Excavations at the Minster Library, York 1997 9781407314570, 9781407323077

Table of contents :
Front Cover
Title Page
Copyright
Dedication
ACKNOWLEDGEMENTS
TABLE OF CONTENTS
FIGURES
TABLES
APPENDICES
LIST OF ABBREVIATIONS
ABSTRACT
CHAPTER 1: INTRODUCTION TO THE PROJECT
CHAPTER 2: A HISTORY OF THE YORK MINSTER CLOSE (NORTH SIDE)
CHAPTER 3: THE ROMAN FORTRESS (FIRST TO FIFTH CENTURY)
APPENDIX A: CATALOGUE OF ROMAN MATERIAL
CHAPTER 4: EARLY MEDIEVAL INTERLUDE (FIFTH TO ELEVENTH CENTURY)
CHAPTER 5: THE ARCHBISHOP’S PALACE (ELEVENTH TO SIXTEENTH CENTURY)
APPENDIX B: CATALOGUE OF MEDIEVAL MATERIAL
CHAPTER 6: THE FORMAL GARDEN OF SIR ARTHUR INGRAM (SEVENTEENTH TO NINETEENTH CENTURY)
APPENDIX C: DOCUMENTATION FOR SIR ARTHUR INGRAM’S FORMAL GARDEN
CHAPTER 7: DISCUSSION
BIBLIOGRAPHY
Back Cover

Citation preview

GARNER-LAHIRE

Excavations at the Minster Library, York 1997 Justin Garner-Lahire

EXCAVATIONS AT THE MINSTER LIBRARY, YORK 1997

Justin Garner-Lahire graduated in British and European Archaeology from the University of York in 1990. He began his professional field career at Sutton Hoo as an area supervisor, then went on to spend several years working in Italy, on sites in Lombardy, Liguria and Milan, as well as research excavations in France, Spain and north Africa. Justin is a founding director of FAS Heritage (est. 1993) and has since led excavations at a wide range of sites in Britain, including at Portmahomack (Highlands) with Martin Carver (1993–2007), York Minster Library (1997), Castle Sinclair Girnigoe (Caithness) (2004–2013), Eilean Donan Castle and Lincoln Castle (2008–2015).

2016

________

BAR 622

Reports of the open area excavation at the York Minster Library conducted in 1997 and associated research are presented in this book. The excavation was complemented by geophysical survey and detailed documentary research in advance of an extension to the Minster Library, a converted medieval archbishop’s chapel. The research unearthed archaeological strata up to two metres deep which have been attributed to eight distinct periods of activity. The periods of activity include evidence for Roman legionary barracks, an 11th-century boundary ditch, a 13th-century solar block of the archbishop’s palace, 17th-century pleasure gardens, and, finally, the conversion of the archbishop’s chapel into the Minster Library.

B A R

BAR British Series 622 2016

Excavations at the Minster Library, York 1997 Justin Garner-Lahire with contributions by Craig Barclay, Paul Bidwell, Martin Carver, Jonathan Clark, Hilary Cool, Brenda Dickinson, Sandra Garside-Neville, Kate Giles, Kay Hartley, Deborah Jaques, Patrick Ottaway, Barbara Precious, Nicky Rogers, Jennie Stopford, Alan Vince†, Felicity Wild

BAR British Series 622 2016

First Published in 2016 by British Archaeological Reports Ltd United Kingdom BAR British Series 622 Excavations at the Minster Library, York 1997

© Justin Garner-Lahire 2016 The Author’s moral rights under the 1988 UK Copyright, Designs and Patents Act, are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

ISBN 9781407314570 paperback ISBN 9781407323077 e-format DOI https://doi.org/10.30861/9781407314570 A catalogue record for this book is available from the British Library

Cover Image: York Minster Library, the former chapel of the palace of Archbishop Walter de Grey (1215–1255) (right), and the surviving arcade of the palace of Archbishop Roger of Pont l’Eveque (1154–1181) (left).

All BAR titles are available from: British Archaeological Reports Ltd Oxford United Kingdom Phone +44 (0)1865 310431 Fax +44 (0)1865 316916 Email: [email protected] www.barpublishing.com

For my parents, Audrey and Peter

ACKNOWLEDGEMENTS The archaeological research programme reported here was undertaken in 1997 by Field Archaeology Specialists Ltd (now FAS Heritage). The programme was sponsored and supported by the Dean and Chapter of York, and was carried out on behalf of the Department of Archaeology, University of York. Derek Phillips, then Minster Archaeologist, acted as archaeological consultant during fieldwork. Scheduled Monument Consent was granted by English Heritage and fieldwork was monitored by Keith Emerick. Planning permission was granted by City of York Council, and fieldwork was monitored by John Oxley. The late Richard Hall, Minster Archaeologist, oversaw the completion of the Field Report on behalf of the Dean and Chapter of York. William Birch & Sons Ltd were appointed the main contractor for the construction of the library extension and provided invaluable logistical support and cooperation during fieldwork. The author wishes to extend his warmest thanks to Martin Carver for his trust in FAS Heritage to undertake the research programme to his high standards of fieldwork. Martin also undertook the task of editing the technical Field Report into this research report. Copy editing and typesetting of this report was undertaken by Cecily Spall and Nicky Toop, FAS Heritage. The author is grateful to the team of professional archaeologists who worked on site: supervisors Andy Copp, Annette Roe, Toby LewisSimpson, Madeleine Hummler, Gigi Signorelli and Steve Timms, excavators Stewart Buchanan, James Brennan, Juliet Giecco, Jay Carver, John English, Candice Hatherley and Adam Exley and finds supervisors Louise Brennan and Cecily Spall. The author would like to thank the following for their specialist input (institutions acknowledged date to 2000: Barbara Precious and the late Alan Vince (Roman pottery; medieval pottery), Sandra Garside-Neville, Jennie Stopford and Kate Giles (building materials), Patrick Ottaway, York Archaeological Trust (iron objects), Hilary Cool (glass and antler objects), Nicky Rogers, York Archaeological Trust (non-ferrous metalwork) and Deborah Jacques, Environmental Archaeology Unit (vertebrate remains), Craig Barclay, York Museums Trust (coin identification), Frances Large, Darren Worthy, Allan Hall and John Carrott, Environmental Archaeology Unit (biological remains) and Erica Paterson and Julie Jones, York Archaeological Trust (conservation). The preparation of this publication, and the overview on the Roman period archaeology authored by Paul Bidwell, was generously supported by Elizabeth Hartley. Figure acknowledgements Preparation of the figures was undertaken by the author and Richard Jackson, FAS Heritage. Patrick Ottaway, John Blair and Christopher Norton kindly gave their permission to reproduce their illustrations, respectively: map of interventions in the fortress at York, Figure 3.1; map of the medieval topography of the Minster area; reconstruction of the Bishop’s hall at Hereford. Trevor Pearson illustrated the finds and the late David Hopkins illustrated the pottery. Specialist acknowledgements Barbara Precious and the late Alan Vince are grateful to the following specialists: Brenda Dickinson (Samian stamps), Kay Hartley (mortaria stamps), Felicity Wild (Samian ware) and David Williams (amphora). Maggi Darling provided general advice. Nicky Rogers wishes to acknowledge the contribution of Hilary Cool to the report on Roman metalwork.

TABLE OF CONTENTS 1

CHAPTER 1

INTRODUCTION TO THE PROJECT

CHAPTER 2

A HISTORY OF THE YORK MINSTER CLOSE (NORTH SIDE) - Kate Giles

10

CHAPTER 3

THE ROMAN FORTRESS (FIRST TO FIFTH CENTURY)

17

Period 1 - c. AD 71-c. 120, The timber barracks and trackway Period 2 - c. AD 120-c. 375, The stone barracks and road Period 3 - c. AD 375-c. 410, Decline and abandonment

17 22 26

The Roman assemblage The Roman pottery - Barbara Precious and Alan Vince Mortaria - Kay Hartley Decorated Samian - Felicity Wild Samian stamps - Brenda Dickinson Roman building materials - Sandra Garside-Neville Ferrous metalwork - Patrick Ottaway Non-ferrous metalwork - Nicky Rogers Coins - Craig Barclay The Roman glass - Hilary Cool Vertebrate remains - Deborah Jaques

36 36 50 56 60 60 63 65 67 67 71

APPENDIX A

CATALOGUE OF ROMAN MATERIAL

74

CHAPTER 4

EARLY MEDIEVAL INTERLUDE (FIFTH TO ELEVENTH CENTURY)

94

Period 4 - c. AD 410-c. 1075, Examining the ‘dark earth’

94

The Early Medieval assemblage The Early Medieval pottery - Alan Vince Antler objects - Hilary Cool

96 96 96

THE ARCHBISHOP’S PALACE (ELEVENTH TO SIXTEENTH CENTURY)

98

Period 5 - c. 1075-c.1240, Robbing of Roman walls and establishment of precinct boundary Period 6 - c. 1240-1618, The archbishop’s aisled hall

98 98

CHAPTER 5

The Medieval assemblage The Medieval pottery - Alan Vince Medieval building materials - Sandra Garside-Neville Floor tiles - Jennie Stopford A carved stone fragment - Kate Giles Post-Roman ferrous metalwork - Patrick Ottaway

106 106 108 109 109 110

APPENDIX B

CATALOGUE OF MEDIEVAL MATERIAL

111

CHAPTER 6

THE FORMAL GARDEN OF SIR ARTHUR INGRAM (SEVENTEENTH TO NINETEENTH CENTURY)

116

Period 7 - 1618-c. 1815, Ingram’s mansion and its garden

118

Period 8 - c. 1815-1997, The reuse of the archbishop’s chapel as the Minster Library The Post-medieval assemblage The Post-medieval Pottery - Alan Vince

125 125 125

APPENDIX C

DOCUMENTATION FOR SIR ARTHUR INGRAM’S FORMAL GARDEN - KATE GILES

126

CHAPTER 7

DISCUSSION

132

Introduction - Martin Carver Roman - Paul Bidwell Early Medieval - Martin Carver Medieval and later - Kate Giles Assessment - Martin Carver

132 132 133 134 136

Bibliography

137

FIGURES Location of Roman fortress and York Minster (fortress plan after RCHME) includes © Crown Copyright 1.1 Ordnance Survey. All rights reserved Location of interventions (evaluation and excavation) includes © Crown Copyright Ordnance Survey. 1.2 All rights reserved Location of survey areas (topographic and geophysical) includes © Crown Copyright Ordnance Survey. 1.3 All rights reserved 2.1 2.2 2.3 2.4 2.5 2.6 2.7

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 3.20 3.21 3.22 3.23 3.24

2 4 6

York, the Minster area, showing the principal features of the medieval topography (after Norton 1998a, 6, Figure 2) Map produced in support of a lawsuit in 1782 (Leeds Archives WYL100/YO/G) Extract from the Ordnance Survey map of 1852 © Crown Copyright Ordnance Survey. All rights reserved Reconstruction of the cathedral of Thomas of Bayeux (after Phillips in RCHME 1985, 142, Figure 25). Reproduced by permission of Historic England Surviving arcade in Dean’s Park (left) and chapel (right) with library extension (distant centre) looking northwest St Sepulchre’s Chapel by Moses Griffith, 1777. Image courtesy of York Museums Trust::http://yorkmuseumstrust.org.uk::Public Domain The Ingram mansion from the west, by E. Abbott. Pen and ink, watercolour and pencil on paper (The Gott Collection, Wakefield Council Permanent Art Collection)

12

Location of York Minster Library with Roman fortress plan beneath, also showing previous excavations within the fortress (after Ottaway 2004, 32, Figure 13) Plan of Period 1 structures and features, timber barracks and track way Plan of Period 2 structures and features, stone barracks Roman stone walls and footings (Periods 2 and 3, Structure 8, F116 and F117) Late Roman hearth (Period 3, F70) Late Roman pebbled surface (Period 3, C1118, C1128) set with eavesdrip drain (F88) and verandah postholes (F76, F73) Projected plan of Period 1 timber barracks and street, includes © Crown Copyright Ordnance Survey. All rights reserved. Projected plan of Period 2 and 3 stone barracks and street, showing Building 7, includes © Crown Copyright Ordnance Survey. All rights reserved. Postulated layouts of stone barracks across north corner of the fortress, includes © Crown Copyright Ordnance Survey. All rights reserved. Examples from Ceramic Groups 1 and 2 (Scale 1:4) Examples from Ceramic Group 2 (Scale 1:4) Examples from Ceramic Groups 3 and 4 (Scale 1:4) Examples from Ceramic Groups 4 and 5 (Scale 1:4) Examples from Ceramic Groups 5 and 6 (Scale 1:4) Examples of ceramics from post-Roman layers (Scale 1:4) Principal Component Analysis of assemblages from York Minster Library, by function Contribution of different forms to analysis displayed in Figure 3.16 Principal components analysis of assemblages by function (combining Samian and other forms) Selected samian fragments (Scale 1:4) Selected samian fragments (Scale 1:4) Roman brick with graffito and Legionary stamped tiles (Scale 1:4) Roman ferrous metalwork (Scale 1:2) Roman non-ferrous objects (Scale 1:2) Selected vessel glass fragments (Scale 1:2)

18

13 13 14 14 14 16

19 23 24 27 27 29 31 35 38 39 40 42 43 51 55 55 56 57 58 62 64 66 69

4.1 4.2

Plan of Period 4 features Antler objects (Scale 1:1)

5.1 5.2 5.3 5.4 5.5 5.6

Plan of Period 5 features Plan of Structure 6 and Structure 7 Medieval foundation (Period 6, F65) showing in the base of robber trench (Period 7, F37) Medieval pier base (Period 6, F21) Plan of Period 6 buildings, includes © Crown Copyright Ordnance Survey. All rights reserved. Fragment of tympanum, Find No.340 (Scale 1:2)

99 101 100 100 105 110

6.1 6.2 6.3 6.4

Plan of Period 7 features Reconstruction of Ingram’s mansion and gardens Post-medieval ornamental path (Period 7, F31) Magnetometer survey results showing topographic survey at 0.20m intervals, includes © Crown Copyright Ordnance Survey. All rights reserved Magnetometer survey interpretation feature map showing topographic survey at 0.20m intervals, includes © Crown Copyright Ordnance Survey. All rights reserved Resistivity survey results (0.5m probe spacing) showing topographic survey at 0.20m intervals, includes © Crown Copyright Ordnance Survey. All rights reserved Resistivity survey interpretation feature map showing topographic survey at 0.20m intervals, includes © Crown Copyright Ordnance Survey. All rights reserved

117 119 118 121

Hereford, reconstructed section of the Bishop’s aisled hall with its wooden arcades (after Blair 1987, 66, Figure 4) Wells, the divided vaulted undercroft of Jocelyn’s hall of the early thirteenth century (after Parker 1863, 148, Plate 5)

135

6.5 6.6 6.7

7.1 7.2

95 97

122 123 124

135

TABLES 1 2 3 4

Data recovery levels Period summary A comparison (in EVEs) of the first to second-century glass assemblages from three York sites Summary of interpretation of geophysical data

7 9 70 120

APPENDICES APPENDIX A CATALOGUE OF ROMAN MATERIAL Catalogue of Roman pottery Catalogue of Roman glass

74 89

APPENDIX A TABLES A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 A13 A14 A15 A16 A17 A18 A19 A20 A21 A22 A23 A24 A25 A26 A27 A28 A29 A30 A31 A32 A33

Breakdown of stratified Roman pottery assemblage by ware and context group by sherd count Breakdown of stratified Roman pottery assemblage by ware and context group by sherd percentage Breakdown of stratified Roman pottery assemblage by ware and context group by EVEs Breakdown of stratified Roman pottery assemblage by ware and context group by weight (g) Breakdown of stratified Roman pottery assemblage by ware, quantified by weight percentage Breakdown of stratified Roman pottery assemblage by form, quantified by EVEs Breakdown of stratified Roman pottery assemblage by function, quantified by EVEs Average weight of selected fabrics by context group Breakdown of residual Roman pottery by ware, quantified by weight percentage Incidence of cross-fitting sherds across the site Breakdown of stratified Eboracum ware into ware types, quantified by sherd percentage Breakdown of stratified Eboracum ware into ware types, quantified by EVE percentage Incidence of Rustic ware decoration by context groups, quantified by weight Incidence of black burnished wares by context groups, quantified by weight Incidence of black burnished wares by form, quantified by weight Incidence of grey wares by context groups, quantified by weight Incidence of colour-coated wares by context groups, quantified by weight Incidence of amphorae by context groups, quantified by weight Incidence of decorated vs plain Samian ware by context groups, quantified by weight Incidence of mortaria by context groups, quantified by sherd count Incidence of mortaria by context groups, quantified by weight Sherd condition, quantified by weight Vessel use, quantified by weight Roman pottery by source Box flue tiles Legionary tile stamps Painted plaster Opus signinum A list of contexts from which the vertebrate remains were recorded, showing the date to which each context was assigned Total number of bone fragments from first-century drain fills C1251 and C1282 Total fragment counts and frequencies of the main domesticate mammals by period Total fragment counts of bones from Roman date groups Numbers of fragments for cattle, caprovids and pigs by body part groups from selected periods

78 78 79 79 80 80 81 81 82 82 83 83 83 84 84 84 85 85 85 86 86 86 87 87 88 88 88 89 90 92 92 93 93

APPENDIX B CATALOGUE OF MEDIEVAL MATERIAL APPENDIX B TABLES B1 B2 B3 B4 B5

Correlation of codes with those used in the Archaeology of York Series Incidence of pottery by (weight (g)) in eleventh to twelfth-century deposits Context group deposit type Incidence of pottery (by weight (g)) in twelfth to thirteenth-century deposits Incidence of pottery (by weight (g)) in medieval deposits

111 113 113 113 113

B6 B7 B8 B9 B10 B11

Incidence of pottery types by context Context group deposit type Incidence of vessel forms by period (by weight (g)) Incidence of vessel forms (by weight) Crested and ridge tile Medieval brick

113 114 114 114 115 115

APPENDIX C DOCUMENTATION FOR SIR ARTHUR INGRAM’S FORMAL GARDEN APPENDIX C FIGURES C1 C2 C3 C4 C5 C6

Map of York (Richards 1685) Reproduced courtesy of Staffordshire Record Office Map of York (Horsley 1697) © Explore York Libraries and Archives / City of York Council, 2015 Map of York (Cossins 1722) © Explore York Libraries and Archives / City of York Council, 2015 Map of York (Drake 1736) © Explore York Libraries and Archives / City of York Council, 2015 A Plan of the City of York (Chassereau 1750) © Explore York Libraries and Archives / City of York Council, 2015 A Plan of the City of York (Jeffrys 1772) © Explore York Libraries and Archives / City of York Council, 2015

127 127 128 128 129 130

LIST OF ABBREVIATIONS AML AY BAR BIHR BL CAR CBA EAU JBAA LAS PRO RCHME YAJ YCL YCA YCAG YML

Ancient Monuments Laboratory Archaeology of York British Archaeological Reports The Borthwick Institute of Historical Research, York The British Library, London Colchester Archaeological Reports Council for British Archaeology Environmental Archaeology Unit Journal of the British Archaeological Association Lincoln Archaeological Studies The Public Record Office, London Royal Commission on the Historical Monuments of England Yorkshire Archaeological Journal The City Library, York The City Archives, York York City Art Gallery The Minster Library, York

ABSTRACT This research report presents the results of an archaeological research programme carried out by Field Archaeology Specialists Ltd on behalf of the University of York in 1997, sponsored and supported by the Dean and Chapter of York. The research opportunity was provided by the need to construct a new extension to the Minster Library. The foundation trenches for the new extension were excavated archaeologically while detailed geophysical and topographic surveys were carried out in the surrounding area of the Minster precinct. The research programme successfully defined a series of buildings interpreted as barracks and a minor street within the north corner of the Roman legionary fortress. The site was prepared and the barracks built initially in timber, c. AD 71, and replaced in stone c. AD120, being occupied into the late fourth century before a period of decline and abandonment. The excavation produced little evidence of sub-Roman or early medieval activity or use of these fortress buildings. The remains of the stone barracks were robbed, probably in the late eleventh century as part of the Norman cathedral building campaign. A substantial eleventh-century ditch which may have acted as a boundary for the medieval cathedral close was also identified. The ditch appeared to have been backfilled during the late twelfth to early thirteenth century prior to a very significant redevelopment of the archbishop’s palace during a period when the Archbishops of York were thought to have turned their attention and resources towards the archiepiscopal palace at Bishopthorpe. A substantial building identified as the archbishop’s solar was defined along with a narrow range which appears to have connected the solar to the archbishop’s chapel, which survives as part of the Minster Library. The destruction horizon associated with these buildings was apparently removed by the construction of extensions to the Minster Library in the nineteenth and twentieth centuries, but they were demolished and robbed to foundation level to make way for the creation of a formal garden manifest as an ornamental path. This period of activity was enhanced by the results of a geophysical survey which also located a large number of anomalies that, on the basis of documentary and cartographic research have been interpreted as the remains of a seventeenth-century ornamental garden associated with the mansion of Lord Ingram.

CHAPTER 1 INTRODUCTION TO THE PROJECT The city of York is renowned for its Roman and medieval monuments and the quality of its archaeological deposits, which have now been under continuous study for more than forty years. Most of the opportunities for archaeological excavation have arisen through new development and are thus mainly located in areas outside the protected monumental zones. Probably no zone in the city is of greater archaeological importance or better protected than that of the precinct of York Minster, site of the headquarters of the Roman town in the first to fourth century, and the great medieval cathedral that survives on the same site (Figure 1.1). The archaeological strata here have a potential for writing the history of the first and second millennium AD that is among the most valuable in Europe.

The fate of the fortress at the termination of Roman government in the early fifth century remains controversial, but there is some consensus that after flourishing briefly as a centre for farming or slaughtering animals as well as light industry, the roof and its columns fell. There was little evidence for occupation between the fifth and seventh century (Period 7), but the occurrence of grave markers dating to the seventh or eighth century suggests the presence of cemetery and a church in the vicinity. Contemporary documents note the existence of an Anglo-Saxon minster from the seventh century, but no trace of it was found beneath the present minster and subsequent hypotheses have placed it both to the southeast and the northwest (see chapter two). By the ninth to eleventh century (Period 8) the cemetery had encroached over the collapsed basilica, but no further north (Phillips and Heywood 1995, Figure 83). Rough constructions were erected among the stub walls of the barracks, where there was evidence for the working of iron, bone, shale and jet. This was seen as a light industrial and residential quarter adjacent to the AngloSaxon minster and contemporary with the AngloScandinavian quarter around Coppergate.

The Minster site was the subject of intensive archaeological investigation in 1966-1973 on the occasion of its underpinning, following the realisation of a subsidence so grievous that the whole building was in danger of collapse. More than 50 excavations were recorded, within a greater or lesser time constraint, and these numerous though disconnected sightings have led to the composition of a synthesis of the sequence of human activity in the centre of the city (Carver in Phillips and Heywood 1995, 177-221).

The Norman cathedral was constructed by Bishop Thomas of Bayeux in 1080 on the footprint of the current minster, an operation preceded by the extensive quarrying of Roman building stone down to the fourth-century floor levels. Subsequent Archbishops of York of the twelfth to thirteenth century refurbished the cathedral and built a palace to the north of it (chapter two). The arcade of Archbishop Roger (1154-1181) survives today. At its east end stood a two-storey chapel built in the thirteenth century by Walter de Gray (1215-1255), which would be converted in 1808 into York Minster Library.

The Roman legionary fortress was founded in AD 71 on a site that was first cleared and levelled. At the centre of the fortress was the headquarters building (the principia), consisting of a large hall (the basilica) and two wings around a courtyard. The principia stood at a T-junction formed by the via Principalis, modern Petergate and the via Praetoria, modern Stonegate. On either side were rows of barrack blocks, each comprising a centurion’s house and the legionaries’ quarters (contubernia). The footprint of the later medieval cathedral stood over half of the basilica and two barrack blocks, so the story was largely built up from these structures. The fortress was first built in timber (late first century) to house the Legio IX Hispana (hereafter 9th Legion), whose stamped tiles were found. It went through two structural phases before the early second century when the principia was rebuilt in stone. By c. AD 120 the barracks had also been realised in stone versions, the tiles of the 9th Legion, whose total disappearance remains a mystery, being replaced by those of the Legio VI Victrix (hereafter 6th Legion). The impression is one of a flourishing garrison. The third century by contrast has an air of inactivity and decay. In the early fourth century, renewal was evident in all structural areas, but by the end of the century the centre had become less military, if just as privileged. Streets had been blocked, the centurion’s quarters were refitted as urban ‘villas’ and the inside of the basilica had ‘put on ritual airs’ (Carver in Phillips and Heywood 1995, 185).

In 1980 a programme of investigations was carried out at York Minster Library by the then cathedral archaeologist, Derek Phillips (Phillips and Heywood 1995, 160-163). Foundations of a Roman stone building (Building 7) were defined, that had stood from the second to fourth century and was in ruins by the ninth. A flat bottomed ditch running parallel to the Roman wall line had been backfilled in the twelfth to thirteenth century. This ditch has since been seen as a boundary ditch, associated with the Minster precinct of the twelfth century or earlier (p11). The opportunity to excavate again in the neighbourhood of York Minster Library allowed several long-term controversial questions to be addressed: when was the whole fortress realised in stone? Was there a third century decline? What happened in the early fifth century? How long did the Roman buildings stand? Was the Anglo-Saxon minster to be found in the area? What was the first of the Bishop’s palaces to be erected?

1

Excavations at the Minster Library, York 1997

Figure 1.1 Location of Roman fortress and York Minster (fortress plan after RCHME) includes © Crown Copyright Ordnance Survey. All rights reserved

This report begins with a description of the investigation (this chapter) and a summary of the history of the Minster close (north side) from the first to the twentieth century (chapter two). Chapter three presents the new information on the structures and assemblages belonging to the Roman fortress. Chapter four explores the evidence for activity between the fifth and eleventh century. Chapter five is concerned with the principal medieval phase, mainly relating on this site to the Archbishop’s palace of

the thirteenth to sixteenth century. The palace was derelict by the Reformation but enjoyed another lease of life after the Dissolution as the seventeenth-century planned garden of Sir Arthur Ingram, before this too was abandoned and the site given over to the nineteenthcentury York Minster Library (chapter six). The chapters that relate the archaeological sequences (chapters three to six) combine both the structures and the

2

Introduction to the Project assemblages. The richest sequence belongs to the Roman, the most elusive to the early medieval, the best documented to the medieval and post-medieval periods. The discussion of the periods is therefore (inevitably) unbalanced. With these constraints in mind, overviews are offered in chapter seven.

trench was found to be located over the southwestern edge of a road, considered by the excavator to be the remains of the via Quintana of the Roman fortress. The depth of the archaeological stratification at this point was found to be c. 2.0m suggesting that the earlier borehole had been driven through an archaeological feature that had been cut into the natural subsoil.

Project Design An excavation was therefore proposed which confined itself to the area to be disturbed by the foundations of the library extension. It was predicted that the results of this undertaking should:

The site (NGR: SE 6029 5233) lies on the northwestern side of the existing Minster Library, close to the northern limit of the York Minster precinct (Figure 1.2). This area lies within a Scheduled Monument (13280 York Minster Precinct; NHLE Entry No 1017777) and the York Area of Archaeological Importance. Archbishop Roger's Palace cloister arcade (Monument NY 256; NHLE Entry No 1257919) is located immediately to the southwest. The excavation was undertaken between the 3rd March 1997 and the 1st May 1997 in advance of the construction of a new extension to the Minster Library, prompted by increasing demand on the storage space and facilities. A new three-storey stone building was designed to replace the Hailstone Room, a small, single-storey extension built in 1960. The Dean and Chapter selected a mass-concrete foundation design that required trenches which were to be excavated archaeologically. Alternative pile foundation schemes were rejected as they were considered to be too destructive and would not have offered a useful research opportunity. The archaeological project was first proposed by Derek Phillips in 1996 (then York Minster Archaeologist) and subsequently revised in its scope and costing by Field Archaeology Specialists Ltd, now trading as FAS Heritage (hereafter FAS). Scheduled Monument Consent and Planning Permission were granted for the archaeological excavation and building of the extension by the Secretary of State and the City of York Council.

1. 2.

Allow useful information to be added to the plan of this part of the Roman legionary fortress. Provide the opportunity for sedimentological study of the ‘dark earth’ deposits.

Implementation The excavation project was originally designed to be undertaken by an excavation team under the direction of Derek Phillips, then York Minster Archaeologist. Due to the timetable requirements of the related construction programme, Mr Phillips was unable to direct the excavation in person. The Department of Archaeology was therefore invited to commission FAS to carry out the direction of the excavation programme with Mr Phillips acting as consultant to the field project. There was no opportunity to alter the agreed area of intervention, but in order to provide a broader context and to increase the research yield of the excavation, the Department of Archaeology requested that a detailed geophysical and topographical survey of the area around the Minster Library be undertaken as part of the project. The Project Design and Data Acquisition strategy (Phillips 1996) was adhered to by the FAS excavation team, although minor improvements to excavation methodology were adopted.

Evaluation The research programme was carried out in the following phases: 1. Topographic and geophysical surveys; 2. Excavation of the foundation trenches; 3. Post-excavation assessment and archive preparation; 4. Post-excavation analysis. Phase 3 included the assessment of the field record and material assemblages in order to establish the potential for further analysis. The assessment was carried out in accordance with Management of Archaeological Projects (English Heritage 1991) and a report prepared (FAS 1999). Informed by the assessment, an updated Research Design was formulated which is presented below.

In August 1991, Northern Archaeological Associates were commissioned to excavate three evaluation trenches designed to test for the presence of medieval and postmedieval structures and deposits within the area of the proposed extension (Northern Archaeological Associates 1991). Trenches B and C, located to the northwest of the Hailstone Room, defined a series of layers, and Trench A revealed structural remains which were considered to be no earlier than the thirteenth century. In 1994, a single borehole was sunk 14.20m below the ground surface to locate and assess the natural subsoil. Undisturbed clay was encountered at 11.85m AOD which suggested that archaeological deposits were present to a depth of 2.60m below ground level. A further evaluation excavation trench was undertaken by Derek Phillips in 1995 to establish whether the archaeological sequence revealed by the 1980 investigation (Cutting YML, Phillips and Heywood 1995) inside the library provided a reliable predictive model for the site of the proposed library extension. This 2.2m x 2.2m trench (ZE) was a continuation of Trench C, excavated to a depth of 1.0m by Northern Archaeological Associates in 1991. The

Survey procedures A site grid aligned to magnetic north with above Ordnance Datum height (AOD) was established with reference to the permanent benchmark in the nave of York Minster. Survey stations were then set out around the Minster precinct using a total station theodolite to allow accurate surveying throughout the area. This site grid was employed during each phase of the fieldwork programme. The results are given in chapter six, p118-24.

3

Excavations at the Minster Library, York 1997

Figure 1.2 Location of interventions (evaluation and excavation) includes © Crown Copyright Ordnance Survey. All rights reserved

4

Introduction to the Project Topographical survey encompassed all the accessible areas of open ground around the Minster and Minster Library, and all surface features such as buildings, boundaries, paths, trees and access to services were mapped (Figure 1.3). Topographic data, including ‘strings’ of data defining significant breaks of slope were also recorded to facilitate the production of a contour surface model of this area. All survey data was acquired using a total station theodolite and processed using LISCAD software.

image of soil resistance at greater depth. The raw data was processed using Geoplot version 2.02. This involved the adjustment of any differences in the average background reading between individual survey grids as well as inconsistencies caused by changing climatic conditions, which were removed to facilitate clear presentation of the data sets. The processed data was transferred to Surfer version 6.2 in which it was prepared for presentation and the resulting greyscale images were imported into AutoCAD.

High resolution magnetometer and soil resistance area surveys were carried out in the Deanery garden to the northeast (Area A), and the grassed area to the southwest of the Minster Library (Area B). The geophysical surveys in Area A were carried out over 10m x 10m survey grids in order to maximise survey coverage within an area containing many obstacles such as trees and flowerbeds and resulted in a total survey area of 1700m². As fewer survey obstacles were present in Area B the surveys were carried out over 20m x 20m survey grids forming a total survey area of 1600m². In both cases, the external edges of the survey areas were set out using a total station theodolite, with intermediate points being positioned using tapes. This procedure ensured an internal grid point accuracy of 0.05m for the survey.

Excavation procedures The excavation procedures employed by FAS are based on a set of principles known as Field Research Procedure (Carver 2009). Within this system, the structure of data is hierarchical. Each stratigraphic unit considered to have been formed by a single deposition is defined as a ‘context’. In addition to the conventional definition of contexts, sets of contexts called ‘features’ and sets of features called ‘structures’ may also be defined. Separate number systems are employed for contexts, features and structures and each are recorded on separate pro-forma record sheets and indices. Pre-determined data recovery levels are applied and recorded during excavation (Table 1). Recovery levels represent coherent sets of definition, excavation and recording procedures on a rising scale of intensity and resource commitment. This procedure is adhered to in order to ensure that the method of data definition and retrieval is standardised at levels agreed by the Project Design and that comparable data-sets are thus retrieved.

A magnetometer survey was carried out using a fluxgate gradiometer with digital storage and data transfer facilities (FM36 with ST1 sample trigger - manufactured by Geoscan Research). Each survey grid was undertaken using the parallel traverse method (unidirectionally) to ensure the capture of good quality raw data. Instrument readings were logged at 0.5m x 0.25m intervals in order to produce a high resolution image of magnetic anomalies which improves the definition of feature form and increases the possibility of defining smaller feature types. The raw data was processed using Geoplot version 2.02. This involved the adjustment of any differences in the average background reading between individual survey grids as well as inconsistencies caused by instrument drift, which were removed to facilitate clear presentation of the data sets. The processed data was transferred to Surfer version 6.2 in which it was prepared for presentation and the resulting greyscale images were imported into AutoCAD and output on a high definition laser printer. A soil resistance survey was carried out using a soil resistance meter with digital storage and data transfer facilities (RM15 Advanced - manufactured by Geoscan Research). The RM15 was used with a MPX15 multiplexer connected to a PA5 probe array fitted with three probes. The use of a multiplexer and multiple probe array allows a series of different readings to be taken at the same point. In this case, two readings at 0.5m probe separation and one at 1.0m probe separation were logged at each point. This method produces two data sets; firstly, a higher resolution data set with readings at 0.5m x 1.0m intervals (0.5m probe spacing); and secondly a lower resolution data set with readings at 1.0m x 1.0m intervals (1.0m probe spacing). The first data set produces a higher definition image of soil resistance anomalies, while the second with wider probe spacing provides a coarser

Environmental sampling and the wet and dry sieving of proportions of excavated deposits was carried out on a systematic basis. Samples were recorded in the sample register. The selection of a deposit for enhanced recovery was dependent upon the perceived analytical potential of the deposit and specifically, whether a context was considered to be of primary or secondary deposition. Three hundred and sixty three deposits were defined during the excavation. Twenty five deposits were selected for enhanced finds recovery through sieving while forty deposits were the subject of environmental sampling (for more detailed information on recovery levels and enhanced recovery see Post-excavation Assessment Report, FAS 2009, Appendix I). Site geometry (the shape, dimensions and position of contexts, features and structures) was recorded in plan and section and was accurately located in relation to the site grid using a total station theodolite. Contexts, features and structures were recorded photographically using 35mm and medium format cameras. Photographic and drawing indices were compiled and cross-referenced with the context and feature indices. Site preparation and constraints Once the old extension of the library had been demolished and associated services redirected, the

5

Excavations at the Minster Library, York 1997

Figure 1.3 Location of survey areas (topographic and geophysical) includes © Crown Copyright Ordnance Survey. All rights reserved

6

Introduction to the Project building contractor (William Birch & Sons Ltd) set out the external outline of the new extension. Modern overburden (0.40m-0.50m), including the concrete raft foundations of the old extension, was then removed throughout the area of the new extension using a backacting mechanical excavator fitted with a wide toothless

ditching bucket. This operation was carried out under strict archaeological supervision (Recovery Level A). The existing basement which houses the boiler for the library was left intact as this was to be incorporated into the new extension.

Table 1 Data recovery levels Level

Operation

Find

Component

Context

Feature

Structure

Machining

Plot 2-D

(not recovered)

Outline plan

Outline plan

Outline plan

Fieldwalking

Plot 2-D

(not recovered)

(not recovered)

Inferred from density plot

Inferred from density plot

Geophysics

(not recovered)

(not recovered)

(not recovered)

Inferred from density plot

Inferred from density plot

Topographics

(not recovered)

(not recovered)

(not recovered)

Inferred from map

Inferred from map

Shovel scraping (definition)

Plot 2-D

(not recovered)

Short description, Short description, outline plan outline plan

Short description, outline plan

Recover by Shovel excavation context

Optional sampling

Short description, Short description, outline plan plan and profile

Short description, plan and profile, photograph (postexc)

Coarse trowelling Plot 2-D (definition)

(not recovered)

Description, outline plan

Description, outline plan

Description, outline plan

Excavation

Recover by context. Optional sieving

Optional sampling

Full description, outline plan

Full description, plan, section, photograph (postexc)

Full description, plan, section, photograph (postexc)

Fine trowelling (definition)

Plot 3-D

(not recovered)

Full description, outline plan, photograph

Full description, detailed plan, photograph

Full description, detailed plan, photograph

Excavation

Plot 3-D. Selective Sample sieving sampling

Full description, outline plan, section, photograph (preexc)

Full description, detailed plan, section, photograph (pre/post exc)

Full description, detailed plan, section, photograph (by phase)

E

Detailed excavation

Plot 3-D. Describe attitude. Sieve all

Keep all

(as LEVEL D) Optional colour plan/section

(as LEVEL D) Full photographic record

(as LEVEL D) Full photographic record

F

(as above) Block removal for Photograph controlled and draw in dissection situ

Keep all

(as LEVEL E) Full photographic (as LEVEL E) record

A

B

C

D

The edges of the excavation trenches for the foundations of the new extension were then set out and located in relation to the site grid from a nearby survey station. The narrow continuous trench which forms the footprint of the exterior walls of the new extension was divided into two areas for recording purposes with Area A to the west of the basement and Area B to the east of the basement. The separate trench to the southeast of the basement was designated as Area C (see Figure 1.2). The machinestripped area was cleaned and archaeologically recorded before hand-excavation of the foundation trench areas commenced.

(as LEVEL E)

Due to the tight schedule of the construction programme it was necessary to give priority to the completion of the archaeological excavation of the foundation trenches in the western half of the area allowing the building contractors staged access to the site. For this reason it was not possible to keep the whole area ‘in phase’ during the excavation resulting in the creation of a more complex archaeological record than would otherwise have been necessary. A temporary scaffolding roof was constructed over the entire excavation area to protect the site and the

7

Excavations at the Minster Library, York 1997 excavation team from adverse weather conditions. Although this shelter certainly prevented any loss of time during spells of bad weather, it also resulted in the reliance on artificial light for deposit definition, something which proved to be a disadvantage during the excavation of medieval deposits.

construction, maintenance and re-development within the fortress from the late first to the late fourth century. The revised objectives for the Roman period were:

The perched water table was encountered at c. 12.75m AOD. Although the pumping of groundwater prevented the flooding of the trenches below this level, both working conditions and deposit definition was poor during the excavation of first and early second century deposits.

2.

1.

3. 4. 5.

Overview of results 6. The results of the survey programme and research excavation are presented as eight periods of activity (Table 2). These periods could, in some cases, be defined as phases that represent distinct changes in activity during a period such as ‘construction’, ‘use’ and ‘disuse’. The definition of these periods from the excavation records has been achieved through a programme of detailed stratigraphic and structural analysis that also assimilated the results of the specialist studies.

To define a detailed sequence of construction, re-development and decline within this area of the fortress. To identify the function of buildings and the internal use of space by phase. To identify possible indicators of a 9th Legion 6th Legion handover. To address problems with the chronology of Roman pottery in York. To investigate aspects of diet, food processing and distribution within the fortress. To define differences in the status of barracks at the centre and periphery of the fortress.

Sub-Roman/early medieval The site record and material assemblages contained some evidence for activity in this notoriously elusive period, which were considered to be of some use in reviewing earlier interpretations of ‘Dark Age’ activity within the area of the Minster Estate. The revised objectives for the sub-Roman/early medieval period were:

The geophysical survey programme successfully defined a wide range of medieval and post-medieval features in the form of a large number of linear and rectilinear anomalies. These anomalies appear to indicate the presence of post-medieval ponds, drains and garden features and possibly the foundations of buildings associated with Ingram’s gardens, while a number of the linear anomalies appear to be the continuation of foundation trenches defined by the excavation and which have been interpreted as the remains of part of the thirteenth-century Archbishop’s palace. The results of the survey are presented in chapter six.

1. 2. 3. 4.

To characterise sub-Roman activity within this area of the fortress. To investigate the apparent invisibility of intrusive features within the ‘dark earth’. To characterise early medieval activity within this area of the site. To review early medieval activity within the fortress.

Medieval The results of the research programme were considered to have the potential to make an important contribution to archaeological knowledge with regard to the late medieval and post-medieval form and development of one of Britain’s most influential Christian centres. Analysis of the archaeological record defined from this period by the excavation and geophysical survey programme clearly had the highest potential in terms of its research yield and was therefore given a high priority in the post-excavation programme. The revised objectives for the medieval period were:

Updated Research Design The character and significance of the Roman structures and deposits encountered during the excavation were accurately predicted, but the late medieval and postmedieval activity defined by the geophysical survey and excavation programme were considered to be far more useful than was envisaged as a result of the evaluation programme. The updated research design refined the academic objectives of the investigation for the Roman period, while setting out more ambitious objectives for the late medieval and post-medieval phases, the former of which was considered to be among the principal rewards of the research programme.

1. 2.

Roman

3.

The small window on the Roman legionary fortress revealed a wide range of structural remains, including buildings, a street and an open drain or alleyway. Evidence for the type of activities carried out, and the use of space within the buildings was found to be sparse, but the related stratigraphy detailed a long sequence of

4.

8

To date, characterise and map the possible precinct boundary ditch. To date and phase the structural elements defined by the excavation. To investigate and characterise medieval activity to the east of the boundary. To define the layout and structural development of the Archbishop's Palace complex in the Minster Library area.

Introduction to the Project Post-medieval

above, recovery was in practice uneven between periods. The Roman sequence was well stratified and much of the assemblage in context, so that a full treatment of artefacts and animal bones was possible (chapter three, Appendix A). The information from the post-Roman and Early Medieval period was often inferential and the material residual; however a few sherds and comb fragments belonging to this period were identified (chapter four). The medieval assemblage was rich, especially in pottery, but while a large consignment of animal bones was related to the later medieval period, the majority was out of context, so no analysis has been offered (chapter five, Appendix B). The nature of the formal gardens of the seventeenth century was achieved through a combination of excavation, surface survey and documentary research (chapter six, Appendix C). The results are discussed and assessed in chapter seven.

The excavation and geophysical survey defined a wide range of post-medieval features within the area of the Minster Library. Limited analysis of these data sets, enhanced by documentary and cartographic research were considered to be worthwhile, providing useful information on the destruction of the palace, the layout of the ornamental gardens of Ingram’s mansion and the activities carried out on the site from the seventeenth century to the present day. Revised objectives for the post-medieval period were: 1.

To date and detail the destruction of the final phase of the Archbishop’s Palace. To define the form and layout of Ingram’s ornamental gardens. To define the eighteenth to twentieth-century use of the area.

2. 3.

Studies of the landscape, buildings, stratigraphic sequence and assemblages of artefacts and animal bones were undertaken to address these objectives. As noted Table 2 Period summary Period

Activity

Date

1

Legionary Fortress: Preparation of site, construction, use and reconstruction of timber barracks

Late first to early second century

2

Legionary Fortress: Demolition of timber barracks, construction and use of stone barracks

Early second to early fourth century

3

Legionary Fortress: Continued use and disuse of stone barracks

Early fourth to early fifth century?

4

Post-Roman and early medieval activity

Early fifth to late eleventh century

5

Medieval activity and boundary

Late eleventh to late twelfth century

6

Archbishop’s palace: Construction, use and modification of stone hall

Early thirteenth to late sixteenth century

7

Ingram’s mansion: Clearance of stone hall, construction and use of gardens

Early seventeenth to early eighteenth century

8

York Minster Library

Early eighteenth to late nineteenth century

Publication Fieldwork was undertaken in 1997 and post-excavation assessment and Field Report completed in 1999 and 2000 respectively. The Field Report was prepared for publication in 2015 including the original specialist reports. The Roman archaeology has been reviewed and placed in a broader research context by Paul Bidwell and the results as a whole by Martin Carver who edited this publication.

9

CHAPTER 2 A HISTORY OF THE YORK MINSTER CLOSE (NORTH SIDE) KATE GILES The discoveries made in the excavations reported here bear most directly on the history of the legionary fortress, the Northumbrian and Anglo-Scandinavian minster, the medieval palaces of the Archbishops and the gardens of Sir Arthur Ingram that succeeded them. Sources relating to the land occupied by the current minster precinct can usefully reflect on topographical features and anomalies that may imply the layout of buildings in these earlier periods. The purpose of this chapter is to provide background to scholarly discussion on the history of the Minster close, and review the longstanding questions that the York Minster Library excavations offered a chance to address.

Heywood 1995, 160-163). The aim of that excavation was to locate a boundary marker, possibly a ditch, which was believed to define the eleventh-century close. The excavations uncovered Roman structures and a medieval ditch infilled during the twelfth to thirteenth century, shortly before the construction of the Library. The Roman fortress Excavations carried out in and around the Minster between 1967-73 confirmed the position and orientation of the headquarters of the Roman legionary fortress (Phillips and Heywood 1995, i, 1). The present Minster is located directly over the principia and barrack blocks of the fortress, but at an angle of 48 degrees to it (see Figure 1.1). The fortress was constructed in timber in the late first century (Period 2), was reconstructed in stone and flourished in the second century (Period 3), declined in the third century (Period 4), was refurbished in the early fourth (Period 5) and declined to informal usage with blocked streets in the late fourth (Period 6) (Carver in Phillips and Heywood 1995, 181-7). During a small excavation at the Minster Library in 1980, Phillips uncovered the foundations of Roman structures at a depth of 12.77m AOD, which were consistent in form and scale with barrack contubernia, and which Phillips concluded were probably contemporary with the earliest stone phase of the fortress as defined by Barracks 1-3 (late first to early second century; Phillips and Heywood 1995, 30).

Sources The earliest surviving descriptions of the area surrounding the Minster come from seventeenth, eighteenth and nineteenth-century visitors to the city (usefully collected in Palliser and Palliser 1979). A rich series of cartographic and pictorial sources depicting the area of the Minster close and individual buildings within it, also survives from the sixteenth century onwards. These include early maps of the city such as those of Speed (1611) and later sources, such as the Ordnance Survey of 1852. Eighteenth-century antiquaries such as Gent (1730) and Drake (1736) were largely concerned to elucidate the development of the liberty, with its four gates and individual properties of interest, such as St William’s College, the Bedern and the Treasurer’s house. Following the Parliamentary Acts of 1813 and 1824, ‘for enlarging and improving’ Minster Yard (BIHR AP 1813/1; AP 1825/1), nineteenth-century historians tended to focus on the improvements made to the Minster area by the sweeping away of the medieval and post-medieval buildings in the precinct, including the demolition of the four precinct gates, and the construction of new buildings such as the New Deanery (Hargrove 1818; Widdrington 1892; James Raine the Younger 1879-94). As the surviving remains of Ingram’s mansion, Peter prison and the Old Deanery were demolished, artists such as Cave (1813) and Halfpenny (1807) recorded their appearance in a series of engravings. During the twentieth century, important contributions to the understanding of the Minster precinct have been made by Hamilton Thompson (1944), Raine (1955), Aylmer and Cant (1977), Leake (1990), Barr (1993) and the Victoria County History (1961). The most useful (and as yet largely unpublished) body of research carried out on the Minster close and Archbishop’s palace remains that of Ron Butler, carried out in conjunction with Eric Gee and David Black as part of the six volume study of the city by the Royal Commission (see Butler 1988, 1997).

Of three alternative models for the nature of post-Roman period that were advanced in the report on the excavations of 1967-73 (Carver in Phillips and Heywood 1995, 194-195), Model B has the most significant implications for the interpretation of the current findings. It proposes that there was a fundamental demilitarisation of the area towards the end of Roman government in the late fourth century, with the basilica eventually functioning as a farm or market. This was followed by a period of dereliction in the fifth to sixth century, and spasmodic Anglian (seventh to eighth century) activity including a cemetery, and a minster located away from the site of the present cathedral (see below). In the ninth to eleventh century the minster area was given over to residence and light industry within the stub walls of the former Roman fortress. The Anglo-Saxon Minster The location and orientation of the Anglo-Saxon Minster in York has been a matter of controversy for centuries (Harrison 1960; Phillips in RCHME 1985; Hill and Brooke 1977; Gee 1977; James in Phillips and Heywood 1995). Bede states that King Edwin was baptised by Paulinus at Easter in AD 627 in a hastily erected (‘citero opere’) wooden church in the city. Some years later, but

Directly comparative material for the 1997 Minster Library excavations was excavated in 1980 (Phillips and

10

A History of the York Minster Close (North Side) before Edwin’s death in AD 632/3, this was encased and replaced by a stone church dedicated to St Peter (Plummer 1896, i, 113-5). However, Eddi’s Life of St Wilfrid (Colgrave 1927) records that when Wilfrid took over the see in AD 669 only forty years later, the church was in a poor state. Wilfrid restored the church, whitewashed it, and adorned it with ‘various kinds of vessels and furniture’ (Colgrave 1927, 34-5; Raine 1879, i, 23-4). It is this seventh-century church which appears to have survived until the Norman Conquest, being enriched by St John of Beverley (AD 705-732) and Ecgbert (AD 732/4-766). The twelfth-century Historia Regum records that in AD 741 the ‘monasterium in Eboraca civitate’ was burned down. However, this may well refer to the buildings of a religious community whose existence is implied by a number of writers. In AD 793 Alcuin could still describe the Minster as a ‘paragon of lofty beauty supported by massive pillars’ (Raine 1879, 356). It continued to thrive over the next three centuries, becoming the burial place of kings and an architectural focus for both ecclesiastical and secular patronage (Fernie 1983, 47; Gem 1993, 29-66).

Trinity, Micklegate (cf Harrison 1957; Miller 1995; Palliser 1996; Morris 1986; also Hall 1978). Norton (1998a) supported a location north of the present minster on the grounds that the medieval Archbishop’s palace is orientated at a peculiar angle which does not correspond to that of the Romanesque cathedral of Thomas of Bayeux (to which it is angled some 45 degrees) or even the Roman fortress (Figure 2.1). He suggests that the line of the medieval Archbishop’s palace preserves the northern boundary of an AngloSaxon enclosure within which the seventh-century Minster was constructed. The boundaries of this hypothetical enclosure can be traced by following the line of the Archbishop’s palace (and those of its successor, Ingram’s mansion) on historic maps such as that drawn up in 1782 (Leeds Archives WYL100/YO/G) and the Ordnance Survey edition of 1852 (Figures 2.2, 2.3). It encompasses a square of eighteen perches, a total area of two acres. Part of the enclosure might have been marked by the ditch excavated near the Minster Library in 1980 by Phillips (Phillips and Heywood 1995, 160-163; and reexamined in the excavations reported here, see p98). However, archaeological evidence gave no precise date of construction for this ditch; it was infilled in the twelfth to thirteenth century, prior to the construction of Walter de Gray’s (1215-1255) chapel.

Confusion over the archaeological evidence for the Anglo-Saxon Minster arose in the nineteenth century, following the devastation of the choir by John Martin’s fire, and its subsequent recording by John Browne (17931877). Browne (1847; 1867) mistakenly attributed work by Archbishop Roger of Pont l’Eveque (c. 1154-1181) to Archbishop Thomas of Bayeux (c. 1080-1100). The foundations beneath this work were therefore identified as part of the Anglo-Saxon Minster. Browne’s error was recognised by Willis (1848), Hamilton Thompson (1912) and Baldwin Brown (1925). However, further excavations in the choir prompted Sir Charles Peers (1931) and later Dr and Mrs Taylor (1965, 707-9) to conclude that there were Anglo-Saxon remains under the choir. This belief had profound consequences for the archaeological agenda framed in 1967, whose principal objective became the location of the rest of the AngloSaxon Minster (Phillips and Heywood 1995, 20).

The Norman Minster and the first Archbishop’s Palace During the rebellion and subsequent ‘harrying of the north’ in 1068-9, the city of York was destroyed by fire, and according to Florence of Worcester (Thorpe 1853) the Minster of the time was severely damaged. Bishop Thomas Bayeux (1070-1100) is reported by Hugh the Chanter to have re-roofed and repaired it and also the refectory and dormitory of the canons (HCY, ii, 350, RCHME 1985, 4). However, the minster was plundered in 1075 when a Danish host of 200 ships attacked York. This act of vandalism is often presumed to have encouraged Thomas to begin work on a new cathedral on a different site from its Anglo-Saxon predecessor, in order that the latter could continue to be used until the new church was serviceable (Phillips in RCHME 1985, 5; also 1986). Thomas’ decision to found the new cathedral on the site of the Roman principia may also have been intended as a symbolic break with the Anglo-Saxon past and association with earlier forces of conquest.

The excavations of 1967-72 did not locate the Minster, but rather a pre-Norman, Anglo-Scandinavian cemetery in five areas on the south sides of the choir, nave and in the south transept of the Minster (Philips and Heywood 1995, 75-92). The existence of an early minster is implied by the presence in this cemetery of seventh or eighthcentury grave markers. From these Phillips (Phillips and Heywood 1985, 5) concluded that the Anglo-Saxon Minster was associated with the cemetery, but probably lay to the north of the present structure; while Carver felt they implied the location of an Anglo-Saxon church southeast of the present minster within the principia courtyard, on the grounds that graves of the AngloScandinavian period were not encountered in the excavations to the northeast or the northwest (Phillips and Heywood 1995, 192). Alternative hypotheses have suggested that the Anglo-Saxon minster was located on the site of St Michael-le-Belfrey, or even on the other side of the river, on the site later occupied by Holy

It is possible that the construction of the first Romanesque cathedral by Thomas was part of a planned package, which included the reform of the canons into a secular chapter along the lines of those at Salisbury, Lincoln and London. The cathedral was started in c. 1080 and the reforms probably a decade later in 1090 (Hill and Brooke 1977, 25-8). These included the subdivision of the archiepiscopal shire to draw a formal distinction between the properties and privileges of the Archbishop (including the site of the archiepiscopal palace) and the houses and buildings of the Minster clergy (Dickens 1953; Harvey 1965; Brown 1980; Rees Jones 1987).

11

Excavations at the Minster Library, York 1997

Figure 2.2 Map produced in support of a lawsuit in 1782 (Leeds Archives WYL100/YO/G)

Figure 2.1 York, the Minster area, showing the principal features of the medieval topography (after Norton 1998a, 6, Figure 2) Although the form and appearance of Thomas’ cathedral has been studied in considerable detail, virtually nothing is known of his episcopal palace. It seems reasonable to suggest that it was probably on the site of its twelfthcentury successor, and that it displayed some of the architectural qualities of Thomas’ cathedral (Figure 2.4) which was both an architectural and structural tour de

force, being both longer and wider than that of Lanfranc at Canterbury (RCHME 1985, 6-7). It is also probable that the cathedral and palace were important symbols which boosted Thurstan’s archepiscopal authority in his long-running and eventually triumphant battle against the primacy of Canterbury (see Hill and Brooke 1977, 32).

12

A History of the York Minster Close (North Side)

Figure 2.2 Map produced in support of a lawsuit in 1782 (Leeds Archives WYL100/YO/G)

Figure 2.3 Extract from the Ordnance Survey map of 1852 © Crown Copyright Ordnance Survey. All rights reserved

13

Excavations at the Minster Library, York 1997 the Transitional style of architecture seen in Roger’s work in the Minster. The fact that the same mason, Simon of Derby, was employed on the chapel of St Mary and the Holy Angels as well as the palace also suggests that there would have been strong stylistic parallels between the two (Lovatt 2002; Becket Materials i, 334-6). Part of a two-storey range apparently abutted by de Gray’s thirteenth-century chapel is shown by Griffith in 1777 with fenestration characteristic of Romanesque architecture (Figure 2.6).

Figure 2.4 Reconstruction of the cathedral of Thomas of Bayeux (after Phillips in RCHME 1985, 142, Figure 25). Reproduced by permission of Historic England The existence of a reasonably substantial late eleventh to early twelfth-century episcopal palace at York is implied by events that occurred under Archbishop Thurstan (1114-1140). In 1132 Thurstan is reported to have led a group of monks from the Benedictine Abbey church of St Mary’s to the archepiscopal palace, where they sheltered during October and November. The monks had been inspired to call for the reform of St Mary’s by a group of Cistercians passing through York on their way to found Rievaulx. Thurstan protected the dissidents, and on 25th December celebrated Christmas at Ripon with them, before granting them land on which they subsequently founded Fountains Abbey. In 1137 the cathedral was badly damaged by fire (Hapgood 1985; Norton 1998b), and Thurstan appears subsequently to have squared the apse and doubled the size of the transepts (Gee 1977, 121). However, there is no historical evidence to suggest that Thurstan was responsible for remodelling the palace.

Figure 2.5 Surviving arcade in Dean’s Park (left) and chapel (right) with library extension (distant centre) looking northwest The records of Roger’s household familia certainly suggest that the palace was of a reasonable size. They include a seneschal, clerk, two chamberlains, a butler (and an assistant called Richard ‘from the buttery’), a steward, lardiner, porter, door keeper, cook, store-room assistant, two further servants, three boy servants, a baker, an almoner, and numerous clerks. Others who expected hospitality but possibly not accommodation within the palace include Roger’s constable, knight, tollgatherer and doctor (Lovatt 2002).

Archbishop Roger of Pont l’Eveque (1154-1181) was a great patron of architecture, responsible for the rebuilding of the choir and crypt of Thomas’ cathedral (Gee 1977, 122; Chronica Monasterii de Melsa (Rolls Ser. 43) i. 215;). The work was probably completed in c.1175, and although it includes conventional twelfth-century stylistic details, it also incorporates features such as keeled rolls and water-leaf capitals which are indicative of the Transitional phase of Gothic architecture, and which are paralleled at Ripon and Byland. The quality of architectural detail and innovation found in Roger’s Minster may also have characterised his work on the quasi-collegiate chantry chapel of St Mary and the Holy Angels (St Sepulchre’s Chapel), which he founded in c. 1180. This was located on the north side of the Romanesque nave, but was demolished soon after the Dissolution in 1547. Unfortunately, there are no records that indicate the form or layout of Roger’s palace, except for the surviving fragments of an arcade that may have functioned as a slype between the principal rooms of the palace complex, and the hall (Figure 2.5). The quality of the architectural detailing on the arcade suggests that it was an impressive structure, and it may well have incorporated elements of

Figure 2.6 St Sepulchre’s Chapel by Moses Griffith, 1777. Image courtesy of York Museums Trust::http://yorkmuseumstrust.org.uk::Public Domain Archbishop Walter de Gray (1215-1255) was one of the most influential Archbishops of the later medieval period,

14

A History of the York Minster Close (North Side) correspond to the remains of Archbishop Roger’s twelfthcentury arcade. This is clearly shown in BIHR CC.Ab.11.85, running along the south side of the building. Further evidence for the position of the great hall is provided by the fact that after finishing its description by noting the presence of three ‘bynnes’ in the pantry and a serving place, the inventory of 1531 describes ‘the entree from the nethre hall to the chapell’. However, this indicates that the internal spatial hierarchy of the hall ran counter to that which might be expected, with the chapel lying beyond the low screens passage entrance to the hall, and the dais at the south-western end of the hall.

responsible not only for the reform of the chapter (Dobson 1977, 48ff.), but also for the reconstruction of the south transept of the Minster (the north being associated with his colleague John the Romeyn, the treasurer (Gee 1977, 131)). Both were also responsible for the reconstruction of the central tower. Walter de Gray built the two storey chapel which in 1808 would be converted into the York Minster Library. In 1231-4 he was granted thirteen oaks by Henry III ‘to build a certain chapel at York’ (Calendar of Close Rolls 1231-4, 238). The building has formal and stylistic parallels with the chapel that de Gray built in 1241 at his new palace at Thorpe St Andrew, just outside York. Both chapels were two storey structures of magnesian limestone, a string course articulating the storeys externally. The York chapel was of four bays, divided by buttresses with moulded plinths and gabled heads. The building was heavily restored in 1806. Records of its restoration are contained in the day book of William Shout (D & C Archives, E10a), and are accompanied by restoration drawings (YML 12).

Sir Arthur Ingram’s gardens In 1618 the palace site, then ruinous, was leased for three lives to Sir Arthur Ingram at an annual rent of £3 6s. 8d. Sir Arthur Ingram ‘did at his own great cost and charges before the ensealing and delivery of the same lease repaire some part of the same whereby the said Palace became decent good and profitable’ (TN/YO/A9, 13).

There are no records to suggest that de Gray substantially modelled the rest of the palace at the same time as the chapel. However, it is possible that the ground floor aisled hall described in 1486 dates from de Gray’s episcopate. If so, it would contrast with de Gray’s palace at Bishopthorpe where (partly as a result of the topography) the hall was a two-storey structure (p134).

Ingram’s mansion consisted of a courtyard which was entered from the south through a pedimented gateway in a low screen wall. The courtyard was surrounded on three sides, with a lesser court formed by buildings along Precentor’s Lane (now Precentor’s Court)(Figure 2.7). On the north side of the courtyard was a block 65 feet long and 50 feet wide. This was a new range that linked the palace site with Precentor’s Lane. In its south wall was the main entrance, facing the medieval gateway of the eastern range, and adjacent to it was a projecting bow window rising the full height of the two-storeyed building. The western range appears to have been of three storeys and was probably built c. 1624 (Butler 1988, 36). A series of projections shown on the plan of 1782 may well have been chimney stacks providing heating for a series of separate lodgings.

The later history of the palace It appears that during the later medieval period York’s Archbishops shifted their attention away from the palace in the city, towards the archiepiscopal residences at Cawood, Bishopthorpe, Sherburn in Elmet and Reste (Gee 1983). However, several sources indicate the later use of the palace building in the Minster precinct. It was the site of a magnificent feast held on the occasion of the enthronement of Archbishop George Neville in 1465. Neville had been attended by five bishops, two dukes, four earls, other magnates, and the heads of the religious houses in the province, and it is likely that a substantial number of these honoured guests attended the feast (YML A 1(2); H 2/3 (Chapter Acts 1427-1504); Dobson 1977, 100). An account of the order of ceremony of a Garter feast held in 1486 also survives. The dimensions of the hall at this date are given as 122 feet by 76 feet (College of Arms, Arundel MS 30, f.214), and the seating arrangement within the aisles is also recorded (Hearne, ed. 1774, vi, 1-14; Johnston 1979, 146-52). The hall consisted of a central nave with aisles, each of which could accommodate two trestle tables, and the hall followed the conventional medieval tripartite arrangement, with a dais end where the king and Archbishop were seated.

Sir Arthur Ingram was not only a great patron of architecture, but also of garden design. Ingram’s garden at Temple Newsam created in the 1630s is known from an illustration by Knyff dating to 1699 (Gilbert 1972; 1981; Strong 1979, 185). It consisted of a broad terrace opening out from the house with a classical style summer house at one end, and a flight of steps leading down to a walled garden. This was laid out in quarters with a fountain in the middle and a raised terrace at the end flanked by tiny pavilions. The York palace garden is therefore significant because it pre-dates Temple Newsam by a number of years, and may therefore have been used by Ingram to try out ideas later employed in Leeds. The gardens were divided into four sections. To the north, and adjacent to the city walls was an orchard and fishponds. To the south, nearest to the Minster, was a bowling green, and between this and the palace was a walled garden, which incorporated the remains of the twelfth-century arcade still surviving today. To the east of the garden was a yard that enclosed the former Archbishop’s chapel. The reconstructed plan identifies

Sources from the sixteenth and seventeenth centuries provide us with a fleeting glimpse of the palace in a dilapidated state. An inventory of 1531 lists the hall and its contents, including tables and benches. It mentions a wall on the south side of the hall, which would

15

Excavations at the Minster Library, York 1997

Figure 2.7 The Ingram mansion from the west, by E. Abbott. Pen and ink, watercolour and pencil on paper (The Gott Collection, Wakefield Council Permanent Art Collection) individual features within the gardens from historical evidence, and relates them to archaeological and geophysical surveys carried out in 1997-8 (see chapter six and Appendix C). The buildings comprising Ingram’s palace gradually fell into disrepair when the Ingram family moved out of York towards the end of the seventeenth century. In 1730 the medieval gateway into Ogleforth was demolished. This was followed in 1772 by that at (Low or Far) Minster Gates. In 1814 and 1825 the Dean and Chapter obtained two Acts of Parliament that entitled them to make compulsory purchases of property in the close prior to its demolition. Copies of the Acts, together with the lists produced by the Church Commissioners of the houses and tenants, survive in the archives of the Dean and Chapter (BIHR AP 1813/1; AP 1825/1). A row of houses on the north side of the Minster was demolished in 1821. The medieval entrance into the close, known as ‘High Minster Gates’, was demolished in 1828. In 1828 the medieval Deanery, located opposite the south transept, was demolished and replaced by a new Deanery built at right angles to the Minster Library. From 1860 onwards traffic was allowed to flow from Lendal Bridge through the Minster Close via College Street to Goodramgate and Monk Bar.

16

CHAPTER 3 THE ROMAN FORTRESS (FIRST TO FIFTH CENTURY) The site of the excavation lay towards the north corner of the Legionary fortress, an area predicted to be occupied by rows of barrack blocks (Figure 3.1). Northeast of this axial road, the barracks run NE-SW, while on its southwest side the barracks run NW-SE. The excavation included the end wall of a barrack on the northeast side and side walls of a barrack on the southwest side, together with the line of the road that separated them. Activity was recorded in three Areas (A, B and C), and over three Periods: Period 1, construction of the timber barrack blocks; Period 2, reconstruction in stone; and Period 3, decline and abandonment. A summary of the assemblage of pottery, building materials and other finds is given at the end of the chapter, with a catalogue in Appendix A.

sand clay with frequent gravel inclusions (C1363). This deposit became gradually thinner towards the southeastern part of Area B and was absent from the southern part of Area A. A similar deposit, c. 0.05m thick and consisting of light brown sand silt with very frequent gravel inclusions (C1328) was found to overly the ‘tread layer’ in Area C. C1363 and C1328 have been interpreted as a single deposition of imported material, laid down to provide a consolidated and well drained layer of made-up ground. Structures 11 and 12 At least two timber buildings were identified within the excavation area, along with a trackway and an open drain or alleyway. The best preserved structural evidence for this phase comes from the southern part of Area A, where the broad, intrusive foundation trenches for the later Roman stone structures were not present. In this area, two similar, parallel-set, NW-SE aligned gullies were defined (see Figure 3.2, F140, F141). F140 was a vertical-sided and flat-based gully (min.L11.35m x W0.50m x D0.50m) which continued beyond both the northwest and southeast edges of excavation. F141 was also vertical sided (min.L5.90m x W0.50m x D0.50m) and continued beyond the southwest and southeast edges of excavation. Both features were filled with mixed deposits of redeposited subsoil and dark grey clay-silt (F140 C1278, F141 C1279). C1279 (F141) contained two sherds of pottery including a sherd of stamped mortarium dated to AD 65-100. Three circular, flat based shallow depressions were identified in the base of F141 set at 1.45m intervals along the centre of the feature. These depressions were interpreted as post settings. Although no similar depressions were defined along the base of F140, both features have been interpreted as construction or bedding trenches for timber posts.

Period 1 - c. AD 71-c. AD 120 Timber barracks and trackway Period 1 has been assigned to activity considered to be associated with the initial preparation of the fortress site and the construction and use of timber buildings within the area. The deposits, features and structures that constitute the three phases of Period 1 form a coherent and well stratified group which represents the earliest remains of human activity on the site (Figure 3.2). Unfortunately, a large proportion of the excavation area at this level had been truncated by the foundation trenches of later Roman, medieval and modern structures. Site clearance The natural subsoil (C1339), which consisted of brownish-yellow clay, was encountered, where it had not been truncated by archaeological features, at c. 12.45m AOD throughout the excavation area. A 0.02-0.10m thick layer of dark greenish-grey clay-silt containing flecks of brownish-yellow clay (C1285 with analogues C1337, C1350) was found to overlie the natural subsoil in all three areas of the site. The interface between the natural subsoil and this overlying deposit was found to be very clear with no sign of merging. The upper surface of the subsoil was also clean and unblemished. Microscopic examination of C1350 revealed that it consisted mainly of inorganic matter with a trace of organic detritus (Jaques et al 1999). This rather sterile deposit has been interpreted as a tread layer resulting from a levelling operation which preceded the initial construction of the Roman fortress. If this interpretation is correct, it endorses Carver’s conclusion that the lack of observed pre-Roman activity in the area was due to its removal through site clearance (Phillips and Heywood 1995, 180). The initial clearance of the fortress site is conventionally dated to AD 71-79.

F140 and F141 lie to either side of F134 which may be interpreted as an external open drain or alleyway lying between two timber buildings. F141 is therefore considered to represent the remains of an external wall and, along with associated features and deposits to the south-west of the drain in Area A, forms the first of these buildings (Structure 11). Likewise, F140 and associated features and deposits to the northeast of the drain in Area A form the second building (Structure 12). Within Structure 11, two gullies (F143, F145) were found to cut the ‘tread layer’ (C1285). F143 was a NE-SW aligned vertical-sided gully (min.L0.55m x W0.40m x D0.25m) which continued beyond the edge of excavation. A circular depression with a diameter of 0.10m, interpreted as a post setting, was defined in the base of F143. F145 runs parallel to F143 and was similar in character, if less substantial (min.0.55m x 0.25m x 0.10m). Both gullies were filled with a disordered

In the northern part of Areas A and B the ‘tread layer’ was sealed by a c. 0.25m thick layer of reddish brown

17

Excavations at the Minster Library, York 1997

Figure 3.1 Location of York Minster Library with Roman fortress plan beneath, also showing previous excavations within the fortress (after Ottaway 2004, 32, Figure 13)

18

The Roman Fortress

Figure 3.2 Plan of Period 1 structures and features, timber barracks and trackway

19

Excavations at the Minster Library, York 1997 mixture of redeposited subsoil and dark grey clay silt (F143 C1281, F145 C1291). The larger of the two gullies (F143) seems to have been a bedding trench for vertical timber posts although no post voids or fragments of decayed wood were identified.

fills of these features (F165 C1352, F173 C1364) consisted of dark grey clay silt with evidence of oxidisation in the form of thin orange lenses of clay silt. These fills did not contain redeposited subsoil, nor was there any sign of C1363, the c.0.25m thick layer of madeup ground cut by these features, having been redeposited in their backfills. Given the nature of these fills, it seems to be more likely that they were related to drainage rather than construction.

A further vertical-sided feature set at 90 degrees to F141 was also defined (F142). This probable gully continued beyond the southwest and southeast edges of excavation. As was the case with the external construction trench, F142 was filled with redeposited subsoil and dark grey clay silt (C1280). This feature was found to be shallower than F141 (min.L0.85m x min.W0.25m x D0.15m) and is considered to have been the foundation for an internal partition wall of Structure 11, however, it is possible that this feature actually relates to a later building.

A similar feature was also defined in Area C and may have been the continuation of F173 from Area B (F160). Its fill consisted of dark grey clay sand, however, occasional flecks of redeposited subsoil were identified (C1338). F160 formed the butt-end of a wide gully (min.L1.20m x min.W0.80m x D0.50m) with shallow sloping sides at the top of the feature becoming near vertical towards the base. On balance, this feature is more likely to have been a drainage gully although the possibility that it represents a construction trench cannot be excluded.

Two of the internal foundations of Structure 11 (F143, F145) were sealed beneath a layer of light greyish-brown silt with frequent inclusions of small fragments of plaster (C1260). The external construction trench (F141) and F142, however, appeared to cut this layer. It is possible that the earlier construction trenches were the result of an error during building work which was immediately remedied as C1260 may have abutted the elevations of clay walls overlying F141 and F142 rather than having been cut by these features. It is equally possible that this stratigraphic situation was the result of an early modification to Structure 11 although the lack of any contemporary preparation layers or floor surfaces does cast some doubt on this possibility.

The second feature in Area C has been interpreted as a probable construction trench (F154), although little remained of the feature as it had been truncated by the modern basement (F10). The short length of F154 which survived within the excavation area (min.L1.20m x min.W0.50m x D0.35m) was vertical-sided and filled with a series of three deposits (C1321, C1322, C1323). The latest of these deposits (C1321) consisted of brown clay with frequent flecks of charcoal which was found to overlie a yellowish-brown sandy clay with frequent gravel (C1322). The lowest fill of F154 consisted of dark grey sand clay (C1323). These deposits were not inconsistent with the mixed fills of other construction trenches, but seem to have been deposited in a more ordered fashion. The second deposit (C1322) was not dissimilar to the earlier layer in this area thought to represent made-up ground (C1328). Given the depth of this feature, it is more likely to represent the remains of an internal partition wall than that of an external wall.

Within the interior of Structure 12, set at 90 degrees to F140, a less substantial gully was also defined (F137). This vertical sided gully (min.L1.3m x W0.40m x D0.15m) had been truncated by later features to the northeast and the southwest. F137 was filled with redeposited subsoil and dark grey silt clay with lenses of light brown clay silt (C1275). C1275 contained a fragment of Roman brick and a sherd of pottery dated to the late 1st century. A single circular depression with a 0.10m diameter was identified in the base of F137. This feature has been interpreted as a foundation trench for an internal partition wall of Structure 12.

A line of three sub-circular features was identified in the eastern part of Area B (F161, F163, F168). All three of these features were filled with blocks of redeposited subsoil and dark grey clay silt with flecks and small fragments of charcoal (F161 C1351, F163 C1355, F168 C1344). Both F163 and F168 were sub-circular with diameters of approximately 0.75m. F163 was found to have gently sloping sides and was 0.20m deep, while F168 had slightly steeper sides and was 0.25m deep. Although F161 had been truncated on its southwestern side by a later foundation trench, its form suggested that it, too, was originally sub-circular with a diameter of c. 1.20m. F161 was found to be 0.50m deep with gently sloping sides. F163 and F168 have been interpreted as postholes which are probably related to the construction of a building as they are both immediately sealed by floor surfaces. F161 may have been related to these smaller features, but its stratigraphic location is less certain and it is possible that it cuts from a slightly higher level in the sequence.

There is evidence for the construction of timber buildings contemporary with Structure 11 and Structure 12 in other areas of the site, however, this evidence is much more fragmented due to the presence of later Roman foundation trenches. For this reason, features and deposits which may represent the remains of buildings have not been grouped under a structure number. All features assigned to this phase in Areas B and C cut through the ‘made-up ground’ deposits C1328 and C1363. Two gullies defined in the northwestern part of Area B (F165, F173) may represent the remains of construction trenches for timber buildings. Very little of these features survived as both had been truncated by later foundation trenches. Both F165 and F173 were steep sided with concave bases and were found to be c. 0.50m deep. The

20

The Roman Fortress A small amorphous spread of compacted clay and black sand silt would appear to be contemporary with F163 and F168. This very disturbed deposit has tentatively been interpreted as a temporary hearth and may, along with the two postholes, relate to the construction of a timber building in the vicinity.

and fragments of plaster and mortar. This deposit was interpreted as a make-up layer for a floor surface. On top of C1258, a 0.05m wide band of yellowish-red plaster (C1256) appeared to define the position of the interior faces of walls belonging to Structure 11. This plaster band was in turn sealed by a 0.05m layer of compacted gravel interpreted as a floor surface (C1255). These deposits were found to have subsided into the backfills of the possibly unused construction trenches of the previous phase (F143, F145).

In the northeastern part of Area B, bounded by the later foundation trenches F148 and F149, a very disturbed and patchy deposit of compacted gravel (C1340) may represent the remains of an interior floor surface related to the initial construction of a timber building in this area. No such floor surfaces associated with the possible building survived to the south-west of the later foundation trench F153 in Area C. However, a layer of clean yellowish brown silt sand (C1317) in this area may represent the preparation for a floor surface.

Two c. 0.30m wide bands of brown clay with very frequent inclusions of fragments of mortar and limestone (F133 C1259, F135 C1262) covered the northeastern and southwestern edges of the primary fill of the drain (C1282) one of which appeared to respect the edge of the band of plaster. C1259, located on the northwestern edge of F134 contained a small pottery assemblage dated to AD 70-100. These deposits may have been simply an accumulation of debris against the walls of Structure 11 and Structure 12 or possibly a deliberate attempt to provide waterproof shoulders for the drain to reduce dampness within the buildings. However, given the position of the plaster strip in relation to the underlying construction trench, an alternative explanation becomes apparent. The plaster strip (C1256) was situated along the centre of the construction cut (F141). The wattle and daub walls or clay ground walls which are associated with post-in-trench foundations are normally at least as wide as the foundation trenches themselves (Davison 1989, 200-221). Even allowing for a certain amount of subsidence, the position of the plaster strip suggests that the post-in-trench foundations had been superseded by low mortared plinths (opus caementicum) or sill walls. The bands of clay and mortar (F133, F135) may represent the remains of such structures, and indicates that the buildings of Phase B were demolished and rebuilt on sill walls.

Drain / Alleyway F134 was a wide, shallow feature (min.L11.5m xW1.60m x D0.20m) which continued to the northwest and southeast of the excavation area. The sides of F143 sloped gently down to a 0.40m wide central channel cut into the natural subsoil. This feature has been interpreted as an open drain or alleyway lying between two buildings. The earliest fill of F134 is considered to have been a primary rubbish deposit (C1282) which contained a large quantity of animal bone and pottery as well a fragment of opus signinum with a worn surface. The pottery assemblage dates this deposit to AD 75-100. Trackway A substantial, possibly linear feature was identified in the western part of Area B (F164). This feature had fairly straight parallel edges aligned NW-SE with fairly steep sloping sides and a concave base (min.L2.10m x W2.00m x D0.35m). F164 was backfilled with a very mixed deposit of greyish brown silt clay and blocks of redeposited subsoil (C1359). This fill contained frequent inclusions of gravel, pebbles and flecks of charcoal as well as a small quantity of pottery dated to AD 70-100. The location of this feature, underlying and on the same alignment as a later street made up of a succession of surfaces, suggests that this feature represents the remains of a deeply worn trackway which was backfilled prior to the laying of the first street surface in this area. The absence of a similar feature in the northeastern part of Area C, an area also occupied by the later formal street, does cast some doubt on this interpretation. However, a very mixed layer of dark grey clay silt with blocks of redeposited subsoil was identified which was similar to the fill of F164 (C1359).

Further evidence which may indicate that these two structures were rebuilt during this phase was identified within Structure 12. Above the backfill of F140 a possible structural feature was defined (F132). This feature consisted of a thin irregular layer of mortar (L1.20m x W0.30m x D0.02) with a straight edge on its southwestern side (C1264). Along this straight edge a 0.01m wide ridge with vertical sides was also defined. This edge ran parallel to the southwestern edge of F140 at a distance of 0.10m to the northeast. A 1.75m long line of limestone fragments and cobbles apparently bonded with clean yellowish brown clay (F166) covered C1264 and continued further to the southeast along the same alignment. Although F132 can be interpreted as the fragmentary remains of the wall associated with the underlying foundation trench (F140), it is possible that it represents the remains of a later sill beam or mortared plinth which provided the base for a timber-framed building.

Reconstruction of Structures? Within Structure 11, a series of deposits, bounded by F141 and F142 and overlying C1260, was identified, which continued beyond the southwestern edge of excavation. The earliest of these deposits, C1258, was a yellowish-red sandy silt containing frequent clods of clay

In the northwestern corner of Area A, a clay deposit has been interpreted as possible surviving wall make-up. C1297 was found to overly F140 and consisted of brown

21

Excavations at the Minster Library, York 1997 clay sand with gravel inclusions. This deposit had been truncated on its northeastern and southeastern sides by later features and continued beyond the northwestern edge of excavation. The southwestern edge of this deposit, however, consisted of a regular, fairly vertical face, c. 0.15m high. Interestingly, no stones or mortar were found beneath or within the deposit.

Drain / Alleyway The drain or alleyway in Area A (F143) accumulated a further substantial deposit of primary refuse during this phase. This deposit (C1251) consisted of dark brown clay silt with very frequent inclusions of charcoal flecks and small fragments of plaster and mortar. C1251 contained fragments of Roman brick and two fragments of opus signinum as well as a large quantity of animal bone, oyster shell, a copper-alloy ferrule (Find No.210, see Figure 3.23, p66) and pottery. The dating of the pottery assemblage suggests that this context may have been deposited as late as the early second century.

No evidence for floor surfaces was defined within Structure 12, as this area seems to have been badly disturbed during the following demolition episode. Use of Structures Structure 11 contained the only surviving evidence for the use and maintenance of the timber buildings. Overlying the gravel surface (C1255), a thin layer of mortar and limestone fragments (C1254) was interpreted as a later internal floor surface. Lying over C1254 a small, thin spread of charcoal and black silt was defined (C1253) which suggests that a brazier or hearth existed within this structure.

Street The earliest street surface in Area B was laid down directly over the backfill of the probable linear feature (F164). This surface consisted of mortar and pebbles (C1354) overlying a compacted pebble and sand preparation (C1356). C1356 contained a small quantity of pottery dated to AD 75-120. A similar, although not identical, situation was revealed in Area C where a mortar and gravel surface (C1347) was found to cover a friable mortar and gravel preparation (C1348). Although the dating of this surface spans Period 1, the surface is considered to belong to the latest phase suggesting that the street existed as an unsurfaced trackway throughout the late 1st century.

Further evidence appeared to confirm the existence and use of timber buildings in other areas of the site. In the north-eastern part of Area B, bounded by the later foundation trenches F148 and F149 a possible floor surface of compacted gravel (C1340) was assigned to Phase B. This surface was covered by a thin layer of dark grey silt clay and lenses of brownish yellow clay (C1342) which contained a single sherd of pottery dated AD 70100 and a copper-alloy brooch dated to the first century (Find No.342, see Figure 3.23, p66). Above C1342 a layer of dark grey clay silt (C1341) was found to contain pottery dated to AD 70-100. C1342 has been interpreted as an occupation layer, while C1341 would appear to have been a preparation layer for a thick (c.0.08m) mortar and gravel floor surface (C1330).

Period 2 – c. AD 120-c. 375 The stone barracks and road Period 2 is characterised by evidence relating to the demolition of the timber buildings from the previous period and the construction and use of more substantial and durable stone structures (Figure 3.3). The remains of the timber buildings in Area A were completely sealed beneath a c. 0.20m thick, mixed layer of light brown clay silt containing very frequent inclusions of mortar and plaster fragments, lenses of charcoal and brown clay (C1238, C1244). Nowhere else within the excavation area were the results of this apparent destruction episode so clear. A dupondius coin of Domitian recovered from C1244 provides a terminus post quem of AD 81-96 for this deposit, although it also contained a large assemblage of pottery which has been dated to AD 100-120. Other finds included a leaf-shaped harness pendant (Find No.189, see Figure 3.23, p66) dating to the late first century. A series of three dumps of building debris were defined in the southern part of Area A (C1222, C1225, C1226). These deposits were very similar in character to C1244 although pottery derived from these contexts has been dated to AD 115-130.

A series of deposits to the southwest of the road in Area C appear to represent a well-preserved sequence of activity within this phase amounting to approximately 0.40m of stratigraphy. Overlying C1317, a probable floor preparation belonging to Phase B, a thin layer of dark grey silt with frequent gravel inclusions (C1314) was defined to the northwest and southeast of the later foundation trench F152. To the southeast of F152 a compact layer of yellowish-brown silt clay and gravel (C1312) was found to cover C1314 and has been interpreted as a possible floor surface. C1312 was in turn covered by a dark brown clay deposit with frequent charcoal inclusions (C1311) which also covered C1314 to the north-west of F152. C1311 would appear to have been a preparation layer as it was sealed by a compact surface of white mortar (C1310). A further preparation layer seems to have been laid down over this surface (C1295) consisting of dark brown silt-clay and frequent crumbs of mortar. C1295 was found to be covered by a mortar and pebble surface.

C1290 in Area B appeared to represent the same destruction episode. This c. 0.15m thick deposit consisted of dark grey clay silt with lenses of clay and contained very frequent inclusions of mortar and plaster fragments and flecks of charcoal. C1290 contained a small assemblage of pottery dated to the late first century.

22

The Roman Fortress

Figure 3.3 Plan of Period 2 structures and features, stone barracks

23

Excavations at the Minster Library, York 1997 In Area C, the destruction episode was represented by an extensive layer of dark brown clay-silt with frequent inclusions of mortar and limestone fragments (C1270). The small pottery assemblage recovered from this deposit has been dated to AD 100-120. Other finds included three copper-alloy strap or buckle fittings derived from Roman armour (lorica segmentata) (Find Nos 226-8, Fig 3.23, p66) and three limestone throwing balls (Find Nos 262264). The street in Areas B and C showed no sign of having been covered with the type of debris characteristic of the above destruction deposits. This would suggest that this destruction episode represents an organised and deliberate demolition of timber buildings followed by the levelling of the site with unwanted building debris. Although charcoal was present in fairly large quantities in these destruction deposits, there was no convincing evidence for the destruction of the timber buildings by fire.

Figure 3.4 Roman stone walls and footings (Periods 2 and 3, Structure 8, F116 and F117) had survived. These walls were centrally placed over a footing of relatively disordered mortar-bonded limestone blocks which in places spread across the full width of the clay and cobble foundations (1.20m). The foundation trenches underlying walls F116 and F117 (F148 C1303, F149 C1304, see Figure 3.2) contained a small assemblage of pottery dated to the late first to the early second century.

Once the earlier buildings had been cleared and the ground surface levelled, broad foundation trenches were cut through all the earlier deposits and into natural subsoil. Foundations which have been interpreted as the footings for external walls are present in all three areas of the excavation (Area A: F167, F169. Area B: F162, F148, F149, F150. Area C: F153; see Figure 3.2). All of these foundation trenches were found to be vertical sided with flat bases and approximately 1.20m wide and 0.75m deep. These trenches had been backfilled with redeposited, puddled subsoil and rounded cobbles.

Apparently within the interior of Structure 8 a short length of wall survived which appears to have been an interior partition wall (F119 C1210). Even though F119 was stratigraphically later than the exterior wall which it abutted (F118) it is considered to be a result of the same phase of construction, as F119, along with its footing and foundation (F151, see Figure 3.2), was built with the same materials and in the same manner as the exterior walls. Although only a short length of F119 remained (min.L1.05m x W0.60m x H0.16m), as it had been truncated by later features, its foundation trench continued into the northeastern edge of excavation. The width of this wall and its foundation (c. 1.0m) are essentially the same as those of the exterior walls. This suggests that F119 was a major load bearing partition wall within a substantial building and appears to have divided Structure 8 into two rooms.

These foundations and associated features and deposits have been grouped into three structures: Structure 8 situated to the northeast of the street; Structure 9 in Area C; and Structure 10 in Area A and Area B to the southwest of the street. Structures 8, 9 and 10 Structure 8 is defined by two surviving walls considered to represent external walls of a stone building (Figure 3.4). F117 (C1208) lies on a NE-SW alignment forming a corner with F116 at its southwestern end and continues beyond the northeastern edge of excavation (min.L3.25m x W0.60m x H0.20m). F116 (C1207) is orientated NWSE (min.L4.25 x W0.60m x H0.25m) and continues beyond the northwestern facing edge of excavation in the northern part of Area B. This wall almost certainly reemerges as F118 (C1209) in the southeastern part of Area B (min.L2.60m x W0.60m x H0.50m). If this assumption is correct, the total minimum length for the southwest facing exterior wall of Structure 8 is 9.55m.

Intrusive later features had all but erased any evidence for the floor surfaces within the two rooms of Structure 8. On completion of the construction of the walls a layer of olive clay silt and limestone fragments was deposited (C1283) in the northern room, along with a layer of yellowish brown clay and limestone fragments in the southern room (C1211). These deposits were presumably laid down in preparation for a floor surface of some kind of which no trace remained. The small assemblage of pottery recovered from these deposits was dated to the early second century.

F116, F117 and F118 appeared to represent a single building episode as the materials and the techniques employed for their construction were identical. Ordered courses of limestone blocks were laid in mortar to form flush pointed vertical faces with a rubble core of limestone fragments and mortar. Although only two courses of F116 and F117 remained, four courses of F118

Structure 9 was defined by two foundation trenches (F153, F152) situated to the southwest of the road in Area C. F153 was interpreted as a foundation trench for an external wall (min.L3.60m x W1.10m x D0.70m). The

24

The Roman Fortress wall assumed to relate to this foundation had been removed by a later robber trench (F82); however, at the southeast end of the foundation a short length of the footing survived. This footing (F99 C1298) consisted of fairly ordered limestone blocks and fragments within a thick layer of mortar (min.L1.7m x W0.95m x D0.20m). It would seem likely that F153 was the continuation of F162 (see Figure 3.2) from Area B which suggests that Structure 9 and Structure 10 formed two parts of the same building.

survived (F114 C1201). The only piece of external wall belonging to Structure 10 to have survived the later robbing episode was F109 (C1195) found over F167 in Area A. This short length of NW-SE aligned wall (min.L0.85m x W0.60m x H0.15m) was made using the same materials and techniques as the external walls of S8. F109 consisted of one course of limestone blocks with a mortar and limestone fragment core overlying a footing of mortar and limestone fragments. The later robber trench (F130) appears to have removed a wall of similar dimensions overlying F169 which, set with F109, would have formed an external corner to Structure 10.

Perpendicular to F153, a less substantial clay and cobble foundation (F152, min.L2.55m x W0.70m x D0.70m) was defined which was truncated to the southwest by the construction trench for the boiler house. This wall line had also been subjected to a later robbing (F71), but as was the case with F153, most of the footing had survived (F112 C1197). F112 was found to consist of a 0.10m deep layer of mortar and disordered fragments of limestone. A large fragment of second-century brick with an incomplete graffito on one surface was recovered from the footing, but there was no convincing evidence to suggest that this wall was a later addition to the building. F152 was interpreted as an internal partition wall which divided Structure 9 into two rooms.

The only surviving evidence for the internal arrangement of Structure 10 was a small sub-rectangular deposit of clay and cobbles (F156 C1329, L0.70m x W0.50m x D0.15m). F156 had been truncated on three sides by later features. However, given that it lies on the construction level of Structure 10, it has been tentatively interpreted as the foundation for a NW-SE aligned internal partition wall. A further structural feature in Area A may also be associated with Structure 10. A NW-SE aligned clay and cobble wall (F129 C1191, L1.65m x W0.60m x H0.10m) was found to overlie a layer of dark grey clay silt (C1245) which in turn covered the foundation trench for wall F109 (F167). F129 had been truncated by the robber trench (F130) which had removed the postulated elevation of F169. Wall F129 appeared to have no foundation and was built over C1245 which has been interpreted as a layer of made-up ground deposited after the construction of the external walls of Structure 10. The northeastern edge of F129 consisted of a vertical face surviving to a height of 0.08m adhering to which three coats of plaster and/or render survived (C1248). The outer coat of plaster was mottled grey with traces of red pigment.

A mixed layer of dark grey silt-clay was found to overlie the foundation trenches in both rooms (C1229). This layer is considered to have been a preparation layer for a floor surface which appears to have been removed prior to the deposition of C1228. Structure 10 comprised a group of features and deposits which had suffered a great deal of disturbance caused by later intrusions. Three substantial clay and cobble filled foundation trenches define this possible building, F167 and F169 (see Figure 3.2) in Area A and F162 in Area B. F167 lay on a NW-SE alignment forming a corner with F169 to the southeast and continuing beyond the edge of excavation to the northwest (min.L1.5 x W1.15 x D0.75m). F169 was set at 90 degrees from the southeastern end of F167 and ran on a NE-SW alignment continuing beyond the edge of excavation (min.L6.90m x W1.10m x D0.75m). Much of the southeastern edge of this feature had been badly disturbed by a large medieval ditch (F64/59), however, 2.4m from the corner formed by F167 and F169 the edge of F169 was found to turn eastwards through 90 degrees and continue into the unexcavated area bounded by Area A. This change of alignment suggests that a further foundation trench, aligned NW-SE had been present in this area. The cobble and clay backfill bounded by this change of alignment was indistinguishable from the backfill of other parts of F169 (C1365).

The remains of a possible mortar floor surface (C1236) may have been contemporary with this wall. C1236 may well be the continuation of a large area of mortar surface defined in the southern part of Area A (C1206). C1236 and C1206 would appear to be stratigraphically equivalent and were certainly similar in character. The clay wall (F129) and mortar surface (C1206) have been interpreted as external features associated with a substantial stone building represented by Structure 10 and possibly Structure 9. The existence of a plastered clay wall attached to the exterior of a stone building may be explained by the presence of three substantial posts, two of which were later removed forming F104 and F102. F102 cuts through the layer (C1193) which seals the mortar surface C1206 which in turn seals a substantial square posthole in the same location (F136). The post of F108 seems to have temporarily avoided the fate of the other two posts. F102, F104 and F108 would appear to define the position of three substantial vertical timber posts, set at c. 1.60m intervals, which would have existed

The clay and cobble filled foundation trench in Area B formed the northeastern limit of Structure 8 (F162). This short length of foundation trench (min.L2.10m x W1.10m x D0.75m) was found to lie on a NW-SE alignment and may well represent the continuation of F153 from Area C. A small part of the mortar and limestone fragment footing overlying the eastern side of F162 had also

25

Excavations at the Minster Library, York 1997 contemporaneously with the extensive mortar surface (C1206). As these postholes are set along the same axis as the clay wall F129, it is probable that these posts supported a roof which created a covered exterior area attached to Structure 10. The clay wall (F129) appears to have formed part of the boundary to this covered area, possibly providing additional shelter to an entrance into the building.

layer of dark brown clay silt which contained frequent lenses of charcoal and ash (C1193). This layer appears to have been a resurfacing of this exterior area with small quantities of occupation debris being regularly deposited and trodden into the surface. This layer contained a small assemblage of pottery dated to the early second century. Partially covering C1193 a similar layer of brown silt clay was defined (C1192). This deposit was almost entirely composed of lenses of charcoal, ash and dark grey silt. C1192 would appear to represent a gradual accumulation of occupation debris within this area.

A broad linear feature situated along the same alignment as the open drain or alleyway of Period 1 was defined in Area A (F123). It was not clear whether this broad gully was created by the subsidence of earlier deposits, or whether the mortar surface was deliberately applied to the sides and base of F123. F123 was then filled by layers which extended across the southern part of Area A.

Maintenance of Street The street surface in Area B appears to have been resurfaced with a layer of compacted gravel and pebbles (C1302) at some point within this phase. A very similar layer of gravel and pebbles was defined overlying the road surface in Area C (C1327) although sealed between these surfaces a small spread of charcoal (C1333) was identified. The earlier street-side gully (F159) had been filled by a deposit of dark brown silt clay (C1335) containing pottery dated to the early second century.

Street The Period 1 street identified in Areas A and B appeared to have been resurfaced during this phase. In Area B a layer of mortar and pebbles (C1325) was found to cover the earlier surface, while in Area C a similar layer set with a narrow roadside gully was identified (F159, min.L5.25 x W0.40m x D0.20m).

Period 3 – c. AD 375-c. 410 Decline and abandonment

Maintenance of Structures 8, 9 and 10 Structure 8 contained very little evidence for further activity as the interior area had been truncated by later features. Sealing C1283 and C1211, which represented the initial floor preparation layers within the structure, further deposits of a similar character were identified (C1203, C1214); these deposits contained pottery dated to the second century.

Period 3 saw the continued use of the Period 2 buildings during the fourth century with some evidence of maintenance and activity. No evidence survived of fourth-century use or maintenance of Structure 8 as the interior of this building had been truncated by later features. The two rooms of Structure 9 witnessed a further accumulation of deposits which have been assigned to this period. In the northern room a thick layer of relatively clean brown clay, interpreted as a floor surface, was defined (C1154). This deposit contained a small assemblage of pottery dated to the late second to mid-third century.

In Structure 9 a very different situation was encountered. Although the Period 2 stone walls had been robbed out in the post-Roman period, the sequence of the interior floor levels remained intact. In the northern room of Structure 9, the first deposit assigned to this phase consisted of light brown clay covered by thin patches of mortar (C1228). This deposit has been interpreted as a floor preparation and the remains of a mortar surface. A series of three thin clay or clay silt layers sealed C1228 (C1219, C1223, C1218). All of these deposits contained frequent charcoal inclusions. C1219 also included the possible remains of a mortar surface. These deposits have been interpreted as floor surfaces into which some occupation debris had been trampled.

Lying in a shallow scoop on the top of this layer, approximately 0.70m from the external wall and 1.20m from the internal wall, a circular hearth was identified (F70). This hearth (Figure 3.5) consisted of a base made of large fragments of brick and tile with smaller fragments set into the edges of the scoop forming the edge of the hearth (C1187). The base of the hearth had clearly been burnt in situ. A deposit of dark grey silt and fine gravel (C1184) was found to cover the base of the hearth, which in turn was covered by a mixed deposit of brown clay silt containing frequent small fragments of brick and tile (C1183). Both of these deposits are considered to be secondary relating to the disuse of the hearth. Archaeomagnetic dating of samples from the base of the hearth produced four possible date ranges; AD 160-200, AD 275-420, AD 1275-1310, AD 1440-1475. Given the stratigraphic position of this feature, it is likely that it was in use during the fourth century within the second date range of AD 275-420. A 0.20m layer of yellowish brown silt-clay in the southern room of Structure 9 (C1185) represents the last surviving floor

In the southern room of Structure 9 a similar, although not identical, series of floor surfaces with traces of occupation debris were defined (C1292, C1273, C1269, C1231, C1227). The finds recovered from these deposits in both rooms have been broadly dated to the second century. Structure 10 contained no further evidence of use or maintenance. The mortar surface (C1206) in Area A was covered by a

26

The Roman Fortress level in this part of the building. This deposit contained sherds of second century pottery. Structure 10 contained no evidence for use or maintenance in this phase.

comprising of cobbles, pebbles and dark grey clay silt. The southwestern side of the street had been truncated by a later pit (F101) although two features probably represent the remains of the opposing road side gully (F113, F124). F113 contained two fills of dark grey clay silt and pebbles (C1199, C1200), while F124 contained a single fill of dark grey clay silt and pebbles (C1239). The small pottery assemblage recovered from the fills of these gullies has been dated to the second and third centuries.

Figure 3.5 Late Roman hearth (Period 3, F70) Drain The southern part of Area A appears to witness a considerable amount of redevelopment work during this period. The substantial posts which are thought to have supported a roof creating a covered external area were removed (F102, F104, F108), hinting at a fairly major refurbishment. The presence of a two less substantial postholes in the vicinity of F102 (F73, diam.0.30m x D0.35m, F76 diam.0.35m x D0.25m) indicates that this structure was replaced or repaired rather than being simply removed.

Figure 3.6 Late Roman pebbled surface (Period 3, C1118, C1128) set with eavesdrip drain (F88) and verandah postholes (F76, F73)

A linear, unbonded stone feature was then built within a shallow trench which coincided with the alignment of the earlier drain in this area (F88 C1152, min.L6.90 x W0.85 x D0.25m). Unfortunately, this feature had been partially removed by a later robber trench (F77), however, it appeared to consist of rows of limestone blocks forming a narrow (0.15m), central channel with a base of flat limestone slabs. This feature has been interpreted as a drain, possibly an eavesdrip drain for the roof of the external covered area.

A narrow linear feature set along the centre of the road was identified within the make-up of the road surface (F121). Two rows of large cobbles formed a narrow channel (min.L2.20m x W0.30m x D0.10m) which was filled with loose gravel and pebbles (C1233). This feature was initially interpreted as a wheel rut due to the very worn appearance of the cobbles forming the side of the channel, but on further examination it seemed to have been intentionally set into the road surface and is considered to represent a further drainage gully associated with the latest street.

Compacted pebble surfaces appear to have been laid down to the northeast (C1118) and southwest (C1128) of the drain. Pottery recovered from C1118 has been dated to the mid-fourth century (Figure 3.6).

The latest street surface and related features defined in Area C mirrored the situation encountered in Area B. The surface consisted of a compact layer of cobbles and pebbles (C1153) which was bounded to the southwest by a possible roadside drainage gully (F100). F100 was found to be a shallow gully (min.L5.30m x W0.40m x D0.20m) filled with dark grey silt clay and pebbles (C1176). The central channel defined in the surface in Area B was also present in this area (F93). The cobble edges of this feature were also found to be very worn.

Street The final resurfacing of the street in Area B consisted of a thick layer of compacted cobbles and pebbles (C1230). This 3.00m wide surface was bounded to the northeast by a possible roadside gully (F84, min.L5.10m x W3.00m x D0.20m). This feature was poorly defined as its backfill (C1140) was very similar to the adjacent street surface

27

Excavations at the Minster Library, York 1997 No evidence for the collapse of walls or roof structures sealed beneath the dark earth layer was defined within the interiors of Structure 8, 9 or 10. Deposits overlying the street surface in Area B, however, do seem to provide evidence for at least the initial decay of Structure 8 and possibly Structure 10 prior to the deposition of the dark earth.

Preparation of the Fortress Site The remarkably clear interface between the natural subsoil and the overlying deposit of sterile greenish-grey clay silt observed during the excavation strongly suggests that this area of the fortress site had been levelled in preparation for the initial phase of construction work. This conclusion appears to confirm the results of environmental studies undertaken as part of the York Minster excavations (Phillips and Heywood 1995, 180). The degree to which this levelling operation had truncated the previous ground surface is uncertain. It would appear, however, that the earlier topsoil and, given the unblemished condition of the upper surface of the subsoil, at least 0.10m of natural subsoil were removed. The resulting flat pre-construction ground surface was found to lie at c. 12.45m AOD. This level suggests that the pre-construction ground surface of the fortress not only sloped down towards the eastern (c. 12m AOD) and southern corners (c. 11m AOD) from its highest level in the western corner (c. 14m AOD) (Ottaway 2004, 33), but also sloped down towards the northern corner beyond the Minster Library.

A spread of thin sandstone slabs, which appeared to have been broken in situ, marked the interface between the backfill of the roadside gully F84 and the overlying layer (C1203). These slabs have been interpreted as roof tiles and, given their close proximity to the external wall of Structure 8 (F116), may well represent the first signs of the collapse of the roof of S8. The worn street surface in Area B was sealed by a thick layer of dark grey silt clay and limestone fragments (C1190). This deposit also contained fragments of thin limestone slabs and ceramic roof tile including one bearing the stamp of the 6th Legion (see Figure 3.21, p62). A deep accumulation of deposits overlying the street in Area B possibly represented the gradual collapse of Structures 8 and 10 (C1171, C1170, C1145, C1107, C1104, C1098, C1099). This series of deposits initially appear to have abutted the northeastern wall of Structure 10 prior to its later robbing. These contexts consist of dark greenish grey clay silt deposits containing a very high proportion of limestone fragments, mortar and ceramic tile fragments. Finds from these deposits included seven limestone throwing balls (Find Nos 327333) and pottery dated to the late fourth century.

The rather thorough site preparation technique which appears to have been employed by the Roman army potentially provides an alternative explanation for the lack evidence for pre-Roman activity as this initial phase of the Roman fortress development would have obliterated all but the most robust of pre-Roman remains. The layer of sand and gravel identified throughout most of the excavation was probably laid down shortly after the completion of the levelling operation. The presence of this imported material is not surprising as without it poor ground conditions would have created serious problems during wet weather due to the impermeable nature of the subsoil.

Discussion Period 1 The construction of the Roman legionary fortress at York (Eboracum) is thought to have started with the arrival of the 9th Legion in AD71. The many, often small scale, archaeological investigations which have been undertaken within the fortress area during the mid- to late twentieth century have successfully defined the basic plan of the fortress, the form and development of its defences and, in a few areas, the detailed layout and development of its buildings. Many of these investigations have concentrated on the defences although the most significant interventions within the interior were located in the front (praetentura) and central parts (principia and latera praetorii) of the fortress (see Figure 3.1). The Minster Library excavation area was located within the rear of the fortress (retentura) and, as a result of the evaluation programme, it was predicted that the excavation would encounter the remains of barracks and the fifth main street of the fortress (via quintana) (Phillips and Heywood 1995, 160). The opportunity to examine the layout and development of buildings and a street within this area of the fortress was, at the time, considered to be the principal rewards of this research project (Phillips 1996).

The Street As predicted, a NW-SE aligned street was identified in the western part of the excavation area. The street, however, was found to be only c. 3.00m wide which, by comparison with other known streets from the fortress, suggests its function as a minor street rather than a main street. This seems to confirm Ottaway’s (2004, Figure 13) postulated layout of the fortress where the via quintana was situated immediately to the rear of the principia. The position of the minor street seems to have been established during or immediately after the initial site clearance operation and continued in use along the same alignment throughout the life of the legionary fortress. Initially, the street appears to have been an un-surfaced trackway evidenced by a broad, linear feature which was most probably created through wear by passing traffic. The date at which the street was first surfaced is uncertain, although it sealed a small pottery assemblage dated to AD 71-100. The earliest surface clearly pre-dates the demolition of the Period 1 buildings, but it is possible that the street remained unsurfaced until the early years

28

The Roman Fortress of the second century.

buildings founded upon sill walls or mortared plinths, set upon the earlier foundations. The presence of a plaster strip situated over the centre of one of the post-in-trench foundations provides the most tangible basis for this conclusion. Similar plaster strips were found conclusively to define timber buildings which replaced earlier post-intrench buildings at 9 Blake Street (Hall 1995, 312; 381). The absence of surviving floor surfaces relating to the post-in-trench buildings does seem to cast some doubt on the latter interpretation, although it is interesting to note that the floor preparation which was contemporary with one of these buildings was characteristic of the type normally associated with opus signinum floors (Phillips and Heywood 1995, 39), several fragments of which were recovered from the adjacent drain fills (p21-22).

The Timber Buildings - Structures 11 and 12 As is often found on fortress sites, the remains of the earliest structures had been heavily truncated by the foundations of later buildings. Fortunately, the remains of the earliest buildings in the southwestern part of the excavation area were relatively well-preserved due to a change of layout in the following period. In this area, two NW-SE aligned post-in-trench foundations separated by a 1.50m wide drain or alleyway were defined (Figure 3.7). These foundation trenches are considered to have supported the external walls of Structure 11 and 12 with internal partition walls built on the less substantial postin-trench foundations. At least one post-in-trench internal foundation was found to be sealed by the earliest surviving make-up layer within one of these buildings. It is not certain whether this feature represents a mistake during construction, or an early modification to the building.

Whether the features and deposits assigned to this period represent a single phase of buildings constructed on postin-trench foundations with plinths for sill beams, or two phases of buildings, the first constructed on post-intrench foundations and the second on sill walls or even mortared plinths (opus caementicum) such as those identified at Colchester (Crummy 1984, 3, 20-21), the principal building material besides timber is certain. The thick destruction layer which covered the entire excavation area consisted mainly of silt clay. The wall make-up of these buildings, therefore, most probably consisted of wattle and daub or clay mud bricks. The buildings were certainly plastered on the interior and appear to have been decorated with white, black, red and

The interpretation of the rather complex and tightly spaced stratigraphic sequence related to the above-ground remains of the timber buildings has not proved to be clear-cut. Although it is possible that these remains simply represent the construction, use and maintenance of Structure 11 and 12, the evidence does seem to suggest that the timber buildings supported by the post-in-trench foundations were dismantled and replaced with similar

Figure 3.7 Projected plan of Period 1 timber barracks and street, includes © Crown Copyright Ordnance Survey. All rights reserved.

29

Excavations at the Minster Library, York 1997 blue-green paint probably in simple designs including thin lines and dots. Gravel and mortar surfaces on clay preparations appear to have been the principal flooring materials. As was also found at Blake Street, there was no indication that posts or horizontal timbers were left in situ, suggesting that they had been removed for re-use during demolition.

from these deposits is also noteworthy (p73), as this anomaly was also identified within an animal bone assemblage from a similar deposit at Blake Street (O’Connor 1987). The results of the analysis of two welldated, primary assemblages from the legionary fortress at York appear to contradict the situation encountered on most Roman military sites in Britain, where cattle remains form the largest proportion of bone assemblages. The suggestion that this refuse was derived from a hospital (valetudinarium) based on the high proportion of bottle fragments within the glass assemblage (p70), does not appear to be relevant given the interpretation of the adjacent buildings. However, the position of the contemporary hospital at Inchtuthil (Pitts and St Joseph 1985, 91) within this part of the retentura should not be disregarded.

It seems likely that the fragmented structural remains identified immediately to the southwest of the street form part of Structure 12 which was more clearly defined to the northeast of the drain. A small part of a probable postin-trench foundation, which was otherwise entirely truncated by later features, is considered to have been the remains of the northeastern external wall of Structure 12. The remains of the probable drainage gully to the northeast of this feature would appear to have been too far from the street to have effectively provided drainage for it and, therefore, may have provided drainage for the building.

Periods 2 and 3 The 9th Legion to 6th Legion handover (lacuna) The remains of the Period 1 timber buildings were sealed beneath a thick layer consisting of clay silt, clay, mortar and plaster. This deposit appeared to be comprised entirely of the materials used in the construction of the earlier buildings and has therefore been interpreted as a destruction layer resulting from the demolition of these buildings and the subsequent levelling of the site with the resulting un-reusable debris. The destruction layer was found throughout the excavation area with the exception of the street surface. The large pottery assemblage from this deposit was dated to AD 100-120 with a dupondius coin of Domitian providing a terminus post quem of AD 81-96. The throwing balls and fittings from lorica segmentata were also recovered from this deposit suggesting that the large finds assemblage was derived from the latest occupation of the Period 1 buildings rather than necessarily being contemporary with the demolition operation.

The remains of Structure 11 and 12 in the southwestern part of the excavation can be interpreted as the rear walls of a barrack pair divided by a narrow alley which also served as the drain for the eavesdrip of the buildings. The surviving part of the probable front wall suggests that Structure 12 was c. 7.50m wide, which, although narrow, is not inconsistent with the typical width of the men’s quarters of a late 1st-century barrack. The drainage gully to the northeast of Structure 12 was situated c. 2.00m from the front wall and may, therefore, have acted as an eavesdrip drain at the edge of a verandah. If correct, this interpretation indicates that a pair of barracks were laid out NW-SE, with their long axis parallel to the via principalis (per scamna), and with the verandah of the first barrack fronting onto the street. Although a structure, evidenced by probable interior floor surfaces, appears to have been situated to the northeast of the street, little can be said with any certainty about its form and function. The drainage gully and postholes in this area hint at the presence of a NW-SE aligned (per scamnum) barrack with its verandah fronting onto the street, effectively mirroring the situation on the opposing side of the street.

The date of AD 100-120 for the abandonment of the Period 1 buildings coincides with the conventional date for the departure of the 9th Legion from York and its replacement by the 6th Legion, between AD 115 and 122. As is the case with many of the earlier investigations within the fortress area, the excavation produced fragments of roof tile bearing the stamp of the 9th Legion (see Figure 3.21, p62), although both of these fragments were recovered from post-Roman deposits.

The fills of the eavesdrip drain between the barracks to the southwest of the street contained assemblages of primary refuse. The earliest assemblage, which is assumed to have been contemporary with the postulated first phase of timber barracks, was dated to AD 75-100. The second assemblage is considered to have been contemporary with the second phase of timber barracks and was dated as late as the early second century.

The buildings The date of the demolition of the Period 1 timber buildings and the construction of the Period 2 stone buildings is uncertain. Although deposits associated with the construction of the Period 2 buildings were found to contain pottery assemblages, the bulk of this material is considered to be residual as there is little doubt that early second-century refuse, derived from the demolition layer, littered the construction site. The limited insight offered by these assemblages does, however, suggest a midsecond-century date for the construction of the Period 2 buildings which would be consistent with the dating of

The apparent status of the refuse, supported by a high frequency of small cups marked with the user’s name, suggests that it was derived from soldiers rather than officers (p50). This conclusion fits well with the interpretation of the adjacent buildings as the men’s quarters of barracks (contubernia). The high proportion of pig remains within the vertebrate assemblage derived

30

The Roman Fortress the stone buildings from Blake Street and elsewhere in York (Hall 1995, 385). The apparent hiatus in activity suggested by the available dating evidence is also consistent with the findings of earlier investigations within the fortress. This decrease in activity appears to date between c. AD 120 and c. AD 150 and is thought to be related to the posting of most of the 6th Legion to the frontier in Scotland during this period (Ottaway 2004, 57). Unfortunately, there was no evidence to suggest whether the Period 1 buildings had been demolished and the area left as open ground for a long period of time, or whether these buildings were demolished immediately prior to reconstruction.

and mortar both of which were centrally placed over 1.00m-1.20m wide foundations of puddled clay and cobbles. The foundation trenches for these walls had been cut through all the earlier deposits allowing the foundations to rest upon undisturbed natural subsoil. Ironically, although the walls had survived the ravages of medieval robbing, the internal floor levels had been all but eradicated by a late medieval ditch. Equally, no traces of internal or external rendering or plastering had survived. The conclusive interpretation of Structure 8 and inference of its layout is problematic due to the small size of the area of this building revealed by excavation. This limitation is further exacerbated by the absence of surviving floor surfaces and internal features. The presence of what would appear to be an internal partition wall, although built to external wall specification, does, however, provide an important clue to the layout and function of this building. The obvious explanation for the presence of this heavily over-specified internal wall is that it represents the remains of a major load bearing wall. This suggests that Structure 8 forms part of a substantial broad building such as a hospital (valetudinarium) or fabrica which, unlike the long narrow barrack blocks, required internal load bearing walls. On closer examination of the dimensions of Structure 8 and its position within the postulated layout of this area of the fortress, another, perhaps more plausible, interpretation of Structure 8 as a barrack becomes apparent.

The Stone Buildings - Structures 8 and 9/10 The excavation defined two substantial buildings which are considered to have been the result of a single building episode (Figure 3.8). To the northeast of the street three stone walls which appeared to form an external corner and internal partition of Structure 8 had partially survived the medieval robbing which affected so many of the Roman stone structures within the fortress. All three of these walls were built of coursed limestone facing (petit appareil) upon a mortared rubble core and were found to survive to a maximum of four courses in height (c. 0.50m). The width of the three walls was found to be c. 2 feet (0.60m) which conforms to standard legionary practice for the construction of external stone walls (Phillips and Heywood 1995, 172). The walls were constructed over footings consisting of limestone blocks

Figure 3.8 Projected plan of Period 2 and 3 stone barracks and street, showing Building 7, includes © Crown Copyright Ordnance Survey. All rights reserved. 31

Excavations at the Minster Library, York 1997 Clearly, barrack blocks were the most likely building type to have been constructed in this corner of the retentura. If Structure 8 did represent the remains of part of a barrack block, the position of the exterior corner implies that it would have been orientated NE-SW with its long axis at right angles to the via principalis (per strigam) and, as was the convention, with the officers quarters fronting onto the perimeter street (via sagularis) to the northeast.

southwestern exterior wall of Structure 8 was found to be a minimum of c. 1.60m as it continues beyond the edge of excavation and does not, therefore, discount the interpretation of Structure 8 as a barrack. The second substantial building of this period, Structure 9/10, was situated to the southwest of the street. Remains of this building were identified in different areas of the excavation but are considered to represent the remains of a single building. Furthermore, Structure 9/10 is also considered to be part of Building 7 identified in the Minster Library Chapel excavation (Phillips and Heywood 1995, 160-163).

The over-specified internal wall divided Structure 8 into two rooms with the internal width of the northwestern room being c. 7.00m. By comparison, this measurement is typical of the internal width of a barrack (excluding verandah). However, barrack blocks were divided into rooms (contubernia) with an internal partition wall running parallel to the long axis of the barrack dividing it into sleeping quarters (paplii) and storage areas (arma). Given the often superficial nature of these internal partitions and the degree to which the interior of Structure 8 had been truncated by later features, only a small part of this partition wall could have survived where it abutted the external wall at the southwestern end of the postulated barrack. No such partition wall was identified during the excavation although, by comparison with the dimensions of excavated contubernia, it would have abutted the external wall in the area which lay beyond the limits of the excavation. A further possible explanation for the absence of a partition wall may also be found through comparison with excavated barrack blocks. The layout of the end or last contubernia is often found to be atypical and it has been suggested that these rooms were used for purposes other than men’s quarters, such as storage (Davison 1989, 33-35), in which case an internal partition wall may not have been present.

The excavation in the Minster Library Chapel defined part of a substantial stone founded building divided into five small rooms with a possible external area to the southwest (Building 7, see Figure 3.8). The construction of this building was tentatively dated to the late first to early second century and was interpreted as a verandahfronted barrack. Although the remains of the building had been badly damaged by post-Roman activity, the wide clay and cobble foundation and mortar footing for the external wall suggested it had originally been a two foot wide (0.60m) stone built wall. The surviving foundations for the internal partition walls were built of clay and cobbles with mortar bonded stone footings. These partition walls were thought to have been the result of a later modification, replacing earlier timber partitions of which no trace remained. Structures 9 and 10 were found to have suffered a similar degree of post-Roman disturbance as Building 7 although a short length of NW-SE aligned wall forming part of Structure 10 had survived later robbing attempts and truncation by a large medieval ditch. Only one course of this petit appareil type wall had survived, but the width (0.60m) and construction techniques of the wall, its footing and foundations were identical to those of the walls forming Structure 8. This surviving part of the stone wall had been built upon a footing of limestone blocks and mortar; both the wall and its footing were centrally placed over a c. 1.20m wide foundation of puddled clay and cobbles. This, and other very similar foundations which formed Structure 9 and Structure 10, were interpreted as the external wall foundations of a large building.

Although the position, construction techniques and internal dimensions of at least part of Structure 8 are consistent with that of a typical barrack, the position of the substantial internal wall appears to support the alternative interpretation of Structure 8 as a hospital or fabrica. If, however, this wall is viewed as being the exterior wall of the contubernia a logical explanation for the continuation of the southwestern exterior wall presents itself. The standard Roman barrack block takes the form of an elongated L-shape, with the officers’ quarters, which are normally an integral part of the barrack block, being wider than the men’s quarters. The front of the men’s quarters were normally provided with a verandah, the substantial timber posts of which are assumed to have supported one side of a rectangular roof which covered the entire building. The construction of an end wall which spanned the full width of the building would not, therefore, be surprising as it would offer a great deal more support to the corner of a substantial roof than would a single timber post.

The northeastern foundation of this building was situated immediately to the southwest of the street. The surviving part of a petit appareil wall ran parallel to this foundation approximately 10.00m to the southwest but the continuation of its foundation was found to turn through 90 degrees towards the northeast, presumably joining the foundation which ran parallel to the street. The badly disturbed remains of a further clay and cobble foundation were found to project from the NE-SW foundation approximately 2.00m from the external corner and continued in a southeasterly direction, apparently on the same orientation as the surviving wall.

Barrack verandas vary in width from 1.50m to 3.75m (Davison 1989, 19), or proportionally appear to have been 1:3.5 of the men’s quarters (Phillips and Heywood 1995, 172) which in this case suggests that the verandah would have been c. 2.00m wide. The continuation of the

The remains of a narrower (0.70m) clay and cobble

32

The Roman Fortress foundation with a rubble and mortar footing have been interpreted as an internal partition wall of Structure 9/10. This foundation was set at 90 degrees to the broader NWSE foundation which ran alongside the street. Although no trace remained of the partition wall which it once supported, this foundation was clearly designed to support a stone wall. There was no evidence to suggest that it was the result of a later modification of the building, and it is more likely that this foundation relates to the initial construction of Structure 9/10.

To the southwest of Structure 9/10, the presence of two substantial postholes and a clay wall appears to confirm that the contubernia were fronted by a c. 2.80m wide verandah, possibly with a porch at the officers’ quarters end. The two postholes which certainly belong to the initial phase of construction of the building appear to continue the alignment created by the front wall of the officers’ quarters. The posts appear to have been set at roughly 1.5m intervals. The size of the postholes, and in one case a post impression in the base of the feature, suggests that they bore timber posts with a diameter of c. 0.20m. A c. 1.5m length of unfounded clay wall was found to lie between the corner of the officers’ quarters and the first timber post of the verandah. The wall was plastered on its interior face and is considered to have been part of a porch arrangement for the entrance into the officers’ quarters from the verandah. The exterior floor surface beneath and to the southwest of the verandah consisted of a mortar surface incorporating a shallow gully which most probably provided drainage for the building’s eavesdrip.

More convincingly than was the case with Structure 8, Structure 9/10 and Building 7 have also been interpreted as having been parts of a barrack. The two short lengths of NW-SE aligned foundation which were situated beside the street are considered to be part of the rear external wall not only of Structure 9/10, but also of Building 7. The external corner at which the sole surviving fragment of stone wall was encountered appears to have been the corner of the officers’ quarters. At this point the building becomes narrower forming the characteristic elongated Lshape of a standard barrack. The postulated barrack, therefore, was orientated NW-SE with its long axis parallel to the via principalis (per scamnum) with the officers quarters fronting onto the via sagularis to the northwest. There was no evidence to suggest that the stone walls of the barracks were anything more than ground walls designed to support a timber superstructure. However, it should be noted that the foundations and stone walls of these buildings were no less substantial than those of Barrack 1 of the York Minster excavations (Phillips and Heywood 1995, 93-104) which was found to have full height stone walls.

Maintenance and modification No indication of later maintenance and modification survived within Structure 8. However, a deposit of fragmented sandstone roofing tiles identified at the base of the exterior wall of Structure 8 were found to cover the latest street surface, suggesting that it had continued in use until the abandonment of this area of the fortress by the Roman army. Likewise, there was no evidence to suggest any major structural modification to the interior of Structure 9/10. Regular maintenance of the interior was witnessed by a series of replacement floor surfaces in two rooms. These surfaces consisted of clay or mortar, although it is likely that the clay deposits were often used as preparation layers for a mortar surface. Although this series of deposits offered little useful dating evidence, a probable fourth-century hearth was found to overlie the final surface. The hearth was situated close to the rear wall of the room and conforms to the classic position of a hearth within the sleeping quarters of a barrack (Davison 1989, 231). No evidence for cooking activity was recovered from the fills of the hearth suggesting that its primary function was for heating purposes.

Although fragmentary, the evidence suggests that the officers’ quarters were approximately 10.80m wide while the men’s quarters appear to have been c. 8.00m wide. The contubernia appear to be partitioned in the conventional manner, with the larger paplio to the rear and the smaller arma at the front. Internally, these rooms appear to have been approximately 3.10m wide; the arma c. 2.20m deep and the paplio c. 4.00m deep. The earliest internal foundations defined in Building 7 consisted of clay and cobbles with a mortar footing. In places, up to three courses of petit appareil wall had also survived. These foundations appear to be very similar to the internal wall foundation of Structure 9/10 although it is interesting to note that the internal longitudinal foundation of Building 7 was of similar proportions to those interpreted as external wall foundations for Structure 9/10.

A probable fourth-century modification was defined within Building 7. A 0.60m wide stone wall built upon a bedding of cobbles and clay was found to overlie earlier floor surfaces. The re-flooring of Room 5B which followed the construction of the wall contained fourthcentury pottery. The construction of this wall resulted in reduction of the width of Room 5B to c. 2.10m.

A sequence of internal floor surfaces was found to have survived on either side of the internal partition of Structure 9/10. These rooms were presumably sleeping quarters, the earliest floor surfaces or floor preparations of which consisted of clay. In Building 7 the earliest floor surface of the probable arma consisted of gravel. No evidence for the finishing of the internal or external walls survived.

The verandah and exterior area to the south-west of S9/10 provides the best preserved sequence of maintenance relevant to this building. The mortar surface which initially covered this area was replaced with a layer of clay at which point at least some of the verandah posts appear to have been renewed. Lenses of charcoal and ash

33

Excavations at the Minster Library, York 1997 accumulate on the surface of the clay layer, presumably derived from the emptying of braziers or hearths in the contubernia. This clay layer was later replaced by a similar layer upon which further deposits of charcoal and ash accumulated. At some point during or after the midfourth century a 0.15m wide drain consisting of unbonded limestone blocks on a base of limestone slabs was constructed, apparently to replace the earlier gully. This feature has been interpreted as an eavesdrip drain and finds a parallel with a similar feature excavated at Carrawburgh (Davison 1989, 227). The verandah and exterior area were then metalled with a compacted gravel and pebble surface. The posts supporting the verandah were also replaced during this operation.

investment. Conclusions The excavation located the postulated minor street which is thought to divide the retentura into two equal-sized strips (Ottaway 2004, Figure 13). To the southwest of the street parts of a stone building were identified which, along with similar remains defined by the Minster Library Chapel excavation, appear to be part of a standard Type A barrack (Davison 1989). This single barrack appears to have been orientated NW-SE (per scamnum) with the officers’ quarters fronting onto the via sagularis and its rear wall facing the street. To the northeast of the street a further stone building was identified. It is not certain whether the building represents part of a hospital or fabrica, or the end of a barrack orientated NESW (per strigam) with its verandah facing to the southeast.

The deposits which covered the final surface on the southwestern side of the street included possible fragments of sandstone roofing slates and roof tile including one fragment bearing the stamp of the 6th Legion. This suggests that Structure 9/10 was in use until the abandonment of this part of the fortress.

In both cases, the length of these possible barracks is uncertain, and therefore it is not possible to assess the effects of the postulated second-century widening of the intervallum (Ottaway 2004, 36) on the dimensions of these buildings. The width of the barrack to the southwest of the street was c. 1.25m narrower than that of the Davygate barracks (RCHME 1962, 40). This suggests that the normal six barrack group may have been constructed in this area of the retentura (Figure 3.9).

The Street The second metalling of the street surface is assumed to have been laid more or less contemporaneously with the construction of the Period 2 stone buildings, although the surface and associated deposits did not contain any dateable material to confirm this. The resurfacing consisted of pebbles set in mortar and, apparently for the first time, was set with narrow drainage gullies. An undated resurfacing with compacted gravel and pebbles was undertaken prior to the final resurfacing. It would appear that no attempt was made to re-cut the drainage gullies during this operation. The final resurfacing of the street is also undated although it may have been contemporary with overhaul of the exterior area to the southwest of Structure 9/10. The evidence suggests that this final maintenance operation resulted in the most sophisticated and durable street surface. The 3.00m wide surface consisted of compacted cobbles and pebbles with what appears to have been a central gutter and kerb of larger cobbles. Fairly substantial drainage gullies were also provided at either side of the street. The cobbled surface seemed to be heavily worn, particularly along the central gutter, which suggests it was well used prior to the abandonment of the fortress. The features and deposits which relate to Periods 2 and 3 were found to contain no primary refuse deposits. The finds assemblages recovered from these deposits were generally much smaller than those derived from the earlier period and were found to contain a high proportion of residual material. With the exception of a dagger recovered from a post-Roman deposit (C1121, Find No.198 see Figure 3.22, p64), very little thirdcentury material was recovered during the excavation. This paucity of third-century material reflects the results of earlier investigations within the fortress and is thought to be associated with a period of decline in public

34

The Roman Fortress

Figure 3.9 Postulated layouts of stone barracks across the north corner of the fortress, includes © Crown Copyright Ordnance Survey. All rights reserved. 35

Excavations at the Minster Library, York 1997 be identified in the hand. Two fabrics, E4 and E5, ironrich and chalky variants of Ebor 1 fabric, are rarer and require microscopic examination in order to refine the identification. This has not always been possible to achieve within the current time-scale and, therefore, these fabrics may be under-represented.

The Roman assemblage The Roman pottery - Barbara Precious and Alan Vince Summary A moderate-sized assemblage of Roman pottery was recovered from the excavation. It consisted of 2189 sherds, 42.08 EVEs and weighed 42.992Kg (see Appendix A for the full catalogue). The sequence spanned the late first to late fourth/early fifth centuries. The earliest phase of pottery use in Eboracum was probably present, both in the stratified sequence and as residual sherds, but the assemblage contained few of the diagnostic wares and forms. The early second century, corresponding to the period between the departure of the 9th Legion and the rebuilding of the fortress in stone, was poorly represented. The Roman levels were covered with a dark earth deposit which, to judge from the ceramic assemblage, was derived from late fourth/early fifthcentury refuse rather than being the result of disturbance of earlier levels.

Unusual amphora sherds have been examined by Dr D F Williams whose comments have been incorporated into the main text. Mortaria stamps and rare or unusual types have been analysed by Kay Hartley and the results are presented separately. Brenda Dickinson has provided a report on the samian stamps and Felicity Wild has examined the decorated samian wares from contexts stratified as Roman. Functional interpretation of the pottery has been based on the work of Greene (1993, 3-124), for example beakers have ‘drinking’ as their main function, while cooking vessels are assumed to have been used as ‘kitchen’ equipment. The latter function is emphasised within the field containing information about the condition of the pottery where the area of the vessel burnt during the cooking process is detailed. This field also lists the degree of abrasion, wear, burning either by cooking or fire, and any deposits that occur on the vessel, such as mortar or lime-scale, as well as whether the vessel had been used in antiquity, which is particularly relevant for the study of locally manufactured wares and imported South Gaulish samian.

Of particular interest was a series of deposits within and overlying drain F134. All three deposits date to the late first to very early second century. They were characterised by the presence of numerous smashed vessels and probably represent primary refuse, reflecting the suggested use of the area as a contubernia within a barrack block. The Roman levels were covered with a dark earth deposit which, to judge from the ceramic assemblage, was derived from late fourth/early fifth century refuse rather than being the result of disturbance of earlier layers.

All the pottery from stratified Roman layers has been quantified using sherd count, weight, estimated vessel equivalent (EVE), and diameter for statistical analyses. Roman pottery from post-Roman layers has been measured using sherd count and weight only, which provides information about the degree of fragmentation. The tables of data used in the text have been assembled using Access. All tables can be found in Appendix A, but are referred to in the text, where applicable.

Methodology In order to make direct associations with comparative sites, the format of the Roman pottery discussion for this site (YML97) has been arranged according to that used for the Roman pottery from the fortress, Monaghan, 1993.

As there is an extant typology of Roman pottery from York (Monaghan 1997), vessels selected for illustration concentrate on those which provide information for the date of major groups. Other vessels have been selected for their intrinsic value. The figures have been arranged within stratified groups by fabric with the local wares first in order of form: flagons; jars; beakers; bowls; dishes; lids and other.

The database archive has been recorded using Microsoft Excel 5.0a and consists of nineteen encoded fields listed as follows: Context; Period; Find number; Fabric; Form; Decoration ; Function ; Condition; Use; Number of Vessels; Publication Number; Drawing Number; Comments; Join; Sherd family; Sherds; Weight; Estimated Vessel Equivalent; Diameter. All the abbreviations and codes used in the database and the text are listed in Table A9.

Ceramic History of the Site (Tables A1-5) First-century deposits

The fabric and form codes used in the database are those detailed in Monaghan, 1997, 976-1024 (form), 10261030 (fabric), with the exception of the samian forms which have been recorded using the conventional forms, for example - 18/31. A binocular microscope x20 has been used for the descriptions of rare or unusual fabrics. The bulk of the early Roman ceramics consists of the oxidised Ebor fabric (E0-11), the majority of which can

The three large groups from Area A form a sequence. There is very little cross-fitting between the deposits (Table A10) and internal differences in the character of the pottery confirm that the three episodes are separate. Given that internal evidence suggests that the dumps overlying the drain date to c.100-120 these earlier deposits must be dated c.71 to c.100 but whether they are

36

The Roman Fortress evenly spread in time or not is impossible to say.

three first-century deposits, C1298, C1270 and C1311.

Ceramic Group 1 c. AD 71/79-100 [CP 1a-1b] (Figure 3.10, Nos 1-4)

A terminus post quem of AD75 for the lowest fill of the drain C1282 is provided by South Gaulish samian plate stamped by the potter Calvus (see below Samian stamps No 1). Further evidence that activity commenced after AD 75 comes from decorated, South Gaulish bowls from C1251, C1290 and C1356 dated to c.AD 75-95.

The assemblage from this period is comprised of the following ceramic groups. They are consistent with a late first-century, early Flavian date, and could be contemporary with the foundation of the York fortress, broadly corresponding to Period 1 at Blake Street (Monaghan 1993). However, the assemblage recovered is much smaller - 68 sherds; 1424g; 2.07 EVE.

A small amount of earlier samian was found in the lowest fill of the drain, C1282, consisting of stamps dating to AD 50-65 and 65-90 (see below, Samian Stamps, Nos 5 and 6, respectively), suggesting that either these vessels were in use for a longer period or were part of earlier rubbish disposal.

Area A (A-ROM1): A small amount of pottery was recovered from a series of gullies C1279, C1278, C1275 and from pit F136, C1266. Area B (B-ROM1): The earliest deposits in area B produced no pottery and the earliest sherd comes from the track, C1359. Area C (C-ROM1): Produced a small group of pottery from miscellaneous layers containing gravel and charcoal C1360; C1361.

The coarse wares are consistent with the above daterange, again mainly legionary-type jars and bowls, together with ring-necked flagons in Ebor. There is a marked increase in rusticated jars, with thick rustication, rather than linear, which suggests a first-century date. However, these vessels continue to be in use into the early second-century. Several sherd links between C1282 and C1251 suggest some redeposition within the two groups.

External dating is provided by stamped wares: a mortarium stamped by the potter Cacumattus, AD 65-100 (see below, Mortaria - No 5, and Figure 3.10, No.4); and two South Gaulish vessels, a dish of Passienus, AD 5570, and a cup of ?Rufus III, of Flavian date, (see below, Samian Stamps Nos 8 and 14). The coarse wares also correspond to a Flavian date, consisting mainly of Ebor ware legionary-type jars and carinated bowls, together with a tazze burnt on the rim. There is a sherd link between this tazze and a deposit associated with the stone building (see Table A10) which suggests that the pottery was redistributed.

Evidence that the later fill of the drain, C1251, may have been backfilled as late as the early 2nd century is suggested by an unusual samian dish in an early micaceous Lezoux fabric, stamped on the interior wall, which is dated from the late first to the early second century (see below Samian Stamps, No 11 and Fig.49, No 25). A fragment of a volute lamp from Central Gaul, and an unusual flagon with a skeumorphic handle, copying bronze forms, (see Figure 3.11, No.27 and Figure 3.10, No.6, respectively) fall within the same date-range. There is also a lid in Ebor (see Figure 3.10, No.14), a form for which there is no convincing evidence before the early 2nd century (Monaghan 1997, 1018). However, evidence for deposition in the early second century, in this case, is limited.

Although imported finewares occur in this group, they consist solely of South Gaulish samian ware (including an inkwell). This group lacks the early imported colourcoated wares, such as Lyons ware, and Lead-glazed ware, noted at Blake Street. Native wares, also present at Blake Street are absent (Monaghan 1993, 736-8, Tables 122127). However, the lack of such wares at the Minster Library may be due to the very small size of the group.

Ceramic Group 3 c. AD 100-120 [CP 2a] (Figure 3.12, Nos 28-34)

Ceramic Group 2 c. AD 75-100/120 [CP 1a - 2a] (see Figure 3.10 and 3.11, Nos 5-27)

This group consists of a large assemblage from the destruction layer over the drain fill, C1244, which appears to be homogeneous in date - 152 sherds; 4023g; 3.49EVE. This appears to be the final detritus of the departing 9th Legion.

This assemblage is the largest from the site - 415 sherds; 10,208g; 16.74 EVE, mainly recovered from the lower fills of drain F134, and the ceramic groups, listed below.

Area A (A-ROM3): The drain fill was buried below layer C1244 which also produced large groups of pottery.

Area A (A-ROM2): A large group of pottery was recovered from the lowest fill of drain F134, C1282 and another large group from its secondary fill, C1251, subsequent to the construction of the stone wall, F166, C1263 (equivalent to F132). An isolated layer, C1296, which is broadly contemporary, produced a small quantity of pottery. Area B (B-ROM2): Pottery was recovered from the street, C1356, a spread, C1357, and a destruction layer, C1290. Area C (C-ROM2): A small amount of pottery came from

A dupondius coin of Domitian dated to AD 81-96 provides a terminus post quem for the spread over the drain fill C1244. However, this group contains pottery forms which are early second-century in date, for example a ring-necked flagon (see Figure 3.12, No.28) which has a slightly more pronounced top ring (FR2), perhaps an incipient example of the later type with a definite thick top ring (FR3). This group also includes a

37

Excavations at the Minster Library, York 1997

Figure 3.10 Examples from Ceramic Groups 1 and 2 (Scale 1:4)

38

The Roman Fortress

Figure 3.11 Examples from Ceramic Group 2 (Scale 1:4)

39

Excavations at the Minster Library, York 1997

Figure 3.12 Examples from Ceramic Groups 3 and 4 (Scale 1:4)

40

The Roman Fortress lid, and there are fewer legionary-type vessels, but this may be due to the lack of rim forms. Sherd links between C1244 and the fills of the drain, C1251 and C1282, suggest that some of the earlier material was redistributed. The group as a whole, although large, appears to be more fragmented than the drain fills. There is no evidence from this group of black-burnished wares or local copies influenced by these types, or Central Gaulish samian, which appear after c. AD 120.

and C1247, and a dump, C1246. As these layers overlie the dumps preceding the stone buildings dated to post c.AD120/140 they must be of this date or later, which corresponds to the date for the stone buildings at Blake Street - 140/160 - 400 (Monaghan 1993, 690). There are few diagnostic sherds, all of which come from C1193. These consist of a roughcast beaker imported from Cologne which dates from the early to mid-2nd-century; a plain, everted rim jar in Ebor and an everted-rimmed jar with a thickened, square rim in local grey ware which date from the first and second centuries but can be as late as the early third century (Monaghan 1997, 983 and 987). Another third-century vessel, the lower portion of a candlestick, in a probable Ebor fabric occurred in C1247, which Monaghan dates from the early third to the fourthcentury.

Ceramic Group 4 c. AD 100-140/60 [CP 2a- 2b] (see Figure 3.12, Nos 31 & 35, Figure 3.13, Nos 36-38) These groups probably correspond with the departure of the 9th Legion and the early occupation by the 6th Legion in c. AD 115/20-130/40. Area A (A-ROM4): This group is slightly larger (110 sherds; 3301g; 2.29EVE), but similar to the dumps from Area B, mentioned above, consisting of a series of spreads and dumps, C1216, C1222, C1225 and C1226. Legionary-type vessels in Ebor are rarer, and a flagon from C1226 has a more pronounced top ring, again possibly an incipient FR3, (see Figure 3.12, No.35), suggesting an early second-century date. There is no evidence for black-burnished wares from this group, but a rim sherd from a dish in Central Gaulish samian from the top layer, C1216, is dated from c. AD 120-140/60.

Area A (A-ROM6 - c. AD 180-250+): These features were cut through by features F104, C1182 and C1180; F102, C1177; and F122, C1224, and were capped by a make-up layer, C1152 and overlying surface, C1128. The assemblage is small consisting of 70 sherds; 1179g; 2.04EVE, but contains several examples of wares dating from the later second century in the mid-third or later. These include a plain flanged dish in local burnished ware (see Figure 3.13, No.44); an indented jar in the same fabric; an undecorated pie-dish with a rolled rim in a BB2 fabric; a local grey ware dish with a bifid rim (JQ3) and a funnel-necked indented beaker in Moselkeramik.

Area B (B-ROM3): As in Area A, pottery was recovered from the dumps which covered the entire trench C1352, C1341, C1342, and layer C1315.

Area B (B-ROM4 - c. AD 140 -200/225): A small quantity of pottery was associated with the construction of the stone buildings in Area B (F116, C1303; F117, C1304]; and F120, C1221) - 73 sherds; 1160g; 0.67EVE. Only C1221 produced material diagnostic of this date which included several cooking-pots with lattice decoration, Central Gaulish samian, and a butt-shaped jar in local grey ware dated from the mid-second to early third century (Monaghan 1997, 980). A plain dish with a curved rim (DD4) in Ebor from C1303 may also be as late as the early third century (Monaghan 1997, 1010). Other material included earlier pottery, including South Gaulish samian.

The dumps from Area B are smaller and more fragmented - 68 sherds; 750g; 0.95EVE. C1315, C1341 and C1342 produced pottery of late first to early secondcentury date, but the upper layer, C1352 produced the first evidence from the site of burnished wares. It is a sherd from a cooking pot with burnished lattice decoration, in local burnished ware (G3), which suggests a date post c.AD120. However, there is also a sherd from a decorated bowl in Central Gaulish samian dated from the Hadrianic to the early Antonine period, giving a date of c.AD120-160 for this context. Second to mid-fourth-century deposits

Ceramic Group 5 c. AD 140/60 - 300/400 [CP 2b - 3/4] (see Figure 3.13, Nos 39-45 and Figure 3.14, Nos 47-57)

Area B (B-ROM5B - c. AD 140 -200/225): Deposits associated with the stone buildings produced small quantities of pottery C1168, C1174, C1211, C1283, C1203, and C1214. The assemblage from this group is larger (119 sherds; 1779g; 2.33EVE) but very similar to that from B-ROM4 with virtually the same forms including form DD4, which is in Ebor fabric E5. Other forms include at least three tazze, one of which joins with Area A C1266, (above, Ceramic Group 1), and an unusual handled vessel, resembling a honey-pot, and a very coarse lid with a thickening below the rim (see Figure 3.13, Nos 39-43).

Area A (A-ROM5 - c. AD 120/140 - 300): Small quantities of pottery were found in deposits associated with the stone buildings in Area A - 35 sherds; 899g; 1.03EVE. The earliest contexts are layers, C1193, C1192

Area B (B-ROM6B - c. AD 170 -200/225): Later layers associated with the stone building C1188 and C1215 were too small to date securely (26 sherds; 274g; 0.56 EVE) but are broadly the same date as the earlier

There appears to be a series of discrete activities occurring during this period, at different times, therefore the following Ceramic Groups concentrate on specific areas. There is a marked change in the composition of the pottery from these groups in comparison with those of first-century date. Ebor is less common and greywares, especially burnished wares are more common.

41

Excavations at the Minster Library, York 1997

Figure 3.13 Examples from Ceramic Groups 4 and 5 (Scale 1:4)

42

The Roman Fortress

Figure 3.14 Examples from Ceramic Groups 5 and 6 (Scale 1:4)

43

Excavations at the Minster Library, York 1997 assemblage. One vessel, a jar or bowl with a rolled rim (see Figure 3.13, No.46), would certainly fit within a later second to early third-century date-range.

1.22EVE. A high proportion of the contexts contained early Roman pottery, including South Gaulish samian and Ebor wares. The earliest occurrence of later Roman pottery, an ‘African’ style platter (PA2) in Ebor fabric, E10 dates from the late second to the mid-third century and came from C1154, (see Figure 3.14, No.53). C1185 overlying C1154, produced a flanged dish with lattice decoration in a BB1-type fabric (B10) which may be in use until the third century (see Figure 3.14, No.52). Also of probable third-century date or later is an unusual, everted-rimmed jar with external ribbing just below the rim (see Figure 3.14, No.51) from C1213, towards the top of the sequence.

Area B (B-ROM7B- c. AD 250 - 300/350): These deposits were cut by gully F87 which produced two small assemblages, from C1157 and C1126 - 31 sherds; 304g; 0.37EVE. This group includes two ‘African’ style vessels, a plate and a much burnt example of a bowl with an exaggerated offset on the interior (see Figure 3.14, No.48) which is dated from the early third to the midthird century (Monaghan 1997, 1000). C1157 produced the only rim sherd of a true Dales ware jar, and a fragment of Nene Valley colour-coated beaker in an orange fabric which is generally dated from the mid-third but continues in use into the early fourth century.

Late fourth-century deposits Ceramic Group 6 c. AD 350-400 [CP 4b] (see Figure 3.14, Nos 54-57)

Area B (B-ROM5A - c. AD 170 -200/230): Street surfaces C1274, C1302 and C1230 all produced small pottery assemblages - 10 sherds; 267g; 0.24EVE. These layers are probably contemporary with B-ROM5B above, but the group is too small; the only diagnostically later sherds being an abraded fragment of Dressel 20 amphora in the later fabric which is current from c.AD 170-300.

Area A: A surface, C1118, a spread, C1119, a post-hole, F72, C1117 and backfills C1120, 1124 which either contained late Roman pottery or were stratigraphically later than deposits that did so (A-ROM8). This small assemblage (41 sherds; 783g; 0.95 EVE, produced the first examples of true ‘Huntcliff’ jars (two examples), together with a sherd of Crambeck late Parchment ware (see Figure 3.14, No.57). These wares are dated by Monaghan to c. AD 360 (1997, 985). A small amount of residual pottery, South Gaulish samian, and Ebor and North French mortaria, also occurred in this assemblage.

Area B (B-ROM6A - c. AD 250/300-400): The street was cut by two gullies, F113 (C1200 and C1199) and F101 (C1178 and C1172). This assemblage, although small 29 sherds; 706g; 0.60EVE, includes a fragment of a funnel-necked folded beaker in Nene Valley colourcoated ware with an orange fabric, and the first instance of calcite-gritted ware jar, which date from the mid-third into the fourth century, together with a single body sherd of probable Crambeck, late Parchment ware.

Area B: Various features and layers producing late Roman pottery were excavated in Area B. They are a scoop, F81 (C1111), layers C1110, C1171 and C1190 and F84 (C1140) (B-ROM9). This group produced a small assemblage - 19 sherds; 216g 0.36EVE. A denarius of Septimus Severus dating to AD 202-10 was found in C1110, and most of the pottery from this group dates from the later second to the mid-third century. The group is unlike the dark earth deposits in Area A in that no fourth century wares occurred in these layers.

Area B (B-ROM7A - c. AD 200-300+): F113 and F101 seem to have been followed by layers and dumps C1147, C1132, and C1146, although the matrix does not show a direct relationship. This is borne out to a certain extent by the pottery which is mostly of Antonine date. The profile of an undecorated ‘dog-dish’ with a slight groove towards the top from C1147, the lowest layer , is certainly of third-century date and may have continued in use into the fourth (see Figure 3.14, No.47). The group, however is very small - 15 sherds; 234g; 0.36EVE.

Area C: Pottery was also recovered from robber trench F71 (C1135), but only consisted of three sherds of Central Gaulish samian.

Area A (A-ROM7): Finally, pottery was found in the backfill of robber trench F77, C1125, and the latest layer associated with the stone building in Area B C1151 (BROM78). Both contexts produced pottery that is typical of late second to early third-century date. This date is consistent with the chronological development in Area A, but the equivalent layers in Area B date from at least the mid-third to the fourth century (see above B-ROM6A).

Ceramic Group 7 c. AD 350-400 [CP 4b] Area A: All three surfaces in Area A were sealed by ‘dark earth’, contexts C1017, C1081, C1090 (A-ROM9). The assemblage was of moderate size - 193 sherds; 2253g; 3.76EVE, and consisted of fresh, relatively large, unabraded sherds, suggesting that they were primary deposits. A terminus post quem of AD 364-75 is provided by a coin of Valentinian I for the lowest of the dark earth contexts, C1090. The pottery from this context included several Nene Valley colour-coated beakers and vessels in Oxfordshire red colour-coated ware as well as at least six ‘Huntcliff’ jars (JH3), together with three examples of late handmade ware (B18 and G18). The latter may have been produced into the early fifth century. C1081 is dated to at least AD 388 based on the presence of a coin of

Area C (c.AD 140-250+): Various deposits within the stone buildings produced small pottery assemblages. These are mainly layers (C1346, C1335, C1334, C1164, C1165, C1228, C1219, C1217, C1154, C1213, C1292 and C1185) (C-ROM3) and a sherd associated with the first of three hearths, F70, C1187 (C-ROM4). The whole assemblage amounts to 113 sherds; 1430g, but only

44

The Roman Fortress Theodosius I. The pottery consisted of virtually the same late forms as C1090, but lacked the late handmade wares. The assemblage from the uppermost layer in this sequence, C1017, also included Nene Valley colourcoated beakers, but only one ‘Huntcliff’ jar and no late handmade ware. However, the only example of a late Crambeck, parchment ware mortarium with painted decoration and a double flange (MD) occurred in this layer.

necked flagons with a prominent upper ring (FR3). Only one example of the latter type occurred on the site and came from a post-Roman layer. There is a single rim from a large flagon or possibly a small amphora from Group 2 (see Figure 3.10, No.7). Most unusual is a flagon with a skeuomorphic knob where the strap handle is attached to the rim from the same group (Figure 3.10, No.6). The rim appears to have been squashed, and may be a ‘second’. The vessel is very thin-walled and is decorated with burnishing inside the rim, vertical burnishing on the neck, and horizontal burnishing on the body wall. There is no exact parallel within the York typology although there is a similar vessel, also with a twisted rim, from 1-9, Micklegate (Queen’s Hotel) 1988.17 (see Monaghan 1997, 997 (FE1). However, the flagon is almost identical to mica-dusted examples from London which are based on bronze flagon forms (Marsh type 3 - Marsh 1978).

The Wares The format for this discussion follows that of the Roman Pottery from the Fortress (Monaghan 1993) in order to make direct comparisons of the data. Tables 1-5 give a breakdown of the Roman pottery assemblages by ware and context group by sherd count; sherd percentage; EVE’s; weight in grams; and by weight percentages, respectively. As Monaghan has given a comprehensive account of the history and function of the wares, it was not felt necessary to reiterate that information here. Therefore this discussion concentrates on the incidence and dating of the fabrics and forms, together with any new evidence derived from the assemblages from the Minster Library site.

Jars are more common than bowls, including the legionary types (JA- jar and BC- bowl). In this respect, the assemblage differs from that at 9 Blake Street where the opposite occurs (Monaghan 1993, 706). However, this is based on the number of sherds rather than EVE’s due to the lack of sufficient data from the site. Other common jar forms are butt-shaped vessels (JB) and everted-rim types (JE), but these are post-legionary forms and do not appear until Group 5. A rare type is the rusticated jar which is usually found in Rustic 1 and 2 fabrics.

Eboracum ware Eboracum ware was the principal ware for kitchen and kitchen-to-table wares throughout the first and to a large extent, the second and third centuries. It is present in some quantity in the earliest levels but reaches a peak during the Flavian-Trajanic period. There is a slight decline in the early first century, but is followed by another peak after c. AD 140/60 (Group 5). However, there is a distinct change between the forms of the early period of manufacture and those of the later, although there is clearly a development, rather than a complete alteration. For example, the carinated bowl (BC) developed into a less defined type (BD) which has been previously discussed by Monaghan (1997). This is parallelled within the local grey ware assemblage (see below, grey). It should be noted that a high proportion is found in post-Roman levels denoting the redistribution of earlier wares up through later stratigraphy (see Tables A1-5). The mortaria are discussed below (see, Mortaria).

Bowls are almost completely represented by the early legionary type (BC), and to a lesser extent the postlegionary type (BD). One exception is an ‘African’ style bowl in the coarse white-slipped variant (W2). It has been so severely burnt that the slip is barely visible (see Figure 3.14, No.48). There is a single example of an early platter (PD2 – see Figure 3.10, No.13) and at least two of the later type (PA- see Figure 3.14, No.53). Two rarer types of platter were found in Group 2 dated from the late first to the early second century. A platter with a groove just below the beaded rim, which is finely made perhaps with a jigger and is very similar to samian form Dragendorff 18 (see Figure 3.10, No.11). The other, with a slightly everted rim is in a fine variant (E3 - see Figure 3.10, No.12). There is no exact parallel although it is similar to 4082, which is also found in E3 fabric (Monaghan 1997, 1017-8).

Monaghan (1997) has isolated several different fabrics within the Ebor range, and all variants, albeit in small amounts, occurred within the assemblage. Iron-rich (E4) and chalky variants (E5) were present before the 2nd century. Burnished (E8 and E10) and red-slipped variants (E9 and E11), together with white-slipped fabrics only occurred in and after Group 5, the stone buildings, dated to c. AD 140/160 -300/400 (Table A11).

Beakers are uncommon and, apart from undiagnostic body sherds, were limited to three types: a simple beadrim vessel with a high shoulder (see Figure 3.10, No.1); a cornice rimmed example in the red-slipped fabric (see Figure 3.14, No.54); and a fine, thin-walled vessel which may be a biconical beaker, although only the rim and neck survive (see Figure 3.12, No.29).

Flagon types just exceed jars in popularity (Table A12). These are predominantly ring-necked flagons with an everted neck (FR2) although the group includes two examples where the top ring is less distinct (see Figure 3.12, Nos 28 & 35). They occur in early second-century groups and may be incipient examples of the later ring-

Tazze are present in small numbers, mainly the early type (TA- see Figure 3.10, No.3 and Figure 3.13, No.39). Rare forms consist of lids (see Figure 3.11, No.22); a single, inturned, globular cup of Italian origin (see Figure 3.12, No.30); two candlesticks, both from contexts dated to the

45

Excavations at the Minster Library, York 1997 later Roman period, and a single unguentarium.

Street (ibid, fig 288, no 2828).

Finally, there are two rare but distinctive vessels for which no parallels have been found. A funnel-shaped object from Group 2 was wheel-thrown, and removed from the wheel by a wire-cutter. The function is unknown, but it appears to have served as a means of dispersing heat and resembles some examples of kilnfurniture (see Figure 3.11, No.15). The other is an object, circular in section, and flattened on one side with a nodular protrusion extending laterally, and came from a late Roman group (BROM9). The flattened base and probable burning suggest that it may have been used as a trivet (see Figure 3.14, No.55).

Table A13 shows the incidence of the type of rustication by period. The thicker, web-like application (RWEB), is most common and is concentrated in late first to early second-century groups. A thick rustication, but not as distinct as the former type, is also present during this period (RUST). Linear rustication (RLIN) does not appear until the early second century, but this was on a single thin-walled vessel. Occasionally this type of rustication occurs on jars in Ebor, and there are three examples (possibly one pot, but the sherds do not join) from Group 2. Black-burnished wares

Mica-dusted ware These wares are rare in comparison to the local equivalent industry (see below, grey-burnished wares). They are not present on the site until the Hadrianic - early Antonine period, but there is insufficient evidence to give a date for the end of importation. Thames estuary BB2 is slightly more common than BB1 from Dorset, but does not amount to more than 33 sherds from the whole site. In addition, there are a few sherds of BB1-type fabrics, and of a distinctively micaceous BB2-type ware.

A single sherd from a mica-dusted beaker in an Ebor fabric (D5) which occurred in context with pottery dating from dating from the later first to the early second century. This corresponds with the dating for these wares from Blake Street (Monaghan 1993, 711). Native wares Three sherds fall within this category, two joining sherds came from the spread over the drain C1244, dated to the early 2nd century. The fragments are from a closed form, sooted on the exterior, in a coarse, handmade fabric. It is fairly high-fired, although this may be the result of cooking processes, and light, reddish-brown in colour, with coarse inclusions of grog and clear and milky subangular quartz, together with occasional calcareous fragments, set in a fine matrix (N1). An Easingwold source is suggested by the inclusions.

As the groups from the Minster Library are much smaller than those from Blake Street, EVE measurement cannot be used to determine the ratio of jars to bowls. Monaghan comments that within the BB2 category at Blake Street the ratio is 4:1 in favour of bowl/dishes (Monaghan 1993, 714). At the Minster Library there are certainly a higher number of bowl/dish sherds than jar forms. Within the BB1 group there is almost twice the number of jar sherds than bowl-dish fragments. This is also the case within the BB1-type category.

The other sherd, a rim fragment of a jar or bowl in a shell- and sand-tempered fabric (N5), came from a late first to early second-century deposit. This vessel has an inturned, undercut rim of ‘native’ tradition, and was handmade (see Figure 3.11, No.23). The rim is virtually identical to 2859 from 9 Blake Street, Period 2, c.AD 71/9-100, but the fabric is described by Monaghan as possibly South Yorkshire grey ware from Cantley (Monaghan 1993, 779-780).

Decoration on the BB1and BB2 jars is limited to acute lattice, but there is more variety in the BB1-type fabric with acute lattice, multiple lattice and diagonal lattice occurring. Bowls/dishes in BB1 have acute lattice, and pointed intersecting arcs as decoration, but although occasional acute lattice is used on the BB2 bowls, the majority are undecorated, indicative of a late 2nd to early 3rd-century date. A vessel in BB1-type has the acute lattice decoration but the rim is atypical, flaring upwards (see Figure 3.14, No.52). Two vessels have been illustrated from the distinctively micaceous category (B5), a grooved dish (see Figure 3.14, No.47), and a cooking pot (Figure 3.14, No.49).

Rustic ware Parallelled at Blake Street, rustic ware is the principal coarseware after Eboracum ware, (Monaghan 1993, 7123). It is rare in the earliest group, but increases in Group 2 dated to the late first to the early second century. Thereafter the fabric declines sharply, which confirms Monaghan’s comments that this ware was linked to military occupation. Only nine sherds of the early, softer, micaceous fabric, all with thick rustication (R2), were found and these occurred in Group 2; the rest being in the higher-fired, sandier fabric, R2. A high proportion of the vessels were sooted on the exterior and, less frequently, on the rim. There is a variety of forms including one with a slight lid-seating (see Figure 3.11, No.18). Other forms are more consistent with the typical jar noted at Blake

Grey-burnished wares Local grey-burnished wares (B3) account for the bulk of this category - 169 sherds, with a further 31 sherds in the darker, probable East Yorkshire fabrics (B14-17). They are absent from groups before c.AD 120/140, the exception being a single example of a legionary type jar (JA) with a small band of burnishing which would have been perhaps better defined as local grey with burnishing. However, this feature was also noted at Blake Street (Monaghan 1993, 714). Apart from this vessel the earliest

46

The Roman Fortress occurrence is in C1352, a dump, which lies immediately below the stone buildings (Table A14). Thereafter, they are most common in groups dating from c.AD 140/160300/400, but declining towards the beginning of the third century. There is a small presence in fourth-century groups where they are most likely to be residual. This is borne out to some extent by the large presence in postRoman layers which contained a high proportion of early and mid-Roman wares.

Later post-legionary forms include examples of a developed form (BD) of the earlier, carinated bowl and several jars with everted-rims (JE), as well as butt-shaped jars (JB), together with everted jars with thickened rims (JQ). Monaghan dates these forms to CP2a-3a (1997, 980-1000). In addition, there are a number of more unusual vessels which occurred within Ceramic Group 5, dated from the mid-second to the third/fourth century. For example, a lid with thickening below the rim (LE) in a coarser fabric more akin to the coarse Ebor fabric, E2, (see Figure 3.13, No.42), a narrow-necked jar (see Figure 3.13, No.40), and a handled jar with a moulded rim, which resembles a honey-pot (see Figure 3.13, No.41).

The East Yorkshire wares appear to have a later bias, with the largest amount in post-Roman layers, but the group is too small to be meaningful. Cooking pots are the dominant form, but only as body fragments, one of which has a cavetto rim with obtuse lattice decoration. As the sherds are relatively small, those with decoration are minimal. Where present, lattice decoration is most common, together with a single example with multiple line lattice. Open vessels are rare with a single occurrence of a flanged bowl and a small quantity of dishes (5 vessels), one of which is a ‘dog-dish’ type.

Other grey wares are rare, consisting of a few sherds in a coarse, sandy fabric (G8), probable East Yorkshire products, both of which appear to have a very late, but uncertain bias. Micaceous sandy grey wares (G6) are equally uncommon and appear to be post-legionary in date. Two vessels have been illustrated in this fabric, a rolled-rimmed jar (see Figure 3.13, No.46) and the base of a later form of tazze (see Figure 3.14, No.41). The latter is highly micaceous with mica accretions around the neck of the base.

There is a similarly high ratio of cooking pots to dishes within the B3 assemblage, again predominantly body sherds. At Blake Street, the ratio was 2:1 in favour of jars, but there are insufficient EVEs from the Minster Library to make a meaningful comparison (Monaghan 1993, 715). The decoration on the jars is mainly acute lattice, together with a slightly higher incidence of multiple lattice decoration. Other closed vessels are rare, but there is a single example of an indented jar (JI) and a BB1-style tankard (KT1), both from a group dating from the late second to, at least, the mid-third century.

Fine grey wares There are only seven sherds in this category, five of which are in a fine, micaceous grey fabric (F2) - body sherds of jar or beaker form. Another probable beaker is in a miscellaneous fine fabric (F1). No decoration occurred on any of the above, but a sherd from a jar also in F1, has burnished horizontal lines on the neck, similar to that on imitation North Gaulish jars. These fabrics first occur in groups dated to the late first to early second century, and are absent by the late second century.

Open forms amount to a maximum of 13 vessels including ‘pie-dishes’, both decorated and undecorated (DP1; DP5; and DP7), a plain grooved dish, and a flanged dish. The latter has bands of horizontal decoration from the same group as the tankard, mentioned above (see Figure 3.13, No.44).

‘White’ wares Pale or white fabrics (P0), other than Crambeck late Parchment ware, are rare, amounting to a small total of 17 sherds. One of these is a flagon fragment in Verulamium region white ware, from the spread over the drain fills (C1244) dated from c.AD 100-120. Ten sherds, also flagon fragments, are in a fabric very similar to that of the mortaria from the Oise/Somme area of North Gaul. They occurred in Flavian-Trajanic deposits, a date-range which fits within the production period for the mortaria. The remaining sherds are unsourced, although at least two sherds may be pale variants of the chalky Ebor fabric (E5).

Grey wares The bulk of the grey ware fabrics is composed of what appears to be the reduced equivalent of the oxidised Ebor fabric (G1). Table A15 suggests that it is uncommon in the earliest groups, but is established by the late first to the early second century. However, there is an unusually high proportion in contexts dating from the mid-second century (Group 5 - stone buildings). It is likely, as with Ebor ware above, that the same industry continued, or resumed supply to the fortress after the establishment of the 6th Legion. There is certainly an early production as the majority of the forms from groups which pre-date c. AD 140/60 are virtually identical to those in Ebor fabrics, for example carinated bowls (see Figure 3.11, No.16 and Figure 3.12, No.33) and legionary-type jars (see Figure 3.11, No.22 and Figure 3.13, No.37). In addition, there is a shallow dish with a slight bead and high flange of late first to early second-century date (see Figure 3.11, No.17) and a lid (see Figure 3.12, No.32) which fits bowl No.33.

North Gaulish grey ware Unlike the assemblage from Blake Street (Monaghan 1993, 717), there were no examples in the standard North Gaulish fabric (G9), and only one sherd in the imitation fabric (G10). This was a fragment of a jar with distinctive burnished horizontal lines, probably a Vase Triconique, which occurred in a post-Roman layer.

47

Excavations at the Minster Library, York 1997 Colour-coated wares

appearance and clearly smashed, suggesting that they were primary rubbish.

Colour-coated wares other than Nene Valley products are extremely rare consisting of only 14 sherds (Table A17). Moselkeramik accounts for five of these, together with products from the Oxfordshire kilns (3 sherds), two roughcast, cornice-rimmed beakers sherds from a Cologne source, and a similar vessel in Ebor fabric C18. Three sherds were unsourced. Apart from the Oxfordshire vessels, a bowl fragment with a demi-rosette stamp and a jar, the remaining fragments were all beaker forms including a funnel-necked, folded beaker with a slightly flaring rim in Moselkeramik.

Crambeck grey ware This ware is much rarer than at Blake Street (Monaghan 1993, 719), consisting of only 13 sherds (Table A16). The majority of these were closed forms and there were no rim sherds. One was certainly a bowl or dish and one with burnished vertical decoration probably came from a jar with loop handles (JL1). The earliest occurrence of this came from features associated with the street dated from c. AD 250/300-400. Three fragments came from post-Roman layers and the remainder occurred in the dark earth deposits.

The earliest occurrence, a Cologne roughcast beaker, was in a group dated from c.AD 120/140-300, and a Moselkeramik beaker occurred for the first time in a group dated from c.AD 180 - 250+. Both of the earliest postulated dates fit well with the production dates of the wares concerned. Stamped Oxfordshire vessels are generally dated to the late fourth century and the presence of the vessel in dark earth layers, dated to the mid- to late fourth century, fits well with the dating for these forms. Other examples occurred in post-Roman layers.

Crambeck late parchment ware This fabric is rare, consisting of nine body sherds of uncertain form but mainly closed forms, jars or flagons. The majority came from post-Roman layers, but two came from dark earth layers dated from mid- to late fourth century and one from a spread immediately below the dark earth of similar date. A less certainly identified sherd came from features associated with the street, dated from c.AD 250/300-400.

Lower Nene Valley ware This ware group is the most common of the colour-coated wares, amounting to 54 sherds, in the three main fabrics (see Table A17). The most common is a white fabric (C1), followed by sherds with orange cores (C3), and two sherds with a grey core (C2). The majority occurred in dark earth and post-Roman layers, but the earliest examples came from layers dated from the mid- to late third century into the fourth. Most of the vessels were small, undiagnostic fragments of beakers, some with barbotine, rouletted, or painted decoration. The majority of the identifiable beakers were funnel-necked folded beakers, including two with a plain rim and applied scale decoration (KF1). There was also a single bag-shaped beaker (KB) and two globular vessels with an elongated neck (KE). Other forms include an everted-rimmed jar and a castor-box and two lids.

Late handmade ware Five sherds in the burnished variant of this ware (B18) came from the site. Most occurred only as the body sherds of jars (some larger vessels) together with a rim from an everted-rimmed cooking pot. There were also three fragments, probably jars, in the unburnished type (G18). Five sherds occurred in post-Roman layers and the remaining sherds came from a dark earth deposit dated by a coin of Valentinian I to c.AD 364-75, but this ware may have continued into the fifth century (see Table A14). Pompeian-red ware This fabric is poorly represented. One sherd in Pompeian fabric 3, from a Central Gaulish source, occurred in a group dated to from the late first to the early second century, within the main production period for this ware. Two further sherds, in Pompeian fabric 6 or 7 (Peacock 1977), came from post-Roman layers. These rare fabrics are noted as being prominent in York, (Monaghan 1993, 720) but are described as being highly micaceous. The two sherds have mica in the surfaces but it is fairly sparse, otherwise they fit the description for fabric 7. The forms fit within the typology for these fabrics, one being a sherd of a dish identical to 2901 from Blake Street (ibid. 784), the other a base from a dish, Peacock form 14 (Peacock 1977, Fig 3, 157).

Calcite-gritted ware Tables A1 and A3 show that this ware was represented by 43 sherds with several rims amounting to 1.64 EVE. Nine sherds could not be identified to an industry, but were probably related to the ‘Huntcliff’ type. Two sherds were securely identified as Knapton fabrics, one of which was a rim fragment from a storage jar. One came from a postRoman layer, the other from dark earth deposits. The remainder were all fragments or rims of ‘Huntcliff’ jars (JH3); an unusual variant being one with a very flattened shoulder (see Figure 3.14, No.57). The earliest occurrence came from features cut into the street (BROM6A dated from c.AD 250/300-400, but this was an unsourced fabric. A coin of Valentinian I, c.AD 364-75, was found together with the first occurrence of a JH3. The majority, almost exclusively JH3 jars were found in the dark earth layers. These vessels were fresh in

Dales ware Only three sherds occurred from the entire assemblage, a body sherd from a layer dated to the late fourth century, and a rim and body sherd from a group dated from c.AD 250-300/350. This fabric was also rare at Blake Street

48

The Roman Fortress (Monaghan 1993, 721) reflecting a lack of third-century wares in general.

Much Hadham ware which occurred in a post-Roman layer. The remaining sherds were undiagnostic and mostly came from post-Roman layers.

Amphorae Samian ware In common with most large sites in Britain, Dressel 20 amphorae form the largest group of the amphorae assemblage (Table A18). These are composed of the earlier, gritty fabric current from the Conquest to the midto late second century in Britain. A small number were stratified within groups dating from c. AD 100 - 140/60, whilst others came from post-Roman layers. An unusual example of this type is a large fragment which has several impressions of a broken stamp on the interior. This vessel was manufactured by pushing out the internal wall with the handle from the same type of vessel. The handle was stamped with a broken die which is imprinted in reverse on the inside of this vessel resulting in four impressions (see Figure 3.13, No.38).

Samian ware is the third largest ware-type after Ebor and grey wares by sherd count. A striking aspect of the samian assemblage is the extremely low amount of Central Gaulish wares manufactured at Les Martres de Veyre (8 sherds) and those from East Gaulish sources (8 sherds) (Table A19). This suggests that there was something unusual about the supply, or requisition for supplies, of these wares. For example, the main period of production for Les Martres de Veyre products is from c.AD 100-120, which approximates to the departure of the 9th legion from York in c. AD 115-120. Furthermore, there is little evidence of late decorated samian from Central Gaulish groups from the Lezoux kilns (pers. comm. Felicity Wild). In contrast, there are a number of examples dating from the Hadrianic to the early Antonine period, which corresponds to the arrival of the main contingent of the 6th Legion. It does not appear on the site in any quantity until after the construction of the stone building (Ceramic Group 5 - c. AD 140/160300/400). East Gaulish samian, although produced from the mid-second century continues to be manufactured into the third. The evidence for third-century pottery from the assemblage as a whole is rare in comparison with early to mid-Roman wares.

The later Dressel 20 fabric, which is generally finer often with a white exterior surface, is generally dated to post AD 170-300 in Britain. This fabric occurred within groups dated to the later Roman period, together with a number from post-Roman layers. Gallic wine amphorae were the next most common type and were present throughout the sequence, again with a higher proportion in post-Roman layers. Other types were rare, but were mainly found within the earlier groups, and include Italian Dressel 2-4 wine amphorae and those carrying types of fish sauce (Camulodunum types 186AC). The ‘carrot’ (a feature of early military sites), ‘Rhodian’, and North African types present at Blake Street were absent from the Minster Library. There is only a single rim from the whole site and the apparent paucity of sherds from these mainly large vessels suggests that they may have been stored in a separate area or that there was a discrete disposal system in operation.

By far the most common component of the samian assemblage is South Gaulish wares, mainly manufactured at La Graufesenque. It is present in very small quantities in the earliest group, but is most common in Flavian/Trajanic levels, after which there is a sharp decline. A high proportion is found after the construction of the stone buildings which suggests that some vessels may have continued in use long after the period of production. There is also a high degree of residuality borne out by the equally high amount found in postRoman layers.

Given the small size of the assemblage from this site in comparison to that from Blake Street the amount of amphorae sherds per 100 other sherds is remarkably similar (see Table A18; Monaghan 1993, Table 117, 721). The exceptions are the earliest period, where there is a larger amount at Blake Street, 4.3 as opposed to 1.9 at the Minster Library, and the post-Roman levels, 1.1 at Blake Street and 5.7 at the Minster Library. There is a high proportion of early Roman wares in the post-Roman layers which suggests that early material was redistributed up through the site.

Despite the small amount of this ware in the earliest levels there is some evidence for wares which pre-date the arrival of the 9th Legion in c. AD 71. Three of the stamped vessels (see below, Samian stamps) have early dates: c. AD 50-65 (Stamp 5); 55-70 (Stamp 8); 60-75 (Stamp 3). In addition, there are ten examples of samian form Dr 29 which generally date from the Neronian to the early Flavian period, although these could have continued to be used later.

Oxidised wares Another distinctive aspect of this assemblage is the uniformity of forms and sizes in the plain wares, which is indicative of some form of control within the supply mechanism. These consist mainly of South Gaulish samian dishes (41 records - 4 for the rouletted version 18R) with the majority having diameters of 16cms, and a smaller proportion ranging from 20-24 cm. There are 28 records of cup form Dr 27, mainly with diameters of 8 or 9cm and very occasionally 12cm. Other forms are rare,

It is worth noting that oxidised fabrics other than Ebor ware are extremely rare (5 sherds and 3 flakes) suggesting that vessels in this type of fabric were almost entirely manufactured from local clays. One of these is a fragment of the stand from a candlestick in a high-fired fabric with a dark grey core, which is similar to some of the standard, Ebor mortarium fabric. The only example of a sourced fabric is a single sherd of a closed form, in

49

Excavations at the Minster Library, York 1997 the most common of these being the dish form Dr 36 with barbotine decoration, a flanged dish (Rt 12) and a single inkwell.

attempt to date any of its many variations, so that the overall date for the form remains at AD 80-150. For further details about it see Holbrook and Bidwell 1991, 198-203.

Within the Central Gaulish group dishes of Dr 31 or 31R are most common (18 records), together with 8 records of form Dr 18/31, whereas cup form Dr 33 accounts for 11 records. Unfortunately there is insufficient rim survival to determine the range of diameters for this group.

2. C1062 (A-PRO2, post-Roman) (106g Diam. 27cm. 14 per cent). A very hard, fine-textured cream fabric with deep cream slip; moderate, ill-sorted inclusions, including quartz, orange-brown and rare black slag-like material. The small amount of surviving trituration grit is black with rare orange-brown grits.

The majority of the decorated wares for both industries consist of samian bowl Dr 37, together with a few earlier Dr 29 and Dr 30 and two examples of decorated beakers Dr 67. They are less common than the plain wares, but are obviously a feature of barrack quarters. In addition, there are two examples of samian mortaria, Dr 45, in Central Gaulish ware.

The incompletely impressed potter’s stamp shows ]AV[ and part of the initial M; complete impressions from the die give MAVRI with criss-cross borders. Stamps from several dies probably have this reading and none give the name in full. Mortaria with stamps from three other dies which give MAVRIIM differ sufficiently to indicate that they may be the work of a different potter. MAVRI is either the genitive of Maurus or an abbreviated form of Maurius. Maurus is recorded as a name (Kajanto 1979).

Particular care was taken to note whether the samian vessels had been used or were fresh and unused. There is only one example from the whole site which appears not to have been used, unlike the assemblage from Blake Street which produced a high proportion of unused vessels (Monaghan 1993). It is interesting to note that there are four examples of graffito from the site, two on South Gaulish cups and on one dish, and one on a Central Gaulish cup (see Figure 3.11 Nos 24-5 and Figure 3.12 No.31). The graffito is crude and scratched on the exterior of the vessels towards the base, possibly indicating that they were stored upside-down. They have not been identified, but are most likely to have been made by the legionaries using the contubernia.

He worked in the Mancetter-Hartshill potteries and his mortaria at Mancetter were associated with those of Sennius (Hartley 1959), seventeen of whose mortaria were found in a gutter deposit at Wroxeter, in a similar context to an even larger deposit of samian ware (Atkinson 1942, 127-129 and 279-280). The samian is dated AD 160-180 and the latest associated coins (worn) are dated AD 155. A stamp of Maurus has been found at Bainbridge, which is believed to have been unoccupied in the period AD 120-160 and none of his work is recorded from Scotland. His rim-profiles suggest that he belonged to the latest generation of potters to stamp their mortaria and the optimum date for his work is AD 150-170+. Incidentally, another mortarium stamped with the same die of Maurus is recorded from York (Monaghan 1997, 932, no.3371) (Figure 3.15, No.59).

Lastly, the site produced a complete barbotine decorated dish (Dr 36) in an orange-red, highly micaceous Lezoux fabric (S3A). This fabric is extremely rare and the vessel had obviously been in use for a long period, judging by the degree of abrasion around the rim and base. Most unusual was the fact that it was stamped on the interior wall rather than the base (see Stamp 11, Figure 3.11, No.26, p39).

3. C1009 (A-PRO3, post-Roman) (88g). An unusual body and base sherd with concentric grooves on the base and neat external ridges and grooves around the upper part of the body. Smooth and hard, very finetextured, cream fabric with few, random, small but illsorted quartz with fewer orange-brown and black inclusions; self-coloured. The tiny, deeply embedded trituration grit was mainly quartz. Heavily worn.

Mortaria – Kay Hartley Stamped and unusual mortaria were examined by Kay Hartley whose report appears below. The following discussion consists of her comments on the vessels that she has seen, together with additions concerning those that were more readily identifiable. (Details of context, phasing and the ceramic date of context are included with any notes published. Fabric examined with hand lens at x20 magnification. NB ‘right facing’ and ‘left facing’ when applied to stamps indicates the relation of the stamp to the spout looking at the mortarium from the outside).

The treatment of the side and base alone would have made this an unusual mortarium. It is almost certainly an import. Unfortunately, it cannot be sourced with certainty from macroscopic examination without more surviving, but the fabric and trituration grit would best fit with production in the potteries of the Atisii at Aoste in Isère. If this is the correct source, the piece would belong to the period AD 50-85, otherwise it is not datable (see Figure 3.15, No.60)

1. C1134 (A-PR01, post-Roman) (110g Diam. 31cm. 8 per cent). Technically Bushe-Fox 22-30 (Fig 19) and fairly similar to Holbrook and Bidwell 1991, fig. 83, no.C40, which was in a Hadrianic-Antonine context. Import from Oise/Somme area of northern France (Hartley 1998, 205-206). Type Bushe-Fox 22-30 was never stamped and there is insufficient evidence to

4. C1005 (A-PRO4, post-Roman) (258g). An incomplete rim-section of a mortarium in self-coloured, fine- but powdery-textured cream fabric; moderate, brown ?slag and fewer quartz inclusions. The fabric is at

50

The Roman Fortress

Figure 3.15 Examples of ceramics from post-Roman period (Scale 1:4) least partially covered with a deposit due to some specific kind of weathering. The whole of the interior has been covered with fine concentric scoring combined with tiny and small flint and quartz trituration grit. Some wear.

number of potters, using basically similar fabrics, who made form Gillam 238, though the rims of some of his mortaria are more curved than is typical for the type. All can be attributed to potteries in the Oise/Somme area of northern France (Hartley 1998, 201-205). There is abundant evidence to date Gillam form 238 mortaria within the period AD 65-100 (see Figure 3.10, No.4).

This is an import from northern France. The precise rimprofile is uncertain, but it is most likely to be a Gillam 238 from the Oise/Somme area, AD 65-100. There is an outside, rather unlikely possibility that it could be a Hartley Group 1 (Hartley 1998, 206-209), but this would only mean that it could be up to ten years earlier in date and that the precise area of production in northern France is unknown. 5. C1279 (A-ROM1, c. AD 65-100 on mortarium) (145g). A self-coloured, very fine-textured, pale brownish-cream fabric, with few ill-sorted quartz and orange-brown inclusions. A small part of the internal surface survives showing concentric scoring combined with tiny to small quartz and flint trituration grit.

6. C1251, (A-ROM2, late first to early second century) (275g Diam. 29.5cm. 16.5 per cent). Eight joining sherds from a mortarium in very hard, red-brown fabric, fired to orange-brown near the surface and with blackish core; matt red-brown slip. There are fairly frequent, ill-sorted inclusions, mostly quartz with rare black and orange-brown material. The tiny to small trituration grit is composed of transparent and milky quartz and is distributed up to the bead with some scattered over most of the flange; it is combined with quite deep concentric scoring on the interior and with fine scoring on the flange. There is little indication of wear.

The slightly worn stamp is from a die of Cacumattus; complete impressions read OF/CACVMATTI, for ‘officina of Cacumattus’. The use of the term ‘officina’ is exceedingly rare in mortarium stamps; Cacumattus is one of only two, perhaps three, makers of mortaria who used it. He is also perhaps unique, outside Italy, in having a journeyman potter, as it were, actually mentioned in the stamps from another die; these read VASSONVS FECIT/OF CACVMATTI. Cacumattus was one of a

This general type of fabric combined with the treatment of the interior and flange is typical of mortaria found on military sites in the north and in South Wales. It is generally equivalent to form Gillam 237, which is AD 60-90 in date; the profile of this example is, however, essentially different. This combination of fabric and scoring was not continued after the early 2nd century so that it could not be later than c.AD 120 and it may well be Flavian. The fabric and treatment of this example

51

Excavations at the Minster Library, York 1997 points to manufacture at York in a military pottery (see Figure 3.11, No.15A).

blackish core and matt brown slip. The frequent, mostly tiny with few largish inclusions, consist mostly of quartz with a little red-brown and black material. A very few quartz trituration grits survive.

7. C1152, (A-ROM6, 180-250+) (18g). A heavily worn body sherd in very hard cream fabric fired to pinkish cream at the surfaces with a buff-cream slip. The ill-sorted and random inclusions include black and brown material as well as quartz. There is some indication that the trituration grit included quartz. This fragment is too small and worn for certain identification, but it could perhaps fit with production in the Rhineland.

The wide, shallow, relatively thin flange might be loosely regarded as Gillam 238 in form, but the G.238s imported on a large scale from the Oise/Somme area of northern France are always in cream to very pale brown fabrics combined with flint and quartz trituration. The fabric of C1221 is totally different, in fact resembling that of 1104 and likely to be from the same pottery in York. Gillam 238s in Romano-British fabrics are rare, but they were produced at Exeter in a workshop which could have been active in the immediate post fortress period, c.AD75-80 and perhaps run by veteran legionary potters (Holbrook and Bidwell 1991, 19 and fig. 86, especially nos.B12-B14 and B20, but nos. B8-B11 have some general similarity). The optimum date for C1221 is Flavian, certainly not later than early second century.

8. C1117 (A-ROM8, mid- to late fourth century) (12g). A flange fragment in fabric identical to 1251 and attributable to the same source. There is no indication on the fragment that this mortarium had grit or scoring on the flange. Military pottery, York. Flavian or Trajanic. 9. C1118 (A-ROM8, mid- to late fourth century) (94g Diam. 30cm. 9 per cent). A self-coloured, finetextured cream fabric with pink core in the rim; slightly powdery to the touch. Fairly frequent but very tiny inclusions, mostly quartz, with some orange-brown and rare black material. The internal surface is too damaged for description. This is a Gillam form 255 imported from the Oise/Somme area of northern France in the Antonine period, c.AD 140-180 (Hartley 1998, fig. 3, inset b, 204206).

13. C1221B, (B-ROM4, 100-140/60) (148g 28cm. 13.5 per cent). A deeply hooked mortarium in slightly brownish-cream fabric with buff-brown slip; packed with small, fairly well-sorted quartz and rare orange-brown inclusions. This is one of the finer variants of the fabrics produced in the potteries south of Verulamium (eg. Brockley Hill, Radlett etc.). c.AD 60-90.

10. C1090 (A-ROM8, mid- to late fourth century) (52g). A very hard, fine-textured, cream fabric few, illsorted and random, quartz and blackish inclusions; selfcoloured slip partially covering the few surviving blackish trituration grits. This is an example of the very hard fabric which could be produced in the MancetterHartshill potteries in the third and fourth centuries. The optimum date for this almost hammerhead-like flanged mortarium is AD180-260, but the texture suggests that a 3rd-century date is more likely; although this form in this fabric was probably most commonly made in this period, it would not be impossible for it to be later.

14. C1172 (B-ROM6A, 250/300-400) (128g Diam. 32cm. 9 per cent). A very hard, brown fabric with thick well-defined, reddish core and a matt, brown slip. The frequent inclusions are almost invisible at x20 magnification, except for a few larger ones; they consist mainly of quartz with few orange-brown and black inclusions. The surviving trituration grit consists of quartz and quartz sandstone and there are traces of concentric scoring on the interior and on the flange; there are also a few grits on the flange. Made in York probably in a military pottery within the period AD 70-120. 15. C1107 (B-PRO2, post-Roman) (95g). A mortarium with incomplete rim-section over-fired throughout to grey except for patches near the surface which preserve the cream to buff-cream colour intended. The brownish stain on the inside is probably also the result of overfiring, as is the extreme hardness. The very moderate inclusions include black fragments and quartz. The surviving trituration grit is black with two quartz fragments.

11. C1104 (B-PRO2, post-Roman) (48g). A fragment from the spout of a flanged mortarium in very hard, quite fine-textured, orange-brown fabric with matt orange-brown slip. The frequent, ill-sorted inclusions are mostly very tiny with some larger; they consist largely of quartz with few orange-brown and black grits. No trituration grits survive. Burnt before fracture. The spout is unusual, probably formed by trailing the equivalent of the bead straight across the flange at right angles to it. If this wide, shallow somewhat tapering flange were in Mancetter-Hartshill fabric its optimum date would be AD180-250, but extrapolating the dating of a mortarium in a well-known fabric to one made in a little known fabric cannot be regarded as reliable, especially when it may be military. Probably made at York, perhaps in a military pottery.

This can be attributed to the Mancetter-Hartshill potteries, probably in the period AD 180-230. The overfiring is such that it might be regarded as a waster, but if the vessel remained complete and of reasonable shape there was no reason not to sell it. In fact, the number of their mortaria in circulation, which are somewhat overfired, (but not to this extent), seems to be rather higher c.AD 160-200+ than at other periods.

12. C1221A (B-ROM4, 100-140/60) (80g Diam. 42cm. 4 per cent). A very hard, reddish-brown fabric with

52

The Roman Fortress Discussion

3.11, No.27; Bailey in Monaghan 1993, 732-5). It is a small fragment from the top of a volute lamp in a white, Central Gaulish fabric with a thin orange-red slip. The sherd is mould-decorated and may represent a figure, but is too small to be certain. The form, probably Loeschke Type IX or X, suggests a late first to early second-century date, which is the date-range of the assemblage (Ceramic Group 2) in which this vessel was found.

There are at least twenty different mortaria from the assemblage from eight different industries (Table A20). The group is rather small, but all the known first-century suppliers are represented (pers. comm. Kay Hartley). Those made in potteries in the Oise/Somme area of northern France are almost as equally represented as those made in York (at least 5 vessels). They are present in the earliest levels, one of which joins to sherds in postRoman levels. These include a stamped example from Ceramic Group 1 (see above No 5), and a later type G255, dated to c.AD 140-180 from Group 5. Imports from the Oise/Somme potteries were probably dwindling from AD 100 and may have petered out by AD 180 (pers. comm. Kay Hartley).

Discussion Site history and function The pottery provides several clues as to the sequence and date of site stratigraphy. In order to use the pottery assemblages to study other aspects, such as pottery supply, trade and rubbish disposal, it is necessary to establish as far as possible how the pottery came to be where it was found, in other words site taphonomy. The simplest model would be that the pottery was used in and around the excavated structures and was then left where it was when it broke. Numerous studies have demonstrated that this simplistic explanation very rarely holds true, and certainly not in a densely and continuously occupied site such as the Fortress of Eboracum.

The mortaria made in York all have matt brown or reddish slip (at least 7 vessels). All are mortaria which are most unlikely to have been stamped, and all are likely to have been made within the period AD 70-120 (see above Nos 6, 8, 11, 12 and 14, and Figure 3.12, No.34). The rim profiles plus the similarity of fabric, combined with the above features, put them into a group which can be attributed to local military production. Such mortaria made at York had a very limited distribution. Local production was never high enough to cut out sales from other potteries. There is nothing here which is likely to have been made at York or in its vicinity after AD 120 (pers. comm. Kay Hartley). They are found mainly in groups dating to the Flavian - Trajanic period, but sherds also occur in Groups 5 and 6 and in post-Roman levels, which suggests that earlier wares were re-distributed up through later stratigraphy.

In order to help establish how the ceramic assemblages came into existence, evidence for post-breakage alteration was recorded (Table A22), together with the number and weight of sherds. From these data it is possible to establish the average weight (and therefore size) of sherds of each major class (Table A8). Evidence for the presence of sherds from the same vessel in different contexts was also recorded (Table A10). Of the 15 vessels with sherds in more than one context, most cross-fit between different deposits in the drain fill (A-ROM2, A-ROM3 and AROM4) and most of these were between the first and second fills. There is evidence that the fills of the drain do contain some progressively later types and by and large sherds reconstruct with others in the same deposits but there was also clearly some mixture of the fills. However, there is no proven case of sherds from a vessel found in the drain fill occurring in late deposits.

Other first-century mortaria were supplied by the Verulamium region potteries, but there are only two examples from the site (No 13, above and see Figure 3.13, No.45). Both date to c.AD 60-90, but were found in Group 5 which dates to c.AD 140/60 - 300/400). They may have been in use over a long period – No.45 is very abraded - or they may be residual in this context. An unusual possible import from the Atisii potteries at Aoste, Isere (No 3 above, and Figure 3.14, No.60) completes the first-century mortarium group.

The completeness of the pottery from the drain suggests that the deposits consist of pottery which was broken and immediately discarded. Of the 12 vessels from the excavation with more than 0.5 EVEs present, 11 come from the drain fill. Most come from the lower fill but one comes from the upper fill and one from the accumulation over the filled drain. Despite this, about 10 per cent of the pottery from the drain (by EVEs) was abraded. This might be evidence that the vessels were thrown into the drain while liquid was still running through it rather than after its disuse. A small quantity was also partially coated with mortar (5 sherds in total). Only one samian vessel was perhaps unused and much of the kitchen ware from the drain showed sign of use. It is likely, therefore, that the drain fills represent domestic debris rather than the debris from stores or material broken in transit.

From c.AD 130/140 the Mancetter-Hartshill potteries were the major suppliers until the potteries at Crambeck started in the late third century (pers. comm. Kay Hartley). There are five examples of these wares, which do not appear on the site until c.AD 140/60, including a stamped vessel (see above No 2, and see Figure 3.15, No.59). The only mortarium from the Crambeck industries, a double-flanged painted vessel of Crambeck type 8, occurs in the dark earth levels. A small fragment of a possible Rhineland mortarium of uncertain date from group 5, complete the assemblage (see No 7, above). Lamps Unlike Blake Street, where a collection of lamps were found, this site produced a single example (see Figure

The earliest deposits produced very little pottery but

53

Excavations at the Minster Library, York 1997 included the other vessel with more than 0.5 EVEs, a flagon.

these wares are probably produced at York itself and the overall proportion of York products hardly changes from group to group (71 per cent, 71 per cent and 78 per cent).

The pottery associated with the stone structures has a very similar abrasion pattern to that from the drains. The vessels were less complete, however, but did include several with EVEs between 0.2 and 0.5. Some material in these deposits came from the same sherd families as sherds in the earliest deposits, proving that the assemblages contain some material redeposited from these levels and several contexts produced assemblages with a mixture of second-century and late first-century types. Three vessels with complete profiles were found, but all three were shallow forms: one was a platter and the other two samian dishes.

Unsourced, but probably British, wares account for 3 per cent, 4 per cent and 0 per cent. With such small quantities from small assemblages there is no way of determining the significance of any differences between the assemblages. Likewise, local wares (Rustic ware and grey burnished ware) account for only 2 per cent, 2 per cent and 0 per cent. Imported vessels are mainly South Gaulish samian, which occurs in roughly similar frequencies in each of the groups (19 per cent, 21 per cent and 14 per cent) and the remaining imported types occur in too small quantities to compare between groups. They consist of Pompeian Redware, Central Gaulish colour-coated wares, South Gaulish Les Matres de Veyre samian, Central Gaulish samian, North Gaulish mortaria and a range of amphorae, all from the western Mediterranean (Dressel 2/4, Camulodunum 186A-C, Camulodunum 186C, Gauloise 4 and Dressel 20). In total these amphorae amount to a minimum of 12 vessels (3 per cent).

The late Roman deposits, by contrast, contain mostly earlier pottery and are dated mainly on their stratigraphic position. It is likely that these assemblages were formed either by the dumping of material excavated elsewhere or by the disturbance of earlier stratigraphy. The dark earth deposits, however, contained mainly contemporary mid- to late fourth-century pottery and included three vessels with EVEs above 0.2. Nevertheless, each of the three assemblages recovered from the dark earth contains residual Ebor ware and Central Gaulish samian.

Status and function of the pottery Three models for the interpretation of the pottery from the Minster Library can be proposed and tested, or at least discussed.

Pottery supply and trade To a great extent, the pottery from the excavation fits the pattern of supply established by Perrin (1995) and Monaghan (1997) for York. During the first, second and early third centuries, York was supplied by a mixture of imported finewares and specialist products with a local industry situated immediately east of the fortress and run by the military, supplying the remainder (Table A24). After this, the quantity of imported wares declined, although supply continued, and most of the pottery was produced by a small number of large-scale industries, principally that in the Lower Nene valley, with much lower quantities from Crambeck or the Black Burnished ware industry in Dorset. Only the drain groups are large enough and span a sufficiently narrow timespan to make any statistical analysis of the assemblages meaningful.

1. That the drain was backfilled in one operation following the departure of the 9th Legion from York or the rebuilding of the fortress in stone. 2. That the drain assemblages derive from a specialised activity, such as a hospital. 3. That the drain assemblages were domestic refuse associated with contubernia rather than a centurion’s quarters. The 9th legion occupied York from c. AD 71 to c.79, at which point there is thought to have been a hiatus in occupation. The next major event in the fortress’ history was the departure of the 9th legion between AD 115 and 120, and its replacement by the 6th legion. Initially, it seems, only a small part of the legion was resident in York, and it is suggested that they occupied the same timber buildings vacated by the 9th legion. Only in the middle of the second century was the legion up to full strength in York and this corresponds to the rebuilding of the fortress in stone.

Despite the number of sherds, weight and EVEs, the drain assemblages represent a small number of vessels; there are a maximum of 174 vessels from the earliest fill, 90 from the next and only 42 from the overlying spread. Differences between the groups might therefore be due to date or to the chance absence of types that were in contemporary use but were not included in the assemblages. The presence of smashed vessels also means that it is quite likely that different lengths of the drain fill would have produced assemblages with varying compositions. Nevertheless, taken at face value, the assemblages suggest that there was progressively more Ebor ware in the drain assemblages (47 per cent, 48 per cent, 71 per cent) and progressively less Rustic ware (12 per cent, 4 per cent and 0 per cent). However, both of

The dating of the first stone buildings is provided by a Hadrianic/Early Antonine vessel (C1221). Thus the drain fill is not sealed until c. AD 130-140 and stratigraphically might therefore be of any date up to this point. As shown above, there are differences in the ratio of Rustic to Ebor wares which suggest that the three drain fills may be successive, although as noted above, the composition of the assemblages is otherwise very similar. The ratio of Rustic to Ebor ware might be an artefact of the small size of the assemblages. However, if the assemblages were the

54

The Roman Fortress result of clearing out of stores and barracks on the departure of the 9th Legion, they might be expected to contain unused material, and this is not the case. It is therefore suggested that the assemblages do represent refuse accumulated during the use of the drain. The closing date of the third group does, however, appear closer to the postulated date of the leaving of the 9th legion than to the rebuilding of the fortress in stone and it is likely therefore that the departure of the 9th legion led to a retraction of use of parts of the fortress and that this part, despite lying on a cardinal street, was little used between c. AD 110 and AD 130/150. If the assemblages accumulated during the use of the drain then they ought to reflect the status and activities of the users of the drain.

component 1 and the silt/clay-filled features have high positive values for component 2. Study of the results shows that the presence of a few unusual vessels is responsible for much of the variation: storage jars, which give high values for component 2, are represented by two sherds, both from the dark earth. The other major contributor to component 2 is drinking vessels (DR) and it is these that separate the dark earth from the remaining assemblages. The high frequency of beakers is, however, a typical feature of late Roman assemblages rather than a site-specific feature. In the early Roman period this function was performed mainly by samian vessels (SDR). Likewise, several of the main negative contributors to principal component 1 vary chronologically: liquid holders (LH), mainly flagons, are generally much more common in the late first and early second centuries than later. Similarly, samian table ware (STA), decorated table ware (STAD) and drinking vessels (SDR) were all gradually replaced by tableware (TAB) in other fabrics. Samian mortaria (SMOR), however, are an exception to this trend and tend to be more common in the late second and early third centuries than earlier. In this case, a single vessel was present, in a late second/third-century deposit and it may well be the presence of this single vessel, plus the presence of non-samian table ware, which distinguished A-ROM6 from the remaining deposits.

Comparative data exist from Blake Street (Monaghan 1993, Period 3). The Blake Street assemblage is considerably larger than that from the Minster Library (3784 sherds, 72.87 EVEs and 62kg). For this reason it might be expected to have a wider range of wares, including more exotic ones which are usually interpreted as indicating a high status assemblage. This is in fact the case. There are three lead-glazed sherds from Blake Street together with a sherd of marbled South Gaulish samian ware and nine Pompeian redware sherds. However, if the percentage of samian or other major wares is compared, the two sites are similar. There is a higher percentage of samian in the earlier fills of the drain than at Blake Street (which had 12 per cent samian by sherd percentage) and a lower percentage in the final fill. In all, however, the composition of the two assemblages in terms of major wares is very similar. Both Blake Street and the Minster Library drain assemblages are likely to represent the debris of soldiers rather than their officers. For the Minster Library, this appears to fit with the frequency of small cups, some marked with the owner/user’s name or mark (Pl I). One would not imagine that marking vessels in this way would be required in the centurion’s quarters. A final suggestion, raised by the study of the glassware (p70), is that the drain fill might represent refuse from a hospital, or at least had an unusual range of forms. Unfortunately, it is difficult to obtain reliable information on the functional composition of Roman pottery assemblages since much is classified as ‘jars’ and ‘bowls’. Both of these forms encompass a range of vessels, some of which were used in the kitchen for food preparation, some for the serving of food (kitchen-totable) and others for use in dining. As an attempt to see whether or not the character of the Minster Library assemblages varied, each specific form was assigned to a functional group (storage, drinking, table, kitchen, kitchen-to-table and so on). The resulting data were then grouped according to stratigraphic groups and analysed using Principal Components Analysis. The results (Figures 3.16 to 3.18) show that the majority of the assemblages form a tight cluster with five outliers: the drain fills (here called A-ROM2, A-ROM3, A-ROM4), a series of silt/clay-filled features of late second or thirdcentury date (A-ROM6) and the dark earth deposit (AROM9). The drain fills show high negative values for

Figure 3.16 Principal Component Analysis assemblages from York Minster Library, by function

of

Figure 3.17 Contribution of different forms to analysis displayed in Figure 3.16

55

Excavations at the Minster Library, York 1997 similar, after which form 37 seems to have rapidly superseded form 29, which ceased to be made c. AD 85. The proportions here, however, do not appear to be indicative of date: the majority have zonal or scroll decoration of the type common to form 29, one (11 below) shows the interior grooving typical of form 30 found on some of the earliest 37s. There were few pieces typical of the Flavian-Trajanic period: the majority clearly belongs to the period of overlap with form 29. The contrast is interesting between this assemblage and that from Blake Street (Dickinson and Hartley 1993, 723), where 29 outnumbered 37 by two to one. With numbers as small as this, however, statistics are unlikely to be significant. Form 30 was absent apart from one rim sherd: the second 67 was a small scrap with rouletted decoration (cf Hermet 1934, pl 90, 4). The larger and more significant pieces are described and illustrated below (Figures 3.19 and 3.20).

Figure 3.18 Principal Component Analysis of assemblages by function (combining Samian and other forms) In an attempt to remove the bias caused by treating samian ware forms as a special type, the analysis was repeated with combined data. Here too the dark earth emerged as an unusual deposit, but only because of the storage jar sherds rather than the beakers. The drain fills likewise continued to be distinguished. In addition, however, the assemblages from the early deposits in the central part of the site (B-ROM1) and the residual pottery from the western and central part of the site (A-PRO, BPRO) were also separated from the mass. In the case of the early deposits it appears that a single inkwell is responsible for this distinction while in the case of the residual groups it is probably the amphora sherds which are responsible, perhaps because weight was used as the quantification method to analyse and residual amphora sherds might resist breakage more than others.

The small amount of Central Gaulish ware, also mainly in tiny pieces, was Hadrianic or Antonine. There were no pieces from Les Martres-de-Veyre. Apart from 16 below, there was nothing of intrinsic interest. A number of sherds have been described without illustration. Figure types are quoted from Oswald 1936-37 (O). I am grateful to G.B.Dannell for assistance in identifying the ovolos. 1. Form 29, South Gaulish. Fragment of upper zone showing rosette in corded medallion between corded spindles. The rosette was used by Labio (Knorr 1952, Taf 3C) and on a form 29 by Mommo in the Pompeii Hoard (Atkinson 1914, 10), the medallion by potters such as Modestus (Knorr 1952, Taf 33) and Sabinus (Mees 1995, Taf 173, 2). Although the parallels are not particularly close, a Neronian-early Flavian date seems likely, c. AD 60-80. (C1341)

The use of PCA here shows that the method is heavily influenced by the presence/absence of rare forms and also reflects province-wide changes in the functions carried out by pottery vessels. Nevertheless, it does show sufficient promise to be worth using to compare contemporary assemblages. However, in order to do this there will have to be considerable care taken to ensure that the methodology employed is comparable from site to site.

2. Form 29, South Gaulish. Two small scraps, joining, of upper zone showing scroll with bird at the end of the tendril. A smaller bird was used this way on two of the bowls from Blake Street (Dickinson and Hartley 1993, Fig 278, 2682; Fig 279, 2694). The scroll here is fairly close to 2682, also to a scroll from Period 2 at Verulamium (Hartley 1972, Fig 85, 23). A similar scroll was used on form 37 with a large rosette ovolo in the Pompeii Hoard (Atkinson 1914, 49), where the bird appears larger and may be the one used here. c. AD 6580. (C1226, C1296)

Decorated Samian - Felicity Wild The decorated sherds reported here came from a total of 36 vessels, of which 25 were South Gaulish. Of these, fifteen were of form 37, seven of form 29, two of form 67 and one (rim sherd only) of form 30. There were scraps from eleven Central Gaulish 37s. Throughout the assemblage, the material was very fragmentary and joins were noted, in many cases, between contexts.

3. Form 29, South Gaulish. Fragment of lower zone showing a straight wreath, row of bottle buds with hound (O.1968) and palisade. The bottle bud was used by Cosius Rufinus (Knorr 1919, Taf 24, 10). The motifs were in fairly common use in the early Flavian period: a bowl by Rufinus in the Pompeii Hoard (Atkinson 1914, 35) shows a similar scheme of decoration. The palisade wreath was used by Rufinus, among others (Knorr 1919, Taf 69D). A form 29 from Rottweil (Knorr 1912, Taf 6, 5) shows both straight wreaths on a lower zone identical to one found on the Blake Street excavations (Dickinson

The most striking feature of the South Gaulish material is the high proportion of early 37s, which outnumber form 29 by more than two to one. Form 37 was first made c. AD 65-70, but is unlikely to have reached Britain in any quantity before c. AD 75. On Agricolan foundations in Scotland the proportions of the two forms are roughly

56

The Roman Fortress

Figure 3.19 Selected samian fragments (Scale 1:4)

57

Excavations at the Minster Library, York 1997

Figure 3.20 Selected samian fragments (Scale 1:4)

58

The Roman Fortress and Hartley 1993, Fig 279, 2689). c. AD 70-85. (C1296)

10. Form 37, South Gaulish, showing a different ovolo with large rosette tongue over zonal decoration, with leaf scroll with bird (O.2233A?) over archer (O.268). The ovolo and leaves may be the same as those on a bowl stamped by Patricius (Mees 1995, Taf 164, 12), who also used the bird (Knorr 1952, Taf 50). A scroll with similar ovolo and smaller leaf occurs on a bowl from Blake Street (Hartley and Dickinson 1993, fig 278, 2647), which shows links with the work of Calvus i, who worked at La Graufesenque with Patricius. Although the quality of the moulding is poor, the zonal decoration and connections suggest a date c. AD 75-95. (C1251)

4. Form 29, South Gaulish, showing a corded palisade in the upper zone, perhaps similar to Hermet 1934, pl 50, 9. There is little to suggest a potter, but a Neronian-early Flavian date seems probable. (C1211) 5. Form 67, South Gaulish. Seven fragments, including one of base with little sign of wear. Decoration shows a leaf scroll with three vertical impressions of the leaf (Hermet 1934, pl 8, 39) with Nile geese (O.2244, 2268) in the upper concavities and a zonal arrangement in the lower, with medallion with bird (O.2260), the same leaves and Nile goose over small hare (O.2076). A similar scroll with rosettes and two fringed leaves (probably different) is on form 78 from Günzberg (Knorr 1919, Textb 22). A variety of similar leaves were used and the types were common to many of the Neronianearly Flavian potters. However, the form and general style suggest a Flavian date. (C1251, C1282)

11. Form 37, South Gaulish, showing the interior grooving on the rim normally associated with form 30, but occurring on some of the earliest examples of form 37. The ovolo, above a chevron wreath, was used by Memor, Mommo and a potter whose name starts Trim-. It is common on first century sites in Scotland, though also occurs at Kingsholm (Wild 1985, Fig 23, 18) on the only example of form 37 from the site. c. AD 70-90. (C1259)

6. Form 37, South Gaulish. Seven fragments of bowl with zonal decoration. The ovolo (Dannell, Dickinson and Vernhet 1998, FG) was used by Sabinus iii on form 30 and Pontus on form 37. A 37 from Cardean (publication forthcoming) probably shows this rather than the smaller version. The general style and motifs here are difficult to parallel on published bowls. The straight wreath beneath the ovolo is from the top of Hermet 1934, pl 12, 85. Hermet illustrates its use thus on an early form 37 (ibid., Pl 81, 10). The four-pronged bud in the scroll (ibid., Pl 14, 86) occurs on a bowl from Verulamium with an incomplete mould signature (Hartley 1972, Fig 91, 76). c. AD 75-95. (C1226, C1244, C1251, C1282)

12. Form 37, South Gaulish. Five fragments, three joining. Decoration shows a horizontal chevron wreath above and below panels containing saltires and a double medallion with griffin (O.880) and hound (O.1964). The types and motifs are all used by Iustus i (Mees 1995, Taf 95, 11 shows the griffin, wreath and heart-shaped leaf, Taf 96, 2 a closely similar saltire, both on form 29). Iustus, however, normally uses a triple medallion, not a double one, as here. c. AD 75-95. (C1244, C1251, C1259) 13. Form 37, South Gaulish, showing ovolo with trident tongue over a scroll with leaf and bird in the upper concavity. The ovolo (Dannell, Dickinson and Vernhet 1998, SJ) was used on bowls signed by Sulpicius. An example from Carlisle (Mees 1995, Taf 194, 2) shows all the features and could be from the same mould. The presence of Sulpicius’ signed work from Wilderspool (Dickinson and Hartley 1992, Fig 16, 47) suggests that he was still producing in the decade AD 90-100. c. AD 80100. (C1175)

7. Form 37, South Gaulish. Five fragments of bowl with zonal decoration. The footstand retains traces of grit and appears unworn. Decoration shows panel with hare (probably O.2072) over triple festoons containing spirals. The festoons and pendant occur on a form 37 from London with mould stamp of Severus (Mees 1995, Taf 190, 2), though with stirrup leaves in the festoons. Severus also used the hare (ibid, Taf 190, 3). The festoons with spirals and probably the same pendant were used on form 37 by Meddillus (ibid, Taf 122, 3). c. AD 75-95. (C1230, C1290)

14. Form 37, South Gaulish. Two fragments in the same style, possibly from the same bowl. One shows the edge of a trident-tongued ovolo and tree with five-lobed leaf, the other, part of an animal (possibly Germanus’ bull, O.1870?) over grass tufts. The five-lobed leaf occurs on a sherd stamped [GER]MANIF from Rottweil (Knorr 1952, Taf 30F) and on bowls attributed to the potter(s) stamping GERMANI.SER (Mees’ Germanus IV) from Ovilava (Karnitsch 1959, Taf 23, 6-10). The work of this potter, or combination of potters, occurs both at La Graufesenque and Banassac. A bowl from Northwich (Wild 1972, Fig 12, 9) shows a tree with similar leaf and ovolo. The grass motif is a common feature of Germanus’ style. c. AD 85-110. (C1352, C1356)

8. Form 37, South Gaulish, with panels showing part of a saltire and hound (O.1925). The motifs in the saltire are on a bowl stamped by Felix (Knorr 1952, Taf 23A) from a mould possibly made by Modestus, cf a bowl from Kingsholm (Wild 1985, Fig 17, 1), though here Modestus' twist motif is certainly broader. Although such saltires were in use under Nero, its use here on form 37 puts the date into the Flavian period. c. AD 70-90. (C1341) 9. Form 37, South Gaulish, showing the ovolo with large rosette tongue used by Frontinus (eg Mees 1995, Taf 65, 3, from London) and occurring in the Pompeii Hoard and on first century sites in Scotland. c. AD 70-90. (C1211)

15. Form 37, South Gaulish. Two joining fragments of the same ovolo with trident tongue, and another, possibly from the same bowl. The ovolo is well known

59

Excavations at the Minster Library, York 1997 from Flavian-Trajanic contexts, though rarely on signed bowls. It is on a form 37 from La Graufesenque with signature of Albanus (Mees 1995, Taf 1, 8), so was clearly used by him and his associates. c. AD 80-110. (C1157, C1315. Possibly C1356)

5. C1282 (1) Masclus i (Masculus) 5b 27g OF SCI (Hull 1958, fig 79, 8) La Graufesenquea. c. AD 50-65. 6. C1282 (4) Memor 5a 27g MO (Hartley 1972a, 224, S15) La Graufesenquea. c. AD 65-90.

16. Form 37, Central Gaulish. Two fragments, not joining but probably from the same bowl, in the style of Sissus I, of whom the vertical borders of double (or multiple) wavy lines are a characteristic feature. A mould from Cannes Museum (Rogers 1999, Pl 113, 10) shows the double border with astragalus and two poorlyexecuted panthers. The panther here (R.4019) is a new type listed in Rogers 1999, used on the style of X.11, not noted for Sissus I. Sissus’ links are with the Quintilianus group, and he is likely to have been at work c. AD 125150. (C1125)

7. C1106 Modestus i 8a 15/17 or 18 [O]FMOE La Graufesenqueb. c. AD 55-70. 8. C1359 (2) Passenus (Passenius) 60a’ 15/17 or 18 P SSIE (60a P SSIEN) La Graufesenquea. c. AD 55-70. 9. C1251 (1) OF M on form 15/17 or 18, South Gaulish. Early Flavian.

17. Form 37, Central Gaulish, showing Docilis’ ovolo (Rogers 1974, B24) and cock (O.2361). c. AD 130150. (Not illustrated) (C1177)

10. C1251 (3) ]MDI? on form 18, South Gaulish. Early Flavian.

18. Form 37, Central Gaulish, showing freestyle hunting scene in the style of Criciro, with his bear (O.1595), small panther (O.1512) and snake on rock filling ornament. c. AD 135-170. (Not illustrated) (C1135)

11. C1251 (4) IXVM I? on form 36, Central Gaulish. Flavian-Trajanic. Stamped inside the base, towards the outer edge. A dish in orange-brown fabric, heavily micaceous, with orange-brown glaze. Both belong to the Lezoux range of the late first early second centuries.

19. Form 37, Central Gaulish, in a light orange, micaceous fabric reminiscent of that of Lezoux in the pre-export period. The decoration, however, is Antonine, showing panel with wide, single festoon over the hound (O.2007A). The style suggests the Paternus v group. The festoon and border of large, squarish beads were used by Laxtucissa (Stanfield and Simpson 1958, pl 100, 24). c. AD 150-180. (Not illustrated) (C1111)

12. C1282 (2) IIIIIII on form 27g, South Gaulish. Early Flavian. 13. C1346 ..F[ on a cup (form 27 or 42), South Gaulish. Neronian or early Flavian. C1359 (1) F.[ pn form 27, South Gaulish. Just

Samian stamps - Brenda Dickinson

14.

Each entry gives: site context number, potter (i, ii, etc., where homonyms are involved, die number, form, reading of the stamp, published example (if any), pottery of origin, date. a indicates stamp attested at the pottery in question. b indicates potter, but not the particular stamp, attested at the pottery in question.

possibly Rufus iii 4a (OF.RVF). Flavian. Roman building materials - Sandra Garside-Neville 113kg of ceramic building material, mortar, plaster, and stone was submitted for analysis. The sample was recorded on the Museum of London Building Material Recording proforma, using fabric samples held by the York Archaeological Trust. All of the fabrics had been recorded in York previously. The ceramic building material archive comprises written records (including Word 6 files), Access 2 databases, and tracings of tegula flanges and drawings of four fragments (three tile stamps and one graffiti fragment). Those fragments which have interesting and diagnostic features, such as types of keying, hobnail imprints, etc were retained. All other fragments are recommended for discarding. The on site collection policy, carried out by the excavators, involved the retention of fragments with obvious features of interest, such as near complete fragments, patterns, edges or holes.

1. C1282 (3) Calvus i 5g 15/17 ]LVI (Walke a 1965, Abb. 2, 10) La Graufesenque . c. AD 75-90. Smaller than normal for Britain (diameter 14.3cm). 2. C1107 Carussa i 3a 31 SS = a (Dickinson 1986, 188 3.28) Lezoux . c. AD 160-190. 3. C1251 (2) Celadus 1b 15/17 or 18 [ La Graufesenquea, Le Roziera. c. AD 60-75.

DI>

4. C1128 Cintusmus i 6a 37 rim CINTVSM[VSF] (Hartley 1972a, 258, D122) Lezouxa. c. AD 150-180. On a bowl with ovolo 3 of Cinnamus ii (Rogers B143).

60

The Roman Fortress One brick fragment from C1197 (second-century) has a piece of incomplete graffiti. It reads:

The bulk of the sample was Roman in date. The fabrics (R2, R3, R6, R6B, R8, and R9) are all very common to York. This material may have come from the legionary tile works at Peasholme Green. It seems possible that fabric R2 is a later fabric as its presence in this sample is confined to third and fourth-century contexts. Fabrics R8 and R9 occur in first-century contexts so may be early fabrics.

] V DVS ]TVS Other marks Hobnail imprints are present on the brick. However the pattern of the nails or the size of the shoes can not be discerned. There is one piece that may have raindrop marks, showing that the bricks were exposed to the elements during the manufacturing process.

Brick was normally used for various purposes within Roman buildings including hypocausts and wall courses. At this site, several fragments were found in fill C221283 of hearth F70, which was possibly part of a hearth structure.

Roman roof tile comprises two forms, tegulae and imbrices.

Sizes

Tegulae (56 items)

For recording purposes, where a fragment had no other features and its form was uncertain, the default was Roman brick. Consequently, it is necessary to disregard thickness measurements that also fall within the thicknesses of known tegula fragments. The thickness measurements of Roman bricks from the York Minster Library site can be said to fall within 39-80mm, though there may be thinner examples. There is a possible cuneatus (solid) voussoir fragment (used in archways or vaults) from C1346 which has maximum thickness of 41mm and a minimum thickness of 32mm.

No complete tegula was recovered. The thickness of the tegulae falls between 22-53mm. The most common lower cutaway, using Betts’ lower cutaway classification (Betts 1985, 160), is D, with a few Bs and Es. Imbrices (17 items) The thickness of the imbrices falls between 15-26mm. One of the fragments has a legionary tile stamp on it (see below).

One complete bessalis brick was found from C1106 (residual, 12th-century). It measured 211x210x70mm. Betts notes that York bessales are larger than average (Betts 1985, 176). This is certainly true of the York Minster Library example which is easily larger and thicker than Brodribb’s national average measurements of 198x198x42mm (Brodribb 1987, 142). This type of brick was commonly used in the pillars of hypocausts, though clearly it was being reused here in the medieval period.

Box flue tile (5 items)(Table A25) There is only a small amount of flue tile. However, there are three types of keying. There is scored tile, with a lattice pattern, and this occurs in a second-century context. Combed and finger-impressed keying is present in third to fourth-century contexts. This seems to support evidence from London that combed keying was rarely used before the second century and that scored keying is an earlier feature of flue tile (Betts and Crowley 1993, 53-4).

Keying Keying is found on several Roman bricks. It takes the form of finger impressed lines running across the tile. There are plain lines, and also others with a lattice work pattern. Keying would be used to help mortar attach itself to the surface of the brick. It is possible that these tiles were used as wall tiles, along with spacer bobbins and Tclamps, to form cavity wall heating. However, some of these tiles are very thick, which would have made fixing them to a wall difficult.

Legionary tile stamps (see Figure 3.21; Table A26) Three tiles bore legionary stamps, two tegulae bore stamps of the 9th Legion and an imbrex bore that of the 6th. Painted plaster (Table A27) The plaster was deposited as early as the first to second centuries. All of the fragments were too small to be able to discern a pattern. However, the colours include white, red, grey, blue-green and black. Several examples were found in situ (C1248, wall F129). There are a couple of black and white fragments from contexts dating from the twelfth to fourteenth centuries which are presumably residual Roman fragments. All of these colours were also found during the excavations of the nearby principia and barracks. These excavations yielded painted plaster associated with the timber barrack phases, as well as the later stone built phases (Weatherhead in Phillips and Heywood 1995, 249).

Signatures Signatures are thought to be the mark of particular brick makers (Brodribb 1987, 99), and are commonly found on brick and tegula. The signatures found on the York Minster Library brick are fragmentary but evidently include loops and arcs, some of which have been previously recorded by Betts (1985, 192-194). Graffiti (Figure 3.21)

61

Excavations at the Minster Library, York 1997

Figure 3.21 Roman brick with graffito and Legionary-stamped tiles (Scale 1:4)

62

The Roman Fortress Opus signinum (6 frags)(Table A28)

come from a suit of so-called lorica segmentata in which iron plates were held together by brass hinges and leather straps. This type of armour was introduced in the first century and remained in use until the third (Bishop and Coulston 1993, 145).

Opus signinum, Roman concrete which contains tile chippings, was found in three first to second-century contexts. One fragment from C1282 had a noticeably worn face.

Padlock case end-plate Find No.337 C1119 Roughly circular plate with a rectangular slot in the centre. Pierced on either side of the slot and set in each hole is a pin; both are now bent flat against the inner face of the plate. The plate is pierced twice for attachment near the edge and there may have been a third hole in an area which is now damaged. D.65; pins: L.31mm. (See Figure 3.22)

Ferrous metalwork - Patrick Ottaway Roman objects Bars Find No.334 C1119 L.85, T.15mm Find No.229 C1302 It appears to widen and curve over slightly from the centre towards one end; thickest at other end where it is broken. Possibly an incomplete chisel. L.140, W.16, T.12mm.

This is a curious object, but it is probably the end-plate from a barrel padlock case. The slot in the centre of the plate was probably for the arm of a U-shaped bolt to which the springs were attached, the other arm being set in a separate tube on the side of the case. The two pins on either side of the hole would have projected into the case and to release the bolt a key would have had to pass over them as well as the springs. Padlocks with this type of mechanism are found in the post-Roman period and a reconstruction drawing appears in Ottaway 1992 (Figure 285). Although barrel padlocks are well-known in Roman contexts, no obvious parallels from a Roman context were found for the type suggested here. The small holes near the edge of the plate were presumably for small rivets which attached the plate to the rest of the case. In the post-Roman period the end-plates were usually attached to the case with brazing metal, but there is none visible on this object.

Punch Find No.219 F121 C1233 A probable smith’s punch of rounded cross section and a rounded tip. L.150, T.21mm. Needle Find No.335 C1126 Head incomplete. L.38, W.2mm. Knife Find No.336 C1125 Oval ring terminal, short stem and stub of blade. L.62; terminal: W. 30mm. Figure 3.22. Fittings Find no.338 C1216 A plate broken at each end and narrows from one to the other; pierced twice at the narrow end, small rivets in situ. L.39, W.16mm.

Buckle Find No.339 C1352 Very corroded D-shaped frame. L.35, W.20mm

Find No.217 C1270 Strip. In two pieces. Pierced twice. L.39, W.15, T.9mm.

Context 1125 1220 1251

Hobnails

Both these objects must be incomplete bindings, brackets or similar items from furniture or wooden structures. Pierced Plate Find no.218 C1185 A rectangular plate which has a section with a curved edge along most of one side as a result of which there is a near pointed tip at one end. The short straight part of this side is bent over a little. The object is pierced in the corner between the opposite side which is straight and the straight end. L.170, W.85, T.3mm. (See Figure 3.22)

Feature 120

Quantity 1 93 14

Dagger Find No.198 C1121 In three pieces; the tip is missing and only a stub of the tang survives. The blade is waisted with curvilinear edges, and has a central rib running down the centre of one or both faces. On both blade faces there is mineralreplaced organic material which is probably the remains of the sheath. The X-radiograph shows a distinct line within the corrosion crust running across the width of the blade approximately 3mm below the top edge of the blade. This may indicate the top edge of the sheath or more likely the edge of the hilt. Another faint line of what appears to be a corrosion product runs roughly parallel above the top edge of the blade. This may also be a trace of the hilt. One non-ferrous tubular rivet from the sheath

This is a plate of distinctive shape which may not now be complete and its original function is uncertain. The curved edge along most of one side suggests, however, that it could have been part of a plate from military body armour which fitted around the neck of a soldier. If this identification is correct, the plate would probably have

63

Excavations at the Minster Library, York 1997

Figure 3.22 Roman ferrous metalwork (Scale 1:2)

64

The Roman Fortress survives in situ near the top of the blade. L.267; Blade: L.257, W at top 83, in centre 70mm. (See Figure 3.22).

military sites in Britain and elsewhere from the Claudian period on into the second century (ibid.). Similar fittings were recovered from the Purey Cust Nuffield Hospital site, located in the retentura of the fortress in York (Cool, Lloyd-Morgan and Hooley 1995, Nos 6529-30). Copperalloy hinged fittings such as Find Nos 224, 226-8 were superseded during the Antonine period by a simpler fastening system (Cool, Lloyd-Morgan and Hooley 1995, 117).

The Roman dagger from C1121, dated to the twelfth to thirteenth century, is a rare example of an iron weapon from the fortress at York in which excavations have hitherto produced little military equipment. As originally constructed the dagger would have had a hilt which fitted over the tang and was probably riveted to it. A crosspiece at the base of the hilt would have fitted over the top of the blade and the lines in the corrosion at the top of the York dagger probably indicate its location.

Although found in an eleventh to twelfth-century backfill, Find No.170 (see Figure 3.15) appears to be a fragment of a Roman openwork belt-plate, although too little is left to establish the original design. Similar examples from Colchester (Crummy 1983, 137, no.4240) and Caerleon (Brewer 1986, 175, nos.36-8) date from the second to third centuries. The arrow-shaped belt mount Find No.167 (see Figure 3.23) may be contemporary. The strap-end tongue Find No.169 (see Figure 3.23) found in a ?fourth-century backfill is paralleled by examples from a third-century deposit at Carlisle (McCarthy 1990, 126, no.65), and from second to third-century levels at Caerleon (Brewer 1986, 177, nos.49-52), where it is noted that they are common finds in Britain and the Continent (ibid.). A complete strap-end with this tongue form was found in Chichester (Down 1978, 302, no.110). Elsewhere in York, the site at 8 Wellington Row produced a similar strap-end tongue, although this has been published as possibly Anglian (Tweddle, Moulden and Logan 1999, Fig.86, no.95k).

Daggers were probably back-up weapons for a Roman soldier, perhaps if his sword were lost, and they were used by both legionaries and auxiliaries (Bishop and Coulston 1993, 76). It has often been thought that daggers were primarily a weapon of the first century and many British examples are of this date. They include those from Hod Hill, Colchester and Gloucester in the British Museum’s collection discussed by Scott (1985). It appears, however, that although the blades of many of these daggers are similar in form and, if slightly shorter, similar in size to the York example, they are markedly narrower, being no more than c. 40mm wide in the centre as compared to 70mm. In a group of third-century daggers illustrated and discussed by Bishop and Coulston (1993, 135-6, fig.95) there are four, one from London, one from Künzing, Germany, where an ironwork hoard including daggers was found, and two from Eining, Germany with blade lengths of c. 240-300mm, but proportionally wider than the earlier daggers. They have length to width ratios of about 4:1 which is much the same as the York dagger would have had when complete. Bishop and Coulston also note (ibid.) that a small number of daggers from the Künzing hoard had pattern-welded blades. In conclusion, in the absence of any other diagnostic features, it is likely that the York dagger is third-century in date.

Two objects appear to be fittings from horse harnesses, and are presumed to be military. The stud with a green glass setting Find No.140 (see Figure 3.23) is directly paralleled by two further examples from York, found in contexts of c. AD 100-160 at 9 Blake Street, a site within the praetentura (Cool, Lloyd-Morgan and Hooley 1995, Nos 6313-4). A similar stud from Caerleon was found in a 1st-century context (Brewer 1986, 175, no.19). The leaf- or heart-shaped pendant Find No.189 (see Figure 3.23) was found in a 2nd-century spread; a similar pendant was found at Carlisle in an early third to late fourth-century context, but is considered most likely to date from the Flavian period (McCarthy 1990, 122, no.51).

Non-ferrous metalwork - Nicky Rogers Military equipment Military equipment forms the largest element of this small non-ferrous assemblage, which also includes personal ornaments, structural metalwork, furniture fittings, lock fragments, sheet fragments, and miscellaneous objects. This comprises copper alloy strapand buckle-fittings and a girdle plate tie-hook from lorica segmentata, belt-plate, belt mount and strap-end fragments, harness pendant and harness stud.

Personal ornament Two copper-alloy brooches were found, both of which are likely to date to the first century AD. The Nauheim derivative bow brooch fragment (Find No.324, see Figure 3.23) dates from the Claudio-Neronian period (Cool and Philo 1998, 29, Nos 1-3); a second possible example was found at 9 Blake Street (Cool, Lloyd-Morgan and Hooley 1995, 6332, 1544).

All the lorica segmentata fittings come from two contexts, both first to second-century in date: the hinged strap-fitting Find No.224 was found in C1296, while Find Nos 226-8 (Figure 3.23), comprising strap- or bucklefittings, and a probable tie hook came from C1270. The very corroded Find No.212 found in a second to thirdcentury backfill may also be from lorica segmentata. Lorica segmentata is a form of armour made up of iron strips articulated on leather straps and with copper-alloy fittings (Bishop and Coulston 1993, 85); it is found on

Dr Hilary Cool comments upon the brooch Find No. 208 (see Figure 3.23): This brooch belongs to the headstud family whose floruit centres on the last quarter of the first century and the first half of the second. The commonest form of these is the Lamberton Moor variant (see for example Crummy 1983,

65

Excavations at the Minster Library, York 1997

Figure 3.23 Roman non-ferrous objects (Scale 1:2)

66

The Roman Fortress no. 65) which was clearly in use during this whole period (Mackreth 1985, 199). During the first century especially, however, there were many other variants of headstud in use and it is to this period that Find No.208 belongs. Several of the features it shows can be parallelled in brooches found in early contexts. The long triangular cell with rectangular cells above on the bow and the perforated catchplate are also found on an example found during excavation of the bath house of the Agricolan supply base at Red House, Corbridge (Daniels 1959, 156 No.11; Snape 1993, 32 No.4). The transversely grooved panel infilled with enamel above the headstud is very similar to the decoration on an example found in the Phase Ic midden at Castleford (Cool and Philo 1998, 40, No.23) for which a date towards the end of the 70s seems appropriate. The most likely period for 208 to have been made thus appears to be c. AD 75-85.

Conclusion The datable finds in this assemblage are consistently first to second-century, and the emphasis is on the military element. It is worth noting that little military equipment had been found previously in York (Ottaway 2004, 56), although one other site which also produced a mainly military assemblage is the Purey Cust Nuffield Hospital site, situated quite close to the Minster Library. Analysis of the finds from the Purey Cust led to the suggestion that the artefacts derived from possible barracks of equitates legionis or from a fabrica where military equipment would have been repaired (Cool, Lloyd-Morgan and Hooley 1995, 1646). Coins - Craig Barclay Find No.211 C1244 Domitian; AE dupondius; c. AD 81-96

As currently preserved the brooch lacks the headloop that is often a feature of these brooches. It does, however, have a slight knob centrally at the top of the bow, these are often found on brooches with loose wire headloops to stop them falling forward. It is possible that originally this brooch had such a headloop which would have been secured inside the hinge casing.

Find No.166 C1005 Vespasian; AE dupondius; Lyons; AD 71 or later Find No.139 C1110 Septimius Severus; denarius; AD 202-10 Find No. 27 C1090 House of Constantine I; AE 3; Trier; c. AD 332-33

Only two other copper-alloy objects may be classed as personal ornaments, and the identification of neither is certain. Find No.83 is a pin shank from undated dark earth, and unfortunately is itself undatable. A possible finger-ring Find No.209 was found in a second to thirdcentury spread.

Find No.26 C1090 House of Constantine I; AE 3; c. AD 330-48 Find No. 25 C1090 Barbarous imitation of House of Constantine; AE 4 prototype; c. AD 330-48

Furniture fittings An incomplete copper-alloy nail Find No.168 found in third-century backfill and a copper-alloy stud Find No.197 (see Figure 3.23) from a second to third-century layer may have been used on furniture (cf Crummy 1983, 115-6, Nos 2982-3050, 3095-3137).

Find No. 45 C1090 Valentinian I; AE 4; AD 364-75 Find No. 28 C1090 Valentinian I; AE 4; AD 364-75

Locks Possibly from a small chest or cupboard, Find No.138 is a copper-alloy tumbler lock bolt fragment (see Figure 3.23); two similar fragmentary lock bolts were found at 9 Blake Street, York in later first to mid-second-century deposits (Cool, Lloyd-Morgan and Hooley 1995, Nos 6356-7), and another was found unstratified at Caerleon (Brewer 1986, 189, No.185). Find No.19 may be the end of a copper-alloy lock-pin from a tumbler lock (see Figure 3.23) (cf Crummy 1983, 124, No.4143; Brewer 1986, 188, No.172). Both Find Nos 19 and 138 were found in medieval levels, but almost certainly date from the Roman period.

Find No.23 C1081 House of Valentinian I; AE 4; AD 364 or later Find No.21 C1081 House of Theodosius I; AE 4; c. AD 388 or later Find No.22 C1081 Barbarous AE 4; 4th C AD The Roman glass - Hilary Cool Vessel glass

Miscellaneous Find No.210 (see Figure 3.23) is a ferrule which appears to have been cut down at one end; it comes from a firstcentury fill. Rings which may have fulfilled a variety of functions were found in contexts ranging from the second to the twelfth century (Find Nos 96 and 220).

The excavations produced 81 fragments of Roman vessel glass of predominantly first to second-century date. There is a small amount of fourth-century material, but the later second to third-century vessels which are normally very common on York sites are virtually absent. Compared to the vessel glass published from the excavations under the Minster (Price in Phillips and Heywood 1995) and at 9 Blake St (Cool, Lloyd-Morgan and Hooley 1995) this is a

67

Excavations at the Minster Library, York 1997 very small assemblage, but it does show some interesting anomalies from the normal pattern of glass use noted within the fortress.

Fourth century vessels are represented by Nos.61-9 in the typical range of greenish bubbly glass of that period, and probably also by the bubbly blue/green fragment No.7 found in a fourth to fifth century context. There are some grounds for thinking that several of these vessels may be late fourth-century forms. The rim fragment No.61 belongs either to a hemispherical cup or conical beaker with cracked off rim (Cool and Price 1995, 88), a type in use throughout the fourth century (see Figure 3.24). The strong yellow/green colour No.61, however, is one that occurs more often in the later fourth century (Price and Cottam 1998, 16). The rim fragment No.62 is unfortunately rather small but appears to come from an open vessel with a vertical side and fire-rounded rim. This is not a form that is common during the fourth century, but cylindrical bowls with fire-rounded rims do start to appear at the end of the fourth century (Cool 1995, 13) and have been recovered from the very late fourth to fifth-century dumps at Wellington Row in the colonia (unpublished YORYM 1987.24). It is also possible that No.63 may have come from a segmental bowl or truncated conical indented bowl (Cool and Price 1995, 104). The latter form is a later fourth-century one and was common in the assemblage from the excavations below the Minster (Price 1995, 360 nos. 40-1, Fig.143). This is perhaps significant in the light of the known 5thcentury occupation at that site. A late fourth to fifthcentury date for all these fragments would not be inconsistent with the stratigraphy as all came from the contexts assigned to the fifth to tenth century. The only other fourth-century vessel form that can be identified is a jug (No.64), but the base type is not sufficiently diagnostic for the precise variety to be identified (see Figure 3.24).

Material that definitely belongs to the first century includes the pillar moulded bowl No.1 and a deep blue vessel decorated with marvered white spots (No.2). The discovery of the former at the Minster is no surprise as pillar moulded bowls are very common on first century sites in Britain. Numerous examples were found during the excavations under the Minster (Price in Phillips and Heywood 1995, 353 Nos 1-6, Figure 141) and at Blake Street (Cool, Lloyd Morgan and Hooley 1995, 1650 Nos5845-90, Fig.734). Vessels decorated with marvered blobs, however, are rarer finds in York. It was a style of decoration that was most fashionable in the mid-first century up to c. 75, and so was going out of use by the time York was occupied. The discovery of substantial parts of a ribbed jug in this type of glass in a ditch fill at Carlisle closely dated to 84/5 (Caruana 1992, 67 No.5, Fig. 10) shows that some blue and white vessels were still in use at that time, but this was probably one of the last ones to survive. Other than this fragment from the Minster Library, the only other fragment from a blobdecorated vessel from York is a greenish body fragment with red, opaque white and blue marvered decoration from the Mount (unpublished YORYM 1974.120). Apart from these two vessels, other first to secondcentury tablewares are very rare in the assemblage. There is a fragment (No.4) probably from a jug of Isings Form 55, a common type of the mid-first to mid-second century (Cool and Price 1995, 120), and a small number of other undiagnostic body fragments (Nos.5-6, 8-17)(Figure 3.24). By far the bulk of the assemblage though consists of fragments of blue/green bottles (50 fragments nos. 1859). Cylindrical (Nos.20-1, 35-8), hexagonal (No.42) and square bottles (Nos.33, 30-41. 58-9) are represented, with the other fragments only being identifiable as bottles or prismatic bottles. These blue/green containers became very common in the later first century. The cylindrical form appears to go out of use early in the second century, but the prismatic forms continue in use into the third century (Cool and Price 1995, 179). An interesting feature of two of these fragments (Nos.58-9) is that they have been flaked like flint for a secondary use as a tool and something similar may have occurred on No.33 (see Figure 3.24). The re-use of thick fragments of vessel glass as tools is not uncommon on Roman sites, but of these three fragments only No.33 was found in a Roman context.Nos.58-9 were found in medieval ones and so they could have been re-used in the post-Roman period.

Overview Given that this is primarily a first to second-century assemblage, it has a rather unusual composition. Military sites of this period do tend to have vessel glass assemblages dominated by bowls and bottles (Cool and Baxter 1999), but the concentration on bottle fragments to the virtual exclusion of all other forms is unusual. This can be appreciated from Table 3 where the Minster Library assemblage (excluding Nos.3, 7 and 61-9) is compared to that from the Phase 2 and 3 contexts at 9 Blake Street (Cool, Lloyd-Morgan and Hooley 1995, 1559-88). The material in these phases relates to the occupation of the site from the foundation of the fortress to the departure of the legion to the northern frontier under Hadrian, and so should be broadly contemporary with the Minster Library assemblage. The small assemblage from the Swinegate sites (ibid, 1608-11) is also shown as this too appears to be primarily a first to second-century assemblage.

The single vessel that can be assigned to the late second to third century is No. 3 which comes from a cylindrical cup with pulled-up knobs (see Figure 3.24). This is a third-century form, commonest in the middle part of the century (Cool and Price 1995, 86). In York, the type is probably best represented amongst the unpublished assemblage from the General Accident Site, Tanner Row (YORYM 1983-4.32), but isolated examples as here have also been found on other sites.

68

The Roman Fortress

Figure 3.24 Selected vessel glass fragments (Scale 1:2)

69

Excavations at the Minster Library, York 1997

Table 3 A comparison (in EVEs) of the first to second-century glass assemblages from three York sites. (+ indicates the suspected presence of a form in quantities unsuitable for calculating an EVE measurement) Cup

Bowl

Jar

Flask

Jug

Bottle

Total EVE

Minster Library

Fragment numbers 70

-

0.2

-

-

0.14

2.1

2.44

Blake Street

690

6.2

6

1.44

2.2

0.65

2.52

19.01

Swinegate

116

1.4

-

0.99

1.4

0.79

0.98

4.7

Site

As can be seen at both the other York sites a range of vessel types are present. As this also occurs in the small Swinegate group, this suggests that the concentration on bottles at the site is not just a function of the small size of the assemblage. The large assemblage from beneath the Minster (Price 1995) is not suitable for presenting in the format shown in Table 3, but it too has a range of first to second-century vessel types represented including a range of drinking vessel and bowl forms. This suggests that the glass vessel assemblage found at the site might be deriving from some specialised type of activity. With so many bottles represented, it would be tempting to say that this was the debris from a kitchen or storage area. However, as the pottery assemblage appears to be relatively deficient in vessels associated with food preparation, cooking and possibly also in storage vessels, this does not appear to be likely.

suspected that melon beads could have been used to ornament horse harness, and good evidence of this was recently published from the cemetery at Krefeld-Gellep where a horse had been buried wearing a necklace or collar of glass and frit melon beads (Pirling 1997, 58-9, Taf 152). The opaque green disc cylindrical glass bead (No.7) is of a type that has generally been assigned to the late Roman period (Guido 1978, 95). A recent re-appraisal of the contexts in which they have been found, however, suggests that the type may not have emerged until the late fourth century and may be indicative of late fourth to fifth-century occupation (Cool 2000). Certainly in York it has been found on sites where the fifth century is attested such as under the Minster (Price 1995, 368 no. 89, Fig.146) and at 9 Blake Street (Cool, Lloyd-Morgan and Hooley 1995, 1650 no. 5832, Fig. 722), though in both cases, as here, not in usefully stratified contexts.

One possibility might be that the assemblage derives from a hospital or some sort of medical facility. There are a number of references in the classical literature to glass vessels being used as medical containers (see for examples Trowbridge 1930, 164 note 93, 169 note 148, 170 note 151, 171 note 162). Glass containers would be ideal to keep medical supplies in as they would not taint and would have been much easier to keep clean than pottery, wood or even metal ones. The idea of the assemblage deriving from a hospital has also been suggested by the pottery study (p54).

1.

Melon bead. Turquoise frit; complete. Heavily weathered and showing much wear around one end. Length 10mm, diameter 12mm, perforation diameter 4.5mm. C1219 Find No.187.

2.

Melon bead. Turquoise frit, approximately threequarters extant in 4 fragments. Length 9mm. C1247 Find No.207.

3.

Melon bead. Turquoise frit; approximately one quarter extant. Length 9mm, C1062 F43.

Three fragments of Roman window glass were recovered. Nos.1 and 2 are typical fragments of the cast variety commonest from the first to third centuries. No 3 seems most likely to be a fragment of blown window glass, though the surfaces are more dulled than is normal with the type. The thinner blown window panes came into use in the fourth century.

4.

Melon bead. Turquoise frit; approximately one quarter extant. Length 11mm. C1352.

5.

Melon bead. Turquoise frit; approximately one quarter extant, retaining turquoise glaze on ridged interior of perforation. Length 11mm. C1238 Find No.186.

1 C1251 Blue/green cast fragment. Area 8cm2. 2 C1107 Light green cast fragment. Area 1.5cm2. 3 C1081 Light green, bubbly blown? Area 2cm2.

6.

Melon bead. Turquoise frit; approximately one quarter extant. Length 18mm. C1005

7.

Disc cylindrical bead. Opaque green glass. Length 3mm, diameter 6mm, perforation diameter 3mm. C1107 Find No.94.

Window glass

Six complete and fragmentary frit melon beads were recovered from the excavations (Nos.1-6). These are very common during the first century and early to mid-second century, and are often found in large numbers on military sites (Crummy 1983, 30). It has long been

70

The Roman Fortress Mammal bones were described as ‘juvenile’ if the epiphyses were unfused and the associated shaft fragment appeared spongy and porous. They were recorded as ‘neonatal’ if they were also of small size.

Vertebrate remains - Deborah Jaques Nine boxes of vertebrate remains were recovered, most of which were dated to the medieval period. An assessment of the material, carried out by the EAU recommended that further work was not warranted, unless there was refinement of the dating framework (Jaques et al. 1999). A tighter chronological framework was achieved for the material from the Roman deposits, which were combined into four groups (first to second century, second to third century, third to fourth century and fourth to fifth century). Of particular interest were C1251 and C1282, two first-century primary deposits associated with the open drain, F134. Material from these contexts and from the other first and second-century deposits was looked at in some detail. Table A29 contains a list of the contexts from which material was recorded and the date group to which they were assigned.

Measurements (unless otherwise specified) followed von den Driesch (1976). Additional measurements, not detailed by von den Driesch, followed those described by the sheep-goat working-party (Davis 1992 and Dobney et al 2007). There were insufficient measurements for detailed analysis and discussion. Although similarities between the bones of sheep and goat often make it difficult to distinguish between them, certain elements can readily be differentiated and recorded to species level with the aid of good comparative specimens and using the criteria outlined by Boessneck (1969). Differentiation was typically attempted for horncore, distal humerus, distal radius, metacarpal, distal tibia, calcaneum, astragalus, metatarsal and all phalanges.

Few early Roman bone assemblages have been recorded from York, and even fewer have been published. Therefore, despite the small size of this assemblage, the tight dating and the location of the deposits from which the bones were recovered, render it of some importance.

Since the assemblages were small, a single standard method of species quantification for bone was employed: this involved simple fragment counts, where calculation of the total number of bone fragments involved the simple counting of all recorded identifiable fragments (number of individual skeletal parts or NISP). Unidentifiable fragments were recorded and quantified separately. At best, quantification using simple variations in numbers of fragments provides data of limited value with regard to absolute numbers of individuals present in the assemblage.

Vertebrate data were recorded electronically directly into a series of tables using a purpose-built graphical input system and Paradox software. Briefly, semi-subjective, non-quantitative data were recorded for the material from each context regarding the state of preservation, colour, and the appearance of broken surfaces (‘angularity’). Additionally, semi-quantitative information was recorded for each context concerning fragment size, dog gnawing, burning, butchery and fresh breaks.

The animal bone assemblage dating to Period 1 (1st/2nd century) provided the largest group of fragments (518, of which 140 were identified to species) and included the vertebrate remains from the 1st-century drain deposits (Table A30). In fact, the latter formed 60 per cent of the assemblage.

Identification was carried out using the reference collections of the Environmental Archaeology Unit. Detailed recording of the assemblage followed the protocol outlined by Dobney et al. (2007) Selected elements (‘A’ bones) were recorded using the diagnostic zones method described by Dobney and Rielly (1988). Remaining elements which could be identified to species (‘B’ bones) were merely counted. Other fragments, (classified as ‘unidentified’) were, where possible, grouped into categories: large mammal (assumed to be horse, cow or large cervid), medium-sized mammal 1 (assumed to be sheep, pig or small cervid), medium-sized mammal 2 (assumed to represent dog, cat or hare) and unidentified bird. As well as counts of fragments, total weights were recorded for all identifiable and unidentifiable categories.

Drain F134 C1251 and C1282 (first century) A small assemblage of vertebrate remains was recovered from these two deposits, amounting to 75 identified and 236 unidentified fragments (see Table A30). The bones were mainly well preserved, with sharp edges; however, both deposits contained a small proportion of fragments which were rather battered in appearance. This was more apparent in the material from C1282. On the whole, colour was recorded as mid to dark brown, although some fragments from C1251 were lighter in colour. The latter tended to be quite eroded and less well preserved. Little dog gnawing was evident and fresh breakage was negligible. Evidence of butchery (see below) was fairly extensive throughout the assemblage, particularly on cattle fragments. The identified fragments consisted mainly of cattle, caprovid and pig remains (see Table A30). A few dog fragments, probably representing the same individual, and single bones of goose and chicken were also identified.

Caprovid tooth wear stages were recorded using those outlined by Payne (1973; 1987), and those for cattle and pigs followed the scheme of Grant (1982). Cattle, pig and caprovid mandibles were assigned to the general age categories outlined by O’Connor (1989) whilst, in addition, recording of caprovid mandibles and isolated teeth were assigned to the age categories detailed by Payne (1973; 1987).

71

Excavations at the Minster Library, York 1997 Concentrating only on the counts for the three main domesticates, it can be seen that cattle predominate, forming 54 per cent of the assemblage (Table A31). Similarly the ‘large mammal’ component (assumed to represent mainly cattle remains) makes up 60 per cent of the ‘unidentified’ assemblage, in comparison to the ‘medium mammal 1’ fragments, which make up only 16 per cent. Bearing in mind the small size of the assemblage, it is apparent that the proportion of pig remains (26 per cent) is quite high.

cavity and evidence for the removal of the spinus was also apparent on some specimens. It has been suggested (O’Connor 1988; Dobney, Jaques and Irving 1996) that these scapulae represent the remains of cured shoulder joints, possibly brined or smoked. The trimming around the glenoid cavity possibly allowed access for the salt into the meat on the bone. Undertaken correctly, meat salted in this way would be preserved and could be stored for some considerable time (Dobney, Jaques and Irving 1996).

An examination of the skeletal elements represented for the three main domesticates (cattle, pig and caprovid) suggests that these deposits mostly contain domestic waste, with a higher proportion of major (scapulae, humeri, pelves and femora) and minor (radii, ulnae and tibiae) meat-bearing bones to non-meat-bearing elements (head and lower limb fragments).

Cattle pelves were cut and chopped around both the acetabulum and the ilium, and split shafts (mainly radii and humerii) were particularly common. Metapodials tended to be chopped transversely, across the middle of the shaft, rather than split longitudinally. Additionally, chopped shaft fragments formed almost 25 per cent of the ‘large mammal’ fraction. This systematic chopping of shaft fragments has been interpreted in a number of ways, but currently it is thought that the main purpose of this technique was for marrow extraction (Dobney, Jaques and Irving 1996; Stokes in Huntley and Stallibrass 2000). Butchery marks were less evident on the remains of caprovids and pigs and no consistent pattern could be recognised.

Total assemblage from first to second-century deposits Combining the fragment counts of the material from the drain fills and those from the remaining first to secondcentury deposits shows little change in the overall proportions of the major domesticates (Table A32). Additionally, roe deer (Capreolus capreolus L.) and raven (Corvus corax L.) fragments were identified from C1335 and C1244 (see Table A32).

Information regarding the age-at-death of the major domesticates was rather sparse as a result of the scarcity of mandibles. Wear stages from two adult cattle mandibles were recorded, indicating individuals aged approximately 8 years (after O’Connor 1989). Most of post-cranial cattle bones were also from skeletally mature animals.

The range of skeletal elements represented for cattle and caprovids shows a similar preponderance of meat-bearing bones, with a smaller proportion of lower limb fragments (Table A33). Mandibles, teeth and cranial fragments are far less abundant. Taking into account the fragments recorded in the ‘large mammal’ category, it is still clear that meat-bearing elements (pelves, shaft and rib fragments) are more prevalent. Pig remains, again, are mainly represented by meat-bearing elements; however, head fragments are present in similar quantities to lower limb elements.

Evidence available from fusion and tooth wear data for caprovids was very limited. Two mandibles indicated age-at-death of approximately 6-12 months and 2-3 years, while most skeletal elements were from adult individuals. Data for pigs was equally scarce, but evidence from two mandibles and the fusion information suggests that most individuals were immature when slaughtered.

This pattern suggests that the refuse disposed of in these deposits represents both domestic and butchery waste. The absence of large quantities of fragments from cattle and caprovid skulls may indicate that the heads of these animals were being sent elsewhere, probably for the utilisation of their horns. The small quantities of bone recovered by sieving do not contain numerous phalanges, from which it can be inferred that these bones are not simply under-represented but are actually absent. This could be because the skins of the animals were being sent on to other craftsmen with the phalanges still attached. It is also possible that heads may have been left attached to the hides explaining their absence. Extensive butchery was noted on cattle fragments throughout these deposits; the pattern showing many of the distinctive features associated with the processing of cattle carcasses in the Roman period.

Deposits from 12 contexts, of second to third-century date, produced a very small assemblage of 26 identified and 70 unidentified fragments (see Table A31). Preservation of this material did not differ significantly from the earlier assemblage. Numbers of fragments were insufficient to provide useful zooarchaeological data; however, cattle fragments were still the most numerous, with pig and horse remains also present. No caprovid fragments were identified but some ‘medium-sized mammal’ rib and shaft fragments were recorded. One red deer (Cervus elaphus L.) tibia, representing an immature individual, was noted from C1166. Two horse mandibles, one almost complete, were recovered from C1211. These bones almost certainly originate from the same individual. Butchered fragments were noted but showed no consistent patterns, although a single cattle scapula was recorded with a trimmed glenoid.

Of particular interest were the scapulae which were recovered from C1251 (drain fill) and C1244 (spread). A number of these had been trimmed around the glenoid

72

The Roman Fortress A total of 176 fragments (77 of which were identified to species) was recovered from the sixteen deposits dated to the third to fourth century (see Table A31). As with material from previous periods, the preservation was good, with a small component of slightly battered and rounded fragments. The latter suggest the presence of a minor component of residual or reworked material.

large part of the Roman military diet (Davies 1971), the high proportion of pig remains at the present site is, however, rather unusual for military sites in Britain and in York has only been noted from early to mid-Roman bone assemblages at 9 Blake Street (O’Connor 1987). The pattern of butchery and carcass reduction, only really hinted at by this assemblage, has been recorded elsewhere from numerous sites. The General Accident site in York (O’Connor 1988) produced high proportions of cattle scapulae from mid- to late second-century deposits showing similar characteristic damage as that displayed by scapulae from the Minster Library. This practice is not unique to York and has been recorded from first-century deposits in Lincoln (Dobney, Jaques and Irving 1996), second-century deposits at Papcastle, Cumbria (Mainland and Stallibrass 1990) and numerous sites on the continent (Lauwerier 1988). Similarly, the systematic chopping of all major elements and the splitting of most long bones for the extraction of marrow is also a phenomenon recorded from many sites in this region (O’Connor 1988; Hamshaw-Thomas and Jaques 1998), elsewhere in Britain (Dobney, Jaques and Irving 1996; Mainland and Stallibrass 1990) and on the Continent (van Mensch 1974; Lauwerier 1988).

Although numbers were small, cattle fragments were the most numerous remains (52 per cent), with an apparently high proportion of pigs (29 per cent, see Table A33). A small quantity of horse and dog remains was recorded, along with a single red deer antler tine. An examination of the representation of the various skeletal elements for cattle shows a higher proportion of head and lower limb elements than seen in the earlier assemblages (see Table A33). Viewing the large mammal fraction, it is evident that a range of elements was present. However, numbers of fragments are too small for useful interpretation. Butchery techniques similar to those recorded from the early assemblages were also noted for this period. Cattle metapodials, previously transversely chopped, show evidence for being split longitudinally, not unlike the cattle radii from the first to second-century group. Additionally, two horncores (one cow, one caprovid) were clearly deliberately removed from the skull.

Although small, the assemblage of Roman bones from the Minster Library excavation shows a similar range of characteristics to others from Britain and Northern Europe (i.e. cattle dominated, systematic and extensive butchery linked with marrow fat extraction, and cured shoulder joints). However, the high number of pig remains is unusual for Britain and may indicate the presence of high status individuals. These could perhaps be officers based at the fort who, unlike the regular troops (of northern European extraction), may well have originated in the southern Mediterranean where pig meat was favoured.

The paucity of age-at-death data precludes any detailed analysis of the age at which the main domesticates were slaughtered. However, a single elderly cattle mandible was present, while a pig mandible and an isolated tooth indicated the presence of two adult individuals. Material from the fourth to fifth century amounted to only 24 identified and 35 unidentified fragments from a total of 10 contexts. A single human incisor was identified from C1125, which indicates the likely presence of reworked material. Clearly the small size of the assemblage renders the vertebrate material from this period of little interpretative value. Discussion Although the assemblage from this site is small, there are a number of useful points which can be drawn from analysis of the vertebrate remains. The bulk of the assemblage is dated to the first and second centuries AD. Analysis of the vertebrate remains hints at a military supply based on beef, with pork and mutton providing only a small proportion of the diet. This contrasts with evidence from most Iron Age and rural Romano-British assemblages (King 1978; Rackham 1985) which suggest a greater emphasis on caprovids. This evidence fits well with the general pattern of the period and the type of site from which it was recovered. According to a survey of animal bone assemblages from Roman and RomanoBritish sites in Britain (King 1978), a preponderance of cattle appears to be a characteristic of military sites. Although documentary evidence suggests bacon formed a

73

APPENDIX A CATALOGUE OF ROMAN MATERIAL Ceramic Catalogue of Roman pottery Fabric descriptions are given using x20 magnification for unusual examples only, otherwise the common name is used as a heading. Details of site information (context and ceramic phase), the fabric and form codes, diameter and drawing numbers are presented at the end of the descriptions in brackets. CERAMIC GROUP 1 c. AD 71/79-100 [CP 1A-1B] (see Figure 3.10, Nos 1-4, p50) 7 Ebor 1

2

3

A beaker with a simple rounded rim and a high rounded shoulder. It is similar to type KA1 (38733874) which Monaghan dates to CP 2b-3a? (Monaghan 1997, 991 & 994), but has a higher shoulder which is more akin to that of Lyon ware beakers with roughcast decoration - for example 2805, from 9, Blake Street, Period 1, c. AD 71 (Monaghan 1993, 773-4). [C1278, A-ROM1, E1, KA, 7cm, see Figure 3.10, No.1, p38]. A legionary type jar with external sooting around the rim and girth. The form is identical to 2813, from 9, Blake Street, Period 2, c. AD 71/9 -100, which is also sooted, but the fabric is grey (Monaghan 1993, 7756). [C1358, B-ROM1, E1, JA?, 18cm, see Figure 3.10, No.2, p38]. A tazze with concave upper and lower portions with a single frill at rim and midriff. The vessel has been burnt on the exterior around the frills. This is the only example from the site of joins between two areas (A and B). Parallels have been noted at 9 Blake Street 2804, Period 1, c. AD 71 (Monaghan 1993, 773-774) and from York Minster - 17, late first/ early second century (Perrin 1995, 327-8). [C1211, B-ROM5B, and C1266, A-ROM1, E1, TA, 18cm see Figure 3.10, No.3, p38].

8

9

10

11

12

Oise/Somme area of Northern France 13 4

A hooked-rimmed mortarium from the Oise/Somme area of Northern France stamped by the potter Cacumattus, dated AD 65-100 (see above Hartley 1999, no 5). [C1279, A-ROM1, M12, ME, see Figure 3.10, No.4, p38].

CERAMIC GROUP 2 c.AD 75-100/120 [CP 1a - 2a] (see Figure 3.10, Nos 5-13, Figure 3.11, Nos 14-27, p38) 14 Ebor 5

6

A poorly defined, ring-necked flagon similar to 2820, from 9 Blake Street, Period 2, c.AD 71/79-100 (Monaghan 1993, 776-7). [C1282, A-ROM2, E1, FR2, 5cms, see Figure 3.10, No.5, p38] An everted-rimmed flagon with an internal groove with a slight ledge where the neck meets the body. There is a skeuomorphic knob where the strap handle is attached to the rim which appears to have been squashed, and may be a ‘second’. The vessel is very thin-walled and is decorated with burnishing inside

15

74

the rim, vertical burnishing on the neck, and horizontal burnishing on the body wall. There is no exact parallel within the York typology although there is a similar vessel, also with a twisted rim, from 1-9, Micklegate (Queen’s Hotel) 1988.17 (see Monaghan 1997, 997 (FE1)). However, the flagon from YML97 is almost identical to mica-dusted examples from London which are based on bronze flagon forms (Marsh type 3 - Marsh 1978, 138-141). [C1251, AROM2, E1, FE1, 7cm, see Figure 3.10, No.6, p38] A large, probably two-handled, flagon rim similar to 4 from York Minster, Timber Levels (Perrin 1995, 3267). [C1282, A-ROM2, E1, FJ?, 12cm, see Figure 3.10, No.7, p38] An almost complete legionary type jar, but with a variant, almost cornice rim for which no direct parallel has been noted. [C1282, A-ROM2, E1, JA, 12cm, see Figure 3.10, No.8, p38] A more typical legionary type jar with a slight neck delineated by a groove, similar to 2818 from 9, Blake Street, Period 2, c. AD 71/9-100, (Monaghan 1993, 775-6). [C1251, A-ROM2, E1, JA, 13cm, see Figure 3.10, No.9, p38] A virtually complete, legionary style, carinated bowl with a slightly down-turned, reeded rim. This is a relatively common form during this period and there are a number of published examples, a close parallel being 2826 from 9 Blake Street, Period 2, c.AD 71/9100 (Monaghan 1993, 776-7). [C1251, A-ROM2, E1, BC1, 24cm, see Figure 3.10, No.10, p38] A platter with a groove just below the beaded rim. The vessel is finely made perhaps with a jigger and is very similar to samian form Dragendorff 18, and is presumably of a slightly later date-range. [C1251, AROM2, E1, PS, 14cm, see Figure 3.10, No.11, p38] A platter with a slightly everted rim in a fine variant (E3). There is no exact parallel although it is similar to 4082, which is also found in E3 fabric (Monaghan 1997, 1017-8). [C1251, A-ROM2, E3, P, 16cm, see Figure 3.10, No.12, p38]. An almost complete platter without a foot-ring with a distinct return at the top of the body wall and a polished interior. There is a slight deposit on the interior which may have occurred after deposition. The vessel was found in two separate contexts. It is identical in form to 2817 from 9, Blake Street, Period 2, c. AD 71/9-100 (Monaghan 1993, 775-6). [C1251, A-ROM2 and C1244, A-ROM3, E1, PD2, 26cm, see Figure 3.10, No.13, p38]. A plain-rimmed lid with sooting on the rim edge. Monaghan comments that there are no York examples with a convincing date before CP2a, early 2nd century (Monaghan 1997, 1018). The stratification of this example may provide evidence for an earlier date (see Chapter 3, p25). [C1251, A-ROM2, E1, LA, 18cm, see Figure 3.11, No.14, p39]. A virtually complete, tapered, funnel-shaped object, which is hollow, and appears to have been slightly over-fired. The object is wheel-thrown and has been removed from the wheel by a wire. The exterior is slightly encrusted with a deposit which may have occurred after use. The function for this object is

Catalogue of Roman Material

15A

Samian

unknown, but it resembles some kiln-props. It may also have been associated with building material usage. [1251, A-ROM2, E1, Y, 4cm, see Figure 3.11, No.15, p39]. A hooked-rim mortarium with deep, concentric scoring on the interior, of Flavian date (see above, Hartley 1999, no 6). [C1251, A-ROM2, M3, ME, 29.5cm, see Figure 3.11, No.15A, p39].

24

Local Grey (G1) 25 16

17

A legionary style bowl similar to No 10, above, but in local greyware, and with a straight rim. It is identical in form to 10, from York Minster, late first/early second century, Barrack 3, contubernia, (Perrin 1995, 327-8). [C1282, A-ROM2, G1, BC1, 22cm, see Figure 3.11, No.16, p39] A shallow, segmental bowl with a broken rim, which was probably beaded, and a high, stubby flange. No direct parallel has been located within the York assemblages, although it is similar to 3944 (Monaghan 1997, 1002-3). [C1282, A-ROM2, G1, BF, 14cm, see Figure 3.11, No.17, p39].

26

A small, almost complete, South Gaulish samian cup, Dragendorff form 27g, which is unstamped. The cup has a worn footring and graffito scratched on the outside, lower wall, just above the footring. The graffito is crude and appears to be illiterate, but probably denotes a name. The form and fabric suggests a Neronian date. [C1251, A-ROM2, S1, 27G, 8cm, see Figure 3.11, No.24, p39] As No 24, above, but stamped by the potter Masc(u)lus (see Dickinson 1999, Samian Stamp Report, below No 5). The vessel has a worn footring and a fragment of crude graffito scratched on the exterior just above the footring. The stamp is dated to AD 50-65. [C1282, A-ROM2, S1, 27G, 9cm, see Figure 3.11, No.25, p39] A broken, but complete, Dragendorff 36 bowl in a micaceous, early Lezoux fabric, unusually stamped on the interior wall. The vessel has clearly been used and is abraded around the rim and footring, and is dated from the late first to the early second century (see Chapter 3, p50). [C1251, A-ROM2, S3A, 36, 16cm, see Figure 3.11, No.26, p39].

Rusticated 2 Central Gaulish, white colour-coated ware 18

19

20

21

22

An everted-rimmed jar with a slight lid-seating and a high shoulder. It is similar in form to legionary style jars with a slight neck and grooves at the shoulder delineating an area of thick, webbed rustication. This type of rustication is generally earlier in date than linear rustication. [C1282, A-ROM2, R2, JR, 14cm, see Figure 3.11, No.18, p39] Similar to No.18, above, but with an everted rim and lacking the grooves delineating the area of rustication. The vessel is sooted on the exterior, and joins between A-ROM2 and A-ROM3. It is similar to 2828 from 9, Blake Street, Period 2, c. AD 71/9-100 (Monaghan 1993, 776-7). [C1282, A-ROM2 and C1244, A-ROM3, R2, JR, 12cm, Figure 3.11, No.19, p39] As No.19, above with a slight groove on the neck above a trace of evenly applied, thick rustication. [C1282, A-ROM2, R2, JR, 14cm, see Figure 3.11, No.20, p39] A rusticated jar with an everted, grooved rim, and a low, sloping shoulder with a groove at the neck delineating a trace of thick rustication. The vessel is sooted on the exterior. [C1282, A-ROM2, R2, JR, 12cm, see Figure 3.11, No.21, p39] A probable rusticated jar with a lid-seated rim and a groove at the base of the neck above a fine trace of rustication? This example has the typical, dark grey slip on the exterior and over the rim, but the interior is unslipped and pale grey in colour. [C1282, A-ROM2, R2, JR, 14cm, see Figure 3.11, No.22, p39].

27

CERAMIC GROUP 3 c. AD 100-120 [CP 2a] (see Figure 3.12, Nos 28-30, 32-34, p40) Ebor 28

29

Native shell- and sand-tempered ware 23

A small fragment from the top of a volute lamp in a white, Central Gaulish fabric with a thin orange-red slip. The sherd is mould-decorated and may represent a figure, but is too small to be certain. The form suggests a late first to early second-century date. [C1251, A-ROM2, C23?, YL, see Figure 3.11, No.27, p39].

A fragment of a jar or bowl with inturned, undercut rim of ‘native’ tradition. The rim is virtually identical to 2859, from 9, Blake Street, Period 2, c.AD 71/9100, but the fabric is described by Monaghan as possibly South Yorkshire grey ware from Cantley (Monaghan 1993, 779-780). [C1251, A-ROM2, N5, J-B, 22cm, Figure 3.11, No.23, p39].

30

75

A globular, ring-neck flagon with an everted rim and a four-ribbed handle. The surviving sherds are relatively small and fragmented, suggesting redeposition. The top ring of the rim has only a slight groove. The underside of the handle and patches of the exterior of the lower body are dark brown-grey in colour suggesting that the vessel was close to another in the kiln. There are faint, narrow lines of burnishing from the girth to the base. The vessel is very similar to 2820 from 9, Blake Street, Period 2, c.AD 71/9-100 (Monaghan 1993, 776-7). [C1244, A-ROM3, E1, FR2, 8cm, see Figure 3.12, No.28, p40] A very fine, thin walled vessel which survives as only a rim fragment and neck. The fragment is too small to be precise but it may be from a biconical beaker, resembling Camulodunum form 20. The vessel is lightly burnished inside the rim. It is similar to 3922, Monaghan 1997, 997 and 999. [C1244, A-ROM3, E1, K, 10cm, see Figure 3.12, No.29, p40] A thin-walled, fragment of an inturned, globular cup or beaker with a bead-rim and a cordon at the shoulder. There is no direct parallel within the York typology but it is very similar in form to the globular beaker, of Italian origins, in Central Gaulish glazed ware published in Greene, 1979 (fig 41 no 12) which he dates from the Neronian to Flavian period in Britain. [C1244, A-ROM3, E1, C, 10cm, see Figure 3.12, No.30, p40].

Excavations at the Minster Library, York 1997 centurion’s quarter, Perrin 1995, 327-8). [C1226, AROM4, E1, CF, 12cm, see Figure 3.13, No.36, p42].

Samian 31

A very abraded, basal fragment from a dish (Dragendorff form 15/17-18) in South Gaulish samian, with crudely, scratched graffito just above the footring, and of probable Neronian/Flavian date. [C1341, B-ROM3, S1, 15/17-18, see Figure 3.12, No.31, p40].

Local Grey (G1) 37

Local Grey (G1) 32

33

A lid with an up-turned rim which is sooted on the interior. The form is unusual and not obvious within the York typology. It may be a shallow dish used as a lid, the diameter of which suggests that it was used with the carinated bowl described below, from the same context. [C1244, A-ROM3, G1, L, 22cm, see Figure 3.12, No.32, p40] An abraded, legionary style, carinated bowl with a reeded rim, which is slightly down-turned.The rounded, rather than carinated, profile is delineated at the girth by two grooves. It is paralleled in the York typology - 3940 (Monaghan 1997, 1001 and 1003) and at 9 Blake Street, Period 3, c. AD 100-140/60, 2863 (Monaghan 1993, 780). The lid above, No 23, appears to fit this vessel. [C1244, G1, BC1, 24cm, see Figure 3.12, No.33, p40].

Dressel 20 Amphora 38

Ebor 39

A hook-rimmed mortarium with a groove on the edge of the flange, and spout. The vessel is discoloured towards the base suggesting that it was fired inside another vessel. The spout has been attached by keying the rim and then applying the spout. The mixed trituration grits include transparent and milky quartz