heraclius (610-641) (Bost 2012, 397-8). The coins were minted at Constantinople, Antioch, Nicomedia and Carthage. The scatter of Byzantine coins found in Britain has not typically been seen to reflect a genuine post-Roman distribution; instead these finds have been interpreted as recent losses and/or of dubious provenance (Campbell 2007a, 74-6). Sam Moorhead’s recent assessment of Byzantine coin data from the Portable Antiquity Scheme offered a more optimistic outlook for potential fifth to seventh century losses (Moorhead 2009). However, as no examples have been found in association with imported Mediterranean ceramics it has not been possible to make definite connections with the Atlantic channels reaching western Britain (Campbell 2007a, 74). As coins arriving in the West are likely to have been melted-down for reuse (Harris 2003, 42), it cannot be confirmed that coinage did not arrive in western Britain.

The data from Place Camille-Jullian is less useful for the end date of the British examples of DSPA. Although the quantity of this ware remained largely the same between Phases 13 and 14 (approximately 17% of the total number of sherds), its final disappearance is obscured by issues of residuality (Soulas 2000, 146). The absence of Rigoir 36 in Britain might indicate that DSPA was not imported late

Regardless, the coins from Place Camille-Jullian, found in conjunction with the imported amphorae and fineware, present more solid evidence for contact – direct or otherwise – with the Byzantine world. Justinianic coins minted at Carthage had previously been dredged from the Gironde estuary, again demonstrating the potential arrival of coinage via the Atlantic routes that brought the East Mediterranean pottery (Lafaurie and Morrison 1987, 69; Campbell 2007a, 75). 5.3.5 Discussion The imported material from Bordeaux – and particularly from Place Camille-Jullian – demonstrate that the finds of Mediterranean pottery in western Britain can be longer be seen as exceptional in the Atlantic. The quantities of Mediterranean imports – which seem likely to increase as further excavations are conducted and published – reveal the city’s enduring integration into Atlantic networks. This would certainly support Reynolds’ suggestion that

Figure 5.3. Chronology of DSPA form repertoire at Bordeaux. From Soulas 2012, 247

99

Links to Late Antiquity 5.4.1 Efforts to identify Continental parallels

Bordeaux operated as an entrepôt on the Atlantic route supplying post-Roman Britain (2010, 109). However, these investigations also point to sustained demand for exotic commodities in Bordeaux itself from the fourth century onward.

The occasional efforts made to establish Continental parallels for the Mediterranean wares imported to Britain have been discussed in Sub-chapter 2.3. These attempts have been much more substantial in relation to E ware. The desire to identify a production source for the pottery and reveal the connections signified by its distribution presented a challenge to British researchers – to match their examples of E ware to vessels recovered at Continental sites. However, these efforts have had limited success. Only a small scatter of E ware vessels from France has been identified, which has not revealed its source.

The increasing focus on East Mediterranean wares from the fifth, and particularly from the sixth century, presents a strong parallel for the contemporary spread of these wares to western Britain and Ireland. However, the assemblage from Place Camille-Jullian shows some deviation from the Insular patterns – specifically the continuing importance of North African amphorae, the higher proportion of ARS to LRC and, most notably, the very late end-date for the cessation of Mediterranean supply. The mid-sixth century seems to mark a point of realignment in the Atantic supply-networks, and the role of Bordeaux in this later phase will be examined more closely.

Thomas’ 1959 article on the imported wares was the first publication to discuss E ware (his ‘Class E’ pottery) at length. As with the Mediterranean material, Thomas made use of Wailes’ research on parallel wares in France to discuss the distribution and likely origin of E ware. Wailes’ investigations in France in the winter of 1958-9 had produced a few examples of E ware from occupation sites in Brittany and the Channel Islands, but had indicated that the pottery was not generally known in France south of the Loire (Wailes 1953, 143). These preliminary observations were communicated in Thomas’ article (Thomas 1959, 98).

5.4 E ware: new evidence for production and distribution The presence of pottery of Mediterranean and Continental origin at sites in Britain and Ireland has been interpreted as reflecting two overlapping, but essentially separate, importation systems. The latter phase is characterised by ‘E ware’, a coarseware thought to originate in western France.

In contrast to the absence of parallel Mediterranean wares, Thomas’ appended ‘1959’ catalogue does include a section on supposed Continental finds of E ware; this was largely based on information supplied by Wailes. The examples comprise a body-sherd from the Channel Islands (from Alderney), a number of sherds of E1 jars from three sites in Brittany (Les Cléons [173], Mene-Geren and Guissény [158]), and a couple of examples identified by Thomas himself from sites in Northern France and the Netherlands (Thomas 1959, 98, 110). Of these, only the examples from Guissény and Les Cléons were retained as definite examples of E ware in Ewan Campbell’s recent catalogue (Campbell 2007a, 47, Figure 35). Thomas had discounted his examples from Northern Gaul and the Netherlands by the compilation of his 1981 catalogue, while additional examples from the Channel Islands included at this date were themselves subsequently dismissed by Campbell (Thomas 1981, 24). Significantly, the pottery from Hamwic that was included in Thomas’ 1959 catalogue as a ‘possible’ or ‘dubious’ example of E ware was rejected by Wailes, who confirmed that the ware was not a feature of regions under ‘Saxon’ control (Wailes 1963, 135-8). There have been no subsequent identifications of E ware from eastern England. A second research trip undertaken by Wailes in 1959 produced further potential examples of E ware from cemetery sites in north-eastern France and the lower Rhineland. These parallel finds were plotted on a distribution map (Figure 5.4), which revealed a scatter of Atlantic finds, an apparent ‘gap’ in western Normandy, and a clear cluster in eastern Normandy/the lower Seine region (Wailes 1963 143).

Figure 5.4. Wailes’ distribution map of E ware, showing a suggested source. Redrawn from Wailes 1953, Figure 173

100

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France There is limited information on two of the other sites listed by Giot. The body-sherd from Longy Bay, Alderney, on Guernsey was described as ‘unlocated’ in Thomas’ 1981 catalogue and has since disappeared from catalogues of E ware (Thomas 1981, 24). Similarly, the potential E1 rim from Mané-Geren/Mene-Geren (Plaudren, Morbihan) was described as ‘mislaid’ by Wooding and illustrated as an uncertain example in Campbell’s 2007 monograph (Wooding 1996, 78; Campbell 2007a, 47). However, E ware was later identified from another site in Brittany – Le Yaudet – which also produced sherds of Aegean LRA2. These three body-sherds were probably from the same vessel, and possibly from an E1 jar or E4 jug (Campbell 2007b, 87). They were found within plough-soil deposits overlying the late Roman levels at the site, and were associated with an early medieval phase of reorganisation of the site. The formation of new plots of land and the building of the first church suggests the presence of a monastic settlement (Cunliffe and Galliou 2007, 196-8). Radiocarbon dates from associated corn-drying ovens have confirmed agricultural activity between the seventh and eighth century (Campbell 2007b, 88).

These examples of ‘E ware’ in north-eastern France were later reviewed by Campbell and were illustrated (after Wailes’ original drawings) in his 2007 monograph, where he determined that they did not represent true E ware, as imported to Britain, but only vessels of ‘white gritty ware’ of a general similarity (Campbell 2007a, 5, 48). By 2007 Campbell’s research had concluded that Continental finds of E ware were ‘few’ and ‘scattered between Brittany and the Charente’. This small dataset included vessels recognised by Campbell himself, others recognised by Wooding, and the last of those proposed by Wailes. 5.4.2 Continental examples of E ware Despite over fifty years of efforts to identify Continental examples of E ware following Wailes’ initial efforts, there has been relatively little progress. The total number of French sites with E ware has barely increased from Thomas’ 1959 list although the focus for distribution has shifted considerably from Wailes’ first appraisal. The most recent dataset of these Continental parallels was published by Campbell in 2007, with subsequent additional data added by 2011 (see Appendix J).

These finds of E ware from Brittany demonstrate the incorporation of a limited number of sites in the north-west of France into the networks that shipped E ware (and other products) along the Atlantic route to Britain and Ireland. This mirrors the few Mediterranean imports also recorded from the north-west. It was suggested in the report from Le Yaudet that the quantity of late Roman Mediterranean imports from ‘Armorica’ was likely to grow, and it may be that additional examples of E ware will also come to light (Cunliffe and Galliou 2007, 194). For now, we are left with a small scatter of E ware on the north-coast of Brittany, described by Campbell to represent ‘offshoots of the Insular distribution’, which clearly does not indicate the focus of its Continental distribution or its probable source (Campbell 2007aa, 48).

The four examples identified by Wailes from the north-west of France were collated by Giot (1973) in his discussion of the dune site at Guissény in Brittany. Giot described Wailes’ failure to find French parallels for the Mediterranean imports in south-west Britain and repeated the resultant interpretation – that ships from the East Mediterranean heading north on the Atlantic route did not stop in Gaul (Giot 1973, 1078). This interpretation was reinforced by the limited amount of DSPA recorded north of the Loire. However, the later Continental category of ‘E ware’ was seen to offer greater potential for revealing connections between the north-west of France and the south-west of Britain. Giot’s article was cited in Thomas’ 1981 publication, as the source of his information on examples of E ware in France, as well as in later publications by Wooding and Campbell.

Further examples of E ware have emerged in the Loire valley. Giot’s article listed the presence of ‘E1 jars’ from the latest phases of the villa at Les Cléons, c. 14km to the south-west of Nantes (Giot 1973, 110). These were again originally identified by Wailes, who commented that the excavation was ‘conducted in such a way that no stratigraphy was observed’. However, a structure post-dating the main villa complex and a seventh century gold coin provided evidence of early medieval occupation (Wailes 1963, 156, 219). Campbell later stated that the finds had been lost, but that the forms corresponded with E ware (Campbell 2007a, 48). His dataset, therefore, included the rim and base of one E1 jar, and the base of one E2 beaker, although an illustration in his monograph also shows a rejected/uncertain rim of ‘E1’ that had been identified by Wailes (Campbell 2007a, Figure 35).

The E ware from ‘La Croix’ beach at Guissény presents a parallel for the finds of imported pottery from dune-sites in the south-west of Britain. Three small sherds were found among a group of 300 coarseware sherds (initially thought to be of Iron Age date) within a layer of ‘kitchen-midden’, together with bones, shell and ash (Fleuriot and Giot 1977, 111). This midden-deposit, first revealed by a storm in 1949-50, was destroyed by the building of a road not long after Giot’s observations. Thomas’ 1981 catalogue lists three E1 jars from the site, based on illustrations in Giot’s article, but Campbell’s dataset only includes one of these, represented by a rim and base-sherd of possibly the same vessel (Giot 1973, 116-8; Campbell 2011). Giot’s description of the variable colour and granular surface of this pottery certainly compares well to E ware (Giot 1973, 119). There are also tantalising hints of other imported pottery at the site, although it is not possible to identify these from the descriptions. The report mentions a scrap of sigillata, said to possibly represent a memento, and some handles from a jug or amphora (Giot 1973, 119).

The imported amphorae and DSPA from the excavations at Rezé – situated south across the Loire from Nantes – were discussed in Sub-chapter 5.2. However, the largest proportion of the ‘late’ pottery from the site represented 101

Links to Late Antiquity

Figure 5.5. Possible E ware from Rezé. Lids numbered by Pirault 1-5, bases 10-13. From Pirault 1997, Figure 16

coarsewares. Most of the coarsewares illustrated or described in the report are clearly not E ware, but Pirault also describes a group of pottery with a heavily-tempered quartz fabric, the forms of which bear some similarity to E ware (see Figure 5.5). His illustration shows five pierced lids which resemble E5 (and seem to have ‘string marks’ on the top), although Pirault compares these with examples from later Carolingian sites. Another four bases illustrated could potentially belong to one of the closed forms (Pirault 1997, 304).

(Randoin 1981, 110). The one illustrated example shows some similarity to E1, and it might be that further parallel examples were present in the assemblage, which was not fully quantified. A possible example of E2 was also noted by Wooding from the excavations at the Monastérie de Saint Martin, Tours (Wooding 1996, 78). It might be that further comparative analysis of the coarsewares from Tours will lead to more parallel examples, but it is apparent that most of the early medieval pots discussed by Randoin (including roulette-decorated vessels) were not E ware. Any examples of this pottery at the town would seem to be imported, possibly supplementing local production, and therefore not revealing for its source or the origin of this system (Campbell 2007a, 48). The developing character of the early medieval pottery at Tours was recently examined by Philippe Husi as part of a wider study of the central Loire basin. He describes the general appearance of white-firing fabrics from the seventh century, but the majority of the illustrated examples do not correspond with E ware (Husi 2013, 25-40). There were no apparent parallels within the illustrated assemblages from other sites in this area. The limited evidence of Mediterranean imports at Tours also indicates that the town was not well integrated into Atlantic networks.

However, as the sherds were not described in more detail (i.e. by colour), and have not been examined by the present author, this identification must remain conjectural. Two examples of E ware have been identified from Tours, suggesting the possible transport of this pottery eastwards up the Loire valley. The excavations at the Château de Tours produced – within a larger assemblage of coarsewares and DSPA – a rim-sherd that was visually identified by Wooding and Campbell to be a small version of E1 (Randoin 1981, 106, no. 18; Wooding 1996, 78). The sherd was associated with ‘Period 4’ at the site – dating to the sixth and seventh centuries. Randoin’s article, suggests further examples of oxidised pottery in white or light coloured fabric with quartz inclusions, from ‘Period 5’ (seventh to ninth centuries), although the mention of feldspar inclusions might argue against the identification of E ware. Randoin describes the principle form of ‘Period 5’ as a jar with a triangular rim of varying shape that is sometimes lid-seated. The base-sherds present show these vessels to have been removed from the wheel by a string

More promising identifications of E ware were made further south, in the Poitou-Charentes Region, although once again the reported quantities are not large. Campbell observed and verified the rim of an E3 bowl found as a ‘stray, surface find’ at a churchyard in Chadenac [186], c. 27 km to the south-east of Saintes (Campbell 2007a, 48). For 102

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France Unsurprisingly, a section of Wooding’s study of contacts along the ‘Western Sealanes’ was devoted to the evidence of E ware and considerable attention given to the question of its origin. He described this level of interest as reflecting an ‘appreciation of its potential in solving chronological and cultural problems of the early middle ages in the Celtic West’, but also observed that the very ‘intractability’ of this problem had proved a ‘lure to researchers’ (Wooding 1996a, 74). In 1990 Thomas described how this ‘enigma’ had ‘spawned a minor industry of dissertations, notes and counter notes’ (Thomas 1990, 2). The discussion presented in this sub-chapter will hopefully go some way to resolving the question, but will not obviate the need for future research and publication.

Wooding the most convincing parallels were from the large early medieval cemetery at Herpès [183], c. 35 km to the east of Saintes, along the Roman road to Limoges [182]. The site was excavated in the late nineteenth century, by which time 1600 burials had been uncovered, and the original context for these finds is, unfortunately, unclear (Wooding 1996a, 77; Haith 2001, 237). The E ware vessels were identified from the collections held by the British Museum. Wooding illustrates two complete examples of E2 beakers, and indicates the presence of one more; Campbell’s dataset only includes one of these E2 beakers but adds a small jug, though without a spout, recorded as E4 (Wooding 1996, 77-8; Campbell 2007a, 48). The cemetery produced a large quantity of grave-goods, including glass, jewellery, weapons and pottery, whereas contemporary cemeteries in this region are usually characterised by sarcophagi and by a general absence of finds (Haith 2001, 236-7). These atypical features have suggested a Frankish, rather than Visigothic, presence in the region. Unusually, the site has also produced Anglo-Saxon finds, potentially demonstrating connections with cross-Channel as well as Atlantic networks, although the overall veracity of the assemblage has been questioned (Haith 2001, 237).

The search for the source Even before Thomas had defined ‘E ware’ as a class, Gerhard Bersu had recognised that the Irish finds of this pottery represented a single, foreign source (Ó’Ríordáin 1947, 73, note 107). However, this early analysis was hindered by World War Two and its aftermath. Thomas initially attributed the ware to France, but conceded that this represented ‘little more than guess-work’ (Thomas 1959, 96). In his first major discussion on E ware Thomas observed that despite the pottery having a similar distribution to the fineware and amphorae, the origin seemed to be completely different. As mentioned, he made use of Wailes’ research to catalogue parallel examples from northern France. The supposed distribution of this ware, as proposed by Wailes, led Thomas to propose a likely production source in the Middle Rhineland (Thomas 1959, 98). As such, he initially considered the ware to represent a forerunner of later Rhenish imports to Anglo-Saxon England – with a shift in the focus of distribution from western Britain and Ireland to eastern Britain indicating a ‘shift in interest’ on the part of the traders sometime between the seventh and later eighth century (Thomas 1959, 99).

Finally, Campbell’s updated dataset from 2011 included possible examples of E ware from a well excavated at the villa site of Jonzac [187], which were not mentioned in his 2007 monograph. These were identified from descriptions and illustrations in a publication by Guériteau, whose research is discussed in more detail below. The vessels comprise two E1 jars and three E4 jugs that were found in association with a brooch of fifth or sixth century date (Guériteau 2006a, 330). However, as these represent complete or semi-complete examples chosen to indicate typological features, it seems probable that more vessels were present within the assemblage. The group came from a sealed deposit associated with the abandonment of the site in the seventh century (Robin 2010, 102). Guériteau’s article also provided Campbell with possible examples of E ware from Îlot Bonnac on the outskirts of Bordeaux, comprising three E2 beakers, one E3 bowl and one E5 lid; it seems that these finds were brought to Campbell’s attention too late to be included in his 2007 synthesis.

However, within a decade this attribution was rejected by Thomas, following his reassessment of the historical and archaeological evidence and petrological analysis conducted by David Peacock (Peacock and Thomas 1967). Peacock’s analysis led Thomas to focus the search on south-western France, particularly on the Charente area. The region around Saintes was suggested as a possible source, with E ware potentially representing a precursor to the Medieval ‘Saintonge wares’ exported to Britain and Ireland (Peacock and Thomas 1967, 38-9). Thomas described sherds that had been assigned a tenth to twelfth century date from Camp du Chaillot, La Jard, ten miles to the south of Saintes, that had a similar ‘texture, colour and fabric’ to British examples of E ware (Peacock and Thomas 1967, 39). These were examined initially in thin-section by Peacock, together with sherds of E ware from Dunadd and Gwithian, but the quartz inclusions were not considered distinctive enough to determine a source. Thereafter, heavy mineral analysis of these inclusions was carried out to determine the geological character of the production

5.4.3 The ‘enigma’ of the origin of E ware The identification of examples of E ware from a scatter of sites in western and north-western France (the Regions of Brittany, Pays-de-la-Loire, Poitou-Charentes and potentially Aquitaine) demonstrated that its distribution was not confined to western Britain and Ireland. However, the number of published Continental parallels remained rather small. Tellingly, the total quantity of vessels involved was less than from many of the sites in Britain and Ireland (Campbell’s 2011 dataset suggests a possible minimum total of 19 vessels from French sites, but includes 25 from Dunadd alone). As a result, although the expanding distribution was used by Thomas, Wooding and Campbell to discuss the possible source of this pottery, these speculations have remained somewhat unconvincing. 103

Links to Late Antiquity However, Campbell’s observations largely disagreed with the Aquitanian source advocated by Thomas, which had, by that point, attained the status of a ‘near certainty’ (Campbell 1984, 35). Campbell published a ‘reconsideration’ of Peacock’s petrological analysis, observing that the original article had offered no basis for a specific link between the results of the heavy mineral analysis and the selection of Aquitaine as a likely source. Campbell also disagreed with the basic premise of Peacock’s technique – that the surface sands (used as a temper in the pottery) would reflect the underlying geology of the source and that proportions of minerals within the samples would be genuinely distinctive (Campbell 1984, 35-8). He remarked that no corroborative evidence of production in Aquitaine had come to light since the publication of this analysis and concluded that the ‘social and chronological interpretations’ resting on this attribution – specifically the model of wine exportation to Ireland – could not be proven (Campbell 1984, 38). Peacock and Hodges both offered a reply to Campbell’s reassessment. Peacock conceded that his original attribution was tentative, but reaffirmed the circumstantial evidence that pointed towards south-west France (Peacock 1984a, 38). Similarly, Hodges admitted that although alternative sources (even an Iberian source) could not be dismissed, the disparate archaeological evidence pointed toward an origin somewhere to the south or west of Tours (Hodges 1984, 40). He did, however, suggest that it would be wise to separate the ongoing search from the development of historical models, contra his 1977 treatment.

source. The four samples from Britain were confirmed to be highly similar, and the later medieval sherds from Camp du Chaillot found to compare favourably. Peacock concluded that a source in the Paris basin or the Aquitaine region was most likely for the British examples of E ware, but that the ‘western seaboard’ of the latter was more likely (Peacock and Thomas 1967, 40-4). These results led Thomas to examine the historical evidence for connections between south-west France – particularly Bordeaux and the mouth of the Garonne – and post-Roman Britain and Ireland, and to envision the Insular distribution of E ware as revealing associations with the expanding influence of the Church in Gaul (Peacock and Thomas 1967, 39). Thomas maintained this position in his 1976 publication on the imported wares, where he described the pottery as representing ceramic production of ‘a continuing Gallo-Roman tradition, under Visigothic domination’, although he conceded that archaeological evidence for export via Bordeaux and the mouth of the Garonne had failed to materialise (Thomas 1976, 251). By this time he was developing the connection between the Insular distribution of E ware and literary evidence for an active wine trade operating between Gaul and Ireland (Thomas 1976, 252). This model was fully expounded in his final, major discussion of the imported pottery (Thomas 1990). However, by this point Thomas had amended his hypothesis, offering the Loire as a more likely point of export than the Garonne, based on the ‘weight’ of historical references and limited archaeological information (Thomas 1990, 22). His new list of suggested source-locations included Nantes, Orléans and Poitiers; Tours had been discounted following the publication of a contemporary ceramic sequence that potentially included E ware, but was dominated by vessels with different characteristics (Randoin 1981; Thomas 1990, 22, note 108; Campbell 2007a, 48). In the conclusion of this article Thomas presented his ‘closing challenge’ for future research as the outgoing president of the Society for Medieval Archaeology. That he proposed the discovery of the source of E ware reflects both his long-standing attention to this issue and his belief in the importance of unlocking the connections revealed by this exchange (Thomas 1990, 22).

Campbell’s 2007 monograph summarised the ‘state of knowledge’ of this inquiry, incorporating his own geological observations with those of Peacock, while considering any Continental parallels and potentially relevant historical sources (Campbell 2007a, 46-9). He concluded that the lower Loire and, in particular, the Poitou-Charentes/ Saintonge region could be considered the most likely candidates, based principally on the more recent identifications at Chadenac and Herpès. Campbell has cited Edward James’ appraisal of Aquitaine as an ‘economic backwater’ during this period, but offered no further explanation as to why the region south of the Garonne should be discounted (Campbell 2007a, 48). Ultimately, he stressed that only the discovery of kilns for this pottery would finally resolve the question of its origin, although ‘the discovery of large numbers of vessels in a single area of France would be a pointer to the production source’ (Campbell 2007a, 47).

Thomas’ early model – founded on Peacock’s analysis – proved to be extremely influential. Richard Hodges’ 1977 examination of the ‘Early French Wine Trade with Britain’ starts with a discussion of E ware and a summary of Continental parallels for the British finds. Although he admitted that the archaeological evidence was slight, Hodges reiterated Peacock’s petrological observations, and, like Thomas, concluded that if Aquitaine was the source of the pottery, a connection with wine seemed probable (Hodges 1977, 241). The distribution of this pottery to sites of varied size and status would therefore indicate that wine-drinking had ‘permeated all levels of society’. Hodges furthermore suggested that the end of the importation of E ware in the eighth century signalled the end of this western-French wine trade, as a consequence of the Arab invasion of Aquitaine (Hodges 1977, 241).

Wooding and Zimmer: E ware in Aquitaine Wooding’s review of documentary evidence for commerce along the ‘western sealanes’ had concluded that there was considerable exchange between Gaul and Ireland in the seventh and eighth centuries, but that two of the most ‘specific references’ indicated connections between the Loire estuary and Ireland, rather than the link with Bordeaux advocated by Heinrich Zimmer (Zimmer 1909; Wooding 1996a 72). This led Wooding to conclude – as Campbell would – that the Loire represented a more likely origin for 104

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France was provided by Leo Swan and was examined by David Williams in 1982 (Wooding 1996a, 73, 78 – citing Warner pers. comm.). Williams’ report of this analysis was never published, and it is not clear from Wooding’s text – or from a parallel reference by Thomas (1990, 21) – what date the sherd was originally considered to be. The sherd is said by Wooding to have had shown similarities in fabric to E ware, but not to have been identical (Wooding 1996a, 78). However, earlier in his text Wooding had noted that in this study Williams used sherds from Abercorn in Scotland as the comparative examples of E ware, but that the authenticity of these finds had been questioned. Charles Thomas had come to doubt that the pottery from Abercorn represented genuine examples of E ware, and the vessels were later disregarded by Campbell (Wooding 1996a, 73; Campbell 2006, 11). As such, it might be that a legitimate find of E ware from St Christoly was unfairly rejected, and as a result, a potentially valid association between E ware and Bordeaux was missed. There is no mention of this possible example in Campbell’s monograph.

the British and Irish examples of E ware than Bordeaux. This hypothesis was reinforced by the (admittedly slight) evidence for Continental examples of the pottery. These examples (discussed in Sub-chapter 5.4.2) were seen to be focused on ‘the regions of the Touraine, Saintonge and Poitou’, while, as Wooding specifically noted, no E ware had been found in the Bordeaux region. He comments that Thomas’ long adherence to Bordeaux as the likely source had diverted attention from the Loire region, and ultimately hindered the resolution of this problem (Wooding 1996a, 79-81). For Wooding, the general interest in the source of E ware represented a ’fixation’ on wine as the driving factor of Atlantic commercial systems, reflecting the enduring pull of the model of trade and contact formulated by Zimmer (Zimmer 1909, 430-76; Wooding 1996a, 74). Certainly, Thomas had repeatedly asserted that the spread of E ware provided the archaeological proof of Zimmer’s model (Thomas 1976, 252; 1990, 16). As Wooding’s assessment of the documentary evidence had concluded that there was no evidence for this ‘wine-trade’ model, it seems that he was keen to reject any associated conclusions, including the potential connection between E ware and Bordeaux (Wooding 1996a, 68).

The second piece of information from Aquitaine concerns a French note on E ware by Anne-Catherine Bélier, published shortly before the publication of Wooding’s thesis (Bélier 1995). Bélier’s note describes the long search by British researchers to identify the source of E ware, summarises the various locations suggested by Thomas and cites his assessment of Peacock’s petrological analysis that led to the proposal of Aquitaine as a probable origin. As well as noting the comparable fabrics of vessels of later date from Charente Maritime, she mentions recent excavations at Saint-Christoly and Place Camille-Jullian – both in Bordeaux – which are claimed to have produced a ‘relative abundance’ of vessels with the same characteristics (Bélier 1995, 261). The associated footnote clarifies that this potential identification is based on the author’s personal observations. Wooding is quick to dismiss this, stating that Bélier had provided no evidence or ‘up to date information’, and it seems that these suggested finds were never mentioned in subsequent British publications on the imported pottery (Wooding 1996a, 78). Neither excavation is mentioned explicitly in Campbell’s 2007 synthesis, although, as noted, his 2011 dataset does include five vessels of possible E ware from Îlot Bonnac, Bordeaux. As will be shown, it seems that Wooding’s dismissal of this emerging evidence from Bordeaux significantly hampered the search for Continental parallels, and, in all likelihood, the identification of the source of this pottery.

Despite rejecting a model of Atlantic exchange that was fundamentally based on wine, Wooding speculated that wine may have formed a minor component of a much more varied exchange, with transported commodities including honey, leather, metals and salt. This trade was conducted through intermittent journeys by boats carrying mixed cargoes – a ‘tramp-steamer’ model of shipment (Wooding 1996a, 96; 1996b, 73-7). Although Wooding’s published thesis mentions metalwork and sculpture as providing evidence of these connections, the distributions of glass and E ware are noted to be particularly significant for determining the routes and context of this commerce (Wooding 1996a, 72-3). As such, Wooding spends some time summarising the various phases of research conducted to identify vessels of E ware in France – following on from Wailes’ initial research. Wooding had himself conducted a ‘limited’ investigation into Continental evidence for E ware in 1987 and 1988. This led him to conclude that E ware was ‘certainly’ present in contexts in western France – although the relatively small number of finds identified by himself, Wailes and Hodges represented only a scatter of parallel examples and not the large assemblage of a production site. Wooding also discusses a number of reported, but not necessarily genuine, examples of Continental E ware which are said to have attained a ‘long-life in oral history’ and to have been assigned more credence than originally intended (Wooding 1996a, 78). These include sherds of ‘E ware’ reported from L’île Lavret, Brittany, which were viewed and rejected by Wooding (1996a, 78).

Coarsewares from the Place Camille-Jullian excavations As part of this project the late Roman and early medieval Mediterranean imports from the excavations conducted at Place Camille-Jullian, Bordeaux in 1989-1990 were reviewed (Sub-chapter 5.3.4). This group of Mediterranean imports is particularly significant for the variety of types present, as well as the presence of a small number of examples assigned a seventh century date. However, the largest category of finds from the excavation was the coarseware pottery (‘céramique commune’) dating to be-

In addition, and by way of illustrating his wider argument, Wooding notes and dismisses two pieces of evidence from Aquitaine. The first of these concerns a sherd recovered from the excavations at Saint-Christoly in Bordeaux, that 105

Links to Late Antiquity some counts of sherds with distinctive elements (types of handle, spouts, bases and so forth) but this does not seem to be a true vessel count and it not clear how most of those listed equate to the vessel forms described in his form inventory. It is therefore possible to suggest that many – and potentially the majority – of the vessels can be equated to E ware forms, but it is not possible to provide any accurate counts for the specific forms present, or to assess relative proportions based on the E ware form typology (E1-E5).

tween the first and eighth centuries. The authors of the final report on the excavations, published in 2012, admitted that a complete study of these wares had not been conducted and that the collated reports were based on partially studied assemblages (Riuné-Lacabe 2012a, 23). The report on the coarsewares of fifth century or later date was produced by Philippe Labrouche. As the team responsible for the final publication were unable to contact the specialist this chapter consists of an unpublished report written in 1991. Labrouche’s conclusions were never updated after his initial assessment and the editor comments that this promising early research was never followed up (Maurin 2012, 293).

Another major complication with the report results from Labrouche’s consideration of fabric and firing conditions. He describes the assemblage as characterised by three fabric groups: fine, semi-fine and coarse, but these are not described in much detail (Labrouche 2012, 295). The ‘semi-fine’ pottery is said to have had a large number of inclusions visible to the naked eye, with a sandy fabric and a slightly rough exterior. The ‘coarse’ fabric group is described as full of inclusions of various sizes, which prick through the irregular surface. Either of these descriptions could feasibly be matched to E ware without the sherds themselves being photographed or examined in person. The colour range is not specified and the hardness is not described. Labrouche does not actually identify the inclusions, apart from mentioning the presence of fragments of grog in the coarse fabric-group (Labrouche 2012, 295). If the main inclusion was, in fact, quartz, the description of the large quantity of inclusions in both types could be seen to correspond broadly with E ware. On consideration, however, the description of the coarse category is, perhaps, a better match. Labrouche’s reference to inclusions that protrude through the surface is extremely similar to the description given by Campbell of the pimply surface of E ware (Labrouche 2012, 295; Campbell 2007a, 32). In addition, Campbell notes that the fragments of iron ore that are observed in variable amounts in the documented E ware fabric have frequently been mistaken for grog (Campbell 2007a, 32). Nevertheless, Labrouche does not always match these fabric groups to the forms identified, so it is impossible to be certain whether there is a clear link between forms similar to E ware and either of his fabric groups.

Although Bélier’s 1996 note mentioned an ‘abundance’ of E ware within the Place Camille-Jullian assemblage, it seems that this connection to the British classification was made after Labrouche’s assessment in 1991; there is no mention of any link to E ware in the published report. However, it is evident from the descriptions and illustrations in Labrouche’s report that much of the pottery from the early medieval phases at the site bears extremely strong similarities to E ware – as described in British and Irish publications, and particularly in the work of Thomas and Campbell. It is likely, therefore, that a significant proportion, possibly even the majority, of the early medieval coarseware vessels from Place Camille-Jullian can be equated to E ware, and presumably represent products of the same source. Unfortunately, it seems that the specialist had possession of the pottery, and as such, it cannot currently be located (Berthault, pers. comm.). As it is not currently possible to examine this material, it seems appropriate to summarise Labrouche’s observations, and to describe the likely, though unproven, associations with E ware. The illustrations shown are all from the original report and appear to have been drawn by Labrouche. The pottery appraised by Labrouche comprises material from ‘Phases 9 to 14’ that were identified at the site (see Table 13). These have been dated from the start of the fifth century to some point in the eighth century, although Labrouche also describes the continuation of some shapes up to the tenth century. He notes that the coarsewares comprised the vast majority of the pottery from these late phases, with the DSP, amphorae and imported finewares representing only a small percentage.

In a similar way, Labrouche discusses the presence of oxidised and reduced sherds within the assemblage, but does not always connect these to the forms described. He notes that it is often very difficult to establish whether the sherds are oxidised or reduced, but nonetheless attempts to connect firing conditions to his fabric groups and discuss how these evolve over time (Labrouche 2012, 295-6). Phases 8 to 11 are said to be characterised by the finer fabric, which is found to have been fired in oxidising and reduced conditions equally. Phases 12 and 13 are said to be dominated by vessels in the ‘semi-fine’ fabric, which by this point are nearly always reduced. He describes a major break between Phases 12 and 13 – which were characterised by semi-fine, reduced fabrics – and with Phase 14 – characterised by vessels in coarse, oxidised fabrics. In fact, sherds in the coarse fabric – possibly equating with E ware – are reported to be nearly all ‘oxidised’. The usefulness

The report provides a quantification table based on the 54 features/pits studied by Labrouche (Labrouche 2012, 294-5). He states that this quantification is based on sherds counts, rather than ‘NMI’ (minimum number of individuals) due to the fragmentary nature of the assemblage and that it was not possible to match most of the sherds to individual vessels. The total number of sherds quantified from these features is 10531, although this does not include sherds from contexts that were considered to have an overly high proportion of residual pottery of fifth century or earlier date. Similarly, sherds from features that contained any sherds of later medieval pottery were discounted. Labrouche’s quantification does provide 106

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France example was recovered. This seems to contrast with their Insular occurrence, where E5 is a rare form (a minimum of eleven examples identified, with most of these from Ireland). The lids are said to have a fabric that is full of inclusions, which is presumably Labrouche’s ‘coarse’ fabric group and most typical of Phase 14. However, he notes the presence of these lids in all levels between Phases 9 and 14, dating from the fifth to the seventh/eighth century (Labrouche 2012, 296).

of Labrouche’s observations can be questioned if much of this pottery is, in fact, the same as E ware. As discussed in Sub-chapter 3.4, E ware is made from an iron-poor, whitefiring clay, while the surface colour is extremely likely to be affected by variations in firing, use and deposition. Campbell notes that some examples of E ware may have a black core, due to the variable flow of oxygen in the kiln (Campbell 2007a, Appendix 2, 7). These would be described as ‘reduced’ by Labrouche’s criteria (Labrouche 2012, 295).

Another of Labrouche’s form inventory comprises ‘coupes carenées’ (also described as ‘écuelles). The earlier, fifth century, group of this class, described to be of a ‘GalloRoman’ tradition, is present in a variety of fabrics. These are succeeded by cups dating to between the fifth and eighth centuries, of which Labrouche identifies three subtypes, the last of which seems to match the descriptions and illustrations of E3 carinated bowls (Figure 5.7 and Figure 3.16).

Labrouche’s analysis focuses on the forms of the vessels which he describes in greater detail. Despite the complications in equating the fabrics, the shapes of many of the vessel-forms that are described from Place CamilleJullian match extremely closely to the E ware typological sequence developed by Thomas and Campbell. The inventory of forms starts with lids, which are extremely similar to the E5 lids illustrated from Britain (see Figure 5.6 and Figure 3.16). Labrouche states that the lids were thrown from the lump, a production technique that was recognised by Campbell for the Insular E ware (Campbell 2007a, 32; Labrouche 2012, 296). The tops of the lids show the same concentric string marks where the vessels were cut-off. The lids are also pierced, matching the example of a perforated E5 lid from Bar Point and the Irish examples referred to by Doyle (Doyle 2009, 23). Lids are stated to be a frequent form at Place Camille-Jullian, although only one complete

Vessels from this sub-group are said to have been recovered from features in Phases 9 to 14 and even later (Labrouche 2012, 297). Significantly, a number of fragments of these bowls bear ‘glassy deposits’ suggesting they were used as crucibles for glass (see Sub-chapter 3.5.2). As mentioned in Sub-chapter 3.4, a similar observation was made for some possible E ware sherds from the Mote of Mark, although these seem to have been from a jar (Campbell 2006, 11; 2011).

Figure 5.7. Selected bowls/‘coupes carenées’ from Place Camille-Jullian – including examples equating with E3. After Labrouche 2012, Figure 2

Figure 5.6. Lids from Place Camille-Jullian – equating with E5. From Labrouche 2012, Figure 1

107

Links to Late Antiquity

Figure 5.8. Selected cups/ ‘gobelets’ from Place CamilleJullian – equating with E2. After Labrouche 2012, Figure 3

The group of ‘gobelets’ or cups described by Labrouche can be equated with the British E2 beakers. Labrouche’s category comprises small, closed vessels with a concave upper zone (see Figure 5.8 and Figure 3.16), which he interprets as probable drinking vessels. He isolates three sub-types by shape/degree of carination, two of which are less common. However, the two sub-types that are illustrated do not seem vastly different and compare well to the variation seen in Insular E2. Campbell describes the E2 beakers as a largely coherent group, having dimensions that fall within restricted parameters (Campbell 2007a, 39; Campbell 2007a, 35, Figure23). This corresponds closely with Labrouche’s observations; he describes the Place Camille-Jullian cups as very regular in size, with heights ranging between 6 and 9cm and diameters between 8 and 12cm. These cups are said to have all been made in a fabric with many inclusions, and to be usually white or light coloured. Labrouche states that they appear, at the earliest, from the mid-seventh century and continue up to the tenth century (Labrouche 2012, 299). The precise basis for this claim is not clear. Certainly, these dates would be late for the established dates for E ware. There has been no suggestion of E2 beakers being any later than the other forms of E ware, but it may be that the recognition of parallel finds from stratified sites in France will reveal phases of production within the E ware repertoire. However, as Campbell comments, even if more chronological information were available from the Continent, the dates of the importation of this pottery to Britain cannot be automatically matched to the dates of its manufacture (Campbell 2007a, 46). It is also not difficult to envisage the continued production of this pottery in Bordeaux beyond the end of the phase of ceramic exports to Britain and Ireland (thought by Campbell to cease by the early eighth century – see Sub-chapter 3.4). Nevertheless, it may be that the evidence from Place Camille-Jullian signals that the discovery of E ware in Britain or Ireland cannot be automatically taken to confirm exchange limited to the sixth/seventh century.

Figure 5.9. Cups/ ‘gobelets’ from Place Camille-Jullian with pouring rims. From Labrouche 2012, Figure 4

of two sherds revealing traces of silver, copper and zinc (Labrouche 2012, 299). Although this sub-form has not been identified in Insular contexts, these vessels demonstrate the utility of E ware for metallurgical/industrial purposes, something suspected by Campbell (2007, 50). The most common of the E ware forms in Britain and Ireland is the E1 jar. Jars were also described from Place Camille-Jullian (described as ‘pots’), and were also noted to be the most common form, although no complete examples were identified, and the overall estimated number of vessels is not given (Labrouche 2012, 299). Again, Labrouche describes sub-types of this form based on rimshape, but as the fabrics are not specified for each group it is not clear which can be potentially matched to E ware. A number of the illustrations and descriptions of these groups do seem to match to E1 jars, the rims of which do show some variance. The most likely is a group of pots described as having a beaded rim offset with a groove (see Figure 5.10, Labrouche’s ‘no. 6’). These were all oxidised and made in the ‘inclusion-rich’ fabric. Like the possible E2 beakers they are also assigned a likely date from the seventh to tenth century. Some of the other shapes are characterised by finer fabrics, which suggest that not all of the pots/jars in the assemblage were E ware. Overall, due to the limitations of the data available it is not possible to identify exactly which of the Place-Camille Jullian vessels can be equated with E1 jars. One further observation can, however, be made: Labrouche notes that some of the pots/jars had been trimmed down to convert into open vessels (Figure 5.11). This draws clear comparisons to the trimmed vessels from Loch Glashan and Port y Candas mentioned in Sub-chapter 3.4.

Labrouche also describes two vessels of this ‘gobelet’ type (E2 beakers?) in a light fabric with many inclusions, but with pouring rims (Figure 5.9). These are suggested to have been used as crucibles, with chemical analysis

Perhaps the most notable potential parallels between E ware and the Place Camille-Jullian vessels are to the forms of pitchers and jugs. Campbell has noted that there are no 108

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France unique to the region of production, and might therefore represent the best evidence of this being the same pottery (see Sub-chapter 3.4). Unfortunately, the fabric for this form is not described, hampering the association with E ware. Labrouche indicates that this jug type was present in Phases 12 to 16 – and therefore, again, from the fifth to the tenth century – although the specific method of attaching the handles seems to be a feature from Phase 14. A final parallel can be made between the form categorised as a ‘pot à bec tubulaire’ which bears clear parallels to Campbell’s form E4B, of which only two examples have been identified within Insular contexts (Figure 3.16). Labrouche describes the vessel as having a trefoil shaped tubular spout which is fitted to the shoulder and attached to the rim. A single example from Place Camille-Jullian was partially reconstructed (see Figure 5.12, Labrouche’s no. 1), which suggested there were no handles. Labrouche notes that the vessel would have been expected to have handles, given its similarity to the DSP form Rigoir 36

Figure 5.10. Selected jars from Place Camille-Jullian – potentially equating with E1. From Labrouche 2012, Figure 7

complete Insular examples of ‘E4’ – jugs with a pinched spout – but this shape can now be equated with the type of jug described by Labrouche as a ‘pichet’ or ‘bouilloire’ (see Figure 5.12, Labrouche’s no. 2). Labrouche notes that there are a number of examples of jugs with a pinched rim that show a particular characteristic. The handles are attached opposite to the pinched spout and under the lip, but frequently the lip has been crushed with a finger to reinforce its attachment (Labrouche 2012, 304). This clearly represents the same production method described by Campbell for E4 jugs, which have a thumbsmear on the upper surface of the rim. As noted, Campbell concluded that this characteristic might be a feature

Figure 5.12. Pitcher and Jug from Place Camille-Jullian – equating with E4B and E4 respectively. From Labrouche 2012, Figure 8

Figure 5.11. Trimmed pots/jars from Place Camille-Jullian. From Labrouche 2012, Figure 5

109

Links to Late Antiquity DSPA and E ware, the latter of which was examined in more detail (Chapelot 2012, 137-9). Discussing the likely origin of E ware, Chapelot cites Bélier’s note that had reported the pottery as present at St Christoly and Place Camille-Jullian, but comments that, as the two sites had not been published this identification could not be verified. Unlike Place Camille-Jullian, the excavations at St Christoly remain unpublished and there does not seem to have been any further reference to E ware from this site. In an associated footnote Chapelot refers to research being carried out by Armelle Guériteau, as part of a PhD project on early medieval pottery in the north of Aquitaine, undertaken at Université Michel-de-Montaigne-Bordeaux 3. He records that she was not aware of any E ware from the two Bordeaux sites mentioned, but had identified various sherds from excavations at Puy-Paulin in Bordeaux [199] (excavated in 2008) that were analogous to E ware (Chapelot 2012, 138, note 154). As mentioned, Campbell’s 2011 dataset had included some potential finds of E ware from a 2006 article by Guériteau – from the villa at Jonzac and from Îlot Bonnac, Bordeaux [200] (Guériteau 2006a; Campbell 2011). It transpires that Guériteau’s research, which is ongoing through work for INRAP (Institut National de Recherches Archéologiques Préventives), has revealed the likely presence of E ware at a number of additional excavations in Bordeaux (locations shown in Appendix B.7). She has graciously provided copies of the ceramic reports from a number of rescue excavations in the city, and has generously allowed the results to be summarised as part of this consideration. The following illustrations have also been included with her kind permission.

(Labrouche 2012, 305; see Figure 3.12). This is significant as Campbell cites the same DSP form when questioning whether E ware form E4B would have had handles; the presence of handles could not be determined by the two, fragmentary Insular examples from Buiston and Scrabo [26] (Campbell 2007a, 39-44). Fifteen vessels of this shape were observed at Place Camille-Jullian in various fabrics, so at present it is not clear that it can be solely linked with E ware. The form is said to appear in the second half of the sixth century, to be relatively common in the seventh and eighth century, and to probably disappear sometime in the eighth century (Labrouche 2012, 44). Labrouche describes other shapes, namely plates, mortaria and tripod-pots, but these were probably not the same as E ware. The coarsewares as a group are described as rarely decorated, and that any decoration present is limited to grooved horizontal lines (Labrouche 2012, 296). This would certainly equate to the description of E ware from Insular contexts (Campbell 2007a, 32). Labrouche also describes the general trajectory of the forms, although the difficulties of associating forms and fabrics limit the usefulness of this for determining the appearance and disappearance of forms connected to E ware. His summary would seem to suggest that the Insular dates for E ware might be rather early, and that the pottery was more connected to the seventh/eighth century (and possibly later) than to the later-sixth/seventh century. However, it might be that this is a construct of Labrouche’s categorisation by fabric and firing conditions. It should also be noted that as a result of the hiatus between the excavation and publication, the chronology of the site is largely based on the finds (especially the fineware pottery) and on stratigraphic considerations, rather than any scientific dating (Riuné-Lacabe 2012a, 23-7; 2012b, 121). As a result, phases following the disappearance of these chronological markers will be less securely dated. However, despite these difficulties, a general picture can be established of a very large assemblage that contained a variety of vessel forms in a number of fabrics – but that contained from about Phase 12 (sixth century) forms that are similar to E ware. Later – from Phase 14 (seventh/eighth century) – pottery that appears directly comparable to E ware comes to dominate the assemblage, by which point DSP and the Mediterranean imports had declined. Labrouche describes Phase 14 as representing a major change in the pottery tradition at the site, witnessing the appearance of new forms and new methods of production. Despite the difficulties with this assemblage and the chronological problems of the excavation, Place Camille-Jullian clearly provides important parallels for the understanding of the Insular E ware, which deserve future consideration.

Place Gabriel Guériteau’s report from the excavations at Place Gabriel [198] in 2002 describes pottery recovered from a sealed well-deposit excavated outside of the city walls, on the banks of the Garonne (Guériteau 2006b, 1). Of the 858 sherds recovered, 711 were identified as of early medieval date, while the rest comprised DSPA and residual Roman material. The assemblage consisted of small sherds as well as complete vessels, which suggested to Guériteau that the pottery represented losses dating to the primary use of the well – connected to the drawing of water (Guériteau 2006b, 7). However, there were problems with the dating of the material. Guériteau identified the pottery as identical to types recovered from deposits from the city dating to between the fifth and twelfth centuries. However, dendrochronological dates for the wooden well-structure indicated that the deposits filling the well dated to after the start of the twelfth century (Guériteau 2006b, 7). Clearly this hampers a straightforward identification of the pottery as E ware. However, as in Labrouche’s report, the descriptions and illustrations of the pottery can be directly matched to the Insular category.

Ongoing research by Armelle Guériteau (INRAP) A recent publication by Jean Chapelot includes a short discussion on the ceramic evidence for late Roman and early medieval contacts between the ‘centre-west’ of France and Britain and Ireland. These comprise céramique à l’éponge,

The suspected early medieval pottery represented a homogenous group of material, characterised by sherds in a coarse, quartz-rich fabric – although Guériteau notes that 110

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France a smaller proportion contained grog (Guériteau 2006b, 2). All of this material was undecorated. She differentiates three colours within this group: white, grey and orange. However, as discussed, colour has been seen to be a less than useful factor for distinguishing E ware, and it may be that all of these represent the same ware affected by different firing or depositional conditions. Some of the pottery was noted to be hard-fired and almost vitrified, which parallels the Insular examples (Guériteau 2006b, 2; Campbell 2007a, Appendix 2, 7). The forms themselves can be directly matched to Labrouche’s illustrated inventory from Place Camille-Jullian as well as the established forms of E ware. Among the complete vessels were: five jugs with pinched rims that can be equated with E4 jugs (see Figure 5.13), two jugs with tubular spouts which can be matched to E4B, one carinated bowl that can be matched to E3, and one carinated beaker matching E2 (Guériteau 2006b, 3-6). Within the remaining fragmentary material, diagnostic elements indicated the presence of a number of jars, potentially equating to E1, as well as lids with pierced holes – E5. The high proportion of jugs present in the assemblage is notable, and presumably relates to their use at the well. Guériteau notes that the seven jugs share a particular characteristic – they are attached under the rim but bonded by one or several finger-marks over the top of the rim (Guériteau 2006b, 6). Again, this matches the jugs from Place Camille-Jullian and the distinctive production technique observed on E ware handles. Although Guériteau suggests dates for these forms, these are partly based on Labrouche’s dates from Place Camille-Jullian, and do not, therefore, represent an external corroboration of his typo-chronology.

Figure 5.13. Jugs from Place Gabriel – equating with E4. Illustration courtesy of Armelle Guériteau

Although the fabrics were divided into four groups based on colour and firing, the general description of the pottery (well-fired, containing sandy inclusions and an absence of decoration) would again suggest that the majority equates to E ware, particularly Fabrics ‘A’ and ‘B’ (oxidised but white/cream coloured and pink/red respectively) and Fabric ‘D’ (over-fired sherds). Within this assemblage vessels equivalent to E2, E3 and E5 were again noted, but in contrast to the Place Gabriel assemblage there were no jugs or pitchers. The highest proportion of vessels was pots, most of which show strong parallels to E1 jars. These are noted to frequently show signs of use during cooking, unlike the cups and bowls. One cup contained a red, powdery deposit that was described as probably ochre, although it would be interesting to compare this to the vessels identified with traces of madder from Britain (Campbell 2007a, 49; Guériteau forthcoming, 5). An unusual form is also described as ‘Pot type 3’, two of which were in the white/cream fabric (see Figure 5.14; Guériteau forthcoming, 6). These vessels have undulating profiles and at least one is illustrated to have a handle scar. Guériteau suggests a possible other example from the city but comments that this seems to be a new form for the period and is unknown in early medieval assemblages from the north of Aquitaine (Guériteau forthcoming, 6). These vessels offer a first parallel for the unique E7 ‘double-carinated’ vessel found at Samson,

Îlot Bonnac A similar range of material was identified by Guériteau within the assemblage recovered from a waste pit excavated by INRAP at Îlot Bonnac, on the outskirts of Bordeaux, in 2005 (Guériteau 2006a; Guériteau forthcoming). As noted, Campbell’s 2011 dataset already included three vessels of E2, one of E3 and one E5 lid, based on Guériteau’s 2006a article. However, the unpublished ceramic report indicates a much greater quantity of material. Of the 679 sherds excavated, three quarters are stated to be of early medieval date. 76 individual vessels were recorded, of which 53 coarseware vessels were identified; these were quantified in a table by MNV – see Table 14.

Table 14. Pottery from Îlot Bonnac, Bordeaux. Adapted from Guériteau forthcoming, 4 Fabric A

Fabric B

Fabric C

Fabric D

Total

E ware equivalent form

‘Pots’

16

6

7

2

31

Mostly E1 jars? 3 vessels possibly matching E7

‘Gobelets’

2

2

3

0

7

E2 beakers?

‘Coupes’

3

3

0

0

6

E3 bowls?

‘Couvercles’

5

2

2

0

9

E5 lids?

E ware fabric?

E ware Fabric?

Vitrified E ware?

111

Links to Late Antiquity ated decoration (including one Rigoir Form 4 dish bearing a medallion with a stamped deer decoration), Guériteau suggested the assemblage dated to the second half of the sixth century. The forms of Atlantic DSP present included Rigoir 4, 6, 16, 18, 25, 29, 30, 56 and possibly 45 (Guériteau 2013, 6-8). Guériteau also notes the presence of one dish of DSP which has the shape and decoration associated with the Atlantic group, but in an oxidised fabric (Guériteau 2013, 13). This demonstrates that any Insular finds of oxidised DSP can still be potentially credited to the ‘Atlantic’ production group. The Rue du Hâ assemblage also contained a late Mediterranean import – the neck of a LRA3 amphora, of the later, two-handled type, which is suggested to be a sixth-century import (Guériteau 2013, 8). This has been identified as a post-Roman import at British sites, including Dinas Emrys (Campbell 1988, 126; Campbell 2007a, 19). Coincidentally, while the potential E ware from the Dinas Emrys was rejected by Campbell, he recently identified the only other piece of imported pottery from the site as a rare example of oxidised DSPA (see Subchapter 3.3.5).

Figure 5.14. Pot with undulating profile from Îlot Bonnac – potentially equating with E7. From Guériteau forthcoming, Plate 2, not to scale

Scilly (Campbell 2007a, 43-5; see Figure 3.16). DSPA was rare at the site, and the only identified form was a Form 29 mortaria. This form has been observed at Insular sites that also have E ware, including Dinas Powys, the Mote of Mark and Dunadd (Campbell 2011).

Puy Paulin

The dating of this site presents an easier association with E ware. Radiocarbon dating of deposit ‘3243’, which produced the majority of the pottery, provided a date bracket of between the mid-sixth and mid-seventh centuries. This fits very well with the dates suggested by Campbell for the ‘floruit’ of Insular E ware. The assemblage from Îlot Bonnac, therefore, seems to represent a good parallel for the dates and types of pottery observed at Campbell’s ‘Continental’ import sites in Britain and Ireland – containing no Mediterranean material, some DSP, but essentially characterised by E ware.

The last of the sites in Bordeaux from which Guériteau has provided information is Puy Paulin, in the centre of the city, where rescue excavations were conducted in 2008. Unfortunately, the development had already destroyed much of the site before the work commenced, and as a result the pottery assemblage represents an arbitrary collection, rather than the sum total of the material present in the deposits. Evidence of occupation and craft-working in the first and early-second century AD was identified, followed by a possible phase of abandonment in the third and fourth centuries (Girond 2012, 3). However, the major feature of the site was a large, high-status house, built sometime between the late-fourth and early-fifth century, within which four mosaic pavements were discovered. This large house was redeveloped through the fifth century, and then seems to have been at least partly destroyed by the beginning of the sixth century. However, there is evidence for continuing occupation in the sixth and seventh century, comprising a building with earth floors built directly on top of the ruins of the ‘domus’ (Girond 2012, 3). No structures were revealed above this, although significant dumping-deposits were identified. The sixth and seventh century deposits were later cut by the foundation of a brick wall of uncertain, but probably medieval, date (Girond 2012, 4).

Rue du Hâ Excavations at Rue du Hâ, Bordeaux [204], in 2007 also produced pottery that might potentially be matched to E ware, although in what would appear to be smaller quantities. These excavations, at the southern outskirts of the city, produced deposits from the first to the sixth century. The Late Antique phase was largely related to a timber building built into the foundations of an Imperial bath complex, and was covered by ‘dark earth’ deposits (Hénique 2007). The pottery studied by Guériteau came from an ill-defined dump deposit from the lowest levels of medieval garden deposits, although the material itself dates to between the fourth and sixth century. She describes the pottery as useful for understanding this transitional period in the city (Guériteau 2013, 1). Of the 1118 sherds studied, 263 vessels were identified. Six fabric groups were isolated, the largest proportion of which was DSPA, while at least one can be potentially linked with E ware. Although vessels equating to E2 and E4 are possibly described, the amount of vessels equating to E ware is not clear from the report. Considering the forms of the DSPA present and any associ-

Guériteau’s report comprised 1291 sherds from levels identified to be of early medieval date, although 546 of these represented residual ‘antique’ pottery (Guériteau 2012, 160). A minimum total of 180 vessels were identified, although it was not possible to match these to the fabric groups identified in the report and therefore to establish the quantity of vessels potentially equating to E ware. The pottery was split into ten fabric categories, based 112

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France Guériteau describes the assemblage as generally typical of the forms and technology of early medieval levels in Gironde and the south of Charente-Maritime and the fabrics and forms of the coarsewares as typical of Bordeaux and its surroundings. She equates the vessels with those from Îlot Bonnac and Jonzac dated to the fifth to seventh centuries, a period which coincides with the date-span of DSP in Bordeaux. These dates are be confirmed by radiocarbon dating (Guériteau 2012, 168). Within this timeframe she observes a general progression from earlier levels characterised by grey coarsewares and DSP being gradually supplanted by rough-white wares. If the latter category is indeed E ware, the dated sequence from the site would appear to equate well with the suggested dates for its importation to Britain and Ireland.

principally on colour, hardness and surface ‘feel’, although these are noted not to represent specific production centres. The first fabric group, described as ‘rough-white’, is perhaps the closest match to E ware, although a second category, ‘rough-orange’, is noted to have the same characteristics. A third group, classified as ‘over-fired’, potentially represents vitrified examples of the same ware. Guériteau notes that these groups were very similar and that distinguishing between them was difficult (Guériteau 2012, 159). As such, it seems likely that they might well comprise the same pottery and potentially all match to E ware, especially given its acknowledged variability in colour. These three fabric groups comprised a total of 347 sherds and accounted for 33% of the pottery assigned an early medieval date. 107 sherds of DSPA were also identified, with Rigoir forms 4 (plate), 5 (dish), 29 (mortaria) and 30 (cup or lid) noted to be present (Guériteau 2012, 165-7).

The complete ceramic quantification from the site, by Laëtitia Pédoussaut, incorporated the earlier Roman pottery, including amphorae. Within this, however, a number of sherds of Late Roman imported amphorae were identified. These were identified as five LRA4 ‘Gazan’ wine amphora of possible fifth or sixth century date, one Lusitanian Almagro 51A of third to fifth century date and one North African Keay 56 amphora, dated to the latefifth or early-sixth century (Pédoussaut 2012, 158). This collection shows parallels to the other Bordeaux amphora groups, of which LRA4 is the most common of the East Mediterranean types.

The majority of the illustrated forms (apart from the DSPA) can again be equated with the established E ware typology, and many of these were made in the three fabric groups that potentially equate to E ware. Jars were the most common shape within the entire assemblage. Guériteau describes the varied rim-shapes within this group, but many can be associated with E1 (approximately 25 of the illustrated vessels) (Guériteau 2012, 165). Dishes are also noted to be quite a common shape; again discounting the DSP, many of these can be matched to Insular E3. Jugs are noted to be rarer, although one example matching to E4 and one matching to E4B are illustrated, both of which are noted to have been in the ‘rough-orange’ fabric. The amount of sherds that could be identified as lids was uncertain, although one pierced example matching E5 was illustrated, again showing the characteristic ‘string-marks’ where it was removed from the wheel (Guériteau 2012, 166; see Figure 5.15). The same feature was observed on a number of bases that could not be matched to specific forms. Unlike the other Bordeaux assemblages, no vessels matching E2 beakers seem to have been recovered, possibly indicating a difference in function; overall, the assemblage seems to have a mixed, domestic function.

Conclusions Guériteau’s various reports clearly demonstrate the presence of pottery at Bordeaux that can be equated to E ware found in Britain and Ireland – based on the illustrations and descriptions of the forms, specific features common to a number of the fabric-groups and the presence of characteristic production techniques. This pottery would seem to characterise the early medieval deposits, particularly in the sixth and seventh century. However, as the fabric groups were not directly correlated with the forms it has not proved possible to state the exact quantity of vessels from these excavations that can be matched to E ware. Guériteau has herself recognised the parallels between the pottery recovered from these rescue excavations and the E ware found at early medieval sites in Britain and Ireland, although this association was not explicit in the reports discussed in this sub-chapter. Furthermore, the pottery from these excavations has not been examined by the present author and this connection has not yet been verified by scientific analysis. Nevertheless, the evidence from these sites clearly compares favourably to Labrouche’s analysis of the coarsewares from Place Camille-Jullian. Research by Paul Tyers and Alan Vince Ian Doyle’s 2009 discussion of the pottery imports in Ireland mentions the parallel examples of E ware noted by Wooding and Thomas, but adds that vessels of a similar fabric are known from ‘post-Roman’ sites in Bordeaux – citing a personal communication from Paul Tyers (Doyle

Figure 5.15. Pierced lid from Puy Paulin showing stringmarks – potentially equating with E5. From Girond 2012, Figure 253, illustration courtesy of Guériteau

113

Links to Late Antiquity These findings are, nevertheless, revealing for the function of E ware. Campbell’s opinion that the ware was primarily a container for the transport of luxury goods, naturally applies to the examples from Insular contexts. The large quantities present at Bordeaux would suggest that the ware had a different – and presumably more prosaic – function in its place of origin. The vessels from Place CamilleJullian show signs of industrial reuse, much like a number of the Insular examples. Campbell was correct that the pottery is unlikely to have had any intrinsic value within ‘early Medieval Atlantic societies’ (Campbell 2007a, 50). Instead, this appearance of this pottery at sites in Britain and Ireland would simply represent the standard pottery available to traders operating from or via Bordeauxmaking use of E1 jars as containers, and potentially, shipping small quantities of E2 beakers, E3 bowls or E4 jugs as paraphernalia for wine-consumption.

2009, 23). It seems that Guériteau’s research in Aquitaine had come to the attention of Tyers and Alan Vince in 2007, as a result of their mutual interest in discovering a source for E ware. Some preliminary analysis was carried out to compare sherds from Bordeaux with British examples of E ware, but this was never published, and their work was cut short by the untimely death of Vince (Tyers pers. comm.). This work was not followed up, although Guériteau has mentioned that a PCR (Projet Collectif de Recherche) has recently been created to investigate early medieval pottery in Aquitaine, and that the question of exports to Britain and Ireland is of interest (Guériteau pers. comm.). 5.4.4 E Ware: a product of Bordeaux? It is extremely likely – based on the discovery of large numbers of vessels that have the same forms as E ware, bear the marks of characteristic and distinctive production techniques, and are believed by a local specialist to be E ware – that pottery classified as E ware in Britain and Ireland is present in large quantities in broadly contemporary deposits at Bordeaux. Although it has not been possible to establish exact quantities, the various reports indicate the presence of at least several hundred comparable vessels (from perhaps several thousand sherds), which represent every one of the forms classified by Thomas and Campbell. The quantities involved appear at least on a scale with – and probably surpass – the total number of E ware found in Britain and Ireland (268 vessels). As mentioned, Campbell indicated that the presence of a large quantity of this pottery in one place would point to a production source (Campbell 2007a, 47). Although no kilns have yet been discovered, it would seem fair to suggest that the source of this ware was Bordeaux, or its immediate environs. As E ware has been described as a homogenous group by Campbell, a single source close to Bordeaux is the simplest explanation for this distribution.

Clearly further research needs to be carried out to confirm the association between Insular finds of E ware and the pottery described by Labrouche and Guériteau, and to establish the full distribution of this ware. The scatter of E ware from the Atlantic seaboard of France has been described, but is it possible that the identification of a source will lead to additional examples? Was this ware common throughout south-west France? As mentioned, Chapelot’s review of ceramic connections between western France and Britain has noted the potential presence of E ware at Bordeaux. He furthermore postulated that the kiln sites might be identified in the Saintonge, providing crucial evidence of ceramic production ante-dating the well-known eleventh and twelfth century industries (Chapelot 2012, 139). However, he also cited a recent study by Brigitte Véquaud on the state of knowledge of early medieval pottery in Poitou-Charentes (Véquaud 2010). This article revealed that although the amount of early medieval material published from this region had increased, there was little that matched well with E ware in either the HautPoitou or Saintonge areas (Chapelot 2012, 139; Véquaud 2010). In contrast, Guériteau has indicated that E ware is present in small quantities in the Charente-Maritime department, while Guitton has suggested its possible presence at Saintes (pers. comms). This research is at an early stage and many questions remain to be answered. Nevertheless, it starts to appear that E ware had a relatively restricted distribution in France – certainly much more so than in Britain and Ireland. Although not specifically an export ware, its distribution beyond Bordeaux (and possibly the Saintes area) is focused outward and toward the Atlantic. Although additional research is necessary to confirm this suggestion, E ware, which has been seen to characterise the Continental phase of imports to Britain and Ireland, might now be better seen to represent Bordeaux’s Atlantic connections. Furthermore, future research on E ware – specifically a consideration of its presence in dated stratigraphic sequences from Bordeaux – might allow the typo-chronological development of this ware to be refined. This would have a major impact on sites in western Britain and particularly Ireland, where the presence of E ware has provided a key chronological marker.

Although it cannot be concluded that every site in Britain or Ireland from which E ware has been found was in direct contact with Bordeaux – individual vessels may well have been redistributed with any associated cargo – it can be argued that the presence of this ware at Insular sites reveals connections that ultimately lead back to Bordeaux. Despite Wooding’s reluctance to link E ware with Bordeaux, based on his rejection of Zimmer’s model of wine trade, this identification fits very well with other categories of material evidence. The British examples of DSP have been linked with the DSPA produced in Bordeaux. Similarly, Campbell noted that Insular finds of E ware were often associated with glass thought to have been produced in Bordeaux or its surroundings (Campbell 1995, 37; 2007a, 138). The recent ceramic evidence indicates that the early attribution of E ware to Bordeaux by Peacock and Thomas was, in fact, correct – and that the finds of E ware in Insular contexts can be directly associated with the Aquitanian port. Whether this can be connected with an active wine trade, with a more varied exchange of commodities, or with cultural or religious contacts operating in the early medieval Atlantic remains to be fully understood. 114

Late Antique and early medieval pottery on the Atlantic Seaboard: the evidence from France 5.4.5 E ware in Spain

supplemented by further finds – from the central-west of France in particular. The information currently available is largely focused on amphorae, and the volume of fineware from Bordeaux is, at present, considerably smaller than from Tintagel. It is possible that the availability of DSPA made the importation of Red Slip finewares less desirable or unnecessary. Outside of Bordeaux, a scatter of amphorae is testament to the fifth and sixth century expansion of East Mediterranean trade. However, the focus on the south-east Mediterranean LRA4 presents one point of deviation from contemporary patterns in western Britain and Ireland.

The recent identification of E ware from the site of Vigo in Galicia, north-west Spain has not only extended the distribution of this ware – representing the first discovery of this pottery to the south of Bordeaux – but has major implications for understanding this ware. E ware cannot be considered to solely reflect a separate exchange system in operation between western France and Britain and Ireland, and which succeeded the phase of Mediterranean importation. This discovery, instead, indicates that these phases of exchange were connected. Pottery from Bordeaux – both E ware and DSPA– were transported southward along the Atlantic seaboard, implying not only a greater complexity of exchange than previously appreciated, but also a degree of continuity in connections between north-west Spain, Bordeaux and Britain between the fifth and seventh century – and even later. These finds are discussed in Subchapter 6.3.2.

This chapter has proposed that large quantities of material which have been recovered from excavations at Bordeaux can be directly equated with E ware. Although this needs to be confirmed by future analysis, it seems that the presence of E ware in Insular contexts can be taken to represent the arrival of commodities exported from the Gironde Estuary. The presence of Mediterranean pottery in western France suggests that at least some of the fifth and sixth century Insular finds could have arrived via an Aquitanian port, although the later-sixth and seventh century evidence is more convincing. By the mid-sixth century the material arriving in western Britain and Ireland – possibly some DSPA but essentially E ware – is likely to have reflected connections originating in Bordeaux.

5.5 Conclusions Contrary to earlier observations, significant parallels for the Mediterranean imports found at early medieval sites in western Britain and Ireland are present in Atlantic regions of France. For now, the evidence largely relates to Bordeaux, but it is likely that this information will be

115

6 Ceramic imports to Atlantic Spain and Portugal in Late Antiquity 6.1 Introduction

6.2 Parallels for the British imports

As argued in Chapter Two, interpretations of the material imported to early medieval western Britain have focused on the exceptional character of the ceramic assemblage. The pottery itself has been considered to have few parallels on the Atlantic Seaboard, and certainly nothing comparable to the total quantity of material from Tintagel. The evidence presented in Chapter Five has revealed a scatter of Mediterranean pottery in the north-west and south-west France, principally East Mediterranean amphorae (LRA1 and especially LRA4). Although many of the LRA4 amphorae might represent imports of the first half of the fifth century, later types were specifically recognised at Bordeaux. Published examples of Mediterranean finewares are fewer, and examples of ARS and LRC have been principally attested from excavations at Bordeaux. As will be considered, much more evidence has been published on late Mediterranean imports from the Atlantic coast of Spain and Portugal. In contrast to Atlantic France, however, the majority of the published Iberian evidence reveals parallels for the fine, Mediterranean tablewares imported to western Britain and Ireland in the fifth and sixth centuries.

6.2.1 Imported Mediterranean finewares in the Atlantic The publication of LRP in 1972 established the sherds of imported fineware discovered in south-west Britain as directly comparable to pottery found in the Mediterranean, and, albeit in much smaller quantities, on the Atlantic seaboard. As mentioned in Chapter Two, Hayes noted the limited presence of ‘Late Roman’ Red Slip wares at selected sites in the Atlantic, including Braga and Conimbriga. His associated map of find-spots of LRC Form 3 presents the British examples as isolated in the northern Atlantic, but unquestionably part of a wider distribution – focused on the East Mediterranean, but including Braga (Bracara Augusta) on the Portuguese coast (see Figure 2.7). By the time of publication of Hayes’ Supplement, the French-Portuguese excavations at the major Roman town of Conimbriga in the Coimbra District of central-western Portugal had been completed, allowing the author to discuss a relatively large group of finewares from the Atlantic region (Delgado et al. 1975; Hayes 1980). Close to one thousand sherds of ARS were found during the excavations, of which 685 were identified as ARS-D (Delgado et al. 1975, 338). However, it was the discovery of 95 sherds of LRC (labelled ‘Phocaean Red Slip’ in Hayes 1980) that demonstrated a clear connection to channels supplying East Mediterranean wares to Britain and Ireland in the later-fifth and sixth century. The Conimbriga report also noted the presence of LRC at Baelo (Baelo Claudia) in southern Spain on the Straits of Gibraltar, as well as Braga, Tróia de Setúbal and Guifões in Portugal – though the latter of these was not included in subsequent catalogues (Delgado et al. 1975, 342). Hayes had previously mentioned the presence of LRC at Tróia, but by 1980 was able to cite a specific publication detailing these finds, which comprised thirteen identified sherds (Maia 1974; Hayes 1980, 525). Hayes considered these ‘recent’ finds from Portugal to provide:

This chapter will consider the growing recognition of late Mediterranean imports in Atlantic Spain and Portugal, and in particular, East Mediterranean LRC. Specifically, it will focus on evidence emerging after the publication of Hayes’ LRP in 1972. As discussed in Chapter Two, although there were occasional mentions of ‘Atlantic parallels’ in British research after this date, they were never fully explored or integrated into interpretations of Atlantic exchange. The character of ceramic assemblages from major Atlantic sites will thereafter be summarised and analysed (minimum vessel counts by type are presented in Appendix K where available). In particular, this chapter will focus on the burgeoning evidence connecting the north-west of Spain to the systems that supplied western Britain in the fifth and sixth century. Most significantly, recent research at Vigo in Galicia has completely altered understandings of the scale and character of pottery distribution in the early medieval Atlantic. The evidence from this site necessitates a complete revision of established interpretations of the Insular imports, and will be discussed in detail. The Iberian evidence will be specifically considered in relation to patterns of exchange in the West Mediterranean outlined by Reynolds (1995; 2010). This analysis will assess whether the imports in Britain and Ireland can be incorporated into broader models of contact between the Mediterranean and the Atlantic.

‘the vital connecting link between the Mediterranean distribution of the ware and its occasional presence, previously unexplained in terms of any normal pattern of distribution, on sites in western Britain’ (Hayes 1980, 525). The quantity of LRC from Conimbriga is certainly roughly equivalent to that from Tintagel (given the quantification issues discussed in Sub-chapter 4.2.3) and the chronology of the group is broadly corresponding, although there are some differences in the forms identified (see Subchapter 6.3.1).

116

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity publication to specifically discuss finds of ‘Phocaean Red Slip’ (LRC) from the Iberian Peninsula. This catalogues a total of ten sites with LRC, all on the southern or eastern coast of Spain, but also mentions the identification of this ware at Conimbriga and Braga (Nieto Prieto 1984, 540, 543). Mayet and Picon’s 1986 publication on LRC has a wider focus, discussing the origins and characteristics of the ware, as well as its entire western distribution. The British and Irish discoveries are recorded (specifically from Dinas Powys, Coygan Camp [87], Tintagel, Cadbury Congresbury and Garranes), although this data is based on Hayes’ summaries, and fails to include the additional Insular examples added by Thomas by 1981 (Mayet and Picon 1986, 130-1). Again, the finds presented are mostly from sites on the Mediterranean coast of France and Spain, though a single vessel is included from Bordeaux. Only two other Atlantic sites with LRC are recorded, and both from the west of Portugal: Conimbriga and Tróia. The overall distribution is stated to confirm the existence of trading channels connecting Britain with the East Mediterranean via Atlantic channels (Mayet and Picon 1986, 132). It should be noted, however, that none of the examples of LRC used for the chemical analysis discussed in the article were from British or other Atlantic sites.

Since the publication of the Conimbriga fineware assemblage – specifically the LRC – the site came to be seen as the only compelling Atlantic parallel for the import sites in Britain and Ireland. As discussed in Chapter Two, Conimbriga was regularly referred to (or alluded to) in subsequent publications on the British finds, and was occasionally suggested to signify the potential for future Atlantic discoveries. Nevertheless, the apparent absence of other sites with significant quantities of fifth or sixth century Mediterranean imports gave the impression that the Insular finds were relatively isolated and unusual, which consequently reinforced the model of direct contact with the East Mediterranean. The material from Conimbriga (together with the very small scatter of LRC from other sites on the Atlantic coast of Portugal) was therefore seen to indicate the routes taken to Britain and even potential stopping-points, but not to represent the primary focus of contact with the East Mediterranean or the final goal of East Mediterranean shipments. However, it is increasingly evident that these interpretations were skewed by a restricted comparable dataset. Although never reflected in the literature on the British imports, since the 1970s there has been a long history of recording late imported finewares (and particularly East Mediterranean LRC) in Spanish and particularly Portuguese publications. Some of the sites recorded in these syntheses have produced only a small amount of material, but this is at least equivalent to the small counts of ARS and LRC from the majority of the British and Irish import sites. In addition, this data reveals crucial details of the distribution of these wares and provides evidence of the broader chronology of the Atlantic import systems.

Significantly, these articles by Nieto Prieto and Mayet and Picon both failed to include the additional Portuguese sites recorded by Maia. As mentioned in Chapter Two, these two publications, together with the Conimbriga fineware report, had provided the comparative data for LRC in Campbell’s thesis (Campbell 1991). Consequently, the British imports continued to be interpreted on the basis of a very limited Atlantic dataset – essentially comprising the assemblage from Conimbriga and the attested presence of LRC at a handful of other sites. Wooding also referenced Nieto Prieto in his published thesis (Wooding 1996a). Campbell’s 2007 monograph added information from Reynolds’ study of ceramic evidence in the western Mediterranean. However, although Reynolds’ list greatly supplemented the record of LRC in eastern and southern Spain, his data from Portugal includes only those four sites included in the 1975 Conimbriga report (Reynolds 1995, Appendix B.2). Despite further finds of LRC being published and documented from sites in Atlantic Spain and Portugal, these were not used as comparative data to assess the British finds.

6.2.2 East Mediterranean finewares (LRC and LRD) By 1978 a second article had been published by M. Garcia Pereira Maia, following the identification of Late Roman C ware at Tróia (Maia 1978). This presented, for the first time, an inventory of sites in Portugal producing this ware. The finds from Braga and Conimbriga, as included by Hayes, are mentioned, but the article also records a small number of unstratified sherds of LRC from six other Portuguese locations: from Cacém in the District of Lisbon (LRC was subsequently identified from two villa sites in Cacém – Cabanas [304]and Casal de Colaride [301]), the sites of Cidade das Rosas [325] and Horta de D. Maria in the Beja district [326], and from Cerro da Vila [338], Loulé Velho [339] and Marim [340] on the south coast (Faro district). Theses are described by form and variant, with all falling between Hayes forms 3B and 3F. However, despite this new evidence of East Mediterranean pottery in Portugal, the sherds were still interpreted in reference to the British examples. For Maia, these new finds not only extended the distribution of this ware – specifically to the south of Portugal – but expressly revealed the ‘chain of commercial connections’ between the Byzantine world and Britain (Maia 1978, 301).

By 1988 the evidence from Portugal had again increased significantly, based principally on research by Manuela Delgado, one of the ceramic specialists responsible for the Conimbriga volume. Delagdo compiled the evidence for LRC from sites in Portugal, adding to the data presented by Maia, although not explicitly making use of the 1978 synthesis. Delgado’s catalogue was based on a review of sherds held in national museums, and on the observation of material emerging from the excavations at Mértola [329] (Beja district) and Cerro da Vila (Vilamoura, Faro district) in southern Portugal (Delgado 1988, 39). A total of 20 sites with LRC were recorded from sites across Portugal, and mapped showing their proximity to the major rivers.

Making full use of Hayes’ new form typology and nomenclature, a 1984 article by Nieto Prieto was the first Spanish 117

Links to Late Antiquity These represented ‘late’ (fifth and sixth century) finds from Roman towns, villa sites and fish-sauce/salting production sites. A quantification of the LRC was provided, giving sherd counts from an overall total of 256. The forms present within each assemblage were also noted, although they were not sub-divided into variants. The vast majority (over 80%) of these were LRC 3, the form which characterises the Insular distribution of LRC. However, certain other forms of LRC were recorded that have not, to date, been identified in Britain or Ireland. These comprised an example of LRC 2 from the villa site of Monte da Cegonha [320], a single example of Form 8 from Mértola as well as examples of LRC 5 from the towns of Braga and Conimbriga and from the villa/fish-processing site of Cerro da Vila on the south coast (Delgado 1988, 45).

sites in southern Turkey (see Sub-chapter 3.3.4), and do not represent imports arriving from Cyprus – previously considered the source of this ware.

The presence of LRC 2 in southern Portugal represents a rare western occurrence of this ware in the late-fourth or first half of the fifth century – and might, therefore, antedate the LRC imported to Britain and Ireland and the main phase of importation of eastern Mediterranean goods into the western Mediterranean and Atlantic.

As stated, Reynolds’ major study on ceramic patterns in the western Mediterranean, published in 1995, records

The list of sites in the Atlantic region with eastern Mediterranean imported fineware was gradually supplemented as additional finds were identified and published. However, the data provided by Delgado and Maia was not rapidly incorporated into wider research on Late Antique pottery and exchange. Miguel Beltrán’s 1990 ceramic guide, Guía de la Cerámica Romana, mapped the Iberian distribution of major Roman and late Roman wares. Although the map of LRC revealed an Atlantic distribution, it included only seven sites in Portugal (Beltrán 1990, 280; see Figure 6.1).

The findings were seen by Delgado to redress an initial imbalance in the distribution of LRC – that apart from Tróia there were no examples of LRC further south than the Tagus River (Delgado 1988, 40). This was rectified by the identification of three sherds of LRC 3 at another site in the Setúbal District, Miróbriga (Chãos Salgados [324]), while two further concentrations of sites with LRC were shown in southern Portugal – one on the Algarve coast (Faro district) and another further inland, surrounding the Lower Guadiana River (Beja district) which runs south to the Algarve (Delgado 1988, Figure 2). These included the sites previously listed by Maia. Importantly, Delgado’s synthesis also records the first examples of LRD (‘Cypriot’ Red Slip Ware) from Portugal, specifically one fragment of LRD 2 from Mértola and two more from Braga (Delgado 1988, 40-3). The latter examples were subsequently identified as LRD 1/2 (Quaresma and Morais 2012, 373). Despite the general rarity of LRD in the West Mediterranean, it was considered by Reynolds to occur in regions where LRC was ‘plentiful’ (Reynolds 1995, 36). He therefore considered the Atlantic seaboard (comprising Conimbriga and Britain), where LRD was considered to be absent, as the ‘exception to this rule’. This observation prompted Campbell to include the absence of LRD in Britain or Ireland as one of the exceptional characteristics of the Insular assemblage (Campbell 2007a, 127). To date, this ware has not been identified in Britain, Ireland or France. However, the discovery of LRD at Portuguese sites with LRC reveals the Atlantic group as a whole to be less distinct from ceramic patterns in the Mediterranean, at least in this respect. This may also hint at the possibility of future discoveries of this fineware in Britain or Ireland. The precise origin of the LRD found in the Atlantic remains to be established; it may be that these Iberian finds can be connected with the known production

Figure 6.1. Beltrán’s western Iberian distribution of LRC. After Beltrán 1990, Figure 128

118

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity only four Atlantic sites with LRC. However, it should be noted that the data in his appendices were stated to represent rim counts from ‘selected sites’, and were not necessarily exhaustive (Reynolds 1995, 3-5, Appendix B.2). The reliance on Reynolds’ data and distribution maps for appraising the scale and characteristics of the British imports can, therefore, be questioned – any resulting interpretations are necessarily founded on an incomplete comparative dataset.

Fulford had previously noted the rarity of LRC at Carthage, commenting that ‘competition aimed at the source of ARS could hardly have been worth-while’ (1984a, 112). Reynolds records the total for Marseille as 9-10 rims (1995, Appendix B.2). Although this figure has risen slightly (see Bonifay et al. 1998, 380-412), Reynolds has recently reaffirmed that the East Mediterranean fineware was a ‘regular, but relatively rare find’ at Marseille in the fifth and sixth century (Reynolds 2010, 117).

The map of LRC distribution presented in Reynolds’ study (see Figure 6.2) revealed a relatively large concentration on the south coast of Andalucía, representing approximately 161 rim-sherds from Belo/Baelo (Baelo Claudia), 158 of which were of Hayes form 3 (1995, Appendix B.2, D.1; Figure 162). A second significant concentration is indicated in south-eastern Spain, specifically representing the deposit from Benalúa, Alicante (69 rims). These south/south-eastern Spanish assemblages are stated to parallel the quantity from Conimbriga (Reynolds 1995, 35). In contrast, LRC is shown to be less common in north-eastern Spain (Catalonia) and in the Balearic Islands – and the north-western Mediterranean in general – and rare at Rome and in western Italy (Reynolds 1995, 35). Further concentrations are presented from a scatter of sites in south-eastern Italy and south-eastern Sicily. The ware is also shown to be absent in Tripolitania, rare at Carthage, but very common in Cyrenaica (Reynolds 1995, 35).

These broad patterns were used to suggest the existence of shipping routes from the East Mediterranean to southern Spain – and thereafter to the Atlantic – via south-eastern Italy and Sicily (Reynolds 1995, 35). The overall distribution of LRC therefore confirmed the likely arrival of East Mediterranean goods at British shores via Atlantic channels – in shipments passing southern Spain and Portugal (Campbell 2007a, 16). Most significantly, the total quantity of material in Britain and Ireland was shown to be broadly comparable to quantities at these larger sites in southern Spain and at Conimbriga, the only significant site on the Portuguese Atlantic. For Reynolds, the distribution confirmed that the supply of LRC to Conimbriga was an ‘offshoot’ of an ‘established’ route to Britain (Reynolds 1995, 135). This conclusion was strengthened by the supposed presence of LRC 10 at Tintagel, which would have post-dated the forms at Conimbriga – although, as noted, this identification was rejected by Campbell.

Figure 6.2. Reynolds’ 1995 map of LRC distribution in the western Mediterranean. From Reynolds 1995, Figure 162

119

Links to Late Antiquity included distribution maps of imported pottery, specifically ARS, LRC and ‘late amphorae’. Significantly, the presence of LRC was recorded from excavations at four sites in the modern region (autonomous community) of Galicia: the villa site at Noville [219] (Mugardos), A Coruña, Lugo [238] and the Roman camp of Cidadela [235] (Naveiro Lopez 1991, 242). Elsewhere he notes that these examples were essentially unpublished; the finds from various excavations at the port-city of A Coruña were later synthesised by López Pérez and are discussed in Sub-chapter 6.3.2, while the LRC from Lugo has recently been described as ‘doubtful’ (Naveiro Lopez 1991, 46; Fernández 2014, 436). This northern Spanish distribution was further extended into the Bay of Biscay by the identification of a rim-sherd of LRC Form 3E, as well as larger quantities of other Mediterranean imports, at Gijón on the coast of Asturias (Fernández Ochoa et al. 1992, 117-22).

By this point, however, the catalogues of LRC in Portugal by Maia and Delgado had demonstrated that the finds from Conimbriga were certainly not isolated, and, crucially, did not represent the only significant non-British assemblage on the Atlantic Seaboard. Instead, this accumulating evidence had started to reveal the widespread distribution of East Mediterranean imported pottery in Portugal, suggesting an under-appreciated complexity to the systems of exchange in the Atlantic. Delgado’s catalogue was superseded in 2001 by the publication of another synthesis, by Élvio Melim de Sousa. This article featured a detailed study of 31 diagnostic sherds of LRC (mostly rim-sherds but also a few base-sherds representing distinguishable vessels) from four sites in Sintra, a municipality within the Lisbon Region, and also provided details of a further six sites in the area surrounding Lisbon (Olisipo) and within the city itself. This concentration, stated to represent a large proportion of the total Portuguese collection, demonstrates the importance of Olisipo and its harbour to the maritime exchange networks operating on the Atlantic in Late Antiquity (Melim de Sousa 2001, 199). The finds from Sintra (from the villa sites of St André de Almoçageme [300], Cabanas and Casal de Colaride, and a single fragment from Armês [299]) were catalogued by form and variant, and again the vast majority were identified as form LRC 3, encompassing variants 3B-F. However, of the 23 distinguishable sherds from St André de Almoçageme, one was identified as a possible LRC 1 and another as LRC 6, while one of the two vessels from Casal de Colaride was identified as LRC 5 (Melim de Sousa 2001, 212-220). Again, these forms of LRC have not been identified in Britain or Ireland.

The inventory of western Iberian sites with East Mediterranean imports was again extended in a recent article by Fabião, who used these finds to frame an argument for economic continuity in the Atlantic in the later-fifth and sixth century (Fabião 2009). Initially focusing on new evidence of economic activity in late and post-Roman Lisbon and around the Tagus estuary (see Sub-chapter 6.3.1), he went on to highlight the expanding Portuguese distribution of LRC as the most studied and most ‘emblematic’ indicator of interaction between the east and West Mediterranean in Late Antiquity (Fabião 2009, 33-4). His distribution map of LRC includes the data from Melim da Sousa’s synthesis, but adds a number of more recently identified/published examples – raising the total number of sites in western Iberia with this ware to 49. Specifically, he notes the presence of LRC at additional locations in the Municipality of Cascais on the coast of the District of Lisbon. He also records more sites in the Faro district/Algarve: ranging from Cacela-a-Velha [331] near the mouth of the Guadiana River to Lagos [335] in the far south-west.

These finds from around Lisbon were supplemented by an updated list of sites throughout Portugal with LRC, which raised the total number of locations to an approximate 37 (Melim de Sousa 2001, 202). This was already considerably larger than the number of sites in Britain and Ireland with LRC (eighteen to date). As in Britain, though, the expanding distribution of this ware appeared to be focused on the coast, or else geared toward locations easily accessible to the Atlantic by river transport (Melim de Sousa 2001, 207).

Sherd or vessel counts are not provided for these Portuguese sites. However, Fabião comments that the quantities of LRC identified from these various locations is unlikely to reflect the true volume of material imported, but instead to reflect the vagaries of research history. In particular, he suggests that the city of Faro (Ossonoba) [341], would be unlikely to have received less of this ware than Conimbriga and that smaller rural sites were unlikely to have received more than coastal redistribution centres, but that this was not currently reflected in the literature (Fabião 2009, 35). The large quantity of LRC later identified at Vigo contradicts any notion that the south of Portugal would automatically be better supplied from the East Mediterranean. However, the relative quantities shown in Appendix H.2 should certainly be considered with caution, and not as a complete or unbiased dataset. The overall western Iberian distribution nevertheless remains focused on coastal locations and is therefore representative of maritime connections and supply – although the presence of LRC at inland locations like Santarém

Beyond Portugal, research had also started to identify the presence of eastern Mediterranean wares, including LRC, in northern Spain. Although discussions of the British imports had occasionally mentioned the presence of equivalent finds in Portugal, no examples had ever been cited from Iberian sites further north than Braga. The inclusion of the north-western Iberian Peninsula into systems of Late Antique commerce was revealed through a major study by Naveiro Lopez, published in 1991. This compiled the various types of evidence (including ceramic data and the remains of metallurgical activity) for pre-Roman, Roman and late/post-Roman exchange networks operating in and incorporating northern Portugal and Galicia – an area roughly equivalent to the late Roman province of Gallaecia. To this end, Naveiro Lopez 120

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity in Britain, it was not only East Mediterranean fineware that was imported to the sites on the Atlantic Seaboard between the fifth and seventh centuries. The presence of African Red Slip ware in the Atlantic also indicates contact with the Mediterranean world in Late Antiquity and can equally be considered as a proxy for the exchange of perishable or recyclable commodities. However, as ARS was imported to Spain and Portugal during the preceding centuries, this ware does not provide such a simple parallel for the British post-Roman imports. Instead, the Atlantic distribution of ARS must be considered within wider patterns of ceramic exchange observed in the late Roman and post-Roman West, which can be seen as a consequence of shifting economic networks, as well as social and political transformations.

[293] and Mértola again indicates the inland diffusion of these wares by riverine transport (Fabião 2009, 35). These larger inland sites would have acted as redistribution centres for smaller, rural sites that continued to be under their influence into the sixth century. Fabião notes that the LRC discovered at the mouth of the Arade River on the Algarve coast (Foz do Arade [334]) represents a rare example of ‘merchandise in transit’ connected to these inland distribution networks (Fabião 2009, 36). This can, perhaps, be paralleled by the LRA1 recently discovered off Cawsand in Plymouth Sound (Sub-chapter 4.3.5). These 49 locations clearly demonstrate that LRC was a more common import in western Iberia than in Britain and Ireland, even if the total quantities from individual sites remain unclear or unpublished. Furthermore, this data seems likely to increase. Fabião admits that his list was not exhaustive, and particularly so for Galicia, where he was dependant on Naveiro Lopez’s inventory. Likewise, the recent finds from Vigo in Galicia have massively increased the recorded number of East Mediterranean fineware imports in the Atlantic (Sub-chapter 6.3.2). Unsurprisingly, given the long history of research on this ware, Adolfo Fernández Fernández – the ceramic specialist responsible for the identification of these finds – included an updated distribution map of LRC in his recently published thesis (Fernández 2014, 440, Figure 225). This added a number of sites on the coast of Galicia (including Bueu) and a smaller number inland. Despite the growing recognition of examples on the Atlantic coast – particularly in the area termed the ‘Rías Baixas’ (which includes Vigo) in south-western Galicia and around the ‘Golfo Ártabro’ in the north-west (including the city of A Coruña) – the finds from Gijón remain isolated on the northern coast of Spain, with no definite finds of LRC heading north or east before Bordeaux. This may reflect a true picture of the distribution of LRC, indicating the importance of sites in north-west Spain as stopping-points or redistributionpoints for ships operating in the Atlantic. Again, however, it might reflect differential research trajectories and it may be that more finds will emerge from France to fill in this ‘gap’ between Galicia and Cornwall.

Broad patterns in the distribution of ARS By the early-second century ARS had supplanted Italian and Gaulish sigillata in western Mediterranean markets, and by the third century it had become the dominant fineware exported throughout the Mediterranean (Hayes 1972, 423). Due to the cost of overland transport, however, its supply was concentrated at coastal locations, and the ware was generally only distributed inland along major rivers or to locations where no local, alternative products were available (Hayes 1972, 414). Although not a common import to Roman Britain (see Sub-chapter 3.3.2) or the Northern provinces in general, it was frequently imported to the Mediterranean coast of Spain from the second century. Nevertheless, the late survival of redslipped sigillata produced in Spain, categorised as Terra Sigillata Hispanica (TSH), is stated to have limited the arrival of Mediterranean finewares in the rest of Hispania until the later-fourth century (Hayes 1972, 415). Hayes comments that finewares continued to be made in Spain on a local scale after this date, but were concentrated on inland areas. However, the study of late Spanish ‘Terra Sigillata Hispánica Tardía’ (TSHT) has developed considerably since the publication of LRP, and Iberian fineware was certainly supplied to sites on the Atlantic coast into the fifth century (Fernández 2014, 299-301). Hayes states that ARS ‘enjoyed a brief vogue’ throughout the Iberian Peninsula from the second half of the fourth century when, significantly, it became more frequently represented in western regions of Hispania (1972, 12, 415). These finds are principally of Hayes’ Stamp Group ‘A’, which he dated to between c. 325 to the mid-fifth century (Hayes 1972, 219; 2008, 70). His 1972 maps of the distribution of ARS with ‘Style A’ stamps and associated forms clearly illustrate the extension of the ware into the Atlantic during this period (Hayes 1972, Map 8, Map 26; see Figure 6.3). By contrast, forms and stamp styles of earlier and later date are shown to have a much more restricted presence in western Iberia.

What has become apparent is that the finds of LRC from Britain can no longer be seen as isolated at the end of a long Atlantic route. In stark contrast to the impression maintained by many publications on the British finds, there are, in fact, numerous Atlantic parallels for the fineware recovered in western Britain. A comprehensive presentation of the information considered in this Sub-chapter – shown proportionally – reveals the British finds to be a relatively small, outlying component of an extensive Atlantic distribution (Appendix H.2). 6.2.3 African Red Slip Ware (ARS) The growing distribution of LRC has been used to characterise the expansion of eastern contact with the western Mediterranean and Atlantic from the mid to later-fifth century. Nevertheless, as seen at the early medieval sites

By the mid-fifth century, the presence of ARS in Spain and Portugal had ‘fallen off’ and was, again, largely limited to the Mediterranean Coast. This reflects a similar picture 121

Links to Late Antiquity Significantly, these observations were extended to sites on the Atlantic seaboard, demonstrating how the fluctuating distribution of ARS in the Atlantic related to broader patterns in the Mediterranean. It should be noted, however, that his 1995 data on ARS in the Atlantic principally comprises the finds from Conimbriga, as well as the British imports. The dated phases described below are taken directly from Reynolds’ chronological groupings (1995, 14-34). ‘Pre-Vandal’ period: the fourth to early-fifth century Reynolds repeats Hayes’ observation that a number of North Tunisian forms became widely distributed across the Mediterranean and into the Atlantic in the fourth and early-fifth century (Reynolds 1995, 14). There had been a resurgence of production in the North from the first half of the fourth century, characterised by the production of new forms bearing stamped decoration and a renewed focus on exportation (Reynolds 12-16). Thereafter, the market in the western Mediterranean was dominated by ARS originating from the various North Tunisian industries – categorised together as ARS-D. Over the same period, the export of Central Tunisian ARS-C to the west declined. Nevertheless, Reynolds notes the presence of fourth to early-fifth century ARS-C in the Atlantic, specifically East-Central Tunisian relief-decorated wares at Conimbriga and Tróia (Reynolds 1995, 15-6). This distribution is also described as ‘exceptional’ for the West. A very few, isolated examples of vessels of this series have been found in Britain, and have been identified as later Roman-period imports (Bird 1977). South-Tunisian ARS was generally rare in the West, although this seems to have increased in the first half of the fifth century (Reynolds 1995, 16).

Figure 6.3. The distribution of ‘stamp style A and associated forms’ shown in LRP. From Hayes 1972, Map 8

in the East Mediterranean where quantities of ARS declined from the early-fifth century and were replaced in the market by LRC and LRD (Hayes 1972, 415-23). In LRP Hayes describes a contemporary, but less dramatic, drop in ARS distribution in the western Mediterranean – and links this to disruption caused by the Vandal conquest of North Africa. More peaceful conditions in the later-fifth century are considered to have allowed a resurgence of African wares in the West, although Hayes comments that this was most apparent following the Byzantine re-conquest of North Africa by Justinian in 533/4 (1972, 423). The establishment of the Byzantine Province of Spania on the south-eastern coast of Spain from 552 is seen to have secured connections with North Africa as well as control of the Straits of Gibraltar (Hayes 1972, 426). However, it seems that by this point in the mid-sixth century Britain and much of the Atlantic was largely cut off from Mediterranean supply.

Early to mid-Vandal period: c. AD 430-475

As with the LRC, Hayes’ LRP includes a more detailed catalogue of Spanish sites with African Red Slip, including a number of Atlantic sites, although the forms are not specified. The presence of ARS at sites such as Conimbriga and Braga is noted, but the majority of these Atlantic examples would pre-date the arrival of this ware at sites in early medieval Britain and Ireland.

As mentioned, Hayes had described a drop in the supply of ARS as a direct consequence of the Vandal conquest of North Africa. However, Fulford’s analysis indicated that this decline in exportation had commenced before c. 425 – therefore in advance of the Vandal takeover (Fulford 1984a, 113; 1989, 3). Bonifay has recently reaffirmed that the decline in ARS exportation is perceptible from the early part of the fifth century – and cannot, consequently, be explained by the rupture of the annona system (2011, 21).

ARS in the West Mediterranean and Atlantic Variations in the supply of ARS were analysed in Reynolds’ 1995 study of the western Mediterranean and in his recent monograph on the Iberian Peninsula (Reynolds 2010). These developed the patterns outlined in LRP, and drew on Fulford’s observations of temporal shifts in the distribution of this fineware – based on the material recovered at Carthage (Fulford 1984a, 112-4; 1984b). Reynolds’ earlier publication specifically outlined patterns of supply to the West in the period before the Vandal invasion of North Africa in AD 429, during the Vandal occupation, and into the late Vandal and Byzantine periods – following the Byzantine re-conquest of south-eastern Spain.

Reynolds outlined a slightly revised model of supply in the mid-fifth century, arguing that some export did continue in the early to mid-Vandal period but only to distinct, regional markets (Reynolds 1995, 17). This period was characterised by the uneven regional distribution of ARS – indicating the complexity of the underlying shipping routes that supplied amphorae, grain and other commodities throughout the western Mediterranean and beyond (Reynolds 1995, 25). ARS exports were re-focused toward specific regions such as southern France (Marseille), south-eastern Spain (Alicante), Sicily and the Balearics, 122

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity production only continued in North Tunisia (Reynolds 1995, 13, 30). ARS-D continued to be exported into the early-seventh century at certain western sites (including Cartagena and Marseille), but by the mid-sixth century supply to western Iberia and the Atlantic had contracted. Reynolds specifies a number of late forms of ARS-D (including 91D, 99C, 105) that were introduced after the Byzantine re-conquest of North Africa (Reynolds 1995, 31). None of these ‘Byzantine’ forms have been identified in Britain or Ireland (apart from the possible 99C from Pevensey Castle) and, as discussed, none of the Insular imports can be securely dated beyond c. 550. These very late forms have been considered to be typically rare in the Atlantic, although examples were later identified from the excavations at Place Camille-Jullian in Bordeaux, and at Vigo.

but by contrast, areas further west – including the Atlantic – became marginalised, as part of a retraction of distribution to the Mediterranean basin (Reynolds 1995, 18). A drop in supply of ARS was observed at Conimbriga in the mid-fifth century, which was reflected by reduced quantities at Baelo on the south coast of Spain. However, the rarity of ARS in the Atlantic during this period was not considered to reflect a lack of demand for finewares – as demonstrated by the appearance of significant quantities of LRC and locally-produced imitation wares at sites such as Conimbriga from the second half of the fifth century (Reynolds 1995, 18, 117). Instead, a break in the supply of ARS left room in certain western markets for East Mediterranean wares at a time when the reach of these industries was expanding. Late Vandal period c. AD 475-535

Examples of ARS in Atlantic Spain and Portugal The later-fifth and early-sixth century witnessed a new expansion of exports of North Tunisian ARS (ARS-D) to the West Mediterranean and into the Atlantic, reaching an apogee in the period AD 500-530 (Reynolds 2010, 101). This increase in volume and distribution was considered by Fulford to reflect a period of ‘revived prosperity and increased overseas contact’ (Fulford 1984b, 260). Notably, this ‘revival’ commenced in advance of the Justinianic reconquests (Fulford 1984a, 114).

This wider consideration of the shifting supply of ARS to the West reveals how successive ‘waves’ of imports can be identified in the Atlantic, which can be used to mark phases of integration and marginalisation into/from Mediterranean systems of exchange. The wide distribution of ARS in the fourth century and early-fifth century was followed by a reduction, but not cessation, of imports towards the middle decades of the fifth century. However, Britain, at the far north of the Atlantic routes, appears to have been completely ‘cut-off’ at this point. From the later-fifth century a greater volume of material was again reaching the Atlantic, but by this date the supply of ARS was challenged, and even surpassed, by East Mediterranean alternatives. These patterns provide a framework to discuss specific examples of this ware at sites on the Atlantic Seaboard.

Over this period ARS production in El Mahrine (ARS-D1) seems to have been generally supplanted by that of Oudhna (ARS-D2) (Bonifay 2003, 123). The ‘E’ style of stampdecoration also appeared – of which style ‘Eii’ represents the only identified stamp amongst the British examples of ARS. In addition, this period witnessed the introduction of a new range of North Tunisian ARS forms – including Forms 99 and 104 – that would become characteristic of late-fifth to sixth century distribution throughout the West (Reynolds 1995, 17). It is these forms that typify the ARS recovered at early medieval sites in western Britain and Ireland.

As with the LRC, examples of ARS have been increasingly catalogued in Spanish and Portuguese publications, revealing an increasing awareness of its distribution from the fourth century. A bibliographic study by Jarrega Dominguez – published in 1991 – catalogued published examples of North African and East Mediterranean finewares imported to Spain in the fourth and early-fifth centuries, the sixth century and (in smaller numbers) in the early-seventh century (Jarrega Dominguez 1991, 5). The later examples in the catalogue demonstrate the supply of this ware to sites in both the Visigothic Kingdom and the Byzantine province of Spania, and are presumed to indicate the supply of other commodities such as olive-oil. Jarrega Dominguez describes the general difficulties in amalgamating this data, stating that the published identifications were frequently questionable and poorly illustrated (1991, 89). He also comments that stratigraphic information was rarely available for these finds, which limited their chronological significance (Jarrega Dominguez 1991, 89-90). Although the majority of the catalogued examples included are from Mediterranean sites, the synthesis nevertheless indicates the presence of ARS and, pertinently, later forms of ARS-D at a number of sites in the north/ north-west of Spain.

By the later-fifth century western supplies of ARS were again dominated by North Tunisian products; by c. AD 500 the South Tunisian industries had disappeared, and Central Tunisian production (ARS-C) had diminished (Reynolds 1995, 30). However, Reynolds observed that the supply of ARS in the late Vandal period did not simply return to the pre-conquest pattern, but remained regionally distinct. Significantly, he describes an increased supply to ‘coastal sites where Vandal supply had been relatively scarce’ (Reynolds 1995, 29). This includes south-western Britain, demonstrating the re-integration of the northern Atlantic into systems of supply by the late-fifth/early-sixth century. The Byzantine re-conquest period AD 533 to the seventh century By the early-sixth century the central and southern ARS production industries were in decline, and by c. 550 123

Links to Late Antiquity Only one example of ARS-C was recorded from the Atlantic region – a possible example of Hayes 73 from Gijón, although this attribution was subsequently questioned (Jarrega Dominguez 1991, 11; Fernández Ochoa et al. 1992, 110). Nevertheless, additional examples of ARS-D were recorded from this site, including Hayes 91A/B and 91C (Jarrega Dominguez 1991, 67, 69). Jarrega Dominguez also includes an uncertain example of the ‘E’ series of ARS from Gijón, but this is not mentioned in subsequent reports on the assemblage, and has therefore been disregarded (see Fernández Ochoa et al. 1992, 110, note 5). Further west, a fourth or earlier-fifth century example of ARS-D was recorded from Galicia, specifically an example of ARS 61A from the monastery of Sobrado dos Monxes [236] (Jarrega Dominguez 1991, 23). The presence of ARS in the north of Spain was supplemented by a possible example of ARS 91A/B from Iruña (Iruña-Veleia) [241] in the Basque Country (País Vasco), although it is not clear whether this example can be tied to Atlantic supply or arrived from the Mediterranean (Jarrega Dominguez 1991, 67, 92-3).

The discovery of a dish of ARS 104A/B stamped with Chi-Rho symbols during excavations of the town-wall prompted Alonso Sánchez and Fernández Ochoa (1988) to consider the wider Iberian distribution of ARS and the research on this ware in the Peninsula. They include a distribution map of ARS-D in Spain and Portugal, which shows finds from approximately 25 sites on the Atlantic Seaboard. However, as the vessel forms are not given, these examples cannot be directly paralleled with the laterfifth and sixth century ARS from Britain. The wide scatter includes finds in southern Spain, Portugal and a distinct concentration in Galicia – confirming the supply of this ware via Atlantic channels – but many of these are likely to relate to the earlier, fourth/early-fifth century distribution of North African fineware. More specific information is contained in the second article (Fernández Ochoa et al. 1992), which considered the Atlantic distribution of laterfifth and sixth century forms of ARS – specifically Forms 91C, 99 and 104. The presence of these forms, in conjunction with LRC, DSPA and East Mediterranean amphorae, provides a more direct comparison for the British imports.

Although Jarrega Dominguez records LRC in Mediterranean Spain and discusses the presence of LRC at sites in Portugal as connected to Atlantic networks, his catalogue does not record any examples of this ware from Atlantic sites in Spain. The finds of ARS from Galicia and Asturias are, nevertheless, interpreted as indicators of the maritime routes that supplied fineware and amphorae of fifth and sixth century date to Britain (Jarrega Dominguez 1991, 93). As such, it is surprising that his distribution maps of ARS appear to suggest that these Spanish examples reached the Atlantic coast via overland routes (Jarrega Dominguez 1991, 109, Figure 4; see Figure 6.4). The absence of Portuguese data from the catalogue seems to have masked the obvious circulation of these finewares via Atlantic channels.

The publication also summarised the long history of the identification of this ware in Spain, and updated the plot of the Iberian distribution of ARS-D. Its coastal distribution remained apparent, with the highest number of sites shown on the Mediterranean coasts of Andalucía, Murcia, Valencia and, particularly Catalonia. Again, inland penetration was concentrated on major rivers, particularly the Guadalquivir in southern Spain and the Ebro in the north-west. Nevertheless, a similar, if reduced distribution was shown in Atlantic regions, with finds shown at coastal sites and distributed inland via the Tagus and Duero Rivers (Fernández Ochoa et al. 1992, 116). This coastal and riverine distribution was also noted by Naveiro Lopez, who recorded the presence of ARS-D at coastal sites in Gallaecia such as A Coruña, but also along the Ulla, Miño/ Minho, and, above all, the Duero (Naveiro Lopez 1991, 45). Imports of ARS are stated to have been identified during excavations of major urban centres, such as Braga, but also from fortified settlements (castros) that produced evidence of ‘late’ occupation and from associated rural burial-sites. This indicates that, as in Britain, these wares became integrated into systems of local redistribution, potentially connected to the formation of regional political networks. However, the majority of the ARS in the northwest region is considered to pre-date the middle of the fifth century, and thus the phase of ‘post-Roman’ importation to Britain (Naveiro Lopez 1991, 45).

By contrast, Beltrán’s 1990 distribution map of ARS had recorded the presence of ARS-C at various sites in Portugal, including Tróia, ARS-D at Santarem, and forms of both regional series at Conimbriga and Mirobriga (Beltrán 1990, Figure 60). These assemblages are discussed in more detail in Sub-chapter 6.3. The finds of ARS-D from Gijón were assessed in more detail in two articles, published around this same period.

The data on ARS from sites in Portugal and Atlantic Spain recorded in Appendix K and shown in Appendix Maps H.1-3, have been primarily drawn from the publications mentioned, but supplemented using additional published syntheses, data-tables and discussions (López Pérez 2004; Prieto 2005 172-3; Viegas 2007; Quaresma 2012, 424, Appendix 7, Table 1; López Pérez and César Vila 2010). Where possible, identifications have been verified or amended with reference to original excavation or ceramic reports. However, a considerable amount of this material

Figure 6.4. The Iberian distribution of ARS-D in the fourth and first half of the fifth century (left) and second half of fifth and sixth century (right). After Jarrega Dominguez 1991, Figure 4

124

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity these products in their own right, or that – as seems more likely – these wares were secondary cargoes shipped with Tunisian grain (Reynolds 2010, 23). He later reiterates that the relatively large quantities of fourth century ARS at Braga and Conimbriga – and the associated scarcity of amphorae – either reflects the direct marketing of ARS to Lusitania, or else indicates that ‘ships plying the Atlantic route carried cargoes of wheat or other non-amphoraeborne goods’ (Reynolds 2010, 47, see also 105, 111).

remains unpublished, and the overall picture of relative quantities of imports arriving over time remains unclear. Specific references to the ARS imports at individual Atlantic sites are considered more closely in Sub-chapter 6.3. However, in many cases these examples have been identified solely as ‘ARS-C’ or ‘ARS-D’, and cannot be closely dated. As such, identified examples of later-fifth and sixth century forms have been mapped separately in Appendix Map H.1.2 to more effectively illustrate parallels for the British post-Roman imports. The rare Atlantic identification of later-sixth/early seventh century forms have been mapped in Appendix H.1.3.

Similarly, Bonifay has questioned whether too much weight has been placed on ARS as being directly connected to the supply of North African olive-oil. Instead, he offers a persuasive argument that ARS was likely to have been exported as a secondary cargo with shipments of grain (Bonifay 2003, 128; 2007, 103). Specifically, the re-emergence of North Tunisian ARS in the fourth-century is connected to a growing reliance in the north-west Mediterranean for supplies of North African grain – at a time when Egyptian grain was increasingly diverted from Rome towards Constantinople (Bonifay 2003, 119-20). If a low ratio of amphorae to fineware can be explained by the prevalence of grain supplies, then the relative infrequency of Mediterranean amphorae at sites in Atlantic Spain and Portugal is less perplexing. This observation was also made by Adolfo Fernández Fernández (2014, 463).

As with the LRC, significant quantities of ARS have been recently identified at Vigo. The characteristics of this group and its chronological significance are assessed in Sub-chapter 6.3.2. Although the presence of fourth and early-fifth century ARS in the Atlantic has long been acknowledged, this increasing information has established a degree of continuity of importation in the fifth century, confirming that the ware was a major component of Atlantic exchange in the late-fifth and (particularly) sixthcentury, and was even imported beyond this period. 6.2.4 Imported Mediterranean amphorae In contrast to the LRC and ARS, there is much less information on imported amphorae of fifth century or later date in Atlantic regions of Spain and Portugal. This pattern is the opposite of the French data discussed in Chapter Five, where more information is currently available on amphorae (see Appendix Maps H.5-8.8). This relative lack of Iberian data is likely to reflect differences in the focus of research between these various regions, and in particular with Britain, where interest in these wares has developed steadily since the 1930s. Fabião has suggested that the small number of East Mediterranean amphorae identified from the west of Iberia results from the unfamiliarity of specialists with these late imports, but that this situation was improving – particularly in relation to Lisbon and the Algarve (Fabião 2009, 37). However, despite increasing interest in these wares, he conceded that the current picture of their distribution was far from complete. As such, the list of examples from Spanish and Portuguese sites on the Atlantic Seaboard presented in Appendix K cannot be considered to be exhaustive, or to fully represent patterns of supply.

The increased supply of ARS to the Atlantic from the latefifth to sixth century would suggest that North African grain exports were a significant feature of Atlantic trade from the late Vandal/early re-conquest period. Reynolds concluded that between the fourth and sixth century sites in Portugal and north-west Spain were able to ‘tap’ supplies of Tunisian and eastern Mediterranean finewares carried by ships heading to Britain, or otherwise ‘on the annona route to Bordeaux’ (2010, 57). However, it may be that the growing quantities of fineware identified in western Iberia indicate that this region was the actual focus of shipment – specifically for consignments of grain – rather than Britain. Certainly, some disparities can be observed between the British assemblage and ceramic patterns in western Iberia, suggesting differences in demand and supply. On the whole, amphorae predominate within the British assemblage, although a few locations such as Cadbury Congresbury and Dinas Powys do have a disproportionately high number of fineware vessels. Reynolds had previously suggested that the contrasting absence of East Mediterranean amphorae at Conimbriga might indicate that the regions ‘commanded different markets’, which were supplied independently (Reynolds 1995, 135). However, this difference could also be explained by the British imports having been redistributed from an Atlantic port. It may be that the relatively large number of amphorae recorded at sites such as Tintagel and Bantham reveal ‘amphora-borne’ commodities – and principally wine – to be more desirable or valuable commodities at the northern edges of the Atlantic networks. Therefore, it could be argued that amphorae recovered from sites in western Britain were either shipped separately to the commodities reaching western Iberia, or else were trans-shipped toward

On the other hand, there are indications that regional differences in the proportion of fineware to amphorae do actually reflect the differential supply of commodities, and it may be that Mediterranean finewares were more frequently imported than amphorae to many of the Atlantic sites. Reynolds has compared the low proportion of Tunisian amphorae to ARS at Atlantic sites such as Conimbriga and Braga to an earlier pattern of North African pottery in Iberia – whereby cooking wares and fineware were common imports in the first and second century, but amphorae were generally scarce (Reynolds 2010, 23). He suggests that this might reflect the marketing of 125

Links to Late Antiquity highly receptive, Insular markets. Nevertheless, the possibility of grain imports reaching post-Roman Britain cannot be completely discounted – potentially corroborating aspects of the story in the Life of John the Almsgiver (see Sub-chapter 2.2.2).

2010, 97). The distribution of LRC and eastern amphorae at Atlantic sites in this period can be compared with a contemporary reduction in imports of ARS (Reynolds 2010, 97). Imports of LRC became a substitute for the less regular supply of African fineware.

‘Late’ imported amphorae in the Atlantic

It was not until the late-fifth century that exports from North Africa (Tunisian Amphorae and ARS) experienced a resurgence in the West Mediterranean – demonstrating a renewed impetus within the Vandal state to develop the production and exportation of commodities such as oil, wine or grain (Reynolds 2010, 86). Reynolds describes the late-fifth and early-sixth centuries as witnessing a ‘general boom in trade’ from Tunisia and the East Mediterranean; this was driven by the needs of the ‘Barbarian’ successor states to feed the populations of cities now under their control, and principally based on grain-supplies (2010, 100). As such, this economic recovery pre-dated the Byzantine re-conquests of North Africa and south-east Spain. The later Vandal period also witnessed the introduction of a major new Tunisian amphora, Keay 62, which came to dominate markets in the West. The distribution of this amphora is seen to demonstrate the extensive exportation of Vandal surpluses – particularly oil – to the West during this period, although Reynolds mentions ongoing debate as to whether wine was a more likely content (2010, 101, 284, note 331).

Despite the general paucity of amphora data from western Iberia, parallels for the fifth and sixth century amphorae found in Britain have been recorded from a number of western Spanish and Portuguese sites, and this data seems likely to increase. As with the LRC, the available evidence is particularly revealing for the East Mediterranean imports, seen to be particularly characteristic of fifth and sixth century Atlantic exchange. Again, a broad chronological consideration demonstrates that the appearance of imported amphorae in Britain can be directly connected to wider patterns of ceramic distribution in the West Mediterranean – and the extension of these into the Atlantic. It should be noted, however, that fluctuations in pottery distribution tend to lag behind the more direct effects of political upheaval. Furthermore, variations in primary export patterns are not necessarily useful for dating the ultimate deposition of an individual pot, especially at sites far from the point of manufacture.

Examples of this amphora have also been identified at a number of Atlantic sites, including Bordeaux and Vigo, revealing some re-integration of Tunisian amphora-borne commodities into Atlantic exchange systems. Nevertheless, African ceramic supplies to the Atlantic seem to be more generally restricted to ARS by this date, mirroring the pattern of exportation to the East Mediterranean (Reynolds 2010, 101).

The late-fourth/early-fifth century witnessed the appearance of certain types of eastern amphorae, principally LRA1 and LRA4, at sites in the West Mediterranean, reflecting a period of generally high economic activity (Reynolds 2010, 85-86).These East Mediterranean amphorae continued to increase their share of western markets in the second quarter of the fifth century, which, as described by Reynolds, can be associated with a contemporary reduction in Tunisian amphorae at sites such as Rome and Carthage (2010, 86). Although there are difficulties in directly connecting these shifts with the Vandal conquest it seems likely that a level of disruption from c. 430 allowed for a general expansion of trade from the East (Reynolds 2010, 86).

This Late Antique Atlantic pattern– characterised by the renewed presence of ARS but relatively few Tunisian amphorae – can be seen to contrast with the West Mediterranean, where Tunisian amphorae continued to circulate in large quantities beyond the mid-fifth century. Reynolds notes that eastern Hispania, in particular, was a major importer of Tunisian amphorae in the first half of the sixth century (Reynolds 2010, 103). Nevertheless, it is important to note that this distinction applies to sites all along the Atlantic Seaboard, and not solely to those in western Britain. As discussed in Chapter Two, the higher proportion of East Mediterranean to North African amphorae in the British and Irish assemblage had argued against their redistribution from a site in the West Mediterranean – where Tunisian supplies predominated. This consequently strengthened the model of direct contact between Britain and the East Mediterranean. However, a wider consideration of shifting patterns in supply reveals a focus on East Mediterranean wares in the period c. 475-550 to be typical for sites across the Atlantic. Rather than being exceptional, the British pattern equates with much of the Atlantic.

Reynolds presents a model whereby eastern merchants, familiar with these markets from the early-fifth century, took advantage of new opportunities later in the century to exploit trade in the West and, significantly, in the Atlantic (Reynolds 2010, 92, 98). The distribution of East Mediterranean amphorae in the West had increased by the mid-fifth century and expanded even further towards the end of the fifth century. By this point these wares were reaching Atlantic markets, including sites in Britain and western France. He specifically observes that the northwest of the Iberian Peninsula had become marginalised from Mediterranean supply, but by the later-fifth century was re-integrated into this ‘new dynamic’ of trade (Reynolds 2010, 47). By the middle of the fifth century East Mediterranean fineware (specifically LRC) was also appearing at sites in the western Mediterranean and the Atlantic, although the majority of examples date to the later-fifth century or first half of the sixth (Reynolds

Both Tunisian and East Mediterranean amphorae continued to be exported to the West Mediterranean in the 126

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity subsequently plotted in a distribution map in the aforementioned article by Fernández Ochoa et al. (1992, 117, Figure 6). This added nine fragments of LRA1 from the excavations at the ‘Termas de Campo Valdés’ bath-complex at Gijón [215], but also noted a possible fragment of LRA4 from earlier investigations at the town (Fernández Ochoa et al. 1992, 136-8). The article also mentions the presence of Mediterranean amphorae further west along the Cantabrian Coast at Cabo Higuer/Cabo de Higuer [220], which demonstrates the shipment of these wares along the north coast of Spain, and consequently extends the known reach of these Atlantic routes toward Bordeaux (see Sub-chapter 6.3.3).

later-sixth and into the seventh century, although for most of the Atlantic sites, amphora imports seem to have declined from the mid-sixth century. This can be compared with contemporary shifts in fineware distribution. Reynolds notes that LRC exports to the West and the Atlantic dropped from the second half of the sixth century (2010, 123). However, East Mediterranean amphorae continued to circulate in the West in the late-sixth and seventh century, even rising in quantity at sites such as Marseille, Naples and Carthage (Reynolds 2010, 124). Marseille is also noted to have continued to receive ARS and Tunisian amphorae into the seventh century, with the proportions of North African vessels continuing to dominate the amphora imports.

Fabião’s updated distribution map (2009, 37, Figure 5) includes ten sites in Portugal and north-west Spain with East Mediterranean amphorae, including two sites in the Faro district of Portugal and two in Lisbon. Specific details of these widely scattered sites are discussed below, particularly for Braga where a number of vessels have been identified. Nevertheless, despite Fabião’s optimism for future identifications, this clearly comprises a more restricted dataset than for the LRC. Once again, many more examples have been identified at Vigo, where a range of imported amphorae has been recorded. Although the full data on the amphorae from the various excavations has not been published, broad patterns in the assemblage are considered below, which shows the group to be highly significant for understanding the British amphorae.

Although there have been suggestions that amphorae continued to be imported to Britain in the second half of the sixth century, there is no direct evidence of this from the finewares – given that the LRC 10 from Tintagel is discounted. This again equates with the broad pattern in the Atlantic. By the later-sixth century much of the Atlantic seems to have become marginalised from the Mediterranean and systems of exchange that continued to operate in the West into the seventh century. Although the presence of late finewares at Bordeaux and Vigo indicate some reduced connection to the Mediterranean, the volume of imports to the Atlantic seems to have markedly declined. Britain, at the far north of the Atlantic routes, was once again cut off from Mediterranean supplies.

6.3 Regional case studies: sites with imported Mediterranean pottery in Atlantic Spain and Portugal

Sites with late imported amphorae Keay’s 1984 study of Late Roman amphorae in the West included only a very few examples from the Atlantic, although he acknowledged that his catalogue was not complete and, therefore, that the amphorae distributions presented by him could not be easily linked to actual patterns of exchange (Keay 1984, 631). In addition to some of the finds from Britain, he records the presence of LRA1 at Conimbriga and LRA4 at Saint-Seurin in Bordeaux, as well as North African amphorae of later-fifth/sixth century date (specifically Keay 62) at Tróia (Keay 1984, 440-442, Figs 192-4, 631-660).

Broad patterns in the distribution of ceramic imports recorded from the Atlantic have been discussed in the preceding sub-chapters, but it is also necessary to consider the specific pattern of imports at individual sites in Spain and Portugal. Variations in the scale, character and chronology of these assemblages reveal points of similarity and difference with the British finds, indicating a hitherto unappreciated complexity to patterns of ceramic supply in the Atlantic. This sub-chapter presents selected case studies of major sites and regional concentrations, or sites of particular significance, but cannot claim to present an exhaustive account of all imported pottery on the Iberian Peninsula. Where provided, minimum vessel counts have been provided in Appendix K, although much of this data is unpublished beyond its inclusion in synthetic articles, catalogues and appendices. It is also likely that this data will continue to increase.

As mentioned, the presence of late imported amphorae in north-west Spain was documented by Naveiro Lopez, who mentioned small fragments of East Mediterranean amphorae of fifth or sixth century date at A Lanzada [255] on the coast of Galicia, and further inland at Cidadela and Taboada [245] (Naveiro Lopez 1991, 67). Although he does not indicate types for these vessels, an accompanying illustration suggests these may be LRA4. Amphorae identified as type ‘Kapitän 1’ are also noted from A Lanzada and A Coruña, although the illustration of the former example bears a strong resemblance the upper part of LRA1 (Naveiro Lopez 1991, 67-8, Figure 15.24). The author also suggests the possibility of late North African amphorae at these two sites, although as the types are not specified the date of these is uncertain (Naveiro Lopez 1991, 70). The East Mediterranean examples from the North West were

6.3.1 Southern and central Portugal As in France, researchers in Spain and Portugal have become increasingly interested in questions of contact and connection in the Late Antique and early medieval Atlantic. Specifically, a growing awareness of the continuation (or re-configuration) of ‘Roman’ economic systems from the late-fourth century into the fifth and sixth century has spurred efforts to identify Mediterranean imports dating to 127

Links to Late Antiquity this period. Notably, research by Catarina Viegas has revealed new understandings of the extent of ‘late’ ceramic imports to the south of Portugal. Additional considerations of evidence for Late Antique economic continuity in Lusitania have been presented by Carlos Fabião (2009) and José Carlos Quaresma (2012, Chapter 4, Appendices 1,2 and 7), and much of the ensuing discussion – and data presented in Appendices H and K – is founded on, and indebted to, these works.

(Viegas 2007, 72). It should be noted that there are difficulties with using fineware imports to date the occupation of sites, especially as these vessels could continue to be used long after supplies had dwindled. Nevertheless, this typical model of abandonment and/or declining economic activity certainly seems to correspond with a widely recognised period of Atlantic marginalisation – whereby the middle decades of the fifth century witnessed a reduction in Tunisian exports to the Atlantic.

Southern Lusitania

In contrast, a number of sites in southern Portugal have produced evidence of continuity of occupation or production beyond this date, demonstrating the surge of East Mediterranean imports into the West in the second half of the fifth century, and the subsequent revival of Tunisian exportation. The presence of East Mediterranean fineware and amphorae and later forms of ARS at these locations demonstrate their incorporation into a phase of renewed economic activity in the Atlantic.

Viegas’ 2007 article characterised patterns of late imported ceramics in the south of the area comprising the Roman Province of Lusitania (particularly sites in the districts of Faro, Beja, Évora and Setúbal), before providing a more detailed discussion of unpublished finds from excavations at two large Roman towns on the Algarve coast – Faro (Ossonoba) [341] and Torre de Ares (Balsa) [337] (Viegas 2007). The specific forms and variants of pottery from these two sites – largely identified from a reappraisal of material excavated in the 1970s – were subsequently provided in her thesis (Viegas 2009).

Within the southernmost district of Faro, Viegas described the presence of late imports at the Roman towns of Faro (Ossonoba) and Torre de Ares, Tavira (Balsa), providing a quantification of classes by ‘MNI’ (minimum number of individuals – equivalent to MNV). The fineware imported to both sites largely follows the ‘typical’ pattern, with most of the ARS-D dating to up to the early-fifth century. However, recent excavations at Faro have also produced evidence of later forms – including ARS 91C, 103 and 104 and sherds with ‘Style E’ stamps – revealing the continuing arrival of imports until the mid-sixth century (Viegas 2007, 74; 2009, Appended Database). This continuity is also demonstrated by a group of LRC, representing a minimum of seven vessels including Form 3 variants F and H. Sherds of DSP were also recovered at Faro, but were identified to the Provençal production, and cannot be seen to represent maritime supply from the Atlantic. Definite finds of DSP of the Atlantic group have, however, been identified on the Atlantic coast of Iberia at Vigo.

Viegas summarises certain general complications with the data currently from Portugal, which reflect typical difficulties encountered across the Atlantic region. The Portuguese imports mostly derive from large villa sites, as well as – more rarely – towns or industrial sites such as fish-sauce production sites (Viegas 2007, 71). The data itself largely comprises finewares, with relatively little information available on amphorae and coarsewares. A general lack of stratigraphic information is also highlighted – specifically that finds rarely derive from secure contexts (Viegas 2007, 71). Finally, a lack of quantified data is said to limit the extent to which comparisons can be made between these sites. This consequently hampers comparisons with the wider Atlantic, including the Insular material. Incorporating information from earlier syntheses (including Blázquez Martínez 2002) Viegas outlines the general pattern of fineware-supply to this region, which can be correlated to the broad patterns described by Reynolds (1995; 2010). Overall, the majority of the villa sites in southern Portugal are seen to be abandoned by the mid-fifth century, as evidenced by a reduction or disappearance of imports of ARS-D. Examples given include the villa sites of São Cucufate [319] and Represas [323], both in the territory of the Roman town of Pax Iulia (Beja) (Viegas 2007, 72). The rarity of LRC, contrasting with the previously high levels of ARS-D, is also considered to corroborate this model of abandonment – although the presence of LRC 3 at both of these sites must indicate that some connection with the Mediterranean continued, at least, beyond c. 450. The villa at Tourega [314] shows a similar pattern – the latest ARS comprising ARS-D forms 58, 59, 61 and 67 – although here, LRC was not identified (Pinto et al. 2004, 19; Viegas 2007, 72). A comparable pattern is encountered at other types of sites, including the fish-sauce production site on the Atlantic island of Pessegueiro [328]

A similar pattern was evident at Torre de Ares (Balsa). Although the majority of the finewares dated prior to the mid-fifth century, the final abandonment of the site could be pushed back to at least the later-sixth century by the presence of ARS 90B (Viegas 2007, 75; 2009, Database). Despite the significant quantity of ARS from the site, the quantities of imported Tunisian amphorae were generally low, leading Viegas to conclude that the fineware vessels were unlikely to be imported with cargoes of olive-oil or fish-sauce (Viegas 2007, 76). Tunisian cooking-wares were also imported to the site, although these were not seen to date beyond the early-fifth century. No East Mediterranean amphorae were recovered, although LRC was present at the site; Viegas assigns a minimum count of nine vessels, of which Hayes 3 was the only form identified (Viegas 2007, 75, 83). The presence of LRC at sites around the lower Guadiana River in south-western Portugal has already been mentioned, demonstrating the inland penetration of these 128

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity of this project, to fully consider regions of Iberia to the west of Portugal. However, the absence of LRC from what had been the capital of Late Roman Hispania, Mérida (the major colony town of Emerita Augusta) suggests that late East Mediterranean imports were not distributed this far up the Guadiana (Viegas 2007, 72). The pattern of ARS at the site also suggests very limited connection to Mediterranean exchange networks by the later-fifth century. Apart from a few fragments of Hayes 91 and 104, later forms and styles of this ware were not present (Blázquez Martínez 2002, 165-6). For now, this would appear to contrast with the textual/epigraphic evidence for ‘Easterners’ at the town (Sub-chapter 2.2.2).

imports from the Algarve coast. Viegas noted that as well as small quantities of LRC, the villa site at Montinho das Laranjeiras [330] at the north of the Faro district had produced a small number of later forms of ARS, including Hayes form 104 (Viegas 2007, 72). Further inland, the presence of LRC and even LRD has been noted at the town of Mértola (Myrtilis) [329], in the south-eastern Beja district. Delgado’s 1992 study of the East Mediterranean finewares from the site catalogued 27 fragments of LRC, of which 17 were noted to be rim-sherds of individual vessels: sixteen of these were identified as Hayes 3, the other as Hayes 8 (Delgado 1992, 129). More recently, the chronological pattern of ARS imports to these sites has also been examined in detail, and used to demonstrate wider socio-economic changes in the region (Fernandes 2013).

Viegas noted that most of the Lusitanian sites had only produced single examples of LRC, although certain sites, including Mértola and Tróia, produced greater quantities (Viegas 2007, 73). The site at Cerro da Vila (Vilamoura, Faro District) has also been attributed a relatively large total of 30 vessels, comprising 29 examples of LRC 3 and one LRC 5 (Quaresma and Morais 2012, 376). A phase of Late Antique activity has been identified at this villa-complex/industrial site, following a destruction event in the fourth century (Teichner 2003, 98). The large production workshops/‘fabricae’ studied at the site have been associated with the production of fish-sauces and, possibly, purple dye during the Roman period. However, the largest of these workshops also seems to have been modified and reused for metalworking, and possibly occupation, between the fifth and sixth century. Several hearth/furnace features were identified within a subdivided warehouse structure, as well as large quantities of iron and copper slag (Teichner 2003, 99). The relatively large quantities of ARS-D and LRC from the site were associated with this late phase of activity.

Fernandes’ assessment revealed a slight downturn in importation of ARS at Mértola from c. 420, although he notes that the arrival of late forms of ARS-C in the mid-fifth century represents some continuity of importation. The major drop in ARS supply appeared to occur in the period c. 460-480, which coincides with the first appearance of LRC– although the quantities of the latter never matched those of the African fineware (Fernandes 2013, 875). Subsequently, the site experienced a resurgence of ARS-D imports in the period c. 480-530 – including ARS 99A and 104A – at which point they reached peak levels. This can be seen to mirror the phase of arrival of ARS at sites in western Britain. Significantly, Fernandes proposes that a comparative analysis of fineware imports reveals a disparity between urban locations such as Mértola and Faro which experience the revival of ARS in the late-fifth/earlysixth century and rural area – specifically Montinho das Laranjeiras – where supplies of imports decline sharply from c. 420 without fully recovering. This is emphasised by the presence of ARS vessels of late-fourth/early-fifth century date from Montinho das Laranjeiras that show marks of repair – usually taken as an indicator of shortages in supply (Fernandes 2013, 875).

At the far west of the Algarve coast, evidence of ‘late’ export-driven activity – specifically the production of fishsauce – has been identified at the Roman town of Lagos (Lacobriga). Excavations in 2002 at Rua Silva Lopes revealed a large production complex at the mouth of the Bensafrim River with several fish-processing tanks (Ramos et al. 2006; 2007). Production at the site, dating from the late-first or second century, was thought to stop at the start of the fifth-century, at which point a number of tanks were intentionally backfilled (Ramos et al. 2007, 86). However, three tanks at the far west of the site showed evidence of a second phase of production. The final cessation of activity around the mid-sixth century was indicated by the presence of ARS 91C, 104 and LRC 3 in the fill of these tanks and the absence of any later forms (Ramos et al. 2007, 86). The site also produced evidence of imported coarsewares dating to this secondary phase of activity. Whether the imported finewares recovered at Lagos indicate a hiatus in production in the mid-fifth century, or simply a reduction in output, the site certainly demonstrates the continuation of production for export into the sixth century. Previously, the fish-sauce industry in this region has typically been considered to have finished by the end of the fifth century (Ramos et al. 2007, 88). Increasing evidence of fish-sauce

It would appear that by the second half of the fifth century, imports in southern Lusitania were generally focused on urban locations, echoing the urban distribution of the later imported amphorae in western France. Both rural and urban sites in this region show a marked decline in imported finewares by the mid-sixth century, reflecting the widespread Atlantic reduction in Mediterranean imports. Nevertheless, a few urban sites are noted to have produced smaller quantities of late-sixth century forms; Fernandes mentions individual instances of ARS 99C, 105A and 107 from Mértola (2013, 875). This broad chronological pattern is seen to equate well with south-western Britain and Galicia (Vigo excluded), where the imported finewares suggest a general reduction in long-distance exchange by the mid-sixth century. Fernandes notes that demand for tin in the East Mediterranean seems to have driven trade with these two regions, and proposes that, similarly, the continuing extraction of copper in the ‘hinterland of Lusitania’ might have spurred trade with the Lower Guadiana beyond the end of Roman control. It has not been possible, as part 129

Links to Late Antiquity were identified – principally comprising twelve vessels of ARS 91B and a single example of 91C (Magalhães 2012, 370). Two fragments of DSP were identified, but the forms could not be identified, and these were not assigned to a specific production group. In addition, there was one example of LRC, identified as Form 3, and possibly variant C. This adds to the LRC sherds identified by Maia from unstratified pottery held in the National Archaeological Museum – which Delgado reported to comprise twelve individual vessels of LRC 3 (Maia 1974, 333; Delgado 1988, 45). Magalhães notes that the later imports from the Workshop 1 area were associated with the reuse of this area as a cemetery (2012, 370). Although these finds indicate some connection to long distance exchange into the sixth century, it seens that by this date the site was no longer functioning as a major exporter to the Atlantic. The fineware evidence from Workshop 2 reveals a similar picture of abandonment by the mid-fifth century (Pinto et al. 2010). Elsewhere, pottery recovered by divers from the harbour at Tróia has also been recorded [315]. Apart from a single example of ARS 91B, all of the pottery was dated prior to the mid-fifth century (Fonseca 2004, 427).

production and associated amphora manufacture reveals that this commodity continued to be a significant element in long-distance exchange in the Atlantic. Lusitanian amphorae of fifth century date were potentially identified at Place Camille-Jullian, Bordeaux, and potentially indicate some degree of fish-sauce importation. There is currently no firm evidence that this commodity was transported further north in the Atlantic by this date, although the possible identification of Iberian amphorae at Tintagel (Sub-chapter 3.2.4) suggests it should not be discounted as a factor of trade with Britain. The major fish-salting complex and settlement at Tróia [316] on the Atlantic coast of south-western Portugal (Sétubal district) presents a different picture to Lagos, indicating some continuity of occupation and activity beyond the fifth century, but not necessarily production. This site can, perhaps, be considered to be more representative of the decline in long-distance exchange networks by the mid-fifth century. The site, at the mouth of the Sado River, was first excavated in the 1950s and 60s. These investigations revealed a large scale fish-salting/sauce production complex that was in operation from the first century AD and came to be one of the largest in the Western Empire. Over twenty workshops have been identified, each consisting of a number of production tanks. The pottery from two of these workshops has been published in detail (Pinto et al. 2010; 2012; Magalhães 2012). A significant settlement also developed at the site. To date, a residential quarter and bath complex have been revealed, as well as an early Christian church and several cemetery plots (Magalhães 2012, 363).

The presence of Tunisian cooking wares at Tróia has also been recently discussed, and again these were not considered to date past the mid-fifth century (Magalhães et al. 2014, 708). The amphora evidence from the site is revealing for exports rather than imports; Almagro 51C and Keay 78 have been identified as the main amphorae types carrying fish products from the site between the third and fifth centuries (Almeida et al. 2014, 661). The latest imported amphorae published from the site comprise a single Tunisian Keay 35B (of probable fifth century date) and one East Mediterranean LRA1 (Pinto et al. 2012, 402, 5). It is clear from this scant, later evidence that once production at the site ceased the site never received the volume of Tunisian and East Mediterranean wares that are documented elsewhere in the Atlantic.

The pottery recovered from one of the workshops, ‘Workshop 1’, was recently reassessed by Ana Patricia Magalhães. The majority of this material had been found in deposits filling the production tanks, which appear to have been used as rubbish-dumps after they had ceased to be used for fish-processing. These deposits produced a very large ceramic assemblage, which Magalhães quantified by minimum vessel count. A few late examples of ARS-C were identified, including the late form ARS 73, which demonstrates the arrival of central Tunisian wares into the fifth century (Magalhães 2012, 365). However, the vast majority of the pottery, representing 533 classified vessels, comprised North Tunisian ARS-D (Magalhães 2012, 363). Of this, almost all of the vessels could be dated from the mid to late-fourth century to the mid-fifth century; the D1 fabric was typical, the forms were dominated by ARS 59, 61 and 67, and the 193 stamped bases were matched to Hayes’ ‘Style A’ (Magalhães 2012, 365-70).

To the south of Tróia, the site of Chãos Salgados (Castelo Velho, Santiago do Cacém – identified as the Roman town of Miróbriga), in the District of Setúbal, witnessed a similar pattern of decline and abandonment after the first quarter of the fifth century, from which it never fully recovered. The ceramic specialist José Carlos Quaresma (who assessed the pottery excavated at the site between 1990 and 2000) connected this downturn in long-distance trade to upheaval in southern Lusitania in the period after AD 409, as a result of Vandal pillaging, the Suevic acquisitions in southern Lusitania from 438 and the subsequent Visigothic conquests of the mid-sixth century (Quaresma 2013, 385, 404).

The evidence from Tróia indicates that fish-processing at the site continued until the mid-fifth century (Magalhães 2012, 364). However, the overall pattern of the ‘Workshop 1’ pottery reveals a significant fall in importation from the second quarter of the fifth century, which can again be connected to a wider phase of Atlantic marginalistion. The site seems not to have experienced a renewed phase of production later in the century, although a few later imports

Quaresma’s analysis of the ARS imported to the site revealed that ARS-A and ARS-C predominated, while ARS-D was less common. 149 vessels were matched to the latter series, the vast majority of which dated to the fourth and early-fifth century (Quaresma 2012, 245-8). The overall pattern of ARS at the site also revealed a major decline in imports from c. 420. Nevertheless, the presence of a 130

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity circuits’ that incorporated the port of Lisbon beyond the conquest of the Alans in c. 411 and its subsequent incorporation into the Suevic, then Visigothic Kingdom (Diogo and Trindade 1999, 87)

few later forms, including ARS 91B, as well as six vessels of LRC 3 (variants C, E and H) and a single Keay 62 amphora confirmed that occupation at the site continued – however reduced – into the first half of the sixth century (Quaresma 2012, 247-8, 343, 366; 2013, 404). Magalhães has suggested that additional information from sites in the District of Setúbal will follow, revealing further evidence for the continuing involvement of south-western Lusitania in long-distance exchange networks into the sixth century (pers. comm.).

The excavations in the Rua dos Correeiros, Lisbon, carried out between 1991 and 1995 reveal a similar picture of change and continuity in Late Antiquity (Grilo et al. 2013). This area was characterised by a series of fish-processing units which were in use between the first and fifth centuries. The discovery of late imported pottery in deposits immediately overlying the disused processing tanks confirmed subsequent occupation at the site, although this could not be related to ongoing production for export (Grilo et al. 2013, 849-50). Nevertheless, this adds to the growing impression that Lisbon retained a connection to long-distance networks of exchange despite the political upheavals of the fifth century. The discovery of four vessels of LRC 3 (three of variant 3C and one of 3C/E) in one layer, as well as a single example of LRC 3F from a later layer specifically demonstrated that occupation continued between the second half of the fifth century and first half of the sixth.

Lisbon Further north, increasing evidence from the region around Lisbon [299-312] is starting to reveal that the port of Olisipo, at the mouth of the Tagus River, continued to function as a centre of industry and Atlantic trade into the sixth century, although very little has been published to date (Quaresma and Morais 2012, 380). However, this information is expected to increase greatly as a result of ongoing research by José Carlos Quaresma of the Universidade Nova de Lisboa, which will provide, for the first time, a full quantification of the late imports from the city. Although the data from Lisbon is currently incomplete, a few reports have revealed certain details of late imported pottery identified during earlier excavations. As mentioned, Melim da Sousa discussed finds of LRC from nine sites in the region surrounding Lisbon, which comprised a significant proportion of the overall, Lusitanian distribution, and which presumably represent the redistribution of commodities arriving at the port. He also specifically recorded that this fineware had been found during three excavations in the city itself: from Rua dos Correeiros [310], ‘Teatro Romano’ [309] and ‘Claustro da Sé’ [308] (Melim de Sousa 2001, 202).

One example of LRA2 was also recovered, and the authors noted that this amphora type had been found in other excavations in the city (Grilo et al. 2013, 851). Until the Lisbon data is fully available it remains unclear how the amphora imports in this region compared with Britain, where levels of LRA2 have been noted as unusually high. The vast majority of the amphora recovered from these late layers at Rua dos Correeiros represented Lusitanian types, including Almagro 50 and 51C. Although these would traditionally have been dismissed as residual finds, their association with East Mediterranean imports potentially indicates that some of these local amphorae were in contemporary use. The authors note that increasing evidence from Lisbon, specifically the as yet unpublished finds from the ‘Sommer Warehouses’, indicates that amphorae – and therefore fish products – may have been remained in production in the lower Tagus region for longer than has been previously appreciated (Grilo et al. 2013, 851).

A reappraisal of pottery recovered from excavations at the Roman theatre in 1966/67 was published in 1999 (Diogo and Trindade 1999). Although the finds did not have contextual information, the group included imported pottery that could be dated beyond the mid-fifth century, including examples of ARS 91B and 91C, and six fragments of LRC – including Form 3 and one example of Form 5A (Diogo et al. 1999, 84-91). Additionally, one example of LRD was recorded, and identified as Hayes form 2. As is typical for assemblages in Atlantic Iberia the evidence for late amphora imports from the site is less clear. Of the thirty types recorded, Lusitanian vessels predominated, although the authors noted that some of these could potentially be of fifth century date. A more likely fifth century import was provided by a single Tunisian Keay 35B amphora (Diogo and Trindade 1999, 86). These finds seem to be associated with a late phase of occupation utilising the theatre structure. In the late-fourth or early-fifth century the theatre was dismantled to reuse the stones (possibly for the city defences) and a number of small dry-stone habitations were constructed within its ruins (Diogo and Trindade 1999, 87). Although this reconfiguration demonstrates changing patterns of urban use in the city, the late imports are testament to the continuation of ‘strong commercial

Elsewhere Quaresma and Morais have commented on the amphorae from contexts excavated in the ‘Sommer Sector’ of Lisbon, dating from the late-fifth and first half of the sixth century (Quaresma and Morais 2012, 372, citing Fabião and Pimento, in press). They report that of a sample of 198 amphorae, eastern imports represented 25% of the group and North African amphorae 12%. The high proportion of East Mediterranean vessels for this period would certainly mirror the pattern at post-Roman sites in Britain. The remainder of the amphorae would presumably be of Lusitanian origin, again suggesting their sixth century production. Excavations at the Governor’s Residence at the Torre de Belém [312] in Lisbon revealed a very large fishproduction complex which seems to have ceased production in the fifth century, by which point it had been reduced 131

Links to Late Antiquity later vessels were identified, specifically single examples of ARS 99 and 110 (Viegas 2007, 72). In addition, finds of LRC (seven vessels of Form 3 and one of Form 8) revealed a connection to commercial networks supplying East Mediterranean goods to the Atlantic coast from the second half of the fifth century (Viegas 2003, 202; 2007, 72). As in Britain, nevertheless, the social, political or economic interactions underpinning these networks of coastal and inland redistribution remain to be fully understood.

in capacity (Fabião 2009, 25-8). Again, however, closer analysis has revealed hints of ongoing long-distance economic connections beyond this date. A sixth century coin of Justinian I (ruled AD 527 to 565) was found in the latest deposits in the complex and interpreted as an accidental loss (Fabião 2009, 26). For Fabião, the presence of this low value, copper pentanummium not only indicated the continuing circulation of coinage connected to small scale transactions in Lisbon, but represented evidence of economic exchange with the Byzantine Empire in the sixth century, rather than solely diplomatic interaction (Fabião 2009, 28). These late connections with the East were corroborated by the presence of scraps of LRA1 at the site. Despite difficulties in establishing that these late finds represent more than occupation, it seems increasingly likely that the port of Lisbon was economically active at least into the sixth century. Fabião commented that the Justinianic coin was not a surprising find in light of increasing quantities of East Mediterranean amphorae and fineware at Lusitanian sites, as well as the recognised presence of Byzantine weights in Portugal (Fabião 2009 27-8). This last category of material suggests the existence of organised trade/active markets, although the known distribution of these (which includes Conimbriga and Braga) was seen as unusual for its concentration on inland and northern regions (Fabião 2009, 28) see Figure 6.5. Although Lisbon has been recognised as a major producer and exporter of fish-products in the Roman period, this activity – as in the Algarve – has generally not been thought to continue past the early-fifth century (Fabião 2009, 28-9). However, Fabião concluded that it was unlikely that all of the production units ceased to operate at the same time, and instead proposed that marine resources were likely to have been exploited into the sixth century – thereby justifying relations with the Eastern Empire (Fabião 2009, 29). The presence of late-fifth and sixth century East Mediterranean imported pottery in the city is taken to provide evidence of these ongoing economic contacts. Fabião has collated the various categories of evidence relating to questions of continuity and collapse in fifth and sixth century Portugal, including archaeology, ceramics, coinage and texts. From this disparate evidence he was able to conclude that Lisbon and its port not only remained an exporter of fish-products, but was directly integrated into a complex network of Atlantic and Mediterranean trade operating at a variety of levels along the coast, from where material was redistributed inland (Fabião 2009, 41). This is exemplified by the presence of late imports at Santarém [293] (the Roman town of Scallabis) in the midTagus region (Santarém District). Although 80km from the port at Lisbon, Santarem was situated at an important crossing point of the Tagus, and therefore occupied a pivotal position on roads linking Lisbon with Mérida, as well as Lisbon with Braga (Viegas 2003, 19). Although the majority of the ARS-D identified from excavations at the town represented forms dating between the mid-fourth and the second half of the fifth century, a small quantity of

Figure 6.5. Fabião’s distribution map of Byzantine weights in Portugal. From Fabião 2009, Figure 6

132

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity Conimbriga examples have concave outer rims (see Figure 6.6). In addition to Form 3, three sherds of the contemporary dish form LRC 5 were identified, although LRC 10 is notably absent (Delgado et al. 1975, 287). This suggests that the site was no longer receiving East Mediterranean goods by the second half of the sixth century – as seems to be the case for Britain.

Although the relative volume of material from Lisbon and its hinterland indicates that the scale of commerce to the city may have been reduced from the fourth/early-fifth century, the ceramic and other evidence suggest that a continuing ‘commercial dynamic’ persisted at Lisbon at least until the mid-sixth century (Grilo et al. 2013, 852). More specifically, Lisbon has considered to have played a role in connecting the central Mediterranean to the North Atlantic route, and in supplying the cities of the Western Seaboard (Fabião 2009, 41; Grilo et al. 2013, 852). Nevertheless, questions of the scale, chronology and nature of these connections must await the full publication of the ceramic evidence from the city.

Beyond the LRC, closer analysis of the complete series of fineware imports found at the site illustrate the continuity of connections between the Mediterranean and the Atlantic in the fourth, fifth and sixth century. The sacking of the site in AD 465-8 by the invading Suevi – known from the contemporary account of Hydatius, Bishop of Chaves – was seen to provide a suitable terminal date for all but a ‘handful’ of the identified vessels (Hydatius, Chronicle; Hayes 1980, 482). However, many of the early conclusions about the site and its assemblage have been revised. The redating of a number of ARS-D forms to the later-fifth and sixth century demonstrated that Conimbriga continued to be occupied beyond the mid-fifth century (Fulford 1984a, 113-4; Wooding 1996a, 31, 45; Reynolds 1995, 153).

Conimbriga As discussed in Sub-chapter 2.3 and 6.2.1, the group of LRC found further north in Portugal at Conimbriga [288] has provided the best known equivalent for the British finds and has typically been used to assess the scale and chronology of the Insular imports. Although previously interpreted within models centred on trade with Britain, it is now clear that this group is neither isolated nor exceptionally large, but forms part of a much wider Atlantic – and particularly western Iberian – distribution. As mentioned, 95 identifiable sherds of LRC were recovered from the French-Portuguese excavations conducted in the in the urban heart of Conimbriga between 1964 and 1971. These were not quantified by MNV, although 21 distinct rimsherds are illustrated in the ceramic report (Delgado et al. 1975, 286-91, 308-310). 93% of the identified sherds are reported to be of Form 3, of which early variants B and C dominate, although examples of 3E and F were also noted. The range of variants compares well with the British group, including Tintagel, although early variant LRC 3B has not been identified in Insular contexts (Campbell 2007a, 14). There is current uncertainty regarding the chronology of Hayes’ sequence, although the general impression from Conimbriga is that much of the LRC can be dated to the second half of the fifth century. The two vessels matched to the ‘small variety’ of LRC 3 also suggest an early date; this sub-form was identified at Bantham, although the

979 sherds of ARS were recovered, of which 273 were identified as ARS-C and 680 as ARS-D (Delgado et al. 1975, 255, 338). The earliest forms of ARS-C indicated the arrival of African finewares from the third century, which continued into the fourth century. As mentioned, Reynolds noted the presence of late-third/fourth century relief decorated wares, specifically four vessels of ARS 52B (Delgado et al. 1975, 256; Reynolds 1995, 15). Later forms of ARS-C were also recovered, including six sherds identified as the fifth century form ARS 82 (Delgado et al. 1975, 257). Nevertheless, the majority of the fineware was connected to the earliest series of ARS-D, reflecting the strength of North Tunisian exports into the Atlantic in the later-fourth and early fifth-century. The most common forms recovered at the site – ARS 58B, 59, 61 and 67 – are associated with this early phase of ARS-D arrivals (Delgado et al. 1975, 262). In contrast, Reynolds observed that forms of ARS dating to the second half of the fifth century were rare in the assemblage, demonstrating the marginalisation of the Atlantic from West Mediterranean networks in the early to mid-Vandal period (1995, 18). However, he concluded that this did not indicate a lack of demand, as reflected by the appearance of significant quantities of LRC (Reynolds, 1995, 18, 24). That a replacement supply was necessary implies a degree of economic vitality at the site, as well as the continuation of ‘Roman’ dining practices. The presence of locally-produced imitations at Conimbriga (discussed in more detail in Sub-chapter 6.3.2) also indicates this continuing demand for slipped tablewares (Delgado 1975, 271). Significantly, a number of later forms of ARS-D were identified, including Forms 91C, 99, 103 and 104 (Delgado et al. 1975, 263-8). These demonstrate that the site was connected to the later phase of North Tunisian importation to the Atlantic, particularly from c. 500. However, as with the majority of Atlantic sites, and Britain, there are no forms that can be definitely dated past the mid-sixth century.

Figure 6.6. Examples of LRC 3 ‘small variety’ from Conimbriga. After Delgado et al. 1975, Plate 78.194 and 98.195

133

Links to Late Antiquity have transformed perceptions of trade in the Late Antique Atlantic (Fernández 2014).

As is common for western Portugal there is much less information on the amphora imports to Conimbriga. Despite the significant group of LRC, Reynolds commented on the rarity of East Mediterranean amphorae at Conimbriga (1995, 75, 80). He mentions the presence of one LRA1 amphora – as previously recorded by Keay – and notes that North African amphorae of fourth-sixth century date were also rare (Keay 1984, 655; Reynolds 1995, 80). The table of imported amphorae found in Lusitania presented by Quaresma records one North African Keay 62 from the site (2012, 366, Table 76). As discussed, the prevalence of LRC and ARS over corresponding amphorae likely indicates that the finewares arrived with other commodities – principally grain.

Braga Chris Wickham has commented that, in marked contrast to the rest of the northern and western Iberia, it was only the cities in the north-west of the peninsula that demonstrated signs of post-Roman/early medieval ‘settlement continuity’. Of these, Braga (Bracara Augusta, the capital of Gallaecia) [275] is said to demonstrate the clearest evidence for the continuance of urban life between the fifth and tenth centuries (Wickham 2005, 663). The city was invaded by the Suevi in c. 411 and formed the capital of the Suevic Kingdom of Gallaecia, until the region was finally annexed by the Visigoths in 585 (Morais 2005, 56). However, despite these changes in political control, the city seems to have retained a strong connection to longdistance exchange via the Atlantic sea-routes.

Finally, Campbell has commented that a particular group of fifth or sixth century glass bowls of late Roman/ Mediterranean tradition (his ‘Group A’) with wheelcut decoration found in western Britain (specifically at Tintagel, Trethurgy and Cadbury Congresbury) were also common at Conimbriga, as well as Bordeaux (Campbell 2007a, 57-8). Although glass vessels appear to have been transported within the same systems of exchange as the pottery, the nature and extent of glass distribution in the Late Antique Atlantic needs to be fully established by future comparative research.

As with Conimbriga, the presence of a couple of fragments of ARS and LRC at Braga was recorded in LRP, and the location was included in published maps of the Atlantic distribution of LRC (Hayes 1972, 432; Delgado 1988, Map 2.; Campbell 2007a, see Figure 2.8). However, as detailed information on the pottery failed to emerge, Braga was never fully considered in comparison with the British import sites. More recent analysis on the pottery emerging from the city has shown that the volume of finewares from Braga is on a similar scale to that of Conimbriga.

6.3.2 Northern Portugal and north-west Spain (Gallaecia – Galicia) There is currently little evidence for late imported pottery in the region surrounding Conimbriga, and, moving north, there are no published references to sites with late imports in the region connecting the provinces of Lusitania and Gallaecia until the Duero (Douro) River. The presence of imported fineware in this area has previously been suggested to demonstrate the transport of this pottery – and associated commodities – inland and upriver (see Subchapter 6.2.3). Towards the mouth of the Duero, the presence of LRC and ARS-D have been recorded at Porto [283], although less information is currently available on the incorporation of this port into systems of Late Antique trade in the Atlantic (Fabiao 2009, 37; Alonso Sánchez and Fernández Ochoa 1988, 369). Recently, additional finds of East Mediterranean amphorae, ARS, LRC and even LRD have been reported from further up the Duero at Crestuma [287], although this assemblage awaits full publication (Silva and Guimarães 2011, 9).

A recent study has reassessed the ceramic finds from rescue excavations conducted over the last thirty years at Braga, providing a first complete description of the East Mediterranean finewares recovered in the city (Quaresma and Morais 2012). A total of 85 identified ‘specimens’ were recorded, of which 35 distinct rim-sherds were illustrated (providing the minimum vessel number recorded in Appendix K). Of these, all but two were of Form LRC 3, of which variants C, E and F were most common – although examples of the early variant LRC 3B and the sixth century LRC 3H were also identified (Quaresma and Morais 2012, 374). In addition, two rim-sherds of LRC 5 were catalogued. Furthermore, two rim-sherds were identified as LRD Form 1/2, one of which bears roulette decoration (see Figure 6.7). The two fragments of LRD, along with two of the LRC sherds, were recovered during excavations in ‘Rua Gualdaim Pais’ [276] together with several ‘late glass bowls’ and sherds of a locally-produced grey-ware imitating DSP. The group was found in a sixth-century layer, overlying deposits associated with the destruction of a fourth century hypocaust (Quaresma and Morais 2012, 375-6).

Further north, there has been a substantial increase in information from the region of northern Portugal and Galicia (Gallaecia) supplementing the earlier research by Naveiro Lopez (1991). Again, much of the ensuing discussion and data in Appendices H and K is based on recent synthetic considerations, particularly those by Fabião (2009), Quaresma (2012, Chapter 4, Appendices 1, 2 and 7), and Quaresma and Morais (2012). This sub-chapter is particularly dependent on information provided by Adolfo Fernández Fernández, whose recent publications on the artefacts recovered from various excavations at Vigo

Apart from these few examples, all of the pottery from Braga is reported to be residual in disturbed, modern deposits. Despite the limited stratigraphic information, the absence of LRC 10 again suggests that connections with the East Mediterranean were in decline by the mid-sixth century (Quaresma and Morais 2012, 380). 134

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity of pottery of later-fifth and early-sixth century forms was also recorded – particularly ARS 97, 99, 104 – although the quantities of these were much smaller, with ten or fewer identified sherds of each (Delgado et al. 2014, 671). Subsequently, quantities of ARS were observed to have markedly declined from the mid-sixth century, although a few, very scarce examples of later-sixth century forms – including 99C and one Hayes 106/ Fulford 67 – were identified (Delgado et al. 2014, 672). The authors also noted a ‘dubious’ example of a seventh century ARS 107/ 93A. In addition to the fineware, a North African coarseware import was identified (described as a jug of ‘type 6’), which was dated to the fifth or first half of the sixthcentury (Quaresma and Morais 2012, 382). As discussed in Sub-chapter 4.2.3, the comparable presence of imported coarsewares in Britain has been contested. Five ‘late’ African lamps were also found at Braga, one of which was assigned a fifth century date (Morais 2005, 58).

Figure 6.7. LRD from Braga. After Quaresma and Morais 2012, Figure 7d

The ARS imports at Braga were also the subject of a recent reassessment, which considered fluctuations in their arrival at the city (Delgado et al. 2014). The vast majority of the late fineware recovered in the city was ARS: this comprised 1233 sherds, including 169 identified as ARS-C and 996 of ARS-D (863 of ARS-D1 and 169 of ARS-D2) (Delgado et al. 2014, 671). However, the prevalence of African finewares cannot be seen to contrast with the Insular pattern where LRC predominates, as the period of the greatest concentration of LRC at Braga (from c. 475) was directly mirrored by a drop in ARS (Quaresma and Morais 2012, 376, Delgado et al. 2014, 672) see Figure 6.8.

Rather more evidence is forthcoming on the amphorae from Braga than from other Portuguese locations (Morais 2005) The site is recorded to have produced a larger proportion of East Mediterranean than North African amphorae, suggesting parallels with Britain, although no examples of LRA2 were identified (Quaresma and Morais 2012, 380). Of the 31 identified amphora fragments, six were North African amphorae, of which single examples of Keay 35A and 36B potentially represent fifth century imports, while another single example of Keay 62 could be more directly associated with the later phase of Tunisian exports appearing from the later-fifth century (Reynolds 1995, 53; Morais 2005, 56). A total of eleven fragments were identified as LRA1, most of which were handles, although two necks/upper handles were illustrated (Morais 2005, 56, 65,

Overall, the majority of the ARS represented D1 forms dating from the middle of the fourth century to the earlyfifth century – particularly ARS 59, 61, and 67 – after which point quantities declined in the middle of the fifthcentury (Delgado et al. 2014, 671, 680). However, a group

Figure 6.8. Patterns of fineware importation at Braga. From Quaresma and Morais 2012, Figure 3

135

Links to Late Antiquity Significantly, the peak of production of the red-slipped vessels at Braga seems to occur in the middle decades of the fifth-century, which was considered to reflect the contemporary low-point in ARS imports (Morais and Fernández 2014, 712). The later, increased distribution of TSBTg was similarly suggested to be due to the drop in Mediterranean imports to the Atlantic from around the mid-sixth century (Morais and Fernández 2014, 718-9). It is clear that the production of regional imitations indicates a strong demand for slipped tablewares in north-western Iberia that continued in spite of periods of marginalisation from Mediterranean supply. Like the arrival of LRC in the mid to late-fifth century, these local products filled a ‘gap’ in the market for ceramic tableware. This demonstrates that there is not a simple association between the decline of long-distance imports and the abandonment of urban centres or the disappearance of ‘Roman’ dining practices. At Braga, for example, the reduced supply of Mediterranean imports from the mid-sixth century was not considered to indicate the end of urban life, but to mark a period of greater ‘self-sufficiency’, characterised by local/regional supply (Delgado et al. 2014, 673).

Figure 2). One of these was potentially a late example, identified as Keay 53B (equivalent to LRA1B2) of probable sixth century date (Morais 2005, 57). Four fragments of LRA4 were found, two of which were handles. One of the two rims was also assigned a potentially late date – the second half of the sixth century (Morais 2005, 57, 66, Figure 3.14). The example would seem to match Pieri’s LRA4B sub-type, dating to the sixth and seventh centuries, although a more specific identification has not been proposed here. Thirteen fragments of LRA3 – representing a minimum of four vessels – were also found, including one almost complete specimen of a rare small variant, Pieri’s LRA3B (Morais 2005, 57). Quaresma and Morais concluded that, on the basis of the later-fifth and sixth century fineware and amphora imports, the initial Suevic conquest and the Visigothic invasion of AD 456 did not signal the abandonment of the city or the loss of long-distance connections (2012, 382). However, the second half of the sixth-century and seventh century can be seen to have witnessed a final retraction of these links with the Mediterranean. As well as the imitations of DSP, five fragments of ‘genuine’ DSP were recovered at Braga; these were identified to the Provençal production group (Quaresma and Morais 2012, 382), although Braga was included by Fernández in his distribution of DSPA (2014, Figure 226). The presence of regional imitations of late fineware, both Mediterranean imports and Gaulish DSP, has been increasingly recognised at sites in north-western Iberia, including at Braga, Conimbriga and Vigo. However, it is only very recently that this phenomenon has become fully documented (Fernández and Morais 2012; Morais and Fernández 2014). At Braga, local versions of Spanish sigillata (TSH) were made from the first century AD, which has recently been re-classified as Terra Sigillata Bracarense. The production of finewares at Braga has now been recognised to continue into Late Antiquity, by which point the vessel-forms produced replicated the wide range of longer-distance/imported finewares arriving at the city: TSHT, ARS, LRC and DSP (Morais and Fernández 2014, 709-10). The earlier series of Terra Sigillata Bracarense Tardía (TSBT) were covered with a red slip (TSBT rojo/r) and frequently stamped or rouletted, but in the later-fifth century these were gradually replaced by grey-slipped vessels (TSBT gris/g) which were less commonly decorated, but sometimes featured incised or impressed lines (Morais and Fernández 2014, 712). These later, grey products (TSBTg) may have been in production into the seventh century, based on finds from Vigo (Morais and Fernández 2014, 715).

A Coruña, Galicia In spite its distance from the Mediterranean, there is increasing evidence to show that Galicia was not isolated from patterns of Late Antique exchange. As discussed in Sub-chapter 6.2, analysis of fineware distributions have shown that the north-west of Spain was periodically incorporated into long-distance networks supplying Mediterranean pottery and (therefore) staples such as wine, oil, grain, as well as other exotic commodities. Furthermore, it has become apparent that Galicia was a crucial zone of connection on the western Seaboard between the Mediterranean and the North Atlantic, as well as to Atlantic regions of France. Reynolds has questioned whether sites such as Braga, Vigo and Conimbriga simply ‘took advantage’ of shipping heading north to Britain or were able to attract a market ‘in their own right’ (Reynolds 2010, 108). Although he concluded that it was the British market that provided the inspiration for these shipments, the growing corpus of material from Galicia suggests that it was in itself a major focus of Atlantic trade between the later-fifth and seventh century. Recent research by José Carlos Sanchez-Pardo has considered the broad archaeological and textual evidence for interregional connections and economic activity in Galicia from the fifth century onward. He asserted that the region experienced a ‘previously unsuspected vitality’ in Late Antiquity, highlighting the continuing occupation of settlements such as Tui [271] and Iria/Iria Flavia [244], and coastal villa sites such as Noville and Moraime [234] beyond the end of Roman control (Sánchez Pardo 2013, 143-4). In particular, he proposed that increased exploitation of extensive mineral resources in Galicia during this period, specifically tin and gold, was directly connected to demand for metals in the Byzantine East (Sánchez Pardo 2013, 151).

Although the known distribution of these wares is currently limited to western Iberia, it is feasible that a few items might have been exported, especially if – as will be argued – the British imports were redistributed from a Galician port (see Figure 6.9). It is therefore theoretically possible that amongst the small and abraded sherds of fineware found in post-Roman contexts in western Britain there are examples of these Iberian ‘imitation’ products. 136

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity

Figure 6.9. The distribution of ‘Terra Sigillata Bracarense Tardía’ in the Atlantic. After Morais and Fernández 2014, Figure 11

Within Galicia, the known distribution of late ceramic imports has continued to rise, adding to the finds catalogued by Naveiro Lopez. In particular, a major synthesis of imported Mediterranean finewares by Ma. Catalina López Pérez revealed patterns in the distribution of ARS-C, ARS-D and LRC within the province of A Coruña. The wider Galician distribution of imported finewares was considered further in a subsequent article (López Pérez and César Vila 2010). In general, African finewares are shown to predominate amongst the imports, and are largely distributed at coastal locations, demonstrating their association with maritime supply (López Pérez, and César Vila 2010, 44). Once again, the vast majority of the examples can be seen to relate to the wave of ARS-D exported to the Atlantic in the fourth and first half of the fifth century, including examples of ARS 59, 61 and 91A/B at Iria Flavia and Cidadela and examples of the latter form at Moraime (López Pérez 2004, 447-57). Nevertheless, the identification of LRC 3 at the same sites points to the continuation of importation into the later-fifth century – reflecting the new connection to the East Mediterranean (López Pérez 2004, 23, 477-8).

Roman port of Brigantium) [221-231]. As discussed in Sub-chapter 6.2, Naveiro Lopez recorded the presence of ARS, LRC and possibly late amphorae. Although little has emerged since on amphorae, significantly more has been published on late fineware recovered from various rescue excavations close to the harbour area of the city. López Pérez has catalogued several examples of ARS-D dating to the second half of the fourth/first half of the fifth century, including ARS 59, 61 and 67 (López Pérez 2004, 48-69; López Pérez and Botella 2010; López Pérez and Muiño Maneiro 2011). The one example of the large dish/ bowl ARS 67 – one of the typical forms of this period – was found during excavations at the monumental Roman lighthouse of Torre de Hércules [221] (López Pérez 2004, 48-69). As mentioned in Sub-chapter 3.3.2, an example of this form at Southwark in London was interpreted as a very late Roman-period import, although it is clear that later variants circulated in the Atlantic beyond AD 410 (Fernández 2014, 169). The only clear example of ARS relating to the later phase of importation was a single rim-sherd of Form 99 from excavations at the villa of O Cantón Grande [231], situated to the south-west of the Old City, on the isthmus leading to the headland (Os Cantóns) (López Pérez 2004, 68-9; López Pérez and Vázquez

Until recently there has been relatively little detailed information from the city of A Coruña (the important 137

Links to Late Antiquity conducted at Vigo (Fernández 2014; see Figure 6.10). Apart from the excavations at a villa site – the ‘Villa de Toralla’ [270] – to the south-west, all of these were situated in the area of ‘O Areal’, the bay/natural harbour to the north-west of the hill-fort settlement of the ‘Castro de Vigo’ (Fernández 2014, 8). Much of the information relates to finds recovered from two large rescue excavations, termed ‘Unidad de Actuación Rosalía de Castro I and II’ (UARCI and II, [261] and [267]), that have been conducted in this area since 1992. Full stratigraphic data was available for these two sites, as well as from the excavations at ‘Marqués de Valladares’ [263] and the ‘Villa de Toralla’, allowing Fernández to fully appraise the fineware in relation to the archaeology. A sequence of 21 datable contexts were established from these four excavations, which, by close analysis of the wares, forms and variants, revealed chronological shifts in the fineware arriving at Vigo from the later-fourth century. Other categories of material were also studied from these four excavations, including, notably, the amphorae and coarseware from UARCII. Although the entirety of the fineware recovered from a second group of excavations was studied, comparable stratigraphic information was not available, and the finds were assessed independently, in relation to this general chronology (Fernández 2014, 8). The pottery from a third group of excavations could only be partially studied; only certain, significant pieces of fineware were considered. No additional categories of pottery, including amphorae, were studied from the two latter groups.

Collazo 2007, 101). This site also produced at least eight individual vessels of LRC, all of Form 3; although variants were not specified, the majority match most closely to LRC 3F, indicating occupation at the villa into the sixth century. These finds bring the estimated minimum total of LRC from A Coruña to thirteen vessels. The overall quantities from A Coruña are clearly much less than from Vigo, suggesting that the latter location was the primary import point for Mediterranean imports to the north-west from, at least, the later-fifth century. Notably, the city has not produced any sixth/seventh century forms of ARS or examples of LRC 10, unlike Vigo, which again demonstrates the primacy of the latter port as a hub of exchange by the later-sixth century. Regardless, it should be noted that, apart from Tintagel, more vessels of LRC have been found at A Coruña than at any of the Insular import sites. Vigo The scale and variety of pottery recovered from various excavations at the city of Vigo, on the coast of south-western Galicia [258-69, 270], has demonstrated not only its importance as a regional port, but its integral role within post-Roman and early medieval systems of exchange operating in the Atlantic. The city of Vigo is situated in the region called the ‘Rías Baixas’, which extends north from the Miño River to the ‘Cape of Finisterre’, and is characterised by four large estuarine inlets – ‘Rías’ – that provide safe natural harbours (in contrast to the hazardous coastline of north-western Galicia) as well as rich marine resources (Fernández 2010, 229). These environmental assets led to the region becoming an important focus for Late Republican and Roman trade in the Atlantic. This trade was initially geared toward supplying luxury items, and particularly wine, to defended riverside sites – termed ‘castros’ – including the ‘Castro de Vigo’, and then subsequently to the ‘semi-urban’ settlements, such as at Vigo and Tui, that developed during the Roman period (Fernández 2010, 229). Fernández has described how the later Roman period witnessed the decline and abandonment of the ‘castro’ sites as well as the appearance of a number of coastal villa-sites, many of which also functioned as centres of industrial activity and production (2010, 234). These villas became a new focus for commodities arriving from the Mediterranean and Lusitania until the early-fifth century. As elsewhere in western Iberia, this point marked a period of upheaval; many of the villa sites were abandoned and a number of the fortified settlements were reoccupied (Fernández 2010, 229). It is after this point that Vigo, on the southernmost ‘Ria’ (of the same name), seems to emerge in a new position as a ‘harbour for the reception and redistribution of African and eastern goods’ (Fernández 2010, 234).

Given this differential treatment, the data presented in Appendices H and K – as drawn from Fernández’s thesis – does not represent the entire assemblage of Late Antique pottery from Vigo, but only a large sample. Nevertheless, the vast quantity of material that has been studied – particularly the Mediterranean fineware – allowed Fernández to construct a comprehensive model of ceramic imports reaching the port. This analysis has revealed the shifting patterns of exchange connecting Vigo with the Mediterranean and the Atlantic between the fourth and seventh centuries. Outlining the development of Vigo, Fernández described how during the first or early-second century, this beachside area of ‘O Areal’ was developed into a complex for the production of salt by evaporation – itself potentially connected to the manufacturing of fish-products (Fernández 2014, 8). Salt production ceased in the later-third or fourth century, possibly as a result of the regression of the sea. After this point the ‘O Areal’ area was redeveloped, becoming, by the fifth century, the central location of Late Antique occupation; excavations have revealed a number of residential buildings converted from the abandoned industrial structures (Fernández 2014, 9). The eastern part of the bay, uncovered by the ‘UARC II’ excavations, seems to be the focus of the late harbour-side settlement, revealing evidence of habitation, industry and, most significantly, commercial activity (Fernández 2014, 8-9). Unfortunately, as Fernández notes, extensive modern

The recent publication of doctoral research by Adolfo Fernández Fernández has provided a highly detailed assessment of the pottery recovered from twelve excavations 138

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity

Table 15 Fineware totals from Fernández’s research at Vigo. After Fernández 2014, 518-20, Tables 48-57.

Figure 6.10. Location of excavations at Vigo with late pottery studied by Fernández. From Fernández 2014, Figure 5

modifications of the port area prevent a full understanding of the Late Antique harbour and the exceptional, protected moorage it afforded ships plying the Atlantic routes (2014, 8).

mately ten times the total number recorded from Britain and Ireland (Fernández 2014, 222). This large quantity of LRC, greater than that of any other site in the West Mediterranean, confirms that Britain can no longer be seen as the focus for the distribution of East Mediterranean fineware in the Atlantic.

Fineware The quantity of fineware studied by Fernández not only demonstrates the massive scale of the material that arrived at the port, but reveals considerable parallels for the British post-Roman import assemblage. A total of 3595 sherds of fineware were studied, representing eight wares of varied origin, from which 1423 individual vessels were identified (see Table 15). The pottery was quantified by minimum number of ‘individuals’ (‘NMI’/MNV), based on the number of distinct rim-sherds (after examination for joining fragments), as well as base-sherds of types which were not otherwise represented (Fernández 2014, 11). All subsequent counts noted in this discussion represent minimum vessel counts (MNV).

LRC Form 3, which represents the entire Insular occurrence of this ware, dominated the assemblage (see Figure 6.11). In contrast with Britain, however, a wide range of forms of LRC were identified: this comprised a single example of the potentially early form LRC 1, examples of the broadly contemporary forms 5, 6 and 8 and, most significantly, 35 vessels of LRC 10. The presence of this late form– unique in the Atlantic– demonstrates the continuing arrival of East Mediterranean finewares at Vigo beyond the mid-sixth century. LRP assigned a late-sixth to early-seventh century date to variants A and B of Form 10, and an early to midseventh century date for variant C, although the date for the introduction of this form was subsequently amended to the mid-sixth century (Hayes 1972, 346; 2008, 86).

In contrast to many of the sites on the Atlantic, and specifically to Conimbriga and Braga, the overall quantity of ARS-D was smaller than LRC (although ARS as a complete category was marginally larger). This does, however, equate with the pattern at early medieval sites in western Britain, where LRC typically outnumbers ARS. The total count of LRC – 605 vessels– is, nevertheless, approxi-

Variant A was the most common at Vigo, comprising 28 vessels, followed by four examples of 10B. Fernández also suggested the possible presence of later variants connected to 10C (Fernández 2014, 256). Overall, LRC 139

Links to Late Antiquity Table 15. Fineware totals from Fernández’s research at Vigo. After Fernández 2014, 518-20, Tables 48-57 Category

Sherd Count

MNV

Percentage (% MNV)

ARS-C

160

52

3.7

ARS-C/E

1

1

0.1

ARS-D, C/D and ‘tardias de pasta marron’

1784

579

40.7

LRC

1203

605

42.5

LRD

18

15

1.1

Unidentified East Med

3

2

0.1

DSPA/DSP

102

65

4.6

TSHT

139

37

2.6

TSBTr

159

45

3.2

TSBTg

26

22

1.6

1423

100

ARS

TOTAL

44.4

Mediterranean shipments to the Atlantic in this late phase.

10 was commonly identified in contexts dated between the late-sixth and the ‘first third of the seventh century’ (Fernández 2014, 256).

Additionally, of the vessels identified as LRC 3, a wider range of variants was present at Vigo than has been recorded in Insular contexts. These include early variants 3A and 3B, which have not been identified in Britain and appear to relate to the earliest phases of exportation to the Atlantic in the mid-fifth century. Similarly, late variants 3G and the intermediate Form 3/10 are further testament to arrivals up to and beyond the mid-sixth century. A new variant, characterised by a recurved rim – ‘borde reentrante’ – was also identified and considered to be late in the sequence (Fernández 2014, 250). Of the LRC 3,

The extended chronology of the LRC sequence at Vigo therefore demonstrates that connections between Galicia and the Byzantine East continued after shipments reaching Britain had stopped. Reynolds has described a general lack of LRC in the West Mediterranean after c. 575, which, made finds of LRC 10 in Britain and at Vigo particularly ‘significant’ (Reynolds 2010, 123). However, as Campbell dismissed the LRC 10 from Tintagel, it is clear that Vigo – and therefore Galicia – was the focus for East

Figure 6.11. LRC forms from Vigo and Britain and Ireland by MNV. After Fernández 2014, Graph 20 and Campbell 2011

140

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity The earliest phase at Vigo (‘sub-horizon A1’ – dating to the second half of the fourth century) was identified at the Villa de Toralla. Here, ARS dominated fineware supply: this principally comprised central Tunisian vessels with applied decoration, as well as earlier forms of ARS-D – particularly ARS 58, 59 and 61A (Fernández 2014, 128). Spanish TSHT was the second most common fineware during this period.

almost half – 262 vessels – were identified as of variant F (Fernández 2014, Graph 21). The Vigo LRC can be seen to exemplify the large volume of East Mediterranean fineware transported to the Atlantic in the period broadly equating with its arrival at sites in western Britain and Ireland. The peak of LRC importation at Vigo was, however, suggested to be from the mid-sixth century – after the established date for the British examples (Fernández 2014, 129).

During the first third of the fifth century (‘sub-horizon A2’), ARS-C became generally rarer at Vigo and was largely supplanted by North Tunisian ARS-D. The presence of ARS 67 in this phase can be correlated with the late Romano-British ARS from London. In contrast with Britain, however, the subsequent phase (‘sub-horizon A3’, dated to c. 440/50) was not characterised by a complete cessation in the supply of fineware. Although quantities of ARS declined massively, Fernández identified some continuity of importation in relation to small quantities of specific forms (including ARS 67 and 73). As at Braga, the Vigo assemblage demonstrated a continuing demand for finewares during this lull in Mediterranean supply; this was met by Iberian productions, TSHT and TSBTr. The latter, regionally-produced, fineware came to the fore at this point, at which time, Fernández notes, new imitations of African shapes were produced, such as ARS 61B (see Figure 6.12) and a composite of forms ARS 59 and 67 (2014, 128, 314-8).

The range of LRC 3 variants recorded from Britain – 3C, D, E and F – are represented at Vigo in contexts dating between the later-fifth and mid-sixth century, although Fernández commented that 3E and F were particularly representative of the period c. 525/530-c. 550 (2014, 129). The possibility that LRC arrived in Britain up to the mid-sixth century goes some way towards contradicting Campbell’s model of the phased importation of Mediterranean wares, whereby an ‘East Mediterranean package’ of wares (arriving from c. 475) was largely replaced c. 525 by a North African ceramic ‘package’ (2007a, 127). The Vigo assemblage is also notable for the quantity of LRD recovered – a minimum of fifteen vessels – which comprises the largest group in the Atlantic. The forms present, principally Hayes forms LRD 2 and 9, also demonstrate the long timescale of East Mediterranean imports to the port. In particular, the four examples of LRD 9 and one identified as LRD 9/10 can be dated to the later-sixth or into the first half of the seventh century (Fernández 2014, 264-7; see Sub-chapter 3.3.4)

The subsequent phase identified at Vigo, ‘Horizon B’, was broken into two ‘sub-horizons’, but as a whole can be considered as broadly equating to the period of post-Roman importation to Britain. Following the pattern observed at many sites in western Iberia –particularly in coastal regions – imports of ARS at Vigo recovered in the later-fifth century, as connected to the new expansion of Tunisian exports. ‘Sub-Horizon B1’ (dated by Fernández to c. 450 to c. 525/30), contained a number of new forms of ARS-D, particularly forms of North Tunisian ‘ARS-D2’ produced at Oudhna and ‘Atelier X’ (2014, 128). Fernández highlights the large bowl ARS 103B as particularly representative of this new phase of ARS imports. Campbell’s catalogue of Insular imports includes two examples of Form 103 (identified as the Carthage form Fulford 64) which equate most closely to variant B. The only British example of ARS with an identifiable stamp is a fragmentary 103B from Cadbury Congresbury (see Sub-chapter 3.3.2). This phase also witnessed wider dynamics in the fineware imports reaching

The overall quantity of ARS at Vigo – 632 vessels – was again vastly in excess of the post-Roman total from Britain (a minimum of 36 vessels). As noted, this figure was roughly the same as the total quantity of LRC, although Fernández observed that this apparent equivalence masked deeper, chronological fluctuations in the pattern of imports arriving at the port. Tunisian finewares arrived at Vigo from the mid-fourth century – with the earliest examples recovered at the Villa de Toralla and the early phase at Marqués de Valladares – while LRC did not appear for another century (Fernández 2014, 222). Analysis of the ARS also revealed the extended chronology of connections between Vigo and the Mediterranean. The largest category of African Fineware was ARS-D, with the most common forms, ARS 91, 99, 103 and 104 paralleling the British examples. Again, however, a much wider range of forms and variants was identified. These can be considered in light of the model discussed in Sub-chapter 6.2.3, where ‘waves’ of ARS were distributed beyond the West Mediterranean and into the Atlantic, and with the detailed sequence of contexts and phases established by Fernández (2014, 128-32). Three phases, termed ‘horizons’, were identified by the Vigo specialist, which were then split into dateable ‘sub-horizons’. Analysis of the pattern of wares seen within this sequence of horizons allows a closer understanding of the British material – see Table 16.

Figure 6.12. Example of TSBT Form 8 from Vigo. After Fernández 2014, Figure 170

141

142

‘Horizon C’

‘Horizon B’

‘Horizon A’

‘C’

‘B2’

c. 560/70-c. 620/30

c. 525/30-c. 550

c. 450-c. 525/30

c. 440/50

‘A3’

‘B1’

c. 400-c. 430

c. 340/50-c. 400

‘A2’

‘Sub-horizon A1’

E ware from Bordeaux.

DSPA imports continue (Rigoir 36).

LRC 3F, 3G. LRC dominates ARS

Late ARS-D (90B, 105 etc).

LRC increases – especially 3E, 3F.

ARS imports continue.

DSPA (esp. Rigoir 1 and 4).

LRC appears - 3B, 3C, 3D. Some LRD.

End of TSHT.

Recovery of ARS-D supply.

TSHT and TSBTr fill gap in fineware demand.

Massive drop in ARS- but some continuity.

Arrival of new East Mediterranean Amphorae, LRA1A, LRA3, LRA4A.

ARS-D imports continue – e.g. Hayes 61, 67.

ARS-C becomes rare.

ARS-C and ARS-D imported to Atlantic. Earlier ARS-D, e.g. Hayes 58, 59, 61A.

Western Britain re-integrated into Atlantic systems c. 475 – 550 Mediterranean wares reach Ireland

E ware arriving from Bordeaux but with revised distribution.

No very late DSPA imports?

Western Britain and Ireland stops receiving Mediterranean goods.

DSPA (directly from Bordeaux?)

LRC

ARS

Keay 25.2, Africana IIIC. Spatheion 1 at Shadwell.

ARS 67 at Southwark, 91 at Pevensey.

Characteristic forms/types

E1 dominates.

High proportion Rigoir 29 but no 36.

No LRC 10. No late (post-Vandal) ARS forms – Pevensey an exception?

Rigoir 4, 16, 29.

Insular examples all Hayes 3.

Hayes 91C, 99A-B, 103, 104.

Britain ‘cut-off’ from Atlantic supply.

Imports of East Mediterranean LRA3 and LRA4 to urban locations.

Last few North African amphora imports arrive in southern Britain.

Late arrivals of céramique à l’éponge.

Rare ARS in Britain.

Comment

Dated Phase

Significant types and forms

Britain and Ireland

Vigo (from Fernández 2014, 128-32, 456, Figure 236)

Table 16. Comparison of the chronology of ceramic importation to Vigo and to Britain and Ireland

Links to Late Antiquity

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity In addition to the Mediterranean and Iberian finewares, the Vigo assemblage contained finewares of Gaulish origin, specifically DSP. Of the 65 vessels, 64 were matched to the Atlantic production, ‘DSPA’, with the majority equating with descriptions of material from Bordeaux (Fernández 2014, 272-3). Wooding had previously noted the apparent absence of DSPA in Spain, commenting that there was a lack of archaeological evidence to corroborate textually attested links between western Gaul and Galicia (1996a, 57). This group of DSPA provides evidence of specific connections between Vigo and Bordeaux. Fernández also reported the possible presence of DSPA at Adro Vello [253] and Moraime in Galicia, and at Braga, although Quaresma and Morais only recorded Provençal products from the latter site (2007, 332). Regardless, this published group from Vigo represents the largest collection of DSPA outside of Aquitaine (Fernández 2014, 273). It is furthermore larger – more than double – the entire Insular collection of 29 vessels.

Vigo, particularly the late arrival of central Tunisian ARS-C, the first arrivals of East Mediterranean LRC and LRD, the disappearance of TSHT, and the substitution of TSBTr with the later series of grey tablewares – TSBTg (Fernández 2014, 128-9). The rise of these grey, regional ‘imitations’ reflects the contemporary appearance of another imported fineware product, DSPA, again one of the wares identified in Britain. A second sub-horizon, ‘B2’ (dated to c. 525/30 to c. 550) can also be equated with the period of Mediterranean importation to western Britain. Large quantities of ARS-D continued to arrive, while imports of LRC increased; this can be seen to contrast with Britain, where the African imports have been seen as typically later than the LRC. The ARS forms identified as characteristic of this period, including 91, 99, 96, 103, 104, directly equate with the African forms found in western Britain (see Appendix C). Some of the later variants of these forms were also identified within the subsequent phase – ‘Horizon C’ – at Vigo (dated c. 560/70-c. 620-30), but none of the British examples can be dated with certainty past the mid-sixth century. One possible exception is the sherd of ARS 104B from Iona (included by Campbell as Fulford 65B), which is the only Insular example of ARS 104 where a variant has been published (Reece 1981, 15). Fernández suggests that this variant is best identified to the second half of the sixth century, or possibly later, although LRFW1 suggested the transition from 104A might occur slightly earlier (Cau et al. 2011c, 6; Fernández 2014, 200). Unfortunately, the sherd from Iona is not illustrated and is noted in Campbell’s database as missing (Campbell 2011). Certainly, the typical very late forms of ARS identified at Vigo within ‘Horizon C’ (dated c. 560/70 to c. 620/30), including 90B 91D, 99C, 104C, 105, 107 and 109, are not present in the British group. These, like the LRC 10, demonstrate the arrival of Mediterranean finewares at the Galician port in the later-sixth and into the seventh century. Outside of southern Lusitania, the only site with similar late ARS forms is Bordeaux – and even there in considerably smaller quantities. The full significance of this late chronology is discussed in Chapter Seven.

Overall, the plate form Rigoir 4 is the most common at Vigo, which is also one of the most common forms identified in Britain. However, certain differences with the complete British group are also apparent, as shown by the relative percentages of forms in Figure 6.13. A greater range of forms has been identified at Vigo than in Britain and Ireland, including two examples each of bowl forms Rigoir 6 and 18. The presence of two vessels of the plate Rigoir 1, again not recorded in the Insular assemblage, might reflect connections between Bordeaux and Vigo operating prior to those between south-western France and western Britain. As mentioned, this form was thought to have declined at Place Camille-Jullian in Bordeaux by the later-fifth century (see Figure 5.3). The proportion of mortaria form Rigoir 29 at Vigo (representing nine vessels) is relatively smaller than Britain, where it is the most common form, but is still higher than Bordeaux (Fernández 2014, 287-8). Campbell has suggested that the Insular prevalence of this form might reflect a specific demand for vessels used for preparing food in a ‘Roman tradition’ (2007a, 27). Fernández proposed that differences in the relative proportions of forms in Britain, Bordeaux and Vigo/northern-Spain might be connected to the varied chronology of importation to these regions (2014, 275). As discussed in Sub-chapter 5.3.4 the British group of DSPA reflects the contemporary repertoire of forms produced at Bordeaux. The more diverse collection of DSPA from Vigo, which includes Rigoir 1 as well as a significant presence of Rigoir 36 (fourteen vessels), can therefore be seen to reflect a longer history of importation. Form 36, a pitcher equivalent to the E ware form E4B, has not been identified in Britain but peaks in the latest early medieval phase at Place Camille-Jullian. At Vigo, although present in contexts dating to the earlier-sixth century, Rigoir 36 was noted to be most representative of the late-sixth and earlier-seventh century (Fernández 2014, 291). It would seem, therefore, that Gaulish fineware arrived at Vigo beyond its latest importation to Britain. The discovery of DSPA in association with Mediterranean imports and E ware at Insular sites led

In spite of the acidic soil at Vigo, detailed examination of forms and fabrics of the ARS allowed Fernández to identify most of the fineware to specific production regions/ workshops. This level of identification has not, to date, been attempted with the British examples, but represents a challenge for future research. Significantly, Fernández was able to distinguish late ‘ARSC/D’ vessels originating from Sidi Khalifa in the Gulf of Hammamet area of Tunisia. Late products from this area had not previously been considered to have been distributed beyond North Africa (Fernández 2014, 208-9, 129). Furthermore, vessels of unclear origin (but previously categorised as ‘Egyptian Red Slip Ware’) were identified, and classified as ARS ‘de pasta marron’ (Fernández 2014, 214). Both of these groups were particularly connected with the latest phases of importation – Horizon C – with forms dated well into the seventh century.

143

Links to Late Antiquity

Figure 6.13. DSPA forms at Vigo and in Britain and Ireland by percentage. After Fernández 2014, 274, Graph 23 and Campbell 2011

vided by Campbell, and is two-thirds of the 150 vessels suggested for Tintagel by Thorpe (Fernández 2014, 355; see Sub-chapter 4.2.3). The majority of the amphorae in these late contexts at Vigo were of East Mediterranean origin, of which LRA1 dominated (see Table 17). All were classified as sub-type LRA1B, only one of which was recorded as the late variant LRA1B2 (Fernández 2014, 340-2). The varied fabric indicated a range of sources in the north-east Mediterranean. The prevalence of LRA1 over a small quantity of LRA2 can be seen to contrast with the Insular pattern, where LRA2 forms a relatively high percentage of amphora imports. However, as discussed in Chapter Four, a bias towards LRA2 is a particular characteristic of the Tintagel assemblage, and does not apply to every Insular assemblage. The focus on LRA1 at Vigo certainly equates with the pattern at Bantham, within South Devon as a whole (as determined in Sub-chapter 4.3), and is more typical for the Atlantic in general.

Campbell to view it as a ‘transitional ware’ between separate phases of Mediterranean and Continental importation (2007a, 31). The presence of DSPA at Vigo in contexts with very late Mediterranean imports reveals that this model cannot be applied to the Atlantic as a whole. Amphorae The various excavations at Vigo have also produced significant quantities of amphorae, although these have not, to date, been studied as comprehensively as the fineware. Although most categories of material were studied from the early contexts that were identified at the Villa de Toralla (fourth to early-fifth century date), amphorae were noted to be scarce from this excavation, and principally of local or Lusitanian/Iberian origin (Fernández 2014, 333). The bulk of the information available on amphorae at Vigo principally relates to late contexts identified at the UARC II excavations; the complete assemblage from these was studied in full. Contexts ‘19-22’ were identified as the latest occupation deposits at the harbour settlement, representing activity dating from the second half of the sixth century (particularly the late-sixth century) to the first third of the seventh century (Fernández 2014, 132, 334). The ultimate abandonment of this area, the closest to the harbour/beach, was suggested to be c. AD 630/640 on the basis of the pottery (Fernández 2014, 113).

Furthermore, Fernández noted that LRA2 was particularly common in one context at the Marqués de Valladares excavations, dated to the first half of the sixth century. Unfortunately, the material from this site has not been, as yet, fully quantified and equivalent data is not available. Nevertheless, this offers a clear point of association for the Tintagel assemblage – it may be that the relatively large quantities of LRA2 recorded in south-western Britain reflect a specific chronological phase of Atlantic exchange (Fernández 2012, 357; 2014, 344; pers. comm.). LRA1, 3 and 4 were also identified at Marqués de Valladares (Fernández 2012, 357).

As such, the recently published group of amphorae studied at Vigo postdates the established date-range of the amphora imports to western Britain and Ireland. This prevents a direct comparison between the two assemblages. Nevertheless, the quantity of amphorae identified from Contexts 19-22 alone (101 vessels identified from 2816 sherds) is equivalent to the lower figure from Tintagel pro-

A wider range of East Mediterranean amphorae were recorded from Vigo than has been identified at the Insular sites, including the late evolved version of LRA2 – LRA13 144

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity Table 17. Imported amphorae from Contexts 19-22 at Vigo. After Fernández 2014, 335, Graph 33, 446, Table 51 Type

MNV

East Mediterranean

Percentage % Of total Amphora

Of total East Med

LRA1

48

47.5

76.2

LRA2

5

5

7.9

LRA13

1

1

1.6

LRA4

4

4

6.3

LRA7

1

1

1.6

LRA10

1

1

1.6

‘Samos Cistern’

1

1

1.6

UnID

2

2

Total East Med

63

62.4

African

18

17.8

South-Spanish

8

7.9

Lusitanian

3

3

Unidentified

9

8.9

101

100

Total

3.2

evidence for late Iberian amphorae being transported in the Atlantic, thereby strengthening the possibility of similar finds amongst the Insular material, especially in the late-fifth or early-sixth century.

and the Egyptian type LRA7. Campbell noted a possible spike of LRA7 amongst the unidentified amphorae at Tintagel, although this awaits confirmation (Campbell 2007a, 22-3). Other possible British examples of postRoman imported Egyptian amphorae (from London, Towcester and Old Sarum) were considered by Tomber and Williams, but the archaeological provenance of these could not be confirmed (2000, 49-51). Four examples of LRA4 were identified from Contexts 19-22 at Vigo, one of LRA4A2 and three of the late sub-type LRA4B2/3 (Fernández 2014, 344-6). The low proportion of LRA4 differs from the typical pattern in the West Mediterranean where this type is usually common (Reynolds 1995, 81-2). Doubtlessly, many of the examples of LRA4 recorded from western France included in Chapter Five predate ‘Contexts 19-22’ and can potentially be equated with late Roman period imports to Britain. Nevertheless, LRA4B has been identified at Bordeaux, and was certainly much more common at Place Camille-Jullian than Vigo. This was one of the differences that led Bonifay to suggest the possibility that some pottery might have reached Bordeaux by different channels to Vigo – including overland from the Mediterranean (Bonifay 2012, 256).

Although the quantified data refers specifically to this late phase of activity at Vigo, Fernández’s broader chronological discussion includes additional information on amphora imports, which further illustrates the connections with the British examples. It should be noted that, as in Britain, the dates of the amphorae cannot be assigned as closely or securely as for the fineware. Although the first East Mediterranean amphorae were stated to appear from the early-fifth century, and were presumed to have been imported throughout the century, it was the late-fifth and early-sixth century that saw the biggest rise in importation. ‘Horizon B’, which compares with the phase of importation to Britain, was stated to be characterised by LRA1 and LRA2, with smaller quantities of LRA3 and 4 (Fernández 2014, 129). Fernández noted that African amphorae might have re-appeared early in this phase following a hiatus in the mid-fifth century – as indicated by the pattern of the ARS (2014, 128-9). These patterns clearly demonstrate significant parallels to the Insular pattern, indicating that rather than being exceptional, the British group of amphorae is broadly representative of typical, contemporary patterns of distribution in the Atlantic. The contrast comes in the following phase of late-sixth and early-seventh century exchange, when Britain and Ireland appear to have been cut-off from supply.

Despite the focus on eastern amphorae, a number of Tunisian amphorae were identified: principally Keay 62, followed by Keay 61, both of which were present at Place Camille-Jullian. Two late Spatheia were also recorded from Contexts 19-22 (Fernández 2014, 336-9). In addition, the presence of a significant proportion of Iberian amphorae indicates continuing economic links with Betica and Lusitania. Two examples of Almagro 51C were considered as potential late Lusitanian products, while Fernández also mentioned the presence of other examples in ‘SubHorizon B1’ (Fernández 2014, 129, 348). This adds to the

Coarseware Beyond amphorae, a highly significant collection of coarsewares were studied at Vigo, again from the late ‘19-22’ 145

Links to Late Antiquity period when Mediterranean wares were reaching Britain (Fernández 2014, 354-8).

contexts. Crucially, this included E ware, representing the first identification of this product on the Iberian Peninsula. This discovery not only extends the known distribution of the ware but demonstrates that the model of distinct phases of Mediterranean and Continental importation to Britain – as outlined by Campbell – is not appropriate when considering the whole Atlantic. E ware was not only exported northward to Insular markets but was also transported southward toward Galicia – presumably with other commodities – at a time when Mediterranean wares were still arriving. The ware represented nearly half of the total coarseware from these late contexts, comprising 44 individual vessels.

Like the DSPA, the presence of E ware indicates commodities being transported south along the Atlantic seaboard from Bordeaux, and thereby reveals connections between the city and the Galician port. If Thomas’ theory of E ware providing a proxy for the trade of Aquitanian wine in barrels was correct (see Sub-chapter 2.2.1), it might be that the East Mediterranean wine imported to Vigo in amphorae was supplemented by Gaulish supplies. A full investigation of the regional and international distribution of E ware has the potential to uncover deeper economic connections, which might not be otherwise attested archaeologically. The full implications of this discovery are considered in Chapter Seven.

As in Britain and Ireland, E1 jars were the most common form, representing 32 of these vessels. Seven E3 bowls and a single base of a possible E2 beaker were also identified (Fernández 2014, 355). In addition, two examples of the E4B pitcher were recorded (see Figure 6.14). These are particularly relevant given the significant quantity of the equivalent DSPA form, Rigoir 36. A possible E4 jug was noted from different context, but overall, no examples of E5 lids or the other rare forms, E6 and E7, were identified (Fernández 2014, 355).

In addition to the E ware, an exceptional collection of imported coarseware/cooking ware, lamps and unguentaria were identified in contexts 19-22, demonstrating the breadth of Mediterranean ceramics reaching Vigo in this late phase. This group included three vessels of African origin; the small number was noted to reflect the general lack of Tunisian coarseware exports by this date (Fernández 2014, 258-60).

Although the Vigo examples do not provide any new dating evidence for this ware, their discovery corresponds with the broad chronology for this ware as established by Campbell, which was seemingly reflected in the Place Camille-Jullian deposits. E ware is shown as characteristic of ‘Horizon C’ and contexts 19-22, therefore arriving at Vigo after c. 550, and consequently after the main

The rest of the coarseware was of East Mediterranean origin, further attesting to the shipments arriving at Vigo from the East and via Atlantic channels. A range of coarse/ cooking vessels of Levantine and Lycian origin were recognised, including cooking-jars and casseroles, as well as a collection of vessels, including cups, jugs and mortaria, identified as from the North of Syria (potentially Antioch) (Fernández 2014, 360-74). The largest group (24 vessels) was cooking ware of Aegean origin, which comprised jars, pans and lids. These were noted to not only represent the most easterly discovery of these products, but to exceed the quantity recorded from the major Mediterranean port of Marseille (Fernández 2014, 364). Of the 32 lamps identified from the various interventions at Vigo, the majority were again of East Mediterranean origin. This prevalence was magnified in the later contexts (Fernández 2014, 37684). Finally, a total of seven individual examples of East Mediterranean unguentaria were recorded from all of the excavations (Fernández 2014, 384-6). These reveal the importation of perfumed oils/unguents, potentially for liturgical purposes. Although the potential presence of imported coarsewares in Britain, and specifically at Tintagel, was downplayed by Campbell (see Sub-Chapter 4.2.3), the range of material at Vigo, in addition to fineware and amphorae, demonstrates that pottery of domestic function or associated with the transport of exotic commodities was circulating in the Atlantic to some degree. It may be that future analysis of British hitherto unidentified vessels might extend the range of pottery seen to be reaching Insular sites. Discussion: The Vigo assemblage

Figure 6.14. Selected E ware from Vigo – including examples of E3 and E4B. From Fernández 2014, Figure 186

146

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity mentioned, examples of imported pottery from the port city of Gijón in Asturias have been the subject of several publications, but the assemblage has remained relatively isolated (Alonso Sánchez and Fernández Ochoa 1988; Fernández Ochoa et al. 1992).

Beyond all of the Spanish and Portuguese assemblages considered in this chapter, the scale and diversity of the ceramic collection from Vigo proves that Britain was not the focus for Mediterranean shipment to the Atlantic between the fifth and seventh century. The quantity of LRC recorded from Vigo dwarfs the quantity recorded from western Britain and Ireland. The assemblage demonstrates significant parallels with the Insular pattern, in terms of the main types of imported fineware present – LRC, ARS and DSPA. However, these associations are largely valid only for a particular phase of activity at Vigo, specifically the later-fifth century to the mid-sixth century. Within this timeframe the ceramic imports to Vigo – showing an increasing focus towards the East Mediterranean, with rising levels of LRC, LRA1 and some LRA2 – can be closely aligned with the Insular imports. A slight difference is the high proportion of ARS at Vigo within this period, which is relatively less common in Britain.

The majority of these finds were made in the ‘Termas de Campo Valdés’ bath complex, in the Cimadevilla/ Cimavilla district of the city, which was originally excavated in the early-twentieth century (Fernández Ochoa and García Díaz 1995, 279). More recent excavations – conducted in the city between 1990 and 1994 – revealed occupation had continued at the complex during the fifth and sixth century, although by this time the structure had lost its original function (Fernández Ochoa and García Díaz 1995, 281). Many of the imported ceramics were recovered from an early medieval layer sealing the disused hypocaust of the baths (Fernández Ochoa et al. 1992, 106). A few additional finds had previously been recovered in association with the town-walls [214] (Fernández Ochoa et al. 1992, 106).

After this phase, sites in Britain and Ireland appear to be cut-off from the Mediterranean supply, and do not receive the late forms of fineware identified at Vigo. The unexpected presence of E ware at Vigo reveals that this pottery was shipped southward from Aquitaine as well as northward to Insular sites. The later phase of ‘Continental’ imports identified in Britain was, therefore, not a separate Insular phenomenon, but a direct development of the earlier Atlantic system of exchange. The apparent emergence of an independent Atlantic network from the mid-sixth century will be discussed in Chapter Seven.

The total group of ARS from Gijón (an approximate minimum of six vessels) compares well to the British finds, being largely restricted to later-fifth/sixth century forms – principally ARS 99A-B, 91C and 104 – and containing none of the very late forms seen at Vigo (Fernández Ochoa et al. 1992, 109-17). Significantly, LRC seems to be much rarer than at the Galician and Portuguese sites, comprising a single rim-sherd identified as LRC 3E (Fernández Ochoa et al. 1992, 109-17).

The role of Vigo within wider Atlantic systems of exchange will also be considered further in the following chapter. Some general conclusions about the function of the site should, nevertheless, be noted. Fernández considered the pottery to demonstrate the intensity of commercial activity at the port of Vigo from the fourth century, and particularly between the sixth and early-seventh century (2014, 475). As discussed in Sub-chapter 6.2, the mid-fifth century witnessed an unprecedented expansion of East Mediterranean goods into the West. Following this reconfiguration of trade, the port at Vigo emerged as the primary entry point for Mediterranean goods reaching the Atlantic and as the main exporter of commodities to the East Mediterranean (Fernández 2014, 475). From here, imported ceramics– and therefore associated commodities such as grain, wine, oil or exotic items – were redistributed to other coastal sites in Galicia, as well as inland. Furthermore, it seems likely that Vigo came to function as a trans-shipment port, where goods were redistributed further north in the Atlantic – including to early medieval sites in Britain. This would certainly contradict the models of transport outlined in Chapter Two.

As mentioned, a few fragments of LRA1 and possibly LRA4 amphorae were also found in the town (Fernández Ochoa et al. 1992, 131-6). The largest group of imports, however, comprised Gaulish DSP. The 26 sherds documented by Ochoa et al. were matched to the Atlantic production group, and at the time, thought to represent its most westerly point of distribution (1992, 125-31, 138). A range of forms were identified, including Rigoir 1, 4 and 16, but there were no examples of mortaria Rigoir 29 or the pitcher Rigoir 36. The unusual prevalence of the bowl Rigoir 16 – typically less common among the Atlantic products, as witnessed at Vigo – was noted to provide a parallel for Britain, and specifically Dinas Powys, where this form is relatively well represented (Fernández Ochoa et al. 1992, 128). The collection also contained a base-sherd decorated with the stamp of a deer. This motif is characteristic of the Atlantic production group, where it is frequently part of a hunting-scene ‘medallion’; Rigoir’s catalogue included numerous examples of deer-stamps from Bordeaux, and the motif has been recorded at Place Camille-Jullian as well as Vigo (Rigoir et al. 1973, 248, 254-6; Fernández Ochoa et al. 1992, 128; Soulas 2012; Fernández 2014, 296; see Figure 6.15). It may be that vessels decorated in this way held some particular value or interest as an export.

6.3.3 Northern Spain In contrast to Galicia, much less evidence is currently available for the northern coast of Spain. The large quantity of material at Vigo is certainly not paralleled further north or east along the Asturian/Cantabrian Coast. As

The presence of imported pottery of Mediterranean and 147

Links to Late Antiquity western Gaulish origin was considered to confirm the direct supply of pottery to Gijón from Aquitaine but, furthermore, to connect the city to a wider network of maritime exchange operating in the Atlantic (Fernández Ochoa et al. 1992, 128, 136). As in Galicia, it is also likely that the pottery arriving at the port, and therefore any associated goods, were redistributed inland. Approximately eight kilometres south-west of Gijón a small group of imported pottery was identified at the Villa de Veranes [217] (Fernández Ochoa et al. 2006). This collection included fragments of the neck and handle of one LRA1 as well as fragments of DSP and imitation grey finewares (Fernández Ochoa et al. 2006, 141, 3). The assemblage was broadly dated to the later-fifth or early-sixth century, and associated with a phase of transformation at the site. Further south, two vessels of DSPA (Rigoir 1 and 4) were also recovered during excavations at the ‘late villa’ at Murias de Paraxuga [232] (Uscatescu et al. 1994, 189). To the east of Gijón there have been relatively few examples of Mediterranean imports. A possible example of LRC was mentioned by Prieto at Iruña (Veleia) in the province of Araba/Álava in the Basque Country/País Vasco (Prieto 2005, 173). At least one vessel of ARS-D (possibly ARS-D2) was also identified from this location by Filloy Nieva 1997, 334). This adds to the earlier possible example recorded by Jarrega Dominguez (Sub-chapter 6.2.3). Examples of ARS-D of fourth to early-fifth century date were also recorded at other sites in Álava in various articles (Filloy Nieva 1997, 334-5; Filloy Nieva and Gil Zubillaga 1997), although it seems likely that these were associated with supply from the Mediterranean via the River Ebro. A more promising Atlantic import is the fragment of the dish form ARS 61B that was found at the cave of Iruaxpe III [239] in the Basque Province of Guipuzkoa/ Guipúzcoa, although this again is unlikely to date far into the fifth century (Azkarate et al. 2003, 328). As mentioned, the discovery of late East Mediterranean amphorae in a marine context at Cabo de Higuer substantiates the shipment of Mediterranean pottery along the north coast of Spain into the fifth and possibly sixth centuries. The site is near to the town of Hondarribia, situated on the modern border with France and at the mouth of the Bidasoa River. A long chronological sequence of pottery was recovered from various underwater investigations carried out between 1961 and 1984. Among the latest finds identified were two vessels of LRA1B (reported as ‘Beltran 82’) and another of LRA2 (reported as ‘Beltran 77’) (Benito 1988, 131-2; Azkarate et al. 2003, 325). A number of syntheses have considered the distribution of DSP in this region, and specifically the possible presence of DSPA (Uscatescu et al. 1994; Filloy Nieva and Figure 6.15. Examples of DSPA with deer stamps from Gijón (top left), Vigo (middle left) and Place Camille-Jullian, Bordeaux (bottom left). After Fernández Ochoa et al. 1992, Figure 18; Soulas 2012, Figure 1; Fernández 2014, Figure 162.4

148

Ceramic imports to Atlantic Spain and Portugal in Late Antiquity arrival of the British imports are founded on a extremely restricted dataset, and must be revised.

Gil Zubillaga 1997; Azkarate et al. 2003; Prieto 2005; Fernández 2007). Uscatescu et al. recorded the presence of vessels of the Atlantic production group in the Basque Country at Peña Forua [233] (Vizcaya), Iruaxpe III and Cabo de Higuer (1994, 194). Interestingly, the latter site produced at least four examples of the mortaria form Rigoir 29, which is particularly well-represented in Britain (Uscatescu et al. 1994, 192).

The British and Irish assemblage does appear to display certain idiosyncrasies – specifically a raised proportion of LRA2 and a higher occurrence of amphorae in general (see Appendix Maps H.5-6). The Insular assemblage, however, is not exceptional, but broadly corresponds to typical patterns of ceramic distribution on the Atlantic Seaboard, specifically between c. 475 and 550. This period witnessed a major expansion of East Mediterranean trade into the Atlantic, resulting in the supply of LRC, LRD and East Mediterranean amphorae to a range of sites in regions equating to Lusitania and Gallaecia and onto the northern coast of Spain. This was supplemented by a renewed phase of late Vandal, North African exportation, as well as late Iberian production. Although wine and oil have typically be seen as the primary imports from the Mediterranean, the high proportion of finewares at these Iberian sites suggests that shipments of grain might have been an important factor.

In addition, possible examples of DSPA were noted at the cave of Abauntz [237] into Navarre/Navarra, and to the south-west at Clunia [273] in Castile and León/Castilla y León. The latter example is unconfirmed, but would represent the furthest inland discovery of a sherd of the Atlantic group (Uscatescu et al. 1994, 195). The publication by Azkarate et al. was much more circumspect, concluding that DSPA was only present in the Basque country at Cabo de Higuer and Iruaxpe III (2003, 8). Further south and into Álava, the reported presence of DSPA has likewise been doubted. Filloy Nieva concluded that although DSP was present at sites in the province, this was limited to the Provençal and Languedoc productions, and could therefore be connected with Mediterranean supply (1997, 335). Nevertheless, the growing awareness of DSPA at sites along the north coast clearly demonstrates a connection with Aquitaine, and suggests the marine transportation of commodities south and east from Bordeaux, which could potentially be connected to a Late Antique trade in wine – a conclusion also reached by Fernández Fernández (2014, 439). This certainly contradicts the initial statement by the Rigoirs that DSPA was not present on the Iberian Peninsula (1973, 207).

This Atlantic system was characterised by the marine supply of coastal locations – from where material was redistributed inland, as part of regional networks of exchange. The port at Vigo came to prominence as part of this system, particularly from the sixth century when it became the primary point of importation to northwestern Spain. It is also likely that Mediterranean commodities were redistributed from Vigo further north in the Atlantic. The excavations at Vigo produced a volume and range of material far in excess of any other Atlantic site, and which, specifically, dwarfs the Insular collection of finewares. In contrast to most of the sites discussed, the excavations at Vigo also revealed an exceptional group of late-sixth and early-seventh century finds – which have not been identified in Britain. This indicates that although western Britain and Ireland may have been marginalised from Mediterranean supplies by the mid-sixth century, connections persisted between the Atlantic and the Mediterranean into the seventh century.

To date, no identifications of E ware have been recorded from sites along the north coast, although the presence of DSPA – in confirming connections with Aquitaine – suggests that this might be a distinct possibility. On the other hand, it may be that by the later-sixth century shipments were directed solely at Vigo, and that this later phase witnessed the disappearance of shorter-distance interactions along the Cantabrian coast. Nevertheless, future broadscale research on coarsewares in the Atlantic has the unique potential to reveal patterns of exchange into the seventh and even eighth centuries.

The sites and assemblages considered in this chapter demonstrate that the presence of Mediterranean imports of fifth and sixth century date in Britain can no longer be taken as evidence of exceptional connections between Britain and the Byzantine East, whether founded on economic or diplomatic motives. Instead, the British finds must be considered within their wider Atlantic context. A new model of Late Antique exchange in the Atlantic will be presented in Chapter Seven, focusing on shorter-distance interactions rather than long-distance, direct transport. The emergence, development and ultimate fate of this Atlantic system will be considered.

6.4 Conclusions This review of imported pottery in Atlantic Spain and Portugal has revealed a vastly different picture of ceramic exchange in the Atlantic to that typically presented in publications on the British imports (Chapter Two). It is clear that the Insular finds are certainly not isolated on the Western Seaboard, but represent the northern extent of a major Atlantic distribution (see Appendix Maps H.1-8). It is also apparent that there has been a long history of research on these wares in Spanish and particularly Portuguese publications, which has never been fully integrated into the British literature. More recently, the Iberian distribution has been extended by an upsurge of information from north-western Spain. It is, therefore, clear that previous models for the 149

7 Discussion and conclusions: Britain in the Late Antique Atlantic 7.1 Introduction

to represent post-Roman imported vessels they were not included in his distribution maps. Typically, the North African amphorae found at these urban sites are of third to late-fourth/early-fifth century types, and cannot be equated with the later, fifth to seventh century African amphorae (Keay 35, 62, 61) identified at sites in the Atlantic and West Mediterranean (see Sub-chapter 3.2.4). The continuing use of the ‘Bv’ category for the post-Roman imports has complicated this distinction.

The findings discussed in Chapters Five and Six – and presented in Appendices H and K – cannot be seen to represent an exhaustive or definitive review of Atlantic sites with imported pottery of fifth to seventh century date, but reflect current understandings based on available, published accounts. As publications from this region increase, interpretations of the Insular material must continue to be revised. However, this assessment has revealed considerable parallels for the material found at early medieval sites in Britain and Ireland – and has, therefore, demonstrated that understandings of ceramic supply to Britain in the fifth, sixth and seventh century must now involve a wider consideration of patterns of distribution on the Atlantic Seaboard.

Bird’s review of ARS in Romano-British contexts did not record any forms that were necessarily fifth century imports, suggesting that the trickle of African fineware imports stopped until the arrival of new forms in western regions in the late-fifth and, particularly, sixth century (Bird 1977). More recently, sherds of ARS-C have been identified at Shadwell in London, but the identified form – Hayes 50/50A – is again of third/fourth century date (Douglas et al. 2011, 177-9). As noted, a North African amphora probably dating to the first half of the fifth century was found at this site. Three small body-sherds of LRA3 were also found; it was not clear if these belonged to the earlier, one-handled type or the two-handled type that characterises the post-Roman imports, but an earlier date was thought likely (Williams 2011b, 80-1).

This concluding chapter will summarise the broad patterns of ceramic distribution presented in Chapters Four to Six, in order to question earlier models of supply to Britain – as outlined in Chapter Two – and, in particular, to challenge the model of direct shipment from the East Mediterranean. This reappraisal will allow new interpretations of contact and exchange between Britain and Ireland, the Continent and the Mediterranean to be proposed. 7.2 Discussion: new models of contact and connection

The presence of ‘Palestinian’ amphorae has been used to suggest a general background of East Mediterranean amphora importation to late Roman Britain (Campbell 2007, 125). As these are largely fragmentary, the specific types are rarely fully identified. These have frequently been described as late-fourth/early-fifth century in date, but following the review of equivalent finds in the Atlantic these can be potentially assigned to the first few decades of the fifth century. A ring-handled amphora (LRA4?) recovered at Billingsgate [*C] in London, for example, has been considered to date to the first-half of the fifth century (Marsden 1980, 80-1). This has been shown in Appendix H.7 for reference.

7.2.1 Late Roman connections in the Atlantic It is apparent that the later-fifth and sixth century witnessed the unprecedented supply of East Mediterranean imports – characterised by new types of amphorae and fineware – to western Britain and Ireland. There remains, however, a level of uncertainty regarding the first half/ three-quarters of the fifth century, and the possible continuation of late Roman patterns of importation. As discussed, the Mediterranean imports found in western Britain have rightly been interpreted to reflect a new, separate ‘dynamic’ in exchange emerging from c. 475. As the Insular finds must now be seen as inextricably connected to patterns of importation to the Atlantic, the question of antecedents to this post-Roman phase can be reconsidered – and specifically, whether western Britain was completely ‘cut-off’ from external supply in the decades after AD 410.

North African and ‘Palestinian’ amphorae (possibly including sherds of a ‘bag-shaped’ amphora) have been found in late Roman contexts at Exeter (Holbrook and Bidwell 1991, 218). This evidence, together with the discovery of North African and Gazan amphorae from post-Roman deposits at Bantham – and the possible early date of the assemblage – has raised the possibility of uninterrupted Atlantic connections to the south-west of Britain (Bidwell et al. 2011, 113-4). The authors of the recent Bantham report proposed a model whereby the sites negotiating the exchange of tin changed between the later-fourth and fifth century – shifting from major urban centres such as Exeter to a more diffuse coastal pattern (Bidwell et al. 2011, 115).

As noted, LRA1 and LRA2 are not thought to have been imported to Roman Britain, unlike North African amphorae and East Mediterranean LRA3 – and, potentially, LRA4. Campbell has described the increasing identification of North African and ‘Palestinian’ amphora sherds within late Roman urban contexts (Campbell 2007, 1256). However, as these examples could not be confirmed 150

Discussion and conclusions: Britain in the Late Antique Atlantic lack of locations in Britain with Mediterranean imports of both late-fourth/early-fifth century date and the later-fifth/ sixth century types (LRA1, 2; LRC; late forms of ARS) (Campbell 2007, 126). An unprovenanced LRA1 was noted by Roberta Tomber from the Museum of London collections, but this was discounted as a ‘genuine London find’ (Tomber 2003, 107). The ARS sherds from Pevensey Castle also present a possible exception, although it is feasible that these vessels arrived via an alternative, Rhineland route (Fulford and Rippon 2011, 125). Thirdly, the apparent lack of Mediterranean finewares of earlier-fifth century date – although again the Pevensey finds might provide an exception. Finally, the evidence emerging from Atlantic Spain and Portugal suggests some continuity in exchange, but an overall reduction in importation from the Mediterranean, in the middle decades of the fifth-century (Fernández 2014, 128, 415-30). It is likely that this pattern will have been mirrored, and potentially exaggerated, at the northern reaches of this Atlantic system.

Although the import-points or consumption sites changed, they argued that the shipping-routes and exchanged commodities did not. The presence of relatively large quantities of the Gaulish fineware céramique à l’éponge at Exeter (c. 30 vessels) has also been taken to indicate late Roman contacts with western France (Bidwell 1979, 188; Bidwell et al. 2011, 114). Along with Argonne ware and Mayen ware this has been recognised as one of the key ceramic indicators for late Roman trade with southern Britain (Fulford 1977). Fulford described its export from western France to Britain as ‘anticipating’ the post-Roman trade evidenced by E ware and DSP (1977, 46). British examples of céramique à l’éponge have usually been associated with the region around Civaux, in Vienne, Poitou-Charentes, where this ware is thought to have been produced (Tyers 1996, 144). This followed an initial suggestion by Raimbault, based on the quantity of material found in this area (Raimbault 1973, 205). However, Bordeaux was later suggested as a more likely origin for the majority of the vessels material found in southern Britain – including those found at Exeter – as well as those recovered in Brittany (Gaillou et al. 1980, 272). Like DSP in the fifth/sixth century, the distribution of this ware was suggested to provide a proxy for the trade in other commodities, principally Aquitanian wine, up the Atlantic coast (Gaillou et al. 1980, 274). Nevertheless, without the discovery of kiln-sites neither source could be confirmed.

7.2.2 An Atlantic Network: new understandings of importation and connection in the late-fifth and sixth century The later-fifth century saw the re-integration of western Britain into Atlantic systems of exchange, resulting in the arrival of later forms of ARS-D and, particularly, LRC and East Mediterranean amphorae. The long history of research on these finds in Britain has accentuated their significance as indicators of long-distance trade and contact. In particular, interpretations have centred on models of direct trade with the East Mediterranean, driven by eastern demand for mineral resources – particularly tin – or by diplomatic relations with the Byzantine state. The review in Chapter Two outlined the dual foundation for these models: the apparent isolation of the Insular finds and specific differences with typical patterns in the West Mediterranean. These factors suggested that the Insular finds were not redistributed from a western – or indeed, Atlantic – port.

Since then, additional research has revealed a greater degree of complexity within this ware and led to the identification of a variant group, characterised by ‘marbled’ decoration and principally comprising the form ‘Raimbault VI’ (Sireix and Convertini 1997; Sireix 2005). Kilns associated with this pottery have been identified approximately 21km west of Bordeaux at Vayres, and it seems that this production site may be the source of many of the examples of céramique à l’éponge found on the Atlantic coast and in Britain (Sireix and Convertini 1997; Sireix 2005, 248-9). The ware has also now been identified in significant quantities in third/fourth century contexts in Bordeaux itself (Sireix and Convertini 1997, 322). These new findings reinforce the presence of specific connections between the Bordeaux region/Gironde and the south of Britain in the fourth century and, therefore, provide a clear precedent for the export – and potentially trans-shipment – of ceramics in the fifth and sixth century. Direct associations between Bordeaux and south-western Britain – based on céramique à l’éponge, glass, DSPA and, now, E ware – are becoming increasingly clear.

The large quantities of equivalent pottery identified from western France (especially from Bordeaux), northern/ north-western Spain and Portugal certainly refute the first argument. Britain was clearly not an isolated destination for Mediterranean supply beyond the Straits of Gibraltar. The relatively high levels of LRC in south-western Britain have frequently been used to epitomise the unusual Insular prevalence of East Mediterranean pottery, but these finds can now be directly linked to the wider dissemination of this fineware into the Atlantic. The distribution of LRC presented in Appendix H.2 is markedly focused on western Iberia. At present, the Insular finds appear somewhat isolated, but it is probable that more information will emerge from western France. Nevertheless, the examples of this fineware in Cornwall, Devon and around the Bristol Channel can be seen to represent the northern reaches of a major Atlantic distribution, dominated by Vigo.

Despite this evidence for continuity in the orientation of Atlantic trading-patterns, certain factors suggest that the western British ‘post-Roman’ imports do represent a separate dynamic, and that there was some break in the supply of pottery to Britain after the early-fifth century. Firstly, the clear shift in the focus of Mediterranean imports from urban sites to fortified, hill-top centres and coastal ‘beachmarket’ sites. Secondly – as Campbell has observed – the

Limitations of the data available from many of the Continental sites have made overall quantities difficult to 151

Links to Late Antiquity in the data have hindered direct chronological comparisons between individual sites. Nevertheless, the broad discussions presented in Chapter Six demonstrate a general focus on East Mediterranean imports in Atlantic Iberia in the period equating with the arrival of these types into post-Roman western Britain – c. 475-550.

assess – particularly for the ARS – but the general distributions of the Mediterranean finewares (Appendices H.1-3) indicate that western Britain was not the focus for shipment. Most notably, the number of fineware vessels recorded from the Insular sites is dwarfed by the quantities from Vigo. As discussed in Sub-chapter 6.2.4, it has been increasingly argued that the supply of these finewares can be seen as an indirect proxy for the shipment of grain. It must be acknowledged that the focus on imported amphorae in Insular contexts (shown in Appendix Maps H.5-6) might reflect an emphasis towards finewares in earlier Iberian research – and that quantities from western Spain and Portugal will increase. Although the later-sixth and seventh century imported amphorae from Vigo were published in detail, we must await the full quantification of the ‘non-fineware’ material from the earlier phases at the site for a more precise understanding of amphora supply in the Atlantic. Nevertheless, it can be proposed that these variations in distribution also reveal underlying differences in demand or the differential targeting of cargoes. Specifically, the focus on LRA1 and LRA2 at Insular sites – reflected by the prevalence of LRA1 in South Devon – signals that demand for East Mediterranean wine, rather than grain, was the key factor driving supply to the northern edges of the Atlantic supply-routes.

As outlined in Sub-chapters 6.2.2-4, a decline in exports from North Tunisia – demonstrated by a drop in ARS – commenced before the Vandal conquests, but continued into the mid-fifth century. This was directly mirrored by an expansion of East Mediterranean wares into the West Mediterranean and, thereafter, into the Atlantic – reaching as far as Britain and Ireland by the later-fifth century. The East Mediterranean producers of wine, oil and fineware had profited by the development of the eastern markets and, as Reynolds observed, were well-positioned to ‘capture’ western markets by the fifth century (2010, 69). The increased presence of imitation products in western Iberia – together with the LRC – indicates the continuing demand for fineware across this period (Reynolds 2010, 65). The later-fifth and, particularly, sixth century witnessed the resurgence of Tunisian exportation, reflecting a renewal of Vandal economic output and trade in surpluses (Reynolds 2010, 69). This revival resulted in the distribution of later forms of ARS into the Atlantic, reaching certain coastal sites in the Visigothic and Suevic Kingdoms, south-western France, and, again, Britain. The Insular finds were largely unprecedented, but it should be noted that at many of the western Iberian sites neither the quantities of this later phase of ARS-D or the LRC ever equalled the laterfourth/early-fifth quantities of ARS. The absence of these late-fifth and sixth century Mediterranean imports at many Atlantic sites has been used to suggest abandonment or economic decline from the early-fifth century – although this might simply indicate that many communities ceased to be integrated into long-distance exchange systems.

Reappraising the Insular assemblage The second factor underpinning the model of direct shipment, the apparent exceptional character of the Insular import assemblage, requires further consideration. A major argument for the model of direct shipment to Britain was Fulford’s observation that East Mediterranean imports (LRC and East Mediterranean amphorae) outnumbered North African imports (ARS and African amphorae) at the post-Roman sites in western Britain. This indicated that the material was not redistributed from the West Mediterranean – where the reverse pattern had been documented (Fulford 1989, 3).

The fifth century expansion of East Mediterranean trade is certainly reflected in Late Antique ceramic assemblages recovered from sites in the West Mediterranean. However, East Mediterranean types as a whole do not frequently surpass the quantities of Tunisian imports. Reynolds’ comparative analyses of published western groups observed a general increase in East Mediterranean amphora imports to sites in the West – including Rome, Carthage and Tarragona – from c. 425-450 (Reynolds 2010, 86, Table 17a). He commented that in southern Gaul – Arles and Marseille – East Mediterranean imports even equalled or exceeded North African imports across this period (Reynolds 2010, 86, n.318; see also Bonifay et al. 1998, 377). As discussed in Sub-chapter 6.2, eastern goods continued to arrive into the West after this phase, although, by the later-fifth and, particularly, the first half of the sixth century, Tunisian supply had reasserted its primacy at most of these sites (Reynolds 2010, 101, Table 18, 21). Peacock’s examination of trends in amphora supply at Carthage revealed a peak of East Mediterranean types – principally LRA1, but also LRA3 and LRA4 – in the

Although the Insular data has increased considerably, this emphasis on East Mediterranean wares has remained, with the ratio for the fineware (LRC to ARS) being close to 2:1. The relative proportions of the Insular amphorae are less clear – complicated by uncertainty over the origin of ‘Bv’ and the unidentified types. However, discounting those identified solely as ‘LRA’, a minimum of 225 vessels have been specifically identified as East Mediterranean types, from an overall minimum of 322. The combined total of those identified as ‘Bv’, ‘thick-walled’ or as North African, 38, is relatively negligible. If, as was suggested in Subchapter 3.2.4, many of the amphorae recorded under the ‘Bv’ or ‘thick-walled’ classification can be matched to an Iberian origin, the proportion of North African imports will be even lower. This would, however, point to the importance of Atlantic, rather than East Mediterranean, connections. Patterns of importation in the Atlantic, as examined in Chapters Five and Six, are more complex, and limitations 152

Discussion and conclusions: Britain in the Late Antique Atlantic clearly had a unique position in Atlantic trade in the fifth, sixth and seventh century, but the role of other ports or coastal sites within this network may be illuminated by future research.

period c. 450-500/535, but the quantities never surpassed African vessels (Peacock 1984b, 118-9, Figure 33). At Marseille, East Mediterranean amphorae retained a significant, though secondary, proportion to Tunisian amphorae into the late sixth/early seventh century – although by this period LRC was much less common than ARS (Bonifay et al., 1998, 377, 387-90, Tables 16-7, 23-5; Reynolds 2010, 220, Table 18).

Campbell’s defining characteristics As discussed in Sub-chapter 2.3.5, Campbell summarised five specific features of the ‘Atlantic’ group of Mediterranean imports which distinguished it from typical patterns in the West Mediterranean. Each of these factors can be reconsidered, to some degree, by the evidence presented in Chapters Four to Six.

Certain sites along the Atlantic Seaboard witnessed both this expansion of East Mediterranean trade and the resurgence of North African exportation, although in the majority of cases this phase of contact with the Mediterranean did not last beyond the mid-sixth century. After this point much of the Atlantic – including western Britain and Ireland – became marginalised from Mediterranean systems. Only a few locations, notably Vigo and Bordeaux, have revealed evidence of later-sixth/seventh century ARS (Appendix map H.1.3), and only Vigo had LRC 10. The amphora evidence is more difficult to assess, but points to a similar decline in importation – outside of Vigo and Bordeaux – from the mid-sixth century. A marked focus towards East Mediterranean amphorae can be observed for the Atlantic region, while Tunisian amphorae are typically rare – in contrast with the West Mediterranean.

Firstly, the apparent lack of Gazan or Palestinian amphorae can be considered. Although the figures from Britain remain small, the identification of LRA4 at Bantham shows that this amphora was a (minor) factor of post-Roman importation to Britain. In general, LRA4 is relatively uncommon in the Atlantic, apart from at Bordeaux – although, as noted, this might reflect integration into a separate chain of supply from the West Mediterranean. The ‘Bag-shaped’ amphorae from Place Camille-Jullian are (currently) without an Insular parallel and, overall, the Atlantic amphora finds show a focus towards the northeast, rather than south-east, Mediterranean – which is likely to reflect the origin of these shipments.

As more data from the Atlantic emerges – and more precise understandings of the relative proportion of ceramic types become possible – these comparisons can be refined. Nevertheless, Fulford’s observation that the imports to Britain and Ireland betray different patterns of supply to those in the West Mediterranean can be upheld – eastern pottery is unusually well represented in Britain and Ireland. The qualification is that this disparity can now be seen to reflect patterns typical for the wider Atlantic region, rather than the exceptional nature of the Insular finds. The scale and variety of the imports to Vigo, particularly the East Mediterranean fineware, amphorae and coarseware, must indicate that the site had some key importance for this distribution.

Secondly, the disproportionately large quantity of LRA2 can be questioned. As discussed in Sub-chapter 4.2.3 this might be partially a construct of the quantification strategies, but nevertheless, remains an atypical feature of the Insular assemblage – particularly of the Tintagel pottery. This amphora type was, in contrast, relatively uncommon at sites in the West Mediterranean (Reynolds 2010, 107). That the most common amphora in Britain is LRA1 does, however, equate with the wider Atlantic – and aligns with the general pattern of East Mediterranean amphorae in the West. The Insular focus towards LRA1 seems to be increasing, and was certainly a defining feature of the South Devon assemblages reviewed in Sub-chapter 4.3. The identification of a variant ‘LRA1 type’ at High Peak that had previously been identified as LRA2 indicates that some of the earlier classifications should be reviewed. Finally, the observation that LRA2 was particularly well-represented at Vigo in the first half of the sixth century suggests that the Insular prevalence of this amphora can be directly tied to a specific phase of Aegean supply to the Galician port (Fernández 2014, 344; see Sub-chapter 6.3.2).

The pattern of importation to Atlantic France will be clarified by future research, but there would appear to be some points of difference from the Insular assemblage and western Iberian groups. In particular – as stated by Bonifay following observations by Fernández Fernández – the focus on LRA4 and the prevalence of ARS over LRC at Place Camille-Jullian suggests that there may have remained some degree of overland supply to south-west France from the Mediterranean (Bonifay 2012, 256). Nevertheless, the presence of DSPA at Vigo – which continued to arrive after the ware stopped reaching Britain – certainly indicates direct connections between the two ports over a sustained period (Bonifay 2012, 256). Furthermore, the considerable quantity of East Mediterranean pottery at Bordeaux attests to continuity of importation via Atlantic channels to Roman, Visigothic and, after 507, Frankish controlled Aquitaine. The scatter of parallel material up the western coast of France is likely to indicate, as in Britain, redistribution within regional exchange systems. Bordeaux

The low proportion of ARS to LRC was also raised as a distinguishing factor, although the relatively small number of fineware vessels from Insular sites suggests that such conclusions should treated with caution. There is, currently, a lack of sites in the Atlantic with full comparable data, but it is clear that the region witnessed a marked increase in East Mediterranean fineware from the later-fifth century – equating with the phase of Insular importation. Certain sites, such as Braga, do show a higher concentration of ARS, but, overall, the high proportion of LRC in 153

Links to Late Antiquity also favourably compared (Reynolds 2010, 108-110). A point of divergence from general patterns in north-western Spain was the unusual prevalence of LRA2 in Britain; this was considered to demonstrate the ‘final stage of a long, direct journey from the Aegean’, again with a possible ecclesiastical dimension (Reynolds 2010, 110).

Britain and Ireland cannot be seen as exceptional, but as reflecting a wider ‘Atlantic’ pattern. The absence of East Mediterranean LRD was presented as another point of difference with fineware patterns in the West Mediterranean. Although the quantities remain relatively small, this ware was imported to western Iberia, including Vigo (see Appendix H.3).

Although the possibility of church involvement in the transport of these wares cannot be discounted, it seems unlikely that the Insular distribution can be accounted for by specific ecclesiastical contacts with Britain – particularly as the finds have been increasingly associated with secular, political sites (see Sub-chapter 2.2). Instead it is more likely that certain locations in Britain and Ireland (Tintagel, Bantham, Dalkey Island?) were integrated into systems of commercial exchange focused on western Iberia and south-western France. Reynolds suggested that Vigo might have been where ships ‘broke their journey en route to Britain’ (2010, 105). However, as Vigo was the focus for Mediterranean commodities in the Atlantic, the Galician/Suevic market can be considered as a key motivation for these shipments. The continuation of importation to Vigo and Bordeaux beyond the mid-sixth century reveals that limited connections between the Mediterranean and the Atlantic continued after the British supply had ceased. Even if trade with Britain was the impetus for this exchange, this clearly did not remain the case.

Finally, the restricted date-range of the imports was stated to contrast with patterns in the West. The established date for the phase of Mediterranean importation to western Britain and Ireland, c. 475-550, can be broadly reasserted, but this can now be seen to equate with a major phase of trade to the whole Atlantic Seaboard. By the mid-sixth century this period of exchange was waning, and most of the Atlantic sites – like those in Britain – did not receive Mediterranean vessels after this date. The mid-sixth century was a point of realignment and transformation of these Atlantic systems, but they did not wholly collapse – as evidenced by the continuing circulation of DSPA and E ware. Conclusions What remains to be questioned is the significance of the sites in Britain and Ireland within this system of exchange, and whether the Atlantic sites represent ‘stopping-points’ on routes targeted towards south-west Britain. As mentioned in Sub-chapter 6.2.4, Reynolds described how Atlantic sites such as Conimbriga and Braga took advantage of established routes supplying Britain. The presence of East Mediterranean and North African pottery in the Atlantic was observed to reflect the extension of general economic trends in the Mediterranean – as discussed above – but also specific, and potentially ecclesiastical, ties with the south-west of Britain (Reynolds 2010, 57, 69). This would indicate that deliberate contact with Britain provided the foundation for Atlantic exchange in the fifth and sixth century. Reynolds furthermore proposed that the late Roman supply of Tunisian amphorae to Roman sites in western Britain, including Exeter, provided a precedent for these trade-routes – again signalling that it was the ‘British market’ that drew ships into the northern Atlantic in the late-fifth and sixth century (2010, 109). It should be noted that these observations were made before the full publication of the material from Vigo. The scale of this assemblage, as well as the increasing volume of material from other Atlantic locations, demonstrates that find-spots of parallel Mediterranean pottery on the Western Seaboard can no longer be interpreted as outliers to the British distribution.

As there is significant evidence of fifth to seventh century imports on the Western Seaboard, and there are sites that share characteristics with the British assemblage, it can be argued that Britain was not the ultimate destination for Mediterranean ships in the Atlantic. It can, instead, be proposed that the Mediterranean imports reaching Britain or Ireland were redistributed from ports in north-western Spain (Vigo), Portugal or, potentially, Bordeaux. As such, connections between communities in western Britain and the emerging polities of post-Roman Gaul and Iberia would have been more significant than long-distance links with the Byzantine East. 7.2.3 Vigo: an ‘Atlantic emporium’ The excavations at Vigo and the detailed publication of its ceramic assemblage by Adolfo Fernández Fernández have dramatically shifted comprehensions of Late Antique exchange in the Atlantic, thereby necessitating new models of supply and connection. In his preface to the Vigo report, Bonifay summarised the implications of Fernández’s analysis – concluding that the volume and diversity of the pottery recovered from the Galician port had not only revised conceptions of the scale and nature of Atlantic trade, but had irrefutably invalidated the notion of direct links between the East Mediterranean and Britain (Bonifay 2014, xi). Much remains to be revealed about the history and development of this port, and future research by Fernández is likely to extend understandings of its regional and long-distance connections. However, his published thesis presented detailed conclusions about the function and significance of the site, which have major rel-

Reynolds did draw parallels between the Insular finds of ARS and those supplied to north-west Spain, suggesting that this might indicate the involvement of the same merchants, or the organisation of the same ‘ecclesiastical bodies’. These two regions are noted to have potentially supplied the same returning commodities, including ‘metals, hides or wool’ (2010, 105). The East Mediterranean imports supplied to Britain and Vigo were 154

Discussion and conclusions: Britain in the Late Antique Atlantic metals would appear particularly important, it is increasingly likely that Galician resources were also exploited in this period. Sánchez Pardo has, furthermore, proposed that increased Galician output by the mid-sixth century diminished the reliance on the Insular sources, accounting for the apparent break in contact between the Mediterranean and south-western Britain (2013, 115).

evance for interpretations of the Insular finds – although these can only be summarised here. Strong connections with Braga (Bracara Augusta) had developed from the fourth century, but from the laterfifth/sixth century Vigo emerged as the primary point of entry for imported to goods to the Suevic capital and its surrounding region (Fernández 2014, 476). Notably, however, Braga and other urban centres in north-western Iberia have not produced significant evidence of the very late Mediterranean imports seen at Vigo, indicating that this regional ‘dynamic’ did not surpass the mid-sixth century (Fernández 2014, 443, 449). From this point until the early-seventh century Vigo seems to have operated in relative isolation as a locus for long-distance exchange and redistribution.

The link between the imported pottery in Britain and tin – as proposed in Radford’s original publication – can be reaffirmed. This also confirms Campbell’s hypothesis that eastern demand for metal resources was the ‘primary motivation’ for post-Roman trade with Britain – although this exchange can now be positioned within an extensive model of indirect shipment, regional interaction and mediated cargoes (Campbell 2007a, 138; Campbell and Bowles 2009, 311). Campbell’s model had focused on demand from the Byzantine state, but Fernández, interestingly, suggested an additional link to routes heading further east. Lead was noted to be particularly prized at locations connected to the silk-road – including Antioch – or connected to routes leading to the Indian Ocean (Fernández 2014, 467). The possibility that communities in early medieval Britain and Ireland were actively, though indirectly, connected to a system of exchange reaching as far as the Indian sub-continent is certainly deserving of future consideration.

From the later-fifth century the port had expanded in ‘inter-regional’ significance as a result of traders seeking to exploit markets further north in the Atlantic – which can be associated with the contemporary expansion of eastern trade (Fernández 2014, 443). Fernández proposed a model of Late Antique trade characterised by two ‘blocks’ or zones of exchange operating on the Atlantic Seaboard – with Vigo positioned as the northernmost destination for Mediterranean shipments reaching the Atlantic and the southernmost point for vessels arriving from Britain or France (2014, 412, 475). Vigo operated as a central ‘hub’ within this network – the point where commodities arriving from the Mediterranean were traded for desirable ‘Atlantic’ resources (Fernández 2014, 475). Although these voyages may have been direct from the East Mediterranean or North Africa they were suggested to have stopped en route, thereby taking advantage of trading opportunities on the Iberian coast (Fernández 2014, 475).

Although it seems likely that traders of diverse origin arrived at Vigo, the recovery of significant quantities of East Mediterranean coarse and cooking ware was proposed to indicate that Vigo had a settled, stable community of Eastern – possibly Syrian – merchants (Fernández 2014, 476). Whether Eastern traders controlled supply further to the north, or whether sailors of British, French or Galician origin negotiated the exchange with sites in western France, Britain or even Ireland is unclear (Fernández 2014, 475). The pot of possible Galician origin found at Bantham provides a potential indicator of these connections, and presents a direction for future collaborative research. The presence of permanent Eastern community at Vigo would, however, obviate the need for an equivalent community at Tintagel or Bantham. Despite its obvious significance, Fernández was unable to identify clear contemporary textual references to the port at Vigo (2014, 476). Ultimately, he concluded that the site functioned as an Atlantic ‘emporium’ at the margins of Suevic – and subsequently Visigothic – political control and without direct ecclesiastical influence. This relative independence, combined with the natural advantages of the port, encouraged Vigo to develop as a centre for international exchange (Fernández 2014, 476). The port can certainly be equated with Hodges’ description of Channel/Baltic emporia as coastal ‘nodes’ of trading-networks, located at a ‘junction most favourable to an international function’ (Hodges 1982, 50-2). Hodges, however, emphasised the level of royal interest/influence in the development of these emporia, and it can be questioned whether the exchange conducted at Vigo was truly autonomous or wholly commercial (1982, 53-8).

Fernández highlighted the potentially mixed nature of cargoes arriving from the Mediterranean, which may have included oil, textiles, books, spices and perfumes (2014, 458-71). The amphorae reflected the key importance of East Mediterranean wine – although, as discussed, it might be that this was particularly relevant within systems of redistribution reaching further north into the Atlantic. Alum was also proposed as a possible import, transported in association with the LRC (Fernández 2014, 458). The continuation of contacts with North Africa – as demonstrated by the ARS – pointed to the considerable shipment of grain, while the rarer African amphorae might have carried oil or wine (Fernández 2014, 462). Nevertheless, Fernández concluded that the ‘primary function’ of this trading community was the acquisition and redistribution of resources from further north in Atlantic, which might include wood, Aquitanian marble, Baltic amber – and potentially slaves (2014, 448, 470-1, 476). Central to this system was the accumulation and trans-shipment of metals – principally tin, but also lead, copper and potentially gold and silver – to the East Mediterranean. Although the Insular sources of these 155

Links to Late Antiquity The reasons for the final cessation of Mediterranean connections with Vigo and Bordeaux in the early-mid seventh century – and therefore the final break between the Atlantic Seaboard and the Mediterranean – can also be seen to reflect broad, contemporary economic patterns and political upheaval. Fernández pointed to the Arab capture of Antioch as potentially significant, especially given his conclusion that the majority of the East Mediterranean consignments to Vigo had originated at Seleucia (2014, 445, 454). The excavation report from Place CamilleJullian describes a comparable decline in Mediterranean importation from the seventh century (Maurin 2012, 417). However, the evidence of E ware now suggests that this period did not mark the end of Bordeaux’s long-distance connections, which continued through the seventh century and, potentially, into the following century. Vigo, similarly, seems to have been increasingly oriented towards a system of exchange focused futher north in the Atlantic (Fernández 2014, 455).

Fernández’s assessment effectively presented Vigo as the major point of redistribution for commodities entering and exiting the northern Atlantic. According to this model sites in Britain would not have had any direct connection with North Africa or the Byzantine East. Instead, the supply of Mediterranean imports – wine amphorae and fineware – would have been mediated through Vigo, before arriving at primary entry-points such as Tintagel and Bantham for consumption and/or secondary redistribution. Despite the long distances indicated by the commodities, this system of Atlantic exchange was characterised by shorterdistance journeys and local interactions. It also seems likely that exchange was conducted at a variety of levels. Understandings of the regional networks responsible for the accumulation of mineral resources and the associated distribution of imported commodities – and the integration of these local structures into inter-regional systems – can also be highlighted as directions for future research. 7.2.4 The transformation and end of the Atlantic network

What is clear is that the ‘phases’ of Mediterranean and Continental importation to Britain and Ireland can no longer be separated. Direct connections between Bordeaux and Vigo, and between Vigo and the East Mediterranean, remained in operation into the later-sixth and early-seventh century. Within this period, E ware – the pottery considered to characterise the Insular ‘Continental’ phase – was transported south from Bordeaux as well as north and west to Britain or Ireland. Thomas’ description of these phases as ‘quite different commercial phenomena’ that were only connected in that they ‘sometimes dealt with the same Insular customers’ can be refuted. Given the likelihood of Atlantic trans-shipment at Vigo and Bordeaux it is feasible that the same communities of traders and sailors orchestrated exchange in the northern parts of the Atlantic across this juncture. Beyond the differences in the commodities arriving – Aquitanian wine and E ware rather than East Mediterranean wine-amphorae and fineware – residents of certain communities in Britain might have experienced little difference. However, the noticeable shift in the distribution of E ware from the pattern of Mediterranean imports demonstrates that many high-status sites in south-west Britain had lost their connection to this evolved system by c. 550. The separate ‘phase’ of Continental importation to Insular sites can, nevertheless, now be envisioned as an increasingly independent Atlantic exchange network, centred on Bordeaux, that was a direct development of a system of mediated contact with the Mediterranean.

The mid-sixth century witnessed a major reconfiguration in this Atlantic network. The majority of sites on the Atlantic Seaboard that had remained connected across the late-fifth and early-sixth century – notably those in western Iberia and northern Spain – ceased receiving supplies of Mediterranean pottery. Western Britain and Ireland, specifically, became marginalised from Mediterranean supply. Fernández suggested that this represented the end of a direct route between Vigo and Britain, which – following Campbell – he connected with the outbreak of the Justinianic plague (Campbell 2007a, 132; Fernández 2014, 452). However, it might be that this demonstrates a shift towards Bordeaux as the redistribution point for trade with the northern Atlantic. Campbell’s argument that the Byzantine merchants ‘killed off their clients’ at sites like Tintagel cannot be easily supported when the port at Vigo continued to operate and connections between Bordeaux and the Mediterranean remained in place (2007a, 132). These transformations can be more generally connected to a widespread decline in economic activity on the Atlantic Seaboard – outside of a few, specific locations. This might be connected to reorganisations in Mediterranean supply-routes – including the annona shipments – following the Justinianic re-conquests in North Africa and southern Spain, or associated with a broader reduction in Mediterranean trade. The end of Vandal control in Africa does not appear to have marked a major revival in Tunisian exports in the West, and although significant connections with the East remained – or even increased – a general picture of economic decline has been identified beyond the mid-sixth century (Fulford 1984b, 261; Wickham 2005, 712; Cunliffe 2008, 433; although see Reynolds 1995, 119). Certainly, Reynolds noted that beyond this point Iberian regions that were ‘not in Byzantine hands’ became ‘almost totally cut off from the Mediterranean world’ – although Vigo has now been established as a specific exception (2010, 69).

7.2.5 New models of Atlantic exchange Within the later-fifth and sixth century Fernández suggested that at least two distinct shipping routes were in operation within this northerly zone – a direct route from Vigo to Britain, and another along the northern coast of Spain to Bordeaux. This, he observed, would account for the scatter of Mediterranean pottery on the north coast of Spain and for discrepancies between the Insular finds and those from Bordeaux (Fernandez 2014, 439). Certainly, the presence of Mediterranean imports at locations on the 156

Discussion and conclusions: Britain in the Late Antique Atlantic it can be proposed that general models of networkstructure developed from complex network theory have potential value for understanding this evolving system of Late Antique/early medieval exchange. In particular, the Atlantic distribution of these ceramic classes can be convincingly equated with the theorised ‘small-world’ and ‘scale-free’ models of interaction (see Figure 7.1). These were effectively used by Sindbæk to visualise networks of Viking trade and communication (2007). Brughmans has recently provided a useful summary of their development and practical application for archaeology (2013, 643-48).

Cantabrian/Asturian coast – Gijón, Cabo de Higuer – indicates the shipment of commodities along this routeway. The comparable distribution of DSPA (see Appendix H.4) similarly reflects return journeys between Bordeaux and Vigo. The presence of E ware at Vigo – which can also now be proposed to directly signify exports from Bordeaux – demonstrates the continuation of this route into the latersixth and seventh century. Although the ceramic evidence shows clear associations between the Vigo assemblage and the Inuslar finds, is it possible that ships sailed directly from the Rías Baixas – passing A Coruña – and continued ‘non-stop’ on a northerly route to coastal trading-settlements in Cornwall and Devon? Harris – citing Sean McGrail’s review of sea-faring techniques on the Atlantic Seaboard – argued that ships reaching western Britain would not necessarily have had to adopt a ‘land-hugging’ approach (2003, 149). Certainly, McGrail had concluded that contemporary sailors would have been able to navigate away from the coast and presented the viability of a direct route between Brittany (Ushant) and south-western Britain, but he had also specifically indicated the operation of a coastal route between the Gironde and northern France (McGrail 1983, 319-22). Naveiro Lopez also highlighted the use of ‘offshore navigation’ by sailors plying the Late Antique Atlantic routes, but again pointed to the importance of intermediate ports – specifically Bordeaux (1991, 132-5).

These popularly utilised models have been used to explain the relatively rapid dissemination of objects or information across long-distances where individuals/communities do not have direct knowledge of each other – and, crucially, where no one has full knowledge of the entire network. The small-world network is characterised by high-clustering (the majority of interactions operate locally) but with a low average ‘path length’ (it is possible to ‘cross’ this network in a relatively few interactions) (Sindbæk 2007, 61; Brughmans 2013, 643). In this way residents of a community in early medieval Devon might have some knowledge of trading activities at Bantham, but limited awareness of exchange conducted in the Atlantic at Bordeaux or Vigo – much less of connections to the East Mediterranean. They were, nevertheless, connected by a relatively small number of links to communities in the East and to the wider economic and social developments of the Mediterranean world.

Ultimately, it seems unlikely that ships took a direct route across the hazardous waters of the Bay of Biscay direct to Britain. The scatter of Mediterranean pottery on the western seaboard of France seemingly attests to these coastal routeways. It is also possible that the apparent differences between the pottery from Bordeaux and from sites in western Britain is a reflection of the chronology of the material at Place Camille-Jullian. As more evidence emerges from western France, these questions can be re-examined. A possible model of shipment along the Atlantic Seaboard has been suggested in Appendix I.1 – adapted from Naviero Lopez 1991, Figure 31; Reynolds 1995, Figure 174; Campbell 2007a, Figure 3; and Fernández 2014, Figure 236 – and based on observations presented in Chapters 4-6. This illustration can, however, only hint at the complex network of regional interactions and systems of supply and redistribution that would have underpinned the long-distance exchange.

The less common, long-distance connections within these networks developed either through occasional ‘weak ties’ or as a result of particular ‘nodes’ (termed ‘hubs’) being particularly – and increasingly – well connected (Sindbæk 2007, 61; Brughmans 2013, 644-5). The latter characterises a ‘scale-free’ network. Sindbæk noted that this type of network was ‘robust against random failure’ but vulnerable to collapse or modification if a hub was compromised – particularly if the network was very hierarchical (2007, 62). The loss of a poorly connected node (site, settlement) from the network would have little effect overall, but the loss of a hub (trading port, emporium) could have extensive implications (Sindbæk 2007, 62). Within the Viking ‘small world’ the loss of a single node might result in the

Exploratory models A preliminary intention of this project was to conduct network analysis on the amassed ceramic data, using an appropriate software programme. However, it became apparent that this would not be possible because of the extreme variability of the data – the results would reflect the regional focus of ceramic research or volume of excavations, rather than genuine patterns in distribution and contact. This can, however, be suggested as a useful method of analysis as more data from the Atlantic region becomes available, particularly for the amphorae. Nevertheless,

Figure 7.1. Small-world systems: one characterised by ‘weak ties’ and the other as a ‘scale-free’ network. From Sindbæk 2007, Figure 1

157

Links to Late Antiquity

Figure 7.2. Circuits of Medieval exchange in the Indian Ocean. From Abu-Lughod 1989, Figure 10

loss of ‘whole sections of the network’ or the abandonment of a particular port or route (Sindbæk 2007, 71).

necessarily direct, but as resulting from routes of transshipment through Indian Ocean and Red Sea ‘entrepots’ – and was potentially mediated by traders operating from Southern Arabia and East Africa (Tomber 2008, 147-74).

Although the applicability of these models has not been verified by testing, the Late Antique Atlantic systems can be broadly suggested to correspond to the theorised ‘smallworld’ model – whereby the majority of sites operated on the basis of local interactions, but were connected to regional centres or trading-ports (hubs) with long-distance connections. As only a small number of these hubs had a large number of intra-regional connections (potentially Lisbon, Tintagel, Bantham, Dalkey Island) or interregional connections (Vigo, Bordeaux) the network could be classed as ‘scale-free’. Thus, the network permitted the long-distance movement of material – despite no agent knowing every link – but was inherently vulnerable.

Systems of later medieval trade in the Indian Ocean have been visualised by Abu-Lughod within three ‘interlocking circuits’ – exchange between which was conducted under the ‘shared “control” of a set of political and economic actors’ (1989, 251-2, Figure 10; see Figure 7.2). Although geographically delineated rather than culturally defined, cultural influences were noted to have spread ‘most freely’ within each of these circuits. Within this system the port emporia functioned as ‘meeting places for merchants and emissaries’ from each zone, and as conduits for social change, including the spread of religion (Abu-Lughod 1989, 253).

This observation has potential value in considering the apparent break of routes between Vigo and Britain –and the Insular break with Mediterranean supply. Did the loss of a particularly well-connected ‘node’ such as Tintagel have a major impact on the overall orientation of the Insular routes? These models might also be particularly useful in considering the end of Mediterranean importation to Vigo, after which point Mediterranean supply to the rest of the Atlantic collapsed and links between the northern and southern parts of the Atlantic Seaboard disappeared.

Significantly, long-distance connections between northwest Europe and the Indian Ocean in the thirteenth century were presented in a corresponding model – characterised by eight ‘interlinked sub-systems’ within a single ‘overarching world-system’ (Abu-Lughod 1989, 32-3, see Figure 7.3). Trading links between cities at either end of this system were not considered to be direct, but ‘broken geographically’ – with intermediate locations serving as important centres of trans-shipment and redistribution (Abu-Lughod 1989, 33-5).

A final speculative model is suggested by the possible link between the Atlantic network and systems of trade reaching the Far East (Fernández 2014, 467). It is notable that Roman and Late Antique exchange between the Mediterranean and India has not been interpreted as

A parallel model can be conjectured for the Mediterranean and Atlantic (see Appendix I,2), with the East and West Mediterranean presented as separate circuits, as well 158

Discussion and conclusions: Britain in the Late Antique Atlantic

Figure 7.3. The eight circuits of the thirteenth century world system. From Abu-Lughod 1989, Figure 1

as – following Fernández’s model – the northern and southern zones of the Atlantic Seaboard. Vigo would therefore be positioned as a bridging-point between the two Atlantic circuits, within which regional interactions operated. This would certainly make sense in relation to the geography of the Atlantic coast and the practicalities of navigating these routes. The increasingly independent operation of the ‘northern zone’ beyond the mid-sixth century and its survival beyond the early-seventh century are particularly comprehensible as the continuing operation of a pre-existing circuit of exchange and contact. If the fifth and sixth century connections between Atlantic and the Mediterraenan can be extended further east, there is potential to define and incorporate additional circuits.

importantly, the assemblage from Tintagel would benefit from a comprehensive reassessment and full quantification and publication. This has considerable potential in developing understandings of the entire British assemblage, but might also reveal previously unrecognised connections between communities in southern or western Iberia and south-western Britain in the fifth and sixth century. The possible presence of Mediterranean coarseware in the assemblage should be resolved and, in particular, the origin of the ‘Bv’ amphorae – and any associated Insular types – could, potentially be established. In relation to this, the possible presence of ‘Galician’ pottery in the south-west of Britain should be a matter for future consideration. These investigations would be reliant on collaboration between those working in Britain, the Atlantic and the Mediterranean.

7.3 Future research and recommendations There is also considerable potential in developing understandings of E ware, both in relation to its production and distribution. If the source of this Insular category can be confirmed as Bordeaux – through concerted typological, chemical and petrological analyses – this would have considerable importance in comprehending early medieval contacts between south-western France, Britain and Ireland. In particular, future research in Bordeaux might allow the chronology of this ware to be refined – which would have enormous significance for sites in western Britain and, particularly, Ireland. The possibility that E ware was more extensively traded should also be appraised, including the prospect of additional finds from northern Spain.

This project has demonstrated that despite over eighty years of research, much remains to be understood about contact and exchange in the Atlantic between the fifth and eighth century. Some specific avenues for future research can be highlighted. There would be value in reviewing older identifications of the Mediterranean imports from sites in Britain and Ireland, following the refinements of the ceramic categories outlined in Chapter Three, and in light of the new data emerging from the Atlantic – particularly from Vigo. The Insular material is relatively small, but is highly dispersed and this would represent a considerable undertaking. Most 159

Links to Late Antiquity world – whether founded on diplomatic or ecclesiastical contacts or commercial motives. Instead, a model of indirect shipment has been presented, which aligns with current understandings of Atlantic trade emerging from Continental research. The very large quantity of material from Vigo has suggested that this site may have been operating as a redistribution point or ‘emporium’ on the Atlantic Seaboard. From this port, a settled community of traders oversaw the importation and trans-shipment of diverse commodities moving north and south along active Atlantic exchange-routes (Fernández 2014, 475-78). Confirming earlier interpretations, this exchange centred on the accumulation of mineral resources. The shipment of grain might have been an important factor in southern regions, but it can be argued that wine was of greater importance further north, and particularly to the communities of south-west Britain and Ireland. Finds of Mediterranean pottery from the Atlantic Seaboard are likely to increase, and it may be that other important sites will be highlighted as hubs of inter-regional trans-shipment or regional redistribution.

Finally, Campbell’s example of incorporating ceramic research with the study of glass should be followed and expanded into the whole Atlantic. The likelihood of glassvessel production at Vigo has major implications for understanding the distribution of glass in western Britain and Ireland – which might have been transmitted by the same channels as the pottery. The ceramic evidence is simply one surviving factor of a more complex network of exchange and contact. The Atlantic connections demonstrated by this project can be reassessed in relation to other categories of material and other modes of interaction. 7.4 Conclusions The evidence from western France, Spain and Portugal has – for the first time since Radford’s 1959 publication – allowed the imported pottery from western Britain and Ireland to be interpreted within its full Atlantic context. Wailes’ early assessment, that parallels for the Insular finds were not present on the Atlantic Seaboard, can finally be refuted. As discussed, this conclusion had considerable impact on subsequent research, and specifically led to the formation of models that emphasised direct links between Britain and the Byzantine East. The evidence presented in Chapters Five and Six has demonstrated that equivalent pottery is present at numerous sites along the Western Seaboard. The volume and diversity of this material reveals that Britain cannot be considered as the focus for distribution, and was unlikely to be the final destination of Mediterranean ships sailing into the Atlantic.

Contact between the Atlantic and the Mediterranean persisted into the seventh century, although Britain had been marginalised from these systems of supply by the midsixth century. From the sixth century Bordeaux developed a new relevance as a centre of Atlantic exchange. On the basis of new evidence from France, the origin of the Insular E ware can be proposed to be Bordeaux – or its immediate vicinity. The Insular finds can, therefore, be suggested as directly attesting exports from Aquitaine, which were shipped – potentially with cargoes of wine – north and west to Britain and Ireland, but also south to Vigo. The separate, though overlapping, phases of Mediterranean and Continental importation to Britain and Ireland must now be seen as the evolution of a single system, and specifically as the late survival of a ‘circuit’ of exchange focused on the northern Atlantic, and centred on Bordeaux.

The Insular distribution of this pottery is increasingly widespread, and the assemblage is potentially more diverse than previously recognised. Patterns of supply to Britain – and indeed the Atlantic – can no longer be determined by the particular characteristics of the Tintagel pottery. Deviations from contemporary patterns in the West Mediterranean can still be observed, but the Insular assemblage can now be shown to manifest characteristic features of a distinctive Atlantic system. The pottery assessed from South Devon can be closely aligned with material recovered at contemporary sites in Spain and Portugal, revealing a clear focus on East Mediterranean imports. Patterns of importation to western Britain and Ireland between the fifth and seventh centuries were directly related to broader economic and political developments on the Western Seaboard – specifically reflecting phases of integration into a persistent, but fluctuating, Atlantic network.

Britain was not an isolated destination for exchange within the Late Antique Atlantic, but was, instead, integrated into a complex, dynamic and evolving Atlantic network. The ultimate origin of the Mediterranean imports cannot be denied, but the ceramic evidence for direct contact between early medieval Britain and the Byzantine world can finally be discounted. Links between Britain and the East Mediterranean existed between the fifth and seventh centuries, but these were indirect and mediated through sites on the Atlantic Seaboard. The social or political connections underpinning this exchange remain to be fully understood, but this model presents a new foundation for understanding contacts between communities living along the Western Seaboard.

The Insular finds are neither isolated nor exceptional, and can, therefore, no longer be used as evidence of special ties between western Britain and the Byzantine

160

Bibliography 8.1 Ancient sources

Amiel, C. and Berthault, F. “Les amphores du Bas-Empire et de l’Antiquité tardive dans le sud-ouest de la France.” Aquitania 14 (1996): 255-263.

Adomnán of Iona, Life of Columba, trans. A.O.R. Anderson and M.O. Anderson, Oxford: Oxford University Press, 1991.

Armstrong, P. “Trade in the East Mediterranean in the 8th century.” In Mango (ed.) 2009: 157-178.

Gregory of Tours, The History of the Franks, trans. L. Thorpe, Harmondsworth: Penguin, 1974.

Azkarate, A., Núñez, J. and Solaun, J.L. “Materiales y contextos cerámicos de los siglos VI al X en el País Vasco.” Anejos del Archivo Español de Arqueología 28 (2003): 321-370. Available at: http://www.ehu.es/gpac/ portal/index.php/eu/agerkariak/117-catlibros/263azkarate-a-nunez-j-solaun-j-l-2003-materiales-ycontextos-ceramicos-de-los-siglos-vi-al-x-en-el-paisvasco-anejos-del-archivo-espanol-de-arqueologiaxxviii-pp-321ss.html [accessed 04.01.15].

Hydatius, “Chronicle.” In The Chronicle of Hydatius and the Consularia Constantinopolitana, trans. and ed. R. W. Burgess, Oxford: Clarendon Press, 1993. Leontius, Bishop of Neapolis, “Supplement to the Life of St. John the Almsgiver.” In Three Byzantine Saints, trans. and eds E. Dawes and N. H. Baynes, Oxford and London: Mowbray, 1977.

Barraud, D. “Les Fouilles de la Place Camille-Jullian et l’exploration archéologique de Bordeaux.” In Maurin (ed.) 2012: 9-11

Procopius, The Secret History, trans. G.A. Williamson, London: Penguin, 1966.

Barraud, D. and Maurin, L. “Bordeaux au Bas-Empire. De la ville païenne à la ville chrétienne (IVe-VIe siècles).” Aquitania 14 (1996): 35-53.

8.2 Bibliography Abu-Lughod, J.L. Before European Hegemony. The world system A.D. 1250-1350. New York and Oxford: Oxford University Press, 1989.

Barrowman, R.C., Batey, C.E. and Morris, C.D. Excavations at Tintagel Castle, Cornwall, 1990–1999. London: Society of Antiquaries of London, 2007.

Agate, A., Duggan, M., Roskams, S. and Turner, S. “Early medieval settlement at Mothecombe, Devon: the interaction of local, regional and long-distance dynamics.” The Archaeological Journal 169 (2012): 343-94.

Bass, G.F. and van Doorninck, F.H. Yassi Ada: a SeventhCentury Byzantine Shipwreck. College Station: Texas A & M University Press, 1982. Batey, C., Sharpe, E., Thorpe, C., “Tintagel Castle: archaeological investigation of the steps area 1989 and 1990.” Cornish Archaeology 32 (1993): 47-66.

Alcock, L. Dinas Powys: an Iron Age, Dark Age and early medieval Settlement in Glamorgan. Cardiff: University of Wales Press, 1963.

Bélier, A-C. “Note sur la céramique E (E-ware).” In Aquitaine and Ireland in the Middle Ages. ed. J-M. Picard (Dublin: Four Courts Press, 1995): 261-3.

Alcock, L. Economy, Society and Warfare among the Britons and Saxons. Cardiff: University of Wales Press, 1987.

Beltrán, M. Guía de la Cerámica Romana. Zaragoza: Libros Pórtico, 1990.

Alcock, L. Cadbury Castle, Somerset: the Early Medieval Archaeology. Cardiff: University of Wales Press, 1995.

Benito, A.M. “Ceramicás del yacimiento submarino del cabo de Higer (Hondarribia).” Munibe 40 (1988): 12363.

Allan, J., Dawson, D. and Kent, O. “Post-Roman Pottery.” In Glastonbury Abbey: Archaeological Excavations 1904-1979. R. Gilchrist and C. Green (London: Society of Antiquaries of London, 2015), 250-93.

Berthault, F. 1997. “Montcaret.” In Carte archéologique de la Gaule 24.1. ed. H. Gaillard (Paris: Académie des Inscriptions et Belles-Lettres 1997): 159-166.

de Almeida, R. “Which amphorae carried the fish products from Tróia (Portugal)?” Rei Cretariae Romanae Fautorum, Acta 43 (2014): 653-61.

Berthault, F. “Les amphores de la place Camille-Jullian à Bordeaux.” Aquitania 16 (1999): 251–93. Berthault, F. “Amphores tardives dans le sud-ouest de la Gaule.” In Bonifay and Raynaud, 2007: 152-3, 363.

Alonso Sánchez, M.A. and Fernández Ochoa, C. “Cerámica Africana D con decoración Paleocristiana en la muralla de Gijon.” Cuadernos de Prehistoria y Arqueología Universidad Autónoma de Madrid 15 (1988): 339-377.

Berthault, F. “Un site fundamental pour l’étude des amphores.” In Maurin (ed.) 2012: 311-25. 161

Links to Late Antiquity Bidwell, P.T. The Legionary Bath-House and Basilica and Forum at Exeter, Exeter Archaeological Reports 1. Exeter: Exeter City Council and the University of Exeter, 1979.

sur la céramique Byzantine, eds V. Déroche and J. M. Spieser (Paris: Ecole française d’Athènes, 1989): 1746. Bonifay, M., Carre, M. B. and Rigoir, Y. 1998. Fouilles à Marseille: les mobiliers (1er-VIIe siècles ap. J.-C.). Paris, Errance.

Bidwell, P., Croom, A.T. and McBride, R. “The pottery assemblage.” In Reed et al. 2011: 93-117. Biek, L. “Tin ingots found at Praa Sands, Breage, in 1974.” Cornish Archaeology 33 (1994): 57–70.

Bost, J-P. “Les Monnaies.” In Maurin (ed.) 2012: 395-410. Bowman, A. “Post-Roman imports in Britain and Ireland: a maritime perspective.” In Dark (ed.) 1996: 97-108.

Bird, J. “African Red Slipware in Roman Britain.” In Roman pottery studies in Britain and beyond: papers presented to John Gillam, BAR supplementary series 30, eds J. Dore and K. Greene (Oxford: British Archaeological Reports, 1977): 269-77.

Briese, M.B. and Vaag, L.E. (eds). Trade Relations in the Eastern Mediterranean from the Late Hellenistic Period to Late Antiquity: The Ceramic Evidence, Halicarnassian Studies III. Odense: University Press of Southern Denmark, 2005.

Blázquez Martínez, J.M. “El comercio Hispano con el norte de África y el Oriente desde el comienzo de la Antigüedad hasta el siglo VIII.” In Vivre, produire et échanger: reflets Méditerranéens, eds L. Rivet and M. Sciallano (Montagnac: Monique Mergoil, 2002): 159-170.

Brughmans, T. “Thinking through networks: a review of formal network methods in archaeology.” Journal of Archaeological Method Theory 20 (2013): 623-62. Brulet, R., Vilvorder, F., and Delage, R. La céramique romaine en Gaule du Nord. Turnhout: Brepols, 2010.

Bonifay, M. “Observations sur les amphores tardives à Marseille d’après les fouilles de la Bourse (19801984).” Revue archéologique de Narbonnaise (1986): 269-301.

Campbell, E. “E Ware and Aquitaine: a reconsideration of the petrological evidence.” Scottish Archaeological Review 3, 1 (1984): 35-41.

Bonifay, M. “La céramique Africaine, un indice du développement économique.” Antiquité Tardive 11 (2003): 113-28.

Campbell, E. “The post-Roman pottery.” In Early Medieval Settlements in Wales AD 400–1100, eds N. Edwards, and A. Lane (Cardiff: University College of North Wales, Bangor and University College, 1988): 124–57.

Bonifay, M. Etudes sur la Céramique Romaine Tardive d’Afrique. Oxford: BAR Publishing, 2004.

Campbell, E. Imported goods in the Early Medieval Celtic West: with special reference to Dinas Powys. Unpublished PhD thesis, University of Wales, College of Cardiff, 1991.

Bonifay, M. “Observations sur la typologie des amphores Africaines de l’Antiquité Tardive.” LRCW1: Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean: Archaeology and archaeometry, BAR International Series 1340 (Oxford: BAR Publishing, 2005): 451-61.

Campbell, E. “New evidence for glass vessels in western Britain and Ireland in the 6th/ 7th centuries AD.” In Foy (ed.) 1995: 35-8.

Bonifay, M. “L’apport des amphores à la connaissance des échanges.” In Bonifay and Raynaud 2007: 96-102.

Campbell, E. “The archaeological evidence for external contacts: imports, trade and economy in Celtic Britain AD 400-800.” In Dark (ed.) 1996: 83-96.

Bonifay, M. “Post-Roman imports in the British Isles: material and place.” Antiquity 82 (2008): 1122-4.

Campbell, E. “Imported glass and Pottery.” In Laing and Longley 2006: 104-13.

Bonifay, M. “Les céramiques sigillées Africaines et Phocéennes Tardives.” In Maurin (ed.) 2012: 251-8.

Campbell, E. 2007a. Continental and Mediterranean Imports to Atlantic Britain and Ireland, AD 400–800, Council for British Archaeology Research Report 157. York: Council for British Archaeology, 2007.

Bonifay, M. “Préface.” In Fernández 2014: xi. Bonifay, M. and Pieri, D. “Amphores du Ve au VIIe s. à Marseille: nouvelle données sur la typologie et le contenu.” Journal of Roman Archaeology 8 (1995): 94110.

Campbell, E. 2007b. “La céramique E.” In Cunliffe and Galliou 2007: 86-8.

Bonifay, M. and Raynaud, C. “Échanges et consummation.” In “Antiquité tardive, haut Moyen Âge et premiers temps Chrétiens en Gaule Méridionale”, J. Guyon and M. Heijmans, Gallia 64 (2007): 93-161.

Campbell, E. 2007c. “Glass.” In Barrowman et al. 2007: 222-8. Campbell, E. Early Medieval Imported Pottery and Glass in the Atlantic Province AD 400-800, Digital Archive. University of Glasgow, School of Humanities, 2011. Available at: http://www.gla.ac.uk/schools/ humanities/research/archaeologyresearch/projects/ earlymedievalimports/ [accessed 02.02.12].

Bonifay, M. and Rigoir, Y.R. “Les dérivées des sigillées paléochrétiennes.” In C.A.T.H.M.A. 1986: 40-2. Bonifay, M. and Villedieu, F. “Importations d’amphores orientales en Gaule (Ve-VIIe siècle).” In Recherches 162

Bibliography Campbell, E. “Early Medieval Glass.” In Agate et al. 2012: 379.

Craw, J.H. “Excavations at Dunadd and at other sites on the Poltalloch estates, Argyll.” Proceedings of the Society of Antiquaries of Scotland 64 (1930): 111-27.

Campbell, E. and Bowles, C. “Byzantine trade to the edge of the world: Mediterranean pottery imports to Atlantic Britain in the 6th century.” In Mango (ed.) 2009: 297313.

Croom, A. Running the Roman home. Stroud: The History Press, 2011. Cunliffe, B. Facing the Ocean: the Atlantic and its peoples 8000 BC-AD 1500. Oxford: Oxford University Press, 2001.

Carandini, A., Anselmino, L., Pavolini, C., Saguí, L., Tortorella, S., Tortorici, E. Atlante I: Atlante Delle Forme Ceramiche, I. Ceramica fine Romana nel Bacino Mediterraneo (medio e tardoimpero). Enciclopedia dell’arte antica. Rome: Enciclopedia Italiana, 1981.

Cunliffe, B. Europe between the oceans: 9000 BC-AD 1000. New Haven and London: Yale University Press, 2008. Cunliffe, B. and Galliou, P. Les Fouilles du Yaudet en Ploulec’h, Côtes-d’Armor, Volume 3: du quatrième siècle apr. J.-C. à aujourd’hui. Oxford: Institute of Archaeology, University of Oxford, 2007.

Carter, G.E.L. “Second report of the Archaeological Section.” In The Transactions of the Devonshire Association 62 (1930): 119-120. C.A.T.H.M.A. La céramique du Haut Moyen Age en France méridionale: éléments comparatifs et essai d’interprétation, La ceramica medievale nel Mediterraneo occidentale (IIIe Congrès, Siena-Faenza, 1984) Firenze, 1986: 27-50. Available at: http://cathma. ass.free.fr/ [accessed 14.01.15].

Cuttler, R., Davidson, A. and Hughes, G. A corridor through time: the archaeology of the A55 Anglesey Road Scheme. Oxford: Oxbow, 2012. Dark, K.R. (ed.). External contacts and the economy of Late Roman and Post-Roman Britain. Woodbridge: Boydell and Brewer, 1996.

Cau, M.Á., Reynolds, P. and Bonifay, M. (eds). 2011a. LRFW1: Late Roman Fine Wares: Solving Problems of Typology and Chronology: a Review of the Evidence, Debate and New Contexts. Oxford: Archaeopress, 2011.

Dark, K. Britain and the end of the Roman Empire. Stroud: Tempus, 2000. Dark, K. “Early Byzantine mercantile communities in the West.” In Through a glass brightly: studies in Byzantine and medieval art presented to David Buckton, ed C. Entwistle (Oxford: Oxbow, 2003): 76-81.

Cau, M.Á., Reynolds, P. and Bonifay, M. 2011b. “Foreward.” In Cau, Reynolds and Bonifay (eds) 2011a: ix-xi. Cau, M.Á., Reynolds, P. and Bonifay, M. 2011c. “An initiative for the revision of late Roman fine wares in the Mediterranean (c. AD 200-700): The Barcelona ICREA/ESF Workshop.” In Cau, Reynolds and Bonifay (eds), 2011a: 1-10.

Delgado, M. “Contribuiçào para o estudo das cerâmicas romanas tardias do Médio Oriente encontradas em Portugal.” Cadernos de Arqueologia 2.5 (1988): 35-49. Delgado, M. Ceramicâs Romanas Tardias de Mértola originárias do Médio Oriente, Arqueologia Medieval, 1 (1992), 125-33. Available at: http://www.camertola.pt/ revista-arqueologia-medieval/arqueologia-medievaln%C2%BA-1 [accessed 28.10.14].

Chaidron, C. “Note sur la découverte de dérivées de sigillées paléochrétiennes dans le nord de la France.” Revue archéologique de Picardie 3 (3-4), (2008): 247-51. Chapelot, J. “Aux origins des châteaux et des bourgs castraux dans la moyenne et basse Charente: entre sources écrites et archéologie.” In Château, ville et pouvoir au Moyen Âge, eds A-M. Flambard Héricher and J. Le Maho (Publications du CRAHM: Caen, 2012): 81-156.

Delgado, M., Mayet, F. and Moutinho de Alarcão, A. Fouilles de Conimbriga, IV, Les Sigillées, Paris: Mission Archaéologique Française au Portugal and Musée Monographique de Conimbriga, 1975. Delgado, M., Fernández, A., Quaresma, J.C. and Morais, R. “Una aproximación a la terra sigillata Africana de Bracara Augusta (Braga, Portugal).” Rei Cretariae Romanae Fautorum, Acta 43 (2014): 671-80.

Collins, R. Early Medieval Europe 300-1000 (Third edition). Basingstoke: Palgrave Macmillan, 2010. Cool, H.E.M. “The parts left over: material culture into the fifth century.” In The Late Roman Transition in the North, eds T. Willmott and P. Wilson, BAR British Series 299 (Oxford: British Archaeological Reports, 2000): 47-65.

Diogo, A.M.D. and Trindade, L. “Ânforas e sigillatas tardias (claras, foceenses e cipriotas) provenientes das escavações de 1966/67 do teatro romano de Lisboa.” Revista Portuguesa de Arqueologia 2, (2) (1999): 83-95.

Cool, H.E.M. “The Vessel Glass from Bantham.” In Reed et al. 2011: 117-9.

Douglas, A., Gerrard, J. and Sudds, B. A Roman settlement and bath house at Shadwell: excavations at Tobacco Dock and Babe Ruth restaurant, The Highway, London. Shadwell Excavations Volume I. London: PreConstruct Archaeology, 2011.

Costen, M. Anglo-Saxon Somerset. Oxford: Oxbow, 2011. Coupry, M.J. “Circonscription d’Aquitaine: Gironde.” Gallia 272 (1969): 343-80. 163

Links to Late Antiquity Doyle, I.W. “The early medieval activity at Dalkey Island, Co. Dublin: a re-assessment.” The Journal of Irish Archaeology 9 (1998): 89-103.

(Galicia, España).” S.F.E.C.A.G, Actes du congrès du Langres (2007): 331-40. Fernández, A. “Rías Baixas and Vigo (Vicus Eleni).” In The Western Roman Atlantic Façade. A study of the economy and trade in the Mar Exterior from the Republic to the Principate, BAR International Series, 2162, eds C. Carreras and R. Morais (Oxford: British Archaeology Reports, 2010): 229-37.

Doyle, I.W. “Mediterranean and Frankish pottery imports in early medieval Ireland.” The Journal of Irish Archaeology 18 (2009): 17-62. Duggan, M. “The Pottery.” In Agate et al. 2012: 370-9. Duggan, M. High Peak, Sidmouth, Ceramic Assemblage Assessment, unpublished report for AC Archaeology and Devon County Council, 2013a.

Fernández, A. “Datos preliminares sobre las ánforas orientales tardías de dos yacimientos de Vigo (Galicia, Espana), con el ejemplo de un contexto de la primera mitad del s. VII.” Rei Cretariae Romanae Fautorum, Acta 42 (2012): 355-362.

Duggan, M. “Post-Roman Ceramics.” In Rainbird et al. 2013b: 38-9. Duggan, M. “A Late Roman 1 Amphora recovered off Cawsand, Plymouth Sound.” Cornish Archaeology 52 (2013c): 239-45.

Fernández, A. El comercio tardoantiguo (ss. IV-VII) en el Noroeste peninsular a través del registro cerámico de la ría de Vigo, Roman and Late Antique Mediterranean Pottery, 5. Oxford: Archaeopress, 2014.

Elliot, E.A.S. “On some earthworks in the South Hams probably concerned in the Irishmen’s raid.” Reports and Transactions of the Devonshire Association 33 (1901): 475–83.

Fernández, A. and Morais, R. “Terra Sigillata Bracarense Tardía.” in Céramicas Hispanorromanas II: Producciones Regionales, eds D. Bernal Casasola and A. Ribera I Lacomba (Cádiz: University of Cádiz, 2012): 131-74.

Empereur, J-Y., and Picon, M. “A propos d’un nouvel atelier de ‘Late Roman C’.” Figlina 7 (1986): 143-6.

Fernández Ochoa, C. García Díaz, P. and Uscatescu Barrón, A. “Gijón en el período tardoantiguo: cerámicas importadas de las excavaciones de Cimadevilla.” Archivo Español de arqueología 65 (1992): 105-49.

Empereur, J-Y., and Picon, M. “Les régions de production d’amphores impériales en Méditerranée orientale, Amphores romaines et histoire économique: Dix ans de recherche.” Actes du colloque de Sienne, 22-24 mai 1986. Publications de l’École française de Rome 114 (Rome: Publications de l’École française de Rome, 1989): 223-48.

Fernández Ochoa, C. and García Díaz P. Excavaciones arqueológicas en Cimadevilla (Gijón), Excavaciones arqueológicas en Asturias 1991-94. Oviedo: Servicio de Publicaciones del Principado de Asturias (1995): 277-285. Available at: http://ria.asturias.es/RIA/ handle/123456789/1765 [accessed 03.01.15].

Excavations.ie. 2014. Database of Irish Excavation Reports. Available at: http://www.excavations.ie/about/ [accessed 10.06.15].

Fernández Ochoa, C., Gil Sendino, F., Villar Calvo, A., Fernández Rodríguez, N., Alvarez Toledo, G. and Morán Fernández, O. “La villa romana de Veranes (Gijón, Asturias). Aportaciones preliminares sobre la transformación funcional del asentamiento en la tardía Antigüedad.” In Formas de ocupación rural en la Gallaecia y en la Lusitania durante la Antigüedad tardía y la alta Edad Media, eds C. Fernández Ochoa, M. Lopes, J. López Quiroga, Cuadernos de Prehistoria y Arqueología de la Universidad Autónoma de Madrid, 31–32, 2005–2006 (Madrid: Universidad Autónoma de Madrid, Departamento de Prehistoria y Arqueología, 2006): 137–94.

Fabião, C. “Las ánforas de Lusitania.” In Cerámicas Hispanorromanas: un estado de la cuestión, scientific eds D. Bernal and A. Ribera Lacomba (Cádiz: University of Cádiz 2008): 725-45. Available at: http:// www.exofficinahispana.org/ialfabetico/ia.php?sec=F [accessed 09.02.15]. Fabião, C. “O Ocidente da Península Ibérica no século VI: sobre o pentanummium de Justiniano I encontrado na unidade de produção de preparados de peixe da Casa do Governador daTorre de Belém, Lisboa.” Apontamentos de Arqueologia e Património 4 (2009): 25-50. Farley, M. and Little, R. “Oldaport, Modbury: a reassessment of the fort and harbour.” Proceedings of the Devon Archaeological Society 26 (1968): 31-6.

Filloy Nieva, I. “Distribución de mercanías en época Romana en Álava. El caso de los recipients.” Isturtiz 8 (1997): 321-357. Available at: http://www. eusko-ikaskuntza.org/es/publicaciones/colecciones/ cuadernos/articulo.php?o=9852 [accessed 04.01.15].

Fernandes, E. “A desagregação do comércio mediterrânico no Baixo Guadiana Tardo-Antigo: análise da terra sigillata africana e foceense tardia.” In Arqueologia em Portugal: 150 anos, eds J. M. Arnaud, A. Martins and C. Neves (Lisbon: Associação dos Arqueólogos Portugueses 2013): 873-81.

Filloy Nieva, I. and Gil Zubillaga, E. “Importaciones Galicas Tardias en Álava.” S.F.E.C.A.G, Actes du congrès du Mans (1997): 335-44.

Fernández, A. “Aportaciones al estudio de la Terra Sigillata Gálica Tardía Del grupo Atlántico de Vigo

Fleuriot, L. and Giot, P.-R. “Early Brittany.” Antiquity 51 (1977): 106-18. 164

Bibliography Ateliers de potiers médiévaux en Bretagne, Documents d’archéologie Française, 55, ed. F. Fichet de Clairfontaine (Paris: Éditions de la maison des sciences de l’homme, 1996): 55-61.

Fonseca, C.P. “A terra sigillata do fundeadouro de Tróia.” Revista Portuguese de Arqueologia 7, 1 (2004): 421-49. Fox, A. “Some evidence for a Dark Age trading site at Bantham, near Thurlestone, South Devon.” The Antiquaries Journal 35 (1955): 55-67.

Giot, P.-R. and Querré, G. “Le tesson d’amphore B2 de l’Ile Lavret (Bréhat, Côtes-du-Nord) et le problème des importations.” Revue archéologique de l’ouest (1985): 95-100.

Fox, A. “Holbeton: Mothecombe.” Reports and. Transactions of the. Devonshire Association 93 (1961): 79-80. Fox, A. “Tin ingots from Bigbury Bay, south Devon.” Proceedings of the Devon Archaeological Society 53 (1995) 11-23.

Girond, S. 4-6 Place Puy Paulin Bordeaux (Gironde): rapport final d’opération archéologique Haut Empire, Bas Empire, Haut Moyen Âge, Moyen Âge Classique. Balma: Hades, 2012.

Foy, D. (ed.). Le verre de l’Antiquité Tardive et du Haut Moyen Age: typologie, chronologie, diffusion, Huitième Rencontre de l’Association Française por l’Archéologie du Verre, Guiry-en-Vexin, 1993. [Guiryen-Vexin]: Musée Archéologique du Val d’Oise, 1995.

Griffith, F.M. “Salvage observations at Bantham Ham, Thurlestone, in 1982.” Proceedings of the Devon Archaeological Society 44 (1986): 39–58. Griffith, F.M. and Reed, S. “Rescue recording at Bantham Ham, Thurlestone, in 1997.” Proceedings of the Devon Archaeological Society 56 (1998): 109–31.

Foy, D. “Les verres des Ve-VIII siècles.” In Maurin (ed.) 2012: 335-71. Foy, D. and Hochuli-Gysel, A. “La verre en Aquitaine du IVe au IXe siècle: un état de la question.” in Foy (ed.) 1995: 151-76.

Grilo, C., Fabião, C. and Bugalhão, J. “Un contexto TardoAntigo do núcleo arqueológico da Rua dos Correeiros (NARC), Lisboa.” In Arqueologia em Portugal: 150 anos, eds J.M. Arnaud, A Martins and C. Neves (Lisbon: Associação dos Arqueólogos Portugueses, 2013): 849-57.

Fulford, M. “Pottery and Britain’s foreign trade in the later Roman period.” In Peacock (ed) 1977: 35-84. Fulford, M.G. “The Red-Slipped Wares.” In Fulford and Peacock 1984: 48-115.

Guériteau, A. 2006a. “Essai de classification typologique des céramiques du haut moyen âge du Nord de l’Aquitaine.” Aquitania 22 (2006): 329-34.

Fulford, M.G. 1984b. “The long distance trade and communications of Carthage c. A.D. 400 to A.D. 650.” In Fulford and Peacock 1984: 255-62.

Guériteau, A. 2006b. “Etude du mobilier ceramique du puits 1429.” In Des rivages garonnais de Burdigala au port de la mer: étude de 2000 ans de rapports entre Bordeaux et son fleuve. Présentations des opérations et synthèse des principaux résultats. Chantier archéologique “Bourse, Saint-Rémi, Jean Jaurès et Gabriel”, Bordeaux, Gironde, RFO de fouille préventive, Janvier 2002-Septembre 2003, F. Gerber (INRAP, SRA Aquitaine, 2006).

Fulford, M.G. “Byzantium and Britain: a Mediterranean perspective on Post-Roman Mediterranean imports in Western Britain and Ireland.” Medieval Archaeology 33 (1989): 1-6. Fulford, M.G. and Peacock, D.P.S. Excavations at Carthage: The British Mission Volume 1, 2. The Avenue Du President Habib Bourguiba, Salammbo: the pottery and other ceramic objects from the site. Sheffield: University of Sheffield for the British Academy, 1984.

Guériteau, A. “Etude de la céramique médiévale, Puy Paulin.” in Girond, 2012: 159-68.

Fulford, M. and Rippon, S. Pevensey Castle, Sussex. Excavations in the Roman Fort and Medieval Keep, 1993-95, Wessex Archaeology Report 26. Salisbury: Wessex Archaeology and University of Reading, 2011.

Guériteau, A. “Etude d’un dépotoir de céramiques du début du Haut Moyen Age.” In Une fenêtre archéologique sur un îlot antique de Bordeaux: le 17 rue du Hâ, Rapport d’opération archéologique préventive, J. Hénique (Hadès, SRA Aquitaine, 2013).

Gaillou, P., Fulford, M. and Clément, M. “La diffusion de la ‘céramique à l’éponge’ dans le nord-ouest de l’Empire Romain.” Gallia, 38, (2) (1980): 265-78.

Guériteau, A. forthcoming. “Etude d’une fosse-dépotoir du début du haut Moyen Age.” In L’Îlot Bonnac (Bordeaux), RFO de fouille préventive (INRAP GSO, SRA Aquitaine, forthcoming).

Gauthier, M. “Circonscription d’Aquitaine.” Gallia 41, (2) (1983): 445-471. Gerrard, J. The ruin of Roman Britain: an archaeological perspective. Cambrdge: Cambridge University Press, 2013.

Guitton, D. “A la recherche de temps perdu! A new approach to domestic ceramics of Late Antiquity (4th-6th centuries) in the heart of Aquitania Secunda (South West Gaul).” In Proceedings of the Ceramics and Atlantic Connections Conference, Newcastle University, March 2014, forthcoming.

Giot, P.-R. “Les Sites ‘Protohistoriques’ des dunes de Guissény’.” Annales de Bretagne 80, (1) (1973): 105-27. Giot, P.-R. and Fichet de Clairfontaine, F. “Quelques aspects de la production céramique en Bretagne.” In 165

Links to Late Antiquity Hodges, R. “Some early medieval French wares in the British Isles: an Archaeological Assessment of the Early French Wine Trade with Britain.” In Peacock (ed.) 1977b: 239-55.

Haith, C. 2001. “‘Herpes’.” In Medieval Archaeology: An Encyclopedia, ed. P.J. Crabtree (New York and London: Garland): 236-8. Handley, M. Dying on foreign shores: travel and mobility in the Late-Antique West, JRA Supplementary series 86. Portsmouth: Journal of Roman Archaeology, 2011.

Hodges, R. Dark Age Economics. The origins of towns and trade. London: Duckworth, 1982.

Harden, D. (ed.). Dark Age Britain. Studies Presented to E.T. Leeds. London: Methuen, 1956.

Hodges, R. “The date and source of E ware.” Scottish Archaeological Review 3, (1) (1984): 39-40.

Harden, D. 1956a. “The glass found at Tintagel.” In Radford, 1956: 70.

Holbrook, N. and Bidwell, P.T. Roman finds from Exeter, Exeter Archaeology Report, 4. Exeter: Exeter City Counc. and the University of Exeter, 1991.

Harden, D. 1956b. “Glass vessels in Britain and Ireland, AD 400-1000.” In Harden (ed.) 1956: 132-67.

Holbrook, N. and Thomas, A. “An early-medieval Monastic cemetery at Llandough, Glamorgan: Excavations in 1994.” Medieval Archaeology 49 (2005): 1-92.

Harris, A. Byzantium, Britain and the West: the Archaeology of Cultural Identity AD 400–650. Stroud: Tempus, 2003.

Hollinrake, N. “Dark Age Traffic on the Bristol Channel, UK: A Hypothesis.” The International Journal of Nautical Archaeology 36, (2) (2007): 336–43.

Harry, R. and Morris, C.D. “Excavations on the Lower Terrace, Site C, Tintagel Island 1990-94.” The Antiquaries Journal 77 (1997): 1-144.

Huggett, J.W. “Imported grave goods and the early AngloSaxon economy.” Medieval Archaeology 32 (1988): 63-96.

Hartgroves, S. and Walker, R. “Excavations in the Lower Ward, Tintagel Castle, 1986.” Cornish Studies 16 (1988): 9-30.

Husi, P. “La Touraine: espace socio-économique ou mosaïque d’entités isolées?” La céramique du haut Moyen Âge dans le Centre-Ouest de la France: de la chono-typologie aux aires culturellles (Tours: Archea, 2013): 25-39.

Hayes, J.W. Late Roman Pottery. London: British School at Rome, 1972. Hayes, J.W. A Supplement to Late Roman Pottery. London: British School at Rome, 1980.

Hutchinson, P.O. Unpublished diaries, digitised from the original manuscripts [online], 1846-94. Devon Record Office and the Peter Orlando Hutchinson Project. Available at: http://www.eastdevonaonb.org.uk/index. php?page=poh-transcripts [accessed 20.12.12].

Hayes, J.W. Handbook of Mediterranean Roman Pottery. Norman: University of Oklahoma Press, 1997. Hayes, J.W. “Introduction. The study of Roman pottery in the Mediterranean: 23 years after Late Roman Pottery.” In Saguì (ed.) 1998: 9-17.

Jackson, M., Zelle, M., Vandeput, L. and Köse, V. “Primary evidence for Late Roman D Ware production in southern Asia Minor: a challenge to ‘Cypriot Red Slip Ware’.” Anatolian Studies 62 (2012): 89-114.

Hayes, J.W. The Athenian Agora Volume XXXII. Roman Pottery: Fine Ware Imports. Princeton. New Jersey: American School of Classical Studies at Athens, 2008.

Jarrega Dominguez, R. Ceramicas finas Tardorromanas y del Mediterraneo Oriental en España. Estado de la cuestion. Madrid: Consejo Superior de Investigaciones Científicas, 1991.

Hencken, H. “An Irish Royal Residence of the 7th to 10th centuries A.D.” Proceedings of the Royal Irish Academy, Section C: 53, (1950): 1-247. Henig, M. “A garnet intaglio from Cefn Cwmwd.” In Cuttler et al. 2012: 156-8.

Jenkins, H.L. “Ancient camp at the mouth of the River Avon.” Devon Cornwall Notes and Queries 2 (1902): 20–3.

Hénique, J. “Bordeaux, 17 rue du Hâ.” In AdlFI: Archéologie de la France – Informations [online], Aquitaine (2007). Available at: http://adlfi.revues. org/7810 [accessed: 09.12.14].

Jones, R. “Ceramics: chemical and organic residue analysis of types Bi, Bii and Bv amphorae from Tintagel.” In Barrowman et al. 2007: 247-57.

Hill, P. Whithorn and St Ninian: The excavation of a Monastic Town 1984-91. Stroud: The Whithorn Trust and Sutton Publishing, 1997.

Jotischky, A. and Hull, C. The Penguin Historical Atlas of the Medieval World. London: Penguin, 2005. Karagiorgou, O. “LR2: a container for the military annona on the Danubian border?” In Kingsley and Decker (eds) 2001: 129-66.

Hill, S. Amphora residue sample 12A10, Unpublished report for SHIPS project, 2013. Hochuli-Gysel, A., Sireix, C., Soulas, S. and Berthault, F. Un ensemble clos du IVe-Ve siècle a Saint-Laurentdes-Combes (Gironde). In Foy (ed.) 1995: 177-85.

Keay, S.J. Late Roman Amphorae in the Western Mediterranean. A typology and economic study: the 166

Bibliography Catalan evidence, BAR International Series 196, parts i and ii (Oxford: British Archaeological Reports, 1984).

intervencíon practicada en el no 8 de la calle Damas.” Gallaecia 30 (2011): 127-33.

Kelly, A. “The discovery of Phocaean Red Slip Ware (PRSW) Form 3 and Bii ware (LR1 amphorae) on sites in Ireland – an analysis within a broader framework.” Proceedings of the Royal Irish Academy 110C (2010): 35-88.

López Pérez, M.C. and Vázquez Collazo, S. “La mesa y la villa romana de O Cantón Grande (A Coruña): approximacíon al yacimiento a través de la terra sigillata.” Gallaecia 26 (2007): 85-108. Lyne, M. Excavations at Pevensey Castle 1936 to 1964, BAR British Series 503. Oxford: BAR Publishing, 2009.

Kingsley, S. and Decker, M. (eds). Economy and exchange in the East Mediterranean during Late Antiquity. Oxford: Oxbow, 2001. Kirwan R. “Notes on the pre-historic archaeology of east Devon.” Report and transactions of the Devonshire Association for the Advancement of Science, Literature and Art 4 (1871): 641-53.

Mackensen, M. Die spätantiken Sigillata – und lampentöpfereien von El Mahrine (Nordtunesien), Studien zur Nordafrikanischen feinkeramik des 4. bis. 7 jahrhunderts. München: C.H. Beck’sche verlagsbuchhandlung, 1993.

Knight, J.K. The end of Antiquity: archaeology, society and religion AD 235-700. Stroud: Tempus, 2007 (Second ed.).

Mackensen, M. “Centres of African red slip ware production in Tunisia from the late 5th to the 7th century.” in Saguì (ed.) 1998: 23-39.

Labrouche, P. “La Céramique Commune du Ve au VIIIe Siècle” In Maurin (ed.) 2012: 293-310.

Magalhães, A.P. “Late sigillata from fish-salting workshop 1 in Tróia (Portugal).” Rei Cretariae Romanae Fautorum, Acta 42 (2012): 363-71.

Lafaurie, J. and Morrison, C. “La penetration des monnaies Byzantines en Gaule Mérovingienne et Visigotique du VIe au VIIIe siècle.” Revue numismatique 6, (29) (1987): 38-98.

Magalhães, A.P., Brum, P. and Pinto, I.V. “The significance of African cooking ware in Lusitania: the case of Tróia.” Rei Cretariae Romanae Fautorum, Acta 43 (2014): 701-8.

Lamboglia, N. “Nuove osservazioni sulla ‘terra sigillata chiara’ (II).” Rivista di Studi Liguri 29 (1963): 145-212.

Maia, M.G.P. “Cerâmica fina oriental de Tróia de Setúbal: ‘Late Roman C Ware.” In Actas do III Congresso Nacional de Arqueologia, Porto 1973 (Porto: Ministério da Educação Nacional, 1974): 333-341.

Lane, A. and Campbell, C. Dunadd: an early Dalriadic capital. Oxford: Oxbow, 2000. Langlotz, E. “Beobachtungen in Phokaia” Archäologischer Anzeiger (1969): 377-84.

Maia, M.G.P. “Contributos para as cartas de distribuição em Portugal da ‘sigillata luzente’ e da ‘Late Roman C Ware’.” In Actas das III Jornadas Arqueológicas da Associação dos Arqueólogos Portugueses, Lisboa 1977 (Lisboa: AAP, 1978): 295-307.

Laubenheimer, F. and Watier, B. “Les amphores des Allées de Tourny à Bordeaux. Aquitania 9 (1991): 5-39. López Pérez, M.C. “El comercio de Terra Sigillata en la provincia de A Coruña.” Brigantium (Boletín do Museo Arqueolóxico e Histórico da Coruña) 16 (2004): 5-541.

Mango, M.M. “Beyond the amphora: non-ceramic evidence for Late Antique industry and trade.” In Kingsley and Decker (eds) 2001: 87-106.

López Pérez, M.C. and Botella, V.T. “La vajilla fina de mesa de época romana en los solares no 10-12 de la calle Tabernas (A Coruña).” Gallaecia 29 (2010) 173-90.

Mango, M.M. (ed.) Byzantine trade, 4th-12th centuries: the archaeology of local, regional and international exchange, Papers of the thirty-eighth Spring Symposium of Byzantine Studies, St John’s College, University of Oxford, March 2004. Farnham: Ashgate, 2009.

López Pérez, M.C. and César Vila, M. “La importación de terra sigillata durante la etapa tardorromana en la fachada atlántica del Noroeste peninsular.” In Metodología de Análisis Aplicada a los Estudios de Cerámica Tardoantigua y Medieval de la Península Ibérica, eds M. Crespo Díez and R. Martínez Peñín (León: Lobo Sapiens, 2010): 41-52. Available at: https://www.academia. edu/8539709/La_importaci%C3%B3n_de_Terra_ Sigillata_durante_la_etapa_tardorromana_en_la_ fachada_atl%C3%A1ntica_del_Noroeste_peninsular_ Metodolog%C3%ADa_de_An%C3%A1lisis_ Aplicada_a_los_Estudios_de_Cer%C3%A1mica_ Tardoantigua_y_Medieval_de_la_Pen%C3%ADnsul a_Ib%C3%A9rica [accessed 09.12.14].

Marsden, P. Roman London. London: Thames and Hudson, 1980. Maurin, L. (ed.) Un quartier de Bordeaux du Ier au VIIIe siècle: les fouilles de la Place Camille-Jullian 19891990. Bordeaux: Ausonius, 2012. Mayet, F. and Picon, M. “Une sigillée phocéenne tardive (Late Roman C ware) et sa diffusion en Occident.” Figlina 7 (1986): 129-42. McGrail, S. “Cross-channel seamanship and navigation in the late first millennium BC.” Oxford Journal of Archaeology 2, (3) (1983): 299-337.

López Pérez, M.C. and Muiño Maneiro, X.A. “La terra sigillata en A Coruña: primeros resultados de la 167

Links to Late Antiquity O’Donnell, M. An Analysis of E ware in Ireland, unpublished MA thesis, University College Cork, 1984.

Melim de Sousa, É. “Contributos para o estudo da cerâmica Foceense Tardia (‘Late Roman C Ware’) no Municipium Olisiponense. Sua representatividade no contexto peninsular.” Conimbriga 40 (2001): 199-244.

Olson, L. Early Monasteries in Cornwall. Woodbridge: The Boydell Press, 1989.

Meyza, H. “Cypriot Red Slip: development of the ware (an attempt at refinement).” In Third International Congress of Cypriot Studies 16-20 April 1996, eds G.K. Ioannides and S.A. Hadjistyllis (Nicosia: Society of Cypriot Studies, 2000): 507-29.

Ó’Ríordáin, S.P. “The Excavation of a Large Earthen Ring-fort at Garranes, Co. Cork.” Proceedings of the Royal Irish Academy, Section C 47 (1942): 77-150. Ó’Ríordáin, S.P. “Roman Material in Ireland.” Proceedings of the Royal Irish Academy, Section C 51 (1947) 35-82.

Moorhead, S. “Early Byzantine copper coins found in Britain: a review in light of finds recorded with the Portable Antiquities Scheme.” In Ancient History, Numismatics and Epigraphy in the Mediterranean World, ed. O. Teken (Ege Publications: Istanbul, 2009): 263-274.

Orton, C. and Hughes, M. Pottery in Archaeology (second ed.). Cambridge: Cambridge University Press, 2013. Panella, C. “Annotazioni in margine alle stratigrafie delle Terme ostiensi del Nuotatore, Recherches sur les amphores romaines.” In Actes du Colloque de Rome (4 mars 1971), Publications de l’École française de Rome, 10. (Rome: École française de Rome, 1972): 69-106.

Morais, R. “From Oppidvm to Dives Bracara: the city trade through the amphorae.” In LRCW2: Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean: Archaeology and archaeometry, BAR International Series 1340 (Oxford: BAR Publishing, 2005): 55-67.

Peacock, D.P.S. 1977a. “Roman amphorae: typology, fabric and origins, Méthodes classiques et méthodes formelles dans l’étude typologique des amphores.” In Actes du colloque de Rome, 27-29 mai 1974, Publications de l’École française de Rome 32 (Rome: École française de Rome, 1977): 261-278.

Morais, R. and Fernández, A. “La producción de vajillas finas de Bracara Augusta.” Rei Cretariae Romanae Fautorum, Acta 43 (2014): 709-20.

Peacock. D.P.S. (ed.) 1977b. Pottery and Early Commerce: characterization and trade in Roman and later ceramics. London: Academic Press, 1977.

Munro, R. Ancient Scottish lake-dwellings or crannogs: with a supplementary chapter on remains of lakedwellings in England. Edinburgh: David Douglas, 1882.

Peacock, D.P.S. 1984a. “E ware and Aquitaine.” Scottish Archaeological Review 3, (1) (1984): 38-9. Peacock, D.P.S. 1984b. “The Amphorae: Typology and Chronology.” In Fulford and Peacock 1984: 116-40.

Naveiro Lopez, J.L. El comercio antiguo en el N.W. Peninsular. A Coruña: Museo Arqueolóxico e Histórico, 1991.

Peacock, D. and Thomas, C. “Class ‘E’ Imported Post-Roman Pottery: a suggested origin.” Cornish Archaeology 6 (1967): 35-46.

Nieto Prieto, F.J. “Algunos datos sobre los importaciones de cerámica ‘Phocaean Red Slip’ en la península Ibérica.” In Papers in Iberian Archaeology, eds T.F.C. Blagg, R.F.J. Jones and S.J. Keay, BAR International Series 193 (Oxford: British Archaeology Reports, 1984): 540-551.

Peacock, D.P.S. and Williams, D.F. Amphorae and the Roman Economy: an introductory guide. London: Longman, 1986. Pecci, A., Salvini, L. and Cantini, F. “Residue analysis of some Late Roman amphora coming from the excavations in the historical centre of Florence.” In LRCW3: Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean: Archaeology and archaeometry, Volume I, BAR International Series 2185 (Oxford: BAR Publishing 2010): 363-367.

Noble, G., Gondek, M., Campbell E. and Cook, M. “Between prehistory and history: the archaeological detection of social change among the Picts.” Antiquity 87 (2013): 1136-50. Nowakowski, J. and Thomas, C. Exavations at Tintagel Parish Churchyard, Cornwall, Spring 1990, Interim Report. Truro: Cornwall Archaeological Unit and Intitute of Cornish Studies, 1990.

Pédoussaut, L. “La céramique antique.” In Girond 2012: 134-59.

Nowakowski , J. and Thomas, C. Grave News from Tintagel, an account of the second season of archaeological excavation at Tintagel Churchyard, Cornwall, 1991. Truro: Cornwall Archaeological Unit and Intitute of Cornish Studies, 1992.

Penhallurick, R.D. Tin in Antiquity: its mining and trade throughout the ancient world with particular reference to Cornwall. London: The Institute of Metals, 1986. Pieri, D. “Les importations d’amphores Orientales en Gaule Méridionale durant l’Antiquité Tardive et le Haut Moyen Age (IVe – VIIe siècles après J.-C.). Typologie, chronologie et contenu.” S.F.E.C.A.G, Actes du congrès d’Istres, 1988 (1998): 97-106.

Nowakowski, J.A., Quinnell, H., Sturgess, J., Thomas, C. and Thorpe, C. “Return to Gwithian: shifting the sands of time.” Cornish Archaeology 46 (2007): 13–76. 168

Bibliography Pieri, D. Le Commerce du Vin Oriental à l’Époque Byzantine (Ve–VIIe siècles): Le Témoignage des Amphores en Gaule. Beyrouth: Inst. Français du Proche-Orient, 2005.

Quinnell, H. Trethurgy: Excavations at Trethurgy Round, St Austell: Community and Status in Roman and PostRoman Cornwall. Cornwall: Cornwall County Council, 2004.

Pieri, D. “Amphores de Méditerranée Orientale durant l’Antiquité tardive (IVe-VII S.): regions de productions et commerce vers la Gaule.” In Bonifay and Raynaud 2007: 149-52.

Radford, C.A.R. “Imported pottery found at Tintagel, Cornwall.” In Harden (ed.) 1956: 59-70. Rahtz, P. A., Woodward, A., Burrow, I., Everton, A., Watts, L., Leach, P., Hirst, S., Fowler, P. and Gardner, K. Cadbury Congresbury 1968–73: A Late/Post-Roman Hilltop Settlement in Somerset, BAR Reports British Series 223. Oxford: British Archaeological Reports, 1992.

Pinto, I.V., Magalhães, A.P. and Brum, P. “Ceramic assemblages from a fish-salting factory in Tróia (Portugal).” Rei Cretariae Romanae Fautorum, Acta 41 (2010): 529-37.

Rahtz, P., Hirst, S. and Wright, S. M. Cannington Cemetery, Britannia monograph series 17. London: Society for the Promotion of Roman Studies, 2000.

Pinto, I.V., Magalhães, A.P. and Brum, P. “Un depotoir du Ve siècle dans l’officine de salaisons 1 de Tróia (Portugal).” Rei Cretariae Romanae Fautorum Acta 42 (2012): 397-406.

Raimbault, M. “La céramique gallo-romaine dite ‘à l’éponge’ dans l’ouest de la Gaule.” Gallia 31, (1) (1973): 185-206.

Pinto, I.V., Viegas, C. and Dias, L.F. “Terra sigillata and amphorae from the Roman villa at Tourega (Évora, Portugal).” In Close Encounters: Sea-and Riverborne Trade, Ports and Hinterlands, Ship Construction and Navigation in Antiquity, the Middle Ages and in Modern Time, eds M. Pasquinucci, and T. Weski, BAR International Series 1283 (Oxford: BAR Publishing, 2004): 117-127.

Rainbird, P. “Oldaport and the Anglo-Saxon defence of Britain.” Proceedings of the Devon Archaeoogical Society 56 (1998): 153–64. Rainbird, P., Hughes, S., Allen, M. J., Duggan, M., Payne, N., Quinnell, H., Simmons, S. and Taylor, R. “Excavations at the Early Neolithic and Post-Roman Site of High Peak Camp, Otterton, East Devon.” Proceedings of the Devon Archaeological Society 71 (2013): 25-53

Piot, C. “La réutilisation des amphores: contribution à l’histoire économique et à la vie religieuse dans le sudouest de la Gaule.” Munibe: Antropologia-Arkeologia 53 (2001): 101-133.

RAMM. William Stewart Mitchell D’Urban (1836-1934), Royal Albert Memorial Museum and Art Gallery Website: Collectors [online], 2012. Available at: http://www.rammuseum.org.uk/collections/collectors/ william-stewart-mitchell-durban-18361934 [accessed: 20.12. 2012].

Pirault, L. “La céramique tardive à la fin du Bas-Empire à Nantes, Saint-Herblain et Rezé (Loire-Atlantique): état des connasissances.” S.F.E.C.A.G, Actes du congrès du Mans, 1997 (1997): 293-307. Pollard, S. “Neolithic and Dark Age settlements on High Peak, Sidmouth, Devon.” Proceedings of the Devon Archaeological Society 23 (1966): 33-59.

Ramos, A.C., Almeida, R. and Laço, T. “O complexo industrial da Rua Silva Lopes (Lagos). Uma primeira leitura do sítio e análise das suas principais problemáticos no quadro da indústria conserveira da Lusitânia meridional.” Setúbal Arqueológica 13 (2006): 83-100.

Prieto, M.Z. “Comercio y distribucíon de cerámicas Romanas en Asturias.” In Unidad y diversidad en el Arco Atlántico en época Romana, III Coloquio Internacional de Arqueología en Gijón, eds C. Fernández Ochoa and P. García Díaz (Oxford: BAR Publishing, 2005): 16389.

Ramos, A.C., Laço, T., Almeida, R. and Viegas, C. “Les céramiques communes du VIe du complexe industriel de salaisons de poisson de Lagos (Portugal).” In LRCW2: Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean: Archaeology and archaeometry, Volume I, BAR International Series 2185 (Oxford: BAR Publishing, 2007): 85-97.

Quaresma, J.C. Economia Antiga a partir de um centro de consumo Lusitano: Terra sigillata e cerâmica africana de cozinha em Chãos Salgados (Mirobriga?). Lisbon: Centro de Arqueologia da Universidade de Lisboa, 2012.

Randoin, B. “Essai de classification chronologique de la ceramique de Tours du IVe au XIe siècle.” Recherches sur Tours 1 (1981): 103-14

Quaresma, J.C. “Les contextes de Chãos Salgados, Mirobriga (Portugal): début du IIIe s.-premier quart du Ve s.” S.F.E.C.A.G., Actes du congrès d’Amiens, 2013 (2013): 385-408.

Reece, R. Excavations in Iona 1964 to 1974. London: Institute of Archaeology, 1981.

Quaresma, J.C. and Morais, R. “Eastern late Roman fine ware imports in Bracara Augusta.” Rei Cretariae Romanae Fautorum, Acta 42 (2012): 373-83.

Reed, S., Bidwell, P. and Allan, J. “Excavation at Bantham, south Devon, and post-Roman trade in South-west England.” Medieval Archaeology 55 (2011): 82-138. 169

Links to Late Antiquity Reynolds, P. Trade in the Western Mediterranean, AD 400700: The Ceramic Evidence, BAR International Series 604. Oxford: British Archaeological Reports, 1995.

Silvester, R.J. “An excavation on the post-Roman site at Bantham, south Devon.” Proceedings of the Devon Archaeological Society 39 (1981): 89-118.

Reynolds, P. Hispania and the Roman Mediterranean, AD 100–700: Ceramics and Trade. London: Duckworth, 2010.

Sindbæk, S.M. “The small world of the Vikings: networks in early medieval communication and exchange.” Norwegian Archaeological Review 40, (1) (2007): 59-74.

Rigoir, J. “Les sigillées paléochrétiennes grises et oranges.” Gallia 26 (1968): 177-244.

Sireix, C. “Bordeaux-Burdigala et la Bretagne romaine: quelque témoins archéologiques du commerce Atlantique.” Aquitania 21 (2005): 241-51.

Rigoir, J., Rigoir, Y. and Meffre, J-F. “Les Dérivées des Sigillées Paléochrétiennes du Groupe Atlantique.” Gallia 31, (1) (1973): 207-263.

Sireix, C. and Convertini, F. “La céramique à l’éponge de la region Bordelaise: la céramique marbrée d’Aquitaine.” S.F.E.C.A.G., Actes du congrès de Mans, 1997 (1997): 321-33.

Riley, J.A. “Typology of the Hellenistic and Roman coarse pottery of Berenice.” In Excavations at Sidi Khrebish Benghazi (Berenice), Supplements to Libya Antiqua, Vol. 2, ed. J.A. Lloyd (Tripoli: Dept. of Antiquities, Ministry of Teaching and Education, People’s Socialist Libyan Arab Jamahiriya, 1979):112-236.

Soulas, S. “Présentation et provenance de la céramique estampée à Bordeaux.” Aquitania 14 (1996): 237-53.

Riley, J.A. “The pottery from the cisterns 1977.1, 1977.2, 1977.3.” In Excavations at Carthage conducted by the University of Michigan, 6, ed. J.H. Humphrey (Michigan: Ann Arbor, 1981): 85-124.

Soulas, S. “Éléments d’évolution de la céramique estampée d’après les fouilles de las Place CamilleJullian à Bordeaux.” S.F.E.C.A.G., Actes du congrès de Libourne, 2000 (2000): 145-153.

Riuné-Lacabe, S. 2012a. “La Documentation Archéologique.” In Maurin (ed.) 2012: 15-54.

Soulas, S. “La céramique estampée tardive.” In Maurin (ed.) 2012: 247-50.

Riuné-Lacabe, S. 2012b. “L’Antiquité Tardive et L’Époque Mérovingienne (Ve-VIIIe siècles): un quartier d’habitations et son abandon.” In Maurin (ed.) 2012: 97-125.

Tabor, R. “Woolston Manor Farm, North Cadbury: an outline report of fieldwork in 2006-7 by the South Cadbury Enivrons Project.” Somerset Archaeology and Natural History 151 (2007): 83-96.

Robin, K. “Une villa ‘aristocratique’ à Jonzac (CharenteMaritime).” Aquitania 26 (2010): 75-107.

Teichner, F. “Aglomeração secundária e centro de produção de tintura no sul da província Lusitânia, Xelb (Silves) 5.” Actas do Encontro de Arqueologia do Algarve 2003 (2003): 85-96. Available at: https:// www.academia.edu/993010/Cerro_da_Vila_-_ aglomer%C3%A7%C3%A3o_secund%C3%A1ria_e_ centro_de_produ%C3%A7% C3%A3o_de_tuntura_no_ sul_da_Provincia_Lusitania [accessed 12.11.14].

Saguì, L. “Il deposito della Crypta Balbi: una testimonianza imprevedibile sulla Roma del VII secolo?’” In Saguì (ed.) 1998: 305-30. Saguì, L. (ed.) Ceramica in Italia: VI-VII Secolo. Firenze: Edizioni All’Insegna del Giglio, 1998. Sánchez Pardo, J.C. “Power and rural landscapes in early medieval Galicia (400-900 AD): towards a reincorporation of the archaeology into the historical narrative.” Early Medieval Europe 21, (2) (2013): 140-68.

Thomas, C. “Excavations of a Dark Ages site, Gwithian, Cornwall: Interim Report, 1953-1954.” Proceedings of the West Cornwall Field Club 1 (1954) 59-72.

Savory, H.N. “Excavations at Dinas Emrys, Beddgelert (Caern.), 1954-56.” Archaeologia Cambrensis 109 (1960): 13-77.

Thomas, C. “Imported pottery in Dark Age western Britain.” Medieval Archaeology 3 (1959): 89–111. Thomas, C. “Imported Late-Roman Pottery in Ireland and Western Britain: Chronologies and Implications.” Proceedings of the Royal Irish Academy, Section C 76 (1976): 245-255.

Shipwrecks and History in Plymouth Sound (SHIPS Project) Amphora (12A10), 2012. Available at: http:// www.promare.co.uk/ships/Finds/Fd_12A10Amphora. html [accessed 27.11.14].

Thomas, C. A Provisional List of Imported Pottery in Post-Roman Western Britain and Ireland. Redruth: The Institute of Cornish Studies, 1981.

Silva, A.M.S.P. and Guimarães, J.A.G. “O Castelo de Crestuma, revelado pela arqueologia.” Boletim da Associação Cultural Amigos de Gaia 73 (2011): 5-13. Available at: https://www. academia.edu/8275775/Silva_Ant%C3%B3nio_ Manuel_S._P._and_Guimar%C3%A3es_J._A._ Gon%C3%A7alves_2011_-_O_castelo_de_Crestuma_ revelado_pela_arqueologia._As_principais_fases_ de_ocupa%C3%A7%C3%A3o_do_s%C3%ADtio_ arqueol%C3%B3gico [accessed 27.11.14].

Thomas, C. Exploration of a Drowned Landscape: archaeology and history of the Isles of Scilly. London: B.T. Batsford, 1985. Thomas, C. 1988a. “The context of Tintagel: a new model for the diffusion of post-Roman Mediterranean imports.” Cornish Archaeology 27 (1988): 7–25. 170

Bibliography Thomas, C. 1988b. “The C.A.U. Excavations at Tintagel Island: discoveries and their implications.” Cornish Studies 16 (1988): 49-60.

2014). Available at: http://archaeologydataservice. ac.uk/archives/view/amphora_ahrb_2005/index.cfm (doi:10.5284/1028192) [accessed: 26.01.15].

Thomas, C. 1988c. “Tintagel Castle.” Antiquity 62 (1988): 421-34.

Uscatescu, A, Fernández Ochoa, C. and García Díaz P. “Producciones Atlánticas de Terra Sigillata Gálica Tardía en la Costa Cantábrica de Hispania.” Cuadernos de Prehistoria y Arqueología de la Universidad Autónoma de Madrid 21 (1994): 183-234.

Thomas, C. “‘Gallici Nautae de Galliarum Provinciis’ – A sixth/seventh century trade with Gaul, reconsidered.” Medieval Archaeology 34 (1990): 1-26.

Vaag, L.E. “Phocaean Red Slip Ware-Main and secondary productions.” In M. B. Briese and L. E. Vaag (eds) 2005: 132-8.

Thomas, C. English Heritage Book of Tintagel, Arthur and Archaeology. London: B.T. Batsford, 1993. Thomas, C. and Thorpe, C. “Comment.” In Batey et al. 1993: 58-60.

Véquaud, B. “La céramique du haut Moyen Âge en PoitouCharentes: état des connaissances (VIe-Xe siècles).” In Wisigoths et francs autour de la bataille de Vouillé (507) ed. L. Bourgeois (Saint-Germain-en-Laye: Association française d’archéologie mérovingienne, 2010): 263-78.

Thorpe, C. “The pottery from the Lower Ward.” In Hartgroves and Walker 1988, 22-5. Thorpe, C. “Ceramics.” In Harry and Morris 1997: 74-82.

Viegas, C. Terra Sigillata da Alcácova de Santarém – Economia, comércio e cerâmica, Trabalhos de Arqueologia 26. Lisboa: Instituto Português de Arqueologia, 2003.

Thorpe, C. “Ceramics: post-Roman imported.” In Barrowman et al. 2007: 231–47. Thorpe, C. and Wood, I. “Early medieval pottery.” In P. Herring, A. Preston-Jones, C. Thorpe and I. Wood, “Early Medieval Cornwall.” Cornish Archaeology 50 (2011): 263-86.

Viegas, C. “Les céramiques tardive dans les sites du sudouest de la Péninsule Ibérique (Algarve-Portugal).” LRCW2: Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean: Archaeology and archaeometry, Volume I, BAR International Series 2185 (Oxford: BAR Publishing, 2007): 71-83.

Timby, J. 2011a. “Pottery.” In Fulford and Rippon 2011: 39-58. Timby, J. 2011b. “Appendix Two: Roman Pottery.” In Fulford and Rippon 2011: 145-50.

Viegas [Taveira], C. A ocupação romana do Algarve: estudo do povoamento e economia do Algarve central e oriental no período romano, PhD Thesis, University of Lisbon, 2009. Available at: http://hdl.handle. net/10451/568 [accessed 22.10.14].

Todd, M. “Lammana.” Cornish Archaeology 22 (1983): 122-3. Tomber, R. “Two unusual amphora types from the Museum of London.” Journal of Roman Pottery Studies 10 (2003): 107-8.

Waagé, F.O. “The Roman and Byzantine Pottery.” Hesperia 2, (2) (1933): 279-328.

Tomber, R. Indo-Roman trade. From pots to pepper. London: Bristol Classical Press, 2008.

Waagé, F.O. “Hellenistic and Roman Tableware of North Syria.” Antioch 4.1 (1948): 1-60.

Tomber, R. and Williams, D. “Late Roman Amphorae in Britain.” Journal of Roman Pottery Studies 1 (1986): 42-54.

Wailes, B. Some Imported Pottery in Western Britain, AD 400-800: its connections with Frankish and Visigothic Gaul, unpublished PhD thesis, University of Cambridge, 1963.

Tomber, R. and Williams, D. “Egyptian Amphorae in Britain and the Western Provinces.” Britannia 31 (2000): 41-54.

Warner, R. “The Clogher Yellow Layer.” Medieval Ceramics (Bulletin of the Medieval Pottery Research Group) 3 (1979): 37-40.

Turner, S. “Coast and countryside in ‘Late Antique’ southwest England, c. AD 400-600.” In Debating Late Antiquity in Britain AD 300-700, eds R. Collins and J. Gerrard, BAR British Series 365 (Oxford: BAR Publishing, 2004): 25-32.

Watier, B. “Les amphores funéraires de la nécropole de Saint-Seurin à Bordeaux.” Information d’histoire de l’art 3 (1973): 113-8.

Turner, S. and Gerrard, J. “Imported and local pottery from Mothecombe: some new finds amongst old material at Totnes Museum.” Proceedings of the Devon Archaeological Society 62 (2004): 171-5.

Webster, C. and Mayberry, T. (eds). The archaeology of Somerset. Wellington: Somerset Books, 2007. Wessex Archaeology. Lellizzick, near Padstow, Cornwall, Archaeological Evaluation and Assessment of Results, unpublished report, Wessex Archaeology Report 65312, 2008. Available at: http://www.wessexarch.co.uk/ reports/65312/lellizzick-padstow [accessed 04.03.14].

Tyers, P. Roman Pottery in Britain. London: Batsford, 1996. University of Southampton. Roman Amphorae: a digital resource. York:Archaeology Data Service, 2005 (updated 171

Links to Late Antiquity Wessex Archaeology. Looe, Cornwall: Archaeological Evaluation and Assessment of Results, unpublished report, Wessex Archaeology Report 68734.01, 2009. Salisbury: Wessex Archaeology.

Bruce-Mitford, R. Mawgan Porth. A settlement of the late Saxon period on the north Cornish coast: excavations 1949-52, 1954, and 1974. London: English Heritage, 1997.

Wickham, C. Framing the Early Middle Ages: Europe and the Mediterranean 400-800. Oxford: Oxford University Press, 2005.

Burrow, I. Hillfort and Hill-top settlement in Somerset in the first to eight centuries A.D., BAR British Series 91. Oxford: British Archaeological Reports, 1981.

Wilkes, J. “Sailing to the Britannic Isles: some perspectives on the remote northwest from the sixth century BC to the seventh century AD.” In Communities and Connections: essays in honour of Barry Cunliffe, eds C. Gosden, H. Hamerow, P. de Jersey, and G. Lock (Oxford: Oxford University Press, 2007): 3-19.

Campbell, E. “New finds of post-Roman imported pottery and glass from South Wales.” Archaeologia Cambrensis 138 (1989): 59-66. Campbell, E. and Lane, A. “Excavations at Longbury Bank, Dyfed, and Early Medieval Settlement in South Wales.” Medieval Archaeology 37 (1993): 15-77.

Williams, D. “Late Roman Amphora 1: a study of diversification.” In M.B. Briese and L.E. Vaag (eds) 2005: 157-68.

Gossip, J. and Jones, A. M. “Excavations at Tremough, Penryn, Cornwall, 2000-6.” Cornish Archaeology 4849 (2010): 1-66.

Williams, D. “Les amphores.” In Cunliffe and Galliou 2007: 38.

Herring, P. St Michael’s Mount, Cornwall: reports on archaeological works 1995-1998, Cornwall Archaeological Unit. Truro: Cornwall County Council, 2000.

Williams, D. 2011a. “A note on two amphora sherds.” In Fulford and Rippon 2011: 146. Williams, D. 2011b. “The amphorae.” In Douglas et al. 2011: 78-81.

Laing, L. and Longley, D. The Mote of Mark: a Dark Age hillfort in south-west Scotland. Oxford: Oxbow, 2006.

Williams, D. and Carreras, C. “North African amphorae in Roman Britain: a re-appraisal.” Britannia 26 (1995): 231-52.

McBride, R. and Timby, J. “The post-Roman imported pottery.” In Allan et al. 2015: 252. Miles, H. “Excavations at Killibury Hillfort, Egloshayle 1975-6.” Cornish Archaeology 16 (1977) 89-119.

Wooding, J.M. 1996a. Communications and Commerce along the Western Sealanes AD 400-800, BAR Reports International Series 654. Oxford: BAR Publishing, 1996.

Young, C.J. Oxfordshire Roman Pottery: The Roman Pottery Industry of the Oxford Region, BAR Reports British Series 43. Oxford: British Archaeological Reports, 1977.

Paz Peralta, J.A. “Aportaciones a la difusión y cronología de la African red slip ware de los siglos V-VII d. C. en dos núcleos urbanos del interior de España: Asturica Augusta (Astorga) y Caesar Augusta (Zaragoza).” Bolskan 21 (2004): 27-43. Available at: http://dialnet.unirioja.es/servlet/articulo;jsessio nid=A1C00CA85F0067DDEFB9B878FB6A301B. dialnet01?codigo=2011693 [accessed 07.01.15].

Zevi, F. and Tchernia, A. “Amphores de Byzacène au BasEmpire.” Antiquités Africaines (1969): 173-214.

Quinnell, H. and Elsdon, S. “The finds from Halligye fogou.” Cornish Archaeology (2009-2010): 145-71.

Zimmer, H. “Der Weinhandel Westgalliens nach Irlan im 1. bis &. Jahr-hundert n. Chr und sein Niederschlag in Irischer Sage und Sprache.” In Über direkte Handelsverbindungen Weestgalliens mit Irland im Alterum und frühen Mittelalter, H. Zimmer (Berlin: Reichsdruckerei, 1909): 430-76.

Rahtz, P. “Excavations on Glastonbury Tor, Somerset, 1964-6.” The Archaeological Journal 127 (1970): 1-82.

Wooding, J.M. 1996b. “Cargoes in trade along the Western Seaboard.” In Dark (ed.) 1996: 67-82.

Rahtz, P. “Pottery in Somerset, A.D. 400-1066.” in Medieval Pottery from Excavations, eds V.I. Evison, H. Hodges and J.G. Hurst (London: John Baker, 1974): 95-126.

8.3 Additional references for appendices Alcock, L. and Alcock, E.A. “Reconnaissance excavations on Early Historic fortifications and other royal sites in Scotland, 1974-87: 4, Excavations at Alt, Clut, Clyde Rock, Strathclyde, 1974-5.” Proceedings of the Society of Antiquaries of Scotland 120 (1990): 95-149.

Toolis, R. and Bowles, C. The Galloway Picts Project: excavation and survey of Trusty’s Hill. Summary report published on behalf of the Dumfriesshire and Galloway Natural History and Antiquarian Society, 2012. Available at: http://gallowaypicts.com/wordpress/ results/summary-report/ [accessed 24.04.15].

Allan, J. and Blaylock, S. “Medieval pottery and other finds from Pigs Paradise, Lundy.” Proceedings of the Devon Archaeological Society 63 (2005): 65-91.

Wilkinson, P.F. “Excavations at Hen Gastell, Briton Ferry, West Glamorgan, 1991-92.” Medieval Archaeology 39 (1995): 1-50. 172

173

LRD

Late Roman D

ARSC/E

African Red Slip C/E

LRC

ARS-E

African Red Slip E

ARS-C

African Red Slip C

ARS-D

ARSA/D

African Red Slip A/D

African Red Slip D

ARS-A

ARS

Abbr.

African Red Slip A

SubType

Late Roman C

African Red Slip

Type

N/A

‘A’ Ware (largely ‘Ai’)

Class A/‘A Ware’ (forms part of Aii)

Radford

Cypriot Red Slip

Phocaean Red Slip

African Red Slip

Hayes

Late D

Late C

Late B (later)

Late A

Late B (early)

Late A and B

Waagé

T.S.Chiara D

T.S.Chiara C

T.S.Chiara A

Terra Sigillata Chiara

Lamboglia

T.S. Cipriota (Late Roman E/D)

T.S. dell’Asia Minore (Late Roman C)

T.S Africana C/E

T.S Africana E

T.S. Africana D1-2

T.S. Africana C1-C5

T.S. Africana A/D

T.S. Africana A1-A2

Terra Sigillata Africana

‘Atlante’ (Carandini et al. 1981)

Uncertain. Cyprus and/ or Southern Turkey

Western Turkey. Various production sites, principally Phocaea/ Foça

South-Central Tunisia

North Tunisia

Central Tunisia

North Tunisia

North Africa

Origin

Fineware, Red Slip Wares (Compiled from Bonifay 2004, 47, Campbell 2007a, 14-16 and Hayes 2008, 67)

NO

NO

-

NO

YES

YES

-

YES

YES

‘Roman’ Britain

NO

YES

-

NO

YES

NO

-

-

YES

‘PostRoman’ Britain

NO

YES

-

NO

YES

NO

-

-

YES

Ireland

NO

YES

-

NO

YES

YES

-

-

YES

Atlantic France

in 5th to 7th century context

Import present?

YES

YES

-

?

YES

YES

-

-

YES

Atlantic Iberia

Appendices

Appendix A: Concordance of ceramic terminology

Dérivées des Sigillées Paléochrétiennes

Type

174

DSPA

DSPL

DSPP

DSP-Languedoc Group

DSP-Provençal Group

DSP

Abbrev.

DSP-Atlantic Group

Sub-Type

D Ware/ Class D

Radford/ Thomas

Gaulish ‘T.S. Grise’

Hayes 1972

Dérivées des Sigillées Paléochrétiennes Atlantique

Paléochretienne Grise, Visigothic Ware, Céramique estampée, Terra Sigillata Gálica Tardía (T.S.G.T.)

Other names

Fineware, DSP (Compiled from Campbell 2007a; Soulas 1996; 2000; Fernández 2007; 2014)

Provence-includes production centre near Marseille

Languedoc/ Narbonne

Bordeaux region

France-Various production regions

Origin

NO

NO

NO

NO

‘Roman Britain’

?

?

YES

YES

‘Post-Roman Britain’

NO

NO

YES

YES

Ireland

in 5th to 7th century context

Import present?

YES?

YES?

N/A

N/A

Atlantic France

YES?

YES?

YES

YES

Atlantic Iberia

Links to Late Antiquity

175

Late Roman 7

Late Roman 5-6

Late Roman 4

Late Roman 3

‘Water Jar’

Bvi

54

Almagro 54, Gaza Amphora

YES

52

46

49

LRA4B (LRA4B1, LRA4B2 and LRA4B3) N/A

48

LRA4A (LRA4A1 and LRA4A2)

LRA4

‘Palestinian, Bagshaped’

Almagro 54

Zemer 53

?

NO

?

?

YES

?

NO

?

?

YES

NO

NO

YES

YES

Late Roman ‘sous module’

LRA3B (LRA3B1-4)

54/Bis

YES

45

NO

?

?

YES

YES

?

YES

‘PostRoman Britain’

NO

NO

?

?

NO

NO

NO

NO

NO

NO

?

?

YES

YES

?

YES

Ireland

NO

YES

YES

YES

YES

?

YES

?

YES

NO

?

?

YES

YES

YES

YES

Atlantic France

in 5th to 7th century context

Late Roman 3 twohandled (later)

Late Roman 3 onehandled (early)

Biv

NO

LRA2C

LRA3

NO

NO

LRA2B

65 NO

43

NO

‘Roman Britain’

Import present?

LRA2A

LRA2

Bi

Other names

NO

53

Keay

LRA1B (LRA1B1-2)

44

Peacock and Williams

NO

Bii

Radford/Thomas

LRA1A

LRA3A (LRA3A1-4)

LRA1

Late Roman 1

Late Roman 2

Sub-Type (Pieri 2005)/ Abbrev.

Type (after Riley)

Amphorae, East Mediterranean (Compiled from Peacock and Williams 1986; Bonifay and Pieri 1995; Pieri 1998; 2005; University of Southampton 2005)

NO

YES

YES

YES

YES

YES

YES

?

YES

NO

YES

YES

YES

YES

YES

YES

Atlantic Iberia

Appendices

Links to Late Antiquity Appendix B: Location maps (individually scaled) B.1: Scotland

176

Appendices B.2: Ireland

177

Links to Late Antiquity B.3: Southern/western Britain

178

Appendices B.4: South-west Britain – showing sites discussed in Sub-chapter 4.3

179

Links to Late Antiquity B.5: Western France

180

Appendices B.6: Spain and Portugal

181

Links to Late Antiquity B.7: Location of investigations in Bordeaux, showing late Roman fortifications and port. After Barraud and Maurin 1996, Figure 2; Girond 2012, Figure 16; Riuné-Lacabe 2012a, Figure 1

182

183

c. 420 to 475, 73B ‘mainly late’

73A-B

73

c. 400 to 450 (From c. 380 in Supplement)

61B

c. 360 to 470

c. 400 to 450 (Finishes c. 380 in Supplement

c. 320 to 420

59B

61A

c. 320 to 380/400

59A

67

61

59

58A and B c.290/300 to 375

58 (ARS-C) (ARS-D)

58B

c. 440 to 500

84 (ARS-C)

58A

A and B

82 (ARS-C)

82A - c. 430 to 475 82B - c. 460 to 500+

c. 420 to 475 - 73B ‘mainly late’

LRP date (Hayes 1972)

73 A and B

Fulford variant

73 (ARS-C)

Fulford form

52A - c. 300 to 350? 52B - 280/300 to late-4th century; late variants early5th century+

Hayes variant

52 (ARS-C)

Hayes form (ARS-D unless specified)

Late 4th century or c. 400 to c. 450

c. 360 to c. 425-450

Late 4th century or c. 400 to 475-500

4th century to 425-450

Carthage Date (Fulford 1984a)

67A – second half 4th century 67B – end 4th, characteristic of first half 5th century 67C – mid 5th-mid 6th century

61A/B1 and A/B2 – start of fifth to mid-5th century? 61A/B3 – first half 5th century – late examples to 480 61A/B4 – first third 5th century 61B1 and B2 – first half of 5th century 61B3 – mid 5th century, late examples to end of 5th century

(‘C5’ production characteristic of second-half of 5th century to start of 6th century at Marseille)

Relevant refinements from Bonifay 2004

c. 340/360 to 460/490

Attested from the early-4th century?

Refinements from LRFW1 (Cau et al. 2011c)

Southwark (1)

52B at Old Ford (1)

Identified in Britain or Ireland? (After Bird 1977; Thomas 1981; Campbell 2011; and additional observations)

(Major/significant late forms including all forms identified in Britain). Fulford’s ‘Carthage’ forms and variants included where used by Campbell (2011).

Appendices

Appendix C: ARS Chronology

90B

184

91

47.2

c. 530 to 600+

91C

c. 600 to 650

c. 450 to 530

91B

91D

Mid to late 5th century

91A

Mid to late-5thcentury to mid-7th century

Second half of 6th century, and possibly later

Early-6th century

88

90

87A – second half of 5th century. 87B-C early 6th century

87

N/A

42

LRP date (Hayes 1972)

c. 425 to 475

Fulford variant

76

Fulford form

c. 420 to 450

Hayes variant

75

Hayes form (ARS-D unless specified)

Floruit 525 to 550

(Form 47.1) c. 370 to c. 500

c. 525 to 550

Late 5th century floruit

mid-5th (?) century to early-6th century (c. 525 to 530)

c. 400-425 to 475-500

Probably 5th century

Carthage Date (Fulford 1984a)

Appear from c. 400/420. Late, small variant of 91B into 6th century

Second half of 6th and start of 7th century

87A/88 – end of 5th to start of 6th century 88A – second quarter to mid-6th century 88B – mid to second half of 6th century 88C – end of 6th to start of 7th century

Relevant refinements from Bonifay 2004

Mid-late 6th century (some possibly of central Tunisian origin)

Potentially into second half of 6th century

Refinements from LRFW1 (Cau et al. 2011c)

Dinas Powys (2) Tintagel (2)

Pevensey Castle (1)

Tintagel (1)

Tintagel (1)?

Pevensey Castle (1)

Identified in Britain or Ireland? (After Bird 1977; Thomas 1981; Campbell 2011; and additional observations)

Links to Late Antiquity

185

103B

64B

c. 560/580 to 620

99C

c. 500 to third quarter of 6th century

c. 530 to 580

99B

64

c. 510 to 540

58

99

c. 490 to 540

99A

55

50/52

96

93B/94/98

After AD 500

c. 470 to 540

LRP date (Hayes 1972)

93B

52

103

Fulford variant

Mostly 5th century

50

Fulford form

93A

Hayes variant

N/A

N/A

N/A

93

Hayes form (ARS-D unless specified)

c. 500 to c. 575

c. 500 to c. 575

c. 475-500 to c. 575

c. 525 to 575

c. 500 to c. 550 Highest percentage c. 525-530

c. 500 to 575+ (peak of popularity passed by c. 575)

Before 450, residual by end of 5th century

Carthage Date (Fulford 1984a)

End of 6th to 7th century

Second quarter of 6th to start of 7th century

End of 5th to mid-6th century

Relevant refinements from Bonifay 2004

Finishes end of 7th century?

Starts second quarter of 6th century

Starts c. 480/490

Refinements from LRFW1 (Cau et al. 2011c)

Cadbury Congresbury (1)

Halliggye (1)? (103B?) Tintagel (2) Cadbury Congresbury (1)? (Hayes 103-6)

Pevensey Castle (1)

Dinas Powys (1)? Tintagel (1)?

Tintagel (1)

Tintagel (1)

Tintagel (2)

Tintagel (1) (E.M. 18.2?)

Cannington Cemetery (1) Dinas Powys 1 (50.3?) Tintagel (1) Whithorn (1) (E.M. 24.1/ Fulford 50.2/3?)

Identified in Britain or Ireland? (After Bird 1977; Thomas 1981; Campbell 2011; and additional observations)

Appendices

186

c. 580/600 to 660+

c. 600 to 650

Early-7th century

580/600 to mid-7th century (610/620 to 680/700 in Supplement)

107

108

109

c. 570 to 600, with late variants to 625+

LRP date (Hayes 1972)

104B

65B

Fulford variant

c. 530 to 580

65

Fulford form

104A

Hayes variant

105

104

Hayes form (ARS-D unless specified)

Present from c. 580600, highest % in first quarter of seventh century, residual by second half of seventh?

c. 550 to c. 575600+(?)

c. 550 to c. 600-650

c. 550 to c. 625-650

65A-B = c. 500 to c. 600-625. Floruit after c. 525

Carthage Date (Fulford 1984a)

109A – end of 6th to mid-7th century

7th century – evolution of Hayes 98

105A – end of 6th to first half of 7th century

Mid-second half of sixth-century

104A1 – end of 5th, first decades of 6th century, 104A2 – second quarter to mid 6th century, 104A3 – end of 6th to mid-7th century

Relevant refinements from Bonifay 2004

Transition A-B around mid to third quarter of 6th century

Starts in late 5th century

Refinements from LRFW1 (Cau et al. 2011c)

Iona (1)

Tintagel (1)?

Cannington Cemetery (1)? Tintagel (2)? Whithorn (1)

Identified in Britain or Ireland? (After Bird 1977; Thomas 1981; Campbell 2011; and additional observations)

Links to Late Antiquity

Appendices Appendix D: Mothecombe ceramic data

YR 2/1) with very dark grey (2.5 Y 3/1) burnished surfaces containing abundant small quartz grits. Absence of obvious slip suggests South-East Dorset black-burnished I originating in the Poole Harbour area of south-east Dorset rather than South-Western BBI distinguished as a separate type at Exeter (Holbrook and Bidwell 1991, 114). Abrasion at edge of rim prevented identification of the form or accurate diameter measurement. Open vessel, possibly a flat-rimmed dish c.AD 120 into the third century (Holbrook and Bidwell 1991, 97).

Pottery discovered prior to the 2004 (not illustrated): • four joining body-sherds of a Late Roman 1 amphora with graffito (three sherds at discovery but since broken). • one body-sherd of Roman or early medieval imported amphora, possibly the ‘thick-walled’ ware identified at Bantham (Bidwell et al. 2011, 100–2) • one body-sherd of an early medieval imported amphora, of unidentified type • fragments of three handles of Late Roman 2 amphorae, representing at least two vessels • two sherds of locally-produced ware, of uncertain date • two sherds of ‘late Iron Age Glastonbury Ware’ reported to have come from hill-wash above the hearths (Turner and Gerrard 2004, 174).

Ceramic spindle-whorls (Figure 4.9): SW1 (Complete, circular, discoid whorl. Maximum external diameter 38mm, perforation diameter 8 mm, thickness 7 mm, weight 10.8 g. Flat, with a slightly rounded edge. Unworn, circular, straight-sided perforation. Made from a vessel with a fine, pink fabric (5 YR 7/4). Samian, (possibly Central Gaulish) although no trace of slippedsurface survives. Mid-first to mid-third century AD (Tyers 1996).

Roman pottery (not illustrated): P1 Central Gaulish black-slipped ware Two abraded bodysherds, possibly from the same vessel. Both have traces of a glossy black slip (10 YR 2/1) internally and externally. Originating in the Central Gaulish sigillata workshops, such as Lezoux, from the mid-second to early third century AD (Tyers 1996, 138)

SW2 Fragment (approx. 50%) of a discoid whorl. Estimated external diameter 36 mm, estimated perforation 8 mm, thickness 7 mm, weight of fragment 4.4 g. Mostly flat (though uneven and tapering toward one edge) with a slightly rounded edge. Straight-sided perforation. One surface is light red (2.5 YR 6/6), the other is light reddish-brown (5 YR 6/4) and uneven, with two noticeable voids and a faint scratched groove. Probably reused from a Roman vessel, possibly an Oxfordshire fabric, and if so, mid-third century to c.ad 400 (Young 1977).

a. fine, light red fabric (2.5 YR 7/6). Body-sherd with traces of barbotine decoration. 23 × 14 mm, wall thickness 3mm, 0.6g. b.as P1a but paler, pink fabric (5 YR 7/4). Curved bodysherd, possibly of a beaker. 27 × 22 mm, wall thickness 3 mm, 1.1 g.

SW3 Fragment (approx. 17%) of a discoid whorl. Estimated external diameter 35 mm, thickness 9 mm, weight of fragment 3.5 g). Largely flat but uneven. Rounded edge, straight-sided perforation. Hard fabric. Uneven colouring, exterior surface light red (2.5 YR 6/8) to brown (7.5 YR 5/3), unoxidized core dark greyish brown (10 YR 4/2). Probably reused from a Roman vessel but otherwise undated.

P2 Oxfordshire red/brown-slipped ware Abraded rimsherd, not enough to measure the full diameter. 35 × 27 mm, wall thickness 5 mm, 4.5 g. Hard, sandy, light red fabric (2.5 YR 6/6) with unoxidized core and no surviving trace of slipped-surface. Vertical rim, slightly thickened symmetrically at top. Groove on exterior beneath the rim at the junction with (not surviving) body. Type C97 mortaria (copying samian form 45), AD 240-400+ (Young 1977, 173). Oxford shire slipped ware is not uncommon in the South West, the industry expanding to supply this area in the fourth century (Young 1977, 239). Oxfordshire slipped mortaria of Young Type 100 have been identified at Tintagel (Thorpe 2007, 230).

Early medieval imported amphorae (Figure 4.10): LRA1 P5 Soft, rough feel with irregular fracture. Reddish yellow (5 YR 6/8). Abundant, ill-sorted, fine to medium (0.25–0.5 mm) quartz, moderate fine black sand. Sparse, very fine gold mica and fine irregular limestone. Moderate voids (up to 0.5 mm). Comparable to the description of ‘Fabric 1’ as identified within the LRA1 at Bantham (Bidwell et al. 2011, 94)

P3 Samian Very abraded body-sherd. 21 × 17 mm, wall thickness 6 mm, 2 g. Fine, if slightly rough, light reddish brown fabric (2.5 YR 6/4) with traces of light red slip (2.5 YR 6/8) on interior and within groove on exterior. Probably Central Gaulish, mid-first to second century AD (Tyers 1996, 113).

a. A total of seven abraded body-sherds (two little more than fragments). Combined weight 31 g. Three of these join: 72 × 26 mm, wall thickness 10 mm, having a distinctive, horizontally running, narrow raised ridge (width 0.4 mm, height 0.2 mm) on the exterior. Another two join separately: 42 × 41 mm, wall thickness 11 mm.

P4 Black-burnished ware Two abraded, joining rim-sherds (probably split since excavation), 45 × 23 mm, wall thickness 6 mm, combined weight 6.8 g. Hard, black fabric (10 187

Links to Late Antiquity b. (not illustrated) A further, very abraded body-sherd in the same fabric with shallow grooves/ribbing on the exterior might belong to the top or bottom of the same vessel. 30 × 24 mm, wall thickness 8 mm, 4.5g

from the same vessel. 28 × 25 mm, wall thickness 8 mm, 3.8 g.

P6 Hard, rough feel with irregular fracture. Reddish yellow fabric (7.5 YR 6/6). Abundant, sub-rounded, ill-sorted, medium quartz/red and white quartzite (0.25–1 mm). Common, fine black sand and iron ore. Sparse, very fine gold mica (up to 0.1 mm) and irregular, medium limestone (up to 1 mm). Compares to LRA1 ‘Fabric 1’ description from Bantham (Reed et al.2011, 94).

P9 Abraded body-sherd in a light red (2.5 YR 6/8), soft, smooth/powdery fabric with moderate, irregular, ill-sorted, medium quartz and red quartzite (up to 1 mm), moderate voids (up to 3 mm) and rare, fine limestone. Exterior yellow (10 YR 7/6), possibly slipped. No visible ridging/ribbing. 65 × 55 mm, wall thickness 9 mm, 19.6 g. The colour, general fabric characteristics and surface finish suggest a possible North African origin, but no specific form can be identified and the sherd might be residual. This sherd draws obvious connections to Bantham where two North African amphorae were identified (Bidwell et al. 2011, 102).

North African Amphora

a. Neck sherd from a Late Roman 1 amphora, good condition. 59 × 58 mm, wall thickness 10 mm, but thickened internally above a notch/groove at junction of neck and body. 41.8 g. Very little of the body survives below the neck, but the angle is relatively gradual, indicating LRA1 rather than the sharp-angled LRA1 ‘Bantham-type’ distinguished at Bantham (Bidwell et al. 2011, 97–100).

Unidentified Late Roman amphora (LRA) P10 (not illustrated) A very abraded fragment of an imported amphora in a hard, quite fine, slightly powdery fabric with moderate, ill-sorted very fine to medium quartz (0.1 to 1 mm), sparse, very fine gold mica and a very coarse red rock fragment (up to 7 mm). Differentially fired, one surface reddish yellow (5 YR 7/6), the other yellow (10 YR 7 6). 37 × 27 mm, wall thickness 12mm, 9.2 g. Roman or early medieval.

b. (not illustrated) An additional, abraded sherd was collected from the beach in 1995. 44 × 43 mm, wall thickness 8mm, 12.3 g. The fabric is as P6a and it could be from the same vessel, although the exterior surface is a very pale brown (10 YR 7/4), possibly due to post-depositional conditions. No external ribbing visible.

Appendix E: High Peak ceramic data

P7 Body-sherd, good condition, 33 × 44 mm, wall thickness 9 mm, 12.9 g. Soft, reddish yellow fabric (7.5 YR 7/6). Abundant, well sorted very fine to fine quartz (up to 0.25 mm). Sparse, fine to medium (up to 0.5 mm) black rock fragments and very sparse, very fine gold mica. One, coarse (2 mm) grey rock inclusion. Although the fabric is noticeably less gritty than P5 and P6, the colour and fabric do not match the distinctive smooth, cream, ‘Banthamtype’ of LRA1. Distinctive, shallow ‘clapboard ridges’ on the exterior and grooving on the interior.

Fabric groups (each representing an estimated minimum of one vessel). For locations refer to Figure 4.11 LRA1 Group 1: Soft, sandy fabric with abundant, ill-sorted, fine to medium black sand and grey rock inclusions, Frequent, irregular voids (probably from limestone), moderate sub-rounded white quartzite. The fabric seems relatively coarse although many of the sherds have very abraded surfaces. Some variation in colour, though most sherds reddish yellow (5YR 6/6 to 7.5YR 7/6).

P8 Soft, rough feel with irregular fracture. Reddish yellow (5 YR 6/6) speckled surface, interior yellowish red (5 YR 5/6), core reddish yellow (7.5 R 6/6). Abundant, ill-sorted fine to medium quartz/red and white quartzite (0.25–1 mm). Common, fine black sand. Sparse, fine, irregular limestone and very fine gold mica. Compares to the description of ‘Fabric 2’ as identified within the LRA1 at Bantham (Bidwell et al. 2011, 94).

This group includes: 36 sherds with a combined weight of 144g, wall thickness varying between 7 and 9mm.

a. Abraded body-sherd 49 × 47 mm (split since excavation), wall thickness 8mm, 13.2 g. Joins to P8b.

All were recovered during Pollard’s excavations in ‘Area B’ (marked as ‘B2’) and include Pollard 1966, Figure 12.8 and possibly Figure 12.6 (although the bottom part of this sherd is now missing). Many have traces of narrow-spaced ribbing or wider, ‘clapboard ridging’ on the exterior surfaces, representing, respectively, sherds from the top and bottom, or from the central section of the body of the amphora. It is likely that these were among or comprised the 23 sherds recovered from Trench B in a patch of blackened sand labelled ‘cooking area II’ (Pollard 1966, 44).

b. Two joining sherds, combined dimensions 32 × 28 mm, wall thickness 8mm, 4.7 g. Joins to P8a. There are closely spaced, shallow grooves on the exterior of P8a and b, the orientation of which suggests these joining sherds may be from near the bottom of the amphora. c. (not illustrated) An additional, abraded body-sherd in the same fabric with the same shallow grooves may be 188

Appendices A further sherd bearing an ‘RK’ stamp (4g) as well as two ‘unassigned’ scraps (1g) have similar fabrics.

‘High Peak’. Apart from one ‘unassigned’ sherd all can be linked to ‘antiquarian’ investigations.

The large body-sherd illustrated by Hutchinson and Pollard (1966, Figure 12.4) also has a similar fabric although there are fine to coarse limestone inclusions rather than voids. The larger inclusions are sparse and certainly smaller than illustrated by Hutchinson or indicated in Pollard (1966, 53). The sherd is reddish yellow (5YR 6/6) with a slightly darker exterior, light reddish brown (5YR 6/4). Weight 86g, wall thickness 7-10mm

A number of these sherds have traces of the ‘clapboard ridging’ characteristic of LRA1 or distinctive raised ridges, some of which seem to have been smudged or flattened before the vessel was fired (as in Pollard 1966, Figure 12.10). It is possible that all these sherds were from the same vessel although only one join was observed. A further six sherds (four of which join), with a combined weight of 51g, from Pollard’s excavations have similar fabric. All are from Area G (and marked G2 or GA2) and have very widely spaced ‘clapboard ridging’ from the central section of a LRA1 amphora. One sherd was illustrated as Pollard 1966, Figure12.9.

In addition, two sherds recovered in 2012 by AC Archaeology have very similar fabric and may be from the same vessel. Group 2:

Group 4: Soft, sandy fabric; which is similar to Group 1. Frequent, fine to medium, sub-rounded voids, moderate, fine to medium sand and white quartz. Sherds mostly reddish yellow (7.5YR 7/6).

Soft, sandy fabric with frequent fine voids or limestone and frequent fine, well-sorted, black sand. Also moderate, medium rounded grey rock inclusions and sparse fine red ironstone. Pink (5 YR 7/4) to reddish yellow (7.5 YR 8/4) but noticeably paler at exterior, very pale brown (10YR 8/4).

This group comprises: 38 sherds with a combined weight of 98g, wall thickness between 7 and 9 mm, all from Pollard’s Area A. 28 of these sherds, weighing 88g, are marked ‘A2’ and were from the ‘Dark-Age ditch’ (although one of these was recovered from the spoil-heap), while 10 sherds, weighing 10g, were from Pollard’s ‘Trench A outer rampart clay-with-flintlevel’.

Three sherds with a combined weight of 34g, wall thickness between 5 and 8mm. The sherds have closely spaced ribbing on the exterior, rather than the combing suggested in Hutchinson’s illustration which might have indicated LRA2. One sherd has ridging characteristic of LRA1.

Again, many of the sherds are very abraded although a number have traces of ribbing or ridging. A number have been glued together, for example those illustrated in Pollard Figure12.5, which have widely spaced ridging and, as indicated by Thomas (Pollard 1966, 53) are from the central section of a LRA1 amphora.

Group 5: Nine joining sherds of a LRA1 amphora recovered by AC archaeology in 2012 (see below) LRA1 ‘Bantham-type’ amphora

Group 3: Group 6: Soft, sandy fabric with abundant, fine to medium limestone and moderate, medium black sand and sparse fine quartz and grey rock inclusions. Colour varies between pink (5YR 7/4), reddish yellow (5YR 7/6) and light reddish brown (2.5YR 7/4), but many sherds have a distinctive, paler ‘speckled’ appearance to the exterior surface, similar to Fabric 3 as described at Bantham (Bidwell et al. 2011, 94).

A hard, smooth, close grained fabric with frequent, fine voids, moderate, fine, well-sorted dark sand and sparse, fine gold-mica. Very pale brown (10YR 8/3) A total of three body-sherds (two of which join) with a total weight of 14g, wall thickness 6 to 8mm. Two of these are joining sherds, and although abraded, show traces of squared, horizontal grooving which seems to have been made with a flat implement. These are marked as from Pollard’s Area B (B1). The third sherd, likely to be from the same vessel, is also from Pollard’s excavation but is marked ‘HP 2A’ or possibly ‘HP BA’. This was illustrated in Pollard 1966 Figure 12.2 (although part of the sherd is now missing) and described by Thomas as being from a ‘buff coloured’ Bi (LRA2) amphora with ‘stylus grooving’ (Pollard 1966, 53-4). As Pollard only refers to 13 ‘reddish’ amphora sherds being found in area A it is likely that these

This group comprises: 15 sherds with a combined weight of 159g, wall thickness between 6 and 10mm. Of these, three may have been illustrated by Hutchinson (also illustrated as Pollard 1966, Figures 12.10 and 12.7). In total, six bear ‘RK’ stamps while one is accessioned as ‘excavated by D’Urban’ and has a stamp that only states 189

Links to Late Antiquity A total of 14 sherds, with a combined weight of 32g, wall thickness 7-8mm, all from Pollard’s excavation. Three of these are glued, joining body-sherds, one of which is marked G2. Another two body-sherds marked G2 join separately. Nine scraps with no obvious joins but assumed to come from the same vessel are marked G2. One of these has a very faint trace of ribbing and it might be that these sherds belong to a LRA1, possibly to one of the other groups. The sherds are very worn and may have been exposed to heat.

‘Group 6’ sherds are the two ‘buff amphora’ sherds recovered from the ‘stone level’ and ‘blackened sand’ layers within the Ditch in Trench B (Pollard 1966, 43). LRA2 Group 7: Hard, sandy fabric with abundant fine to medium sub-angular voids, fine, rounded black sand, sparse sub-rounded quartz, and very sparse gold mica. Colour is reddish yellow (7.5YR 7/6).

Group 10: Hard, very smooth, close grained fabric. Very fine inclusions, abundant very fine, well-sorted sand and sparse mica. Pale colour; light grey (2.5Y 7/2) to pale brown (2.5Y 8/3).

One, abraded body-sherd, weight 24g, wall thickness 8mm, bearing an ‘RK’ stamp. Although it was illustrated in Pollard 1966, Figure12.3, this sherd was mistakenly assigned to Carter’s excavation. The sherd is identified as Bi (LRA2) by Thomas and described as relatively light and thin (Pollard 1966, 53). There are three closely-set, thin and shallow grooves on the exterior of the sherd which seem scratched, rather than deeply combed. The end of the sherd with the fourth groove visible in the 1966 illustration is now missing. The hard fabric and grooving suggests LRA2 but as this is a small body-sherd, this cannot be classified with certainty.

A total of four sherds with a combined weight of 96g, wall thickness 5-7mm. One of these has a stamp bearing ‘High Peak’ but no initials and is catalogued by RAMM as excavated by D’Urban. This thin sherd has very faint, very shallow, raised ridges on exterior and a faint groove on interior. A second, larger sherd with an ‘RK’ stamp has wide, horizontal, shallow grooves on the exterior as well as a diagonally scratched groove made with a rounded implement prior to firing. Either of these might be the ‘large buff sherd of thin hard fabric’ (not illustrated) assigned by Thomas to his class Biii and attributed at the time to Carter’s excavation (Pollard 1966, 53). The remaining two sherds also have ‘RK’ stamps; one of these has no ribbing but seems to be from towards the shoulder of an amphora while the other has a faint groove on the exterior. It is not certain that these sherds are from the same vessel.

Group 8: Fine grained, hard fabric, with moderate, very ill-sorted, sub-rounded, medium to coarse quartzite, rare very fine gold mica and fine limestone. The fabric is light red (2.5YR 6/6) but the exterior is paler, pink (5YR 7/3) and possibly slipped.

A further body-sherd, 20g, wall thickness 7mm, bearing an ‘RK’ stamp could not be confidently matched to another group due to abrasion, but seemed to have an uneven exterior surface.

A single body-sherd, weight 11g, wall thickness 7mm with horizontal, squared grooves on part of the exterior and traces of ridges on the interior. The sherd was illustrated by Hutchinson and in Pollard 1966 Figure 12.1, at which point it was identified by Thomas as Bi (LRA2) and described as having ‘stylus grooving’ (Pollard 1966, 53). Again a tentative identification of LRA2 is possible.

Unidentified sherd (UnID) Group 11:

There is a further, small, ‘unassigned antiquarian’ bodysherd in a similar, though more micaceous, fabric in the RAMM assemblage with a trace of a possible ridge on the exterior. Weight 4g, wall thickness 8-9mm

Hard, coarse fabric with rough feel and uneven fracture. Frequent, coarse dark and light grey coarse sub-rounded rock inclusions, medium angular white quartzite and red ironstone. Pink (7.5YR 7/4)

Unidentified Late Roman amphora (LRA)

One distinctive body-sherd in a coarse fabric, 15g, wall thickness 10mm. No ribbing on exterior, interior surface very worn. Possibly also from an amphora, but if so, from a thicker vessel in a distinctive coarse fabric.

These groups could not be matched with certainty to a specific type of Late Roman amphora due to a lack of diagnostic elements and have been classified as Unidentified LRA.

Sherds recovered by AC archaeology in 2012 One very abraded small body-sherd, 3g, with faint traces of ridging. Wall thickness 7mm.

Group 9: Hard, but crumbly fabric with very rough feel. Abundant, ill-sorted fine to medium quartz. Light brown to brown (7.5YR 6/4 to 5/4).

One abraded body-sherd, 6g, of a late Roman 1 amphora (LRA1), with traces of ridging suggesting that the sherd is 190

Appendices marine growth suggests that the amphora was, at least for some time, buried upright (in the position it was discovered) with the upper portion partially exposed. It was not possible to make a fresh break, but the fabric has a hard, sandy feel and, a buff colour (Munsell grey, 5Y 5/1) and appeared to contain moderate, fine, rounded black sand and angular white grains (possibly limestone) with some fine red rock fragments and rare, sub-angular quartz.

from toward the central section of the vessel. Wall thickness 7-8mm. Both sherds have are reddish yellow (5YR 6/6) and have a soft, sandy fabric with ill-sorted inclusions. Frequent, fine to medium sub-rounded voids (presumably from limestone inclusions). Abundant, fine, black sand, moderate, fine, red ironstone and fine, sub-angular, quartz and white quartzite. Both of these sherds are likely to belong to the same vessel as a Late Roman 1 amphora partially recovered by Pollard during excavations in the 1960s, in ‘Trench B’, and have consequently been classified here with ‘Group 1’. The sherds were recovered in 2012 from a trench dug adjacent to Pollard’s Trench B, and found within the recently identified ‘outer ditch’. It seems, therefore, that what Pollard identified as ‘dark age cooking area II’ on the terrace (which produced the bulk of the ‘Group 1’ sherds) might instead represent deposits from the edge of this ‘outer ditch’ (Rainbird et al. 2013, 47).

The weight of the surviving portion is 1398g with a height of 154mm. The diameter of the body at its widest point is 228 mm while the diameter of the rim is 97mm (exterior) and 77mm (interior). The neck is 110mm high with an internal diameter of 101mm at the junction of the neck and body. At the shoulder the wall thickness is 10mm, but elsewhere is 6-7mm. The two grooved and slightly twisted handles, as typical for LRA1, measure up to 35mm wide by 29mm deep and join the neck high up just under the rim. There are traces of horizontal ribbing on the exterior of the body, starting just below the point where the handles are attached. The interior surface also has grooves typical for this type of amphora (Williams 2005, 159).

Nine joining sherds, 34g, of a vessel with fairly closely spaced, rounded, horizontal ribbing on the exterior (see Figure 4.12). Wall thickness 7mm.

Amphora residue sample 12A10 analysed by Steve Hill (Hill 2013)

Soft, close-grained fabric with a smooth, powdery feel and fine inclusions. The sherds are very abraded. Moderate, sub-rounded voids, sparse, fine limestone and very fine gold mica. The fabric and ribbing suggest that these sherds are from a LRA1. The narrow-spaced ribbing suggests that the sherds are probably from towards the top or bottom of the vessel, which is also indicated by the curvature of the sherd and the angle of the ribbing. The fabric is reminiscent of the ‘LRA1 type’ fabric recognised at Bantham (Bidwell et al. 2011, 97-9), although the sherds are slightly more ‘yellowish’ in colour (10YR 8/4) and are not the same vessel as ‘Group 6’ assigned within the wider site assemblage. As these sherds do not appear to match to any within the earlier collections held at RAMM, they have been assigned as ‘Group 5’ within the wider site assemblage. They were recovered from basal fills of a linear feature at the north of the site, which might also be part of the ‘outer ditch’ or relate to an entrance to the site, which has not been identified to date (Rainbird et al. 2013, 48).

‘Analysis of the residue observed on the interior surface of the Cawsand amphora was carried out at Plymouth University using photoacoustic spectroscopy, a method based on the absorption of modulated infrared light leading to the local warming of the absorbing gas. The generated pressure waves are detected by a pressure detector (i.e. microphone) producing a signal proportional to the absorption (Hill 2013). Photoacoustic spectroscopy was used as a scanning technique to characterise the sample against library materials at Plymouth University, although it would require authenticated standards for a fully validated comparison. Mass spectrometry would have been more robust, but would have required more work and increased costs. The uncarbonised parts appear to be a clear red resin under the microscope. The sample responded well with the photoacoustic cell and produced a good spectrum with some phenolic activity but interestingly no traces of triglycerides. The latter would be present in olive oil. There is also some unsaturation. Tartaric acid is the usual signature for wine as it occurs in all grapes; tartaric acid was identified in the sample. The best fit for the spectroscopic information would therefore be wine. If the resin was hydrocarbon based, it would have absorbed some of the contents in transit and this might explain this result. Ascorbic acid residues can be found in white wine but not in red, although this might not necessarily survive in ancient white wine samples. The absence of ascorbic acid in the sample possibly suggests red wine’. (Hill 2013)

Appendix F: Cawsand amphora (See Figure 4.13 and Figure 4.14) Vessel description The colour is variable; the exterior surface is light browngrey with darker grey patches (varying Munsell light grey, 5YR 7/1 and pale yellow, 5Y 8/2) but is noticeably more orange-brown toward the rim and paler on the body below the handles. This colour difference on the vessel and the

191

192

-

Cawsand Amphora Duggan 2013c

(20)

-

-

Trethurgy Quinnell 2004; Campbell 2011

Cadbury Congresbury Rahtz et al. 1992 Campbell 2011

South Cadbury Alcock 1995 Campbell 2011

-

-

High Peak, Sidmouth Pollard 1966 Duggan 2013a

Dinas Powys Campbell 2011

1(1)

Mothecombe Totals Duggan 2012

Comparative sites

-

Bantham Totals pre-2001 Bidwell et al. 2011

South Devon Sites

-

Tintagel: Thomas 1988 Thorpe 2007 Campbell 2011

Undated Amphora

Tintagel

Appendix G

7(43) 5

3 (175) 3

2 (47) 2

3 (51)

1 (1)

(76) 4/5+

5 (18)

(28?) 20 (482)

35 (375) 29

LRA1 (Bii)

-

-

-

-

1

-

6 (91)

-

LRA1 ‘Banthamtype’

7(101) 4

2? (215?) 2

(15) 2

1 (30)

-

(7) 2?

2(3)

(1?) 1(1)

40 (914) 34

LRA2 (Bi)

1(7) 1

1 (9) 1

1 (2) 1

-

-

-

-

-

5? (19) 3

LRA3 (Biv)

-

-

-

1 (1)

-

-

-

0 2 (15)

-

LRA4 Gaza/(Bvi)

-

-

1 (12)

-

-

-

1? (1)

(4) 5 (34)

27+? (118) 23

‘Thickwalled’

‘Bv’

1

1? (1)

2 (10)

‘North African’

0 1

? 1

1

1(110)

-

(1) 2+?

1? (1)

(14) 16 (59)

34? (288?) 10

UnID. LRA

15(151) 11

7+ (547?) 7

6? (76) 7

6 (192)

-

(84) 9/10+?

10? (24)

(47) 52 (692)

141 150 (1714) 99

Total early medieval amphora

Links to Late Antiquity

Appendix G: Quantities of amphorae from sites in South Devon, Tintagel and selected sites in south-west Britain.

Sherd counts in parentheses. ‘Bv’ figures split between ‘thick-walled’ and ‘North African’ amphorae where specified. Tintagel vessel counts for ‘Unidentified LRA’ from Thomas 1988 includes possible coarseware. Tintagel sherd counts from Thorpe 2007 refer to 1990-9 totals only, estimated vessel counts refer to entire assemblage. Bantham pre-2001 totals based on figures in Bidwell et al. 2011, 110.

Appendices Appendix H: Ceramic distribution maps. After Campbell 2007a; 2011: Doyle 2009; Thomas 1981; Fernández 2014; Rigoir 1973 and all references listed in Appendix K. Any transparency is used for clarity, but is not indicative. H1.1 ARS-D: ‘Atlantic’ distribution Not scaled

193

Links to Late Antiquity H1.2 ARS-D: sites with specified late-fifth/sixth century forms Not scaled

194

Appendices H1.3 ARS-D: sites with specified late-sixth/early-seventh century forms Not scaled

195

Links to Late Antiquity H.2 LRC: relative Atlantic distribution Proportional scaling. Figures by MNV or presence attested

196

Appendices H.3 LRD: relative Atlantic distribution Proportional scaling

197

Links to Late Antiquity H.4 DSPA: distribution Not scaled

198

Appendices H.5 LRA1: relative Atlantic distribution Proportional scaling

199

Links to Late Antiquity H.6 LRA2: relative Atlantic distribution Proportional scaling

200

Appendices H.7 LRA4: relative Atlantic distribution Proportional scaling

201

Links to Late Antiquity H.8 ‘Late’ North African amphorae: Atlantic distribution Not scaled

202

Appendices H.9 E ware: relative distribution Proportional scaling

203

Links to Late Antiquity Appendix I: Schematic/speculative maps of supply and exchange After Abu-Lughod 1989; Naviero Lopez 1991; Reynolds 1995; Campbell 2007a; Fernández 2014 and discussions in Chapter Seven Appendix I.1 Schematic of Atlantic supply routes

204

Appendices Appendix I.2 Schematic of Atlantic circuits of exchange. Showing northern and southern ‘Atlantic Seaboard circuits’ (labelled 1 and 2) and speculative West and East Mediterranean circuits (3 and 4)

205

Links to Late Antiquity Appendix J: E ware minimum vessel totals after Campbell 2011 and Doyle 2009. Data from Campbell 2011 and Doyle 2009, with additional data for Continental sites (see Appendix K)

Site ID No.

Site

Country

Total

36

Rathmullan

Ireland

2

37

Lisdoo

Ireland

1

38

Marshes Upper No. 3

Ireland

2

1

Craig Phadraig

Scotland

1

39

Moynagh Lough

Ireland

2

2

Dun Ardtreck

Scotland

1

40

Colp West

Ireland

8

4

Dunollie

Scotland

4

41

Knowth Site M

Ireland

1

5

Dundurn

Scotland

1

42

Knowth

Ireland

2

6

Clatchard Craig

Scotland

1

43

Painestown

Ireland

2

7

Iona

Scotland

1

45

Ninch

Ireland

7

8

Ederline

Scotland

1

46

Randalstown

Ireland

3

9

Ardifuir

Scotland

1

47

Smithstown

Ireland

1

10

Dunadd

Scotland

25

48

Flemington

Ireland

1

11

Loch Glashan

Scotland

5

49

Castletown Tara 1

Ireland

1

12

Dumbarton Rock

Scotland

3

50

Rosepark

Ireland

4

13

Little Dunagoil

Scotland

2

51

Collierstown

Ireland

1

14

Buiston

Scotland

2

52

Killucan

Ireland

2

15

Dundonald Castle

Scotland

1

53

Garretstown 2

Ireland

2

16

Lochlee

Scotland

1

54

Lusk

Ireland

3

17

Kildalloig Dun

Scotland

1

55

Roestown 2

Ireland

2

18

Trusty’s Hill

Scotland

1

56

Lagore

Ireland

5

19

Mote Of Mark

Scotland

12

57

Ratoath

Ireland

1

20

Whithorn

Scotland

16

58

Gracedieu

Ireland

1

21

Kiondroghad

Isle of Man

3

59

Raystown

Ireland

1

22

Port y Candas

Isle of Man

1

60

Summerhill Demense

Ireland

2

61

Ballinderry no. 2

Ireland

1

23

Teeshan

Ireland

1

62

Clonmacnoise

Ireland

3

24

Gransha

Ireland

1

63

Dalkey Island

Ireland

P+

25

Langford Lodge

Ireland

2

64

Mount Offaly

Ireland

2

26

Scrabo Hill

Ireland

2

65

Rathgureen

Ireland

1

27

Dunnyneill Island

Ireland

1

66

Lugacurren

Ireland

1

28

Clogher

Ireland

5

67

Iniscealtra

Ireland

1

29

Ballyfounder

Ireland

1

68

Killederdadrum

Ireland

1

30

Armagh, Castle St.

Ireland

1

70

Derrynaflan

Ireland

1

31

Slievegrane Lower

Ireland

1

72

Kedrah

Ireland

1

32

Downpatrick

Ireland

1

73

Lisleagh 1

Ireland

1

33

Spittal Ballee

Ireland

1

74

Lisduggan North No.1

Ireland

2

34

Lough Faughan

Ireland

1

75

Reask

Ireland

1

206

Appendices Site ID No.

Site

Country

Total

Site ID No.

Site

Country

Total

77

Garryduff No. 1

Ireland

9

201

France

P++?

78

Caherlehillan

Ireland

3

Bordeaux, Place Camille Jullian

79

Church Island

Ireland

1

204

Bordeaux, Rue du Hâ

France

P?

80

Garranes

Ireland

2

267

Vigo, UARC II

Spain

44

82

Ballycatteen

Ireland

3

88

Carew Castle

Wales

2

89

Longbury Bank

Wales

5

90

Portclew Chapel

Wales

1

91

Caldey Island

Wales

1

92

Linney Burrows

Wales

1

93

Hen Gastell

Wales

3

94

Lesser Garth

Wales

1

96

Dinas Powys

Wales

12

98

Carhampton

England

1

118

The Kelsies

England

1

120

Trethurgy

England

1

125

Bantham

England

3

127

Gwithian

England

6

134

Tean

England

9

135

May’s Hill

England

4

136

Dial Rocks

England

1

137

Dolphin Town

England

1

138

Bryher

England

1

139

Bar Point

England

2

140

Samson

England

12

157

Le Yaudet

France

1

158

Guissény

France

1

169

Tours

France

P?

172

Rezé

France

P??

173

Les Cléons

France

2

183

Herpès

France

2

184

Saintes

France

P+?

186

Chadenac

France

1

187

Jonzac

France

5

198

Bordeaux, Place Gabriel

France

P++?

199

Bordeaux, Puy Paulin

France

P++?

200

Bordeaux, Îlot Bonnac

France

P+?

207

Randalstown

Collierstown

Lusk

Gracedieu

Dalkey Island

Mount Offaly

Iniscealtra

Blanchfieldsland

Derrynaflan

Rock of Cashel

Reask

46

51

54

58

63

64

67

69

70

71

75

Lisnacreevy

35

Ninch

Clogher

28

45

Whithorn

20

Stalleen

Mote Of Mark

19

44

Dumbarton Rock

12

Colp West

Dunadd

10

40

Iona

7

Site ID No.

Rhynie

208

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Ireland

Scotland

Scotland

Scotland

Scotland

Scotland

Scotland

Country

3

4

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

3

3

1

1

1

1

1

1

1

1

1

3

1

1

1

1

1

1

2

2

2

2

2

1

1

1

1

1

2

P+

1

3

1

3

7

8

5

16

12

3

25

1

Doyle 2009

Doyle 2009

Doyle 2009; Campbell 2011

Doyle 2009

Doyle 2009; Campbell 2011

Doyle 2009

Doyle 1998; 2009

Doyle 2009; Campbell 2011

Doyle 2009; Campbell 2011

Kelly 2010

Doyle 2009; Campbell 2011

Doyle 2009

Doyle 2009

Doyle 2009; Campbell 2011

Doyle 2009

Warner 1979; Doyle 2009

Hill 1997

Campbell 2006; Laing and Longley 2006

Alcock and Alcock 1990

Craw 1930; Lane and Campbell 2000

Reece 1981

Noble et al. 2013

Primary data sources

By MNV. British data after Campbell 2011, expanding on Thomas 1981; Irish data supplemented by Doyle 2009. ‘Bantham-type’ included with LRA1.

Links to Late Antiquity

Appendix K: Atlantic sites with Mediterranean imports and DSPA (E ware totals shown from these sites) and Continental sites with E ware.

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

DSPA

LRD

LRC

ARS-D

Site

Cadbury Congresbury

Carhampton

Lundy

Cannington Cemetery

Glastonbury Abbey

Glastonbury Mound

Glastonbury Tor

Athelney

Woolston

South Cadbury

98

99

100

101

102

103

104

105

106

Longbury Bank

89

97

Coygan Camp

87

Dinas Powys

New Pieces

86

96

Dinas Emrys

85

Llandough

Cefn Cwmwd

84

95

Deganwy

83

Hen Gastell

Russelhill

81

93

Garranes

80

Site

Caldey Island

Caherlehillan

78

209

Country

England

England 3

P?

P?

9

LRC

England

England

England

2

2

4

1

9

3

2

1

1

1

5

1

1

2?

1

1

1

3

3

1

1

2

8

LRA1

England

England

England

England

England

4

1

1

2

1

1

2

1

4

1

1

2

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1+

2

4?

1

1?

North African

Wales

Wales

Wales

Wales

Wales

Wales

Wales

Wales

Wales

Wales

Ireland

Ireland

Ireland

Ireland

LRA

91

Kilgrovan

Site ID No.

76

1

12

3

1

5

2

3

Alcock 1995

Costen 2011; Campbell 2011

Campbell 2011

Rahtz 1970

Rahtz 1974

Allan 2011

Rahtz et al. 2000

Allan and Blaylock 2005

Hollinrake 2007; Campbell 2011

Rahtz et al. 1992

Alcock 1963; Campbell 1988; 1991

Holbrook and Thomas 2005

Wilkinson 1995

Campbell 1989

Campbell and Lane 1993

Campbell 1988; 2011

Campbell 2011

Savory 1960; Campbell 1988

Cuttler et al. 2012

Campbell 1988; 2011

Doyle 2009

Ó’Ríordáin 1942; Doyle 2009

Doyle 2009; Campbell 2011

Doyle 2009

Primary data sources

Appendices

E ware

‘Thick-walled’

‘Bv’

LRA?

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

DSPA

LRD

ARS-D

Pevensey Castle

Druce Farm

Tintagel

High Peak

Lydford

Lellizzick,Harbour Cove

Harbour Cove, Padstow

St Merryn

Killibury

Mawgan Porth

The Kelsies

Hay Close

Trethurgy

Perran Sands

Looe Island

Cawsand

Mothecombe

Bantham

St Michael Caerhays

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

Site ID No.

108

Site

Ham Hill

210

England

England

1

2

1

2

P?

26

5

1

2

1

1

1

2

1

2

1

2

1

1

2

34

1

3

2

16

1

1

1

2

10

1

2

1

1?

23

5

1?

‘Thick-walled’

England

England

England

England

England

England

England

England

England

England

England

England

6

29

LRA1

1?

4

LRA2

England

LRC 28

1

LRA

England

ARS-D

16

3

1

‘Bv’

England

England

England

England

Country

107

3

1

1

Thomas 1981; Olson 1989

Fox 1955; Silvester 1981; Reed et al. 2011; Bidwell et al. 2011

Fox 1961; Turner and Gerrard 2004; Agate et al. 2012; Duggan 2012

Duggan 2013c

Todd 1983; Wessex Archaeology 2009

Thomas 1981; Olson 1989

Quinnell 2004

Campbell 2011

Thomas 1981; Olson 1989

Thomas 1981; Bruce-Mitford 1997

Miles 1977; Thomas 1981; Olson 1989

Thomas 1981; Olson 1989

Thomas 1981; Olson 1989

Wessex Archaeology 2008

Thomas 1981; Olson 1989

Pollard 1996; Rainbird et al. 2013; Duggan 2013a

Radford 1956; Thomas 1981; 1988a; Thorpe 2007

Ladle pers. comm.

Fulford and Rippon 2011; Timby 2011

Burrow 1981

Primary data sources

Links to Late Antiquity

E ware

North African

LRA?

‘Bag-Shaped’

LRA 7

LRA4

LRA3

DSPA

LRD

Soissons

Limé

Bayeux

Reims

Limetz-Villez

Château-Thierry

Les Mureaux

Châlons-sur-Marne

Paris, General location

Paris, Collège de France

145

146

147

148

149

150

151

152

153

154

Samson

140

Mercin-et-Vaux

May’s Hill

135

144

Tean

134

Rouen

Halliggye

133

143

Grambla

132

Cherbourg

St Michael’s Mount

131

142

Chun Castle

130

Wailly-les-Arras

Tremough

129

Site

141

Phillack

128

Site ID No.

211

France

France

France

France

France

France

France

France

France

France

France

France

P?

P?

1

1

ARS-D

France

France

England

England

England

England

England

England

England

England

England

England

Country

Gwithian

1

1

1+

LRC

127

P

1

P

2

P

P

P

P

1

3

P

P

1

1

1

1

1

1

4

1

1

2

1

1

2

1

1

1

12

4

9

6

Pieri 2005

Chaidron 2008; Brulet et al. 2010

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008

Chaidron 2008; Rigoir et al. 1973

Chaidron 2008; Brulet et al. 2010

Chaidron 2008

Thomas 1985

Thomas 1985

Thomas 1985

Quinnell and Elsdon 2009-10

Olson 1989

Herring 2000

Thomas 1981

Gossip and Jones 2010

Thomas 1981; Olson 1989

Nowakowski et al. 2007

Primary data sources

Appendices

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

DSPA

LRD

Moncelon

Aubigné-Racan

Blois

Tours

Couffi

Nantes, General Location

Rezé

Les Cléons

Bourges

Curçay

167

168

169

170

171

172

173

174

175

Le Tuileau

162

166

Châtenay-sur-Seine

161

Mornay

Chartres

160

165

Melun

159

Orléans

Guissény

158

164

Le Yaudet

157

Ouzouer-le-Marché

Jouars-Pontchartrin

156

163

Île Lavret (Bréhat)

Site ID No.

155

212

France

France

France

P

P

P

P

1

1

1

1

1

2

P??

P?

1

1

E ware

France

France

France

P++

P

P

P

P

P

P

P

P

P

P

DSPA

France

France

France

France

France

P?

ARS-D

France

France

France

France

France

France

France

France

France

France

Rigoir et al. 1973

Rigoir et al. 1973

Giot 1973; Campbell 2007a

Pirault 1997; Pieri 2005

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973; Randoin 1981; Riley 1981; Pieri 2005; Campbell 2007a

Rigoir et al. 1973

Pieri 2005

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973; Brulet et al. 2010

Rigoir et al. 1973

Rigoir et al. 1973

Chaidron 2008

Chaidron 2008

Chaidron 2008

Giot 1973; Fleuriot and Giot 1977; Campbell 2007a

Cunliffe and Galliou 2007; Campbell 2007b

Chaidron 2008

Giot and Querré 1985; Pieri 2005; Campbell 2007a

Primary data sources

Links to Late Antiquity

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

LRD

LRC

Country

Site

Petit-Bersac

Neuvicq-Montguyon

Périgueux

Clos de Chardonnet

Les Grandes Vignes

Saint-Georges de Montagne

Libourne

Saint-Laurent-desCombes

189

190

191

192

193

194

195

Saintes

184

188

Herpès

183

Jonzac

Limoges

182

187

Chateâu d’Oléron

181

Chadenac

Saint-Jean-d’Angély

180

186

Saint-Bertrand-deCommiges

179

213

France

France

France

France

France

France

France

France

France

France

France

France

France

France

France

France

France

France

P

P

P

P

P

P

P

P+?

P

P

P

P

P

P

P

DSPA

Saint-Fort-sur-le Né

Poitiers

178

France

France

2

2

3

1

5

1

P+?

2

E ware

185

Yzeures

177

Site ID No.

Loudun

Site

176

Hochuli-Gysel et al. 1995; Pieri 2005; Campbell 2007a

Amiel and Berthault 1996

Amiel and Berthault 1996

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973; Amiel and Berthault 1996

Rigoir et al. 1973

Rigoir et al. 1973

Guériteau 2006a; Robin 2010; Campbell 2011

Campbell 2007a; 2011

Rigoir et al. 1973

Rigoir et al. 1973; Guitton forthoming; pers. comm.

Wooding 1996a; Campbell 2007a

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Primary data sources

Appendices

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

LRD

LRC

ARS-D

Country

Bordeaux, Rue du Hâ

Andernos

Neujon

Lamothe

La Molège

Montréal du Gers

Lectoure

Dax

Auch

204

205

206

207

208

209

210

211

212

Bordeaux, Place Camille Jullian

201

214

France

France

France

France

France

France

France

France

P

1

1

P++

P++

6

P

P

P

P

See individual sites

LRD

France

13

P+?

LRA1

France

France

France

France

ARS-D

Bordeaux, General Location

Bordeaux, Îlot Bonnac

200

France

P

DSPA

203

Bordeaux, Puy Paulin

199

France

2

1

1

1

1

3

12

5

LRA4

Bordeaux, SaintChristoly

Bordeaux, Place Gabriel

198

5

1

4

3

1

6

2

P+

19

1

2+

North African

France

France

P?

P++?

P+?

P++?

P++?

E ware

202

Bordeaux, SaintSeurin

197

Site ID No.

Montcaret

Site

196

Amiel and Berthault 1996

Amiel and Berthault 1996

Amiel and Berthault 1996

Amiel and Berthault 1996

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Rigoir et al. 1973

Guériteau 2013

Rigoir et al. 1973; Gauthier 1983; Berthault 1999; 2012; pers. comm.; Pieri 2005; Maurin (ed) 2012; Bonifay 2012; Labrouche 2012; Soulas 2012; Girond 2012; Guériteau 2006a;b; 2012; 2013; forthcoming

Gauthier 1983; Pieri 2005; Berthault pers. comm.

Pieri 2005; Maurin (ed) 2012; Berthault 1999; 2012; Bonifay 2012; Labrouche 2012; Soulas 2012

Guériteau 2006a; Guériteau forthcoming; Campbell 2011

Girond 2012; Guériteau 2012

Guériteau 2006b

Watier 1973; Amiel and Berthault 1996; Pieri 2005; Campbell 2011; Berthault 2012; pers. comm.

Amiel and Berthault 1996; Berthault 1997

Primary data sources

Links to Late Antiquity

‘Thick-walled’

‘Bv’

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA3

LRA2

LRC

Country

215

A Coruña, Rúa da Franxa 18

A Coruña, Intervención en la Casa Martelo

A Coruña, Rúa de Franxa 9-11

A Coruña, Praza de María Pita

A Coruña, Rego de Auga no.60

223

224

225

226

227

Noville

219

A Coruña, Brigantium, General location

Mugardos

218

222

Villa de Veranes

217

Site

A Coruña, Torre de Hércules

Coaña

216

Spain

Spain

Spain

Spain

ARS-D

Spain

2

1

LRC

2

1

P

P

1

1

See individual sites

1

1

P

P

3

LRD

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

2

5

1?

9

P

DSPA

221

Gijón, Termas de Campo Valdés

215

Spain

P

2

1?

1

LRA1

Cabo de Higuer

Gijón, Muralla de Cimadevilla

214

Spain

1

1?

LRA4

220

Necropolis of Cadabarcos, Bares

Site ID No.

213

López Pérez 2004

López Pérez 2004

López Pérez 2004

López Pérez 2004

López Pérez 2004

López Pérez 2004

López Pérez 2004

Benito 1988; Fernández Ochoa et al. 1992; Uscatescu et al. 1994; Azkarate et al. 2003; Prieto 2005; Fernández 2007; 2014

Naveiro Lopez 1991; López Pérez 2004; Fabião 2009

López Pérez and César Vila 2010; Fernández 2014

Fernández Ochoa et al. 2006

Alonso Sánchez and Fernández Ochoa 1988

Jarrega Dominguez 1991; Fernández Ochoa et al. 1992; Prieto 2005

Fernández Ochoa et al. 1992; Prieto 2005

Naveiro Lopez 1991; López Pérez and César Vila 2010; Fernández 2014

Primary data sources

Appendices

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA3

LRA2

Country

Sobrado dos Monxes

Cueva de Abauntz

Lugo

Cueva de Iruaxpe III

Albeiurmendi

Iruña

236

237

238

239

240

241

Peña Forua

233

Cidadela

Murias de Paraxuga

232

235

A Coruña, Villa Romana de O Cantón Grande

231

216

Spain

Spain

2

1

1

P

P?

3

1

8+

2

1?

P+?

2?

P?

8?

2

P?

LRA2

Spain

P

1

3

3

3

P

1

P?

LRA4

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

2

LRC

Moraime

A Coruña, Calle Tabernas no. 10-12

230

Spain

Spain

DSPA

234

A Coruña, Dársena de la Marina

229

Site ID No.

A Coruña, no.8 de la Calle Damas

Site

228

P

Jarrega Dominguez 1991; Fernández Ochoa et al. 1992; Uscatescu et al. 1994; Filloy Nieva 1997; Filloy Nieva and Gil Zubillaga 1997; Prieto 2005

Filloy Nieva 1997

Uscatescu et al. 1994; Azkarate et al. 2003; Prieto 2005; Fernández 2007; 2014

Naveiro Lopez 1991; López Pérez and César Vila 2010; Fernández 2014

Uscatescu et al. 1994; Prieto 2005; Fernández 2007; 2014

Jarrega Dominguez 1991

Naveiro Lopez 1991; López Pérez 2004

López Pérez 2004; López Pérez and César Vila 2010; Fernández 2007; 2014

Uscatescu et al. 1994; Prieto 2005; Fernández 2007; 2014

Uscatescu et al. 1994; Prieto 2005; Fernández 2007; 2014

López Pérez 2004; López Pérez and Vázquez Collazo 2007

López Pérez and Botella 2010

López Pérez 2004

López Pérez and Muiño Maneiro 2011

Primary data sources

Links to Late Antiquity

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA3

LRA1

LRD

ARS-D

Country

217

Oyón

Adro Vello

Astorga

A Lanzada

Bueu

O Facho de Donón

252

253

254

255

256

257

Cacabelos

248

Perezuelas

Zambrana

247

251

Torres Oeste

246

La Iglesia

Taboada

245

250

Iria Flavia

244

Camponaraya

La Miel

243

249

Kobairada

Site ID No.

242

Spain

Spain

Spain

Spain

P

P

P

4

1

2

1

P

P

1

P

P

9

1

3

P

P

1

P?

DSPA

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

P

P

López Pérez and César Vila 2010; Fernández 2014

López Pérez and César Vila 2010; Fernández 2014; pers.comm.

Naviero Lopez 1991; Fernández Ochoa et al. 1992; López Pérez and César Vila 2013; Fabião 2009; Fernández 2014

Paz Peralta 2004

Fernández 2007; 2014

Filloy Nieva 1997

Filloy Nieva 1997

Filloy Nieva 1997

Alonso Sánchez and Fernández Ochoa 1988

Alonso Sánchez and Fernández Ochoa 1988

Filloy Nieva 1997

Alonso Sánchez and Fernández Ochoa 1988; Fernández 2014

Fernández Ochoa et al. 1992; Fernández 2014

López Pérez 2004; López Pérez and César Vila 2010; Fernández 2014

Filloy Nieva 1997

Filloy Nieva 1997; Fernández 2007; 2014

Primary data sources

Appendices

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

LRD

LRC

ARS-D

Country

Site

Vigo, Parcela 13

Vigo, Villa de Toralla

Tui

Cerveira

Clunia

Sto. Amaro

Braga, General Location

Braga, Rua Gualdim Pais

270

271

272

273

274

275

276

Vigo, Parcela 23

266

269

Vigo, Rosalía no. 5

265

Vigo, Parcela 14

Vigo, Hospital no. 5

264

268

Vigo, Marqués de Valladares

263

Vigo, UARC II

Vigo, Aliviadero de la c/Colón

262

218

Country

Portugal

Portugal

Portugal

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

Spain

605

15

64

P

5

P++

P

P

P

1

34

P

2

P?

1?

2

4

Data given as a total for Vigo, general location

48

Data given as a total for Vigo, general location

579

2

4

1

3

17

North African

267

Vigo, UARC I

261

Site ID No.

Vigo, Areal 6-8

Site

260

ARS-D

Spain

LRC

Vigo, General location

LRD

Data given as a total for Vigo, general location

DSPA

259

Spain

LRA1

Vigo, Túnel de O Areal

LRA2

258

44

Morais 2005; Quaresma and Morais 2012

Delgado 1988; Melim de Sousa 2001; Morais 2005; Fabião 2009; Quaresma and Morais 2012; Delgado et al. 2014

Alonso Sánchez and Fernández Ochoa 1988

Fernández Ochoa et al. 1992; Uscatescu et al. 1994; Prieto 2005; Fernández 2007; 2014

Alonso Sánchez and Fernández Ochoa 1988

López Pérez and César Vila 2010; Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Fernández 2014

Primary data sources

Links to Late Antiquity

E ware

‘Thick-walled’

‘Bv’

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

Lousada

Fraçao

Recezinhos

Freixo

Mozinho

Porto

Vila Nova de Gaia

Boelhe

Cividade de Eja

Crestuma

Conimbriga

Egitania

Póvoa de Cós

Parreitas

Cardílio

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

Site ID No.

Falperra

219

Portugal

Portugal

Portugal

Portugal

ARS-D

P

P

P

P++

1

P+

P

P

P

P

P

1

LRC

Portugal

P

P

P

P

P

P

1

P?

LRD

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Country

277

1

P

P

1

Quaresma 2012

Quaresma 2012

Delgado 1988; Melim de Sousa 2001; Fabião 2009

Quaresma 2012

Delgado, Mayet and Moutinho de Alarcão 1975; Keay 1984; Delgado 1988; Reynolds 1995; Melim de Sousa 2001; Quaresma 2012

Silva and Guimarães 2011

Alonso Sánchez and Fernández Ochoa 1988

Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Alonso Sánchez and Fernández Ochoa 1988; Melim de Sousa 2001; Fabião 2009

Alonso Sánchez and Fernández Ochoa 1988

Delgado 1988; Melim de Sousa 2001; Fabião 2009

Delgado 1988; Melim de Sousa 2001; Fabião 2009

Alonso Sánchez and Fernández Ochoa 1988

Alonso Sánchez and Fernández Ochoa 1988

Delgado 1988; Naveiro Lopez 1991; Melim de Sousa 2001; Fabião 2009

Primary data sources

Appendices

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

DSPA

Site

Azeitada

Aldeia do Penedo

Monforte

Povos

Elvas

Armês

St André de Almoçageme

Casal de Colaride

Quinta da Bolacha

Miroiços

Cabanas

Alto da Cidreira

Freiria

Alto do Caparide

Lisbon, General location

Lisbon, Teatro Romano

Lisbon, Rua dos Correeiros

Casal do Clérigo

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

Site ID No.

220

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

5

8

P

P

5

6

P

P

1

1

5

P

1

2

23

1

1

8

LRC

Portugal

Portugal

Portugal

P

P

P

19

1

P?

LRD

Portugal

Portugal

Portugal

Portugal

Portugal

Country

Santarém

ARS-D

293

1

P 2

Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Grilo et al. 2013

Diogo and Trindade 1999; Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Fabião 2009; Quaresma 2012

Fabião 2009

Melim de Sousa 2001; Fabião 2009

Delgado 1988; Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Fabião 2009

Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Alonso Sánchez and Fernández Ochoa 1988

Quaresma 2012

Alonso Sánchez and Fernández Ochoa 1988

Melim de Sousa 2001; Fabião 2009

Quaresma 2012

Melim de Sousa 2001; Viegas 2003; Fabião 2009; Quaresma and Morais 2012

Primary data sources

Links to Late Antiquity

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

DSPA

Site

Beja

Monte da Salsa

Represas (Beja)

Chãos Salgados

Cidade das Rosas

Horta de D. Maria

Aljustrel

322

323

324

325

326

327

Convento de São Francisco

318

321

São Romão

317

Monte da Cegonha

Tróia

316

320

Fundeadouro de Tróia

315

221

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

P

149

P

532

6

32

ARS-D

São Cucufate

Tourega

314

Portugal

Portugal

1

1

6

2

P

P

2

1

P

P

P+

P

LRC

319

Comenda

Site ID No.

313

Site

Lisbon, Torre de Belém

Country

312

1

P

P

1

P

1

Alonso Sánchez and Fernández Ochoa 1988

Delgado 1988; Maia 1978; Delgado 1988; Melim de Sousa 2001; Fabião 2009

Delgado 1988; Maia 1978; Melim de Sousa 2001; Fabião 2009

Delgado 1988; Melim de Sousa 2001; Fabião 2009; Quaresma 2012; 2013

Delgado 1988; Melim de Sousa 2001; Viegas 2007; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Fabião 2009

Delgado 1988; Fabião 2009; Quaresma 2012; Quaresma and Morais 2012

Delgado 1988; Melim de Sousa 2001; Blázquez Martínez 2002; Viegas 2007; Fabião 2009

Fabião 2009

Fabião 2009

Maia 1974; 1978; Delgado 1988; Melim de Sousa 2001; Fabião 2009; Pinto et al. 2010; 2012; Magalhães 2012; Magalhães et al. 2014

Fonseca 2004

Pinto et al. 2004; Viegas 2007

Melim de Sousa 2001; Fabião 2009

Fabião 2009

Primary data sources

Appendices

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA4

LRA3

LRA2

LRA1

DSPA

LRD

222

Cerro da Vila

Loulé Velho

Marim

Faro

338

339

340

341

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

Portugal

London, Shadwell

London, Southwark

London, Billingsgate

*A

*B

*C

1

124

P

135

9

P

P

P

7

1

1

2

9

P

3

P

P

P

P

P

17+

1

1?

LRA4

England

England

England

Possible post-AD 410 imports in eastern Britain

Torre de Ares (Balsa)

Foz do Arade

334

337

Torrinha

333

Milreu

Cruzinha

332

336

Cacela-a-Velha

331

Site

Lagos

Montinho das Laranjeiras

330

Portugal

P?

LRC

335

Mértola

329

Site ID No.

Portugal

Country

Pessegueiro

ARS-D

328

P

P

1

P

Marsden 1980

Bird 1977

Douglas et al. 2011; Williams 2011b

Viegas 2007; 2009; Fabião 2009

Delgado 1988; Melim de Sousa 2001; Fabião 2009

Delgado 1988; Melim de Sousa 2001; Fabião 2009

Delgado 1988; Melim de Sousa 2001; Teichner 2003; Fabião 2009; Quaresma and Morais 2012

Delgado 1988; Melim de Sousa 2001; Viegas 2007; 2009; Fabião 2009

Fabião 2009

Ramos et al. 2007; Fabião 2009

Melim de Sousa 2001; Fabião 2009

Ramos et al. 2007

Fabião 2009

Fabião 2009

Melim de Sousa 2001; Viegas 2007; Fabião 2009

Delgado 1988; Alonso Sánchez and Fernández Ochoa 1988; Delgado 1992; Melim de Sousa 2001; Fabião 2009; Quaresma and Morais 2012; Fernandes 2013

Viegas 2007

Primary data sources

Links to Late Antiquity

E ware

‘Thick-walled’

‘Bv’

North African

LRA?

LRA

‘Bag-Shaped’

LRA 7

LRA3

LRA2

LRA1

DSPA

LRD