Contextual Archaeology of Burial Practice: Case studies from Roman Britain 9781407311968, 9781407322681

Table of contents :
Front Cover
Title Page
Copyright
Table of Contents
List of Figures
Acknowledgements
Preface
CHAPTER 1: INTRODUCTION
CHAPTER 2: DIGGING DEATH: THE DISTRIBUTION OF BURIAL EVIDENCE FROM ROMAN BRITAIN
CHAPTER 3: THE EVIDENCE FOR ROMAN BURIAL RITUALS, WITH SPECIAL REFERENCE TO CREMATION
CHAPTER 4: BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS (VERULAMIUM) AND NEIGHBOURING MINOR CENTRES
CHAPTER 5: BURIAL TRADITIONS AND URBAN SOCIETY IN WINCHESTER (VENTA BELGARUM)
CHAPTER 6: DEATH IN THE COUNTRY
CHAPTER 7: THE GEOGRAPHY OF FUNERARY DISPLAY
CHAPTER 8: THE FORM OF BURIAL DISPLAY: A WIDER CONTEXT
CHAPTER 9: CONCLUSION
CONCORDANCE
APPENDICES
BIBLIOGRAPHY

Citation preview

BAR 588 2013 PEARCE CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE

B A R

Contextual Archaeology of Burial Practice Case studies from Roman Britain

John Pearce

BAR British Series 588 2013

Contextual Archaeology of Burial Practice Case studies from Roman Britain

John Pearce

BAR British Series 588 2013

Published in 2016 by BAR Publishing, Oxford BAR British Series 588 Contextual Archaeology of Burial Practice © J Pearce and the Publisher 2013 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 9781407311968 paperback ISBN 9781407322681 e-format DOI https://doi.org/10.30861/9781407311968 A catalogue record for this book is available from the British Library BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 2013. This present volume is published by BAR Publishing, 2016.

BAR

PUBLISHING BAR titles are available from:

E MAIL P HONE F AX

BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK [email protected] +44 (0)1865 310431 +44 (0)1865 316916 www.barpublishing.com

CONTENTS Table of Contents ...............................................................................................................................................................................ii List of Tables .................................................................................................................................................................................... iv List of Figures .................................................................................................................................................................................... v Acknowledgements .......................................................................................................................................................................... ix Preface ................................................................................................................................................................................................ x CHAPTER ONE Introduction........................................................................................................................................................... 1 CHAPTER TWO Digging death: the distribution of burial evidence from Roman Britain ................................................... 13 CHAPTER THREE The evidence for Roman funerary rituals, with special reference to cremation ..................................... 27 CHAPTER FOUR Burial traditions and urban society in St Albans (Verulamium) and neighbouring minor centres .......... 40 CHAPTER FIVE Burial traditions and urban society in Winchester (Venta Belgarum) ....................................................... 61 CHAPTER SIX Burial traditions and rural society in southern Roman Britain ..................................................................... 79 CHAPTER SEVEN The geography of funerary display: a wider context............................................................................ 111 CHAPTER EIGHT The form of funerary display: a wider context ..................................................................................... 130 CHAPTER NINE Conclusion .............................................................................................................................................. 145 CONCORDANCE ................................................................................................................................................................ 155 APPENDIX ONE Data for figures in Chapter 2................................................................................................................... 164 APPENDIX TWO Descriptions of features discussed in Chapter 3 ..................................................................................... 166 APPENDIX THREE Data for figures in Chapter 4 .............................................................................................................. 176 APPENDIX FOUR Data for figures in Chapter 5 ................................................................................................................ 178 APPENDIX FIVE Data for figures in Chapter 6 .................................................................................................................. 180 APPENDIX SIX Data for figures in Chap. 7; tombs and assemblages discussed in chapters 7 & 8.................................... 195 BIBLIOGRAPHY ................................................................................................................................................................ 213

i

EXTENDED TABLE OF CONTENTS CHAPTER ONE Introduction 1.1 1.2 1.3. 1.3.1 1.3.2 1.4

Introduction and aims A context for the study of funerary rituals: interpreting social and cultural change in the Roman provinces Death and burial in Roman Britain and beyond Current interpretations of burial practice Perspectives for a contextual archaeology of burial The structure of the discussion

1 1 4 4 8 11

CHAPTER TWO Digging death: the distribution of burial evidence from Roman Britain 2.1 2.2 2.3 2.4 2.5 2.6

Introduction Sample compilation and classification The distribution of Roman period burial evidence in Britain Factors influencing the distribution of burial evidence The implications of the distribution of burial evidence Conclusion

13 13 17 22 23 26

CHAPTER THREE The evidence for Roman burial rituals, with special reference to cremation 3.1 3.2 3.3 3.3.1 3.3.2 3.4 3.5 3.6

Introduction Documentary, artistic and archaeological sources of evidence for the cremation pyre The form of the pyre Documentary evidence for pyre terminology and form Archaeological evidence for pyre form Cremation burials and other features in Roman cemeteries The pyre as focus of display in Roman Britain Conclusion

27 27 28 28 29 31 35 39

CHAPTER FOUR Burial traditions and urban society in St Albans (Verulamium) and neighbouring minor centres 4.1 4.2 4.2.1 4.2.2 4.2.3 4.2.4 4.3 4.3.1 4.3.2 4.3.3 4.3.4 4.3.5 4.4. 4.4.1 4.4.2 4.4.3 4.5 4.6 4.7

Introduction St Albans (Verulamium) Iron Age and Roman cemeteries at St Albans Burial practice at King Harry Lane, St Stephens and other cemeteries of the 1st2nd centuries AD The spatial context of burial in late Iron Age and early Roman St Albans Burial and society in late Iron Age and early Roman St Albans Baldock Iron Age and Roman cemeteries at Baldock Iron age burial practice at Baldock Burial practice at the Wallington Road cemetery Burial practice at Walls Field and other Roman period cemeteries The spatial context of burial at Baldock Braughing-Puckeridge Iron Age and Roman cemeteries at Braughing-Puckeridge Burial practice at the Skeleton Green cemetery Burial practice at the Braughing ‘B’ and ‘A’ cemeteries Iron Age and Roman cemeteries at Welwyn Burial and small town society at Baldock, Braughing and Welwyn Conclusion

40 40 40 42 46 47 47 47 49 49 51 52 53 53 54 55 55 57 58

CHAPTER FIVE Burial traditions and urban society in Winchester (Venta Belgarum) 5.1 5.2. 5.2.1 5.2.2 5.3

Introduction Early Roman cemeteries at Winchester Burial practice at the Victoria Road East cemetery Burial and society in early Roman Winchester Late Roman burial practice at Winchester ii

61 63 63 66 67

5.3.1. 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 5.4

The Victoria Road West, Andover Road (Eagle Hotel) and Hyde Street (Late) cemeteries in the northern suburb The Lankhills cemetery in the northern suburb Late Roman cemeteries in the eastern suburb: Chester Road, St Martin’s Close and others Late Roman cemeteries at Oram’s Arbour and the western suburb The spatial context of late Roman burial at Winchester and the ‘managed cemetery’ Burial and society in late Roman Winchester Conclusion

67 70 72 74 74 75 77

CHAPTER SIX Burial traditions and rural society in southern Roman Britain 6.1 6.2 6.2.1 6.2.2 6.2.3 6.2.4 6.2.5 6.3 6.3.1 6.3.2 6.3.3 6.3.4 6.3.5 6.4 6.4.1 6.4.2 6.4.3 6.4.4 6.4.5 6.5

Introduction: current models of social and tenurial distinctions in rural burials Burial practice in rural Roman Hertfordshire The distribution of Iron Age and Roman burial evidence from Hertfordshire Burial practice in early Roman rural Hertfordshire Burial practice in late Roman rural Hertfordshire Burial and settlement space in rural Roman Hertfordshire Burial and society in rural Roman Hertfordshire Burial in rural Roman Hampshire The distribution of Roman burial evidence from Hampshire Burial practice in early Roman Hampshire Burial practice in late Roman rural Hampshire Burial and settlement space in rural Roman Hampshire Burial and society in rural Roman Hampshire Burial and settlement space in rural Roman Britain The cemetery in rural Roman Britain Infants and other burials on settlement interiors Burial and boundaries The context of rural funerary monuments An alternative model for the relationship of burial to settlement space Conclusion

79 80 80 82 83 82 84 86 85 87 91 92 94 95 95 100 102 106 107 109

CHAPTER SEVEN The geography of funerary display: a wider context 7.1 7.2 7.2.1 7.2.2 7.3 7.3.1 7.3.2 7.3.3 7.4 7.5

Introduction Burials on the urban periphery The creation and monumentalisation of urban cemeteries in Roman Britain Streets of tombs in Roman Britain? Funerary display beyond the city Introduction: a framework for analysis Display and the dead I. The civitas of the Catuvellauni Display and the dead II. Hampshire and the civitas of the Belgae The context of burial display – settlement type and civitas location Conclusion

111 111 111 115 118 118 119 123 125 128

CHAPTER EIGHT The form of funerary display: a wider context 8.1 8.2.1 8.2.2 8.3.1 8.3.2 8.4

Introduction The composition of early Roman burial assemblages ‘porticus et balnea et conviviorum elegantia’: the interpretation of early Roman burial display The composition of late Roman burial assemblages ‘cingulum ponere’: the interpretation of late Roman burial display Conclusion

130 130 136 140 142 143

CHAPTER 9 CONCLUSION 9.1 9.2 9.3.1

Introduction Burial evidence – distribution and character Differentiating the dead

145 147 147 iii

9.3.2 9.4

Space and society: burial context Last words

150 153

CONCORDANCE. Concordance to Iron Age and Roman cemeteries and burials from Britain referenced in text and appendices

155

APPENDICES APPENDIX 1. Data for figures associated with chapter 2 1

Data for figures 2.1-2.40

164

APPENDIX 2. Descriptions of features discussed in chapter 3 2.1 2.1.1 2.1.2 2.1.3 2.1.4 2.2

Pyre sites from Britain and other provinces ‘Permanent’ pyre sites from Roman Britain ‘Permanent’ pyre sites in other Roman provinces ‘Temporary’ pyre sites from Iron Age and Roman Britain Temporary pyre sites from other provinces A sample of busta

166 166 167 168 173 174

APPENDIX 3. Data for figures in chapter 4 3.1 3.2

Note on unpublished cemeteries from Baldock and St Albans Data for figures 4.6-4.8, 4.11-13, 4.18-4.21, 4.24-4.28

176 176

APPENDIX 4. Data for figures in chapter 5 4.1 4.2

Note on unpublished cemeteries from Winchester, with particular reference to late Roman period Data for figures 5.4-5.7, 5.10-12, 5.15-5.17

178 179

APPENDIX 5. Data for figures in chapter 6 5.1 5.2

Data for figures 6.9-6.12 Descriptions of the relationship of burial to settlement features from a sample of rural sites in southern Britain discussed in Chapter 6.2-4

180 180

APPENDIX 6. Data for figures in chapter 7 and burial assemblages discussed in chapters 7 and 8 6.1 6.2 6.3 6.4 6.5 6.6

Data for figure 7.1 Burial assemblages from the civitas of the Catuvellauni and environs Other burial assemblages from southern Roman Britain Burial monuments from the civitas of the Catuvellauni and environs Burial assemblages of the ‘East Hampshire Tradition’ Stone, lead-lined and large timber coffins and burial monuments in late Roman Hampshire

BIBLIOGRAPHY. Abbreviations and References

195 195 202 205 209 212 213

LIST OF TABLES Table 2. 1 Table 3.1 Table 8.1

Buried populations and ‘real’ populations compared The identification of different features related to the Roman period burial process Samian ‘services’ in early Roman burial assemblages from Britain

iv

24 33 133

LIST OF FIGURES CHAPTER 2 Figure 2.1 Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6 Figure 2.7 Figure 2.8 Figure 2.9 Figure 2.10 Figure 2.11 Figure 2.12 Figure 2.13 Figure 2.14 Figure 2.15 Figure 2.16 Figure 2.17 Figure 2.18 Figure 2.19 Figure 2.20 Figure 2.21

The number of burial / cemetery records from five counties in different data sources (SMR, JRS/ Britannia, Philpott) Dated and undated burial / cemetery records in different data sources (SMR, JRS/ Britannia, Philpott) Dated cemeteries from three counties (Hertfordshire, Hampshire and Cambridgeshire) in different data sources (SMR, JRS/ Britannia, Philpott) Cemetery types from three counties (Hertfordshire, Hampshire and Cambridgeshire) in different data sources (SMR, JRS/ Britannia, Philpott) Burial / cemetery excavations recorded from 1920 to 2010 in JRS / Britannia Percentage of burial / cemetery types recorded from 1920 to 2010 (JRS / Britannia) The percentage of dated cemeteries in different data sources (JRS, Britannia, Philpott) The percentage of dated burials in different data sources (JRS / Britannia) The percentage of cemeteries of different types in different data sources (JRS, Britannia, Philpott) The percentage of burials in different types of cemeteries (JRS / Britannia) The percentage through time of different cemetery types (Britannia) The percentage through time of different burial types (Britannia simplified) The number of cemeteries from different settlement types (JRS/Britannia) The number of burials from different settlement types (JRS/Britannia) The percentage of dated cemeteries at different settlement sites (JRS / Britannia) The percentage of dated burials at different settlement sites (JRS / Britannia) The percentage of cemetery types from different settlements (JRS/ Britannia) The percentage of burial types from different settlements (JRS / Britannia) The distribution of Roman period cemeteries by county (JRS / Britannia) The distribution of Roman period cremation cemeteries (JRS / Britannia) The distribution of Roman period inhumation cemeteries (JRS / Britannia)

14

Verulamium and neighbouring minor centres (dashed line shows possible civitas boundary) The distribution of Roman period burials, St Albans (after Niblett 2000: 97, Fig. 10.1) Late Iron Age and early Roman burials at King Harry Lane (adapted from Stead and Rigby 1989) The Folly Lane burial complex and environs to south (after Niblett 1999: 15, Fig. 8) Roman burials on Watling Street excavated in the 1930s and 1980s excavations at St Stephens, St Albans (adapted from Niblett, St Albans Museums Field Archaeology Unit archive) The quantities (g) of cremated bone in burials from a sample of RomanoBritish cemeteries Numbers of different ceramic forms, St Stephens The number of burials with different numbers of ceramic accessory

41

14 14 14 17 18 19 19 19 19 19 19 20 20 20 20 20 20 21 21 21

CHAPTER 4 Figure 4.1 Figure 4.2 Figure 4.3 Figure 4.4 Figure 4.5 Figure 4.6 Figure 4.7 Figure 4.8

v

41 42 43 44 45 45 45

Figure 4.9 Figure 4.10 Figure 4.11 Figure 4.12 Figure 4.13 Figure 4.14 Figure 4.15 Figure 4.16 Figure 4.17 Figure 4.18 Figure 4.19 Figure 4.20 Figure 4.21 Figure 4.22 Figure 4.23 Figure 4.24 Figure 4.25 Figure 4.26 Figure 4.27 Figure 4.28

vessels, St Stephens The distribution of LPRIA and Roman burials, Baldock (adapted from Burnham and Wacher 1990: 284, fig. 96) The Wallington Road cemetery, Baldock (after North Herts Archaeological Service archive) Numbers of different ceramic forms, Wallington Road The number of burials with different numbers of ceramic accessory vessels, Wallington Road The number of burials with different numbers of ceramic accessory vessels, Walls Field Burials and pyre areas at Royston Road (Area 15), Baldock (adapted from Frere 1989: 299, fig. 22) Roman period Braughing with cemeteries (adapted from Burnham and Wacher 1990: 104-5, figs 27-28) The Skeleton Green cemetery, Braughing (after Partridge 1981: 246, Fig. 90, 246) Braughing B cemetery (after Partridge 1977: 74, Fig. 29) Numbers of different ceramic forms, Skeleton Green The number of burials with different numbers of ceramic accessory vessels, Skeleton Green Numbers of different ceramic forms, Braughing B The number of burials with different numbers of ceramic accessory vessels, Braughing B Iron Age and Roman period burials from Welwyn (adapted from Rook 1986: 107, Fig. 28) Numbers of different ceramic forms, Welwyn The number of burials with different numbers of ceramic accessory vessels, Welwyn The percentage of burials with ceramic accessory vessels from urban and minor centres The average number of ceramic accessory vessels, burials from urban and minor centres Samian as a percentage of ceramic accessory vessels, burials from urban and minor centres The percentage of burials with glass accessory vessels, urban and minor centres

48 50 51 51 51 52 53 53 54 54 55 55 55 56 57 57 58 58 58 58

CHAPTER 5 Figure 5.1 Figure 5.2 Figure 5.3a Figure 5.3b Figure 5.4 Figure 5.5 Figure 5.6 Figure 5.7 Figure 5.8 Figure 5.9 Figure 5.10 Figure 5.11 Figure 5.12 Figure 5.13 Figure 5.14 Figure 5.15

The cemeteries of Roman Winchester (adapted from WMS archive) The northern cemetery area at Winchester (adapted from WMS archive) The Victoria Road East cemetery from the first to early third centuries (adapted from Ottaway 1992: 77, Fig. 3.9 and WMS archive) Detail of Victoria Road East cemetery, mausoleum and four post monuments in Trench X (adapted from Ottaway et al. 2012: 61, Fig. 24) Cremated and inhumed populations, Victoria Road East Numbers of different ceramic forms, Victoria Road East The number of burials with different numbers of ceramic accessory vessels, Victoria Road East The number of burials with different artefact types, Victoria Road East The Victoria Road West cemetery from the third to early fifth centuries (adapted from WMS archive) The Lankhills cemetery, Clarke and Oxford Archaeology excavations (after Booth et al. 2010: 43, Fig. 2.30) The number of burials with different artefact types, Victoria Road West The number of burials with different artefact types, Lankhills (Clarke) Average NAT of burials with different attributes, Lankhills (Clarke) Chester Road phases of burial (adapted from Ottaway et al. 2012: 18081, fig. 84; WMS Archive) Monument F57 and associated burials, St Martin’s Close (after Morris 1986: 344, Fig. 4) The number of burials with different artefact types, Chester Road vi

61 62 63 64 65 65 65 67 69 69 70 70 70 73 73 74

Figure 5.16

The proportion of burials with grave goods in late Roman cemeteries at Winchester The average NAT per burial in late Roman cemeteries at Winchester

77

The distribution of LPRIA period burials in Hertfordshire The distribution of Roman period burials in Hertfordshire A rural settlement at Boxfield farm: enclosures and the early Roman cremation cemetery (after Going and Hunn 1999: 8, fig. 4 and 30, fig. 15) Settlement and burial in Area 1, Foxholes Farm (after Partridge 1989: Fig. 4, facing 22) Wood Lane End mausolea and associated features (after Neal 1984: Fig. 1, 194) The distribution of Roman period burials in Hampshire Cemeteries and burials at Owslebury (adapted from Collis 1970: 249, Fig. 1, bd.)

81 81 84

Figure 6.8a

The Late Iron Age and early Roman cemetery, Owslebury (adapted from Collis 1977: 29, Fig. 5)

88

Figure 6.8b Figure 6.9 Figure 6.10 Figure 6.11 Figure 6.12 Figure 6.13

Burials on Site P at Owslebury (after Collis 1968: 24, Fig. 3) The number of ceramic forms at Owslebury The orientation of burials at Owslebury The orientation of late Roman rural inhumation burials in Hampshire The number of late Roman rural burials with different artefact types, Hampshire The distribution of late Roman burials [squares] and human skeletal fragments (open triangles) at Balksbury (adapted from Wainwright and Davies 1995: Fig. 8, 8) The distribution of burials at Burntwood Farm, Site R6 (adapted from Fasham 1980: 42, Fig. 4) The conquest period burial, associated pit and late Roman structure, Thruxton (after Cunliffe and Poole, 2008d: 39, fig. 4.23) The distribution of burials in relation to principal Iron Age and Roman features at Frocester Court (after Price, archive)

89 90 90 91 91 92

Figure. 6.17

Roman period settlement and burials at Alington Avenue (after Davies et al. 2002: 57, Fig. 24)

96

Figure 6.18

Settlement and burials of all periods at Roughground Farm (from Allen et al. 1993: Fig. 69, 98) Fourth century settlement at Catsgore, with burials added (adapted from Leech 1982: Fig. 5, 8)

96

Figure 6.20

Duckend Farm and Duckend Car Park sites, Stansted. The distribution of the principal clusters of cremation burials (boxed) in multi-period palimpsest (adapted from Havis and Brooks 2004: 263, Fig 169)

97

Figure 6.21

Mid-first to mid-second century features from Area B, Wavendon Gate (adapted from Williams et al. 1995: 26, Fig. 16) Settlement, cemetery and shrine at Bancroft, (A) 1st century BC to 1st century AD, (B) 1st to 2nd century AD, (C) mid 4th-early 5th century AD (after Williams and Zeepvat 1994: Fig. 5, facing p.6, Fig. 6, facing p.8) Iron Age, Roman and Anglo-Saxon burials at Area 2, Westhampnett (after Fitzpatrick 1997a: Fig. 6) Burials and shafts at Keston in relation to Period VI settlement (adapted from Philp et al. 1999: 184, 191, 195, Figs. 67, 69-70) Period VI north cemetery at Keston (after Philp et al. 1999: 46, Fig. 21) Settlement and burials at Bradley Hill (infant cemetery in building 3) (adapted from Leech 1981: Fig. 2, 179) The excavated area, courtyard enclosure with burials and cemetery at Lynch Farm (adapted from Jones 1975: Figs 3 and 4, 97 and 99) Late Roman settlement and cemetery at Stanton Low (adapted from Woodfield and Johnson 1989: Figs. 4-5, 140-41)

98

Figure 5.17

77

CHAPTER 6 Figure 6.1 Figure 6.2 Figure 6.3 Figure 6.4 Figure 6.5 Figure 6.6 Figure 6.7

Figure 6.14 Figure 6.15 Figure 6.16

Figure 6.19

Figure 6.22 Figure 6.23 Figure 6.24 Figure 6.25 Figure 6.26 Figure 6.27 Figure 6.28 Figure 6.29

Late Roman cemetery at Claydon Pike, setting and cemetery by trackway (adapted from Miles 2007: 170, 185, Figs 6.1 and 6.13)

84 85 86 88

93 93 95

97

99 100 101 101 103 104 105 105

CHAPTER 7 Figure 7.1

Numbers of funerary inscriptions from the towns of Roman Britain (RIB vii

112

Figure 7.2 Figure 7.3 Figure 7.4 Figure 7.5 Figure 7.6 Figure 7.7 Figure 7.8

I and III, Britannia to 2011) Figure 7.2 Funerary monuments and associated burials, Great Dover St, London late 1st-2nd century AD (adapted from Mackinder 2000: 11 and 15 figs 8 and 12) Funerary monuments at the Derby Racecourse cemetery (adapted from Wheeler 1985: Fig. 98, 223) Burial assemblages from the civitas of the Catuvellauni and neighbouring areas with six or more accessory ceramic vessels Burial assemblages from the civitas of the Catuvellauni and neighbouring areas with three or more artefact types as grave furniture The distribution of first to third century funerary monuments within the civitas of the Catuvellauni The distribution of burials in the East Hampshire Tradition The distribution of stone and lead coffins and monumental burials in late Roman Hampshire

116 116 121 122 122 124 125

CHAPTER 8 Figure 8.1 Figure 8.2 Figure 8.2a Figure 8.2b Figure 8.2c Figure 8.2d

Cremation burial at Avisford (from Roach Smith 1848: Pl. XLIV) Plans of 1st-2nd century AD cremation burials approximately to same scale (for descriptions see Appendix 6.2 and 6.6) Burial 2, Winchester Grange Road (after Biddle 1967: 232, Fig. 5) St Albans ‘Tripod burial’ (after Niblett and Reeves 1990: 445, Fig. 3) Burials 7 Clothall Rd Baldock (after Stead and Rigby 1986: 72, fig. 32) Burial 566 Winchester Victoria Road (after Ottaway et al. 2012: 261, fig. 102)

viii

134 135 135 135 135 135

Acknowledgements Numerous debts must be acknowledged here, both for assistance with the writing of this book and for living with its production and that of the thesis on which it is based. That thesis was written at the University of Durham and was supported by a British Academy studentship. I am grateful to the following for discussing and providing information from unpublished excavation data and other publications, Sites and Monuments Records (now ‘Historic Environment Records’), and otherwise unobtainable references during the writing of the thesis and subsequently: Trevor Ashwin (Norfolk Archaeological Unit), Bruno Barber (Museum of London), Wendy Barrett (Southampton City Council), Justine Bayley (Ancient Monuments Laboratory), Ina Bauer, Rosemary Braithwaite (formerly Hampshire County Council), David Bowsher (Museum of London), Stuart Bryant (Hertfordshire County Council), Gil Burleigh (formerly North Hertfordshire District Council), Sue Byrne (Gloucester Museums), Maureen Carroll (Sheffield University), Edith Evans (Gwent and Glamorgan Archaeological Trust), Andrew Fitzpatrick (formerly Wessex Archaeology), Keith Fitzpatrick-Matthews (North Hertfordshire District Council) Susan Fox (Roman Legionary Museum, Caerleon), Patrick Garrod (Gloucester Museums), Malcolm Gomersall (formerly Winchester Historic Museums Service), Frances Griffiths (formerly Devon County Council), Christine Hamlin (University of WisconsinMilwaukee), Stuart Hartgrove (Cornwall County Council), Valerie Hope (Open University), Bruce Howard (formerly Hampshire County Council), Fachna McAvoy (Central Excavation Unit), Paul McCulloch (Wessex Archaeology), Tom McDonald (Hertfordshire Archaeological Trust), Jackie McKinley (Wessex Archaeology), Quita Mould, Ros Niblett (formerly St Albans Museums), Steve Parry (Northamptonshire Archaeology), Robert Philpott (National Museums Liverpool), Edward Price (Frocester Court), Ken Qualmann (formerly Winchester Historic Museums Service), Val Rigby (formerly British Museum), Mark Stevenson (formerly North Hertfordshire District Council), Dale Trimble (formerly Historic Lincolnshire, now Archaeological Project Services), Colin Wallace, Simon West (St Albans Museums) and Andrew Westman (Museum of London). The Digital Atlas of Roman and Medieval Civilizations (Harvard University) is gratefully acknowledged as the source of the Roman road data used in figures 6.2, 6.6, 7.4-7.8 (http://dighist.fas.harvard.edu/content/darmc-digital-atlasroman-and-medieval-civilization). I would like to acknowledge the enduring support and advice of my supervisor Martin Millett and thank him for his constructive criticism of the thesis and my examiners, Greg Woolf and Colin Haselgrove for comments, advice on publication and subsequent encouragement. To Manuela Struck and Jake Weekes for reading earlier drafts of papers related to the thesis and for supplying many references, to Jenny Price, Simon James and Brian Dobson for comments on individual thesis chapters, I also owe thanks, as I do to Paul Booth, Rebecca Redfern and Hella Eckardt for later advice and references. Former research students at the Department of Archaeology, Durham University, provided comments, practical help and encouragement (Rebecca Gowland, Nic Holland, Ardle MacMahon, Louise Revell, René Rodgers, Jeremy Taylor, Steve Willis, Hugh Willmott). Sally Worrell (4.2), Hugh Willmott (6.8b, 6.13) drew individual figures and Katie Robbins created maps (fig. 6.1, 6.2, 6.6, 7.4-7.8). To her and Kosmas Dafas I’m especially grateful for invaluable help during final preparation for publication. Alan Bowman and David Thomas generously encouraged my continued development of research interests in Roman funerary archaeology and beyond whilst working as their research assistant on the Vindolanda tablets project and colleagues. Periods of research leave at King’s College London have allowed me to work on updating the project. At KCL too I thank Henrik Mouritsen in particular for discussion of funerary matters. Other commitments have delayed its publication longer than I would have hoped and David Davison and colleagues at BAR are sincerely thanked for their patience and tolerance of deferred submission. These acknowledgements also give an opportunity to recognise the bequest from my grandmother, the late Kathleen May Hunter, which cushioned the limbo of the ‘writing-up’ year, and to whose memory this book is dedicated. To my parents, Richard and Primrose Pearce, much more is due than conventional filial gratitude for moral and practical support and encouragement. Sadly my mother did not live to see this book, though it would have caused her no small relief to have seen it published. Sally Worrell bore the brunt of the writing of the thesis and its later re-working, and I thank her in particular for her encouragement, tolerance, and for not taking it too seriously.

ix

Preface This publication is the revised version of my doctoral thesis, Contextual archaeology of burial practice: case studies from Roman Britain, which was submitted in January 1999 at the University of Durham and passed in April of the same year. I have updated the thesis by taking account of work that has taken place since its completion, which proved a substantial task not least because of the welcome growth in Roman funerary archaeology as a field of study. I have updated references to those sites from which I used data when unpublished which are now published and I have specifically aimed to incorporate, as well as respond to, the results of new work, in particular in the following areas: (i) to update the survey of cemetery fieldwork reported in JRS and Britannia that forms the basis of discussion in chapter 2; (ii) to include further published sites from my sample areas (chapters 4-6); (iii) to consider cemeteries relevant to interpreting the relationship between tombs and their landscape context (chapters 6 and 7); (iv) to use recently published assemblages relevant to interpreting the composition of burial assemblages in chapter 8; (v) and in chapters 7 and 8 to include some limited further references to recent publication or excavation of continental European cemeteries, though with no pretence at systematic coverage; In preparing the thesis for publication revisions have generally been incorporated within the original structure. On the advice of my examiners I have reduced the discussion of epigraphic evidence for the process of cremation in chapter 3, in order to prepare this as a separate article. Chapters 4 to 6 have also been re-organised with a view to simplifying their organisation. Again on the advice of my examiners I have also included in chapter 4 revised material from my MA thesis on burial practice at St Albans which was a pilot study for the doctoral thesis. Introduction and conclusion have been rewritten to allow for the other changes made. The illustrations have been reduced in number, especially for chapter 6. A note on references to individual cemetery excavations from Britain In order to avoid repetition of references, sites with burials from late Iron Age and Roman Britain are referenced by name only in the main body of the text. Bibliographical references to these sites are given in the Concordance, listed in alphabetical order under the names used in the text. Where burials derive from the cemeteries of a particular town their name is usually prefixed by the name of that town, for example ‘Winchester Hyde Street’. In the Concordance the reader is also directed to the relevant appendices if more detailed information from a particular site has been exploited. Cross-referencing between chapters is by the chapter and section number (e.g. 3.6.5), to appendices by prefixing the number with ‘appendix’ (e.g. Appendix 5.1.1). The data which support the figures as well as some arguments in the main text are gathered in the appendices.

x

CHAPTER 1: INTRODUCTION 1.1 Introduction and aims The dead were a recurring presence in the lives of Romano-Britons, not only through funerals and commemorative ceremonies. Cemeteries and monuments lined the approaches to towns or saluted the rural passerby, while graves were also dispersed across the landscape in half-remembered association with hedges, banks and ditches around farms or fields. Yet they, and the rituals which saw them to the grave and beyond, have not intruded with equal force into the archaeological study of Roman Britain. This in contrast to the central place of burial in the archaeology of many other periods, including those which frame the Roman era in Britain, the late pre-Roman Iron Age and the early Anglo-Saxon period. The aim of the research presented here is therefore to propose routes for integrating the study of death rituals into more general debate. Rather than analysing burial data, in particular assemblages of grave goods, in isolation, it advocates a shift towards study of burial in context, considering interment in relation to other aspects of funerary ritual and examining it in tandem with other types of archaeological evidence. This, it is argued, will help to achieve that integration and to make a more prominent contribution to the interpretation of provincial Roman society. The focus of this study lies primarily on one Roman province but it considers this province in the light of wider research traditions in the mortuary archaeology of the Roman world.

specific themes to be pursued, and how they are organised through the book, are then discussed (1.4). 1.2 A context for the study of funerary rituals: interpreting social and cultural change in the Roman provinces In order to provide useful insights into British and other provincial Roman societies the analysis of burial practice must be situated within the broader study of social and cultural change. To this end, recent developments in the interpretation of that broader change are briefly reviewed and their implications for the study of burial practice explored. The last three decades have witnessed the explicit overturning of approaches to an archaeology of Roman Britain centred on political and military history. The principal preoccupation of archaeological study of the Roman period in Britain and also other north-western provinces, especially of the early Roman period, is with the nature of the cultural change formerly subsumed under the term ‘Romanization’ and the contribution of different agents to that change. Views of an interventionist imperial policy to promote or enforce the adoption of Roman culture, with the army as a potential instrument of such a policy, have been discarded. The explanation for the adoption of Roman culture because of its innate ‘civilised’ superiority has been discredited for its identification with nineteenth and twentieth century British / western imperial values, although the extent of this identification is debated (Freeman 1996; 1997; 2007; Hingley 1996a; 2000). Instead the focus has now shifted to alternative agents and mechanisms.

Among the different possible meanings of a contextual archaeology (Hodder and Hutson 2003), for present purposes it is taken to mean the archaeology of a social practice, burial, studied within its particular social and cultural context. The proposition that death and burial should be interpreted in relation to their social context is not of course specific to a post-processual perspective. It was one of the major achievements of the New Archaeology to demonstrate the potential insights into social structure that could be derived from study of burial practice, on the general principle that variation in burial form was related to the organisation of the burying society rather than solely to differences in afterlife beliefs or ethnic affiliation (Binford 1972b; Chapman et al. 1981). However the contextual position is this, that different treatment at burial does not directly reflect the position of an individual within the burying society but represents a transformation of it, dependent on contextual attitudes to the dead (Parker Pearson 1999a). This study therefore examines burial practice in Roman Britain from this perspective. In this first chapter the broader context of the study of Roman Britain to which the analysis of burial practice may contribute is identified (1.2). The frameworks for interpreting Roman period burial practice in Britain as well as other Roman provinces, in particular in relation to cultural change and social structures, are then evaluated in the light of wider developments in the archaeological interpretation of mortuary ritual (1.3). The

Particular emphasis has been given to the co-opting of local elites on whom the structure of the empire and, with particular relevance here, the diffusion of a Romanised lifestyle in Britain and elsewhere have been argued to depend (e.g. Alcock 1994; Brandt and Slofstra 1983; Brunt 1976; Macmullen 1990; Millett 1990a; 1990b; Roymans 1996a; 1996b; Roymans and Derks 2011; Terrenato 2005; Woolf 1998). This argument emphasises the active, strategic adoption of Roman material culture and associated behaviour by nascent provincial elites exploiting an opportunity to reinforce their position within local social hierarchies. In this argument, variation in the intensity of this process depends on the structural properties of local societies. For Britain Millett (1990a; 1990b) has argued that the lack of social competition consequent on the ‘closed’ character of elites, i.e. the continued dominance of the same high status groups from the Late Iron Age (late second century BC to AD 43) into the Roman period, explains the limited adoption of the epigraphic habit or of euergetism expressed in public buildings compared to other provinces. In the developing discussion modifications and criticisms of such models have been proposed. In particular it has

1

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

power without sufficient consideration of the ideological context of their adoption, with ideology defined as;

classified as the inferior ‘Other’ (Said 1978). The use of the term ‘Romanization’ to describe or explain cultural change is argued to perpetuate such an ‘imperial’ system of values for the evaluation of archaeological evidence, implying superiority of Roman over native cultures and a one-way transfer of culture (Mattingly 1997a; Webster 1996a).

‘…a system of ideas, values and norms which structure human behaviour and thought. In this sense ideology cannot be isolated from the religious, social, political, or economic aspects of Romanization, for it is central to them all.’ (Millett et al. 1995: 2)

This perspective has had various consequences. The response of provincial elites to Roman power has been reconsidered with greater emphasis on the interaction with local value systems. Woolf (1995) has argued that here the task is to define context-specific areas of social practice around which cultural identities cohered. These are more easily identifiable in parts of the eastern empire, for example language for Greeks or diet for Jews, but the almost total dependence on archaeological evidence makes their identification more difficult in the western empire and as yet rarely attempted. One exception is Roymans’ (1996b) exploration of a value system based on martiality / pastoralism in the lower Rhine area that endures into the Roman period through the particular heavy recruitment of Batavian auxiliary units in the Roman army. However as Haselgrove (1987a: 116) envisaged, by tribute or plunder conquest could remove traditional accoutrements of authority, leaving a tabula rasa where Roman connections were perhaps the symbolic quarry to be exploited by default.

This direction is congruent with a more general recent emphasis in archaeology, to seek explanations centred on human agency created in particular historical contexts. Against the traditional Marxist characterisation as false consciousness, ideology has been defined as a lived perspective on the world that not only suppresses but also creates subjects; giving power to as well as power over, although as Treherne argues, ideology is not simply reducible to a general world view (McGuire and Paynter 1991b: 5-7; Miller and Tilley 1984b: 7-8; Treherne 1995). A reconsideration of the ideological domain has stimulated re-evaluation of the apparent laissez faire character of the ‘Roman’ part of this cultural exchange. For example it has proved difficult to account for the contemporaneous development of very similar urban forms across the western provinces, regardless of their history, by independent acts of emulation (Lomas 1997; Millett 1990b: 40). Woolf (1995; 1998) has instead argued for the generation of such similarities by an indirect dissemination of values from the centre of the empire through two processes: the systematisation of the institutions of imperial government, necessitated by the expansion of empire, and by the reformulation of an elite Roman identity in the late Republic through a new ethic of civilised conduct, designated as humanitas. The values were internalised by local elites, re-fashioning and representing themselves through a Roman-style education and through commissioning monumental public architecture furnished with Roman iconography and inscriptions in newly created urban centres (Revell 2009), though other areas of social practice and their material components have yet to be so intensively investigated. In this argument Gallic elites had to acquire a ‘cultural competence’ compatible with humanitas, in order to allow the sort of interaction through which relationships of friendship and patronage with high-ranking Romans, for example governors, could work (Woolf 1998: 63). In the literary sources provincials can be observed developing their relationships with metropolitan elites by adoption of their behaviour, as Brunt put it (1976: 169):

The occurrence of many categories of ‘Romanised’ material culture has been documented across what must have been quite wide variations in social status, for example from Britain the widespread distribution of terra sigillata in Roman Britain, or the use of mortared masonry with associated decorative media for houses in town and country. However the mechanisms behind this diffusion have been much less extensively investigated. Whittaker (1997: 155) considers that direct Romanisation acts at an elite level whilst that of the poor happened through ‘osmosis’. Millett has proposed the dispersal of Romanised material culture as the more active local emulation of elite behaviour: ‘…progressive emulation of this symbolism down the social hierarchy was self-generating, encouraging others to aspire to things Roman, thereby spreading the culture’ (Millett 1990b: 38). To what degree the distribution of given attributes satisfies the conditions proposed by Miller (1982), for example of differences in their chronological horizon in different social milieu, or the restriction on resources used in emulation, has not yet been fully worked out as Freeman (1993) has indicated and in well-known cases, for example the distribution of terra sigillata, shows patterns that would not be straightforwardly predicted by such a model (Willis 1997; 2004). More problematically ‘emulation’ is only one possible reason for which new behaviour or materials may be adopted. To interpret it as such is to follow what is often an elite reading of the changing practices of non-elite groups, rather than to

‘At every stage in Roman history the aristocrats who ruled at Rome found it most natural to support men like themselves elsewhere.’ Another related critique comes from post-colonial examination of imperial encounters, which has shown the dependence of imperialism on central cultural values, in relation to which other value systems encountered are 2

INTRODUCTION

consider its specific contextual importance (Campbell 1993).

cultural and social process in discussion (James 2001: 199-202). As Hill (2001: 15-16) and Pitts (2007: 697-98) have noted, it has distracted attention from other dimensions of identity, including class and gender. After a ‘formative period’ of cultural encounters in the decades immediately around the conquest of Roman Gaul, Woolf (1998) suggests that ‘Roman’ culture came to differentiate within Gallic society more than it demonstrated cultural affiliation. Revell (2009) takes a similar perspective on urban public life of the midimperial period. She explores urban public buildings not as abstract symbol of affiliation to a Roman cultural model but rather as spaces which served as a medium for creating identities, especially those of aristocratic males through enabling certain socio-political activities (such as public speaking, bathing, and watching spectacles etc.). With the shift to a perspective that sees identities as continuously constituted rather than fixed and essential then these other dimensions come more clearly into view. Among other things this responds to Barrett’s (1997b) advocacy of a reading of archaeological evidence from the ‘bottom-up’ perspective of everyday practice, in order to explore how the routines of living and their associated material culture brought identities into being, rather than ‘top-down’ abstract processes such as Romanization. The fullest application of such an approach, Gardner’s (2007) examination of identities in late Roman Britain, widens the focus beyond elite male identities constituted through urban living.

It is in this area that post-colonial theory has perhaps had greatest impact for the study of the Roman empire, in seeking to identify contestation of Roman political or cultural authority. On analogy with the colonies of postmedieval European empires, the existence of resisters and resistant value systems to those of Rome has been advocated, within an imperial context of which the oppressive and exploitative characteristics have been emphasised. Attention has been drawn to the existence of alternative and resistant value systems differentiating Roman and non-Roman, elite and non-elite (Mattingly 1997a; 2006; 2011; Webster 1996). Resistance has been identified in both violent rebellion and in the ‘discrepant experience’ of everyday practice, especially among nonelites, as ‘hidden transcripts’ by which subordinate populations acquiesce to elite power at points of contact but find greater freedom in alternative arenas. In Roman Britain these have been located in the private sphere, in the home and in aspects of ritual (e.g. Hingley 1997; Mattingly 2004; Webster 1997a). Most fruitfully perhaps some post-colonial perspectives, informed by the broader changing context for the interpretation of material culture, stress the ambivalence of ‘Roman’ material culture and its polyvalency dependent on context. Webster (2001) has argued that ‘creolisation’ may be an appropriate term to describe the potential, as in much later colonial settings, for Roman material culture to be given new meanings through use in unfamiliar contexts and in new ways or combinations. Indeed this ambivalence is already present in ‘Roman’ culture itself, which is arguably constructed dialectically (Millett 1990a: 1). Terrenato (1998; 2005) argues for the notion of bricolage to be helpful here, i.e. the knowing adoption of selected aspects of the ‘Roman’ package to create new form and practices. Like others (e.g. MacMullen 1990) he also emphasises the fault lines between elite and non-elite groups in the adoption of Roman culture, proposing structured oppositions in spheres affected to greater or lesser degrees by contact with Roman power; so public male elite life is conditioned to a much greater degree than non-elites and the private sphere (in which he included burial practice):

What then is the significance of these developments for the study of burial? As others have noted (and as will be further discussed in the next section), in recent debate over cultural change some important areas of evidence have received much less attention than others, especially those which traditionally fell under the heading of ‘lifeways’, such as diet, costume, art, language and literacy and those pertaining to the sphere of ideology as defined in Hawkes’ (1954) ‘ladder of inference’, including religion and burial practice (Freeman 1993; Millett et al. 1995: 3; Woolf 1992). Burial practice in Britain and beyond has not seen the re-examination and re-theorisation to the same degree as the study of provincial religion (e.g. Derks 1998; van Andringa 2002; Webster 1997a). The opportunity therefore exists to use this previously neglected source of data for the study of cultural and social change. From the discussion considered in the previous paragraphs, the adoption of externally-inspired changes to rituals, ‘Roman’ or otherwise, is unlikely to be automatic and must be read in its local context. The same debate obliges a consideration of the possibly varied understandings of rituals and, more to the point, their associated symbols in these contexts. Rituals and symbols associated with local elites may have a richness of possible interpretation because of the diverse contexts in which they sought to negotiate their position, towards local societies and external agents, but the values and strategies of non-elite groups also require examination. Here burial evidence gives a likely opportunity to remedy the limited attention paid to certain dimensions of identity. However the use of labels

‘The world of interactions opened up by the emerging empire was experienced only very indirectly by these peoples [i.e. the majority nonelite population]. Their fine wares might now have come from two thousand rather than two hundred miles away, but this did not disrupt their perception of the world or challenge their identity in any major way.’ (Terrenato 2005: 68) This suggests too that the ‘Romanization’ debate, focused on the encounter between metropolitan and local elites (especially Roman), especially as analysed for the formative period of provincial Roman culture in the first centuries BC and AD, has over-privileged a particular 3

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

such as ‘elite’ or ‘non-elite’ not only raises practical problems of identification in a burial context as much as in others (cf. Gardner 2007). It also risks the reification of categories which through burial like other arenas are continuously brought into being and re-made as moments; albeit symbolically charged, in the flow of social relations.

some more recent works it enjoys a slightly higher profile, but even in these exploitation is limited (e.g. Gardner 2007; Mattingly 2006; Millett 1995b; Todd 2004). A primary cause must be, first of all, the enduring abundance of the Roman period archaeological record in general, of which burial represents only one relatively unspectacular component, in contrast to other periods dominated by mortuary archaeology. Nevertheless burial evidence is now plentiful, albeit with a skewed distribution (chapter 2) and this seems unlikely as the sole explanation. The dominance of the ‘intellectualist’ perspective, as for the study of burial practices in the classical world more widely, is a more significant cause (Morris 1987; 1992: 15-18). Burial rituals in Britain have been interpreted primarily as evidence for afterlife beliefs, whether native, Roman or a fusion of the two, in the later period with the additional element of ‘eastern’ religions (Alcock 1980; Black 1986; Macdonald 1977; Henig 1984: 190-205). The paucity of other evidence has given burial a more central if controversial significance for assessing conversion to Christianity in late Roman Britain (Green 1993; Thomas 1981; Watts 1989, Petts 2003: 135-57). As is more widely the case for the ancient world, atypical if intriguing rituals – Ian Morris’ (1992) ‘quirks and oddities’ - garner more attention than the commonalities of burial practice, especially the decapitated inhumation burials which are a recurring feature of late Roman cemeteries and which have been considered as possible ‘deviant’ burials (Taylor 2008; Tucker 2012). The publicity achieved for the burials of individuals whose lives or deaths are susceptible to sensationalism, most recently the unwanted infants of the Yewden ‘brothel’ (Eyers 2011) further illustrates Morris’ observation.

1.3 Death and burial in Roman Britain and other provinces 1.3.1 Current interpretations of burial practice In this section current approaches to the study of burial practice in the Roman provinces, especially those bearing on cultural change and social structure, are evaluated to identify reasons for the relative neglect noted above and to develop an appropriate perspective on burial data. This builds on Jones’ critique of approaches to burial practice in Britain and other western provinces (Jones 1993a: 1993b) and Morris (1987; 1992) discussion of approaches to death and burial in the wider ancient world (see also Pearce 2000). It considers the interpretative traditions for provincial Roman burial as applied to the study of Britain and explores avenues for its fuller integration within a broader field. From sporadic attention (e.g. Reece 1977a) the burial rituals of the Roman provinces, especially those in western Europe, have become the regular subject of conference proceedings from the early 1990s onwards, including (intermittently) discussion of British evidence (e.g. Faber et al. 2004; Fasold et al. 1998; Ferdière 1993a; Heinzelmann 2001; Pearce et al. 2000; Scheid 2008; Struck 1993a; Trousset 1995; Vaquerizo 2002; Vidal 1992). The number of doctoral theses written in the last two decades wholly or substantially on burial material from Britain and neighbouring provinces illustrates the accelerating interest in this field of study (e.g. Cooke 1998; Gowland 2002; Hatton 1999; Hamlin 2007; Keegan 2002; Moore 2009; Philpott 1991; Puttock 2002; Redfern 2006; Swift 2000; Tucker 2012; Weekes 2005; White 2007). Many of the latter exploit the now much greater sample of well documented human skeletal remains, itself a product of the very significant growth in excavated cemetery sites over the same period (2.2).

Where burial has been integrated into broader narratives, a traditional emphasis has lain on connecting it to textually attested (or represented) population movements at the beginning and end of the Roman periods indicated in historical narratives. Hawkes and Dunning’s (1930) linkage of burials of the Aylesford tradition to the Belgic migration recorded by Caesar remained influential into the middle decades of the 20th century. For the late Roman period discussion of ethnicity and cultural identity in burial practice has focused on burials furnished with grave goods which may signify service in the late Roman army or civilian bureaucracy, and elements of female costume which may indicate an origin beyond the Rhine-Danube frontier, in particular in a cluster of burials from Winchester Lankhills, as well as small numbers at other sites (Cooke 1998; Cool 2010). Their significance in debate is out of proportion to their numbers; compared to continental Europe, especially the zone from northern France to Hungary, such burials are very rare and Waffengräber are generally not documented (Böhme 1974; 1986; Swift 2000).

Nonetheless burial archaeology remains a relatively neglected area of study for the province in that while burials may be frequently excavated and published, limited attention has been paid to the insights that they may provide into wider provincial Roman societies. Leaving aside the key exploitation of objects preserved complete in burials for typo-chronology, discussion of burial practice has most often been located within the descriptive accounts of provincial life (e.g. Alcock 1996; Liversidge 1968; Salway 1981); in many of the major syntheses or collections of essays for the province it is little used or absent (e.g. Faulkner 2000; Frere 1987b; Hanson 1994; Hingley 1989; James and Millett 2001; Jones 1991a; Millett 1990a; Potter and Johns 1992; Rogers 2011; Salway 2002; Todd 1989). Occasionally in

The shifts in conceptualisation of both scenarios illustrate both an increasing empirical knowledge and changes in 4

INTRODUCTION

interpretive perspectives. With the demonstration that the different elements of the ‘Belgic;’ package (coinage, burial ritual, ceramics) were not contemporaneous (e.g. Birchall 1965) changes to late Iron burial practice have been subsequently discussed instead in terms of the manipulation of symbols within the framework of status legitimation. From the abundant grave good assemblages of mid-first century to Augustan Belgic Gaul, in particular deposits of multiple amphorae and Roman finewares, Haselgrove (1987a; see also BöhmeSchonberger 1993) has argued that innovation in burial practice represents one dimension of a re-orientation of the symbolic behaviour of elites in mid first century BC Belgic Gaul after conquest removed traditional symbols of authority (1.1):

1959; 1969) to an increasing acknowledgement of local variability, as the recent regional case studies show (e.g. Fasold et al. (1998) Pearce et al. 2000; Struck 1993). For the study of geographical and, by extension, cultural ‘origins’ from burial rituals, as Reece observed (1982a), it is necessary for burial rites to be well documented across the range of possible origin areas. There remain significant lacunae in the ‘map’ of Roman burial rituals, in particular the restricted evidence for Rome and central Italy in the late Republic and early empire, though here data are increasing, especially from recent developmentrelated excavation around Rome (Heinzelmann 2001a; Buccellato et al. 2003). On the other hand where evidence is more abundant it can exacerbate the difficulty of identifying geographical or ethnic origin for a particular practice. Two key examples from Britain may be given here, one of a ritual widely found and the other of practice at a particular cemetery. Manuela Struck (1993b) identified four possible areas from which the bustum burial tradition in Britain (3.4) might have been derived and between which it was difficult to discriminate as the most likely source. In the third century AD cremation cemetery at Brougham, Cumbria, burial practice showed a striking difference from the province as a whole as well as the frontier region (3.5). Some rituals such as the cremation of horses on the pyre and objects such as bucket pendants and some bead types, suggested connections to the middle and lower Danube. Other artefacts, for example Hemmoor buckets, had a wider distribution in continental graves.

‘At burial sites such as Goeblingen-Nospelt the message is re-iterated in everything from largescale wine consumption to the cremation rite: the power and continued success of the ruling groups was indissolubly linked to the Roman alliance’ (Haselgrove 1987a: 116) Creighton (2000; 2006) has suggested too that burial rites documented in pre-conquest Britain asserted connections to Rome by client kings, though not all accept this. Fitzpatrick (2000; 2007) points out the local diversity of rituals into which these Roman artefacts were incorporated. It is worth noting too that large-scale deposition of grave goods, especially ceramics and amphorae, typical of such burials does not resemble contemporary practice in Rome for elite and imperial funerals which are characterised by the placing of material on the pyre rather than in the grave (Flower 1996; Price 1987).

With occasional exceptions such as Brougham, for Roman Britain commentators have emphasised general continuity from pre-Roman ritual forms, with the main difference being an expansion of the repertoire of grave goods to include newly available objects, for example lamps and glass unguent vessels (Philpott 1991: 217-19; Struck 1995; 2001). This is a view widely held for the north-west provinces; until the spread of inhumation from the late second century onwards burial practice is considered as not fully Romanised’ (van Doorsaeler 1967: 67; Morris 1992: 68; Wightman 1985: 188). The regionally diverse character of Iron Age burial in Britain may also have influenced the emphasis on characterising regionality (rather than ethnically specific variability) in burial practice for the Roman period. This has generally used individual characteristics, for example the placing of the body, the presence of specific grave goods, or choice of a particular monument type to identify regional trends (Philpott 1991; Struck 2000). Jones employed multivariate analysis to argue that different burial rituals can be associated with cultural rules shared by communities at local and regional as well as at wider imperial levels. (Jones 1983; 1987; 1991b; 1993b).

For the ‘intrusive’ burials of late Roman Lankhills and comparable examples, re-examination armed with new data, including a much better understanding of late Roman personal ornament and its distribution, has been more significant than theoretical repositioning for doubting the original interpretation (see further 5.3.2). For rituals and grave furnishing at Lankhills these have served to weaken the coherence of the groups identified by Clarke and to nuance the geographical and cultural connotations of individual objects (Cool 2010; Pearce 2010). The key development has been the undermining through stable isotope analysis (oxygen and strontium) of skeletal remains of any relationship between burial practices, whether ‘intrusive’ or local, and geographical origin (Eckardt 2010). For the intervening period of the early and middle Roman periods (i.e. the first to third centuries AD) the association between burial and ethnicity and the identification of burials of indigenous and exogenous individuals has been less extensively debated in the study of Britain than for the German and Danubian provinces. Here too there has however been some shift away from the more confident identification of ethnic affiliation from burial rite (e.g. van Doorsaeler 1967; Nierhaus

However although the empirical knowledge of burial types is becoming more nuanced and refined and its complexity and diversity more fully acknowledged, less consideration has been given to the demonstration that the identities given to the dead are not the direct equivalents of those which defined them in everyday life 5

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

(Hodder 1982a: 142-43). The presentation of identity through mortuary ritual has been shown to be an active construct; the identification of social practices such as burial as explicit ethnic markers, must therefore be argued for and not assumed (Jones 1995). Struck’s (1995) analysis of the continuing abundance of grave goods in early Roman period burials in Britain not only as continuity of traditions, but also as an assertion of cultural difference, a manifestation of resistance, remains unusual in considering the possible active manipulation of Roman period burial rituals. Its argument echoes the claims sometimes made for the continuity of pre-Roman monument types in a provincial setting. For example Benabou (1976: 281-84, 494-96) interpreted the third century AD popularity of the cupola tomb type in North Africa, with its resemblance to a widespread pre-Roman form, as asserting cultural resistance. Similar claims were made in the past for the preference for tumuli, the most frequently documented monuments in Britain, Gallia Belgica and Noricum / Pannonia as revivals of local prehistoric forms (Wigg 1993a).

intrinsic value, for example copper alloy vessels or precious metal jewellery (e.g. Nuber 1972; Rottloff 1995). Van Lith and Randsborg’s (1985) identification of glass as a symbol of investment in burial from its recurrent association with larger grave assemblages is a rare example of more systematic analysis of these attributions of value. While a general equation between lavish display in burial in the form of large grave good assemblages and high social rank is widespread, more systematic analysis is less frequent. The identification of ranking through burial was the principal concern of Jones (1983) survey of first to third century cemeteries in the western provinces, of which individual elements have been published (1977; 1984a; 1984c). This study remains the most extensive application of a processual approach to Roman period burial evidence. Jones’ approach followed Binford’s proposition that: ‘…we would expect that the facets of the social persona symbolically recognised in mortuary ritual would shift with the levels of corporate participation in the ritual and hence vary directly with the relative rank of the social position that the deceased occupied in life.’ (Binford 1972b: 229)

However the equation of ‘traditional’ with ‘resistant’ in a Roman setting is problematic, given the authority attached to the former in Roman culture (Jiménez 2008). A capacity for multiple resonances in indigenous and Roman perspectives seem more plausible than resistance to explain the choice of monuments; in the case of tumuli the potential echo of late Republican metropolitan tombs as much as the Bronze Age barrows of north-west Europe has been noted by others (van den Hurk 1984; Morris 1992: 51). A similar capacity for engendering multiple readings also seems appropriate for the many regional monumental traditions, attested especially in Mediterranean provinces. These typically see new construction and artistic techniques as well as tomb forms of Roman / Italian origin combined with an existing repertoire of funerary monuments, often developed in earlier ‘Hellenising’ phases (e.g. Cormack 1997; de Jong 2010; Jiménez 2008; Mattingly 2003; Stirling and Stone 2007; Schmidt-Colinet 1997; von Hesberg 1992). A similar potential for diverse readings has also been identified in funerary art. For example the carved tower tombs of mid-imperial Moselle region use Roman sculptural techniques and motifs but represent local elites in regional rather than Roman dress and participating in productive and commercial processes in a way that may have been disparagingly read by their Italian counterparts of a century before (Freigang 1996). General discussion of culture change (see 1.2 above) and these examples of polyvalent readings for monuments suggest an interpretive perspective which may also be applied to burial rituals too.

In other words burial provided a more or less direct presentation of the roles of the deceased individual and the analysis of a cemetery or group of cemeteries a measure of social complexity, based on the number of different dimensions of the social persona which could be identified. Jones identified burial norms across the western provinces subdivided into local variants (see above). Within these local variants, often identified at the level of the individual cemetery, through multivariate analysis he defined groups of burials which he associated with different social statuses. The absence of associated inscriptions for most of the sample or other direct or indirect status indicators made it impossible to specify these. The principal conclusion that provincial Roman society was stratified is perhaps unsurprising, even if this particular approach has not been repeated on other samples it has engendered a more widespread quantitative approach to the analysis of cemeteries (e.g. Bridger 1996; Cool et al. 2004; Millett 1993; Struck 1995). Beyond the analysis of individual cemeteries (e.g. Quensel von Kalben 2000) multivariate approaches have received rarer subsequent application, though correspondence analysis offers new opportunities in this regard (e.g. Biddulph 2005; Cool and Baxter 2005). Spatial location has also been used to explore status difference. For example Martin-Kilcher (1976; 1993a) has mapped status differences in rural society onto burial practice, assigning prominent tombs to elites, spatially marginal and discreet for villa workers and slaves; similar models have been offered for rural sites in Britain (6.1). These interpretations are however generally predicated on a relatively simple elite: non-elite dichotomy and have not yet engaged with the range of hypotheses for rural social structures offered in recent discussion of settlement space in Britain (see 1.3.2) or

As well as cultural identity, burial evidence has also been used for exploring social structures. It is a commonplace to relate burials privileged by abundant and diverse furnishing to elites (e.g. Mattingly 2006: 377-8; Philpott 1991: 230; Struck 2000). Similar conclusions are sometimes drawn from the presence of particular objects, usually on the basis of archaeological scarcity or perceived 6

INTRODUCTION

wider models of Roman rural society (e.g. Garnsey and Whittaker 1997).

recently excavated instances from LPRIA and early Roman north-western Europe, for example at Folly Lane, St Albans, and Stanway, Colchester and in the Ardennes and Moselle regions (e.g. Abegg 1989a; Metzler et al. 1991; Metzler and Gaeng 2009; Wigg 1993c). In larger cemeteries used over a long period of time distinguishing features related to individual ceremonies is not so easy but the sequence of ritual can still be reconstructed in more general terms (e.g. Haffner 1989b; Polfer 1993). Westhampnett, east London and Springhead Pepper Hill represent examples of Late Iron Age and Roman cemeteries from Britain where this approach has been fruitfully applied.

Criticisms of Jones’ and others’ approaches are possible within a processual framework. His analyses were in the main directed at large and well-recorded but often isolated cemeteries. Only occasionally, as for the small town at Braughing (Herts.) or Ampurias, for example, did the opportunity exist to compare practice between cemeteries around one site or in one region. A single cemetery or the cemeteries of a single settlement are unlikely to contain all the variation in the burial practice of a particular society. A regional approach is therefore necessary even if this compromises data quality (J. Brown 1981; 1995; Goldstein 1995). Techniques of mortuary analysis which depend on the existence of large sample sizes can only rarely be applied to many of the burials from rural contexts, encountered as singletons or in small numbers even in extensive excavations.

In general also approaches to social status have been primarily focused on archaeological rather than skeletal evidence, though the latter was fundamental to early processual studies of burials (e.g. Shennan 1975). Though hedged around with multiple complexities, analysis of skeletal material, at least from inhumation burials, has very substantial potential further to illuminate social structures where variation in health can be plausibly attributed to differential access to resources based on social status. On the basis of larger and better documented samples, osteological indicators from inhumation burials for the impact on skeletal development of diet and disease are now revealing significant differences within and between Roman period cemeteries, as well as between the Roman and other periods (Roberts and Cox 2003; 2004). Assessment of dietary variability is now also supplemented by the study of stable isotopes (carbon and nitrogen) (e.g. Chenery et al. 2011; Müldner and Richards 2007). Analysis of inhumation burials from Dorchester (Durnovaria) over the first to fourth centuries AD illustrates the potential of such analyses; individuals buried in stone and lead-lined coffins and mausolea enjoyed better diet and lower mortality risk, though the latter is also conditioned by other variables including age and settlement context (Redfern 2008; Redfern and Dewitte 2011; Richards et al. 1998).

It also remains to be proven that the burial form within which Jones analysed variation was that followed by all social groups. Morris (1987) has for example demonstrated that the proportion of the population receiving a ‘normal’ burial practice in Greece fluctuates markedly between the Geometric and Classical periods. Given the lacunae in the burials of Iron Age date in Britain, it remains to be demonstrated that the visible forms of burial attested in the Roman period are entirely representative, as also other parts of north-western Europe where pre-Roman burial traditions are only sporadically documented (Collis 1977c; Ferdière 1993d; Hessing 1993c; Pion and Guichard 1993; Whimster 1981). Jones’ analyses were based primarily on grave good assemblages. These are usually the most reliably recorded aspect of burials, but it is an archaeological commonplace that interment in the grave, sometimes with furnishing is not the only or even the most significant part of the ceremony for the destruction of resources. The Roman period is no exception; textual evidence for ‘normalised’ Roman period funerary rituals, show an extended period for the funeral itself, including laying out, procession, cremation, burial and associated sacrifice and commensalism, as well as similar activity in later commemoration (Scheid 1984; 2005: 161-88; Hope 2009). Excavation increasingly shows that other ritual episodes of ritual prior or subsequent to interment may be archaeologically accessible, especially the act of cremation (see chapter 3). Here developments in research into Roman period burial practices mirror those in other periods (McKinley 1997). In sites where there is only one or a small number of graves and good preservation different deposits can be ordered in a convincing sequence of ritual from the systematic analysis of cremated bone and collection of fragmentary artefactual material and pyre debris. The cremation and burial sites identified as those of the emperor Galerius and his mother from Gamzigrad, Serbia, provide a spectacular example (Srejovic and Vasic 1994), but there are also

Like most processual analyses of burial practice, Jones’ and related analyses concentrated on the social status and cultural dimension of identity (Parker Pearson 1999a: 7284). The bio-cultural and life-course approaches also enabled by the same skeletal data, assessing age and gender both in terms of lived experience and ritual treatment, have given significant insights into these dimensions of identity (Redfern and Gowland 2011; Gowland 2007; Redfern 2008). Studies of individual cemeteries of late Roman date where furnishing is common have shown that provision of some objects with the dead is closely conditioned by age and gender and that ‘status’ cannot be identified in isolation from these variables (e.g. Cool et al. 2004; Gowland 2001; 2002; Martin-Kilcher 2000). Though the sample is smaller than from other provinces, texts and images from commemorative monuments also have significant potential to illuminate these dimensions of identity. Whether responsibility for commemoration is directly related to household structures may be doubted for not taking sufficient account of epigraphic funerary 7

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

conventions and ritual context (Morris 1992). However the epitaphs of women and children have yielded significant insights related to frontier communities; the predominance of women in the corpus of funerary monuments from Hadrian’s Wall is argued to reflect the use of commemoration to legitimate de facto marital relationships for auxiliary soldiers with the trappings of connubial authority in word and image (Hope 1997)

monumentalisation established within the general ‘epigraphic habit’ (e.g. Hope 2001; MacMullen 1982a; Woolf 1996; 1998) or the variability by region in the chronological horizons of funerary monuments, as seen for example in Gaul (Ferdière 1993a; Hatt 1986; Landes 2002; Moretti and Tardy 2006). But while the active role of Roman commemorative monuments in dialogue with their audience has been extensively explored, again much less attention has been paid to burial rituals here and the significance and potential readings of the material culture associated with the dead in burial rituals, beyond traditional argument for objects serving somehow to enable a certain style of afterlife existence.

This last observation introduces a further key element to the critique of approaches to status. As significant as methodological problems are the theoretical objections to interpretations of this kind, especially to the premise that variation in the treatment of the dead directly reflects the varied statuses of the living. A plethora of ethnographic and historical examples demonstrate this not to be the case; rather than being a passive reflection of a society from which its organisation can be ‘read’, mortuary rituals are actively used in the negotiation of social position, especially in periods of change (Hodder 1982b; Parker Pearson 1999a: 84-87). Burial rituals in the city of Rome provide myriad examples, from the political theatre of the funerals and tombs of Republican aristocrats and emperors to gift-giving through provision of burial places for households and clients in the late Republican and imperial city (P. Davies 2000; Patterson 2000; Price 1987; Purcell 1987).

1.3.2 Perspectives for a contextual archaeology of burial practice In the previous section two connected areas of interpretation have been the main focus of concern, the relationship of burial practice to cultural identity, especially the adoption of Roman practices, and to social status. The critique of processual approaches to mortuary rituals suggested caution in identifying elites and nonelite groups and emphasised the capacity for burial ritual to be an arena where status and other dimensions of identity might be negotiated; this section examines how this may be accessed through archaeological evidence.

Early post-processual critiques argued that burial rituals serve as occasions for dominant groups to assert and reinforce their world view but they also offer opportunities to non-dominant groups to claim recognition for their achievement of riches or honour (Parker Pearson 1982; Shanks and Tilley 1982: 131-32). Again Roman examples illustrate this point well, both in the metropolis and in local civic contexts where the curial class is underrepresented and freedmen distinguished by the number and elaboration of their monuments, as well as their choice of funerary portrait and other iconography (Hackworth Petersen 2006; Kampen 1981; Mouritsen 2005; Zanker 1975; Zimmer 1982).

Some methodological difficulties have been identified above, i.e. that any study of burial practice must problematize the degree to which the modes of burial usually considered typical of the Roman period are indeed so, and that the archaeologically visible stages of burial rituals need not be wholly representative (see further 1.4 below). In the critique advanced above it was emphasised that burial rituals are manipulated by participants as part of broader social and symbolic strategies. Twinned with this emphasis on agency is that on context, i.e. that the way in which funerary rituals are both understood by and act on the living is contingent on the place of the dead in particular cultural contexts.

There is of course a danger of reductionism if excessive emphasis is placed on instrumentalist manipulation of power relations (Barrett 1991; Parker Pearson 1999a: 86). Tarlow (1999) notes that the affective dimension is neglected in this instrumentalist view of ritual, though in the absence of appropriate evidence this is difficult to reconstruct in specific settings. A more straightforward observation to assess is that burial also need not always play a prominent role in social competition; Huntington and Metcalfe’s (1992) ethnographic studies show the dependence of funerary ceremonial on the ‘history’ of the burying group, both its ability to fund funerary ceremonial, and the coincidence of the death and / or funeral, irrespective of the status of the deceased, with occasions for display. As well as this intrinsic variability, Cannon (1989) demonstrated cross-culturally the existence of cycles of display and restraint in the mortuary sphere and again Roman examples fit this pattern well, as do the differing local chronologies of

It is thus essential to place burial practice in its context in order to assess the degree to which it is used as medium for negotiating status or other differences. Others have argued that this can be achieved by comparison across the conventionally separated elements of the archaeological record (Hodder 1984; Morris 1987; 1992; Parker Pearson 1993; 1999). Many prehistoric datasets, including that exploited by Parker Pearson in his 1993 study of Iron Age Jutland, are dominated at different periods by either cemeteries, settlements or monuments (see also Bradley 1984). The major advantage of the Roman period is the abundance of archaeological evidence from a wide range of contexts, as well as documentary material. It is this possibility that is exploited here, drawing comparisons across these different realms of study, but the scope is narrower than that suggested by Parker Pearson. For example the relationship between age, gender and funerary rituals receives little attention in this thesis (see above). Instead, 8

INTRODUCTION

in the following paragraphs I wish to explore two key aspects of context for discussing the place of burial in the construction of identities and the creation of social relationships, the material character of the ceremony, and its spatial setting, where the ceremony takes place and / or the monument is erected. Although they are considered separately for analytical purposes, it will become clear that space cannot be considered outside other aspects of the treatment of the dead.

1989; 1991; Barrett 1993: 115-20) have stressed how the artefacts placed with the dead are signifiers of which the connotations resonate beyond the sphere of burial. These objects evoke other areas of social practice, and the ceremony draws its power from these recurring allusions, in the case of Bronze Age funerals to feasting and warfare: ‘…these Beaker funerals were singular events, in which particular identities for the deceased had to be produced and fixed for the deceased in the minds of the onlookers’ (Thomas 1991: 40)

The first area of analysis is to consider how, in the context of societies known to be strongly hierarchical, communities are internally differentiated by their burial ceremonies, both at the intra- and inter-cemetery scale. It is a commonplace of Roman archaeology that settlement is highly hierarchical, but there is further scope to explore the complexity and dynamics of the burying communities for cemeteries associated with different site types within regions. Non-epigraphic burial evidence has been much less used to understand these (Burnham et al. 2001: 72) or to examine the hypotheses for rural social organisation based on settlement evidence, for example J. T. Smith (1978; 1982; 1997) and Hingley’s (1989) characterisations of rural societies that become increasingly stratified over time have yet to be examined in relation to burial data, beyond some generalising status-related characterizations (1.3.1). In question here is not the attribution of status labels to particular burial types, problematic both in principle and in practice, but rather the investigation of how difference is articulated. The significance of differentiation among burials can be assessed both by comparison to status-related differentiation in other spheres, through material culture and domestic or public space both by comparison of funerary rituals on an intra- and inter-cemetery basis. Parker Pearson’s (1993) approach to contextualising cemeteries also includes identification of the context in which items of value were deposited, so as to assess burial as a sphere of display in comparison to other arenas, such as votive deposits or public architecture. Again the Roman period lends itself to this because of its abundance of contextual data, though it is easier to compare particular aspects of burial rituals to other spheres, for example the ceramics deposited different contexts, than to compare globally between contexts where very different forms of evidence are involved. But in this respect structural Marxist approaches like processual analyses also encounter a further objection. Why does the expression of difference in the ritual assume the form of a particular artefact or monument (Hodder 1982b; 1984)? Quantitative comparison and assessment of the relative importance of different depositional environments must therefore be supplemented by attention to the possible connotations of the objects and animals placed in burial assemblages.

Or, as proposed by Gosden and Garrow (2012: 196) for the study of decorated metalwork in Iron Age burials: ‘Material culture was generally placed in the grave to have an effect – on the onlookers at the funeral, on the ‘gods’ in the present, on the dead person in the afterlife’. On this point Treherne (1995) extended the ‘beautiful death’ of Homeric epic as characterised by Vernant (1982) to a much wider span of European pre- and protohistory as a response to the ‘Angst’ of death by the presentation of the male corpse as the epitome of a particular male virtue. The corpse was carefully groomed, and furnished with sets of artefacts related to personal care, shears, razors and toilet sets, and weapons and armour, either as full sets and panoplies or as individual artefacts acting in metonymic relationship to the whole assemblage. In these cases the display of the dead does not allow a specific attribution of the status of the deceased but can be exploited as a guide to areas of behaviour central to social reproduction beyond the mortuary sphere. For the study of prehistoric societies evidence for these other areas may be lacking, but ethnographic examples show how the presentation of the dead through dress and objects serves to bring to mind the normative qualities by which the living are measured. For example weapons, animals and farm produce evoke the achievements as farmers, hunters and warriors of men and baskets, calabashes and soup pots the fertility and household economy of women amongst the LoDagaa in Ghana (Goody 1961: 80-83). In this perspective the mise-en-scène of the burial deposit in the grave, grave goods and human remains, can be presented as an equivalent, if ephemeral representation to Roman funerary portraiture, which represents through a combination of portrait image, attribute or action the (idealised) collective values of groups. These are not necessarily elites, and often comprise social groups which in some respects are transitional or ambivalent, for example the freedman, auxiliary soldier or gladiator (Hope 1997; 2000; Mouritsen 2005; e.g. Zanker 1975). In the case of burial the ‘fixing’ of identities for the deceased in the minds of onlookers is effected by the artefacts, giving cues for reading for the participants in the ceremony in the form of associations with other areas of social life. Roman archaeology offers rich possibilities

Approaches to burial furnishing in prehistory offer an avenue of interpretation for Roman period burial assemblages. Developing earlier work on burial symbolism by Pader (1982), recent study of Bronze Age and Iron Age burials (Gosden and Garrow 2012; Thomas 9

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

for exploring the range of resonance and allusion in burial assemblages because of the wealth of contextual information. Areas of social practice alluded to in the burial assemblages can be compiled from the recurring presence of certain artefacts and then contextualised in relation to depositional practice in other contexts and to literary and artistic evidence, in order to establish the potential range of connotations that the different objects and their combinations may carry. The contextual richness also allows us to explore the polyvalency in these sets of objects. Discussion of late Iron Age and early Roman burials in Gaul and Britain has focused for the former period on the competitive display of access to Roman imports or connections to Roman power of the buried or buriers, and for the latter on the continuity or even active evocation of traditional ‘Iron Age’ forms of social life (see above). Both interpretations privilege one connotation amongst a complex of signifiers and give less emphasis to the changing forms of social practice, especially in the sphere of eating and drinking, that the contents of the burials have suggested.

between tomb and road frontage, and thus between monument and viewer within the well preserved cemeteries of Rome and Campania. From the first century BC to the second century AD the emphasis shifted from the outward external projection of individual or family status to a greater emphasis on tomb interiors as the space of decoration, alongside some withdrawal from the street frontage as the primary focus of display. This characterisation has subsequently been extended beyond this specific setting (e.g. von Hesberg 1992; Heinzelmann 1998) but, as Roth-Congès (1990) notes, it has not been established to what the sequence observed in central Italy applies elsewhere, and other spatial relationships between tomb and public have been mooted (e.g. Leveau 1987b: 287-88). Later cemeteries feature little in von Hesberg and Zanker’s (1987b) characterisation of the changing relationship of tomb to the street frontage (1.3.1). While the locations with proximity to saintly bones had usurped the road frontage as the privileged burial location by the fifth century AD, at least in parts of the Mediterranean and more occasionally in the north-western provinces (Duval and Picard 1986), we should be wary of absorbing the intervening centuries into this picture. Given the frequency with which monuments were erected in the countryside, there is also significant scope to extend this approach beyond the city. The setting of monuments associated with villas has generally received so far rather limited attention; for the north-west provinces work by Lafon and Adam (1993) and Krier and Henrich (2011) represent rare exceptions.

The other key aspect of context to be developed further is space. The assessment of the distribution of the attribute in question in relation to site type, the latter providing a social context, is well established and will form part of the regional approach, argued above to be insufficiently developed in previous work, in order to explore the distribution of burial display in relation to the settlement hierarchy. The current orthodoxy holds that burial display was focused on a rural rather than urban setting, a supporting argument for the characterisation of Roman towns in the province as having limited socio-political significance (see further 7.1). This characterisation is revisited here in order to establish the social geography of burial display.

The preservation of British cemeteries necessitates the adoption of less fine-grained perspective, asking the broader question of the degree to which, in the cities of Roman Britain, the frontage of the roads radiating from towns is the preferred location for the erection of monuments or other forms of burial display. Esmonde Cleary (1987: 168) has drawn attention to the possible preference for certain locations, for example the routes to London, but there is scope for further examination of the relationship between cemeteries, monuments and their settings, especially where the relationship of road frontage to cemetery is reliably established.

The analysis of the spatial context also requires consideration of the relationship of burial to its setting in an experiential sense. Parker Pearson’s (1993) model for establishing context entails the assessment of the spatial relationship between cemeteries and settlements to determine their relative proximity over time and thus the importance of the dead as measured through their visibility to the living. This approach is part of a broader shift towards a phenomenological perspective on cemeteries which considers how the dead were encountered in landscapes and thus not only the frequency with which or scale on which they imposed themselves on the living but also how their presence gave meaning to the landscape (Tilley 1994). Again here there is potential common ground with classical archaeology, especially Von Hesberg and Zanker’s (1987b; 1992) re-examination of the commonplace of the streets of tombs illustrates the potential of Roman data for such an approach. The setting of urban cemeteries in the north-west provinces has been predominantly exploited as evidence for cities’ topography (e.g. Bedon et al. 1988; Esmonde Cleary 1987). Von Hesberg and Zanker’s (1987b) re-examination of the spatial setting of tombs associated with Roman cities elicited subtle but important changes in the relationship

There also is scope to extend this approach to nonmonumental burials and to consider more than only the audience of passers-by. Large scale excavation of Roman rural sites in Britain normally reveals individual or small groups of burials dispersed across settlements, often in close association with other features. These have had little attention in archaeological analysis, often because the small sample size does not lend itself to statistical analysis; the main exceptions identity them as those of subordinate groups, workers, slaves, tenants and so on (Collis 1977b; Philpott and Reece 1993) (see further 6.1). However ethnographic examples indicate that burials close to or within houses and other settlement features need not indicate the low status of these individuals but rather gives insights into the organisation and 10

INTRODUCTION

conceptualisation of settlement space (e.g. Moore 1986). Directly and indirectly such instances have influenced a reconsideration of the human remains, whole or part bodies, regularly found on Iron Age settlements in Britain, the deposition of which is now interpreted as significant deposits in the lifecycle of a community, rather than as the burial of low status or deviant individuals (Hill 1995; Cunliffe 1992). The problematic requirement of large burial samples for the application of specifically ‘mortuary theory’ can be sidestepped by an attention to spatial context and association (cf. Hill 1995). There is scope therefore to re-examine the relationship between burial and settlement features from the premise that space does not so much reflect social relationships as constitute the active medium through which these are created and transformed (Parker Pearson and Richards 1994b). This dimension has been explored through the study of villa architecture and art (E. Scott 1990b; S. Scott 1994; 1995), but while the importance of depositional evidence for such an approach has been signalled (e.g. Chadwick 2012; Hingley 1990a; Meadows 1995; Rippengal 1993) it has yet to be extensively exploited for burials, with the significant exception of E. Scott’s study of infant burials (1988; 1991). In this context we should not only be seeking to explore connections with other contemporary features, since prehistoric monuments were selected for votive deposits and sometimes sites of burial in the Roman period (Vermeulen and Bourgeois 2000; Williams 1998a).

provincial and wider setting. The content of the chapters is briefly outlined in the following paragraphs. The distribution of Roman period burial evidence from Britain, in particular England and Wales, according to burial type, date, settlement context and region is set out in Chapter 2. The chapter explores the factors which are responsible for the very uneven distribution which is documented and considers its implications for the study of provincial society. It argues that lacunae are not only due to taphonomic factors but also to the slow rate at which an archaeologically visible burial practice is adopted in the early first millennium AD. This slow emergence of an archaeologically visible burial form is discussed as a significant but previously neglected aspect of cultural change. Chapter 3 investigates another characteristic of burial evidence, namely the different types of data potentially available in excavation for the sequence of funerary rituals, with a particular focus on the pyre and cremation. It was apparent that there were various problems in the identification and interpretation of the different features to be anticipated on cemetery sites and their relationship to the burial process. An increasing sensitivity to the formation processes behind cemetery deposits makes it difficult to take older classifications of features, especially pyre sites, at face value. The focus of this chapter lies on the pyre site, probably the richest source other than the grave for mortuary activity which may be accessed archaeologically, and the conclusions on the cremation process that may currently be derived from its study, drawing on evidence from the wider Roman world to put the British evidence in context. The chapter evaluates the range of features which might be anticipated from cemeteries, with particular attention to the contexts from which pyre material might be derived. Expectations concerning pyre sites are argued to be dependent on assumptions about the organisation of cremation, especially the notion of a municipal or public ustrinum, which documentary evidence shows to be problematic. Differences in recovery strategies and types of evidence available make it difficult to sustain any generalisations about regional difference in Britain on the relative emphasis of pyre and grave as a context for burial display.

1.4 The structure of discussion My original intention was to translate these aims into practice by examining ritual sequences, from preinterment activity to commemoration, and spatial contexts of burials and monuments as regional case studies. However as I feared this soon proved impractical since no single region satisfied the quality of data required from a sufficient number of sites to do so and it became clear that the sample would have to be drawn more widely. It soon also became clear that the relationship between burial process and surviving evidence from excavations required further consideration, with particular reference to the cremation pyre. And regardless of the specific evidence for rituals which happened to be available from particular sites, it also soon became clear that burial evidence itself as a whole was very uneven in its distribution in a way which had not fully been appreciated previously and did not seem to be entirely explicable through taphonomy. Both topics, the distribution and character of burial data, especially from cremation cemeteries, required prior assessment. To accommodate this the book is organised as follows. Chapters 2 and 3 therefore investigate the character of burial evidence available from the province and its typicality; chapters 4 to 6.3 present regional case studies, exploring the variation in burial practice in relation to social dynamics in three contexts, urban, ‘small town’ and rural; chapters 6.4 to 8 put their results in a wider context, comparing burial practice at a regional rather than site level and putting the case study results in a

The case studies of mortuary rituals, their associated material culture and the spatial setting of burial begins in chapter 4. No sample area in Britain currently offers a range of urban, minor centre and rural burial data to satisfy fully the requirements of a regional approach to burial practice; samples from two counties, Hampshire and Hertfordshire, allow study and comparison of the burial practices across a range of settlement types, including public towns, i.e. the civitas capital of Venta Belgarum and the municipium of Verulamium (St. Albans), minor centres (especially Baldock and Braughing) and the countryside. The main emphasis lies on the early Roman period but some analysis is extended into the fourth century AD, especially with regard to the 11

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Hampshire sample. Where available, information is exploited from different points of the cremation ceremony, the efficiency of burning, the deposition of artefacts on the pyre, the collection of cremated bone, and the grave goods. Information on the sequence of ritual is less available from inhumation cemeteries and is here restricted primarily to examination of grave, container for the body and grave goods.

these mortuary displays, exploring the connotation that these had for different participants. In contrast to the Iron Age, it argues that the major effect is achieved not straightforwardly by spectacular deposition of prestige goods but by an allusive style of display that emphasises different qualities at different periods. For the early Roman period this is otium, especially the acquisition of cultural savoir faire with particular reference to the convivium, for the later Roman period through the differentiated presentation of the body in dress and ornament. The conclusion summarises the results of the different strands of analysis and reflects on the possibilities for future work from the perspective developed through this book.

Chapters 4 and 5 characterise burial practices in the urban and small town cemeteries of the sample areas, exploring differentiation within burial practice and (briefly) setting burial practice in the context of other aspects of social process seen through the archaeological evidence, in order to establish the place of burial within social dynamics at the sites in question. The first sections of chapter 6 (6.2-3) assess the small corpus of rural burial evidence from the sample areas from the same perspective. The chapter evaluates the validity of current models for characterising status difference in rural societies through burial evidence, especially as expressed spatially. A focus on spatial separation of burials based on status is found however to be of limited value. Drawing on a wider sample of burial evidence from across the province in 6.4-6.5, this chapter identifies recurring preferences in the places chosen to bury the dead on rural settlements which give wider insights into the structuring and experience of rural settlement space. Building on the analyses in the previous chapters, 7 and 8 explore on a broader level the spatial relationship of burial display to political and social geography and the character of burial display. In 7.2 urban cemeteries at St Albans and Winchester are compared as foci of display to those of other cities in Britain and beyond. Different histories are identified between urban centres in this regard; a monumentalising funerary culture is, on current evidence, restricted to a small group of cities differentiated by the circumstances of their foundation. In 7.3-4 the relative importance of towns, minor centres and the countryside as places of burial display is considered within the sample areas, widening the discussion to the context of the civitas. The burial practices established at large well recorded urban, minor centre and rural cemeteries allows the definition of regional characteristics against which individual and small groups of burials and less well recorded assemblages can also be exploited. The distribution of burials which meet quantitative ‘thresholds of display’ or which are characterised by qualitative attributes, for example burial monuments and stone or lead coffins, is explored and interpreted, revealing funerary display to be better characterised as occurring in zones rather than in association with particular site types. In chapter 8 the character of burial display, especially through the objects deposited with the dead, is examined for its insights into Romano-British society. It focuses in particular on the grave assemblages from ‘privileged burials’ which can be distinguished on a quantitative basis in the sample regions and explores the character of 12

CHAPTER 2: DIGGING DEATH: THE DISTRIBUTION OF BURIAL EVIDENCE FROM ROMAN BRITAIN population. The implications of this archaeological invisibility for the analysis of burial ritual are considered.

2.1 Introduction Most analyses of burial practice in Roman Britain, as well as in other provinces, have proceeded directly to the investigation of particular aspects of mortuary ritual, for example the treatment of the body or the deposition of particular artefacts (1.3). Limited attention has been paid to how far the distribution of burials with particular attributes relates to the global burial sample available for study, the degree to which that sample is itself representative of Romano-British society, or to the factors which have generated and potentially skewed that sample. Discussion of the nature of the evidence is instead usually focused on what types of information can be recovered during excavation or during post-excavation analysis, on how we should record, manipulate and interpret mortuary data (chapter 3). A core tenet of mortuary analysis in the 'New Archaeology' (Chapman et al. 1981) which remains central whatever the theoretical perspective adopted, is that the character of the overall sample of burial data conditions the degree to which we may extrapolate from it to an understanding of the society of which it is the ‘residue’. The implications of this point need further consideration in the study of Romano-British burial practice.

2.2 Sample compilation and classification The aspiration to establish a database of all Roman burials in Britain remains unfulfilled (e.g. Chambers 1980). Even prior to the boom in development-related archaeology, to establish an inventory of Roman period cemeteries from an area comparable in size to Britain was an enormous undertaking (van Doorselaer 1967). The closest approximation to a distribution map of all Roman period burials in the National Archaeological Record (NAR – now incorporated within the National Monument Record - NMR) was published by Leech (1993), but acknowledged to be only a preliminary assessment from records collected up to 1991. The Historic Environment Records (HERs), the geo-referenced inventories of archaeological sites maintained by local authorities in England and Wales, should provide the most comprehensive and up-to-date data but are not yet compiled to a comparable high standard, nor are even the best exhaustive (in preparing the thesis I used their predecessors, the Sites and Monuments Records). Even where available, they individually require timeconsuming manipulation to produce appropriate data and only a handful of local and regional syntheses, have been undertaken, for example post-war Royal Commission county survey volumes on Cambridgeshire, Dorset, Gloucestershire, and Northamptonshire and other more recent examinations of individual counties or regions (e.g. Andrews et al. 2009; Booth 2001; Booth et al. 2007; 2011; Faull 1977; Faull and Moorhouse 1981; Foster 2001; Leech 1980; Taylor 1993). In some cases these are now complemented by the overviews in the English Heritage regional research framework documents, some of which provide extensive updates on burial excavations undertaken in the context of development-led archaeology (e.g. Chadwick 2009; Philpott 2006).

Since 1920 c. 15,000 Roman burials have been excavated in Britain. At no single site does the number compare with the major continental cemetery projects, for example Krefeld-Gellep or Wederath (Pearce 2002), but Britain is nevertheless among the better documented provinces for published Roman period burials. However the distribution and provenance of the sample are as important as its overall size. Robert Philpott (1991) demonstrated in outline that excavated burial evidence is biased to the late Roman cemeteries of towns in southern England, an observation since re-iterated by others (e.g. Esmonde Cleary 1992: 28; Millett 1995b: 125; Roberts and Cox 2004). However Philpott’s survey, for reasons to be discussed below, introduced its own biases into understanding of the distribution of burial evidence. This chapter aims therefore to offer a general characterisation of the distribution of Roman period burial data from Britain according to selected variables. The collection of data for analysis and their classification are first described. The general characteristics of the excavated sample are then outlined, including the representation of burial types and distribution in space and time and in relation to different types of context (e.g. urban, rural, military etc). The reasons for biases in the distribution of these data are then considered, including taphonomic factors and the impact of both research- and development-led archaeology. While post-depositional factors have undoubtedly had a significant impact on the availability of data, it is also possible to argue that the distribution of evidence is affected, as in the Iron Age, by the continued practice of an archaeologically invisible burial rite by a significant part of the Romano-British

However alternative sources exist for studying the relative distribution of burial evidence. The text and appendices of Philpott’s (1991) survey of RomanoBritish burial practice reference 923 cemetery sites. Philpott made no claim to have produced, and indeed deliberately eschewed a comprehensive listing of all known Romano-British cemeteries (1991: 2). His distribution maps were compiled ‘on the basis of a particular trait, which appeared to be reliably, though not necessarily comprehensively documented.’ (R. Philpott pers. comm.), and these were systematically collected from England alone. Only burials with particular grave good types, burial container or grave construction were used in analysis. References to poorly known, unquantified and badly recorded groups of Roman graves were omitted, but other significant categories were also excluded, for example cremation burials only accompanied by ceramics and unfurnished inhumation burials, the most frequently occurring burial types. The 13

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 2.2 Dated and undated burial / cemetery records in different data sources (SMR, JRS/ Britannia, Philpott)

distribution of cemeteries in Philpott’s sample is therefore biased to areas and periods where certain types of grave furnishing were more common. Philpott acknowledged (pers. comm.) that the absence of unfurnished inhumations was likely to cause the later Roman period to be under-represented in his survey. An advantage of this sample is however that its cemeteries are more precisely and reliably dated because of the prioritisation of grave goods as a focus of analysis.

100% 80% 60% 40% 20% 0%

An alternative source of data is the reporting of cemetery excavations in the annual summaries of fieldwork in Roman Britain from the Journal of Roman Studies (JRS) and Britannia. References to cemetery or burial excavations have been collected from JRS (Vols. 11 to 59, 1921-1969) and Britannia (Vols. 1 to 41, i.e. 1970 to 2010) from 1921 to 2010. These summaries have previously been used to explore changing patterns of excavation on other types of (non-burial) site (e.g. Evans 1995; Hingley 2000: 149-50; Wilkes 1989). These record the excavation of c. 13,250 burials from 1100 excavations. This is not, of course, an exhaustive source. In the earlier JRS surveys the lack of systematic archaeological monitoring of development work and of a comprehensive network of correspondents must mean that many burials went unreported. Comparison with urban and minor centre cemeteries published during the same period suggests that these reports underestimate the number of burials by c. 2,000. However in contrast to the Philpott sample the combined JRS / Britannia evidence allows a fairer representation of areas or periods where unfurnished burial was the norm. For more recent years (since the introduction of PPG16 in 1990) the Archaeological Investigations Project (AIP) provides an alternative source of data on fieldwork interventions and also demonstrates that reports on some projects are absent from the Britannia fieldwork summaries. The grey literature project has however shown the AIP also to require extensive filtering and supplementing with other sources before a complete overview even of recent fieldwork can be established (Fulford and Holbrook 2011).

Early

Inter

Late

U

Figure 2.3 Dated cemeteries from three counties (Hertfordshire, Hampshire and Cambridgeshire) in different data sources (SMR, JRS/ Britannia, Philpott) 100% 80% 60% 40% 20% 0%

Early

Inter

Late

Figure 2.4 Cemetery types from three counties (Hertfordshire, Hampshire and Cambridgeshire) in different data sources (SMR, JRS/ Britannia, Philpott) 100% 80%

Figure 2.1 Numbers of burial / cemetery records from five counties in different data sources (SMR, JRS/ Britannia, Philpott)

60% 40% 20% 0%

He. Ha. De. Ca.

C.

Co. 0

50 Phil.

100 JRS/Bri.

150

200

I.

C & I

U

The degree to which the datasets drawn from Philpott and JRS / Britannia are representative of the total available sample is assessed by comparison with data from five counties held in local authority inventories (Fig. 2.1). Figures from Devon and Cornwall, for example, confirm that the poorest represented counties in the JRS / Britannia or Philpott samples have a generally scant

SMR

14

DIGGING THE DEAD

Roman period burial record. The Cornish figure represents the maximum possible number of Roman cemeteries in the county, since long-lasting burial practice and limited dating evidence impede the attribution of many burials to a specific point from later prehistory to the early medieval period (as elsewhere in northern and western Britain). Comparison of samples from better-represented counties, Cambridgeshire, Hertfordshire and Hampshire, suggest that the Philpott and JRS / Britannia samples represent between approximately 15% and 30% of the total number of cemeteries recorded of likely Roman date. However since the inventories often record in multiple entries cemeteries which are represented in the Philpott and JRS / Britannia samples as single sites, these percentages represent an underestimate.

to it in Philpott’s appendices and where necessary by the publication. Data from JRS and Britannia have been quantified differently. Each reference in JRS / Britannia to the excavation of one or more burials is counted as a separate ‘cemetery’. This will produce some overrepresentation of large urban cemeteries excavated in several seasons, such as Dorchester Poundbury, Winchester Lankhills or the East London cemetery. References to excavation of infant burials have not been included in the cemetery listing as their reporting within the excavation summaries is very inconsistent and the phenomenon of infant burial on settlements has been extensively studied elsewhere (Scott 1999; Struck 1993c;). Total numbers of burials from larger cemeteries will include infant burials but these are too small to affect significantly the comparisons made here.

Comparison of cemeteries by date shows that undated cemeteries are under-represented in the Philpott and JRS / Britannia datasets compared to the HER (Fig. 2.2). When dated cemeteries only are compared (Fig. 2.3) there is less divergence. Little difference is also visible in the type of burials represented in the different samples (Fig. 2.4) (see below for the definition of terms). The principal exception is Hertfordshire, where the divergence may be explained by the large number of long used cemeteries in the county classified as ‘intermediate’ (see below) in provisional reports in JRS / Britannia but later more securely dated. The higher proportion of late inhumations in the Philpott sample from Hertfordshire contradicts the general tendency of late inhumations to be underrepresented in this source. This appears to be due to a relatively high frequency of provision of grave goods in the small sample of late Roman inhumation burials from this county (6.2.3). Within certain limits therefore, the Philpott and JRS / Britannia datasets can be used to assess patterns in the distribution of burial data. Where appropriate these must be compared to the fuller data sets to establish absolute patterns.

‘Early’ (E) cemeteries comprise those dated to the first and second centuries AD, though the use of some began in the first century BC and may extend into the third century AD. ‘Intermediate’ (Inter.) cemeteries span the second to third century transition, beginning either in the first or second centuries AD and ending in the mid to later third to early fifth centuries AD. The category includes cemeteries in which burial was sporadic or continuous over these periods. ‘Late’ (L) cemeteries are those dated between the third and early fifth centuries. The late category may sometimes include post-Roman burials, but cemeteries with only post-Roman burials have been omitted (see further 6.4.2). The final date category is Roman but otherwise unknown (U). The dates assigned to cemeteries follow those given by Philpott or by JRS / Britannia. The re-dating of older cemetery excavations has not been attempted. More secure dating of local or regional pottery and the increasing awareness of the possibility that samian vessels, the chief dating indicators for many cemetery assemblages, may sometimes be buried many decades later than their production suggest that the date of some burial assemblages may require re-evaluation (Wallace 2006). This is certainly possible where the material itself or adequate records still exist (e.g. Fabrizi 1984; Lyne 1994; Monaghan 1983) but is outside the scope of this study. Re-dating of this type is however unlikely to affect seriously the general conclusions based on the coarse dating parameters used in this study.

The following data for each cemetery were compiled from the Philpott and JRS / Britannia sources, site name, location (by county), date, type and context. The number of burials was also recorded from the JRS / Britannia entries only. For characterising distribution the sites were assigned to the counties in existence between 1974 and 1998. This was preferred to an attribution to civitas (cf. Struck 2001) because of the difficulty of establishing the boundaries of the latter. To avoid manipulation of a large number of very small samples the Welsh sample has been considered as a single unit. Here as elsewhere the number of Roman period cemeteries is likely to have been underestimated because of long-lived burial traditions. Scotland was not examined because similar problems are particularly acute (Alcock 1992); the burial data from the early first millennium AD in Scotland have been closely reviewed elsewhere (Maldonado Ramirez 2011: 82-97).

Of greater potential significance for distorting understanding of cemetery date is the use of conventional ‘early’ dating for cremation and ‘late’ dating for inhumation burials in the JRS / Britannia entries where no other apparent dating evidence is available or mentioned. It is now clear that substantial numbers of inhumation burials across the province are of early Roman date (see below) and in northern and western England the continuing practice of unaccompanied inhumation from the Bronze Age to the late or postRoman periods impedes dating where other evidence is not available (Crow and Jackson 1997; Faull 1977; Faull and Moorhouse 1982; O’Brien 1999). The same applies to Wales and Scotland (E. Alcock 1992; H. James 1992;

Terms are defined as follows. ‘Cemeteries’ from the Philpott sample may comprise one or more burials. The date and type of a cemetery, whether cremation, inhumation, or both, were established from the references 15

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

The number of burials reported in JRS / Britannia has also been recorded; in order not to count the same burials twice either numbers of burials excavated in each year or summary statements on the total number recovered over several seasons have been used. The excavation summaries do not consistently provide numbers of burials and quantification is hampered by vague references to ‘several’, ‘many’ or ‘hundreds’ of burials, especially in early JRS volumes. A much larger proportion of JRS (28%) than Britannia references (16%) lack information on numbers. Comparison of numbers of burials in the JRS / Britannia entries for larger (over 50 burials) urban and minor centre cemeteries with the numbers reported from those cemeteries in publication or archive suggests that the total from this site type is under-represented by c.1500 burials, i.e. more than 10%. Numbers of burials are also prone to distortion by a few extensively excavated sites. For example Ospringe, Water Newton, Springhead Pepper Hill and, in particular, Baldock provide over half of the minor centre burials and over two thirds of burials from military/vicus sites come from Caerleon Lodge Hill, Derby Racecourse, Brougham, Brough and Low Borrowbridge. This is a less significant problem for the urban sites, of which there several large cemeteries, and for villa and rural sites of which samples of larger than one hundred burials are scarce: Laxton and Wasperton represent rare exceptions.

Maldonado Ramirez 2011; Pollock 2006). The more widespread application of radio-carbon dating of cemeteries, mainly applied in southern England, has demonstrated a use that begins or continues into the midlater first millennium AD date of cemeteries which might otherwise have been considered as only late Roman (e.g. Bell 1990:80; Booth et al. 2007: 226-7; Chambers 1973; 1976b; 1987: 66; Hearn and Birbeck 1999; Holbrook and Thomas 2005; Leach et al. 2001; Rahtz et al. 2000: 422; Roberts 2001; Watts and Leach 1996; Weddell 2000). More occasionally early Roman dates are also given (2.3). The particular difficulty of dating individual unaccompanied burials is well illustrated by an example from Ferrybridge, where radio-carbon dating of a crouched burial in a Roman period enclosure, from which no post-Roman evidence was otherwise recovered, suggests a 7th century AD date. Use of narrower dating brackets has produced a more nuanced understanding of the distribution of burial data over time, both in Britain (e.g. Struck 2001) and neighbouring provinces (e.g. van Doorsaeler 1967: 29-30; Polfer and Thiel 1997). However these analyses depend on a quality of data only available from a small number of cemeteries, and also have to take into account differential visibility of different periods caused by fluctuation in the general availability of the diagnostic chronological indicators, especially samian (Willis 2004) and by the relationship of ceramic assemblages from burial to those from other sites; Tuffreau-Libre (2000) for example has noted the absence of samian from a number of first century AD northern French cemeteries despite its presence on contemporary local settlements. Whilst the coarseness of the date ranges used for this analysis must be acknowledged, the method adopted here retains the largest possible sample for study.

The cemetery context is classified according to five types, (i) urban, (ii) minor centre, (iii) military/vicus, (iv) villa and (v) rural. Other categories, for example temple or prehistoric site were discarded during compilation as the number of cemeteries was very small and the references were subsumed within other categories. ‘Urban’ sites comprise the coloniae, municipia and civitas capitals, ‘minor centres’ the small towns, roadside settlements or local centres listed by Burnham and Wacher (1990), Millett (1990a: 152-3) and R. F. Smith (1987: 126-9). The ‘military / vicus’ category includes fortresses, forts and the settlements attached to them, the latter those sites listed by Sommer (1984) but excludes cemeteries possibly linked with early military phases of urban sites and minor centres, as it is very difficult to distinguish military-associated burials from others on such sites (Jones 1984b; Struck 2001). These have been classed with the relevant urban or minor centre. It is also not easy to distinguish urban and rural cemeteries. The attribution in the excavation summaries has usually been followed but since urban cemeteries could extend for up to several hundred metres outside large towns, isolated or small groups of burials recovered on the fringes of modern successors to Roman towns, here classified as rural may have been those of town dwellers. To separate ‘villa’ and ‘rural’ burials begs the thorny question of rural site typologies (Hingley 1989). Burials are attributed here to villas if the JRS / Britannia or subsequent report explicitly relates them to a ‘villa’ when a building of ‘architectural pretension’, usually of stone masonry, is implied. The ‘rural’ category includes villages, individual or small groups of structures, production sites, fields, track and droveways and individual or small groups of burials with no other

My dating categorisation introduces some further biases, in particular a possible smoothing of chronological trends by the ‘intermediate’ category. This category does not take account of different numbers of burials from different periods or of the continuous or sporadic use of the cemetery, although this is balanced with discussion below of the relative importance of different phases within the category. The classification system may also contribute to the reification of the difference between the early and late Roman periods, but in many urban cemeteries at least the late second or early third century represents a significant hiatus in cemetery use (Esmonde Cleary 1987: 175-76; see also chapters 4 and 5 below). Four categories of cemetery type have been used, cremation (C), joint cremation and inhumation (C&I), inhumation (I) and unknown (U). The number of joint cremation and inhumation cemeteries has an artificially high representation, as in many instances one type predominates throughout the use of the cemetery or in a particular phase of use. Where appropriate the balance of cremation and inhumation burials within this category is scrutinised in further detail.

16

DIGGING THE DEAD

contextual information. Rural sites may be underrepresented as individual or small number of burials from non-cemetery excavations are perhaps less likely to be noted in fieldwork summaries reported to Britannia than those from larger urban cemeteries.

recorded in similar proportions, but if the small samples from the middle years of the 20th century are disregarded, from the mid-1930s twice as many inhumations as cremation cemeteries have been consistently reported (Fig. 2.6).

2.3 The distribution of Roman period burial evidence Romano-British graves have been ‘excavated’ from the Roman period onwards but only with the regular publication of excavation summaries in JRS can the frequency of cemetery excavations be quantified (Fig. 2. 5). It has followed the same trends as excavations of other types of Roman site during the 20th century, diminishing in the war years and rising in the latter third of the century to a level over three times as great as in earlier decades (cf. Hingley 1989: 5; 2000: 149; Wilkes 1989: 247-9). The cemetery sample does not reflect the slight decrease in the number of all Roman period excavations in the 1980s (cf. Wilkes 1989: 247-9). In the ten years from 2001-2010 at least 2280 burials were excavated from 248 cemeteries, though in decreasing numbers in the later years of the decade. The last two decades have been especially rich for Roman funerary archaeology because of the focus on the margins of historic centres in redevelopment work in towns and the scale of infrastructure projects and associated excavation of Roman rural settlements and landscapes; funerary evidence was documented on average in 10% of interventions across sample areas in the period 1990-2004 (Fulford and Holbrook 2011: 336-8). In the earliest reports from JRS cremations and inhumations were

Many cemeteries in both the JRS / Britannia and Philpott samples are not more closely dated than ‘Roman’, especially in the JRS sample (47%) (Fig. 2.7). The different samples otherwise show similar period biases, though later Roman cemeteries are less well represented in the JRS sample than in the others. The difference between the early and later periods becomes more marked in the numbers of burials rather than cemeteries, with the majority dating to the late period. The proportion of undated burials is smaller than of undated cemeteries, as isolated or small groups of burials account for most undated examples, 23% in the JRS/Britannia sample (Fig. 2.8). The high proportion of dated burials in the ‘intermediate’ period is a product of a small number of long-used large urban and minor centre cemeteries within this group. Investigation of the more precise dating of burials in the intermediate category suggests that they divide approximately evenly between the early and late periods. With this adjustment, and omitting burials of unknown date, 38% of burials date to the early and 62% to the late Roman period in this sample. In the more recent and more fully documented Britannia data early Roman burials account for 31% and late Roman for 69% of the dated instances.

Figure 2.5 Burial / cemetery excavations recorded from 1920 to 2010 in JRS / Britannia

160

140

120

100

80

60

40

20

0

17

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 2.6 Percentage of burial / cemetery types recorded from 1920 to 2010 (JRS / Britannia)

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

C

I

C&I

U

burials in the Britannia sample. A better quality of dating evidence as well as a realisation of the variety of burial practice, has increased awareness of the possible early date of inhumations. The lower representation of early inhumation examples in the JRS sample (not plotted) reflects an earlier dependence by excavators on inhumation as an indicator of late Roman date.

Inhumation cemeteries predominate in the JRS / Britannia and Philpott datasets (Fig. 2.9). The greater likelihood of cremation cemeteries to have been furnished with grave goods explains their slightly higher representation in the latter sample. When numbers of burials are counted the proportion of burials in joint cremation and inhumation cemeteries assumes a much greater importance (Fig. 2.10). If the burials in this joint category are assigned to the different categories, cremations account for c. 29% of the total and inhumations 66%. In the Britannia sample alone 55% of cemeteries and almost three quarters of burials are of inhumations.

Whilst inhumation can be documented as an early Roman burial rite all across the province in every type of settlement context (Struck 2001) there is also a regional dimension to its significance. Inhumation is the majority burial practice of early Roman southern Dorset, through the frequent presence of dated grave goods, a continuity of the so-called ‘Durotrigan’ tradition. From the presence of only two cremation burials in a sample of fifty cemeteries from Somerset and north Dorset, and from the occasional presence of first and second century AD ceramics with otherwise findless inhumation burials, Leech (1980: 341) argued that inhumation was probably always the majority tradition in a broader area of early Roman south-west England. There are increasing indications that the same is true of other southern and central English counties with few cremation burials and, previously, a high proportion of undated inhumations. Recent excavations in Gloucestershire / Avon for example have also demonstrated the predominance of inhumation as a burial tradition on rural sites as well as

When the number of cemeteries of different type from each period is quantified, taking only the Britannia data for which there is more information on date (Fig. 2.11), the high number of joint cremation and inhumation cemeteries obscures the relative importance of the different rituals. The cremation and inhumation burials from joint cemeteries were therefore separated (Fig. 2.12). Both numbers of cemeteries and numbers of burials sustain the orthodoxy of the prevalence of cremation in the first two centuries AD, being replaced by inhumation as the dominant burial practice at some point during the third century (Jones 1981). However inhumation burials can be shown to account for a more substantial fraction of the early Roman burial record than hitherto appreciated, 29% of cemeteries and c.24% of

18

DIGGING THE DEAD

on urban sites (e.g. Gloucester London Road, Tockington Park Farm; Bishop’s Cleeve; Hucclecote). Even in southeast England excavations in London’s eastern cemetery and in small town cemeteries at Baldock (Herts) and Pepper Hill (Kent) demonstrate that inhumation can sometimes comprise a high proportion of early Roman burials (see also Farley and Wright 1979). In summary the evidence from Britain suggests an increasingly

localised and complex relationship between cremation and inhumation. This echoes the variety of practice in other provinces (e.g. Faber et al. 2004; Vidal 1992) and in the metropolis itself (e.g. Catalano et al. 2001). A wider application of radio-carbon dating of unaccompanied inhumations (see above, 2.2) will help to add further nuance to this picture.

Figure 2.7 The percentage of dated cemeteries in different data sources (JRS, Britannia, Philpott)

Figure 2.10 The percentage of burials in different types of cemeteries (JRS / Britannia)

100%

100%

80%

80%

60%

60%

40%

40%

20%

20%

0%

0% JRS

Brit.

Early

Inter.

Phil. Late

C.

U

I. JRS

Figure 2.8 The percentage of dated burials in different data sources (JRS / Britannia)

C & I Brit.

U

Comb.

Figure 2.11 The percentage through time of different cemetery types (Britannia)

100%

100%

80%

80%

60%

60%

40%

40%

20%

20%

0%

0% JRS

Brit.

Early

Inter.

Comb. Late

C.

U

I. Early

Figure 2.9 The percentage of cemeteries of different types in different data sources (JRS, Britannia, Philpott)

C & I

Inter.

Late

U U

Figure 2.12 The percentage through time of different burial types (Britannia simplified)

100%

100%

80%

80%

60%

60%

40%

40%

20%

20%

0%

0% C.

I. JRS

C & I Brit.

U

Early

Phil.

Inter. C.

19

Late I.

U

U

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Rural sites are, by a narrow margin, the best represented of site types in cemetery samples (Fig. 2.13), contrasting with their low representation (9%) in all fieldwork recorded in JRS and Britannia to 1995, as extrapolated from data given by Hingley (2000: 150). In contrast to other aspects of rural settlement archaeology (Hingley 1989: 4-5; 2000: 150; Taylor 2007), villas do not predominate in this sample, although given the diversity of contexts included in the ‘rural’ sample too much

weight should not be placed on this. However a strong bias to urban sites emerges from quantification of numbers of burials where the share taken by the rural and villa samples is greatly reduced (Fig. 2.14). The latter are only usually recovered in small numbers in excavation of rural settlement sites (Chapter 6) whereas the urban sample is derived from a relatively small number of large scale cemetery excavations.

Figure 2.13 The number of cemeteries from different settlement types (JRS/Britannia)

Figure 2.16 The percentage of dated burials at different settlement sites (JRS / Britannia)

600

100%

500

80%

400

60%

300

40%

200

20% 0%

100

Urban

0 Urban

Mil/Vic. Min. Cen.

Rural

Mil/vic. Min. cen. Early

Villa

Inter.

Rural

Late

Villa

U

Figure 2.17 The percentage of cemetery types from different settlements (JRS/ Britannia)

Figure 2.14 The number of burials from different settlement types (JRS/Britannia) 7000

100%

6000

80%

5000

60%

4000

40%

3000

20%

2000

0%

1000

Urban

0 Urban

Mil/Vic. Min. Cen.

Rural

Mil/vic. Min. cen. C.

Villa

I.

C & I

Rural

Villa

U

Figure 2.18 The percentage of burial types from different settlements (JRS / Britannia)

Figure 2.15 The percentage of dated cemeteries at different settlement sites (JRS / Britannia) 100%

100%

80%

80%

60%

60% 40%

40%

20%

20%

0%

0%

Urban Urban

Mil/vic. Min. cen. Early

Inter.

Late

Rural

Mil/vicus

Villa

U

C.

20

Min. centre I.

U

Rural

Villa

DIGGING THE DEAD

Figure 2.19 The distribution of Roman period cemeteries by county (JRS / Britannia)

Figure 2.21 The distribution of Roman period inhumation cemeteries (JRS / Britannia)

Figure 2.20 The distribution of Roman period cremation cemeteries (JRS / Britannia)

The number of cemeteries recorded in the JRS / Britannia dataset associated with different types of settlement varies by date, although it should be remembered that the dates of a large proportion are unknown (Fig. 2.15). There is a general bias to the late Roman period, with the exception of military / vicus and rural sites. Quantification of numbers of burials again illustrates the predominance of the late period in the urban and villa samples (Fig. 2.16). Given the high proportion of urban burials within the overall sample, the late urban bias is responsible for the overall bias to the late Roman period in the JRS / Britannia sample (Fig. 2.8). The proportion of intermediate and late burials is greater than cemeteries in the rural sample (Fig. 2.16), but the most striking difference lies in the massive increase in representation of the early period from the military/vicus sites and of the intermediate period on military /vicus and minor centre sites. The proportion of early and intermediate burials from military/vicus sites is heavily affected by large samples from Brougham, Caerleon Lodge Hill and Derby Racecourse, while the minor centre sample is dominated by excavations at Baldock, Ospringe, and Springhead Pepper Hill (see above). The higher proportion of undated burials in the rural and villa sample is due to the recovery of isolated or small groups of burials without stratigraphic or artefactual evidence to permit more precise dating (6.4). Inhumation cemeteries predominate on urban, villa and rural sites by a ratio varying between 2:1 and 8:1, while cremation cemeteries are better represented at minor centres and especially military/vicus sites (Fig. 2.17). The comparison based on numbers of burials (cremation and 21

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

samples occasionally deriving from some minor centres, including Water Newton, Ashton and Ancaster. Similar differences between south-eastern and the rest of southern and central England are visible in the distribution of early and late cemeteries, although to a lesser degree because of the omission of the large number of undated inhumation cemeteries. The use of the county as the smallest unit of analysis disguises significant local variation, as discussed in the analyses of burial data from Hertfordshire and Hampshire (6.2, 6.3).

inhumation burials from joint cemeteries are again separated and re-assigned) shows similar tendencies (Fig. 2.18). The exceptional patterns at military / vicus sites and minor centres are created by the same few large cemeteries discussed above. The total number of cemeteries and the number of cemeteries of different date and type were calculated for each of the English counties and Wales as an index of regional variability. The distribution of burials in the Philpott and JRS / Britannia samples is very similar and only the latter, which better reflect the impact of fieldwork in cities and on infrastructure projects, are shown here (Fig. 2.19), confirming current assumptions concerning the distribution of Roman period burial evidence (2.2 above). Three broad zones can be differentiated (see also Taylor 2007: 36, Fig. 4.1). The best represented counties are those from south east England, especially Kent, Greater London, Hertfordshire and Essex (although parts of the south-east lack burial data, for example East Sussex and Surrey) and the least those from northern and western England and Wales. Between the two an intermediate zone from Dorset and Hampshire stretches north-east to Lincolnshire. In Philpott’s zone the better representation of North Yorkshire and Dorset is a product of regional variability in grave furnishing, the high representation of the former being a product of the many furnished burials recorded at York in the nineteenth century and of the latter a consequence of rich furnishing in the Durotrigan burial tradition which persists into the late first and early second century AD. Conversely the lower representation of Humberside, Northamptonshire, Lincolnshire and Norfolk in the same sample is due to less frequent furnishing of inhumation burials.

2.4 Factors influencing the distribution of burial evidence A range of extrinsic factors, including burial practices, grave protection and depth, aspects of post-depositional burial environment such as pH, drainage, temperature and oxygen availability, flora and fauna, post-depositional disturbance and erosion, and bias in excavation and reporting have influenced the formation of this like other funerary samples (Garland and Janaway 1989; Henderson 1987; Waldron 1987; Mays 1991; 1994a, 1994b). The effects of individual factors, as well as their cumulative impact, cannot yet be clearly modelled but their general scope and impact can be considered at different stages from burial to publication; Original population  Burial practice  Taphonomic process  Excavation and Recording  Analysis and Publication.

The increase in numbers of cemeteries excavated in the last 30 years (Fig. 2.5) is, unsurprisingly, not uniform across England and Wales. The JRS/Britannia data show an increasing focus of excavation on south-east England in the later 20th and early 21st century; for example almost one third of the Roman cemetery excavations from 1999 to 2005 took place in London and Kent, and a substantial further number on the Essex-Cambridgeshire border. By contrast in the JRS pre-war sample different biases are evident, for example the large numbers of burials noted in Cambridgeshire and Dorset, the likely product of strong local archaeological fieldwork traditions as well as of the visibility of local burial rites with frequent artefact deposition.

The influence of Roman period population distribution of course influences the distribution of burial data, but it is argued in the following paragraphs that it is not currently appropriate to use numbers of cemeteries or burials as even a crude demographic indicator. The relative representation of different areas and periods is very much dependent on the relative archaeological visibility of different burial practices. The possibility of a continuing tradition into the Roman period of a non-archaeologically recognisable burial style will be considered below. The relatively constant ratio of cremation and inhumation cemeteries recovered from the middle of this century onwards (Fig. 2.5) suggests that whatever the causes behind recovery biases, their combined effect continues to operate at a fairly constant rate. The factors potentially affecting the visibility of different burial types will not necessarily have had a consistent impact. While human bone from cremation burials is better preserved than inhumations in a wider variety of burial environments (Mays 1998: 209), the burials themselves are often shallower and, especially when not in a durable container, more vulnerable to post-depositional disturbance (Ferdière 1993d: 436). Without such a container or grave goods they may also have been less visible to antiquarian or earlier archaeological investigation than inhumations

The general distribution map masks differences in the availability of burial evidence of different types, dates and contexts. Figs. 2.20 and 2.21 show a concentration of cremation cemeteries in south-east England, especially Essex, and a more even distribution of inhumation cemeteries across southern and central England. The density of burial evidence in Gloucestershire, mainly inhumations, has been produced by relatively large-scale fieldwork on the cemeteries of both Gloucester and Cirencester in recent decades. Elsewhere in central England rural cemeteries predominate, with larger 22

DIGGING THE DEAD

cemeteries remains to be identified and the corpus of burial evidence from military sites in Wales and Scotland is exiguous (Arnold and Davies 2000: 133-140; Collard and Hunter 2000; Petts 2009; Pollock 2006). The principal evidence for cemeteries associated with military sites is concentrated in a very small number of sites, of which two of the three largest published samples come from sites further south (Caerleon Lodge Hill, Derby Racecourse). In these same regions the limited attention to rural settlement means that the burial sample associated with non-military settlements is even smaller (e.g. Roberts 2001: 279).

(Philpott and Reece 1993: 418). More recent careful excavations of some urban (East London; St Albans St Stephen’s), minor centre (Baldock), and rural (Owslebury) cremation cemeteries have suggested the potentially high frequency of this burial form. On the other hand older discoveries of inhumation burials without grave goods or obvious connection to a Roman period site are perhaps more likely to have passed unreported. The visibility of burial types has also been demonstrated to be affected by selection in research. The significance of unaccompanied cremation and inhumation burials and thus of late Roman graves is underestimated by Philpott’s focus on burial furnishing. The overrepresentation in his sample of early cremation cemeteries has been shown by the comparison with the JRS / Britannia samples (Figs. 2.7 and 2.9). The difference made to archaeological visibility by burial furnishing, where skeletal preservation is relatively poor is illustrated by comparison of the massive numbers of Anglo-Saxon burials recorded in Norfolk and Suffolk (Lucy 2000: 2) with the small number of Roman period cemeteries from these counties.

The representation of ‘rural’ sites is higher in the cemetery sample than in the total excavated sample analysed by Evans (17%). This is partly a product of different methods of compilation of data: Evans does not include field systems for example, and his ‘other’ category includes sites which are classified here as ‘rural’. Conversely villas are much better represented in the general excavation sample (14%) than among excavated cemeteries. Excavation practice arguably affects the visibility of burial evidence from both villa and rural sites. Only excavations which intensively examine the spaces in which rural burials are located, in particular on settlement margins (6.4.3), are likely to recover burial evidence. The lack of burial evidence from villas must be due to the bias towards principal buildings in older excavations (Hingley 1989). Large-scale development-led excavation of rural sites may sometimes have privileged major archaeological features and relationships over single or small groups of small features such as burials.

The cemetery distribution described above broadly can be partly attributed to preservation factors, acidic soils accounting for the relative absence of evidence from northern and western Britain (e.g. Philpott 1991: 5; Philpott and Reece 1993: 418). Like all generalisations about the ‘Highland’ and ‘Lowland’ zones, this requires qualification. Inhumed burials have also not survived in localised areas of southern and eastern England with more acidic or very free draining soils, this perhaps contributing to the relatively small number from Hertfordshire (6.2). Systematic analysis of regional bias in human bone preservation has only so far been surveyed for certain areas (e.g. Mays 1991; 1994a; 1994b). Conversely ploughing for arable agriculture has had a much more significant impact on the archaeology of eastern and southern Britain than on that of the northwest, although this has often brought cemeteries to archaeological attention in the process of their destruction.

The regional distribution of Roman period burial evidence is in broad agreement with the distribution by county of all excavations (totalling 10,308) on Roman sites in England held on the RCHME database by 1994 (RCHME 1995: 316, Fig. 4), but there are some discrepancies which deserve further attention. Northern counties with a more substantial number of excavated Roman sites, for example Cumbria, Northumberland, Derbyshire and Lancashire also lack burial evidence. This can again be attributed to the research agendas of military archaeology (see above) and to the problem of attributing burials to the Roman period within a context of mortuary rituals that do not significantly change over two millennia (2.2). The over-representation of Hertfordshire in the burial sample appears to be due to a particular antiquarian and archaeological concentration on burial practices within the county.

The bias which has had the most clearly demonstrable impact on the availability of burial is the nature and distribution of archaeological fieldwork. Comparison of the distribution of burial data by site type (Figs 2.132.14) to Evans’ (1995) quantification of the percentage of excavations on 3016 different site types recorded in Britannia from 1969 to 1989 shows the predominance of urban sites in both samples but military sites, especially those from northern English counties, account for a much higher proportion of all excavations (25%) than of the cemetery sample. The recovery of burial evidence, as opposed to the buildings of forts and fortresses, has had a low priority in the research agenda for military archaeology. In synthetic works especially, little attention has been paid to burial evidence, even though much of the epigraphic information on the frontier garrisons and vicus inhabitants comes from a mortuary context. From many Hadrian’s Wall forts, for example, the location of

2.5 The implications of the distribution of burial evidence The above discussion has given an overview of the distribution of burial evidence and has discussed several of the processes that contributing to it, key aspects of which remain poorly understood. While the complexity of these processes and their interrelationship should be acknowledged, some implications of this distribution of evidence for the study of burial practice can be addressed.

23

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

is not the total sample of burials to have been excavated, but it is difficult to extract further quantitative data for this type of analysis from earlier records (2.2). Infant (under one year old) burials are substantially underrepresented in Romano-British cemeteries, although their representation in cemetery populations varies substantially (Watts 1989; Pearce 2001). They might be expected to account for 25-30% of all burials in a preindustrial population at the margin of demographic viability (Hopkins 1966). No quantitative allowance is made for infant burials here but their under-representation in this sample should be noted.

The relationship of the sample to the original burial population from Roman Britain can first be estimated (Table 2.1). The following exercise uses Millett’s (1990a: 185) population estimates for late Roman Britain (rounded to the nearest 50,000) and a death rate of 25/1,000 p.a. based on some pre-modern populations and used here only as a basis for generating an estimate for the number of burials made during the four centuries of the province’s existence (R. Jones 1977; Morris 1987: 74; Wrigley 1969: 162-3). No account is taken here of potentially different mortality rates between urban and rural populations. With the exception of the Dorchester comparison (see below, 3) the numbers of burials are derived from the combined JRS / Britannia sample. This Table 2. 1 Burial populations and ‘real’ populations 1. Total burial population RB population x expected mortality rate = annual death rate: Annual death rate x length of Roman period = predicted burial population Number of excavated burials (JRS / Britannia) as proportion of predicted population

(3,650,000 / 1,000) x 25 = 91,250 91,250 x 400 = 36,500,000 13,000 / 36,500,000 = 0.035%

2. Urban burial population (‘urban’ sites and ‘minor centres’) RB urban population x expected mortality rate = (240,000 / 1,000) x 25 = 6,000 annual death rate: Annual death rate x length of Roman period = 6,000 x 400 = 2,400,000 predicted urban burial population Number of excavated burials from towns (JRS / 8307 / 2,400,000 = 0.35% Britannia) as proportion of predicted population Number of late Roman excavated burials from towns 4450/ 1,200,000 = 0.37% (JRS / Britannia) as proportion of predicted population 3. Late Roman Dorchester - burial population (sources for number of documented burials: Davies et al. 2002; Davies & Thompson 1987; Dinwiddy & McKinley 2009; Farwell & Molleson 1992; Green et al. 1981; Startin 1981) Dorchester population x expected mortality rate = Minimum (2,000 / 1,000) x 25 = 50 annual death rate: Maximum (5,000 / 1,000) x 25 = 125 Annual death rate x length (years) of late Roman Minimum 50 x 200 = 10,000 period = predicted Dorchester burial population Maximum 125 x 200 = 25,000 Number of excavated burials from late Roman Minimum 1,700 / 10,000 = 17% Dorchester as proportion of predicted population Maximum 1,700 / 25,000 = 6.8% 4. Rural burial population (rural + villa totals) RB rural population x expected mortality rate = annual death rate: Annual death rate x length of Roman period = predicted burial population Known rural burial population as proportion of predicted population

(3,300,000 / 1,000) x 25 = 82,500 82,500 x 400 = 33,000,000 2,747 / 33,000,000 = 0.008%

Note. For ease of calculation a rounded figure has been given for the total number of burials reported in JRS / Britannia. That this does not represent the total number of excavated Romano-British burials should not be forgotten; for example c.200 burials can be estimated as ‘missing’ from the urban sample reported in JRS / Britannia (2.2). This does not however affect the orders of magnitude in the calculation.

24

DIGGING THE DEAD

Rather than interpret these data in demographic terms, an alternative argument is cautiously advanced here, i.e. that a contributory factor to the low number of burials hitherto excavated from rural sites in Roman Britain is the continuity of archaeologically invisible burial practices from the Iron Age into the Roman period. The evidence suggests that as in the Iron Age (Whimster 1981) many lacunae exist in the burial record of the Roman period. There are areas with generally good preservation conditions and a relatively high intensity of archaeological activity where the burial record from rural sites is heavily biased to inhumation burials, usually of late Roman date. This applies, as argued above, to a belt of counties from Hampshire, Wiltshire and Somerset to Lincolnshire and East Yorkshire. Arguments from negative evidence are dangerous and the biasing factors affecting the visibility of burial evidence have been outlined. Two caveats come immediately to mind. Individual excavations may reveal large numbers of early burials in areas where rural burial evidence was previously limited or absent, for example at Owslebury (Hampshire) (Collis 1977b) or at sites in rural Avon and Gloucestershire (e.g. Masser and McGill 2004; Parry 1999; Thomas et al. 2003). Inhumation is not by itself a reliable dating indicator. In most areas of Britain it can also be documented as an early Roman burial tradition; in some areas accounting for a very significant proportion of first and second century burials (2.4). However in certain regions these caveats need not apply. In the Upper Thames valley, for example, frequent large-scale excavation of rural sites has taken place, revealing a substantial corpus of inhumation burials and cemeteries. Radio-carbon dating has been consistently applied to such cemeteries and found them to be universally of late and post-Roman date. In this area there are many early Roman settlements, but almost no burials (Booth 2001; Booth et al. 2007: 226-7). An analogous situation many pertain in Wiltshire, where although there are many undated examples, artefactual evidence suggests a late Roman date for most inhumation (Foster 2001: 165).

In total the JRS / Britannia sample (rounded to 13,000) represents 0.035% of the estimated original number of burials (Table 2.1). Changing the population estimate to the higher or lower estimates others have preferred will not change the order of magnitude. Given the biases identified above in the distribution of evidence, it will be anticipated that the proportion of the original population represented among burials will be higher in certain contexts. If early and late Roman periods are considered separately the latter is characterised by a slightly higher proportion of original burial numbers (c.0.04%). The most striking effect is achieved by considering urban and rural burials separately. The number of burials from urban cemetery excavations is equivalent to 0.35% of the estimated original urban burial population, a proportion that rises slightly for the late Roman period. The estimates for late Roman Dorchester, the number of burials being derived from recent well documents excavations, show that the proportion (6.8%-17%) is even higher at urban sites where cemetery excavation has been most extensive. Dorchester is not atypical in a much higher representation of late than of early cemeteries. Conversely the bias to excavation of urban cemeteries means that the ratio of rural burials to the predicted original burial population is over forty times lower than for urban burials. The small size of the burial sample per se need not frustrate broader extrapolation but it is the combination of sample size plus the biases within it that counsel caution here, particularly with regard to analysis of population size. One should look for demographic trends in Roman Britain in the rural areas where the vast majority of the population lived, but the currently available sample of rural burials represents a very small evidence base with a substantial regional bias. Even within the better studied contexts, for example urban cemeteries, caution is necessary. Leaving aside considerations of excavation bias, an interpretation of increased numbers of burials from late Roman towns as urban population growth is unacceptable, since ‘urban’ cemeteries need not have included only townsfolk (Millett 1990a: 142). It is not currently possible to distinguish the burials of town and country dwellers. Rural and urban burial rites are similar (Esmonde Cleary 1992) and the pathology-based methodology offered by Molleson (1992) is unsatisfactory as urban and rural populations are unlikely to have had distinct occupations to the degree to which she suggests. The rate of population change which could be extrapolated from changing numbers of burials in some areas would also be of a degree which would not be supported by any other form of archaeological evidence or by feasible demographic models (cf. Morris 1987). For example the number of late Roman cemeteries recorded in the JRS / Britannia sample declines in Essex and Hertfordshire by c.300% and rises in Somerset by c.1400%. Although Roman period cemeteries have occasionally been used for demographic reconstruction (e.g. van Doorsaeler 1967; Wightman 1985) the caution exercised both within and beyond the Roman period (Morris 1987; 1992) is therefore more appropriate.

The shift towards burial of the dead in formal graves in separate cemeteries has been argued to be one manifestation of the increasing emphasis on the individual in the social practices and material culture of the later Iron Age (Hill 1995: 122-4). This survey of the distribution of Romano-British burials has given an indication of the possible time depth of this process in the sphere of burial, extending well into the 1st millennium AD in some parts of rural Roman Britain. The duration of this process varies by region and the trends in the manifestation of a visible burial tradition are not always linear. For example East Yorkshire has a well characterised Mid-Late Iron Age and early medieval burial tradition (Stead 1991; Lucy 1998) but much less is known of Roman period burial, in contrast to the settlement archaeology, for which Roman period evidence is abundant. In Kent the Channel Tunnel excavations have shown a very strong skew to the late Iron Age and early Roman period in terms of human burial evidence from rural contexts (Booth et al. 2011: 312-4). It is difficult to differentiate the impact of the 25

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

The biasing factors on the distribution of evidence include both the character of burial environments and the distribution of research- and development-led excavation. The characterisation of burial evidence given above is based on samples of which the shortcomings have been noted; fuller collection of data in a more fully processed form may well nuance the patterns noted above, at last on a regional or context basis for example through the ‘grey literature’ projects (Fulford and Holbrook 2011). To a limited extent an expanded sample will result from gradual accretion of excavation and publication, but without a more targeted approach to maximise opportunities to expand the sample of burial the existing biases in the dataset will be perpetuated.

adoption of Roman customs from the general trend towards archaeologically visible burial in the late first millennium BC and first millennium AD. The same problem characterises burial as other areas of study, i.e. to distinguish the effects of incorporation in the empire from more general patterns of long term change (Reece 1990). In evaluating the impact of Roman culture on local burial traditions, the rituals, grave furniture and monument type have been accorded primary significance by most scholars (1.3.1; e.g. Alcock 1980; Black 1986; Philpott 1991) but the emergence of archaeologically visible individual burial should perhaps receive equal if not greater weight. The assumption that archaeologically visible Roman burials represent a sample drawn from the burial practices of the whole population is also the basis for differentiating burials with a view to documenting social hierarchies (e.g. Jones 1983; Philpott 1991). In this context unaccompanied cremation or inhumation burials may be interpreted as those of the lowest social stratum. However on the evidence assembled here, archaeologically visible burial in Roman Britain represents, at some times and places, the practice of a minority. Thus the characterisation of those few, often isolated unaccompanied inhumation burials as those of the lowest social stratum deserves re-consideration, especially in areas where they comprise the majority of visible burials as is often the case outside south-east Britain. Instead they arguably represent a privileged minority to whom a burial rite requiring greater investment of energy was extended. Although on the basis of grave good deposition late Roman burial traditions in Britain may appear impoverished, on this evidence it is a period during which an archaeologically visible burial is extended to an increased proportion of the population.

On the basis of the survey it has been argued that the numbers of cemeteries and burials from Britain are not a suitable indicator of change in population size and distribution. On a more positive note further insights into ritual change have been derived from it. The chronological relationship between cremation and inhumation is more complex than previously acknowledged. In particular inhumation burials comprise a substantial minority of late Iron Age and early Roman burials in south-east Britain as well as other areas. It has also tentatively been proposed that they do however provide further evidence for a slow and regionally variable emergence of an archaeologically visible burial tradition, in which the Roman conquest does not mark a significant threshold. Whimster’s prediction (1981: 195) that ‘analysis of post-conquest native burial traditions will eventually reveal many of the same lacunae that we have had to face in the pre-Roman period.’ is supported by the evidence collected here. In much of north-western Europe too archaeologically visible burial practice is difficult to identify in the late Iron Age and the burial traditions identified through archaeology may not be those of the majority of the population. The variation documented over time and space in north-east Gaul illustrates the considerable regional variability in the emergence of such a practice (e.g. van Doorsaeler 1967: 29-30; Hessing 1993b; Pion and Guichard 1993; Polfer and Thiel 1997; Willems 1984).

2.6 Conclusion Several coarse-grained trends have been suggested in the distribution of cemetery evidence by date, type, site association and region, using primarily samples of data taken from Philpott (1991) and the JRS / Britannia fieldwork summaries, with greater emphasis being placed on the latter. The distribution of burial data from Roman Britain is clearly uneven. Late Roman inhumation burials predominate in the excavated burial sample. Many more burials have been recorded from Roman towns and ‘minor centres’ than from other site types. Although as many cemeteries have been recovered from rural as from urban sites, only half as many burials have been recovered from a rural context: since the majority lived in the countryside proportionately the rural population is therefore much more limited. More excavation of burials and cemeteries has taken place in south-east England than in other areas and the regional distribution of burials of different type and date is uneven, cremation burials of early Roman date being more common in south-east and, to a lesser extent, north-west England and inhumation burials, often of later Roman date, having being more frequently excavated in central and southern England.

The gradual emergence of archaeologically visible burial, a likely combination of change in burial ritual and location, has significant implications for the social interpretation of burial, in particular for the establishment of social hierarchies. Relative social status, whether realised or claimed, is usually read off from variation in the treatment of the body or grave furniture, but variability must be read in relation to context. If, as is argued here, ‘simple’ unfurnished cremation or inhumation is the only form of burial evidence from a particular area, interpretations as burials of lower-status individuals cannot be easily sustained. This point is particularly relevant for rural settings, where single or small groups of such burials n site margins have been so construed in previous scholarship. This is explored further in considering the relationship between burial and settlement space in rural Roman Britain in chapter 6 (6.4.7). 26

CHAPTER 3: THE EVIDENCE FOR ROMAN BURIAL RITUALS, WITH SPECIAL REFERENCE TO CREMATION 3.1 Introduction The previous chapter considered the distribution of Roman period burial data from Britain and its implications. This chapter shifts focus to address another characteristic of data availability, namely the stages of funerary ritual to which the recorded archaeological evidence pertains, with particular emphasis on cremation rituals from Roman Britain and beyond. Interment, where it does take place, is the culmination of what may have been an extended sequence of funerary rites. The potential richness and complexity of such rites is clear from ethnographic or historical accounts in general (e.g. Huntington and Metcalfe 1992). Roman documentary and visual evidence too reveal aspects of the ceremonies that see emperors and less often others to their ‘long home’, as well as subsequent commemorative rites (Davies 2000; Hope 2009; Maurin 1984; Scheid 1984; 2005: 161-88). Such evidence derives predominantly from Rome and, with exceptions (e.g. Montserrat 1997), is not met elsewhere save in unusual circumstances. In the northwest provinces the ‘Testament of the Lingon’, a tenth century manuscript transcription of a will in which the conduct of the cremation and commemoration of a Gallic magnate are specified, is the sole surviving description of funerary rituals not related to an imperial subject (Le Bohec and Buisson 1991).

for the plant and animal materials consumed on it (Bakkels and Jacomet 2003; Blänkle et al. 1995; Kreuz 2000; Lepetz 1993; Marinval 1993). Artifact assemblages from pyre sites can also be compared with graves; the differences in ceramic forms or in the frequency of glass between pyre and grave illustrate the opportunity better to assess the significance of interment against other stages of ritual (e.g. Ames-Adler 2004; Blaizot et al. 2009; Kaiser 2006; Polfer 1996). In the study of Roman burial practice the pyre is nonetheless the poor relation of the grave. Despite frequent archaeological use of the term ustrinum, its meaning in a literary or epigraphic context has yet to be analysed in detail and its archaeological correlates remain obscure (Fevrier, in Béraud and Gébara 1987: 29). As others have noted (e.g. Blaizot et al. 2009; Polfer 2000), the poor condition of artefacts from pyre sites in comparison to grave assemblages ensured a lack of interest on the part of earlier generations of antiquaries and archaeologists but even where well preserved pyre sites have sometimes received only cursory attention. In this respect the Roman period does not differ from others in which cremation was frequently practised, for example Bronze Age or early mediaeval Europe. A necessary pre-requisite to the study of the pyre site and its relationship to the grave is an ability to identify it archaeologically. In the first part of this chapter I use the archaeological, literary and epigraphic evidence to assess the range of archaeological features classified as pyre sites, both in Britain and other provinces, and to consider how they are to be distinguished from potentially similar features excavated within cremation cemeteries. The consideration of pyre sites also requires some scrutiny of the documentary and archaeological evidence for the assumption that a common characteristic of Roman cemeteries was the ‘ustrinum publicum’ or ‘öffentlich Verbrennungplatz’, the permanent municipal cremation facility which has become a commonplace, at least of German-speaking funerary archaeology of the Roman provinces.

There is however an increasing sensitivity to the possibilities of reconstructing a fuller ritual sequence through archaeological evidence. Exceptional environments, for example mummification (e.g. Rossignani et al. 2005), ‘plaster’ burial (Green 1993), waterlogging (e.g. Atlantic House) or other factors promoting anaerobic conditions, for example airtight coffins (Boscombe Down; Roden Downs, Spitalfields; see also Chap. 8.3.1) occasionally allow a fuller understanding of rituals associated with Roman period inhumations through better preservation of organic materials. The close observation of skeletal elements and their relationships in the grave, provides potential clues to burial ritual from the evidence for the decay process in more normal preservation conditions (Duday 2009). A priori however cremation may offer greater recurring potential for such a study, especially the cremation ceremony itself, as was noted in chapter 1 (1.3.1). The systematic study of cremated human bone deposited in burials has given new insights into pyre technology and ritual (e.g. McKinley 1998a; 1994b; 2000; in Cool et al. 2004: 283-310; Boston and Witkin 2006: 23-35 for recent work on Roman period examples in Britain; Ubelaker and Rife 2007 report other analyses). Analysis of other components of pyre residue, left at the pyre site or deposited in the grave, from the sites of single burials or larger cemeteries demonstrate the possibilities for reconstructing the form of the pyre, objects burnt on it and in associated ceremonies from unprepossessing remains. Botanical and faunal remains are a fruitful recent source of study for the substance of the pyre and

The conclusions drawn from the study of pyre sites will then be used briefly to assess the significance of the pyre as a stage of funerary rituals in Roman Britain, in particular previous claims for regional variability in its significance, using information from the various contexts identified as sources of pyre residues. 3.2 Documentary, artistic and archaeological sources of evidence for the cremation pyre With the exception of imperial pyres, there has been little interest since the nineteenth century encyclopaedic treatments in the evidence for cremation sites. With well known exceptions, most famously Virgil’s account of the burial of Misenus (Aeneid VI. 212-235), information on cremation and the pyre site in the literary sources is scarce, not only because such things could be taken for granted, but also because literary funeral scenes evoked 27

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

1993). It has become clear that the variety of cremationrelated features has not been sufficiently appreciated in the past, thus casting uncertainty on previous identifications of pyre deposits (3.3.2). Having formerly been neglected as sources for the pyre and on local environments, information on the carbonised remnant of the pyre structure and associated material has been noted in a small but growing number of studies (e.g. Blänkle et al. 1995; Bridger 1996; Campbell, in Cool et al.2004: 269-271; Gale, in Fitzpatrick 1997a: 77-83; Booth et al. 2011: 253).

an atmosphere of piety (Foulon 1987) or political theatre (Flower 1996; Price 1987) which did not need attention to the detail of funerary arrangements at the less public elements at the side of grave or pyre; whr tis occurs reference to real or fictional cremation focuses more often on the breach than on the observance (Noy 2000; Scheid 2005: 164-5). The proscriptions around the behaviour of the libitinarius (undertaker) may also explain the impropriety of alluding to them in a record of ceremonies of state and of family, unless with the intention of shock (Dumont 1995; Hinard 1995b). The leges Libitinariae from Puteoli and Cumae, probably of the first half of the first century BC (Bodel 1994: 74-76), and collegium regulations (e.g. CIL XIV 4114) are more informative on the contract between undertaker and client or the financial responsibilities of the collegium and its membership than on the practical arrangements for funerals. Inscriptions which explicitly refer to cremation, in particular those mentioning ustrina, are few and limited mostly to Rome and Italy. The instances assembled by Boatwright (1985: 493, n.39) illustrate the usage of the term, as well as its rarity, but they have yet to be systematically gathered and analysed. Chronological distance leads some ancient authors to comment in greater detail on funerary rituals no longer practised at the time of writing, as demonstrated by Servius’ explanation of cremation related vocabulary in his commentary on the Aeneid (In Verg. Aen. III. 21-23. 6-12; XI. 201), but the value of these observations is uncertain. In contrast to other ancient art (e.g. Garland 1985: 21-34; Jannot 1987; Steingräber 1986), pictorial representation of funerary ritual, especially of pyres, is rare in a Roman context. With occasional exceptions much of the limited evidence depicts the pyres of emperors (von Hesberg 1978; Price 1987).

3.3. The form of the pyre 3.3.1 Documentary evidence for pyre terminology and form The major distinction in archaeological terminology, derived from Latin sources, contrasts bustum and ustrinum, the former in-situ burial on the site of the cremation, the latter the site of cremation but not burial (Fellmann 1993). The distinction relies primarily on Festus’ second century AD epitome of the Augustan etymological work by Verrius Flaccus (Festus s.v. bustum). This, like discussion by Servius (Ad Aen. XI.201) of cremation-related vocabulary, can be shown to be ambiguous, as Bridger (1996: 220, n. 1142) has noted and seems artificial when compared to attested usage. Bustum can denote the pyre (e.g. Verg. Aeneid 11. 201; Luc. 8. 740) but more often the tomb or ensemble of tomb and monument (Verg. Aeneid 12. 863; Mart. 5.37.14; Cic. Att. 7.9). Where used in literary sources, bustum does not appear to indicate in-situ cremation and burial, although few ceremonies are described in sufficient detail to assess this definitively. The term ustrinum is hardly ever employed in literary sources (Boatwright 1985: 493), and where encountered is used in a figurative rather than literal sense (e.g. Apul. Met. 7.19). The pyre is most frequently referred to as rogus (e.g. Mart. 8. 75.9-10; Plin. Nat. 35.49; Tac. Ann. 1.8.5); pyra (Verg. Aeneid 6.216), ignis (Luc. 8.738) and, rarely, ara (Ov. Tr. 3.13.20-21) are alternatives. The opening lines of Book XI of the Aeneid, describing the funeral of Pallas, illustrate the terminological variety.

The nature of pyre sites also militates against their archaeological visibility. The most common ethnographically attested pyre comprises a framework of timbers set at right angles to one another, braced by upright posts at the corners or sides and stuffed with kindling material, sometimes with a pit or flue for ventilation beneath (McKinley 1994a: 80; Wahl and Wahl 1983). Experimental pyres show that only a small residue of ash and charcoal is left over after a cremation. The effects of heat on the ground surface depend on soil type and usually penetrate to a depth of a few centimetres at most (Gaitzch and Werner 1993; McKinley 1997: 134). Surface debris and heat-modified soils will be highly susceptible to subsequent truncation. Survival is only more likely where the pyre either takes a more permanent form, includes earth-cut features or is protected by overlying deposits. Preservation beneath barrows is the likely explanation for the better representation of Bronze Age pyre features than those of other periods in Britain (McKinley 1997: 132). From ethnographic observation, experimental work and more careful excavation a more critical perspective has recently questioned the extent to which we understand the formation processes behind archaeologically known pyre sites (Gaitzch and Werner 1993; McKinley 1994a; 1997; Polfer 2000; Struck 1993c; Vernhet 1987; Wahl and Wahl 1983; 1984; Witteyer

Bustum, rogus and occasionally pyra are more common as cremation-related terms in inscriptions than ustrinum. As in the literary sources, bustum more often refers to the tomb (e.g. VI 16193) than to the pyre, which again is recorded more commonly as rogus (e.g. CIL VI 51613), pyra (e.g. CIL VI 6976) and ignis (e.g. AE 1959, 20). Some inscriptions explicitly contrast the rogus as pyre with the bustum as tomb (e.g. CIL VI 10097; XIV 2765); one (CIL VI 10237) contrasts ustrina and bustum as pyre and tomb. In the occasional cases where epitaphs refer to burial on the site of the pyre, the formula hic crematus/a est is used (e.g. CIL VI 21516). The limited documentary evidence available generally relates to the temporary structure of the pyre. An injunction not to build the pyre with wood shaped by the axe is a sumptuary law in the Twelve Tables (Roman Statutes. Law 40: Twelve Tables X.2; Flower 1996: 11828

EVIDENCE FOR ROMAN BURIAL RITUALS

the provision of cremation facilities for the masses. Bodel (1994: 38, 107, n.153) argues that the small number of other references to these are based on this.

9). Other literary references suggest that the pyre could take altar form and might be built from cypress and painted logs, and that combustion could be assisted by pitch and papyri (Daremberg and Saglio 1896: 1395; Marquardt 1886: 381, n.6). The tiered structures represented on coins and diptychs are of little relevance for understanding the form the pyre ordinarily took (Price 1987). The layered grid of logs below the twisting flame on cist covers from Altinum is likely better to correspond to the common reality (Tirelli 2001: 245-246)

3.3.2 Archaeological evidence for pyre form Although the bulk of epigraphic evidence for pyre sites derives from Rome, there is limited archaeological evidence from the metropolis; the location and form of the monumentalised ustrinum of Augustus and other imperial ustrina continue to be discussed (Boatwright 1985; Cupitò 2001; Pollini 2012: 216-8). The structural evidence and burnt debris from the likely pyre site of the emperor Galerius from Gamzigrad (Serbia) suggests that ancient descriptions and visual representations do not exaggerate the magnificence of imperial cremation (Srejovic and Vasic 1994). Evidence from elsewhere in central Italy is also limited. At Ostia some columbaria and other group monuments in several of the city’s cemeteries include structures identified as ustrina (Floriani Squarciapino et al. 1958) whilst at Pompeii burnt areas within funerary enclosures, such as those C and E on the road outside the Porta Nocera, represent the only evidence so far discovered are better documented than most of the sites claimed as ustrina (D’Ambrosio and De Caro 1987: 202; Jashemski 1970: 10-5; Kockel 1983: 39-41).

In inscriptions the ustrinum is frequently recorded as a single element within substantial tomb complexes, sometimes part of a large funerary garden (e.g. CIL VI 10346; VI 10237), more rarely as a distinct entity and the main subject of the dedication (e.g. CIL I 3406; VI 4417). The latter are likely to have been related to a complex close by, an association supported in the case of CIL VI 4417 by nearby inscriptions (CIL VI 4414-4416). The recorded dimensions frequently correspond to the commonest tomb plot size at Rome, ten or 12 feet square, for example CIL VI 4417 and VI 30040 (Purcell 1987: 36-38). Occasionally ustrina were positioned directly on the street frontage (CIL VI 11706) but more often they were behind the tomb (CIL VI 10237 ll. 12-13) or distanced from other elements of the complex (CIL VI 10346, ‘trans via’). This less conspicuous placing perhaps explains the scarcity of epigraphic evidence. Cremation was only one phase of the funerary ritual, whereas the tomb was the place of enduring contact between living and dead. That repeated burning would disfigure any structure built for cremation may also have inhibited the recording of its construction. Occasionally the inscriptions mentioning ustrina have been found in situ (e.g. CIL X 8284) but the context has never been well enough recorded to indicate the form they took (Cupitò 2001: 49-50). Prohibitions on their being established in connection to individual tombs might suggest that the term could sometimes refer to an ephemeral structure (CIL VI 4410; 16746). Archaeological evidence for pyre facilities attached to monumental tombs in the cemeteries of Rome and its environs is slight (see below).

Studies by Polfer (1996; 2000), Faber (1998: 170-75) and Blaizot (2009) are the most important syntheses of archaeological evidence in the Roman provinces, though others also provide useful references and discussion (Airoldi 2001: 115, n.2; Bel 2002: 78-79; Fasold 1993a; Kaiser 2006; Ludwig 1988; Tirelli 2001: 244-5, n. 12; Witteyer 1993). These discussions distinguish between pyre sites constructed out of permanent materials, earthcut features and surface spreads and dumps of pyre debris. The former have been documented on an occasional basis in several provinces and comprise square, rectangular or occasionally circular small walled enclosures built in stone or brick, and sometimes with multiple walls in concentric form. Their function is not well understood. Most were poorly preserved and reported, and even where better surviving, often only limited inference can be made as to their function from their documentation. The dry stone construction of some may have been intended to mitigate the effects of heat (Wels, Linz, Fréjus St Lambert, and Briord). Some examples included a surviving platform on which cremation likely took place, for example at Reichenhall or Linz, although it seems more probable that cremation at Linz took place beside rather than above this platform. The walls may have supported a platform on which a pyre was constructed, but as the superstructure of these features has not survived this is impossible to state definitively. Perhaps masonry-lined pits, as at Carnuntum and St Paul-Trois-Châteaux enhanced ventilation beneath the pyre in a similar way to the pits of busta (Blaizot et al. 2009: 104-5). The erection of the pyre within the larger structures to prevent collapse would certainly have been possible (Wels, Innsbruck, Salzburg). Layered burnt debris has been recovered from the interior of most of these features; sometimes it was deposited in nearby pits

Some ustrina are elements of monuments for individuals or small groups, often of freed slaves (e.g. CIL III, 02912; V 3554; AE 2001: 757). The ustrinum is usually mentioned on inscriptions attached to the monuments of corporate groups, collegia or familiae. Heads of households and patrons of collegia are explicitly attested as providing an ustrinum with the rest of a monument for families and freedmen (e.g. CIL VI 24374; VI 10346). Such Provision fits within the wider context of funerary patronage (Patterson 1992; Purcell 1987) and despite the frequency of reference to public or municipal crematoria in the archaeological literature there is little supporting literary or epigraphic evidence (e.g. Kraus 1999; Meiggs 1973: 459: Polfer 1996: 20; Ruprechtsberger 1983: 23; Sági 1954: 109; Topál 1981: 76; Zsidi and Furger 1997: 299, 308). The reference by the third century AD writer Porphyry (Sententiae 1.8.11) to ustrinae publicae of two or three centuries earlier, which he locates next to the puticuli on the Esquiline is the only direct evidence for 29

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

location of these pits away from the street frontage at Mainz, in contrast to the monuments adjacent to it, also supported the interpretation that these features were not tombs. Larger features of this type include examples from Sampont, Seebruck, Tongeren, Overpelt and Kempten. They may be areas repeatedly re-used for cremation, perhaps formed by the continued re-excavation of pits beneath pyres but the detail presented in publication is not usually adequate to assess this. The lack of examination of cremated human bone from most such features makes it impossible to establish a minimum number of individuals cremated. At Seebruck and possibly Kempten pyre debris seems to have been deposited in nearby pits. The distinction between ‘pyre sites’ and dumps of redeposited pyre debris cannot be definitively made in most cases, especially when the evidence which might differentiate them is not fully published. Layers of pyre debris, cremated bone and burnt soil at Destelbergen, Velzeke and Springhead Pepper Hill are unusually clear examples of the accumulation produced by several decades of repeated deposition of pyre debris in features near the cremation site.

and ditches (Vernes, Gravelotte). Larger structures, as at Cologne or Louroux, are more likely to have enclosed a pyre area than to have been part of the pyre structure. The presence of urned cremation burials within these features at Salzburg and Innsbruck suggests that some combined pyre site and monumental tomb or indeed that some (e.g. Carnuntum) are not pyre sites but Brandschuttgräber (3.3.2) within monumental tombs. The structures identified as ustrina in cemeteries by the Porta Romana and Porta Laurentina at Ostia seem to be of similar form. A small number of features of this type have been noted in Romano-British cemeteries, but none are well recorded. The identification of ‘permanent crematoria’ from St Albans St Stephens and Colchester Gurney Benham House (Black 1986: 210) is not based on adequate evidence, although ‘burning chamber 3’ at the former site may not have been dissimilar to some of the continental examples discussed above. Other features identified as ‘permanent pyre sites’, at Springhead, Corbridge and possibly Densworth are platforms, either of tile or clay and cobble, the size of which would allow for the construction of a pyre above them. Earth-cut features are much more numerically significant as probable pyre sites. These occasionally take the form of possible flues, for example at Westhampnett, Pepper Hill, Baldock Wallington Road, Colchester Stanway (enclosure 3) and Oakley Cottage. This would also seem the most obvious interpretation of the narrow trenches placed at right angles which form feature TXXX at Histria. Post holes have also sometimes survived in association with possible pyre sites, perhaps from posts to brace the pyre, including at Histria. In some cases (e.g. Colchester Stanway enclosure 4) these are easier to resolve into a coherent plan than others (e.g. Altforweiler; Roden Downs; Septfontaines-Dëckt). They may also be related to other funerary structures, such as shelters or platforms for the corpse prior to cremation, as argued for example at Clemency (Metzler et al. 1991: 139). However the majority of evidence for pyre-related cut features comprises pits identified as pyre sites from burning to their bases and sides and from pyre debris and cremated human bone in their fills, with cremation taking place over or within them. These features appear to fall into groups of rectangular or subrectangular pits of two different sizes, those equivalent in size to busta (see below), roughly two metres long by one metre wide and larger features between four and six metres long and two to four metres wide. Some of the smaller features may be one-off pyre sites, a suggestion made by Nierhaus (1959: 22-8) and systematically investigated in the excavation of the Mainz-Weisenau cemetery by Witteyer (1993; see also Blaizot et al. 2009). Here features of this type lacked evidence of grave goods and the quantity of cremated bone recovered was much less than that which would be expected from an adult cremation (3.3.2). There was also clear evidence for the sorting of their fill and for frequent intercutting; Graves by contrast rarely disturbed one another. The absence of intercutting suggests that otherwise similar features at Krefeld-Gellep are more likely to be busta, as does the presence in some instances of grave goods and of collected cremated bone. The

Pyre sites are also indicated by burnt surfaces in cemetery sites or by surface spreads of burnt soil, charcoal, ash, cremated bone and the remains of pyre goods. Such deposits are particularly susceptible to post-depositional truncation so it is not surprising that these should be represented only by occasional traces, identification of which may not be conclusive. Features from Enclosure 3 at Colchester Stanway or the irregular burnt surfaces at Colchester 6 Lexden Road illustrate this ephemeral character. Continental examples where traces of burnt ground surfaces survive are limited in number (Blaizot et al. 2009: 91-3). Better preserved and more substantial examples from Britain include spreads of pyre debris from York Trentholme Drive, Baldock Royston Road and in the east London cemetery. Recently excavated rural examples include Maddington Farm Shrewton and Totterdown Lane, Horcott. Substantial spreads of debris up to twenty metres across and half a metre deep from neighbouring provinces include features from Altforweiler, Dillingen, Septfontaines-Dëckt, Vattevillela-Rue, Lazenay and Wederath. Pyre debris may also be deposited in small quantities into cut features: the many such features documented at Wederath illustrate this clearly (see also Blaizot et al. 2009: 211-15). These will be difficult to distinguish from re-deposited pyre debris or Brandschuttgräber / Brandgrubengräber (see below), as at Oakley Cottage, East London, Brougham, Low Borrowbridge, Dreiborn or Lellig. In occasional cases pyre debris is found in association with other human skeletal material outside formal cemetery contexts, for example in the complex deposit in the south-west gate at South Cadbury dated to the mid-first century AD. Analysis of charcoal has provided a growing if still limited source of evidence for the raw materials of the pyre: samples from Wederath and Krefeld-Gellep for instance have indicated that beech, and to a much lesser extent, oak provided the pyre timber (Kaiser 2006: 5). 30

EVIDENCE FOR ROMAN BURIAL RITUALS

evidence, the archaeological evidence would not seem to support the idea of the ‘public’ or ‘municipal’ cremation facility: analysis of the distribution of cremation-related features in the northern cemetery at Kempten leads Faber (1998: 75) to a similar conclusion.

Blaizot et al. (2009: 112-16) note the difference between oak and beech, the predominant materials used in central France, and the much greater diversity of raw material in the remains of pyres from Mediterranean France. For Britain the high representation of oak charcoal from pyre debris in busta and other pyres, followed by ash, in at Pepper Hill and other Channel Tunnel Rail Link sites typifies the common pattern. Against an emerging background norm unusual variants can be more easily identified, for example the use of cherry wood at Pepper Hill and Kempsford Farm, perhaps for its aromatic as well as calorific qualities, or the very diverse range of woods used at late fourth century AD Lankhills, a potential indicator of fuel scarcity (Challinor in Booth et al. 2010: 441-3; Booth et al. 2011: 253; Challinor in Andrews et al. 2009: 127-8; Smith, in Stansbie 2012: 185). In these and other cases cereals and grassland plants seem to have provided material for kindling, though this is not easily distinguished from vegetation growing close to the pyre and accidentally consumed (e.g. Challinor in Booth et al. 2010: 441-3; Hinton in Fitzpatrick 1997a: 85-7; Green in Millett 1986: 77; Murphy 1990; 1992; Murphy and Fryer in Niblett 1999: 388).

3.4 Cremation burials and other features in Roman cemeteries As has already been indicated, identification of pyre sites is further complicated by the existence of other cemetery features of which the archaeological traces are similar. Particular confusion is caused by the deposition of pyre debris in graves. While studies of Romano-British burial practice have usually defined cremation burials by the container for the cremated bone and/or associated grave goods (e.g. Jones 1983; Philpott 1991), cremation burial typologies in neighbouring provinces are based on the relationship between the deposition of cremated bone and of pyre debris, taking fuller account of this phenomenon (1.3.1). The categories have often however been imprecisely defined by different researchers, as Bridger has exhaustively demonstrated (1996: 220-26). The most commonly accepted usage is that of Bechert (1980), simplifying the earlier formulations of Nierhaus (1959) and Van Doorsaeler (1967: 89-98), although his labels are sometimes tautologous and extend the meaning of Latin terminology beyond its attested usage (van Doorsaeler 1983: 917-18). A burial typology must be sufficiently flexible to accommodate local variety whilst also allowing the burial data to be used in synthetic study of variability in grave forms. This is best achieved by very explicit definition of burial types within the use of these general labels.

Polfer (1996: 20) has suggested that the more substantial permanent structures discussed above were required for the higher volume of cremation in urban cemeteries, while ephemeral ad hoc structures served the less intensively used small rural cemeteries. As he admits, this distinction is not always observed empirically. Rural examples of pyres in permanent materials include Briord, Louroux, Gravelotte and Vernes, while some very extensively excavated cemeteries with plentiful evidence of pyres lack any features of this type, for example Wederath or the many excavated cemeteries at Lyon (Blaizot et al. 2009). In other cases permanent structures seem to have been connected to individual monuments or enclosures within a larger cemetery, for example at Ostia or St Paul-Trois-Châteaux. By contrast the more ephemeral pyre features are recorded from many urban and minor centre cemeteries, notably at Wederath. Here there are several large concentrations of Aschenflächen, spreads of ash indicating the probable sites of pyres, along routes through the cemetery and at its eastern extremity, but cremation also took place within or next to many individual Grabgärten. The presence of multiple, seemingly contemporary cremation areas interspersed amongst burials, is also recorded in other large cemeteries, for example at Hooper Street site in the east London cemetery, the south-western cemetery from Tongeren and the northern cemetery at Kempten; at other sites the pyre area indicated by these spreads or deposits of burnt material seem to have remained separate from the area of burial, for example at Baldock Royston Road and Seebruck-Bedaium, and to a degree at Springhead Pepper Hill, though from the second century AD later burials overlie the pyre site. At several large rural cemeteries also cremation and burial areas were separate (e.g. Chantambre, Schankweiler, Septfontaines-Dëckt). In summary it is difficult to identify an association between pyre structure and settlement context. Like the epigraphic

The terms and categories in Bechert's scheme (1980: 256) are therefore the basis for what follows, although criticisms are noted. Bechert clearly distinguished in theory between the processes which might produce different types of grave deposit but there has been less attention to the archaeological correlates of these processes; work by Blaizot and colleagues (2009) represents a recent systematic exception. This brief review will detail the expected characteristics of the different types of feature which can be confused with pyre sites and assess the criteria by which they can be distinguished. The first, the bustum, is defined by in-situ burial on the site of the pyre. Its literary usage is as a synonym for burial rather than as a distinctive ritual, cautioning against reification of it as a burial form with a particular association with Roman culture (see above). Indeed Struck (1993b) has already identified several areas of the empire in which bustum is an archaeologically recognisable burial form in pre-Roman cultural traditions. When used in an archaeological context the term is best used in its descriptive sense only, but this poses its own problems. The description that follows is based on graves from cemeteries mainly outside Britain (appendix 2.2). Of the possible British examples most have been insufficiently documented to be certain of their identification as graves rather than other features (Struck 31

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

cremated bone from the pyre residue (e.g. Ota 2007; Blaizot et al. 2009: 158-60).

1993b; Booth et al. 2010: 503-4), though important groups have been more closely documented at Pepper Hill and in Oxford Archaeology excavations at Lankhills.

Within the fill, or distributed between fill and possible container, we might expect the full amount of cremated bone, which for an adult will vary between 1.5 and three kilograms (McKinley 1989a: 66; Mays 1998: 220), though variability according to age and sex is not well established. However the expected amount has rarely been recovered in the few busta from which the human bone has been analysed and is often lower than in other types of cremation burial in many examples, even where very precisely recorded (Appendix 2.2), prompting questioning of the interpretation of features claimed as busta (e.g. McKinley 2000; Gonçalves et al. 2010). However a combination of ancient and recent factors will have influenced this quantity, including the possible falling of some debris from the pyre outside the pit, the removal in antiquity of cremated bone, or a lack of whole-earth sampling in excavation; hand recovery of cremated bone will miss both the smaller bone fragments and the less well burnt darker material. Preservation conditions may influence recovery, although their impact on cremated bone is poorly understood (Mays 1998: 209).

Evidence of the wooden structure of the pyre itself has occasionally been recorded, in post holes on the edges of pits, or remains of timbers either within the fill or at its base. The survival from a pyre site or bustum at Beckfoot, Cumbria, of an oak bier or bed, as well as a charred feather stuffing, is exceptional. The pits for busta are usually as large as, or larger, than those of adult inhumation burials and rectangular or sometimes oval or irregular in plan, with near vertical sides and flattish bases. Traces of burning to a depth of several centimetres have often been noted on their edges, sides and less frequently bases; examination of the surface of the grave cut in a Lisbon example shows burning to a depth of 15 cm. From differential burning it may sometimes be possible to distinguish whether the pyre burnt over or within the pit. At La Favorite, Lyon, for example Tranoy (1987: 44) distinguishes between two types, one with evidence for burning all the way down the sides to the base, the other with strips of burning and baking of the soil around the edge of the pit. The former may indicate construction of the pyre within the pit, the latter over it. In some examples from Moesia sufficient evidence survives to calculate the number of layers of timber employed (Ota 2007: 78). Erection within the pit would have inhibited ventilation, especially as pyre debris accumulated, but the distribution of traces of burning suggests both possibilities.

The variant Flächenbustum form, i.e. in-situ cremation on the ground surface rather than in or over a pit, is much rarer. Most instances have been recorded beneath barrows in the Hunsrück-Eifel area and in Noricum and Pannonia (Struck 1993b: 84). The Flavian period barrow burial at Riseholme, beneath which a thick (>0.2m) deposit of pyre debris was documented, is a rare example from Britain, but for which documentation is limited.

Within the fill experimental pyres have taught us to expect a layered deposition of charcoal and burnt human bone still in approximately correct anatomical position after the collapse of the pyre structure (Gaitzch and Werner 1993; McKinley in Fitzpatrick 1997a: 65). However inadequate recording or publication may obstruct the identification of bustum stratigraphy. This layering has only occasionally been recorded at cemeteries considered here, for example grave 1 at La Calade, and to some extent at Pepper Hill and regularly at Lankhills where documented. The positions of cremated bone and sometimes remnants of dress occasionally suggest that the body still lies as it was on the pyre (Bura 2001; Hatt 1964: 78; Vanvinckenroye 1963: 128-29) but this important criterion for distinguishing busta from other features has in general been investigated only in the most recent excavations (Blaizot et al. 2009: 118-126). McKinley’s (1997: 136) analysis of Bronze Age pyre sites from Linga Fold suggests that sorting cremated bone from pyre debris leaves the vertical stratigraphy relatively intact but disturbs the interrelationship of anatomical elements in the horizontal plane. Close observation of the stratigraphy has suggested that the cremation residue could be more significantly manipulated before the grave was filled. In busta from Neuss, Krefeld-Gellep, MainzWeisenau it was often noted that after cremation the pyre debris was sorted through and the grave goods, selected human bone and fragments of pyre goods placed on the pit base, with pyre debris re-deposited above them. Similar practice occurs elsewhere, separating some

More commonly, in Britain at least, the body was cremated on the pyre but buried elsewhere. The burials may be subdivided into two forms, (i) of the cremated bone only, within a container, usually a ceramic vessel (urned burial) or as a heap which had perhaps originally been in an organic container, or scattered in the grave (unurned burial), and (ii) with the pyre debris (Brandschuttgrab). The latter category is subdivided by Bechert into the Brandgrubengrab, the burial of cremated bone and pyre debris together unsorted, and the Brandschüttungsgrab, the burial of the sorted cremated bone with some pyre debris. Jones (1991b: 117) and Bridger (1993b) have both noted the scarcity of Brandschuttgräber from Britain and well recorded examples are even rarer. Published Brandschüttungsgräber include examples from York Trentholme Drive and the Mount, Derby Racecourse 1, South Shields and Lankhills 60, Low Borrowbridge and Brougham. Examples of Brandgrubengräber include those from Ben Bridge, Knob’s Crook and Holborough, as well as Low Borrowbridge and Brougham, though in these cases the identification must be provisional because the full amount of cremated bone deposited is often not known. In these as well as other instances it is difficult to distinguish burials from small deposits of pyre debris (see above). Bridger attributes the scarcity of Brandschuttgräber in Britain to inadequate recording, but 32

EVIDENCE FOR ROMAN BURIAL RITUALS

includes fragments of artefacts placed on the pyre. The variety can be illustrated by examples from Hoepertingen, in which the pyre material deposited in grave I contained a concentration of cremated animal bone (Roosens and Lux 1974: 44) and Cucuron, where the 1.5m deep fill of Brandschüttungsgrab 1 mainly comprised clean charcoal (Hallier et al. 1990: 152).

this need not be so in every case. Even where a greater sensitivity exists to the significance of the presence of pyre debris it has not been recorded, for example in Switzerland, where urned burials are the most common type (Berger and Martin-Kilcher 1976: 154). The pyre debris recovered from these graves can only represent a fraction of the amount of debris generated by cremation. This is made clear by the variation in the amount of pyre debris deposited with Brandschuttgräber in cemeteries where they are well represented and the general label masks significant local diversity (Bridger 1996). It is difficult to identify the reasons behind the selection of debris to include with the burial. The weight of cremated human bone from pyre debris separate from the urn is often not recorded, but in almost half the graves from Ergolding, Bavaria (Struck 1996), no cremated human bone was recovered from the pyre debris and only very small amounts were recovered from others, although the recovery method, which is not recorded, will have affected these figures. The pyre debris sometimes

A low amount of cremated human bone deposited in Brandgrubengräber has been noted even where the graves are very large, for example less than 40g from many graves at Sint-Denijs-Westrem (Vermeulen 1992: 230) and less than 100g in burials in the Escaut valley (van Doorsaeler and Rogge 1985). Comparison with other grave types in the same cemeteries at both Stettfeld (Wahl and Kokabi 1988: 36) and Ergolding (Struck 1996: 29) showed that the average amount of cremated bone in Brandgrubengräber was lower, although if truncated lack of burial container will make cremated bone in burials of this type more vulnerable to disturbance.

Table 3.1 The identification of different cemetery features *’Full’ means an amount equivalent to that found in other types of graves in the same cemetery (3.2, 4.2.2). Dimensions of Brandschuttgräber are based on samples from cemeteries in Germany and Belgium Type

Features

Bustum

Structural evidence for pyre, heavy burning on sides (and base) of cut. Grave cut only. Possible burning of sides/base of cut from deposition of hot material. Possible burning of sides/base of cut from deposition of hot material. Structural evidence for pyre, heavy burning on sides (and base) of cut. Cut features of varying size or surface spreads, sometimes evidence for burning of ground surface Possible burning, usually under barrow.

Urnengrab Brandgrubengrab Brandschütungsgrab ‘One-off’ pyre Pyre debris deposit

AschenGrube

Pyre debris Yes

No Yes

Yes, often in small quantity Yes

Dimensions c. 1 x 0.5m to 3 x 2m. Many at upper end of range. Very varied From 0.7 x 0.5m to 3 x 1.4m. Many at upper end of range From 0.5m² to 1.5 x 1.25m. Most at middle of range

Container Sometimes

Quantity of human bone Full, possibly in anatomical relationships.

Grave goods Sometimes

Yes No

Full* Full*

Sometimes Sometimes

Yes, or heaped bone

Full*

Sometimes

No

Little, possibly poorly burnt.

No

Yes

Very varied

No.

Varied, a high proportion may be poorly burnt.

No

Yes

2m x 1m

No

Animal bone only.

No

33

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

One further variant of cemetery deposit should also be mentioned. Individual or groups of complete objects, for example ceramics, deposits of coins and animal burials may be the product of similar activity; examples from Britain (e.g. Friary Field; East London; Ketton; Willersey; see also chapter 5) are part of a wider provincial phenomenon (e.g. van Doorsaeler 1967: 108110; Ebel 1989: 90-91; Lepetz 1993: 42; Nierhaus 1969: 253; Nuber and Kokabi 1993). Assemblages of complete ceramics not associated with cremated bone are commonly interpreted as cenotaphs (e.g. Allain et al. 1992: 39; Cool 2011: 296-7; van Doorsaeler 1967: 105-6; Noël 1968: 107; Simmonds et al. 2008: 136-7; Wahl and Kokabi 1988). Such deposits are just as likely to be products of cult practice rather than burial, especially when present in such large quantities, for example at Sampont near Arlon (e.g. Noël 1968). Deposits of this kind are also reminiscent of the ‘special’, structured’ or ‘placed’ deposits in pits, shafts and wells on RomanoBritish settlements, usually read as the product of sacrifice related to the sphere of domestic, agricultural or craft process rather than wider civic rituals. Again they typically comprise individual examples or assemblages of whole or near complete objects, especially of ceramics and other vessels and animal carcasses or articulated limbs or other elements, sometimes recurring at several points in the fills of such features. Evidence of burning or deliberate breakage is widespread and human skeletal remains also occur intermittently among them. Their frequency and significance was emphasised by Wait (1985) and Merrifield (1987) and examples have been recognised in an ever-widening range of contexts (Chadwick 2012: 294, with extensive references; Fulford 2001). Examples in close proximity to burials include Folly Lane, St Albans (4.2) or Keston (6.4.3).

As well as pyre sites and burials other features and deposits, seemingly the product of rituals associated with burial or commemoration have also been excavated within cemeteries, several of which are also potentially confused with pyre sites and busta or Brandschuttgräber. Particularly significant is the category of Aschengruben defined by Wigg (1993b). These features are often similar in size to graves but their fill, containing ash, charcoal, burnt ceramics, wood and bone artefacts, coins, molten glass and metal, burnt animal bone, plant and other food remains but no human bone, is likely to be related to funerary ceremonies other than cremation or burial. Occasionally the sides or base may also be burnt through deposition of hot material. This term was originally applied to a type of deposit best known from the Treveran area, with well documented examples at Wederath, Siesbach (Abegg 1989a) and Büchel (Eiden 1982). Analogous features have also been recorded beneath barrows in Belgium, sometimes comprising spreads of burnt material (Massart 2007). Recent publication of the Wederath Aschengruben illustrates the need for close examination for the presence of cremated human bone in such deposits as it was eventually found in 84 of the 434 excavated examples (Kaiser 2006: 26). Aschengruben and similar deposits are part of a spectrum of objects and assemblages from Roman cemetery sites which are not the direct product of the burial process but of associated rituals. Larger assemblages of this type can be illustrated by deposits of burnt and unburnt ceramics, animal bone and other artefacts filling ditches and pits from a number of cemeteries in central and southern France, including Roanne (Vaginay 1987: 110-11), St Paul-Trois-Châteaux (Bel 2002: 49-53), St Marcel-surIndre (Allain et al. 1992: 26-28), Nuits St Georges (Planson 1982) Lyons and elsewhere in central and southern France (Blaizot et al. 2009: 240-251). At Caerleon Lodge Hill a large first and second century pottery assemblage associated with a layer of charcoal but only a minute quantity of cremated bone was recovered from section 157 of the west-east ditch. At Low Borrowbridge several pits contained concentrations of charcoal but no evidence for cremated human bone or ceramics. Where ground surfaces have not been truncated they may also contain similar evidence, albeit usually in smaller quantities and more fragmented, probably derived from commemorative feasting and ritual, for instance in northern Italy (Ortalli 2001) as well as Pompeii (van Andringa and Lepetz 2005) and St Paul-Trois-Châteaux and other sites from central and southern France (Bel 2002: 64-74). The wide distribution of evidence of this type suggests its significance as a component of mortuary rituals; what remains much more difficult is to differentiate the specific processes responsible for individual deposits, for example libations for the dead, sacrifices, or the remnants of meals consumed by the living during burial or subsequent commemoration, or to attribute them to a culturally specific ritual context, e.g. commemorative festivals in the Roman calendar such as the Parentalia.

There are therefore a range of grave and non-grave deposits to be expected from a cemetery site that relate to funerary and commemorative rituals. Differentiating them and determining their formation processes are not unproblematic, especially as the non-burial deposits have often not been as fully documented as graves. Additionally even substantial features sometimes lack evidence from their fill to reveal their purpose, for example the pits closely associated with the cremation burials in the Oxford Archaeology excavations at Lankhills, Winchester (5.3.2). Deposits from previous or later occupation areas in the same space as burials or derived from nearby activity or rubbish dumping, perhaps interleaved with burial evidence, may also prove difficult to differentiate from some residues of ritual process, often the case on urban margins (for example Cirencester Bathgate or the east London cemetery). In well-preserved and recorded evidence from cemeteries the distinction between the different types of Brandschuttgräber should be fairly clear as it depends on one key characteristic, whether or not the cremated bone has been separated from the pyre debris. However there are several sources of possible confusion between the remains of the pyre itself, deposits of pyre debris elsewhere, and graves including busta, 34

EVIDENCE FOR ROMAN BURIAL RITUALS

Britain, including the larger Westhampnett and King Harry Lane cemeteries and a small number of ‘Welwyn’ type burials. The sources for pyre ritual from the Roman period are more varied, including features of all the types described above, pyre sites and pyre-related features, debris deposited in busta and Brandschuttgräber and the animal bone and small fragments of pyre good collected with cremated bone. The regional and chronological variation in the degree to which cremation was adopted as a ritual and in the excavation of cremation cemeteries has been outlined in the previous chapter. The distribution of cremated bone assemblages available for study is southern and urban in emphasis. The only large samples available north of the Severn-Wash line derive from Derby Racecourse, Brougham and Caerleon Lodge Hill. Evidence available from southern Britain dates mostly to the first and second centuries AD, but cremation at several of the cemeteries in northern Britain persisted into the fourth century AD (2.3). The distribution of busta and Brandschuttgräber to some extent favour northern Britain (Struck 1993b; 2001), although both types, especially Brandschuttgräber, have also been identified in southeastern cemeteries, including Baldock Royston Road, St Albans St Stephens and the east London cemetery, although even in recent excavation the problems of classifying deposit types remain acute (e.g. McKinley in Barber and Bowsher 2000: 264). Well documented pyre sites or pyre debris deposits with even partially recorded artefactual assemblages however are rare. The two principal examples from larger cemeteries currently available for consideration here are from Trentholme Drive and east London, with limited evidence available from the former.

Brandschuttgräber, and especially Brandgrubengräber, as well as with Aschengruben, all of which will be characterised by the presence of burnt debris. Analysis of cremated bone and attribution to human or animal provenance is obviously essential to the identification of Aschengruben. Distinguishing between the other feature types depends on the characteristics of the feature itself, its fill, especially the cremated bone, and less certainly the relationship to other features (Table 3.1). Evidence for the pyre, whether the (occasional) preservation of pyre timbers or in negative form as post holes or heat discoloration will permit a distinction between busta and pyre features and graves, although the deposition of hot pyre debris in graves may also burn the grave cut, as for example at Knob’s Crook. The presence of a container for cremated bone and grave goods serves to distinguish burial from non-burial features. Where this is lacking the amount of cremated bone has been considered the arbiter for distinguishing grave from pyre or debris deposit. However, as has been argued above, the amount of cremated bone from busta and Brandgrubengräber is often very low and may be insufficient, in isolation, as a defining characteristic of a burial. The distribution or condition of the cremated bone is likely to be a better guide to identification of a feature. Its spatial setting may also provide further circumstantial evidence. Limited evidence suggests that pyres are more frequently disturbed by later deposits than graves and where the cemetery spreads along a road they are perhaps less likely to be positioned prominently yon the road frontage. Given the detailed evidence required, there will be a limit to which evidence from older excavations can be re-analysed. Many busta referenced by Struck in her survey (1993b) could perhaps be challenged as potential pyre sites and in many instances data necessary for definitive identification will not have been recorded.

Cremation burials in south-east Britain in the late Iron Age were often furnished with rich grave good assemblages (Stead 1967) but evidence is accumulating for the pyre as a significant stage of display. The fragments collected and buried with cremated bone and found with the small number of pyre debris deposits are the main sources of evidence for treatment on the pyre, the latter from a handful of recent excavations. Those recovered from pyre debris and graves of mid 1st century BC date at Westhampnett, Sussex, comprised predominantly dress fittings and animal parts, as well as some ceramics. Ornaments and dress associated items were also deposited in the grave and pots were more frequently recovered as grave than pyre goods. Fragments deposited with the cremated bone from King Harry Lane several decades later included animal bone, copper alloy and molten glass. The latter two categories were rare, and associated with only a small number of burials; the glass vessels and the identifiable bone and copper alloy objects, for example the wine strainer and pyxis, were not represented among the grave goods and hint at more impressive ceremonies than the relatively modest grave good assemblages would suggest (4.2.2).

3.5 The pyre as focus of display in Roman Britain The focus of this final part of this chapter lies on Britain and the relative importance in time and space of the pyre as a stage of funerary ritual. Philpott (1991) has argued for significant regional and context differences in the provision of grave and pyre goods. His survey demonstrates clearly that grave goods are plentiful in south-east England, especially in areas with a pre-Roman tradition of cremation burial, but much rarer in Wales and northern England. He argued that this scarcity could be accounted for by an alternative emphasis in burial rites: ‘A further tendency observed in the north, at early military sites elsewhere and in the most heavily Romanised urban centres is the destruction of offerings on the pyre rather than placing them intact in the grave’. (Philpott’s (1991: 220) The types of evidence available from the excavated burials and cemeteries vary significantly, but there is limited direct evidence for the pyre itself. Knowledge of pyre treatment in the pre-Roman period is confined to a handful of cemeteries of first century BC and early first century AD in the ‘Aylesford’ tradition of south-eastern

‘Welwyn’-type graves from Baldock California (1st century BC) and St Albans Folly Lane and Stanway (conquest period) suggest that spectacular destruction of artefacts could sometimes form part of the associated 35

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

and occasionally other sites in the same region, for example Birdoswald.

ceremony (4.2.1-2, 4.3.1). At both Folly Lane and Stanway the fills of destroyed burial chambers included large assemblages of broken ceramics, and many of the objects deposited in the Lexden tumulus were in broken or fragmentary form. Other deposits suggest that this custom was distributed more widely in Britain, including the Melsonby hoard from Stanwick, north Yorkshire, if correctly re-interpreted (Fitts et al. 1999) and the phenomenon is documented across north-west Europe (e.g. Fitzpatrick 2000).

At Brougham many graves also included unburnt objects deposited with the dead, in particular ceramics. A small number of ceramic vessels were burnt but pots were much more common as grave goods, accounting for 280 of the 296 vessels deposited (Cool et al. 2004: 446). Of other grave goods glass vessels were the most frequent, usually drinking vessels rather than the containers placed on the pyre. Animal bones were also sometimes deposited unburnt, but their scarcity is likely to be a consequence of high soil acidity. Personal ornaments were recorded in small numbers among the grave goods as well as on the pyre.

In the Roman period striking differences can be documented in the degree to which material was destroyed on the pyre, but our view is heavily conditioned by the types of evidence available in individual cases. The major source for assessing cremation rituals associated with a military garrison in northern Britain is the second to fourth century cemetery at Brougham, Cumbria. Although the vicissitudes of excavation in difficult circumstances and subsequent attrition of the archive have affected the study of this cemetery, impeding differentiation of burial and nonburial features, painstaking analysis has nonetheless restored a very rich view of the burial rituals performed, although comparison with the limited documentation from other sites shows it is unlikely to be typical.

There is other evidence from smaller groups of burials in northern Britain for both pyre as well as grave serving as foci during the funeral. Pyre residue from two Brandschüttungsgräber at South Shields included burnt bone and the nearby dump of (probable pyre) debris contained fragments of glass and ceramics. One of the two contained fragmentary unburnt ceramics and the other two intact flagons and copper alloy fragments. At High Torrs (Galloway) a late second or early third century burial probably of non-local origin had been furnished with joints of meat on the pyre, cattle and cattle or sheep / goat. Probable grave goods comprised an iron ring with onyx intaglio, samian pottery, and casket fittings.

Almost 300 cut features, both graves (including urned burials, Brandschüttungsgräber and probably Brandgrubengräber) and non-burial deposits, were excavated on the site, a substantial proportion of which included redeposited pyre debris. ‘Pyre goods’ were found with a very high proportion of burials, though joining pieces of bone veneer from different graves suggest that the proportion may be exaggerated by contamination when remains were collected from a common pyre; nevertheless the pyre was clearly a key focus of display. In some instances the ceremony must have been spectacular: for one burial (102) a complete horse carcass was burnt. The diverse ‘pyre goods’ include fragments of ceramics, glass (where recognisable, bottles and flasks), copper alloy and silver vessels, animal bone (from carcasses to individual joints of meat), personal ornaments, most frequently beads but also a few precious metal pieces, and other items, including sword scabbard slides. Hobnails and nails from biers and wooden boxes were also documented: some of the charcoal fragments may be from the same source (Campbell, in Cool et al. 2004: 269). The most frequent item was bone veneer (found in 92 (32%) of the deposits), which Greep (in Cool et al. 2004: 274), has argued to derive from caskets or beds or couches on which the dead had been laid. Although bone inlay has been occasionally recovered from graves in other provinces, metal fittings for furniture are more frequent (Martin-Kilcher 1976: 60-61). The two other documented instances of funerary couches with veneer in Britain, Folly Lane and the Colchester ‘child’s burial’, both date to the first century AD and few examples from continental Europe date to later than the first century AD (Béal 1991; Letta 1984; Obmann 1998). If the Brougham veneers do derive from couches or biers, this tradition would seem to be particular to Brougham

In other cases however graves were poor in objects destroyed on the pyre or deposited in the grave. Of the up to 71 Brandschuttgräber, possibly of third century date, excavated at Low Borrowbridge, Cumbria, the burnt material from the pyre was limited to small quantities of sherds, nails from the pyre and hobnails as well as two instances of animal bone. Objects buried after cremation were equally scarce, only two being furnished with ceramic accessory vessels, three with brooches, one with a bead necklace and two with coins. Objects with second to fourth century cremation deposits from Petty Knowes cemetery at High Rochester comprised small numbers of hobnails, nails and ceramics, some of which were burnt. Artefactual material was also scarce among the twenty nine highly truncated second and third century busta, Brandschuttgräber and other cremation-related features at Lanchester, comprising a small number of sherds, hobnails and nails. A similar impression emerges from the small portion of the first and second century cemetery excavated north of Corbridge, although damage to the burials and loss of the cremated bone make it dangerous to press evidence from this site too hard; pyre goods comprised fragments of molten glass while grave goods were also few but included exceptional individual artefacts, in particular an enamelled hexagonal copper alloy perfume flask. South of the frontier cemeteries associated with the army also reveal significant variety. There were no Brandschuttgräber in the cremation cemetery of the late first and second centuries at Caerleon Lodge Hill, so the 36

EVIDENCE FOR ROMAN BURIAL RITUALS

ceramics including incense cups, ten glass vessels of which one was identifiable as an unguent bottle, occasional personal ornaments, four coins, and oyster shells. The ceramics were not fully reported, save for the samian vessels and incense cups. Roughly half of the samian assemblage derived from decorated vessels, which are rare as grave goods in Britain (8.2.1). The evidence from Wroxeter and Lincoln is much more limited. Wright’s (1872: 235) account of cremation burials at Wroxeter records the recovery of fused and molten unguent bottles from many graves, but glass and ceramic grave goods were also relatively common. The few objects placed with the dead as found in four cremation burials, including three Brandgrubengräber, of possible Flavian date at Lincoln Monson Street had almost all been burnt on the pyre, including unguent bottles, other glass vessels, a copper alloy mirror and small quantities of animal bone.

absence of buried pyre debris, as well as the small proportion of cremated bone buried in comparison to other cemeteries (cf. Fig. 4.6) influences the archaeological visibility of pyre goods. Burnt animal and plant remains were recovered from 9% and 4% of burials respectively. Uncremated animal and plant materials are unlikely to have survived the acid soil conditions. Unburnt ceramics, typically represented by small numbers of fragments, were recovered from a larger proportion of graves (55%) than burnt (10%). Hobnails were recorded from 22% of burials, in a few cases having been burnt with the deceased on the pyre. Glass vessels were recovered from 6% of graves and showed no evidence of having been subject to heat, although a small number of glass counters and beads had been. Excavations at the Coed site south-east of the fortress have revealed much more substantial grave assemblages, so the Lodge Hill cemetery is not fully representative of burial practice at Caerleon (Arnold and Davies 2000: 135; Pollock 2006: 26-7).

In cremation cemeteries from southern Britain, the main sources of evidence for pyre ceremonial are the artefact and other fragments, primarily of joints of meat, which were collected with the cremated bone. Pyre debris is, seemingly, less frequently deposited in graves and documented pyre sites are rare, including the urban and minor centre cemeteries considered in chapters 4 and 5 as well as other large urban cemeteries. Individual sites however indicate the possible significance of the pyre as a focus of display.

Of the 39 first to mid-fourth century cremation burials at Derby Racecourse which survived damage by later inhumation burials, only one contained a substantial grave good assemblage, others being rarely furnished. Grave fills and burnt deposits within the walled enclosure by contrast yielded an assemblage of burnt and broken ceramics in which amphorae, flagons and non-local fine wares were represented in much higher proportions than in contemporary settlements nearby. Burnt glass fragments were also recovered. It is difficult here to differentiate objects with which the dead were furnished on the pyre from the residues of funerary feasting, but the richness of the assemblages is striking. Cremated animal bone, pig and chicken were recovered with approximately one third of burials.

The main illustrations of pyre rituals are taken from the East London and Pepper Hill cemeteries, supplemented with reference to recent evidence from Colchester and Cirencester. Here pyre features and graves suggest the frequent destruction of food, dress items and vessels on the pyre, without a marked difference between what was burnt and what was buried. The commonest species represented in the London faunal assemblage were pig and chicken, but a wider variety of species was identified than at other cremation cemeteries in southern Britain, including not only sheep / goat and cattle but also a variety of fish and birds (Philpott 1991: 198; White 2007). Burnt animal remains were found with a quarter of burials at Pepper Hill, pig and chicken again being the most popular species; in one case a whole pig seemingly was burnt on the pyre (11637). The evidence for plant remains is the first to be documented on a significant scale from Roman period cemeteries in Britain. Leaving aside wood charcoal, plant weed macrofossils previously recovered from late Iron Age or Roman cremation deposits are most likely to derive from local vegetation or kindling material (see above). In London charred lentils, peas and Celtic beans were deposited in features with cremated human bone. The cereals recovered may be derived from kindling or part of a ‘semi-cleaned’ crop placed on the pyre. The grape pips and peppercorn recovered in the plant macrofossil sample must have derived from exotic foodstuffs placed on the pyre but water logged fruit pips including fig, grape and others may have derived from feasting deposits or accidental incorporation. Another London burial (Great Dover Street) has also revealed the burning of exotic plant

Among the second to third century cremations from York Trentholme Drive were ten certain Brandschüttungsgräber, five with pyre goods. The thirtyeight other scattered patches of pyre debris cannot be confidently classified and might derive from several of the feature types discussed above. Only one grave certainly contained grave goods but the majority of grave groups were not preserved intact. Pyre goods recovered included ceramics and glass, animal bone, burnt bone counters, a coin and a large deposit of iron nails. The most substantial assemblage in grave 32 comprised the burnt remnants of a beaker, flagon, glass vessel, a bone counter, nails and bones from a cremated fowl. In a grave from the Mount cemetery burnt samian was recovered from the debris of a Brandschüttungsgrab. The cremated bone was contained in a grey ware jar and around it lay large fragments of eleven plain samian vessels. Sherds from the same vessels showed very different degrees of burning, which Dickinson attributed to their shattering and dispersal on the pyre, and Wenham to deliberate breakage prior to burning. The former seems more likely to explain this fragmentation (cf. Polfer 1996: 118-20). Within the large pyre debris deposit from Trentholme Drive the artefact range was similar to the burnt material described above, consisting of hundreds of nails, 37

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

excavation (e.g. Lexden Road) indicate sometimes generous pyre treatment in the western cemetery. The very numerous cremation burials and pyre debris deposits from the ‘alienated land’ sites south of the Circus show considerable diversity in the provision of food and of objects on the pyre as well as in the grave for early to mid Roman burial ceremonies.

material, including pinecones, either as foodstuff or for its smell or ritual significance, also found in the rural cemetery at Mucking alongside date stones. At Pepper Hill burnt weeds and cereals were documented in two graves, present by accident or use as kindling, but one (11801) included a more diverse assemblage with grapes, lentils, figs, horsebean, pea or similar, again deposited uncooked. As yet neither the range nor quantity of species stand comparison with rich deposits from cremation cemeteries in Gaul and Germany (e.g. Bakkels and Jacomet 2003; Kreuz 2000; Marinval 1993).

Beyond the towns substantial assemblages of burnt pyre material have also occasionally been documented from individual or small groups of burials. The Brandgrubengrab and pits of pyre debris from the Flavian period burial at Knob’s Crook included burnt and broken pieces of bronze from vessel(s) and fittings, a glass vessel, eight samian vessels and burnt soapstone objects including a button, beads and a possible bracelet, but no grave goods. At Mucking (cemetery II) a possible pyre debris pit of late second century date, including carbonised remnant of stone pine and date as well as a large ceramic assemblage represents the remnant of a substantial ceremony. The burial and pits of debris from the late second or early third century burial at Holborough included burnt animal remains, glass, ceramics and an iron stool, with broken fragments from at least five amphorae at the side of the main burial. The objects recovered with cremated bone and pyre debris within graves from other rural sites, for example animal bone and dress items from Stansted (burnt bone toggles from graves 12 and 13, a burnt bow brooch from 32) and from Strood Hall (glass beads from burial 1585), illustrate the likely more typical objects placed on the pyre with the dead. In the two modest Wiltshire sites at Maddington Farm and Wayside Farm only animal bone is documented in the pyre debris-bearing contexts.

In London ceramics, especially amphorae and drinking vessels, were better documented in the pyre debris than among the grave goods. Unrecognisable fragments of burnt and molten copper alloy and glass otherwise dominated the finds assemblage from the pyres, though the few recognisable items, for example pins, hobnails and fragments of bone inlay and metal mounts are similar to those also deposited as grave goods. At Pepper Hill both pyre and grave were characterized by more modest artefact assemblages. The ceramics found in the pyre site fills (eight vessels) as well as burnt sherds in the graves and redeposited pyre debris comprised a modest sample. Traces of the destruction of metal items, most likely to have been personal ornaments, were more numerous, either as fragments (e.g. brooch fragments in three pyre site fills), or staining, The latter, a product of contact with iron and molten copper alloy objects was documented in 21 and 23 burials respectively, or c. 13% of the cremation burial sample. Where the associated skeletal element could be identified they were mostly attributable to the head and upper body. Pyre sites as well as graves and debris deposits also contained nails, in one case hobnails, in others from biers or, especially in the case of smaller ones, perhaps from boxes. One fragment of burnt animal bone inlay from a box or bier was found. Fragments of two worked bone pins, and occasional scraps of molten glass comprised the remaining objects derived from the pyre. In the grave goods by contrast ceramics for eating and drinking were by far the most frequently attested items.

In conclusion, the examples rehearsed in the preceding paragraphs show that the destruction of material on the pyre was a recurring feature of many burial ceremonies in the province, whether as part of the presentation of the dead, for example the furniture or dress objects, or perhaps as consumed by participants in the ritual. The repertoire of objects overlaps to a significant degree with those also buried with the dead, though at individual sites, most strikingly Brougham in Cumbria, the suite of objects and animals cremated was strikingly different to the provincial norm and has significant echoes of rituals elsewhere in the empire, notably on the Danube (Cool et al. 2004: 463-66). However a regional or site type-related variation in privileging the pyre or grave as a focus for the destruction of foodstuffs or objects cannot be easily sustained. Although the characterisation of LPRIA burial practices has in the past concentrated on grave goods, there is increasing evidence for cremation itself as the occasion for significant destruction of commodities. In the Roman period pyre debris and associated artefacts are more archaeologically visible in individual cemeteries in northern than in southern Britain, perhaps because of the greater frequency of busta and Brandschuttgräber in the former area. However an enormous variability can otherwise be seen between individual cemeteries in the relative emphasis given to the different phases of ritual in cemeteries, the Brougham cemetery being exceptional in

At Cirencester a ditched enclosure excavated on Old Tetbury Road, close to the route of the Fosse Way included a series of features containing the residues of an impressive mid-late first century AD cremation ceremony, the principal relevant context being a pit c. 2m in diameter, and neighbouring features, all containing pyre debris which as well as cremated human bone incorporated charcoal, animal bone, nails from a bier or box(es), much heat-distorted and molten glass and copper alloy fragments and pottery, probably burnt and in poor condition. The latter was dominated by continental finewares, including samian and sherds from at least two amphorae. Material from Colchester demonstrates a wide spectrum of possibilities; while few assemblages are as rich as the so-called ‘Child’s burial’ with its many fragments of burnt bone inlay from a funerary couch ((Eckardt 1999: 77), antiquarian evidence such as the many fused glass unguent bottles in burials from the Joslin collection (Philpott 1991: 117) and recent 38

EVIDENCE FOR ROMAN BURIAL RITUALS

The epigraphic attestations of ustrinum, although relatively infrequent and deriving primarily from Rome, provide insights into the organisation of cremation. They often were parts of larger funerary complexes and were provided for familiae and collegia, suggesting that like other aspects of burial provision cremation often took place within the context of patronage of these smaller groups rather than for urban communities on a larger scale. The distribution of archaeologically documented cremation sites within cemeteries suggests that often several pyre areas were in contemporaneous operation, sometimes separate from, sometimes among the burials, usually away from the road frontage. There was little evidence for an association of the more permanent stone built pyre sites with urban sites, suggesting, in combination with the analysis of the documentary evidence, that the model of a public or municipal cremation place is not appropriate for these.

the frequency and type of materials burnt on the pyre with the dead. Pyre treatment in southern Britain remains less easy to reconstruct because of the less frequent inclusion of pyre debris in graves; the cremated animal bone accidentally collected with human remains gives us a view of the common accompaniment of the dead on the pyre. The glimpses of other pyre furnishing in varied evidence indicates the diversity of practice and the potential significance of this stage of the ceremony, including spectacular individual instances of destruction of objects by fire in town and country. As things stand a combination of differences in survival and recording of pyre debris deposits and of the traditions practised in individual cemeteries, perhaps related to the origins of burying communities, as argued for Brougham, frustrate generalisations about the material significance of the pyre. 3.6 Conclusion This chapter has used documentary and archaeological evidence to discuss the non-burial features from cemetery sites, with a particular focus on the pyre site and other contexts which may contain evidence for the ceremonies allied to cremation and commemoration. Such features have received limited attention through excavation, but accumulating evidence indicates their importance in reconstructing and interpreting burial rituals, as well as in relativizing the importance of grave goods, the traditional focus of analysis.

In the final section of the chapter, the material importance of cremation as a stage in funerary rituals in Britain was assessed through the evidence of the artefacts burnt with the dead. The distribution of data of different types heavily conditions our perception of the importance of the pyre site, but notwithstanding spectacular individual instances, it was argued that emphasis in previous scholarship on a northern and military preference for the pyre as the focus of display cannot be supported. The limited evidence instead suggests significant local and variability. With the accumulation of a much larger dataset it will be possible to generalise in this area; beyond the site-specific level its reality as a spoiler or qualifier for arguments from grave goods alone is better substantiated than as an aspect of ritual with systematic variation in time and space.

Documentary evidence has been shown to have limited potential for reconstructing the form the pyre and its associated ceremonies took. The distinction between bustum and ustrinum adopted by archaeologists from Festus is little reflected in wider Latin usage, and does not help us anticipate what form a bustum might take, if indeed we can be certain that it was always perceived as a distinctive burial type. Of the epigraphically attested ustrina we know relatively little. We remain ignorant of whether they were enclosures within which temporary pyres were erected or whether they took a more permanent form to enable burning. The remnant of Roman period pyre sites mostly comprises the pits over which pyres would have been constructed, both perhaps for ventilation and as a repository for pyre debris. Possible flues were documented at a handful of sites. As for the larger pits, it is not clear whether or not pyres were set up within or above them. Perhaps they represent repeated re-use of the same area. The interpretation of the more durable structures is problematic. The most likely explanation is that pyres of wood were set up above or within them, but more elaborate construction may also have been intended to give them a monumental form. The most frequent form of evidence for the pyre is the often hitherto poorly recorded burnt areas and spreads or dumps of burnt debris, which can potentially be confused with other cemetery features including graves containing deposits of pyre debris and other features related to commemorative ritual and feasting. Criteria have been offered for differentiating different feature types, but only limited re-analysis is possible of features where only limited evidence is recorded (see also Weekes 2008). 39

CHAPTER 4: BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS (VERULAMIUM) AND NEIGHBOURING MINOR CENTRES 4.1 Introduction The nature of burial evidence having been considered in previous chapters (2 and 3), the focus now shifts to examining burial rituals in the first sample area. This chapter documents the burial rites practised at St Albans and at three neighbouring minor centres to its north-east, Baldock, Braughing and Welwyn (Fig. 4.1). Together these settlements provide one of the most substantial regional funerary datasets available for Britain in the Late Iron Age and early Roman period (2.3) and, as others have noted, are the product of a common ritual tradition (Jones 1983; Philpott 1991). After a review of the evidence, the different stages of funerary rituals are established for burials of late Pre-Roman Iron Age to early third century AD date, with the main focus lying on those of Roman date. The small sample size from all of these sites except for Baldock means that late Roman burials are not considered in detail (cf. 6.2). The evidence generally dictates a focus on grave goods, although it is sometimes possible to comment on pyre rituals and monuments. Cemeteries have usually been selected for closer examination where 50 or more burials with adequately recorded assemblages are recorded, but important individual or smaller groups of burials are also considered. Here and in chapter 5 burials are considered ‘urban’ if they derive from one of the collective cemeteries of these settlements, as revealed by excavation, or are located in their immediate environs, i.e. up to c. one kilometre from the centre of settlement areas. The fuzziness of the distinction between ‘urban’ and ‘rural’ burial is considered further in a later chapter (7.3). In the different settlements studied the chapter explores the ways in which burial, through its rituals and spatial setting, served to negotiate social identities in these individual communities, in particular through differentiated resources consumed in burial. The conclusion (4.7) compares the analyses of the urban and minor centre cemeteries to assess the relative importance of burial in this regard. The analyses in this chapter, together with that of rural burials in the hinterland of these sites (6.2), serve as a basis for assessing the distribution of burial display across different types of settlement (7.3.3) and for exploring the character of that display, especially as seen in artefacts deposited as grave goods (8.2).

fieldwork, c. 900 burials of first to early third century AD date have been excavated since the 1960s, most from the southern cemeteries of the town (Niblett and Thompson 2005: 141). Not all are yet published, the key unpublished excavations being those at the St Stephen’s cemetery, the only early Roman cemetery from the city so far to have been subject to extensive fieldwork. The lack of funerary, as well as other monumental stone inscriptions, from the town must be in part attributed to the robbing of masonry-built tombs in an area poor in building stone, but the significance of this factor is difficult to assess. Substantial numbers of burials have also been excavated from nearby minor centres. The LPRIA and Roman cemeteries at Baldock have been more extensively examined than at any equivalent site in Britain. Of the c.1800 burials excavated, those revealed from Walls Field by Westell (1931) and by work directed by Ian Stead in the 1960s have been published (Stead and Rigby 1986). One cemetery has been published and interim discussions and a digital draft of the final report exist for other excavated from the 1970s to the first decade of the 21st century (Burleigh 1993; 1995a; 1995b; Burleigh and Fitzpatrick-Matthews 2007; 2010). More modest published samples of early Roman burial evidence are available from Braughing and Welwyn. To a greater extent at St Albans and to a lesser extent at the minor centres the fieldwork on the settlement areas also provides an archaeological context for the cemeteries. Excavations at other minor centres in Hertfordshire, for example Ware and Cow Roast, focused on the inhabited areas, provide additional evidence for the form taken by minor centres (Bryant and Niblett 1997). 4.2 St Albans 4.2.1 Iron Age and Roman cemeteries at St Albans Figures 4.2 show the distribution of Roman cemeteries around St Albans; in several cases to the south of the city their use began in the late pre-Roman Iron Age. A majority of burials so far recorded date to the late Iron Age and early Roman period, in particular the first century AD. These have been excavated in the predominantly cremation cemeteries south and south-east of the town, especially the King Harry Lane and St Stephen’s cemeteries, and small groups and individual burials in the Verulam Hills Field cemetery, south-east of the London gate.

Despite the richness of this sample its limitations should be stated. The prehistoric background to late Iron Age and Roman burial practice in Hertfordshire is difficult to establish, since not only burials but few EIA and MIA sites of any type have been excavated. By contrast there is abundant late Iron Age burial evidence, much of which is associated with the ‘oppida’ which have dominated excavation in the county (Bryant and Niblett 1997; Bryant 2007). St Albans (Verulamium) possesses one of the largest samples of late Iron Age and early-mid Roman burial data from any urban site in Britain. Of the c. 1200 burials recorded by antiquarian research and more recent

The first burials in the King Harry Lane cemetery, dated to the very late first century BC or first years of the first century AD are the earliest known from Iron Age St Albans. Like the other known pre-Roman cemeteries, the King Harry Lane site lies downhill from the monumental banks and ditches that (partly) demarcate the valley slopes of the Ver and the river crossing associated with the St Michael’s enclosure, beneath what later becomes the Roman forum. The first use of this cemetery is

40

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

contemporaneous with the creation of several other elements in the Verulamium complex, including the latter as well as the Prae Wood and Gorhambury settlements. The cemetery may be a communal burial space for groups occupying several of the dispersed settlements of the oppidum, although it is not the only such. There is a

discrete, primarily cremation cemetery of unknown size at Verulam Hills Field and a scattering of cremation and inhumation burials associated with the St Michael’s enclosure’ (Haselgrove and Millett 1997; Niblett and Thompson 2005: 36-37; Pearce 1997).

Figure 4.1 Verulamium and neighbouring minor centres, with possible civitas boundary shown (alternating dashes and dots)

Figure 4.2 The distribution of Roman period burials, St Albans (after Niblett 2000: 97, fig. 10.1)

41

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

In this phase the dead were deposited in spaces separate from those of the living. The conversion of the settlement with its dispersed foci into the planned space of a Roman city perpetuated this separation (with the exception of infant burials) but reconfigured it to place the dead around the planned and inhabited area of the city, in close association with the creation or formalisation of a road network. The latter seems to have ended the use of some cemeteries and instigated the development of others. The main use of the King Harry Lane cemetery area ended in c. AD 60, although it is not clear whether this is a direct consequence of the construction of the Silchester road which bisects it. However this area south of the Roman town continued to be used for burial. At roughly the same time as burial at King Harry Lane ended and burial at Verulam Hills Field was interrupted, the use began of the St Stephen’s cemetery on both sides of Watling Street, though it may have pre-conquest beginnings (Niblett 1999: 400). At St Stephen’s some of the earliest burials were placed by Watling Street over a kilometre from the initially small pre-Flavian centre of the Roman town and 450m from the third century walls, perhaps implying that as at other cities the designation of cemetery space may have allowed for significant urban expansion (cf. Esmonde Cleary 1987: 137), though the higher ground adjacent to Watling Street may have been preferred for its visibility (see 7.2.1). The early Roman cemeteries to the north of Verulamium are poorly known, with the exception of the Folly Lane mortuary enclosure set out probably in the post-conquest period (c. AD 45-50). Here

by the late first century AD a sanctuary had been constructed, with small groups of cremation burials in its environs. The King Harry Lane and St Stephen’s cemetery will be the major focus of discussion, with some reference to the smaller groups of late first century BC to early third century AD burials. As noted above, evidence of late Roman burial practice at St Albans is slender, but excavations in the city’s late Roman cemeteries, mostly north-east of the river Ver, suggest that it resembles that at other late Roman cities. Most burials were of unaccompanied east-west oriented inhumations, a very small number being differentiated by placing in stone or lead coffins or within monuments; the lack of associated objects means that dates attributed to burials often span the late third to fifth centuries (Niblett and Thompson 2005: 142-3). 4.2.2 Burial practice at King Harry Lane, St Stephens and other cemeteries of the 1st-2nd centuries AD The earliest burials at King Harry Lane were mainly deposited within the subrectangular ditched enclosures which subdivided the cemetery, while later burials were more often deposited in the areas beyond (Fig. 4.3). Burial was divided by the excavators into four phases. There is some disagreement on their precise date, but the main use (phases 1-3) seems to last from the last decades of the first century BC to c. AD 60, or perhaps a little earlier (Haselgrove and Millett 1997: 291). The rate of burial per year peaked in phase three (c. AD 40-60).

Figure 4.3 Late Iron Age and early Roman burials at King Harry Lane (inhumation burials shaded (after Stead and Rigby 1989)

42

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

Figure 4.4 The Folly Lane burial complex and environs to south (after Niblett 1999: 15, Fig. 8)

The evidence for burial rituals derives from the graves themselves, since pyre sites were not recorded. The majority ritual is cremation, although 17 burials were inhumed, several being placed in the ditches that demarcated individual enclosures. Cremation was generally complete, with 74% of burials recorded as ‘well-burnt’ from the uniform white colouring of the cremated bone (Stirland, in Stead and Rigby 1989, 241). Fragments of cremated animal bone, primarily pig or chicken, were recorded with 22% of burials. Some of the phase 1 burials at the centre of the enclosures with large grave good assemblages also included molten metal fragments, one of which was identified as a wine strainer (Niblett 2000: 101). Glass vessels were also recognised by molten fragments in four graves from phase 3 and one from phase 4. Collection of cremated bone varied, as at most sites, between a few grams and over two kilograms, with the majority of burials (82%) represented by less than one kilogram, i.e. involving only partial collection of remains from the pyre. The area of the burial pit dug at King Harry Lane, in the small minority of cases where it could be determined, declined slightly over time, from almost 0.5m² in phase 1 to less than 0.4m ² by phase 3. Since the surviving grave goods only occupied part of this area, this may indicate a decreasing frequency in deposition of perishable grave goods. This phenomenon

may also relate to factors influencing the placing of the dead. The cremated bone was placed on the western side of the pit in one third (15) of cases in the 46 burials where position could be established. In 331 of 455 burials the cremated remains were deposited in pottery vessels, especially beakers and jars. Where not so contained, the regular clustering of bone indicated the probable existence of a perished organic container. Wooden covers were placed over the burial in a small number of cases. Ceramics were the most common grave goods, with 211 of the 472 burials containing between one and ten vessels, though only 40 contained three or more. The largest ceramic assemblages, including those associated with several ‘founder burials’ at the centre of the ditched enclosures, date to phase 1. Over time the average number of accessory vessels declined, with burials outside the enclosures containing fewer than those within (Millett 1993). In earlier phases beakers, jars and platters were the most common forms, many being Gallic imports. Samian is scarcely represented until phase 4, i.e. from the midfirst century AD onwards. The 241 brooches excavated on the site form the other dominant category of grave good, with other items (e.g. bracelets, mirrors, coins) being represented in a single or a handful of cases.

43

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 4.5 Roman burials on Watling Street excavated in the 1930s and 1980s excavations at St Stephens, St Albans (after Niblett, St Albans archive)

Except for the enclosure ditches evidence of grave monuments was not recovered. However the lack of intercutting and the near circular arrangement of burials in some enclosures suggest that surface markers must have signalled the place of most if not all burials. Upcast from the ditches is likely to have been used to create a mound, though direct evidence of this did not survive.

cart or chariot, a carnyx, horse gear, and an ivory-inlaid chair or couch, molten copper alloy and four kilos of silver, probably from dining vessels, ceramic tablewares including samian, Gallo-Belgic wares and local imitations and amphorae, exceed anything known from St Albans or contemporary burials elsewhere (Appendix 6.4). The construction of a temple in the centre of the enclosure in the later first century AD is also exceptional.

At St Michael’s and Verulam Hills Field burial practice of the mid-first century AD was similar, though grave good provision was less generous, but the funeral conducted at Folly lane in the middle of the first century AD differed in scale and complexity (Fig. 4.4). A massive ditched enclosure (c. 2 ha) contained not only the pyre site and burial but also a shaft within which a timber chamber may have stood for some time to house precremation rites for the dead. The elaborate character of these is indicated by the items smashed and thrown into the collapsed mortuary shaft or consumed on the pyre. The quantity and diversity of burnt and broken pieces including faunal remains, chain mail, fragments from a

With the evidence from the St Stephen’s cemetery analysis of burial ritual can be continued up to the early third century AD (Fig. 4.5). 94 grave groups from this cemetery of mid-first to early third century date were published by Davey (1935) from chance finds in the 19th century and excavations in the 1930s on King Harry Lane, east of the Iron Age cemetery. To the north-east 340 further burials of the same date range were excavated in four campaigns during the 1980s by Hertfordshire Archaeological Trust; these were much closer to and on either side of Watling Street (Niblett 2000; Niblett and Thompson 2005: 138-141). The ritual practised is predominantly cremation, although a small number of 44

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

Figure 4.6 Percentages of burials with varying quantities (g) of cremated bone (St Albans King Harry Lane; St Stephens; Skeleton Green; Braughing ‘B’; Baldock Wallington Road; Winchester Victoria Road

individuals were also inhumed. Some non-burial cremation-related features have been identified on the site. Possible brick-built pyre sites were excavated by Davey (3.3) and there are also features containing charcoal, fragmentary artefactual material and small quantities of cremated human bone which might represent pyre debris deposits (Appendix 2.2.1). In most cases the high efficiency of cremation can be deduced from the consistent white colour of the cremated bone. Only 12% of cremations included fragments with grey or blue colouring suggesting incomplete combustion of the organic component. The evidence for pyre rituals otherwise mainly comprises the scraps of animal bone or artefacts inadvertently collected with the cremated bne. 43% of individuals were cremated with animal parts, including sheep/goat, pig and chicken. The presence of molten copper alloy fragments and staining of cremated bone in a small number of burials suggest that ornaments had been deposited on the pyre or that individuals were cremated dressed. Fragments of burnt glass were occasionally noted. The cremated bone collected and buried varied between a few grams and over two kilograms, i.e. quantities equivalent to those at other sites (Fig. 4.6). Where they could be recognised grave cuts were small, usually only just large enough to take the cremated bone and grave goods.

Figure 4.7 Numbers of different ceramic forms, St Stephens (Davey and HAT excavation combined) 100 90

The majority of cremation burials (over 80% in the 1980s excavations) were deposited in pots, mostly Verulamium Region jars. Of the rest some were deposited un-urned and a handful in wooden caskets of which the lion head fittings survived, a container type also used in other Roman period burials in Hertfordshire and Cambridgeshire (Philpott 1991: 13-14).

80 70 60 50 40 30

The commonest grave goods were ceramic vessels, glass being much rarer, deposited only with 11 of the burials excavated in the 1980s. Forty five burials were furnished with grave goods of other types, most in a handful of instances (e.g. lamps, mirrors, brooches, bone pins, hobnails), though coins were found in 15 burials. Among the ceramics flagons and beakers in locally made fabrics were the commonest types (Fig. 4.7). The number of vessels varied between one and ten, but few burials contained more than three (Fig. 4.8). At the time of examining the archive, dates were available for burials based on spot dating of individual ceramics, especially Verulamium Region Ware. Examination on this basis did not detect a change in the number of vessels deposited in burials over time. Higher numbers of ceramics in the burials excavated in the 1930s may reflect failure to recognise unurned burials without grave goods rather than a zonal difference in the cemetery.

20 10 0 Fl.

Bkr

Bow l

Jar

D/Pl.

Cup

Lid

Vessel

Figure 4.8 The number of burials with different numbers of ceramics, St Stephens (SS1 = Davey excavation; SS2= HAT excavations) 250

200

150 SST1 SST2 100

50

0 0

45

1

2

3

4

5

6-10

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Evidence was found for a limited range of enclosures and burial markers. South of Watling Street were two rectangular enclosures, one with a burial at its centre and another of cill beam construction around which a group of first century burials clustered, but without any evidence for burials in its interior. Niblett (2000) suggests that this may have served for display or exposure of the corpse prior to burial. To the north of Watling Street was a six post structure beneath which the cremation burial with the largest grave good assemblage from the cemetery was recovered. Four or six post-hole structures were also documented above four other graves. In one instance a burial appeared to have been marked by a sheep skull placed on a post. In the excavations of the 1930s a flint foundation was noted in association with one burial. Otherwise the lack of inter-cutting makes it likely that other burials were marked.

tomb finial, found in the bed of the Ver north of the town (Blagg and Hunn 1984) and a possible mausoleum detected on an aerial photograph north-west of the road that runs to the north-east gate (Niblett and Thompson 2005: 145). A substantial third or fourth century AD flint and mortar tomb (4.2m x 2.6m) with a painted ceiling has been documented in the Verulam Hills Field c. 100 metres outside the south-east gate (Niblett and Thompson 2005: 143).

The other evidence from early Roman Verulamium shows similar burial practices. The first and second century cremation burials occurring in small clusters south of the Folly Lane sanctuary (Fig. 4.4 shows the largest group of 11 burials), the first-second century burials at Verulam Hills Field and second-third century burials from the King Harry Lane site are not dissimilar to the St Stephens sample, though furnishing is more modest. Pyre goods comprise cremated animal bone, predominantly sheep or goat, but also burnt ceramics, molten glass unguent bottles and occasionally other items. There was considerable variety in the way the bone was deposited in the grave, sometimes with debris from the pyre. Some burials lacked evidence for a container and where presented they included not only ceramic vessels but also, in individual instances, an amphora and wooden casket. Ceramics, mostly single vessels, dominate grave good assemblages. Items in the fills of shafts in the same zone as cremation burials south-west of Folly Lane enclosure entrance indicate their ritual character (animal skulls, near complete ceramics, face pots, and, in one instance a defleshed cranium). The excavator has suggested that the cutting and filling of these shafts may have a connection to mortuary rites, though their date range is generally later than that of the burials and their filling may be the product of rituals related to the sanctuary to the north rather than directly to nearby burials.

4.2.3 The spatial context of burial in Late Iron Age and early Roman St Albans The documented Late Iron Age cemeteries lie within the monumental dykes running along the south-west side of the Ver valley and downslope from them. The King Harry Lane cemetery is near the hypothesised prehistoric precursor of the Silchester road, but was not structured around it. Instead ditched enclosures form its nucleus, several containing primary burials around which others clustered. The layout of burials seems therefore to have been structured by features internal to the cemetery. The Folly Lane complex impinges much more significantly on its surroundings through its monumental enclosing features and its situation on the crest of the hill north-east of the town, as well as the mound on which the pyre was built (Fig. 4.2). Verulamium’s earliest streets and the Folly Lane enclosure were set out at roughly the same time in the late 40s or 50s of the first century AD. The fragmentary evidence for pre-Boudiccan urban layout suggests a rectilinear network of streets, within which were timber buildings, masonry structures and, possibly, a ‘proto-forum’, with the pre-Roman central enclosure ditch and a bank and tower at the river Ver serving as boundaries. The Folly Lane enclosure must have been one of the most substantial and imposing elements of the city that was coming into being. The construction of a temple over the pyre site and the re-cutting of its ditch in the late first century AD coincided with the postBoudiccan monumentalisation of the city. The excavator has suggested that the Folly Lane sanctuary was connected to the monuments of the city’s centre through a processional route involving the theatre and temple northwest of the forum, the Branch Road baths (Niblett 1999: 416-7; Creighton 2006: 135-48). While this is necessarily speculative, it serves to emphasise the site as a significant monumental component of the urban landscape.

There are occasional exceptions to this relative homogeneity of practice. A Flavian burial (‘William Old’) to the south of the town and east of the King Harry Lane cemetery contained a very substantial assemblage of grave goods, including a large number of ceramics, primarily samian dishes and cups. Burial 27, of Antonine date, accompanied by at least eight vessels and a strigil, outside the Folly Lane enclosure, is much more richly furnished than its neighbours. An infant burial in a lead coffin at Verulam Hills Field included textile fragments associated with gold thread, as well as items of apotropaic significance, including a phallic amulet, beads and shells. Apart from the timber structures documented at St Stephens, the limited evidence for funerary monuments comprises a possible fragment from a tower

On current evidence the area along the Colchester Road did not otherwise become a monumentalised street of tombs. The small clusters of cremation burials south of the temple enclosure are perhaps associated with extramural settlement rather than a more extensive cemetery. To the south of the city on the St Stephens site, burials lined either side of Watling Street as it climbed out of the Ver valley towards the south. A narrow strip of burials on its northern side cut earlier settlement and a broader, probably discontinuous zone of burial extended for several hundred metres to the south of the road (Fig. 4.5). The apparent thinning of burials away from the road frontage is in part misleading since early twentieth century housing probably destroyed much evidence (Frere 1987a: Fig. 15, 328). There were, seemingly, other 46

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

axes through the cemetery: if the course of the road excavated in 1935 is projected northwards it traverses the area free of burial to meet the causeway over the roadside ditch. The few more substantial structures, dated to the first century AD, were concentrated on the road frontage (see above). The enduring attraction of the road line is shown by the early third century deposition of a lead coffin at right angles to the course of the road, cutting the lip of the roadside ditch. In general however both burials with larger grave good assemblages and four post structures were distributed in the cemetery without any particular concentration. The earliest burials were dispersed across the whole of the cemetery which must therefore have developed from several discrete foci. The pyre sites identified by the possible cremation pits south of the burials adjacent to Watling Street and the brickbuilt ‘crematoria’ identified by Davey probably also served individual groups (3.3.2). The significance of Watling Street as a structuring feature for cemetery layout should not therefore be exaggerated.

The burial rituals documented at the Folly Lane complex echo these local traditions but in every respect exceed them, especially in the scale of the monument and in the spectacularly destructive burial ceremony. Speculation over the ‘regal’ identity of the individual buried is possible (e.g. Niblett 1999: 396-401), although individuals without a political role in their own right could be used in death for the projection of dynastic power. The Folly Lane evidence is perhaps more important as an insight into political theatre, some participants even perhaps recalling the spectacular cremations of members of the imperial family at Rome (Creighton 2006), or more probably remembering the contemporary and earlier ceremonies at Colchester and in Gallia Belgica (Crummy et al. 2007; Fitzpatrick 2000). The embedding of a monument associated with a particular individual or group into the space of the newly created town can also be compared with other provincial cities (7.2.2). The significance of the complex endured through its conversion into a sanctuary; perhaps the integration of rituals associated with it into public life allowed those who claimed a connection with a potent ancestor to exploit an association with the monument to their advantage.

There is limited contextual evidence for other burials south of the city. The richly furnished Flavian-period ‘William Old’ cremation burial, c. 100m from the later town gate, 12m east of the Silchester Road, is not closely associated with a known cemetery, but its environs by the road have not been investigated (Fig. 4.2). Further south the ‘Lindum Place’ burials, also richly furnished and, in one instance, set within an enclosure, were associated with a rural settlement in the city’s hinterland.

So far there is little evidence for subsequent emulation of the Folly Lane rituals or setting. The most striking aspect of post-conquest burial practice up to the third century AD is the limited evidence for differentiation between burials. The quantity of material burned or buried is more modest than before. Very few burials are distinguished by more than a joint or two of meat on the pyre and one or two accompanying pottery vessels or other items. Graves are cut to a minimum size sufficient to take the cremated bone and grave goods. Comparisons with the fills of pits and ditches in settlement areas reveals the common availability of the artefact types deposited in graves (e.g. see Frere 1984b). The percentage of samian in the burial assemblages is a little higher than in general circulation but this can be explained by the predominance of tableware in funerary ceramics in general (cf. Willis 2004: 7.2.4). With occasional exceptions tomb markers are modest in materials and scale: there is little evidence for monumentalised Gräberstrassen. This modesty is striking when compared to the settlement, since this period saw the construction and reconstruction of public buildings from the Flavian period to the later second century AD, including the forum-basilica, theatre as well as temples and bath houses (Niblett and Thompson 2005: 146-52). The increasing use of masonry, painted plaster, opus signinum and mosaic floors in the second century AD indicates the investment of wealth in private housing too (Niblett and Thompson 2005 157-8; 162-3).

4.2.4 Burial and society in late Iron Age and early Roman St Albans In contrast to neighbouring oppida (see below) there are no ‘Welwyn’ type burials of mid to late first century BC date at pre-Roman St Albans: few if any of the burials so far excavated predate the first century AD. The earliest burials at King Harry Lane, especially at the centre of the ditched enclosures, are distinguished by larger quantities of grave goods and occasional evidence for the destruction of material on the pyre. Millett (1993) interprets the subsequent decline thereafter in number of ceramics deposited, especially in burials outside the ditched enclosures, as evidence of social dislocation accompanying the process of urbanisation. In his view these later burials were those of individuals less well integrated into local social networks able to provide for funerary rituals. An alternative scenario can however be proposed, remembering that in any case only a minority of the population may have been buried in an archaeologically visible form (cf. 2.4). The burial rite arguably shifted from differentiating to expressing homogeneity. There is little evidence for age or gender based differentiation in burial ritual, especially after cremation. Though likely to have served several of the dispersed settlement nuclei that constituted the oppidum, the cemetery played host to rituals in which the assimilation of the deceased to a common group of undifferentiated ancestors perhaps prefigures the process of settlement agglomeration at Verulamium (Pearce 1997).

4.3 Baldock 4.3.1. Iron Age and Roman cemeteries at Baldock At Baldock an extensive Iron Age settlement developed into a Roman period minor centre, recently estimated as covering up to 48ha at its maximum extent (Burleigh et al. 2006). Approximately 1800 burials have been excavated from more than 20 cemetery areas, dating from 47

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

the first century BC to fifth century AD (Fig. 4.9). The scale of investigation of its cemeteries is almost unsurpassed for Roman Britain, although a population of a few hundred individuals could easily account for this number of burials. Full publication will allow detailed comparison of burial practice over the late Iron Age Romano-British period and any discussion in advance of that must be provisional. Here detailed analysis is offered only of two cemeteries, Wallington Road and Walls Field, although information from others is exploited in discussion. Since only a small number of the many cemetery areas at Baldock are referred to, site nomenclature is taken as far as possible from Burleigh and Fitzpatrick Mathews (2007), which revises the names and codes employed in earlier accounts (Burnham and Wacher 1990: 281-8; Stead and Rigby 1986).

1995a; Burleigh and Fitzpatrick-Matthews 2007: 44-6). Existing land divisions and routeways appear to structure the location of first century BC burials and as at St Albans the separation of burial from settlement space prefigured Roman practice. To the north-east of the settled area several groups of Iron Age cremation and inhumation burials, some placed within ditched enclosures, were located along a ridge immediately north of an earlier pit alignment which was transformed in the first century BC into a continuous bank and ditch (Burleigh 1995a: 105; Burleigh and Fitzpatrick-Matthews 2010: 24-26). The enclosures were adjacent to route-ways crossing the settlement and cemetery areas (e.g. the California enclosures), and burials in at least two of them were marked by barrows. As well as small numbers of burials associated with individual enclosures (e.g. Meria Road) they were also documented to the east of the settlement at Wallington Road and probable examples are also known from Walls Field. The ‘Tene’ burial remains the sole pre-Roman burial recovered on the south-western side of the town.

So far no cemeteries equivalent in size to King Harry Lane have been show to exist at Baldock but small groups or individual cremation and inhumation burials of late Iron Age and occasionally earlier date have been recorded from several areas of the oppidum (Burleigh

Figure 4.9 Roman Baldock, with individual LPRIA and Roman burial areas indicated (adapted from Burnham and Wacher 1990, 284, fig. 96) (CAL = California; CP = Convent of Providence; IWE = Icknield Way East; MR = Mercia Road; RR = Royston Road WLR = Wallington Road WAL = Walls Field; Cl.R. = Clothall Road; TE= The Tene)

48

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

In the early Roman period the relationship of settlement to cemetery showed broad continuity from the preceding period but with some change. Some burial areas were abandoned or encroached on by the expanding Roman period settlement, while others continued in use, most bounded by ditches (Burleigh and Fitzpatrick-Matthews 2010). To the north-east settlement spread over Iron Age barrows and enclosures at California. A road was driven through the centre of a probable burial enclosure (Downlands Enclosure A). However further to the north LPRIA cemeteries at Royston Road, Icknield Way and to the east at Wallington Road continued in use. To the east the principal period of burial at Walls Field and Clothall Road began in the early Roman period, seemingly also the case for the small groups of early Roman burials excavated south of the town. Burial continued until the early fifth century in some of the cemetery areas focused on Royston Road and Icknield Way and resumed closer to the centre of the settlement in the California area in the late Roman period. A large inhumation cemetery to the south of the Tene also probably dates to the late Roman period. Within the settlement thirty five infant inhumation burials were recovered from enclosures A, B and C (Burleigh and Fitzpatrick-Matthews 2007: 119-20) and adult inhumation burials were also occasionally excavated from early and mid-Roman settlement contexts and in ditches and in the final part of well infill sequences in the late Roman period (Stead and Rigby 1986: 87; Burleigh and Fitzpatrick-Matthews 2007: 115). The latter were interpreted by Stead and Rigby as evidence for the dereliction of the settlement, but can instead be seen in the light of long term intra-settlement deposition elsewhere (6.4.2). Deposition within settlement contexts of fragmentary human remains, predominantly skull fragments, also seems to have occurred on an occasional basis in the early Roman period. Examination of their context suggests that these are better classed as ‘structured deposits’ (cf. 3.4) than as the product of disturbance of earlier burials. In three pit fills (A111, A414 and B50) fragments coincided with the presence of complete pots. Stead and Rigby (1986: 258) noted the relative frequency of the latter and drew an explicit parallel with ceramic deposition in burials but did not note this presence of associated human bone. The location of such deposits at junctions of features and entrances suggests further care in deposition (cf. 6.4.3).

large bronze cauldron, two fire dogs, two bronze dishes, two wooden buckets with copper alloy fittings, a single Dr1A wine amphora and a small amount of cremated bone including bear phalanges. The continuing use of the tomb as a focus of cult and burial activity into the Roman period suggests that it may have been marked by a monument. The ritual conducted for the barrow burial surrounded by a very substantial square ditched enclosure at California cannot be fully documented because a corn drier was later dug into it, but as well as a small part of the burial with a copper-alloy bound bucket, a grogtempered pedestal jar and the chopped up remains of two pigs, an adjacent feature contained pyre debris and grave goods, including bones (horse, cattle, sheep/goat, pig, fowl, amphibian and fish), bronze and iron fragments from chain mail and perhaps furniture fittings. The very little preserved human bone suggested this was an adult burial. From the mid first century BC onwards, and in increasing numbers towards the end of that century and in subsequent decades much more modestly furnished cremation and inhumation burials were made within enclosures of a similar size to those at King Harry Lane and as unbounded groups of burials, some of which continued to develop into the Roman period. Burials in phases 1 and 2 at the Wallington road cemetery (see below) illustrate the more typical rituals practiced in the Late Iron Age as well as offering a sample of Roman burial practice, predominantly for the first and second centuries AD. 4.3.3 Burial practice at the Wallington Road cemetery The Wallington Road cemetery lies on the eastern side of the Iron Age and Roman settlement, on the ridge on which several groups of burials were sited in the late Iron Age. In its environs are several strip enclosures (Stead and Rigby 1986: 30, Site S); the ditches which enclosed many of the burials may relate to this broader enclosure system. In the first and early second centuries AD they bounded the cemetery on its north-western and southeastern sides but during the second century they silted up and burial took place beyond them. Deep ploughing had damaged a significant proportion of burials and the cemetery was excavated in difficult salvage conditions in early 1982, producing c. 25 inhumation and 150 cremation burials. The excavators estimate that 90% of the cemetery was examined, although scattered burials to the south-east may imply a larger original cemetery area. The cemetery was in use for burial from the mid first century BC to the early fourth century AD, although the frequency of burial varied, the peak of use occurring during the second century AD (Fig. 4.10).

4.3.2 Iron Age burial practice at Baldock In the Late Iron Age a diversity of burial practice is attested in cemeteries associated with Baldock and skeletons and individual bones have also been recovered from non-cemetery contexts. The earliest deposits are those in a solution hollow adjacent to the enclosure around the LPRIA barrow burial at California, which includes human bone fragments as well as occasional complete bodies dating from the fifth century BC to the LPRIA (Burleigh 1995a: 105; Burleigh and Fitzpatrick Matthews 2007: 111-12). There are two lavish cremation burials of the first century BC. The assemblage in the Tene burial, dated to the first half of the first century BC, a very early, if not the earliest of the ‘Welwyn’ type within the Aylesford-Swarling tradition, comprises a

Burials of phase 1 and 2 (i.e. from c. 50 BC – AD 70) were predominantly inhumations, but from the late first century AD cremation was the exclusive rite. Most inhumations were oriented north-east to south-west, or south-east to north-west. Where skeletal evidence survived the head was laid to the east. The orientation of the enclosures in the area in which the cemetery was situated may have influenced this. Inhumation burials were rarely furnished. The colour and condition of bone indicates that most cremated individuals were fully burnt 49

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

and the quantities collected are similar to those from contemporary sites at neighbouring settlements (Fig. 4.6). There is no evidence for any particular body part having been preferentially deposited or excluded from burial. In a small number of cases charcoal was found with cremated bone but in general the latter had been carefully cleaned of pyre debris. Nor was the latter generally included in any quantity in grave fills, though this was the case in other Baldock cemeteries (Burleigh 1993: 43). The presence of some pyre debris was noted in analysis of the cremated bone, including occasional charcoal fragments and cremated animal bone of cow, sheep / goat and unidentified bird with 10% of cremations. Other pyre goods comprised a bone pin, lead object and glass bead, and, more commonly, nails and iron staining (in 27% of graves). McKinley and the excavators argue that many of these derive from cremation in a coffin or on a bier. The possible pyre site to the north of the cemetery contained burnt animal bone, possible elements of a bier or other furniture and very small quantities of mostly residual ceramics (App. 2.1.3).

cooking jars in local fabrics, many of which showed signs of use. Some had been trimmed before burial but it was not always possible to assess this because of the extensive damage to many burials, in some cases leaving only the base of the vessel containing cremated remains. This damage means that the number of burials without grave goods is almost certainly exaggerated in this analysis. Ceramics were the predominant grave good, flagons, beakers and samian dishes being the most common (Fig 4.11). Some accessory vessels were either seconds or had been damaged prior to burial. There were changes in the representation of different forms over time. The proportion of jars and flagons decreased over time, that of dishes increased and later diminished and that of beakers, especially colour coated beakers, came to predominate in phase 5. The vessels deposited with the burials varied in number from one to five, more than half being provided with at least one (Fig. 4.12). With the exception of nails other artefact types were recovered from only a very small number of graves. The most frequently occurring were hobnails, recovered from seven burials. The general lack of inter-cutting suggests that the position of most burials was marked; truncation by ploughing and difficult excavation circumstances meant how this was done could not generally be established, although some of the individual postholes noted in excavation may have served to mark burials.

In the few cases where a burial cut was recorded it was oval or, less commonly, sub-rectangular in shape. The cremated bone was usually placed in the centre of the grave; in nine of the 14 burials in which it was placed offcentre, the southern and eastern sides of graves were preferred. Almost all cremation burials were urned in

Figure 4.10 Wallington Road cemetery, Baldock: left, general plan of excavated burials; right, detail of burials from dashed area, shaded = inhumations (after North Herts Archaeological Service archive)

50

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

Figure 4.11 Numbers of different ceramic forms, Wallington Road

burials in the centre of the site separated two groups, both of which were in use from the late first to probably the fourth century AD. The vast majority of burials were cremations but some inhumations, south-north and westeast oriented, disturbed earlier graves. These are not considered here. No information was published on the cremated bone or the burial fill. Fabrizi’s reanalysis (1984) suggested that the excavator’s dependence on samian vessels for dating exaggerated the number of first century burials and underestimated those of third century date, but confirms that most date to the second century. In view of the predominance of the second century among the burials change in practice over time is not further studied here.

45 40 35 30 25 20 15 10 5

With rare exceptions cremation burials were deposited in ceramic jars. Of the ten unurned burials two were possibly in caskets. Casket fittings were recovered from other graves but may have derived from grave goods rather than containers. The majority of burials were accompanied by ceramics but other grave goods were rare. The ‘equipment’ category comprised primarily iron lampholders and lamps from five and casket fittings from six graves respectively. The glass assemblage comprised bottles, flasks and ‘decanters’ which illustrations show to be a label applied to both bottles and flasks. Glass vessels were generally deposited in burials containing larger ceramic assemblages, usually twice the average assemblage size.

0 Jar

Flagon

Beaker

Dish

Cup

Other

Figure 4.12 The number of burials with different numbers of ceramic accessory vessels, Wallington Road 60

50

40

30

Analysis of ceramic forms is hampered by the lack of comprehensive illustration or detailed descriptions. The illustrations reveal that the often-used term ‘vase’ applies to jars, beakers and bottles. The most frequently occurring accessory vessel forms comprised dishes (and a small number of cups), flagons and beakers. The remainder of the identifiable vessels were jars, occasionally bowls and a number of miniature vessels. As at Wallington Road there was a higher proportion of beakers in the accessory assemblage of later burials. The most common combination of forms was beaker, dish and flagon. The number of pots deposited varied from one to nine but few burials contained more than three (Fig. 4.13).

20

10

0 0

1

2

3

4

5

Figure 4.13 The number of burials with different numbers of ceramic accessory vessels, Walls Field

The evidence from Wallington Road and Walls Field by no means covers the entire range of burial practice in early Roman Baldock. While cremation predominates in other cemeteries, the extensive excavations to the north of the town in the Royston Road area (including sites at Sale Drive / Yeomanry Drive) and along Icknield Way and Stane Street show the continuing widespread use of inhumation. Across these cemeteries there is evidence for less effort being invested in burial than was documented above; colour of cremated bone suggests less complete combustion of organic elements in some instances, and of the regular deposition of inhumation burials without coffins. The unurned deposition of cremated bone is much more frequent, though the shape of the bone deposit suggests that in many cases organic containers have perished. Furnishing with grave goods is generally

4.3.4 Burial practice at Walls Field and other Roman period cemeteries The Walls field cemetery lies to the south of the settlement, between two of the routes approaching Baldock from the south east (Fig. 4.9). A 61 x 14 m area within it was excavated in the late 1920s. An area free of 51

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

less frequent than in either of the cemeteries examined above (Burleigh 1993: 45; Rosten 2007).

west. The emphasis on placing the dead by the routes into the settlement continued into the Roman period, but little trace can be detected of preferential selection of the road frontage as the location of lavish burial practice or monument building, for which there is in any case little evidence. The evidence from the Royston Road offers a more direct and extensive image of the topographical setting for burial (Fig. 4.14). To the immediate east of the settlement area, separated from it by a ditch were the pyre sites, to the east of these was the burial area, then on the north-eastern margin of the cemetery quarry pits which served as a dump for pyre debris and redeposited material from burials. Thus the burial ceremonies distanced the space of the dead from that of the living, from the pyre to eventual deposition on the furthest margins of the cemetery (and ultimately to oblivion by the destruction and scattering of the burial). This cemetery provides some further negative evidence for the significance of the cemetery road frontage as a space for monumentalisation. Its location on the immediate edge of the settlement, close to a junction with the Icknield Way and prominent on the ridge above the settlement lent it a situation for display which had been exploited for the Iron Age barrows. In the Roman period this was not the case: the average accessory vessel assemblage here was the lowest at Baldock (Burleigh 1993: 45). Similarly at Icknield Way Roadside immediately to the north burials cutting the ditches of the road were equally modest. Evidence of burial markers in this prominent location is almost absent, though the truncation of cemetery surfaces should not be forgotten.

The largest burial assemblages from Baldock derived from small groups of burials on the western and southern side of the town near the routes to St Albans and Braughing. To the south of Baldock three mid-first century AD burials excavated on Clothall Road and a single early second century burial at Convent of Providence are all characterised by cremated bone placed unurned or in a wooden box or casket, uncremated and / or cremated bone from several animal species and many vessels, including multiple samian deposits (7.3.3; 8.2). The Clothall Road burials, which cut an earlier ditch and one of which was cut by a later ditch, seem more likely to be part of a small group of burials, perhaps on the edge of a settlement compound but neither their context nor that of the Convent of Providence burial is well understood. A possibly late third century cremation burial of a juvenile with many lamps and pedestalled cups, along with other non-burial deposits of ceramics and animal remains was excavated at the Tene, close to the first century BC Welwyn-type burial. 4.3.5 The spatial context of burial at Baldock There is substantial continuity in the placing of the dead at Baldock from the Iron Age to Roman periods. In the Late Iron Age the placing of burial enclosures by trackways made passing of the spaces of the dead a recurring element of daily routine. The barrows on the ridge to the north-east would have been quite prominent, especially when viewed from the settlement to the south-

Figure 4.14 Burials and pyre areas at Royston Road (Area 15), Baldock (after Frere 1989: 299, fig. 22)

52

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

unfurnished cremation burials and a deposit of unburnt skeletal fragments from a minimum of fourteen individuals with evidence for excarnation, mixed with animal bone and pottery in fills in ditch F1, Station Road. This deposit dates from the late first century BC to mid first century AD. Well documented early Roman burial evidence from Braughing is confined to three cemeteries on the western side of the settlement, known as Skeleton Green and Braughing ‘A’ and ‘B’. The limited evidence from the Fordstreet site suggests that contemporary burial practices to the north-east of the town were similar. There is limited evidence for the spatial context of burials.

4.4. Braughing-Puckeridge 4.4.1 Iron Age and Roman cemeteries at BraughingPuckeridge A combination of excavation and survey data has shown Braughing also to have been an extensive late Iron Age complex extending over a considerable area (c. 40 ha) that developed into a Roman minor centre (Fig. 4.15) (Potter and Trow 1988; Bryant and Niblett 1997: 276; Burnham and Wacher 1990: 103-11). Later Iron Age burial evidence from Braughing is scattered but slight and attests to a variety of practices, including a possible Welwyn type burial (Whimster 1981: 373), a few urned

Figure 4.15 Roman period Braughing with cemeteries (stippling = area of occupation; squares = masonry buildings; triangle = burials) (adapted from Burnham and Wacher 1990: 104-5, figs 27-28)

Figure 4.16 The Skeleton Green cemetery, Braughing (after Partridge 1981: 246, Figure 90)

53

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 4.17 Braughing B cemetery (after Partridge 1977: 74, Fig. 29)

4.4.2 Burial practice at the Skeleton Green cemetery The cemetery lay 50m west of Ermine Street on the western margin of the Roman settlement (Fig. 4.15). Trial trenching suggested that contemporary settlement lay on higher ground to the north and east (Stead 1970: 38). In the late first century AD the earliest burials were cut through flint cobble platforms from buildings which had gone out of use half a century earlier. It is not stated in the report whether the earlier east-west road was still in use but its continued use is implied by the different cemetery phase ditches respecting it. The main period of use of the cemetery was the late first to mid-late second century AD. The excavator proposed that burial began in the southern enclosure, closest to the road and surrounded by the horse shoe shaped ditch and later spread north, necessitating an expansion of the enclosure (Fig. 4.16). However although the ‘heirloom factor’ makes dependence only on samian for burial dating dangerous (2.2), the distribution of Flavian date samian vessels shows that both areas of the cemetery could have been in use for burial from the later first century. Other burials (20; 28; 32) within the horseshoe-shaped enclosure are dated (Partridge 1981: 46) to the earliest phase without the argument for their date being presented. The case for a horizontal stratigraphy in the cemetery seems therefore not proven.

The majority of graves were earth cut and sub-circular or sub-rectangular in shape. The only exception was the wooden cist in 28, within which a wooden box containing the cremated bone and grave goods had been deposited. Most burials were placed in the grave in jars in local fabrics. Of these five were also placed inside wooden boxes, and one deposit of cremated bone was perhaps bagged before placing in a wooden box. Five burials were deposited in caskets with lion head fittings (see 4.2 above). Ceramics were by far the commonest grave goods. Hobnails, coins, glass and personal ornament were recovered with almost 10% of burials. The most spectacular assemblage was that of at least eleven glass vessels in 49. The ‘equipment’ category comprised mostly nails, of which the specific purpose could not be established. Among the different vessel forms (Fig. 4.18) flagons, beakers and dishes predominate, with many burials being accompanied by two or three vessels (Fig. 4.19), typically a flagon, beaker and samian dish or flagon and dish, as Philpott (1991: 34) also noted. Figure 4.18 Numbers of different ceramic forms, Skeleton Green

All the cremated individuals were efficiently burnt (Wells, in Partridge 1981: 290). The quantity of cremated bone deposited was similar to that at other minor centre sites in southern Britain (Fig. 4.6). Only a part of the total expected quantity of cremated bone was recovered, despite good preservation. As Wells noted, the near absence of skull bone, sometimes despite the collection of a very large proportion of the cremated bone (e.g. 1294g in 39), is a striking characteristic of several graves but one difficult to assess because quantities of cremated bone from different body areas are not reported. Little pyre debris was recorded with the cremated bone, although grave fills were not described. Save for a fragment of burnt samian, the only known pyre goods comprise cremated animal bone, predominantly cattle, sheep and fowl, recovered from 27% of burials.

45 40 35 30 25 20 15 10 5 0 Flagon Beaker

54

Bow l

Jar

Dish

Cup

Lid

Vessel

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

Figure 4.19 The number of burials with different numbers of ceramic accessory vessels, Skeleton Green

Figure 4.20 Numbers of different ceramic forms, Braughing B

90 20

80

18

70

16

60

14

50

12

40

10

30

8

20

6

10

4

0 Flagon

2

Beaker

Bow l

Jar

Dish

Cup

Vessel

0 0

1

2

3

4

6

Figure 4.21 The number of burials with different numbers of ceramic accessory vessels, Braughing B

A relationship between the location of this cemetery and the earlier road to the south is possible though not proven but in any case this is not a major route. Within the cemetery burials with large ceramic and non-ceramic assemblages, 49 and 33, as well as most of the casket burials were deposited in a semi-enclosed area in the northern part of the site rather than near to this route.

16 14 12 10

4.4.3 Burial practice at the Braughing ‘B’ and ‘A’ cemeteries Cemetery B lay 200m and the much smaller cemetery A c. 80m west of Ermine Street on the western fringe of the Braughing settlement (Fig. 4.15). Since the two sites are over 300m apart it is unlikely that they formed part of a continuous cemetery. Of the 104 cremation burials excavated at Braughing B, a few were possibly of late first or early fourth century date, but most dated to the second or early third centuries AD (Fig. 4.17).

8 6 4 2 0 0

1

2

3

4

5

6

The few burial assemblages in the Braughing ‘A’ cemetery were exceptional for their containers, two bronze decorated wooden caskets and one glass bottle, as well as the high number of ceramic and glass vessels deposited in some burials. The small scale of excavation makes it impossible to assess the size of the cemetery from which this group derived, although human and animal bones and ceramics in the small section of ditch excavated may suggest that they lie within a larger cemetery.

The vast majority of burials at Braughing B were efficiently cremated (Wells, in Partridge 1981: 290). Skull bones were absent from a number of undamaged burials, but again since the quantity of bone from different body areas was not reported it is difficult to assess the significance of this. Whether or not pyre debris was deposited with the burials is unknown as grave fills were not reported. Over a third of burials (37.5%) contained cremated animal bone, cattle, sheep and fowl bone predominating.

4.5 Iron Age and Roman cemeteries at Welwyn Understanding of the Iron Age and Roman period is more limited at Welwyn than at the other centres considered. For the Iron Age discrete settlements and burials are documented with more limited evidence for the major linear earthworks that characterised the contemporary oppidum centres discussed above (Bryant and Niblett 1997: 275). The burials, in particular the tomb excavated at Welwyn Garden City after its disturbance by a pipeline, are central to characterisation of the AylesfordSwarling burial tradition of Late Iron Age south-east England: this tomb gave its name to the most lavish burials of this tradition and to the first phase of such burials, i.e. those that preceded the introduction of

Grave cuts were circular, oval or subrectangular earth cut features, one (65) of which was possibly lined with wickerwork. With the exception of one burial in a face pot the rest were deposited in jars in local fabrics. The predominant grave goods were ceramics. The different ceramic forms were evenly distributed between graves (Figs. 4.20 and 4.21), most commonly in combinations of flagon, beaker and dish, flagon with dish or beaker, or beaker with dish (Philpott 1991: 34). The ‘equipment’ category consists entirely of nails and corroded iron objects.

55

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

imported table wares from Gaul and Italy in the last decades BC. The burial was that of an adult male, whose heaped cremated bone were found claws from the bear skin in or with which his corpse had probably been burnt. Five Dr 1B amphorae and 36 other ceramic vessels were found in the tomb, as well as other a bronze dish and strainer, silver cup and wooden vessels silver, glass gaming pieces, beads and bracelet fragments, a nail cleaner, razor and other objects. The excavation of six more modest burials in its environs shows it to have been part of a larger cemetery of unknown extent.

same cemetery and are considered with other rural sites from Hertfordshire (6.2). Of the 95 burial groups recorded Rook considered 39 to be reliable, but did not specify which. Burial groups recovered during machining (south-west of the dashed line, Rook 1973, 2, Fig. II) are omitted from analysis as are those where the burial description indicates damage, leaving 80 for study. The discrepancy with Rook’s figure for reliable assemblages suggests caution in inference from these data. Dated samian suggests the use of this part of the cemetery from the Flavian to late Antonine periods, although the cemetery as a whole continued in use until the late Roman period. The lack of detailed information hinders systematic examination of cremation ritual. According to Wells’ analysis (in Rook 1973) the majority of individuals were efficiently cremated. In comparison to other sites small quantities of cremated bone were recovered, but this may well reflect salvage conditions. 25% of burials included burnt animal bone. Few species could be identified with certainty, but the range included pig, cattle, sheep and fowl. 83 % of (the eighty undamaged) burials were urned, usually in jars in local fabrics. Unurned burials may have been deposited in other containers, but only in one case (46) was this certainly a wooden box. Grave goods apart from ceramics are extremely scarce. The commonest represented ceramics are similar to those from Braughing ‘B’, as are the common burial assemblages (Figs. 4.23 and 4.24).

More work has been done on nearby villas than on the minor roadside centre which developed during the Roman period (McDonald forthcoming; Rook 1986: 110). One large Roman period cemetery has been investigated, across the river from the settlement area on the road to Braughing (Fig. 4.22). Rook (1973) suggests that it originally contained several thousand burials, but this estimate depends on the extent to which isolated observations derive from a single cemetery. The principal information derives from Rook’s 1967 salvage investigation. How typical it is of the settlement is unclear, although a poorly recorded boxed cremation burial from Welwyn Grange with a large assemblage of personal ornaments, ceramics and glass vessels suggests that its burials do not span the full range of variability. A nearby third century mausoleum and associated inhumation cemetery do not seem to have been part of the

Figure 4.22 Iron Age and Roman period burials from Welwyn in relation to the modern village (hatched) and findspots (circles) of Iron Age and Roman material (adapted from Rook 1986: 107, Fig. 28) (1 = gravel pit, cemetery?, 2-4 cremation and inhumation burials and teple mausoleum, the Grange; 5 pre-conquest and Roman period cremations, Mill Lane; 6 = LIA Welwyn burials)

56

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

Figure 4.23 Numbers of different ceramic forms, Welwyn

compounds, though direct spatial connections are not as clearly demonstrable as at other minor centres (cf. Burnham and Wacher 1990: 31). Burial practice became more homogenous than in the preceding period. Most individuals older than infants were cremated. Efficient cremation, colouring most of the cremated bone white, was typical. Minor differences in the proportion of cremated bone collected for burial are likely to be the product of different preservation and / or excavation (Fig. 4.6). The only recurring evidence for material placed on the pyre with the dead is of joints of meat. The varying proportion of burials with cremated animal bone between different sites may be partly due to preservation, but the higher percentage of burials with evidence of cremated animal bone at Braughing ‘B’ is not explicable by a difference in the quantity of cremated bone recovered. Most individuals across all sites were buried in a pot, commonly a locally made jar. Caskets and wooden boxes were only used in any number at Skeleton Green, .the percentage of such burials at the latter (20%) being unusually high.

60

50

40

30

20

10

0 Flagon

Beaker

Bowl

Jar

Dish

Cup

Vessel

Figure 4.24 The number of burials with different numbers of ceramic accessory vessels, Welwyn No. of burials

Grave good assemblages were generally modest and dominated by ceramics (Fig. 4.25 and 4.26). The sample considered above does not capture the full range of variation in this aspect of burial practice, since other cemeteries at Baldock show a much lower rate of vessel deposition than the example considered in detail above (Burleigh 1993). The majority of vessels were in local fabrics, with the principal exceptions of samian (Fig. 4.27) and colour-coated vessels which formed an increasing proportion of assemblages in later phases of the Baldock cemeteries. Flagons, beakers and then dishes were the most popular forms, though with some variations in their proportions. The presence of other artefact types is more difficult to compare meaningfully given their much lower frequency. They were documented in more than a few instances only from Skeleton Green (32%) and Baldock Walls Field (12%), at both glass vessels being better represented than at other cemeteries (Fig. 4.28). Others have already noted the difference in grave furnishing between these minor centre cemeteries (Burleigh 1993; Jones 1983), but the apparent homogeneity in earlier stages of ritual must be remembered in evaluating its significance.

25

20

15

10

5

0 0

1

2

3

4

5

4.6 Burial practice and small town society at Baldock, Braughing and Welwyn Varied burial rituals have therefore been documented at these sites in the Late Iron Age, including inhumation, cremation and deposition of human remains in other contexts. It is not clear what proportion of the population at these sites received formal burial within discrete cemetery areas, but the development of separate spaces for the dead certainly began in the late Iron Age, most clearly at Baldock. As in the Aylesford-Swarling tradition more generally, so at Baldock and Welwyn burial hierarchies marked by numbers and types of grave goods can be identified; as at King Harry Lane, though to a more marked degree, the earliest burials in local sequences were often the most spectacular, manifested in the ‘Welwyn’-type tombs of the early-mid first century BC. At Baldock the wealth of contextual information also allows us to note both a date close to the beginning of the occupational sequence and, at least in the case of the California cemetery, topographic prominence.

A few burials are exceptional in the quantity of items deposited or the unusual character of grave goods. Examples include the mass of glass vessels in grave 49 at Skeleton Green, the large grave good assemblages within Braughing ‘A’ and to the south and west of Baldock. The limited context information indicates that these were placed in smaller groups of burials rather than large communal cemeteries. Their date ranges from the midfirst to early third centuries AD. Where evidence allows some assessment there is little indication for the placing of such deposits in more spatially prominent areas. The modest character of the burial assemblages from the Royston Road site at Baldock, on the immediate periphery of the settlement and adjacent to the route running north-east from it suggests also that no particular prestige was attached to the street frontage.

The funerary topography of the Roman period shows a continuation of the trend to separate the living from the dead. Large communal cemeteries on settlement peripheries and radial roads are documented as well as smaller burial plots, perhaps connected to individual 57

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 4.26 The average number of ceramic accessory vessels in burials from urban and minor centre cemeteries

With the occasional exception of post holes, there is very little direct evidence for markers or monuments from any of these sites. Again lack of suitable building stone may mean that funerary monuments have been robbed out. Older monuments sometimes appear to have endured as significant landmarks. The ‘Tene’ burial at Baldock may have continued to be a focus of burials and ritual deposits, although the construction of a Roman corn-drier in the centre of a barrow (California) at Baldock shows that this was not always the case (see also 6.4.3).

2.5

2

1.5

1

As at Verulamium it is possible to assess burial rituals in relation to other archaeological evidence. None of these minor centres is marked by monumental public buildings on a scale equivalent to those of Verulamium, though at least two substantial stone buildings have been excavated at Braughing. The only known public building is a temple site recorded on aerial photographs west of the town. Several shrines are also known from Baldock, including a stone built example recently revealed by aerial photography in Baker’s Field (Burleigh and Fitzpatrick Matthews 2007: 411-12). Timber buildings, often set in individual enclosures, dominated housing with so far little evidence for characteristics such as mosaic floors or painted plaster documented in increasing quantities from the second century onwards at St Albans (Burnham and Wacher 1990: 108-10, 284-6). Items deposited in the tombs reflect those commonly in circulation, though burials emerge as privileged contexts for the deposition of some artefact types, for example samian pottery or glass vessels in some instances (cf. Willis 2004: 7.2.6). The precious metal votive leaves recently excavated at the nearby sanctuary of Senuna indicate alternative foci for the presentation and deposition of portable items of value, although this sanctuary may have a funerary dimension (Burnham et al. 2006: 411-13; Jackson and Burleigh 2007).

0.5

0 WLR

WLF

SKG

BRB

WLN

SST1

SST2

Figure 4.27 Samian as a percentage of ceramic accessory vessels in burials from urban and minor centre cemeteries 35 30 25 20 15 10 5 0 WLR

Figure 4.25 The percentage of burials with ceramic accessory vessels from urban and minor centre cemeteries (WLR Baldock Wallington Road; WLF Baldock Walls Field; SKG Braughing Skeleton Green; BRB Braughing ‘B’; SST1 St Albans St Stephens 1 (Davey); SST2 St Albans St Stephens 2 (HAT))

WLF

SKG

BRB

WLN

SST1

SST2

Figure 4.28 The percentage of burials with glass accessory vessels from urban and minor centre cemeteries 12

10

100 90

8

80 70

6

60

4

50 40

2 30 20

0 WLR

10

WLF

SKG

BRB

WLN

SST1

SST2

0 WLR

WLF

SKG

BRB

WLN

SST1

SST2

4.7 Conclusion In these neighbouring settlements the same changes in burial ritual took place during the first century BC and first century AD, i.e. a shift to majority, though by no

58

BURIAL TRADITIONS AND URBAN SOCIETY IN ST ALBANS…

means exclusively cremation burial from diverse forms of burial of the dead, including cremation, inhumation and the placing of body parts or bones in other depositional contexts. In general a hierarchy can be documented in which burials are differentiated by the number and type of items burnt or buried with the dead and sometimes by the monument with which the funerary activity is enclosed and marked, with the earliest burials often the most spectacular, as manifested by quantity and type of material burnt on the pyre or buried in the grave, or in associated structure. There is some variation in the phasing of this process, the Welwyn-type burials at Welwyn, Welwyn Garden City and Baldock being the earliest in local burial sequences, while the Folly Lane complex post-dated the more ‘modest’ earlier burials documented at King Harry Lane across the Ver by several decades.

assemblage was also lower in the St Albans cemeteries (Fig. 4.27), which accounts for the difference in forms represented, dishes in particular as well as cups being less common at St Stephen’s than in minor centre cemeteries. Willis (2004: 9.4) has noted this as a wider phenomenon distinguishing burials in minor centres from those to towns. This distinction is not however visible in the deposition of other grave goods, for example the percentage of burials with glass vessels (Fig. 4.28). Other items were rarely deposited, the most obvious difference being the higher proportion of burials with coins in the Verulamium cemeteries. There is however no evidence in this sample for consistent co-deposition of coins, lamps and glass unguent vessels as grave goods, which is sometimes taken to indicate the burials of individuals of extra-provincial origin or a shift towards a more Roman style of furnishing (Struck 1995; cf. 5.2.3). The general lack of inter-cutting indicates that most burials must have been marked, but only at St Stephens and Baldock does evidence survive for their form, which generally seems to have been modest timber structures, though in some cases groups of burials were collectively marked with enclosure ditches, albeit less often than in the pre-conquest period.

There are also common spatial trends in the manifestation of discrete cemetery areas from the early first century BC onwards, with a tendency for these spaces to be separated from those of the living by a combination of enclosures and physical distance, though they could be prominent features in their landscape setting. As well as the sites discussed above, at Welwyn and Welwyn Garden City known burials are widely scattered across the settlement complex, possibly at the former in association with elite enclosures (Bryant and Niblett 1997). There are obvious differences in the degree to which burials agglomerate in cemeteries: at present King Harry Lane is unique in its size, the multiple enclosure cemetery being equivalent to many of the individual enclosures documented elsewhere.

At only a small number of individual cemeteries does burial seem to continue significantly beyond the conquest (see especially at Baldock), but in general spatial differentiation between living and dead is a process already documented in the late Iron Age. In the Roman period the ‘concentric’ relationship of burial around settlement characterised all of the sites considered, both in cemeteries of significant extent containing hundreds if not thousands of burials and in small plots containing many fewer. These cemeteries continued in use until the early third century AD when some cemeteries began use. When published the extensive excavations of cemetery areas at Baldock will allow us to assess the periodicity of cemetery use with much greater precision. The creation of a planned urban centre, linked by roads to other cities, seems to be clearly connected to the restructuring of cemeteries at St Albans, although the chronology of this process remains to be further unclear. At the minor centres the differentiation between settlement and cemetery face is a little fuzzier in some instances, with adult burials within the settlement area being documented though rare. In two cases the sample considered allows us to examine the relationship between road frontage and burial directly. Funerary monuments were documented intermittently along the road frontage at St Stephens, whereas no evidence of markers was recovered at Royston Road Baldock, but the St Stephens cemetery was not a monumentalised Gräberstrasse. There was only very limited indirect evidence for burials marked with monuments away from these sites. Within this sample the Folly Lane complex, which endured in use as sanctuary into the third century AD., remains unique in scale as well as in form.

The conquest does not mark an abrupt change in burial rituals at any of the cemeteries examined. Cremation remained the dominant ritual for those afforded an archaeologically visible form of burial, with inhumation sporadically documented until it became common from the third century AD. Burials are regularly distributed in small clusters within zones on urban margins, sometimes mainly devoted to burial, sometimes with other closely associated activity. However the choice of burial areas was determined, the most striking aspect is the limited differentiation in the rituals of Roman burial practice, with a small number of exceptions. The condition of the cremated human bone shows efficient cremation in the vast majority of cases. There is little evidence for placing materials other than joints of meat with the dead on the pyre. Burial of the cremated bone in a ceramic jar was everywhere the commonest form of burial, with other types of container represented in occasional instances. Ceramics in local fabrics were the most frequently deposited grave goods, with other items only occasionally being represented. Urban cemeteries differ from those of minor centres in the number and type of ceramics buried. The percentage of burials with ceramics from St Albans was lower than at all of the minor centre cemeteries studied in detail and the average number of ceramics per burial was lower (Figs. 4.25 and 26), though at cemeteries from Baldock not analysed in detail here furnishing with ceramics was more limited (Burleigh 1993). The proportion of samian in the grave good

The patterns identified so far offers limited support to the model of greater ‘investment’ in burial in minor centre cemeteries as opposed to the major towns (1.3). The form rituals took in the early Roman period and the degree to 59

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

which burials were differentiated is similar in the cemeteries of St Albans and the minor centres. The evidence considered here does not indicate a differential adoption of innovations in burial rituals between these settlement types. There is a consistent difference in the scale and composition of ceramic assemblages studied between St Albans and the small towns, though in other aspects of ritual the similarities are more apparent tan the differences. It must be emphasised too that only one Roman period sample from St Albans is available for detailed study and the Baldock cemeteries remind us of the significant variability in practice between the cemeteries of minor centres. At all the settlements studied, a small number of burials were identified which were distinguished by the quantity, type and diversity of grave goods, though with the exception of Folly Lane, only very limited evidence for tomb monuments has so far been detected. There is therefore limited evidence that a more hierarchical social structure is revealed at Roman Verulamium than at minor centres elsewhere in the civitas of the Catuvellauni. Worth noting however is the variation in the relative significance of burial compared to other spheres. In Verulamium the evidence for alternative forms or display has been indicated, in particular public architecture and, by the later second century AD, larger and more elaborately decorated private houses. These however are less significant in the minor centres and burial takes on a correspondingly greater significance. This assessment of urban and minor centre cemeteries provides a basis for the regional distribution of display to be examined in chapter 7, along with evidence from the rural hinterlands of these sites (6.2). In the next chapter however the second of the urban samples is examined.

60

CHAPTER 5: BURIAL TRADITIONS AND URBAN SOCIETY IN WINCHESTER (VENTA BELGARUM) burials are not available from neighbouring minor centres, this survey together with the results of the examination of rural burials from Hampshire (6.4) provides a basis for defining a regional threshold of burial display and exploring the distribution of burials which exceed it (7.5). It also provides a further sample of burials for exploring the character of burial display based on its artefactual composition (8.2-8.4)

5.1 Introduction This chapter explores a second sample of urban burials from the perspective applied in the previous chapter. Extensive excavations from the late 1960s have given Winchester (Venta Belgarum) one of the most abundant sets of Roman urban funerary data in Britain, although the different cemetery areas are very unequally represented in recent excavation (Esmonde Cleary 1987: 153-6; Kjølbye-Biddle 1992; Ottaway et al. 2012). In contrast to the sample discussed in the previous chapter, the evidence available is concentrated in the late Roman period, but there is a sufficient early Roman sample to make study of change through the Roman period possible. Again cemeteries were selected for more detailed analysis where 50 or more burials with adequately recorded assemblages were available, but important individual or smaller groups of burials have also been examined. The distinction between ‘urban’ and ‘rural’ burial is also not straightforward to draw; the distance from the city’s cemeteries of one-two kilometres or more of the richly furnished individual and small groups of burials in the city’s hinterland of Millett’s (1987) ‘east Hampshire Tradition’ suggests a location in the burial plots of extra-mural farms or villas (6.3; 7.3.2). As well as burial rituals the relationship of burial and monument to their wider context of settlement and roads at Winchester is considered. Since large numbers of

Although there is a relatively abundant corpus of burial data from Iron Age Hampshire (6.3.1), little is known of the burial rites of the Iron Age settlements on St Catherine’s Hill or Oram’s Arbour that preceded Roman Venta Belgarum (Biddle 1983; Qualmann et al. 2004). The corpus of Roman period burials is much richer. Over 1200 burials have been excavated from the northern cemeteries of the town, in particular associated with the roads to Silchester and Cirencester. Smaller samples are available from the cemeteries to the east and west of the city (Fig. 5.1). Most of the city’s Roman cemeteries excavated since the 1960s have been published (KjølbyeBiddle 1992; Esmonde Cleary 1987: 150-56; Booth et al. 2010: 6-7; Ottaway et al. 2012). Within the Roman period city a small number of infant burials has also been recorded (e.g. Cunliffe 1964: 43; Kjølbye-Biddle 1992: 211; Zant 1993: 33-4; Ford et al. 2011: 363).

Figure 5.1 The cemeteries (hatched areas) of Roman Winchester (adapted from WMS Archive)

61

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 5.2 The northern cemetery area at Winchester (adapted from WMS Archive). Large stippling = cemetery areas. Hatching = areas without burial evidence

The only early Roman cemetery so far to be excavated on a substantial scale (c. 185 burials), Victoria Road, lies c. 50m to the north of the city’s north gate (Fig. 5.2). The extent of this cemetery is unknown; the limited scale of recent excavation on the Evans Halshaw garage site (Hyde St) makes it impossible to assess whether the two mid 1st century AD cremation burials excavated here are part of the same burial area extending towards the Silchester road or isolated burials. There are few other reliably excavated burial assemblages contemporary with this cemetery, a handful of late Iron Age or early Roman date being noted by Ward-Evans (in Hawkes and Dunning 1930: 183) but not described in any detail. 19th century references to cremations from Water Lane suggest the existence of early Roman cemeteries east of the city (Ottaway et al. 2012:317-9). First century cremation burials with large grave deposits at Nun’s Walk, Highcliffe and Winnall and Grange Road are all located a significant distance from the settlement area (a kilometre or more) and do not seem to be part of the urban cemeteries. Their distribution and character are considered within discussion of the East Hampshire Tradition (6.3.2, 7.5 and 8.2-8.4).

North of the Roman town extensive excavation of fourth century burials has taken place further from the north gate at Andover Road - Eagle Hotel (38 burials) and Victoria Road West / Hyde Street (c. 190 burials) and along the Cirencester road; n particular more than 800 burials have been excavated from the Lankhills cemetery. Again the limited extent of excavation makes it impossible to assess whether the two third or early fourth century AD cremation burials excavated on the Evans Halshaw garage site indicate the cemetery’s extending towards the Silchester road. To the east of the city excavations at Chester Road and St Martin’s Close, as well as other small-scale work, has produced a corpus of almost 200 burials, in an area where little evidence has otherwise been recovered for extra-mural activity (Rees et al. 2008: 17). To the west the burials excavated from the Oram’s Arbour ditch are primarily those of infants, although there is some evidence for a more extensive cemetery along the road to Old Sarum (Ottaway et al. 2012:133-73). Little excavation has taken place in the southern cemeteries. Changes in rituals within the late Roman period in these cemeteries are discussed, but the focus of analysis lies on characterising the period of the very late third to early

62

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

and ditch were documented in the excavation) and, initially, to the east by an Iron Age and early Roman parallel ditch and track, worn to form a hollow way, c. 20m west of the later Silchester road and also leading to the entrance to the Oram’s Arbour enclosure. The base and fill of the Iron Age ditch were cut by phase 1 inhumation and later cremation burials and by the end of the first century it was probably no longer a significant topographic feature. In the late first century the first of three north-west south-east ditches on the north-eastern part of the site cut across the line of this track. At its eastern end the outermost and latest of these turned southwards, perhaps to run parallel to the Silchester Road. In its later phase the cemetery therefore comprised a strip roughly 20m wide running parallel to the Cirencester Road. An absence of burials east of the late Iron Age and early Roman track suggest that it marked the eastern limit of the cemetery, but the whole of the cemetery area was heavily disturbed by later Roman and subsequent occupation and activity. The distribution of the earliest burials suggests that all excavated areas of the cemetery were used from the beginning.

fifth centuries as a whole. The chronology of individual cemeteries is discussed in appendix 4.1. 5.2 Early Roman cemeteries at Winchester 5.2.1 Burial practice at the Victoria Road East cemetery The excavated portion of this cemetery was in use between the third quarter of the first and the late second or early third centuries AD (Fig. 5.3a), after which the area was occupied by ephemeral buildings (Ottaway et al. 2012: 33-96). The majority of burials have been attributed to two main phases, of later first to mid-second date, with a small group dated to the mid-second to early third centuries. The use of the cemetery was most intensive in the first phase. The southern edge of the excavated area lay c.20m north of the junction of the Silchester and Cirencester roads, outside what had been the entrance to the pre-Roman Oram’s Arbour earthwork enclosure and the probable site of the Roman city’s north gate. The cemetery was bounded to its west by the Cirencester road (the edge of the road, a roadside path

Figure 5.3a The Victoria Road East cemetery from the first to early third centuries (adapted from Ottaway 1992: 77, Fig. 3.9 and WMS archive)

63

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Fig. 5.3b Detail of Victoria Road East cemetery, mausoleum and four post monuments in Trench X (adapted from Ottaway et al. 2012: 61, Fig. 24)

Inhumation (66) and cremation (67) burials were roughly equal in number in the first phase but the latter increased in number over time, accounting for 32 of the 46 phase 2 burials). Infant inhumations, either singly or in small clusters and cremation burials were distributed across the site, but adult inhumations concentrated in the northeastern part of the site close to and within the late Iron Age ditch where they were the earliest burials in the sequence.

the legs of two had possibly been tied before burial. A further inhumation (480) comprised the disarticulated bones of an adult male and female, over which the sacrum, pelvis and femur of a horse had been placed. Over 60% of the inhumations were placed with their heads between east and south, though adult burials were more often placed with the head to north or south. Orientation is partly conditioned by the north-south alignment of the Iron Age ditch with which early examples were associated. With two exceptions, there was no evidence for inhumations having been buried in coffins and grave goods were scarce. A miniature iron shovel was recovered above an infant inhumation which lay over the upper body of a decapitated adult. Two infant inhumations were furnished with coins, one of these also with a vessel.

There was a clear division in the burial ritual practised at Hyde Street according to age. Most inhumation burials were of infants under two years of age (Fig. 5.4), generally aged between birth and three months. The proportion of these within the inhumation sample increased from 75% in phase 1 to over 90% in phase 2. Since infants of any period have only been recovered in small numbers from Winchester’s intra-mural area (see above) it seems likely that many were also buried outside the town. Whether this area was preferentially used for the deposition of infant burials because of its position near the gates is difficult to assess because of the lack of excavated contemporary cemeteries. Certainly the proportion of infant burials within this cemetery population is higher than the proportion to die in the first year at the extremes of viability of pre-industrial populations (cf. Hopkins 1966). In the later Roman period infant burials are similarly dominant among those buried in the silted-up Iron Age enclosure ditch west of the town (5.3.4).

101 cremation burials were recorded but no pyre-related features were identified in the excavated area. The uniformly white or light grey colouring of the cremated bone suggests efficient cremation in most instances. Charcoal analysed from the Evans Halshaw garage cremations primarily comprised oak. The principal exception is an uncremated skull and an articulated vertebral column and ribs on a bed of charcoal and cremated bone in grave 489. Lack of soil discoloration suggests that cremation may not have taken place in this grave (3.4). Rather than inefficient burning, this may be a further example of seemingly deliberate semi-cremation occasionally noted elsewhere, for example at Baldock Royston Road, Derby Racecourse, Guilden Morden and Suddern Farm (Danebury environs). The amount of cremated bone excavated from undisturbed burials was similar to other sites in southern England (Fig. 4.6). The very small amount of skull bone in two graves (550 and 560) suggests that bone from certain parts of the body may have been kept from burial.

The shallow depth of most grave cuts and location in an area of intensive burial subjected inhumations to extensive damage. Infant burials were mostly deposited in the foetal position and adult burials extended and supine. As well as one crouched burial, other adult inhumations were in unusual positions; three adult females were deposited prone, one was decapitated, and

64

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

Figure 5.4 Cremated (white) and inhumed (grey) populations, Victoria Road East

Figure 5.7 The number of burials with different artefact types, Victoria Road East

80

60

70

50

60

40

50

30

40

20

30

10

20

0 HN

10 0 Adult

Juv.

Vessel Coin

Pers. orn.

Glass

An. Equip. bone

Fragmentary burnt artefacts were deposited in twelve graves, including ceramics, lamps, glass, bone inlay, perhaps from caskets or biers (cf. 3.5) and burnt nails, probably from wooden objects placed on the pyre. Lamps were only recovered among the pyre goods and most glass vessels from the cemetery had also been burnt. The largest assemblage of pyre goods included eight burnt samian vessels, other burnt ceramics, molten glass and burnt nails (438). Three burials were Brandgrubengräber and three Brandschüttungsgräber (3.4). The latter were located in close association in the western part of the cemetery on the road frontage.

Inf.

Figure 5.5 Numbers of different ceramic forms, Victoria Road East 30

25

20

15

Graves were all earth-cut pits, usually the minimum size to receive the deposit. Approximately 60% of burials were urned in both phases, grey ware jars in local or regional fabrics being the most common container (83%). One third of graves included ceramics as grave goods (Fig. 5.5). Local and regional fabrics dominated the ceramic assemblage, samian accounting for most of the non-local fabrics (16%), but ten of the 18 samian vessels were recovered from a single grave (566) which was also the most richly furnished in terms of the total number of ceramics. Some ceramics were damaged before burial, for examples by holes pierced in cremation containers (524, 536, 558). As in other cemeteries some burial vessels were kiln wasters or of a soft and crumbly fabric, perhaps destined for burial only (cf. Philpott 1991: 36; TuffreauLibre 2000). An approximately even number of graves contained flagons, beakers, dishes/platters and cups, but dishes and cups were the most popular forms in the overall accessory vessel assemblage because of their high representation in a small number of graves. In the very small number of graves with four or more accessory vessels there is in often an equal number of each, supplemented by a flagon (Fig. 5.6). No consistent layout was observed in the placing of vessels or other objects within the grave.

10

5

0 Flagon

Beaker

Bow l

Jar

Dish

Cup

Lid

Vessel

Figure 5.6 The number of burials with different numbers of ceramics, Victoria Road East 80 70 60 50 40 30 20 10 0 0

1

2

3

4

5

6-10

11-

Almost half the graves contained some other type of artefact, almost all nails, some of which probably derived from wooden objects, including boxes, with a structural or decorative role, and others were probably residual 65

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

local Iron Age ritual traditions (6.3.2). The Brandschüttungsgräber and burials in their immediate vicinity were characterised by the presence of pyre goods rather than grave goods and by the burning of objects on the pyre more than their deposition in the grave, including artefact types that were otherwise rare in the cemetery, lamps, unguent bottles and coins and imported finewares. Lamps, unguent bottles and coins are a consistently identified combination in contemporary burials in central Italy and the Rhineland and are sometimes taken to indicate the burials of individuals of extra-provincial origin or a shift towards a more Roman style of furnishing (Fasold and Witteyer 1998). These burials may therefore derive from a non-local and perhaps non-insular tradition. Ottaway (1992: 79-80) suggested that these graves might be those of Roman soldiers. Although the evidence for an early fort at Winchester (Wacher 1995: 291: Biddle 1975: 296-7, 1983) is not yet compelling, we might anticipate the burials of soldiers from 1st and 2nd century Winchester on other grounds; the city’s only complete inscription reveals the presence of a beneficiarius in the city in the late first or second century (RIB 88) and small finds evidence may suggest the presence of soldiers in small numbers (Cool, in Ford et al. 2011: 306). Combined with the evidence for a greater range in ceramic furnishing, the diversity of burial practice may suggest that in the cemetery’s first phase there was a greater marking of identity through differing rituals and a more competitive use of burial furnishing as a means of display, though in other aspects of the ritual (e.g. cremation) there is limited differentiation. In its second phase (from the later first century AD) burial ritual became more homogeneous. Inhumation was reserved almost exclusively to infants, urned deposition without pyre debris predominated among the cremations and the number of ceramic grave goods declined, though the grave with the largest collection of non-ceramic grave goods (466) and mausoleum F272 date to this second phase.

finds (Fig. 5.7). In a few cases such boxes are likely to have held the cremated bone. Hobnails were occasionally recorded. Other non-ceramic artefact types included personal ornaments, especially brooches, bracelets and beads, toilet articles including four mirrors and gaming pieces. Many of these objects associated with care of the body come from a single richly furnished second century AD burial (466) which included ivory and amber personal ornaments. Not all the instances of animal bone in the grave fill represent deliberate deposits, but animal parts had been buried with the dead in at least eight cases. The main change over time in burial furnishing concerns ceramics. While approximately a third of burials contained ceramics in both phases, the average number of vessels per grave declined from 1.23 to 0.61. All bar one of the deposits of more than five vessels occurred in phase 1. The form preference however showed little change. The proportion of burials with non-ceramic artefacts however remained constant; the largest such assemblage with the widest variety of non-ceramic artefacts, 466, dates to phase 2. The lack of intercutting suggests that most burials were marked but direct evidence survived only occasionally. A small number of graves were marked by timber post-built structures including burial 600 on the north-south ditch and a small cluster of three immediately south-west of the mausoleum adjacent to the road (Fig. 5.3b). Tiles from the fill of burial 466 may perhaps have derived from a structure above it of which no trace otherwise survived. One certain masonry monument (F272) lay close to the Cirencester road, 3.2m square, the surviving walling being of mortared flint and tile, perhaps with a clay floor within. A single unurned and unaccompanied cremation burial could be associated with the primary phase of the mausoleum. As for other ritual activity possibly associated with burial, deposits of single animals and artefacts were documented within the cemetery, some perhaps deposited in relation to individual graves (for example a flagon deposited adjacent to the well-furnished burial 566), others seemingly isolated on the eastern and southern margins of the cemetery. The large non-grave good ceramic assemblage recovered during excavation may relate to funerary activity but various formation processes, including contemporary occupation in the cemetery environs and rubbish dumping during and after its use, when suburban settlement extended into the cemetery site, are likely to have contributed to its accumulation.

The Victoria Road cemetery extended along the frontage of the Cirencester road, although understanding its development is complicated by the existence of the Iron Age hollow way to the east of the site. The earliest burials did not show a preference for placing by the roadside. Clearer constraints on the cemetery organisation emerged by the second century, by which time ditches to the north-east of the site and the road to the south-west defined a strip for burial aligned on the latter. Neither larger ceramic nor non-ceramic assemblages were deposited in a roadside position; here the isolated position on the eastern edge of the cemetery of one richlyfurnished burial, 466, with evidence for a possible marker, is notable. The only attribute shared between burials located close to the road is the presence of pyre goods, in three cases with Brandschüttungsgräber. Only from the late first century does the cemetery assume some aspects of a monumentalised Gräberstraße, but these are unspectacular, comprising one certain masonry-built tomb and a handful of adjacent graves marked by posts. Lack of evidence from the city’s other early Roman cemeteries makes it difficult to develop this discussion

5.2.2 Burial and society in early Roman Winchester The burial ritual documented in the Victoria Road cemetery shows its greatest diversity in its first phase. This diversity is documented in both burial ritual (inhumation and a variety of cremation burials) and in the varying quantities of grave goods, especially ceramics. This diversity suggests diverse burial traditions among those using the cemetery. The inhumation tradition, including the close association in one case between the burial of human and animal remains, may be connected to 66

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

in the city, in particular the extensive drainage of the lower valley slopes and bottom, the setting out of the street grid, and construction of public buildings, including a forum by the end of the first century AD, as well as the earthwork circuit of which the first phase dates to the Flavian period (Qualmann 1993: 75-6; Wacher 1995: 291-4), i.e. contemporary with the initial use of the cemetery. Such buildings are likely to have been be paid for by local aristocrats, but like other towns Roman Winchester, admittedly excavated on a more limited scale than many, lacks substantial evidence of monumental inscriptions to celebrate the erection of these buildings (Blagg 1990). Nonetheless a fragment from the Brooks site (RIB 3043) bearing the largest lettering yet found on a stone inscription in Britain, suggests the scale both of some buildings and of the advertisement of their construction.

further; the limited evidence elsewhere includes an observation of a structure over a cremation burial associated with a substantial extra-mural building to the west of the town, albeit with very limited documentation (Haverfield 1900: 287). The absence of inscribed tombstones suggests that funerary monuments were uncommon, though as at St Albans the lack of good local building stone means that any erected will have been prone to later re-use. The location of the Victoria Road cemetery is of wider interest as it is the only Roman example so far excavated in Britain to be located close to the likely pomerium, its southern tip being only c. 50m from the hypothesised later north gate of the city, and a very short distance from the junction north of it where the Cirencester and Silchester roads met, created at the same time as other early elements of the urban plan at Winchester (see below). In other Roman cities in continental Europe such a location is favoured for burial display, typically where spaces given for burial as a civic honour by the ordo would be placed (7.2.2). However although the first phase of the cemetery saw some differentiation in burial ritual, no consistent placing of the richest burials on the street frontage could be identified and there was in any case very limited evidence for monument construction. Neither the limited documentation of older finds and the better documented small interventions (e.g. Evans Halshaw garage) elsewhere among the city’s early Roman burials suggest that significant variability within the city’s early Roman cemeteries has been omitted. Comparison to rural burials suggests that competitive funerary display in the urban cemeteries proper was limited, since the cemetery includes only two burials from which the number of accessory vessels places them above the threshold defining the ‘East Hampshire Tradition’ and they are among the smallest assemblages within this grouping (6.3.2).

5.3 Late Roman burial practice at Winchester 5.3.1 The Victoria Road West, Andover Road (Eagle Hotel) and Hyde Street (Late) cemeteries in the northern suburb Further burial areas of late Roman date lie west and north of the early Roman cemetery discussed in the previous section, of which the most important site is Victoria Road West (trenches IV and V). Burials here are dated to between the late third and early fifth centuries, following a phase of earlier occupation by a roadside path, ditches and buildings to the north-east of feature 12. A few burials in the secondary silt layers of the road ditch (parallel to and north-east of F12), damaged by later ditch digging, may date to an earlier period (Fig. 5.8). The earliest burials from the principal use of the cemetery were interred immediately south-west of the roadside ditch and parallel to the Cirencester road in the late third century AD. In later phases inhumation burials were laid further to the west and oriented west-east at an oblique angle to the road. The division of the cemetery’s use into three principal phases was established from stratigraphic relationships and burial alignment and an absolute chronology for these phases from the few datable grave goods and comparison with the development of burial practice at Lankhills. Only a small number of burials were dated to the first phase (12) (c. AD 270-320), more to the second (62) (mid 4th century) and third phases (43) (late 4th-early 5th centuries).

Comparison with the contemporary settlement helps further to assess the significance of differentiation among the burials. Very few substantial ceramic and finds assemblages have yet been published from the city, and those from the suburbs are generally not contemporary with those from this cemetery. Albeit unevenly distributed, samian was used in burial rituals in a similar proportion to its general occurrence in urban settlement assemblages, accounting for c. 10-20% of vessels (EVEs) (Willis 2011: 183-84). In terms of small finds, as at St Albans, with very few exceptions (e.g. burial 466) grave furnishing at Victoria Road comprised objects which were widely available (e.g. Ford et al. 2011: 237; Rees et al. 2008: 390). Within the city, contemporary private housing takes a form typical of early Roman cities in Britain, i.e. of modest timber strip buildings. The sequences from The Brooks (Zant 1993) and Wolvesey Palace sites (Biddle 1975: 321-5) begin in the Flavian period with closely spaced timber built strip buildings (see also Massey 2006: 25), within which it is difficult to see significant differentiation. Much more significant evidence for the articulation of status difference in a material form comes from the public works and buildings

Small numbers of cremation burials were recovered from all phases of use. They were usually urned and deposited in small shallow cuts, save perhaps for feature 95 for which the cut was very deep, well formed and much larger than of an inhumation burial, though its status as a burial rather than associated deposit is not certain (Ottaway et al. 2012; 113). The majority of burials were inhumed. Most graves were dug with care and grave sides were vertical or close to vertical and grave cuts were rectangular in shape, occasionally tapering at the east end. Two graves, both adjacent to and parallel with the road, were larger than average and had been stepped at 1m depth, from which a normal size grave had been dug. 67

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

for the setting of a large post, may indicate a possible ritual connection. On the other side of the road were four approximately contemporary similar features, in one of which several dog carcasses and near-complete ceramic beakers had been deposited (Ottaway et al. 2012: 96, 107-9). Much more directly linked to the burials were two other large-inhumation like pits, G109 and G127, also contained substantial numbers of complete vessels adjacent to graves 107 and 108 respectively (Ottaway et al. 2012: 113). These

All the shallow and/or irregular-shaped graves dated to the final phase. Corpses were laid out supine, with legs extended and arms either at the sides of the body, slightly flexed to lie on the pelvis, or folded across the trunk. Positions in the final phase showed more variety and included all the burials in which bodies had been placed prone or on their side. The presence of coffins was established from the number and distribution of nails, and more occasionally from wood stains. A coffin was certainly used in 32% of burials and possibly in a further 26%. The proportion of coffined burials varied over time; while there was possible or certain evidence for a coffin in all the 9 phase 1 burials and over 80% of the 62 dated to phase 2, in phase 3 (35 burials) none of the bodies were certainly placed in coffins, although damage may sometimes account for the lack of evidence. Numbers of nails suggest that some coffins must have been very elaborate, especially in phase 1, for which the average number of nails per coffin (58) is much higher than in phase 2 (14). Graves had occasionally also been packed with stone.

At the Andover Road Eagle Hotel, south-west of the Victoria Road site and further west from the Cirencester road (Fig. 5.2), similar rituals were documented in a smaller excavation area. Other than a short stretch of north-south aligned ditch in the north-west corner of the excavation areas no other boundary features were excavated. Stratigraphically the earliest burial on site was a single north-south orientated adult male inhumation in a lead-lined timber coffin in a very deep grave shaft (>2.2m), with a coin of Constantine close to his right hand. The remaining 37 excavated burials of fourth century date were aligned east-west, with heads to the west. With the exception of five buried prone (one of which had been decapitated) and four on their sides the rest were buried extended and supine. In at least 14 out of 32 cases evidence for a timber coffin survived; some burials had also been packed with stone. A very small number were furnished three with a coin, one with a bone comb and up to five with hobnailed footwear, though this was certainly the case in only one instance.

One third of burials were accompanied by grave goods (Fig. 5.10). As the ‘equipment’ category comprises mainly iron and bronze fragments of unknown function, some of which were probably residual, it probably exaggerates the proportion of furnished burials. The proportion of furnished burials declined over time from over 60% in phase 1 to 30% in later phases. The phase 1 burials, both inhumations and cremations, were distinguished from later burials by the frequency of large ceramic deposits, elaborate coffins made with hundreds of nails and stepped grave construction. A deposit (109) of five vessels and other artefacts in a wooden box placed at the end of phase 1 burial 108 may be interpreted as an associated offering (cf. 3.4). Save for the more frequent recovery of grave goods from coffined burials in phase 2 (69%) than for the phase 2 average, further associations between grave structure, burial container and grave goods were not identified.

54 graves, perhaps dated to the second half of the fourth century were noted at Hyde Street to the north-east of the Victoria Road cemetery (Fig. 5.2) but many were only observed in builders’ trenches. 30 were partially or wholly excavated, most in a single open area excavation. With the exception of a single cremation burial cut into the backfill of an earlier inhumation, corpses were normally laid out in the same positions as at Victoria Road but four had been deposited prone and one decapitated. Only 10% of fully excavated burials had any evidence for a coffin and 20% were furnished, three with hobnailed footwear, two with coins and one with a double-sided bone comb. Two thirds of burials in the excavated and observed sample were oriented west-east like other sites.

The predominant orientation of burials was with head to the west but there was significant change over time (Fig. 5.8). The predominant north-south and north-west southeast orientation of the few phase 1 burials was parallel to the road alignment. West-east was the dominant orientation in phase 2 (over 95%) and remained the most frequent orientation in the final phase, though it characterised only a minority of burials.

Late Roman burial practice in other parts of the northern cemetery excluding Lankhills was similar to that described in preceding paragraphs. For example at the Cattle Market site up to 60 south-west north-east oriented inhumations were excavated in the 1930s. Collis (1978: 142-49) also reports small groups of similar east-west oriented inhumations, usually unaccompanied. At the Evans Halshaw garage two urned cremation burials associated with a ditch to the rear of structures bordering the Silchester road were of similarly modesty to those in other cemeteries (Fig. 5.2); the burnt bone pins associated with the cremated bone in 7071 suggest the dressing of the corpse for the pyre.

Associated with this period of burial were several deep pits of which the fills had elements reminiscent of structured deposits. Three substantial pits several metres deep (F18, 43 and 46) were excavated in the ditch which ran parallel to the Cirencester Road. Above the lower fills, characterised by abundant amphibian bones, the fills of these in general seem to be related to rubbish disposal, in particular related to butchery and metal-working, but certain idiosyncrasies, for example the likely use of F46

68

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

Figure 5.8 The Victoria Road West cemetery from the third to early fifth centuries (inhumation burials grey, cremations black) (adapted from WMS archive)

Figure 5.9 The Lankhills cemetery, Clarke and Oxford Archaeology excavations (adapted from Booth et al. 2010: 43, Fig. 2.30)

69

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

5.3.2 The Lankhills cemetery in the northern suburb To the north of the sites discussed above the Cirencester road follows the rising ground. At c.500 m from the north gate site is the Lankhills cemetery excavated in two major campaigns, 451 burials by Clarke from 1967-72 and a further 332 by Oxford Archaeology, north of the area excavated by Clarke (Fig. 5.1-2, 8). The quantified observations are based on the burials published by Clarke, with additional reference being made to the Oxford Archaeology sample where analysis allows expansion on or qualification of Clarke’s results (Appendix 4.1). The use of the Lankhills site for burial dated from very early in the fourth century AD to the early fifth. The cemetery was probably bounded to its west by the Cirencester Road and to the east initially by feature 12/ 450, a substantial north-south oriented ditch (1318) to the east, and earlier features on a similar alignment. To the north the Oxford Archaeology (OA) excavations revealed a ditch which was not breached by burials and marked a likely northern boundary. The area used for burial expanded from west to east during the fourth century, although the western area was still in intermittent use in the latest period with infilling of areas left vacant by earlier interments. The precise dating of different areas of the cemetery may be debatable but the broad outline of the spread of burial from west to east during the course of the fourth century is convincing (Fig. 5.9). The area east of F12 (Clarke’s area O) was the latest in use. Both cremation and inhumation were practised as burial rituals for the duration of the cemetery’s use. Six cremations were recovered in Clarke’s excavations as well as a further three in earlier salvage. The context of two unurned cremations in a layer of topsoil in the better preserved burial sequence in feature 12 may suggest that similar burials may have been lost, though the OA excavations documented very little stray cremated human bone (Booth et al. 2010: 500). Two urned cremations were recovered and an inhumation-sized Brandgrubengrab (60) and bustum (359). The dating of four cremations to the second half of the century demonstrates the persistence of the rite as a minority practice. In the 25 cremations in the OA sample, some of which clustered in the north-western part of the cemetery, forms of cremation burial were similarly varied; evidence for pyre and grave furniture included faunal remains, ceramics, and dress items, including one instance of a crossbow brooch and two of belt fittings (3.4). Several of these cremations showed evidence for incomplete combustion of organic remains (bone colour variability), in some cases linked to a diversity of woods among the pyre debris, a potential sign of difficulty of accessing fuel resources (Challinor in Booth et al. 2010:441-2). The predominant burial mode was inhumation. Most grave pits were rectangular in plan and in cross-section, containing burials in similar positions to those documented at Victoria Road. More were buried on their side, placed prone or decapitated in Clarke’s area O than 70

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

forms of personal ornament, equipment for daily life such as spindle whorls, knives and vessels in other materials (Booth et al. 2010: 506-7).

elsewhere, suggesting that this was becoming more common in the latest phase of use. Exceptions to the normal grave form include 14 ‘step graves’, as at Victoria Road large pits stepped in at a depth of c. 0.5m with normal size graves cut in their base. Wood fragments occasionally suggested that the ‘steps’ had supported a wooden platform over the grave. In two instances vessels may have been placed on these platforms. Most graves were carefully shaped though the proportion of these declined over time. Of the seven irregular shaped graves six dated to the latest phases. Gullies interpreted as bedding trenches for hedges delineated rectangular enclosures around four graves. The profiles of better preserved graves in feature 12 suggest that most burials were marked by mounds. Four further stepped graves and one enclosure were documented by Oxford Archaeology (Booth et al. 2010, 35-36 39).

The grave goods were the focus of Clarke’s analysis, and subsequent discussion of the cemetery has focused on his identification of ‘intrusive’ groups of burials. From the presence of particular artefact types, especially personal ornament, buried with men (in particular crossbow brooches and belt fittings taken to be markers of military or bureaucratic status) and women and from their being worn rather than simply deposited with the dead, he identified two such groups. To one he ascribed an origin in Hungary, either within or beyond the Danube, the other to northern Germany, these groups being settled in the late fourth or early fifth century as soldiers or officials in Roman service, a theme on which others have elaborated, as summarised by Booth et al. (2010: 509-10). The latter group has commanded little acceptance and the empirical basis for the identification of the former has been contested. Baldwin (1985) questioned its degree of internal consistency and like Philpott (1991) emphasized the diversity of late Romano-British burial practice against which Lankhills must be compared, better characterised than at the time of Clarke’s survey (see also Booth et al. 2010: 508-9). Provision of worn personal ornaments, for example, is now documented elsewhere, ain London’s eastern cemetery (Philpott 1991: 142-4). It can in any case be difficult to establish, especially for the crossbow brooches and belt sets, whether they were worn or unworn (see e.g. Esmonde Cleary 1983; Cool, in Booth et al. 2010: 283-4). Others have argued with varying emphasis that some artefacts claimed as nonBritish in origin (Cool 2010: 36-41; Swift 2000: 69-77) and the general character of grave ritual (Cooke 1998: passim) support Clarke’s argument for external origins, especially in late Roman Pannonia, for some of the individuals buried at Lankhills. Crucially for this argument the analysis of strontium and oxygen isotopes from skeletal remains has shown there to be no association between individuals identified from their burial rituals as ‘local’ and ‘intrusive’ and their geographical origins as inferred from the isotopic signatures, though this need not of course preclude a presentation or perception of personal or group origins elsewhere (Evans et al. 2006; Eckardt et al. 2009b; 2010) (see also 1.3.1).

The predominantly west-east orientation of graves does not apply to the whole cemetery. Burials with heads to the south-west and west-south-west were concentrated in the western part of the site (Fig. 5.9), perhaps aligned at right angles to the Cirencester Road, projected to run 2030m to the west of the site. Burials with a predominantly western alignment in the centre and eastern part of the site were generally at right angles to feature 12, while on the eastern margins of the areas excavated by Clarke (area O) and Oxford Archaeology burials were characterised by a much greater diversity of orientation. 83% of burials were coffined although the proportion was less than half in the mostly later burials in area O. The presence of brackets indicated more elaborate timber coffins in four cases. 8.4% of graves were lined with flint and tile; the frequency of stone packing in area E suggests that it was more frequent in the final third of the fourth century. A similarly high proportion of coffined burials was noted for the OA burials. Analysis of surviving traces of wood showed only oak to have been used; again a small number were distinguished by more elaborate construction, as revealed by higher numbers of nails or their much larger than average size (Booth et al. 2010: 323-30, 481-3). Consideration of the various forms of evidence for containment of the body within a shroud revealed no clear pattern (Booth et al. 2010: 475-6). 54% of all graves and 63.2% of intact graves contained grave goods of a wide range of types and materials; a similar percentage of burials with grave goods was recorded in the OA excavations (Booth et al. 2010: 484). (Fig. 5.11). The most frequently occurring categories were hobnailed footwear, ceramic vessels and coins, other objects being represented in much smaller numbers. Flagons, flasks, jugs and beakers were the most frequently represented ceramic forms, as was also the case in the OA graves (Booth, in Booth et al. 2010: 258). Vessels related to the consumption of liquids also predominate among the glass (Cool in Booth et al. 2010: 269-72). The OA excavations require some revision to Clarke’s observations on change over time; in the second half of the 4th century AD the objects deposited with the dead demonstrated a growing diversity, including various

The emphasis on ethnic affiliation and geographical origin has meant that less attention has been paid to Clarke’s argument for the presence of a broader group of high status individuals buried in the cemetery, some of whom were the same as or spatially associated with the ‘intrusive’ graves. He identified these from stepped grave construction and from the presence of artefacts made from certain materials, including ivory, pewter, glass and silver as (implicitly) possessing greater value. This can be further assessed by comparing the ‘number of artefact types’ (NAT) of burials with these attributes to those of others in the cemetery. The NAT calculation follows the method used by Hedeager (1978: 218), with glass, metal and ceramic vessels and different types of personal 71

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

on the surface of graves in the fill of feature 12 must again be a function of better preservation of graves within this feature. The backfills of graves investigated in 20022005 also occasionally yielded objects which might be placed in this category, including horse skulls in three cremation burials (Booth et al. 2010: 499-500). Clusters of large shallow pits, up to 4m in diameter, were also identified in the OA excavation, interleaved with cremation and inhumations burials, but their largely sterile fills did not reveal their purpose which may be reasonably assumed to have been associated with burials (Booth et al. 2010: 226-33, 504-5).

ornament treated as separate categories. The technique provides a rough summary measure of the grave goods whilst avoiding the subjective weighting of grave goods involved in ‘wealth score’ attribution (Arnold 1980: 108), though it does mean that a gold ring and pair of hobnailed shoes are treated as equivalent. It also allows poorly recorded or damaged burial assemblages to be used in quantitative analysis. The subdivision of assemblages into artefact ‘types’ however requires subjective judgements not discussed by Hedeager. In the case of Lankhills the method gives a high NAT to female and child burials because of the many items in the suites of worn or unworn personal ornament. Some categories used are not mutually exclusive, for example graves with ‘female’ ornaments and with silver artefacts overlap in the case of some rings and pins.

5.3.3 Late Roman cemeteries in the eastern suburb: Chester Road, St Martin’s Close and others The topography of Winchester’s eastern suburb, including the course of the road to Neatham, is poorly understood (Esmonde Cleary 1987: 150; Ottaway et al. 2012: 317-20), but most known burials are late Roman in date and have been discovered from St Martin’s Close to the north to the slope of St Giles’ Hill to the south (Collis 1978: Haverfield 1900). It is unclear whether these findspots relate to separate burial areas or are part of a larger continuous cemetery (Fig. 5.1).

As graves with grave goods have a minimum NAT of 1, the NAT of graves with the attributes identified by Clarke should be compared to an average NAT of those burials with grave goods. The analysis (Fig. 5.12) shows that the attributes identified by Clarke as indicative of higher status are associated with graves with higher NAT values, including burials with some types of female ornament and brooches and knives. The NAT of stepped graves is also significantly higher (2.06) than the site average of 1.11. Burial differentiation at Lankhills, both the Clarke and OA samples can therefore be couched in the following terms. In the first half of the fourth century certain burials were distinguished by a number of associated characteristics both in rituals and grave form. The former included more elaborate coffins, higher numbers of artefacts and the presence of certain artefact types, as well as post-burial deposits placed in graves (see below), the latter including examples with stepped construction and location within enclosures. The coincidence of these traits was noted by Clarke (1979: 336). In the second half of the century a small number of burials were characterised by diverse grave furnishing and by burial with or wearing ornaments of differing type. The strong association between age, gender and grave furnishing has been noted by others, with substantial quantities of personal ornament in particular being associated with young (female) children and adolescents, and of cross-bow brooches and belts generally with adult males (Cool in Booth et al. 2010, 306-8; Cool 2010; Gowland 2001). In individual cases the personal ornament is especially striking for its size, materials or rarity, in particular in the case of burial 1846 from the OA excavations; objects associated with this unsexed adult included silver belt fittings, a large gilded crossbow brooch with inscription (utere felix vene vivas) and copper alloy spurs, found by the right knee. In particular the burial of boots (?) with spurs is very difficult to parallel (Cool in Booth et al. 2010: 291).

Evidence from Chester Road was recovered primarily from the western part of one trench (III), from which one hundred and ten burials were excavated, the eastern side having been destroyed by 19th century housing. The area was used as a cemetery from the late third to early fifth centuries (Fig. 5.13). For part of its history the cemetery was divided by a ditch and later a fence running east-west across the southern half of the site. Burials took place on both sides of this boundary throughout the use of the cemetery, with no apparent difference between those on either side. In the final phase a trackway ran north-northwest south-south-east on the western margin of the site, overlying earlier graves. Contemporary graves appeared to be oriented at right-angles to it. This seems to be a minor route within the eastern suburb. The high frequency of stratigraphic relationships between graves allowed seven burial phases to be established. The limited numbers of artefacts present led the excavators to extrapolate dates for these from Winchester’s other cemeteries. Cremation burials were recovered from several phases but inhumation was the main burial mode. Most graves were well shaped in all phases, the principal exceptions being three unusually deep instances. Six of the seven irregular shaped graves date to the final phases, which were also characterized by more frequent intercutting. Occasionally graves were packed with flint and stone. 25% of burials were provided with grave goods, again the proportion changing little through time and hobnails being the predominant accompaniment (Fig. 5.15). Burial 530 of a 17-25 year old female was distinguished from the rest in several respects. It was the only burial with many ornaments, wearing several worked bone bracelets and a single bronze example, and was placed in a coffin in one of the three exceptionally deep grave cuts, its sides packed with flint. Burials from which there was definite evidence for the presence of a

There are several examples of post-burial deposition of ceramics, coins, ornaments and animals. For example the fill of the ‘cenotaph’ (400) included two dogs and five coins. A group of personal ornaments had been deposited in the fill of grave 100, and two pewter bowls in 408. The frequency of evidence for coins and ceramics deposited 72

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

coffin were more likely to contain grave goods (43%) than those from which there was no evidence of a coffin (6.2%), but further associations between grave structure, burial container and grave goods were not identified. Though the majority of burials are west-east orientated there is significant change over time, this characterizing only 42% of burials in phases 1 and 2 but over 85% in later phases. Corpses were laid in the same range of positions as at Victoria Road. Three burials were

deposited prone and a maximum of four on their right side, spread over all phases. Coffins were certainly used in 33% of graves and possibly in a further 21%, these proportions not changing over time. This proportion is probably lower than in other cemeteries because of the damage to graves inflicted by intensive re-use of burial space.

Figure 5.13 Chester Road phases of burial (adapted from Ottaway et al. 2012: 180-81, fig. 84; WMS Archive)

Figure 5.14 Monument F57 and associated burials, St Martin’s Close (after Morris 1986: Fig. 4, 344)

73

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 5.15 The number of burials with different artefact types, Chester Road

massive enclosure ditch for the Iron Age settlement. This seems to have been a preferred site for burial, with other isolated burials or small groups reported elsewhere. The 45 Romsey Road excavations suggest the existence of a more substantial late Roman cemetery further to the west on West Hill near the Old Sarum road (Fig. 5.1).

16 14 12 10 8 6 4 2 0 HN

Vessel

Coin

Pers. orn.

Glass

An. bone

Burials were deposited in the Oram’s Arbour ditch from the late third to early fifth century AD. They were characterized by a population profile markedly different from other Winchester cemeteries (Ottaway et al. 2012: 170). The vast majority were of infants, over half less than three months old. At both the Carfax and New Road sections through the ditch the depositional sequence consisted of small groups of burials interspersed by periods of disuse in which the ditch gradually silted up, with only a very small proportion being coffined. At Carfax some separation was exercised between adult and infant burials, four adult burials being placed beside rather than in the ditch. West-east was the most common burial orientation. On the western side of Oram’s Arbour an entrance was excavated, through which a minor road ran in the Roman period (Ottaway et al. 2012: 171-2). On the inner edge of the north ditch of the entrance one infant, one child and two adults had been buried. In salvage excavations at 22-34 Romsey Road at the southern end of the western ditch section only adult burials dated to the late second or early third century were recovered. Adult burials were more likely to have been deeply and carefully cut, and the bodies were more commonly placed in coffins and provided with grave goods than the infant burials (Ottaway et al. 2012: 1656).

Equip.

A little to the north-east fifty two burials were identified at St Martin’s Close but only thirty one were excavated and most of these only in part, though more extensive excavation took place around a probable funerary monument which had more than one phase of use. Graves were dated to the second half of the fourth century from comparison with Lankhills (Fig. 5.14; appendix 6.2). One third were definitely placed in coffins, seven were also packed with flint or lined or roofed with limestone roof tiles. A lead-lined coffin with a ‘plaster’ burial was documented in a deep shaft within the mound (F57): this appears to be a secondary burial in what, to judge from the foundations (F55) and the painted plaster must have been a substantial mausoleum, being at least 3m long. The predominant orientation was west-east. Otherwise rarely occurring objects were present, including a box with strips of bone inlay containing a bone comb, a silver pin with bird-shaped head, and gold thread.

5.3.5 The spatial context of late Roman burial at Winchester and the ‘managed cemetery’ The general layout of Winchester’s known cemeteries has been described above (5.1). From Victoria Road East there was only very limited evidence for the attraction to the road frontage of monuments or large assemblages (5.2.2). An absence of evidence impedes the tracing of developments in the relationship between street frontage and burial during the third century A.D.; the large graves of the first phase at Victoria Road, probably dated to the late third century, aligned on the Cirencester road c. 20m away, suggest the continued attraction of the road frontage for display by the roads near the north gate, but the influence of the adjacent boundary ditch may have been more significant. The placing of the deeply cut grave at the Andover Road (Eagle Hotel) site also suggests that privileged burials could be made at some distance from gate and road here. In subsequent phases burials in this cemetery were located away from the road frontage and orientation was not influenced by the road line. At Chester road, east of the city, the apparent orientation of burials at right angles to a trackway in the final phase of development suggests that a close relationship between route and burial may have persisted into the late fourth or early fifth century, but this trackway may have served only as pathway within the cemetery, the line of which was influenced by the orientation of existing burials.

Elsewhere in the eastern cemetery small groups of late fourth century unaccompanied inhumations were excavated in Water Lane and on Blue Boar Hill, but some exceptional burial treatments were also documented. Two lead coffins were recovered, 200-300m to the south of St Martin’s Close in St John’s Street. One comprised a ‘double leaden coffin additionally protected with iron bars’ the other a lead-lined coffin from which a gold coin of Constantine was recovered, if reliably documented then an exceptional burial find (Esmonde Cleary 1987: 152; Haverfield 1900: 287). What looks like the loop of a Type IIA belt buckle (cf. Hawkes and Dunning 1961: Fig. 17g) was found in a grave on Blue Boar Hill but the context must be regarded as doubtful as the other objects recorded from the grave were a first to second century fibula and a medieval rowel-spur (Esmonde Cleary 1987: 152; pers. comm.). 5.3.4 Late Roman cemeteries at Oram’s Arbour and the western suburb Occupation evidence in the western suburb is confined to the substantial but little reported buildings destroyed by the railway cutting (Haverfield 1900: 287) and the area to the south of the Old Sarum Road which was farmed and then served for dumping of bone waste at the end of the first century AD (Wilson 1975a: 278-9). Of the small groups of burials recovered, most information derives from sections through the Oram’s Arbour stretch of the 74

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

seems to be the assumption that settlements of greater size and complexity will generate a more highly regulated cemetery space, but examples from ancient Rome (Hopkins 1983; Purcell 1987) and nineteenth century London (Curl 1980) show that while urban populations can generate intense demand for space for burial on urban peripheries, the regulation of that space is often a response to, rather than an anticipation of, such pressure. Nor is the converse of this assumption supported. Complex ‘organised’ cemeteries, for example the ‘row grave’ cemeteries of early medieval northern Europe exist without either large nucleated populations or civic authority.

The shift during the fourth century to west-east as the predominant orientation of burial in all Winchester’s cemeteries suggests that the influence of topographic features in the immediate vicinity of burial generally diminished, significant perhaps only insofar as it affected local variation within a general preference for west-east alignment. A stronger relationship between local topography and burial alignment can be noted in the rural sample (6.3). It is clear from the distribution at Lankhills of larger grave good assemblages and structurally more elaborate graves, such as those marked by enclosures, that the road frontage did not determine the location for marking burial by an enclosure or monument. The Lankhills cemetery lies close to the top of the incline followed by the Cirencester Road as it leaves Winchester and some burials during the fourth century may have been placed to take advantage of wider visibility further up the slope. The mausoleum at St Martin’s Close northeast of the city lay some distance from the road (c. 100m) though its siting on rising ground might also have enhanced its wider visibility from or view of the town below. However given the limited investigation of cemetery areas, especially of topographically more prominent parts of the extra-mural space, such as St Giles Hill rising steeply from the east gate, only limited conclusions can be drawn about preferences in monument location.

At Winchester, like most other provincial Roman cities, the burial of adults beyond the occupied area and presumably beyond the pomerium, which may follow the line of the later defences, is documented from the city’s beginnings. Although the organisational principles behind rectangular inhumation graves on a similar orientation suggest themselves to the eye more readily than that of scattered cremation burials, indicators of controlled use of space can be identified among the cremations and inhumations at Victoria Road. The absence of intercutting suggests that graves were marked and that those markers were respected. Rarity of intercutting also characterizes early Roman cemeteries from St Albans and neighbouring minor centres (Chapter 4) and other civitas capitals (e.g. Chichester St Pancras). Some late Roman ‘managed cemeteries’ can in any case be interpreted as ad hoc developments from groups of burials on the boundary ditches of settlement enclosures like those at Ilchester; examples include Butt Road (Colchester), Newarke Street and Clarence Street (Leicester) and Poundbury. Like the earlier cremations late interments seem to be made from multiple points of origin within the space given over for burial. The development of the late Roman ‘managed’ cemetery is an exaggerated phenomenon and the concept is of limited use for assessing the relative vigour of civic institutions and risks anachronism.

The development of Winchester’s cemeteries over time provides an opportunity to comment on the ‘managed cemetery’ hypothesis. From initial suggestions by Thomas (1981: 232-3), the ‘managed cemetery’, characterised by ‘a degree of small minded fussiness about order, arrangement and alignment’, has become a commonplace of the archaeology of fourth century Britain (Esmonde-Cleary 1992: 33; 2000: 128; Philpott 1991, 1993b; Philpott and Reece 1993: 421; Watts 1991). Philpott (1993b: 417) has argued that ‘semi-professional undertakers’ replaced family organisation of the funeral and introduced regular layout and practice. Standardisation of practice, as well as the choice of new sites for these late Roman cemeteries is argued to manifest an extension of civic control to the cemeteries of larger settlements. A similar view allows Burnham and Wacher (1990: 31) to extrapolate from cemetery organisation to the vigour of civic life:

5.3.6 Burial and society in late Roman Winchester In (late third and) fourth century cemeteries at Winchester various forms of cremation burial are documented in the northern cemeteries, but as the rite of a very small minority. Extended supine inhumation, typically in a wooden coffin, oriented west-east with the head placed at the west end, in a grave of which the integrity was generally respected, is the dominant burial form in all the cemeteries considered. The range of grave goods is common to all the cemeteries; by the late Roman period ceramics were less frequently deposited and hobnailed footwear and personal ornament became more frequent. Where remains of the pyre survived, presentation of corpses appears similar to those in the grave (Evans Halshaw, Lankhills). Lack of intercutting indicates that many burials must have been marked but direct evidence for this has rarely been recovered.

‘organised disposal of the dead in large communal cemeteries, set aside for the purpose, is a useful indicator of more complex communities, so that the appearance of isolated burial plots can be instructive; at Ilchester the late burials within the extra-mural plots might well be indicative of the later Roman decline in standards within a contracting settlement.’ Visual comparison of the plans of early Roman cemeteries such as Victoria Road or St Stephens’s (St Albans) with Winchester’s late Roman cemeteries renders the ‘managed cemetery’ a superficially attractive hypothesis. However the development of such a phenomenon can be disputed on both general grounds and from specific data sets. Implicit to the argument

These common characteristics set up a background against which to compare and assess differences from the

75

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

The differentiation in burial practice which has attracted greatest attention from the cemeteries of late Roman Winchester is the possible presence of ‘intrusive’ burials at Lankhills, identified by a combination of object types and style of deposition. Clarke (1979: 387) suggested that the men buried with cross-bow brooches and belts, considered to be badges of service within the late Roman army or bureaucracy, were officials running the gynaecium in the city, if the reference in the Notitia Dignitatum to the procurator gynaecii (in Britannis) Ventensis, is to Winchester (Not Dig. XL Wild 1967). This characteristic has however been de-contextualised from the study of cemeteries in Winchester. The Lankhills evidence is undoubtedly significant for the study of mobility in late Roman Britain (5.3.2), but the focus on origins does not give sufficient consideration to how cultural or status differences may have been perceived or negotiated. Although the burials concerned are distinctive in some respects, in many others they share the rituals practised for the majority of burials in the cemetery. Many aspects, including grave form and orientation, the container for the body and the way the body was laid out would have been familiar to those who had attended the funerals taking place in the cemeteries of Winchester and other Romano-British cities. The burial of a group of males with the specific symbols of late Roman authority, the crossbow brooch and belt set, is more unusual both at Lankhills and within a broader Romano-British context, though it too fits within a local and wider context where funerary display is increasingly structured around impressive personal ornaments and dress (see further 8.5). The burials in the ‘intrusive’ groups share the phenomenon at Lankhills for deceased older children and young women to be associated with the most substantial assemblages of personal ornaments, whether won or unworn (Gowland 2001; Cool in Booth et al. 2010: 307).

norm. There were distinctions in ritual practised, by grave form, whether by survival of evidence of a container for the body, and by the material and construction technique for that container, by quantity or type of grave goods, in particular personal ornaments, and occasionally by enclosure or monument. Some between-cemetery differences can be established, for example between the different cemetery areas in the northern suburbs, where provision both of a coffin and of grave goods was much more frequent at Lankhills than in other areas; in the light of the argument above, as much weight should arguably be put on the former as on the latter for differentiating the resources expended on the funeral. To Esmonde Cleary (1987: 155) the ‘poorly dug graves’ and ‘low quality’ grave goods at Oram’s Arbour suggested that ‘this land of marginal utility may have been used for the burial of the marginal in society’. However the key distinction between this site and others relates primarily to age; here are many of the infants otherwise absent from most late Roman cemeteries in Winchester. The treatment of the few adult burials resembles that in other cemeteries. In the eastern cemetery a distinction is hinted at by the difference between burials on the Chester Road site, which frequently intercut and were rarely furnished, and the small groups at St Martin’s Close and St John’s Street, associated with otherwise rare grave goods, placed in more unusual burial containers and marked by monuments. As if not more significant than the between cemetery differences are those within the different areas; in all cases a small minority of burials were distinguished by exceptional grave construction, some being stepped or unusually deep or large, or the presence of a monument. In several cases these different characteristics coincided to differentiate a small number of bodies, at Lankhills and occasionally elsewhere (e.g. St Martin’s Close) some of these burials are associated with objects which are scarce and / or symbolically laden. A recurring minority rite not usually associated with traits just mentioned is for the individual to have been decapitated or laid prone rather than supine. While the perception of late Roman burial rituals and cemeteries is often of homogeneity, consideration of the full range of evidence for ritual treatment rather than grave goods alone therefore reveals differentiation in burials within and between cemeteries which may be just as significant as in the early Roman period. It is not yet possible to comment in detail on any possible association between health status, and its possible social dimension, and burial treatment. Two hints at status difference in the lived experience of individuals buried at one cemetery may however be mentioned from the results of the Oxford Archaeology project at Lankhills A much higher instance of fractures was observed among the individuals buried prone than among the sample as a whole (Booth et al. 2001: 478). Analysis of stable isotopes suggested that the individuals buried in coffins in the OA excavation had greater access to marine animal protein than was typical, while those buried prone had less (Cummings and Hedges in Booth et al. 2010: 417-19).

On this evidence there is greater evidence for burial display at Lankhills at different stages of the ritual than in other cemeteries, the presentation of the body (if dressed perhaps for the laying-out), the container in which the body is borne or buried, the grave goods deposited with the dead and the construction of some graves, singling them out what is otherwise likely to have been a plethora of low mounds. At Lankhills display through grave goods, especially personal ornament and dress fittings and the average NAT per grave is more intense than other cemeteries in the northern suburb (Fig. 5.16 and 5.17). Although grave goods were sometimes deposited in number with burials outside this cemetery, for example among the earlier burials at Victoria Road West, the persistent deposition of grave goods with a high proportion of graves into the late fourth century differentiates Lankhills from Winchester’s late Roman cemeteries. The deposition of later offerings in the fill of or on the surface of the grave and elaborate grave constructions and enclosures may have been more frequent, the product of post-burial rituals here, but this may reflect differences in the preservation of grave fills.

76

BURIAL TTRADITIONS AND URBAN SOCIETY IN WINCHESTER

Figure 5.16 The proportion of burials with grave goods in late Roman cemeteries at Winchester

from the earlier period. In the extra-mural area, especially the northern suburb which has been much more extensively excavated, the evidence is of more ephemeral occupation (Ottaway et al. 2012: 91-6, 100-09). Within the walls however fourth century houses were built (or rebuilt) in brick or masonry rather than timber and mosaic floors were laid. Some of these were perhaps of a more substantial courtyard form (James 2006: 27-46; Zant 1993: 83-127; Ford et al. 2011: 59-70). The emphasis on architectural investment lies squarely on house rather than tomb, but there is convergence in both settings towards the routines and rituals of smaller groups rather than the wider urban community. Limited published information from Winchester on domestic finds assemblages compounds the difficulty of assessing the availability of artefacts placed with dead caused by different pathways to deposition between settlement and burial. However although as earlier grave goods comprised many items which were in common circulation, selected because of their function (e.g. the preference in ceramic and glass vessels for forms related to drinking), some, for example silver and ivory personal ornaments, were rarer. In occasional cases, unlike the earlier period, individual burials were furnished with objects otherwise very rarely attested and signifying likely high rank; the furnishing of grave 1846 at Lankhills with gilded crossbow brooch, silver belt fittings and spurs is the clearest example of this.

60 50 40 30 20 10 0 Chester Road

Hyde Street Late

Lankhills

Victoria Road

Figure 5.17 The average NAT per burial in late Roman cemeteries at Winchester 2

1.5

1

5.4 Conclusion At Winchester the burials of individuals of extraprovincial origin in the Lankhills cemetery have garnered the lion’s share of archaeological interest in the city’s Roman cemeteries, but this has distracted attention from the opportunity to assess how burial practice changed over 400 years within a Romano-British urban community, how social distinctions may have been presented in burial in that period and how those changes compare to other aspects of archaeological evidence from the city; for the latter, given the limited data available, we must also extrapolate from the situation at other Roman towns. The burial rituals practised over time shows a characteristic sequence for southern Roman Britain, with cremation dominant in the first to early third centuries AD alongside a minority of inhumation burials, infants being the only group to which it is usually applied. There is an apparent gap in burial evidence during the third century AD, but from the later part of that century until the fifth century AD inhumation dominated. It is difficult to know how much weight to place on the greater number of later Roman burials documented from Winchester’s cemeteries, a common characteristic of Romano-British burials, and it is not yet possible to distinguish between the possible factors responsible, including population size, preference for burial location, the changing archaeological visibility of the burial mode (cf. 2.5) or the chance distribution of the areas so far excavated.

0.5

0 Chester Road Hyde Street Late

Lankhills

Victoria Road

This should be qualified by noting that not all possible indicators of prestige in late Roman burial are yet documented at Lankhills. ‘Plaster’ burials and lead-lined coffins, rare in Winchester compared to other late Roman towns in Britain (cf. Green 1993; Toller 1977), have been recorded on the Andover Road (Eagle Hotel) site and in the eastern suburb. No parallel to the masonry monument at St Martin’s Close has yet been documented at Lankhills. On other indicators the space for competitive assertion of status had changed in the late Roman period. As in most Romano-British towns, there is less evidence of investment in public buildings, with masonry from major buildings found reused and some streets disused (Biddle and Biddle 2007: 191). The defensive circuit represents the major evidence for public works so far documented of late Roman date, the Flavian earthwork being fronted by a stone wall in the third century AD, perhaps, and possibly studded with towers in the fourth, though evidence for the latter is limited (Biddle and Biddle 2007: 192-3; Qualmann 1993: 73; Wacher 1995: 297-9). The character of private building shows significant changes

Changes in burial practice in later Roman Britain are sometimes described as a process of homogenization (Philpott 1991: 224-26). On the evidence from Winchester this characterization cannot be fully sustained 77

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

and is too focused on grave goods. It is impossible directly to compare early and late Roman periods directly because of the change in burial practice. Nevertheless from this case study it is difficult to argue that the sphere of burial became less competitive in the late Roman period. Change from cremation to majority inhumation meant that the pyre no longer offered the occasion for the destruction of materials, though on current evidence this was not in any case common in the early Roman period. It is impossible to assess how the other elements preceding the burial changed in nature, such as laying out the dead or the procession to the tomb, though ornaments worn by some individuals might derive from their being dressed and displayed at this earlier stage. In the later Roman period there was a greater differentiation in the container for the body, with occasionally lead and more commonly massive wooden coffins distinguishing a small group of burials. The range of objects deposited with the dead remained broadly the same but the emphasis shifted from ceramics to personal ornament, more commonly for women and children than for men, though outside Lankhills the proportion of burials accompanied by grave goods of any type was low. The significance of the lack of grave goods, especially ceramics is difficult to assess and the relative expense, for example, of making a wooden coffin against deposition in an urn with a locally made pot can only be guessed at. Nevertheless the possibility should at least be raised that the former would be more costly. In both periods burials must often have been marked, to judge from the lack of intercutting in most cemeteries. Very few such markers survive, but there is slightly greater evidence for monumental tombs from the later Roman period. From the Hyde Street evidence the road frontage was exploited to a limited degree in the early Roman period as a space of display. During the fourth century AD the focus of monument building seems to have shifted to the interior of what very extensive cemetery areas. Individual monuments may have been sited with some concern for their wider visibility or impact, but this requires fuller testing. In summary, a small proportion of individuals in Winchester’s cemeteries were consistently differentiated from the rest by a privileged burial treatment. This is agreement with the results of the analysis of the St Alban’s cemeteries, though these covered a more limited period. Aspects of that treatment in both periods sometimes derived from non-regional or extra-provincial traditions, but nevertheless adhered to, and had the potential to impress within local forms of display. When other aspects of archaeological evidence are taken into account, the assertion of difference seems to shift over time from monumental civic buildings to the routines of daily life and the rituals of death.

78

CHAPTER 6: DEATH IN THE COUNTRY 6.1 Introduction: current models of social and tenurial distinctions in rural burials

May suggested too that the burials dispersed across the settlement at Dragonby were not ‘part of a regular tradition’ and were those of ‘low status or diseased’ individuals (1996: 125-6). Jones’ (1987: 828) contrast between ‘ordered’ and ‘disordered’ rural cemeteries implies an analogous distinction and similar arguments have been made for other sites (e.g. Mackey 1997: 31). The use of existing features for burials has been explained by the convenience of lesser effort to excavate, often on land by previous use or poor quality rendered unusable for agriculture (R.F. Smith 1987: 115-18; Swan 1984: 50). Philpott and Reece (1993) further argued that difference in burial location reflects tenurial position; while landowners and their families maintained a formal burial space over several generations to legitimate their right to the land, tenants and workers devoted less effort to maintaining a formal cemetery on land to which they lacked a long term attachment. Although this connection is not drawn explicitly, the proposal recalls the influential ‘Saxe-Goldstein’ hypothesis, relating disposal of the dead in a formal bounded cemetery area with the legitimation of rights to critical contested resources (Chapman 1981; Goldstein 1981; Morris 1991; Parker Pearson 1999a: 31).

This chapter shifts the focus from town to country and has two main aims. Starting with samples of rural burials from Hampshire and Hertfordshire it evaluates current characterizations of rural societies based on the evidence of burial practice, focusing on the treatment and in particular the placing of the dead in and around rural settlements. It also provides a building block for subsequent analysis of the broader distribution of display in burial, comparing rural practice with that in civitas capitals and minor centres (chapter 7). Despite the growing interest in the character of rural Romano-British society (Taylor 2007 with references), the exploitation of burial evidence has so far been limited (1.3) and has focused on intra-site differences in a small number of examples. A general model of status distinction in rural burial practice based on ritual and spatial criteria has been intermittently advocated for the Roman provinces, of which Martin-Kilcher’s (1993a) is the most developed version, specifically focused on villas and building on earlier observations (cf. van Doorsaeler 1967: 24-26). This locates the graves of the villa owner and family in enclosed ‘jardins funéraires’ or similar, often highly visible through their prominent position in relation to the main buildings of the villa or its principal access routes, often enhanced by monuments. This contrasted with graves for subordinates, such as villa workers or slaves, in small, poorly furnished and largely invisible tombs outside the villa enclosure, often to the rear. Classifications of rural burial in Britain have drawn similar distinctions between burials enclosed in formal cemeteries and those dispersed across settlements. The latter are the most frequently encountered type of burial, in rural Britain at least, and are typically interpreted as those of individuals of low status (Collis 1977b; Jones 1987: 828; Philpott 1991: 232; Philpott and Reece 1993: 422). From the location of furnished and unfurnished graves at Owslebury, Collis (1977b) extrapolated an influential template for status-related difference based on placing and furnishing (6.3.2). In this lower status is reflected by interment in individual or small groups of burials outside formal cemetery areas and scattered around settlement sites, often within or close to nonburial features. To Philpott the implication of minimal effort by exploiting existing features prompted an identification of individuals buried at such locations as social outcasts:

On initial consideration there are empirical weaknesses in the data that are argued to support such models and the spatial distinction proposed above is difficult to identify. While it is not difficult, especially in some regions of Gaul and Germany, to identify monuments probably for high-status groups which meet Martin-Kilcher’s description, her survey identified only a very small number of sites where the systematic variation she proposed could be detected. Even among these some examples can be contested. At Köln-Mungersdorf for example the separate cremation and inhumation cemetery are not contemporary (Fremersdorf 1933) and at Courroux only a single cemetery was located. The numerous extensively excavated villa settlements from the Hambach forest west of Cologne offer a substantial opportunity to test this model but in interim analysis only at two of the five sites with more than twenty burials (Hambach 69, 303) was Gaitzch (1986; 1993) able to detect significant distinctions in numbers or types of grave goods between different groups of burials. Nor, for Britain, did Philpott and Reece propose specific examples to illustrate their hypothesis. Beyond Collis’ example at Owslebury (6.3.2) it has not yet been established whether rituals for burials placed in close association with other settlement features such as boundary ditches are in fact consistently different from those conducted within more formally demarcated cemetery areas. The small sample of rural burials in Britain, especially for the early Roman period, has already been noted and may indicate a continuing tradition of archaeologically invisible burial for perhaps the majority of the population (2.3). The study of variation within archaeologically documented burial therefore examines only a likely minority of the population.

‘the substantial group of burials in disused features such as ditches, corn-drying ovens, pottery kilns or wells may be the result of indifference or laziness on the part of the grave-diggers, violent or illicit death, or disapprobation on the part of the family or community.’ (Philpott 1991: 232).

79

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

The changing emphasis of settlement archaeology, in which boundary features for settlement sites have acquired much greater prominence also provides an opportunity for closer examination of the placing of burials (cf. 1.4). The longstanding interest in the process of landscape division through settlement and field enclosures and trackways has been given renewed impetus for the Roman period by consideration of its social dimension, in particular the way in which group identity was negotiated in relation to landscape and property (Chadwick 2008; Taylor 2007). With occasional exception however the character of spatial associations between boundaries, burials and other deposits has not been closely examined (Chadwick 2012; Esmonde Cleary 2000). Given the good contextual information now available for many burials within their settlement context, there is scope here for examining further the space of the dead in relation to the wider construction of space as a medium through which social relationships were negotiated (1.3.2).

purpose are not mapped. Figures 6.1 and 6.2 generally reflect the distribution of Late Iron Age and Roman archaeological data of all kinds. The majority of excavated Iron Age and Roman burials in Hertfordshire derive from cemeteries at Verulamium and the minor centres discussed in chapter 4. Outside these there are clusters of burials to the south of Welwyn, Baldock and Ware, as well as a small group near Cow Roast, although the environs of towns appear largely devoid of burials, up to ten kilometres around St Albans and two to five kilometres at Baldock and Ware. Field walking and aerial photography in the environs of St Albans certainly suggest that rural settlement is not absent from this area (Hunn 1996). The absence of burials is more plausibly explained by fieldwork bias than by the burial of the rural dead from this zone within urban cemeteries. The greater quantity of rural burial evidence from the centre and north of the county mirrors that of other LPRIA and Roman period archaeological evidence. With the exception of St Albans archaeological fieldwork has had a greater intensity in north and west Hertfordshire, on the Chiltern dip slope and the valleys which cut it, rather than the clay plateaux to the east, as well as on certain development corridors, particularly the A1 (Ermine Street), a bias also manifested in the distribution of villas (Bryant and Niblett 1997: 280; Niblett 1995b: 8, 74). Areas such as Stevenage, Hemel Hempstead, Hertford and Watford also lack the tradition of antiquarian and archaeological interest that characterises St Albans and Letchworth/Baldock. A general association between burial discoveries and Roman roads also emerges, burials seeming to concentrate on roads from St Albans to Welwyn and Braughing, Dunstable to Baldock, Cow Roast to St Albans and Ware to Enfield.

The first sections of this chapter (6.2 and 6.3) assess burial rituals in the sample areas already studied in chapters 4 and 5, using modern counties for the purposes of comparison. For both Hertfordshire and Hampshire the distribution of burial data of the Late Pre-Roman Iron Age (LPRIA) and Roman period is briefly assessed, demonstrating the sub-regional and local variation in the availability of evidence (cf. 2.3). Burial practices in the rural cemeteries in these sample zones are then summarized, in order to assess how and to what degree burial practice is differentiated in a rural context and in order to allow subsequent comparison with the civitas capital and minor centres (chapter 7) which were discussed in chapters 4 and 5. As will be seen both rural samples are small but indicate recurring and significant relationships between burial and associated settlement space; this aspect is then extended through a broader survey of evidence from rural settlements from the first century AD to the first decades of the fifth century (6.4). The published sample of rural settlements with associated burials has grown very considerably in the first decade of the twenty-first century (2.3). The sites considered in this wider survey do not represent an exhaustive collection of relevant data, but comprise significant examples mainly derived from excavations from the 1960s onwards, principally from central and southern Britain; appendix 6.2 describes the evidence from the individual sites referenced. The survey identifies recurring characteristics in placing the dead around rural settlements, in particular associations with boundary features, and seeks to interpret those associations.

In comparison to the many well recorded urban and minor centre grave groups (chapter 4) the evidence for rural burial is limited. Villas, particularly around St Albans and in the south-west of the county, have dominated study of Hertfordshire’s rural landscape (Branigan 1973; Hunn 1992; 1994; Neal 1978; Niblett 1995b). However even large-scale excavations such as Gorhambury (Neal, Wardle and Hunn 1991), Gadebridge Park (Neal 1974), or Dicket Mead / Lockleys (Rook 1986) have recovered only a very small sample of directly associated burials. In this respect Hertfordshire corresponds to the national pattern (2.3). Although many non-villa sites have been identified, few have been extensively excavated (Burnham et al. 1995: 354-5; Going and Hunn 1999; Hunn 1996: Morris and Wainwright 1995; McDonald 1997; Moss-Eccardt 1988; Partridge 1989). Smaller groups or individual burials scattered across non-villa settlements have also been recorded, as illustrated at Thorley, M1 Widening Junction 8, Boxfield Farm and Foxholes Farm. ‘Villages’, for example Slip End, Ashwell (Burleigh 1976), have received little attention. Some older records suggest the existence of substantial rural cemeteries, cremation cemeteries north-east of Great Wymondley and at Hinxworth New Inn, as well as mixed cemeteries at Slip End, Ashwell and Foxholes (Hitchin), but with too little information to use in analysis.

6.2 Burial in rural Roman Hertfordshire 6.2.1 The distribution of Iron Age and Roman burials Figures 6.1 and 6.2 plot the distribution of Late Iron Age and Roman burials in Hertfordshire (excluding St Albans and the small town sites). These distribution maps (and figure 6.6) include only excavations of burials or funerary monuments. Small enclosures known from aerial photographs which have a possible but untested funerary 80

DEATH IN THE COUNTRY

Figure 6.1 The distribution of LPRIA period burials in Hertfordshire (K. Robbins)

Figure 6.2 The distribution of Roman period burials in Hertfordshire (K. Robbins)

81

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

One so far exceptional site is the shrine at Ashwell. Discovery of votive plaques dedicated to a previously unknown goddess, Senua, prompted excavation of a sanctuary site which revealed votive deposits including small quantities of cremated human bone (Jackson and Burleigh 2007; Burleigh and Fitzpatrick-Matthews 2007: 413-4).

detect any principle behind the cemetery layout, especially as many shallower burials have probably been destroyed by ploughing. The excavators suggest that large pit GAP, at some distance from other burials and seemingly lined with carbonised wood, was a ‘poor man’s mausoleum’, but no artefactual material or cremated bone was recovered from the fill; a non-burial deposit might alternately have been placed here.

6.2.2 Burial practice in early Roman rural Hertfordshire Like other aspects of the Early and Middle Iron Age archaeology of Hertfordshire, evidence for burial practice is limited (Bryant 1995). Deposition of part skeletons or skeletal fragments on settlement sites, the common pattern for Iron Age southern Britain, is documented mainly from Baldock and from Braughing in the LPRIA and Roman period (4.3.1; 4.4.1). From the mid to late first century BC cremation predominated as the burial ritual until the third century AD, inhumation thereafter. Cremation cemeteries and those of LPRIA and early Roman date dominate the rural as well as the overall sample from the county (Figs. 2.4, 6.1). Occasional instances of isolated human bones are also recorded; for example a skull was recovered with a stone ‘face’ in a well fill from the villa at Northchurch and a likely structured deposit (cf. 3.4) in a pit fill of early Roman date from the M1 Junction 8 site included a fragment of human skull as well as much animal bone and charcoal.

The second example is the (unpublished) late first and second century AD cemetery at Cross Farm, Harpenden, the settlement context of which is unknown. The recovery of a scatter of calcined bone during fieldwalking prompted further examination of the site but only part of the cemetery has so far been examined (in 1992) and its full extent is unknown. The excavated portion of the cemetery revealed two clusters of burials for which pyre and burial rituals can be examined. Burnt animal bone recovered from three graves comprises the sole evidence for pre-burial treatment, though it is likely that the representation of animal bone is affected by the collection of the remains of the dead; the cremated bone quantities from these three graves were amongst the largest from the site. The range of weights of cremated bone was very similar to that at urban and minor centre cemeteries (Fig. 4.6). Grave cuts were mostly of sufficient size for the urn; in one case a layer of flints had been placed around the urn base (23), and in others (21 and 22) urns had fallen into the centres of pits, possibly when other artefacts had decayed. Most burials were deposited in jars and accompanied by at least two accessory vessels. Samian accounted for 27% of the twenty six vessels in the accessory assemblage, similar to the proportion at the minor centre cemeteries (Fig. 4.27). The most common accessory vessel forms were flagon and dish. Two glass vessels and a finger ring with intaglio comprised the only non-ceramic grave goods. Occasional evidence was noted of deliberate damage to objects in well preserved burials. A flagon in 25 lacked its rim and there were missing pieces from the rims of glass flasks in 21. It is not clear whether poorer artefact assemblages are the product of original depositional practice or subsequent disturbance. No difference in furnishing was identified between the burial clusters.

For the early Roman period two examples offer the opportunity to examine rural burial practice in some detail, to be supplemented with evidence from other sites with fewer burials or limited records. The first is from Boxfield Farm, Stevenage, comprising twenty five cremations, dated from the Flavian to Antonine periods, most to the latter, and is associated with a non-villa settlement (Fig. 6.3). A single fragment of melted glass, a burnt samian sherd, one fragment of possible bird bone and a single further possible instance of animal bone comprised the evidence for pyre goods. Charcoal and burnt flint were often deposited in burials, though what type of Brandschuttgräber these were cannot be definitively established since the volume of material is not specified (cf. 3.4). The heavy plough damage to burials obstructs assessment of the quantity of cremated bone and comment on grave layout: at most 0.1m of deposit survived in most cases. Most cremation burials were deposited in jars. In some cases iron fragments may indicate the original presence of wooden boxes, and there was one example of a probable casket with five lion-head studs (GBA). The typical burial assemblage comprised a dish and drinking vessel; flagons are less well represented than on minor centre sites (cf. chapter 4). Samian comprised 43% of the 32 accessory vessels, a higher proportion than at any urban site, as the excavators also noted (Going and Hunn 1999: 33) (cf. Fig. 4.27). The only non-ceramic grave good recovered apart from iron fragments was an unusual form of glass cup (Isings 27). This and some samian vessels were probably damaged before burial. There was little differentiation between grave good assemblages. The maximum number of accompanying vessels was four (GAX). It is difficult to

Other information on contemporary burial practice is slight. Ceramic assemblages from the handful of graves at Little Wymondley, Foxholes Farm, and M1 Junction 8 are similar to those at Boxfield and Cross Farm. The typical accompaniment for cremation burials at Foxholes Farm was a flagon or beaker, although assemblages are poorly preserved. Cremation burials from a non-villa settlement at Thorley and Baldock bypass were rarely furnished. Limited documentation from other sites, for example Slip End and Foxley Hill (Ashwell), Hinxworth New Inn, Barley Homestall Farm, Great Wymondley, and The Crown, Kelshall, suggests the deposition of similar grave furniture. Spectacularly different are two late second century burials from Turnershall Farm near Wheathampstead, both with substantial quantities of 82

DEATH IN THE COUNTRY

grave goods, found 50m south of a villa which was destroyed in the mid second century AD. The only perhaps comparable assemblages are the many glass vessels reported with one or more burials in the Watford barrow, as well as the substantial ceramic assemblages from burials associated with the enclosure at Lindum Place, St Albans, a few hundred metres from the town. More modest assemblages were documented at the barrow burials at Youngsbury and Pickford Hill as well as at Boxmoor villa, all cremation deposits in glass vessels.

corporate group (Black 1986). An example of the latter – the dendrofori - is attested for St Albans a few miles away in a dedication on a beaker placed in a burial in Dunstable (RIB II.8 2503.114). 6.2.3 Burial practice in late Roman rural Hertfordshire Extended inhumation, typically with little grave furniture, is likely to be the most common late Roman burial rite; radio-carbon dates for such burials from Queen St, Hitchin, shows this ritual to endure into the post-Roman period and only wider application of radio-carbon dating to unfurnished inhumation burials will allow the history of changing rituals into the mid first-millennium to be more confidently written (Davis 2005). The general scarcity of inhumation burials in Hertfordshire, late Roman or otherwise (2.3) is in part a product of the relative hostility of soils in much of the county to bone preservation, for example the clay with flints overlying the chalk of the dip slope of the Chilterns to the north and west of the county and the tertiary sands and clays in the river valleys (Niblett 1995b: 8). The only well recorded late Roman cemetery lies adjacent to the Welwyn Hall mausoleum. Other records of inhumation cemeteries, for example Hinxworth Place (Hinxworth) and Sawbridgeworth are too brief to support detailed analysis or comprise isolated burials. The decline in the availability of evidence in the late Roman period continues into the early mediaeval period; especially from the mid fifth century onwards cemeteries are notoriously scarce in Hertfordshire, as students of the post-Roman period have long appreciated (Malim et al. 1996: 112; O’Brien 1999: 118), but the lack of evidence from the late Roman period has not previously been noted.

In contrast to the relative paucity of burials there is an abundance of burial monuments from rural Hertfordshire, in particular barrows, though none have been subject to adequate modern investigation (Dunning and Jessup 1936; Jessup 1962). The standard of documentation varies widely, but we can rarely be sure that assemblages are fully reported and many barrows are not securely dated to the Roman period (Dunning and Jessup 1936; Foster 1986: 192-4). Given the variety of Roman re-uses of prehistoric barrows (Williams 1998a), the simple record of Roman material associated with barrows is insufficient to prove a Roman date of construction; in one recently excavated example within the study area Bronze Age barrows were re-used as the site of modest Late Iron Age and Roman cremation burials, placed in their siltedup ring ditches (Baldock Bypass). Recent study of the Moulton Hills and Chronicle Hills in neighbouring Cambridgeshire illustrates the difficulties associated with dating barrow burials (Baxter 2006; Taylor and Arbon 2007). Comparative maps which include the many possible as well as confidently dated examples may therefore exaggerate the numbers (e.g. Becker 1993: 361, Abb. 1). Of the barrows excavated within Hertfordshire (Appendix 6.6), the Roman period date of Youngsbury and Pickford Mill is established from burial assemblages, from antiquarian reports of Roman material at Hoddesdon and Watford and from the location adjacent to a Roman road at Six Hills, although the Bronze Age barrows at Badbury Rings caution against this characteristic as a reliable criterion of Roman date without supporting evidence (Fowler 1964). One of the burials in the enclosures excavated south of St Albans at Lindum Place was at the centre of an 8m square enclosure, and from the evidence of postholes in the two corners excavated may have had a timber monument raised over it. The most architecturally elaborate monuments are ‘templemausolea’ from Wood Lane End, Rothamsted and Welwyn Hall (Fig. 6.5). The primary burial in the latter was buried in a Greek marble sarcophagus. The concentration of this monument type, which also includes the Folly Lane monument (4.2.1), is not paralleled elsewhere in Britain, though all offer different variations of the relationship between temple and burial, and a funerary function at Wood Lane End and Rothamsted is not proven. At Wood Lane End other buildings associated with the large enclosure (c. 70 x 80m) within which the tombs were set may have been related to probable cult activity, including a bath-house and another building conjectured to be a possible schola for a guild or other

The single excavated cemetery for which information is available is the small late fourth or early fifth century cemetery established close to the demolished mausoleum at Welwyn Hall, overlying earlier field boundaries. The possible fifth century date is suggested from the extremely worn and fragmentary state of the fourth century vessels deposited as grave goods. The cemetery comprises twenty six east-west orientated inhumation burials, one of which had been surrounded by a penannular ditch. One fifth of burials had been placed in wooden coffins. Evidence for grave goods was recovered from over half the burials, principally ceramics, many of which were fragmentary or broken before deposition, and also hobnails, ornaments and animal remains. Other information is very limited. A single stone coffin was recovered at Ayot St Lawrence, and two, one with a lead lining, at Park Street (Appendix 6.2). Limited information from other sites suggests that provision with grave goods may have been relatively frequent, for example from the Water pipeline diversion site on the Baldock bypass. The silver pins recovered from female burial 2 at Park Street indicate more lavish burial furnishing in some instances. Evidence of monuments is less plentiful than for the earlier period. Parts of grave superstructures were excavated at Park Street; there was also evidence of a possible mausoleum at Gadebridge.

83

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

excavations at Thorley, Bishop’s Stortford, have revealed individual cremation and inhumation burials distributed across several settlement enclosures, with a particular concentration in and around enclosure B, spread in time over at least two centuries. Part and whole animals were also deposited in and around the perimeter of this enclosure.

6.2.4 Burial and settlement space in rural Roman Hertfordshire The predominant location of burials is on settlement margins, for example at the Gadebridge Park and Park Street villas or at Little Wymondley and Foxholes, where they were immediately outside the south-west corner of the enclosure and bounded by an earlier ditched feature (Fig. 6.4). In the M1 Junction 8 widening scheme site cremation burials were also in the corners of adjacent ditched enclosures. At Boxfield they were outside the settlement enclosure and possibly close to a droveway, within a larger system of enclosures dated from the midfirst to fourth centuries AD (Fig. 6.3). At Welwyn Hall late Roman burials were associated with field boundaries as well as with the earlier temple-mausoleum. The

Burials were also associated with other features. At Gadebridge Park an adult west-east oriented inhumation burial lay less than a metre from the stokehole of a corndrying oven of the mid second to mid-third century and a disarticulated skeleton was recorded from a well fill at Little Wymondley.

Figure 6.3 A rural settlement at Boxfield farm: (left) enclosures with early Roman cremation cemetery located on north-west side and (right) plan of cemetery (right) (after Going and Hunn 1999: 8, fig. 4 and 30, fig. 15)

Figure 6.4 Settlement (prehistoric and Roman) and burials in Area 1, Foxholes Farm (adapted from Partridge 1989: Fig. 4)

84

DEATH IN THE COUNTRY

Figure 6.5 Wood Lane End mausolea and associated features (after Neal 1984: Fig. 1, 194)

A chronological relationship can be tentatively suggested between burial and significant changes in settlement layout. Saunders (1961: 122) noted that the burials at Park Street were roughly contemporaneous with the rebuilding of the villa in the late third or early fourth centuries. The Welwyn Hall mausoleum was constructed at roughly the same period as the building of the Dicket Mead complex across the river, with the Greek names in graffiti from the villa and the Greek origin of the sarcophagus buried in the mausoleum giving a suggestive but unprovable connection between the two (Rook et al. 1984).

for a relatively brief period. The Welwyn Hall mausoleum was constructed in the mid third century and demolished during the fourth, although its influence on local topography persisted as an orientation point for a field boundary and the late Roman inhumation cemetery discussed above. The use of the entire Wood Lane End complex was confined to the second century AD, during which time Building 1 was demolished and its material probably re-used in building 6. 6.2.5 Burial and society in rural Roman Hertfordshire Any synthesis based on this fragmentary and limited data must be very provisional. Given the limited context information it does not allow any sustained comparison between villa and non-villa sites. In view of the small numbers of burials so far documented it is difficult to assess how far the normative rituals, as they appear archaeologically, apply to the population as a whole. The evidence for the settlement context of burials is limited, but encompasses a mixture of villa and non-villa settlements. Within the early Roman period sample the rituals practised and the degree of differentiation between them are similar to those documented at St Albans and the minor centres (4.7), with some larger and more diverse grave good assemblages existing in the rural sample. The much higher percentage of samian vessels in burials than settlement deposits in a rural setting documented in the two main samples has been noted as a wider phenomenon, and contrasts with the small percentage constituted by samian among ceramics deposited in rural settlements (Biddulph

These monuments were often visible in the broader landscape. The Welwyn mausoleum appears to have been visible from the main Welwyn cemetery (4.5), the settlement beyond the river, as well as the Lockleys and Dicket Mead villas. The Wood Lane End tower was likely to be visible over a wide area of the plateau west of St Albans (Neal 1984), indeed from as far away as Folly Lane on the other side of the valley of the Ver (Hunn 1994: 46). The impact of both structures was increased by their great height and by their rendering in white. Both monuments were sited close to although not directly upon roads, although the course of the Silchester road from Verulamium which passed to the south of the Wood Lane End enclosure is not securely established (Neal 1983: 74). Barrow monuments also appear to have been situated beside roads, for example Six Hills at Stevenage. Where this could be established monuments seem to have stood 85

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

2005; Willis 2004: 7.2.7-8; 2011: 188-9). A larger, more diverse and in occasional cases more architecturally complex corpus of funerary monuments than (is firmly documented) at St Albans or the small towns was also noted. The diversity of practice and of material investment in burial noted in the rural sample as a whole, if not at any single site, contrasts with the more limited differentiation visible in rural houses of the same period, which have more limited evidence for Roman-style amenities or decorative techniques and are smaller and less complex in layout than the late third and fourth century villas. On these various indicators early Roman burial practice therefore appears as a more significant sphere in a rural context for competitive display than it does in town or minor centre. Analysis of the later period is inhibited by lack of data but burial ritual does not, on current evidence, mirror the increasing differentiation in house forms, especially as represented in local villas which like elsewhere in Roman Britain, reached their apogee in terms of size, complexity and provision of amenities (for example Gorhambury, Neal et al. 1991).

6.3 Burial in rural Roman Hampshire 6.3.1 The distribution of Roman burial evidence from Hampshire As in Hertfordshire urban development and road building, antiquarian and recent archaeological activity and soil types condition the archaeology of the LPRIA and early Roman period in Hampshire. The first two factors have produced intensive examination of sites along the M3 corridor and A303 widening scheme, and at some urban sites, particularly around Winchester, Alton / Neatham, and the environs of Basingstoke, although there is little evidence from the heavily urbanised coast of the county from Southampton eastwards (Champion and Champion 1981: 39; Cunliffe 1993: 179-80; Fulford 1996; Massey 2006). The archaeology of the chalk downlands in the north of the county is much better known than the river valleys to the south, of which only relatively recent excavation is recovering settlements of the late Iron Age and Roman period (e.g. Adam et al. 1997; Crockett 1996). Unusually for the Roman period a research project (Danebury environs) makes a substantial contribution to the rural burial corpus (Cunliffe and Poole 2008a).

The spatial relationship of burial site to settlement is similar to the province as a whole and is considered with these later in this chapter (6.4).

Figure 6.6 The distribution of Roman period burials in Hampshire (K. Robbins)

86

DEATH IN THE COUNTRY

Roman burials in the county, as mapped on Fig. 6.6, are more abundant in (local government) districts in the north of the county (Basingstoke, East Hampshire, Test Valley and Winchester), with clusters around Winchester, Alton, Andover and Basingstoke. Districts in the south-west and south are under-represented, especially the New Forest and also Southampton, Eastleigh Fareham, Portsmouth, and Havant. The distribution of burials of all periods is the similar. For example the majority of long barrows (RCHME 1979: frontispiece) and Iron Age burials on settlement sites (Whimster 1981: figs. 4, 7 and 10) have been discovered on downland in the northern half of the county. The distribution of early Anglo-Saxon burials is also biased to the same areas (Hawkes 1986: Fig. 6; O’Brien 1999: 149), the main differences being the tighter focus of early mediaeval burials on Winchester and the substantial cluster of burials around Southampton, with other undated burials from Bitterne and Southampton also likely to belong to the early mediaeval period (Cotton and Gathercole 1958: 30; Wendy Barrett pers. comm.). The distribution is also affected by the different preservation environments within the county for human bone, which have been characterised in a wider consideration of burial environments in Wessex (Mays 1991). The extensive chalk downland provides relatively good preservation condition for inhumation burials in particular, but the tertiary clays and sands in the Hampshire basin and the heathland of the New Forest offer less favourable environments.

Soberton and occasional post-Roman burials, the burial practice of villa occupants remains very poorly known (cf. 2.3), even for the late Roman period when villa complexes are at their most extensive (Cunliffe 1993: 255; Johnston 1978; RCHME 1983; Cunliffe and Poole 2008a). 6.3.2 Burial practice in early Roman Hampshire For the majority of the Iron Age the visible form of burial practice in Hampshire, like much of central southern England, was the deposition of human body parts, whole corpses, articulated limbs or fragments of bone, usually with other material, in a variety of depositional contexts around settlements (Wait 1985; Whimster 1981; Wilson 1981). Recurring depositional associations occur between human bone, the part or whole remains of animals and other artefacts (Cunliffe 1992; Hill 1995), though taphonomic process prior to deposition differs between human bone and other material, at least in two key Hampshire examples, Danebury and Winnall (Madgwick 2008). Occasional evidence exists for discrete burial in the Early and Middle Iron Age, in particular the inhumation burials buried in quarry hollows outside the inner enclosure ditch at Suddern Farm, Over Wallop (Cunliffe and Poole 2000b: 152-70) and an MIA cremation burial from Owslebury. In the Late Iron Age deposition of complete rather than part bodies accounts for an increasing proportion of the mortuary evidence. Usually these are deposited within settlements but occasionally occur in graves in spatially demarcated cemeteries, the principal example being Owslebury (Wait 1985: 116; Whimster 1981: 191). The classification of some contexts of this date as a grave or other feature is problematic, as illustrated by the furnished burial of two women in pit 5 with multiple animal carcasses at Viables Farm, of likely 2nd or 1st century BC date. Late Iron Age cremation burials with extensive ceramic assemblages like those at Owslebury and Hurstbourne Tarrant were initially assumed to be outliers of the Aylesford Swarling tradition (Hawkes and Dunning 1930; Whimster 1981: 151-2). However it is difficult to identify a homogeneous regional burial tradition in central southern England compared to those to the Thames estuary or the Durotrigan rite. Instead a heterogeneous group of burials with large or unusual artefact assemblages is contemporary with the Aylesford and ‘Durotrigan’ traditions, for example the Owslebury ‘warrior’ inhumation burial of the late second century BC and, a little later, the cremation burials with ceramic assemblages at Owslebury and Hurstbourne Tarrant and a mirror burial from Latchmere Green near Silchester. This heterogeneity is more generally characteristic of coastal southern Britain in the eastern channel area (Hamilton 2007).

There is also a bias to site type in recent archaeological activity. The domination of Iron Age settlement archaeology by the study of hillforts has been mitigated by recent examination of a wider range of open and enclosed sites. Relatively extensive downland settlements, occupied over long periods, though not always continuously, from the Early Iron Age into the Roman period, bounded and / or subdivided at certain periods of their history by large ditched enclosures and thus highly archaeologically visible have attracted significant archaeological effort. Several have been very extensively examined on the line of the M3, at Owslebury and in the Danebury Environs programme. The distribution of villas again shows a similar bias to the chalk downland, though fewer have been recently examined, exceptions being Sparsholt, East Meon and sites excavated during the Danebury Environs programme (Johnston 1978: Fig. 22; 1981: 48; Cunliffe and Poole 2008a). Chalton remains the only excavated example of a village site from Hampshire (Cunliffe 1973). Scattered references to burials to the east of Andover have been classed as rural here, but evidence at the junction of the Winchester-Cirencester and SilchesterSalisbury roads indicates the likely existence of a roadside settlement (Jennings 2000: 129-30). The majority of burial evidence derives from the large downland enclosed sites, both in the LPRIA and the Roman period. Save for infant burials, lead coffins from Twyford and Bishopstoke and the stone coffins from

87

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Fig. 6.7 Cemeteries and burials at Owslebury (adapted from Collis 1970: 249, Fig. 1, b-d.) Left. 1st century BC and 1st century AD. Centre. 2nd century AD. Right. 3rd-4th century AD)

Figure 6.8a The Late Iron Age and early Roman cemetery, Owslebury (adapted from Collis 1977: 29, Fig. 5)

In the early Roman period, with the principal exception of infants, burials were distanced from the areas of the living, generally being placed on the margins of, or outside, rural settlements (see below). The deposition of body parts and skeletal fragments within settlements diminished or disappeared; where large deposits of Roman period faunal remains, with which fragmentary human material may be confused, have been examined, the practice of partial deposition seems to have finished around the time of the Roman conquest. For example at Old Down Farm (Andover), the Middle Iron Age (MIA) human remains assemblage comprises whole inhumations and skeletal fragments, but in the LIA and early Roman period only whole inhumations were recovered. The absence of fragmentary human bone from faunal samples from Neatham or Ructstalls Hill is further negative evidence against the continuing deposition of fragmentary remains within settlement sites in the Roman

period. The age profile of the human skeletal fragment assemblages within the settlement area at Winnall Down (Fasham 1985: 120) also suggests changes from the Iron Age to Roman period. In the MIA sample, fragments from adults and under one year olds were recovered from within the settlement area in a ratio of 7:5; in the Roman period this changes to 5:16. Nevertheless occasional evidence of adult body parts on settlements persists into the late Roman period, though it is not always clear whether this represents primary or secondary deposition. At Cowdery’s Down a skull fragment probably from an adult male, as well as two infant burials were recovered from the terminal of late Roman enclosure ditch 2. At Balksbury single large fragments of adult skulls were recovered from the late Roman fills of Gully 8 and pit 14, within the settlement area and close to another late Roman burial. Examples from Owslebury are referred to below. 88

DEATH IN THE COUNTRY

Figure 6.8b Burials on Site P at Owslebury, inhumations 22 and 23 of 3rd-4th century date (after Collis 1968: 24, Fig. 3)

Discrete burial became the early Roman norm. In the LPRIA and early Roman period individual inhumations are documented (Neatham, Old Down Farm (Andover), Thruxton, Woolbury) but cremations account for most burials. In the late Roman period inhumation appears to have predominated, although the date of change is hard to document as few of the Hampshire inhumations are closely dated (cf. 2.5). There is some local variability in the prevalence of cremation and inhumation, the latter being more common in the south and west of the county (Fig. 6.6). Roman period rural ‘cemeteries’ generally contain few burials. This may sometimes be attributed to disturbance or to incomplete excavation, but even in more extensive excavations numbers of burials are small, as is well illustrated by Danebury Environs excavations (e.g. Houghton Down, Woolbury, Thruxton, Grateley). The first century BC to second century AD cremation burials in an enclosure on the northern periphery of the Owslebury complex remains the best documented and largest rural burial group and discussion begins with it, but its typicality is not clear, especially for the early burials; the area south-east of Winchester has recently come to prominence through a concentration of unusual and prestigious finds of late Iron Age date including gold torcs and Roman denarii (Hill et al. 2004)

was recovered in varying quantities from all occupation periods, burials of adults being concentrated in cemetery 1 from the first century BC to second century AD, while burials of late Roman date seem to have been scattered across the site (Figs. 6.7 and 6.8). In the former period cremation is predominant (for adults), in the latter inhumation, although the dating of inhumation burials is problematic as they lack grave goods and are rarely stratigraphically related to other features. The date of several is often based on neighbouring features. Given the plough damage to the site it is unlikely that the complete sample of original burials survived to be excavated. Change in the number of burials appears to be a factor of changing preferences in the location of the dead; occupation evidence does not imply equivalent changes in population. Amongst the cremation burials the colour and condition of the cremated bone testify to efficient burning. Only small amounts of cremated bone seem to have been buried, although this is also affected by poor burial preservation. The only exception was double burial 1, where a much larger proportion of both cremated individuals was excavated; cremated sheep, pig and bird remains from this burial comprise the sole evidence of pyre ritual. Pyre debris was not deposited in graves.

The principal occupation of the rural settlement at Owslebury dated from the third century BC to fourth century AD. The site comprised several enclosures integrated into a system of trackways (Fig. 6.7), initially a single banjo enclosure, from the first century BC to the first century AD a ‘multiple defined entrance’ settlement with several trackways radiating from a central area and finally a mid- and late Roman phase in which the subdivision of the site by ditched enclosures and defined entrances was not so obvious (Collis 1970: 256). Three forms of burial practice were recorded, cremation, inhumation and the occasional deposition of unburnt skeletal fragments in settlement features. Burial evidence

The majority of cremations of all periods were unurned; the small number of urned burials dated to the first century AD or later. The most frequently occurring grave goods were ceramics. In most cases these were deposited intact but in one case (41) some must have been broken before deposition or placed in the grave as fragments. Plough damage had disturbed many assemblages. Using the maximum figures for individual grave groups, there was an average of 7.2 ceramics per grave, rising from 4.9 in the first century BC to 7.1 in the first century AD. The largest ceramic deposit, of at least 36 vessels in burial 1, dates to the second century AD. The earliest ceramics 89

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 6.9 The number of ceramic forms in burials at Owslebury

were in local fabrics. In the first half of the first century AD they were complemented by small numbers of imported vessels, Gallo-Belgic wares and from the mid first century occasional samian vessels (Collis 1990: 215). Samian comprised only 4% of all vessels, and five of the seven examples were recovered from one mid-first century AD grave. The soft fabric of most vessels in burial 1 may reflect their manufacture specifically for the grave. A shift in vessel form is visible between the first century BC and first century AD. Jars were the commonest form in the first century BC burials, while dishes and the majority of cups were recovered only from first century AD burials (Fig. 6.9). Flagons were rare in comparison to the urban assemblages from Winchester (5.2.1) and elsewhere. Most assemblages included a combination of dishes and drinking vessel with jar or bowl. There were larger ‘services’ accompanying burial 45, two flagons, two samian dishes and two samian cups, and burial 1 in which the vessel forms closely correspond to those identified in samian burial groups (8.2). Other grave goods were rare and mostly associated with the larger ceramic assemblages. Burial 1 contained several pieces of unburnt decorated bone inlay, possibly from a box (cf. 3.5.3) and burial 41 an iron razor, one or possibly two whetstones and a pig’s jaw. The three brooches with an adult male cremation burial (45) further undermine the hypothesis that the provision of three brooches necessarily indicates female costume of the ‘Menimane’ type (Wild 1985).

25 20 15 10 5 0 Fl.

Bkr

Bowl

Jar

D/Pl.

Cup Vessel

Figure 6.10 The orientation of burials at Owslebury (n = 32)

N NE E SE S SW W NW

From the first century BC to second century AD inhumation was mainly reserved for infants and occasionally older children, up to two years old (66). The most frequent orientation was north-south but preferences changed over time (Fig. 6.10). All burials with head to the east date from the third or second century BC to first century AD, but those with head to the west or north-west date to the second century AD and later. The position in which inhumation burials were laid was largely age dependent, infant burials being most commonly crouched or flexed and adults extended. Evidence of coffins was recovered from five graves, all of the second-third century AD or later and all with one exception of adults. With the exception of the weaponry from burial 39 (later second century BC) no grave goods were recovered with inhumations.

Otherwise the corpus of early Roman rural burials from Hampshire remains small. Other instances comprise individual or, occasionally, small groups of burials. There are only a very few unfurnished cremation or inhumation burials (e.g. Popley, Ructstalls Hill; Houghton Down; Woolbury; a secondary burial at Hurstbourne Tarrant; South Wonston) and a handful of burials with modest grave good assemblages including ceramics and ornaments (Popley, Old Down Farm (Andover); Old Down (East Meon); Ructstalls Hill; Finkley; Oakridge; Thruxton). The Thruxton burial, with a deep pit a few metres away containing substantial animal bone deposits, is a reminder that the pyre or grave goods may be misleading in terms of the material expended in ritual.

Collis (1977b) identified an increasing hierarchy in burial practice over time at Owslebury, which in the second or third century AD divided into ‘rich cremations’, coffined or extended inhumations and ‘irregular inhumations’ distinguished by their lack of coffin and burial in settlement features rather than formal cemeteries, but since it is difficult to date many inhumation burials closely, an alternative characterisation not contradicted by the burial data is of a phase of concentration of (adult) burials in cemetery 1 on the settlement periphery, followed by dispersal of the placing of the dead in the middle and late Roman periods.

The best known burials are those of the ‘East Hampshire’ or ‘Central Southern’ tradition identified by Millett (1986; 1987) and distributed across eastern Hampshire, Surrey and West Sussex. These are defined by large ceramic assemblages; Millett (1987) uses a minimum of ten or more vessels to distinguish the burials of this tradition). Millett (1990a: 23) considered the grouping analogous to the Welwyn / Lexden group of well furnished pre- and ‘cusp’ of conquest burial, although the earliest of the ‘East Hampshire’ assemblages is

90

DEATH IN THE COUNTRY

contemporary with the final phase of the latter. Examples span a date range from Claudian (Alton) to late second / early third century AD date (Daneshill). While characterized as a ‘rural’ burial tradition (Millett 1987), many occur in the cemeteries of Winchester and Chichester and in the immediate hinterland of the former (7.3). Elsewhere they are associated with minor centres (Neatham) and non-villa sites (Daneshill, Owslebury), some in cemeteries alongside other less well furnished burials (Haslemere, Owslebury), and others spatially separated (Neatham, Alton and Daneshill). The apparent high frequency of burials with large quantities of grave goods may be misleading since the presence of artefacts will have increased their chance of observation and recovery.

from the typical burial of Iron Age crouched inhumations in this region with the head placed to between north and east (Whimster 1981: 191-2). In this characteristic the rural burials differ from the Winchester cemeteries by the fourth century AD (5.3). Figure 6.11 The orientation of late Roman rural inhumation burials in Hampshire (n = 41)

N NE E SE S SW W NW

These are all cremation burials. The limited evidence of artefacts placed on the pyre includes burnt animal bone (Nun’s Walk, Winchester) and ornaments (Haslemere 1905.1). The commonest ceramics in grave good assemblages are the many cups and dishes, sometimes in samian (e.g. Sparsholt, Winchester Highcliffe, Alton 3), more commonly in local fabrics, which occasionally imitate samian forms (e.g. Winchester Grange Road 2, Owslebury 1). Other artefacts deposited include dining equipment and food or feasting debris (e.g. Winchester Grange Road 2, Winchester Milland (Portal Rd)), as well as personal ornaments and items related to the manipulation of personal appearance (e.g. Alton). The significance of assemblage composition is considered further in chapter 8 with reference to many examples from this group. In individual cases the depositional sequence shows some complexity, with some artefacts placed in layers above the burial itself (Alton 5) or split between several depositional contexts (e.g. Neatham burial 5). This may indicate extended burial rites. There is little direct evidence for above ground monuments, but the linear layout of burials at Alton and Neatham, and the general lack of intercutting suggests that graves were marked.

Figure 6.12 The number of late Roman rural burials with different artefact types, Hampshire 25 20 15 10 5 0 HN

6.3.3 Burial practice in late Roman rural Hampshire Inhumation was the principal burial practice of the third and fourth centuries AD, although a small number of cremation burials have also been documented (e.g. Owslebury). Singletons and very small groups predominate in all well-documented cases; the only samples where more than ten burials have been recovered are Portway Industrial Estate, Andover, and Brook. Most individuals were buried extended and supine, a small number crouched or flexed (e.g. Balksbury 75) and in one case placed in an irregular position to accommodate the body to the dimensions of the pit selected as the burial space (Grateley). Three decapitated burials have been recovered from rural sites (Choseley Farm, Cowdery’s Down and Andover Southern Distributor Road). The high degree of variability in burial orientation (Fig. 6.11) seems likely to be a product of the close relationships of burials to features in their immediate vicinity (see below). However a general avoidance of placing the head between north-east and south-east, and a preference for the north or north-north-west can be identified, a shift

Vessel Coin

Pers. Glass An. Equip. orn. Bone

Evidence for deposition in wooden coffins was recovered from c. one third of the reliably documented rural inhumations. To judge from the thickness of the timber stain and the number of nails, the coffins at Burntwood Farm were of particularly massive construction. Unusually elaborate iron fittings for a wooden coffin, not so far paralleled in a Romano-British cemetery, were documented for burial 70002 at Brook. A small number of lead and stone coffins have also been recorded (Appendix 6.7). Graves at Burntwood Farm were distinguished by exceptionally large grave cuts, sometimes over three metres in length, but otherwise graves were earth cut features of sufficient size for the coffin. One of the late Roman burials at the Southern Distributor Rd site in Andover was of a stepped form similar to examples from Winchester. By what means burials were marked is not usually clear but again the lack of intercutting and the careful layout of aligned burial groups suggest that most were. The limited evidence for burial monuments takes very diverse forms.

91

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

At Brook one probable cremation burial was surrounded by a square ditched enclosure, but this was not excavated. Burials at the Portway West industrial estate were marked by flint cairns. The clearest case of a funerary marker is at Thruxton, where a late 4th century AD fence enclosed an inhumation burial and associated shaft c. 250 years after the original interment; this is associated with the use of a single room within the adjacent aisled building as a cult space which may well be related

grave goods being absent for example in all save one of the late Roman graves from Owslebury, and from the poorly preserved sample at Thruxton (if of Roman date), but more common in other small well recorded groups (e.g. Middle Wallop; Snell’s Corner; Burntwood Farm; Odiham). 6.3.4 Burial and settlement space in rural Roman Hampshire It is from large enclosed non-villa sites that the majority of rural burial evidence derives in the LPRIA and the Roman period, these being the best attested settlement types in the county (Massey 2006: 23-4). Save for infant burials, lead coffins from Twyford and Bishopstoke and stone coffins from Soberton, the burial practice of villa occupants remains poorly documented. Substantial discrete cemeteries are so far rarely reported, the first century BC to second century AD enclosure at Owslebury remaining somewhat exceptional (Fig. 6.8a), though the partially examined mid-late Roman cemetery at Brook, with at least 60 burials, shows that other examples remain to be discovered. The majority of interments were recovered as single or small groups of burials in association with boundary features, usually the ditches and gullies which defined settlement and other enclosures, but also field boundaries and occasionally landscape features of greater antiquity. This association takes several forms.

The commonest grave goods, hobnails from footwear and ceramic vessels (Fig. 6.12), parallel those in contemporary cemeteries at Winchester (5.3). Over half the sample of 60 burials with sufficient available documentation contained grave goods, though only 11 more than a pair of hobnailed shoes or a vessel. Dress items are infrequently recorded, though a female burial from Easton contained a bracelet on the arm and a comb and wooden box near the skull. This pattern is also seen in the Brook cemetery, where the six late Roman burials excavated contained hobnails, two ceramics and one a bracelet. A few other burials, poorly recorded, include other items; a group of miniature vessels was placed in the Binsted stone coffin and four glass bottles in a lead coffin at Bishopstoke. The high percentage of burials accompanied by grave goods (53%) reflects the close dependence on grave goods for an attribution of date and cannot therefore be taken as representative. Comparison at the level of individual sites shows some variation,

Figure 6.13 The distribution of late Roman burials (squares) and human skeletal fragments (open triangles) at Balksbury (adapted from Wainwright and Davies 1995: Fig. 8, 8)

92

DEATH IN THE COUNTRY

Figure 6.14 The distribution of burials at Burntwood Farm, Site R6 (adapted from Fasham 1980: 42, Fig. 4)

Figure 6.15 The conquest period burial, associated pit and late Roman structures, Thruxton (after Cunliffe and Poole, 2008d: 39, fig. 4.23)

Burials were often placed close to or deposited within boundary ditches, whether of a settlement as at Balksbury (Fig. 6.14), Martin’s Down, Owslebury (Figs 6.7-8), Oakridge, Popley, Suddern Farm, Winnall Down or Woolbury or probable fields as at Burntwood Farm (Fig. 6.14). In this sample some burials were placed to the rear of sites (e.g. Winnall Down; Cowdery’s Down 2 and 3), but deposition in or close to entrances is as if not more

common. On site P at Owslebury burial 24 was deposited at the entrance to a pre-Roman enclosure and burials 22 and 23 lay just beyond the entrance to the first century AD enclosure (Fig. 6.8b). A child at Martin’s Down and two infants at Cowdery’s Down were deposited in ditch terminals at the entrance. Burials were also placed at entrances to internal sub-divisions of settlements. A cremation in a cist was deposited beneath the eastern 93

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

doorway of the possible villa at Finkley and two cremation burials were located between enclosure ditches 3 and 4 at Daneshill, possibly the entrance to a rear portion of the enclosure. A perhaps analogous location for burial is at junctions of other features, for example burial 54 at the junction of field boundary and trackway ditch at Burntwood Farm.

since the relative date of different features in the landscape palimpsest of surrounding landscapes could not have been established. Infant burials comprise the most frequent exception to burial on boundaries and account for the majority of burials from the interiors of sites from the LPRIA to the late Roman period (Houghton Down Owslebury, West Park, Rockbourne, Micheldever Wood, Sparsholt, Ructstalls Hill). In some cases they were grouped on internal boundaries (e.g. Owslebury). Only in joint burials are they more often deposited on site peripheries (e.g. Old Down Farm (Andover), Choseley Farm). At most sites the small numbers of burials impede the recognition of the age at which an individual would be accorded separate burial in the company of adults; at Owslebury it lay between 12 and 18 months.

The location of burials on site margins sometimes appears to be influenced by other features on their peripheries. An association of infant burials with ‘corn driers’, structures of which the purpose may have been both to dry grain and germinate it for malting (van der Veen 1989; Campbell, in Cunliffe 2008a: 69-70) has been noted by Scott (1990a; 1991), but adult burials have also been found in close proximity. Two adult inhumation burials were aligned on the north wall of a fourth century AD corn drier at Choseley Farm, although the graves cannot be closely dated. At Rockbourne the inhumation burial which cut the wall of the corn drier must have postdated use of the installation.

A strict interior/exterior distinction is not always observed: at Winnall Down for example the complete infant burials were found at greatest distance from the site, in two cases, 6289 and 8581 in the areas used for deposition of human remains in the MIA, as opposed to other non-infant burials placed on or in the ditch of the Roman period enclosure (Fasham 1985, 134-6). At Cowdery’s Down also infant burials were recovered from a ditch terminal at an enclosure entrance. Conversely adult burials occasionally intruded on settlement interiors, although it is often difficult to assess whether sites were still occupied at the time of burial. An adult inhumation was placed within the top of a well filling at London Road, Holybourne. Graves at Houghton Down and Balksbury (Fig. 6.13) appears to lie within the settlement area. Especially striking is the deposition of at least 24 adults and three children at various points in several episodes of a well infill sequence between the late third and seventh centuries at Oakridge, although little is known of the site context. The excavator suggested that the human burials and many animal carcasses represented the hasty disposal of plague victims, but the apparent deposition of bodies in different groups rather than in a single episode in the infill of the well counts against this. The associations with complete pots and animal carcasses, sometimes in large numbers, is reminiscent of the structured or placed deposits identified in Roman settlement contexts (3.4). At late Roman Grateley the awkward placing of the body within what seems to be a re-used storage pit, with fragmentary infant remains as well as an animal mandible and abundant iron fittings, perhaps deposited as part of a structure or frame, recalls local Iron Age structured deposits more than normative late Roman burial practice (cf. Hill 1995).

In some instances burials were placed outside enclosure ditches, the distance varying from immediate proximity (Oakridge, Popley, Ructstalls) to many tens or hundreds of metres (Old Down Farm (Andover), Cowdery’s Down). The importance of the distinction between burial and settlement space is well illustrated on site P at Owslebury where the regular form of the late Roman rectangular enclosure is interrupted to keep earlier burials outside the boundary ditch. At Burntwood Farm burials were located on a Roman period field boundary. The graves were cut in the fourth century, parallel to and between three and five metres north of Feature 8, a ditched boundary. A late or sub-Roman period line of post-holes (Line 2) was later established later along the line of the graves, sometimes cutting their fills, rather than the earlier boundary (Fig. 6.13). The ‘off-site’ area as a space for burial is likely to be under-represented, although burials from such areas have sometimes been discovered beyond the limits of formal excavation (e.g. Cowdery’s Down, Oakridge). Burials were also ‘distanced’ in time as well as in space by the selection for burial of features of earlier date, for example fills of Iron Age ditches (Suddern Farm, Old Down Farm (Andover), Micheldever Wood), earlier Roman period enclosure ditches (Balksbury, Owslebury, Martin’s Down, Oakridge, Winnall Down), although regrettably few reports record the stage during the formation of the fill at which the burials were placed. Features of greater antiquity were also sometimes chosen for burial, for example Neolithic and Bronze Age barrows at South Wonston, the Bronze Age boundary feature at Odiham, and long barrow at Giant’s Grave, Old Winchester Hill. At Snell’s Corner Iron Age, Roman and Anglo-Saxon burials were aligned on a Bronze Age round barrow. At Thruxton (fig. 6.15) the site of a midfirst century AD burial and associated enclosure again saw burial three to four centuries later. The distinction between features one or two centuries or one or two millennia old should not perhaps be over-emphasised,

6.3.5 Burial and society in rural Roman Hampshire Similar conclusions to those from Hertfordshire can be drawn from the Hampshire evidence, though as before sample size is small and it is not yet clear how far the forms of burial ritual discussed are representative of the population as a whole. The form of early Roman burial ceremonies is similar to that documented in Winchester at Hyde Street (5.2), but in general burial assemblages were much more substantial in a rural setting. Within the small 94

DEATH IN THE COUNTRY

sample there are very few instances with modest furnishing comparable to those from Winchester Hyde Street. There are limited parallels in the urban sample to the very large ceramic deposits among the burial ceramic assemblages, commonly in Winchester’s environs, sporadically around Neatham. Again the percentage of samian in some burials is much higher than that deposited in contemporary rural settlement contexts (6.2.5). Although lacking equivalent monuments to the Herts sample, again a comparison may be drawn between the limited burial evidence and the relatively undifferentiated character of rural housing, though the early Roman sample of the latter is limited (Cunliffe 2008a: 178-84; Massey 2006: 23-4).

century AD laying of the Bacchus mosaic inside the aisled hall on the settlement site was contemporary with the construction of the double fenced enclosure around the adjacent much earlier burial and the associated deposit. This provides a further instance of the making of material connections to the dead of centuries before in the negotiation of status (cf. 4.2.4). 6.4 Burial and rural settlement space in Roman Britain 6.4.1 The cemetery in rural Roman Britain The discussion of the Hertfordshire and Hampshire samples suggests that status difference is not expressed among rural burials in full or straightforward agreement with the models outlined at the beginning of the chapter, even in these regions where grave furnishing is relatively abundant and funerary monuments are used. The sample of rural burials in both cases is limited, being in part constrained by the character of fieldwork in both areas. In both cases nonetheless it is plausible to argue that for at least some of the Roman period only a minority of individuals were buried in an archaeologically visible manner assessed here, as for the Roman rural burial sample in general (2.5). The burials discussed here are unlikely to represent the exclusive and, perhaps, even the most common form of funerary ritual sometimes practised in the countryside.

For the late Roman period the percentage of burials with grave goods was higher than in the urban cemeteries, with the exception of Lankhills. Unlike Lankhills the possible signifiers of rank, such as cross-bow brooches and belt sets, or other more rarely occurring grave good types were lacking from the rural sample; the main evidence for material investment lay in grave scale or container for the body. The villas documented in the environs of Winchester (e.g. Sparsholt) and further afield in Hampshire (e.g. in the Test valley) reach their most monumental form until the later Roman period (Cunliffe 2008a: Johnson 1978; Massey 2006: 24-5). Given this increasing elaboration of settlement space, the significance of burial as an arena for negotiating status had arguably declined. One striking exception which integrates burial with the elaboration of villa complexes is at Thruxton, near Andover; the fourth

Figure 6.16 The distribution of burials in relation to principal Iron Age and Roman features at Frocester Court (adapted from Price, archive)

95

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure. 6.17 Roman period settlement and burials at Alington Avenue (after Davies et al. 2002: 57, Fig. 24)

Figure 6.18 Settlement and burials of all periods at Roughground Farm (square = inhumation burial) (adapted from Allen et al. 1993: Fig. 69, 98)

Differences in preservation and in the scale and intensity of excavation affect the numbers of burials recovered, but in many cases even of extensive excavation, scattered individual and small groups of burials closely associated with other features, tracks, ditches, houses, work sites, occur more commonly than larger ‘cemeteries’. The

Owslebury model is itself not a straightforward case as the differences between burials are as likely to represent variability over time as they are to reveal differences within a contemporary population. Differences between roughly contemporary burials only emerge when burials are studied across a wider region rather in individual 96

DEATH IN THE COUNTRY

groups, and this only intermittently (see further chapter 7.3). Within the spectrum of locations observed from isolated single burials to very extensive cemeteries, it is difficult to identify consistent status-related spatial difference in burial or to support the characterisation of deposition of in ‘convenient’ or ‘expedient’ settlement features as a low status marker. The Hampshire sample showed that burials in these ‘expedient’ boundary locations comprise the near totality of the known burial sample, especially for the late Roman period, and were as, if not more, likely to be placed in coffins and furnished with grave goods as contemporary burials in urban cemeteries. There was also evidence for general rules determining their orientation as well as likely influence by neighbouring features.

in similar spaces from other regions characterised in appendix 5.2. In general cremation and inhumation burials in such contexts share the rituals of those found in other cemeteries in the same regions, indicating no lesser ‘formality’ or investment of energy and effort in their rituals. What comes more clearly into focus in the case studies is the interest of spatial context, in particular two aspects. The first is the nature of the ‘cemetery’ itself; in the small sample so far examined the cemetery as a cluster of burials in a demarcated area accounts for only a part, and not obviously a majority part, of the burial corpus. The second is the variability within the association with settlement boundary features which deserves further exploration. In this section the existence of the ‘cemetery’ as an entity in a rural context is more widely assessed, using other data from southern and central Britain, in some cases from fieldwork projects on a larger scale than those assessed so far.

This pattern needs fuller substantiation with other samples of data, but impressionistically the same characteristics applies in many cases for the rural burials

Figure 6.19 Fourth century settlement at Catsgore, with burials added (adapted from Leech 1982: Fig. 5, 8)

Figure 6.20 Duckend Farm and Duckend Car Park sites, Stansted. The distribution of the principal clusters of cremation burials (boxed) in multi-period palimpsest (adapted from Havis and Brooks 2004: 263, Fig 169)

97

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 6.21 Mid-first to mid-second century features from Area B, Wavendon Gate (adapted from Williams et al. 1995: 26, Fig. 16)

Across this wider sample isolated and small groups of burials associated with settlement features are ubiquitous both in smaller-scale and more extensive excavations. Examples from across the province include the following Biglis, Bower Road (Sneeth), Cambourne Jeavons Lane, Catsgore (Fig. 6.18), Dalton Parlours, Dragonby, Eastrea, Kempston Church End (phase 2), Kempston Marsh Leys; Nash, Newport Pagnell, Parlington Hollins, Rotherley, Rudston, Snettisham Bypass, Tutt Hill, Wakerley, Winterton, and Woodcuts. Frocester Court (Fig. 6.16) provides a nice example of this phenomenon, where small clusters of burials occur across the site linked by similar rituals but neither ritual nor location is perpetuated over a long period of time. Many of these also have a close relationship to neighbouring features, for example the long line of burials along the boundary ditch at Alington Avenue (Fig. 6.17), or the burials that precede the fourth century phase at Poundbury. Others are partly or wholly demarcated by enclosures, some already in existence and others specifically intended to contain the space of the dead, for example at Roughground Farm (Figure 6.18). Examples of cemeteries of early Roman date with several tens of burials can be documented mainly from south-east Britain, including Bancroft (Fig. 6.22), Duxford, Godmanchester Rectory Farm, Mucking, Strood Hall, Sutton Valence, Westhampnett (Fig. 6.23), Guilden Morden or Litlington, as well as among ‘Durotrigan’ cemeteries in continuing use in the post-conquest period, for example Jordan Hill, Dorchester Fordington Bottom, Maiden Castle and Poundbury. Philpott and Reece (1993: 420-1) proposed that such more formally demarcated cemeteries became more numerous in the late Roman period and were perhaps a rural equivalent of the ‘managed’ cemeteries in towns and small towns (cf. 5.3.5). Those they cite, including Bradley Hill (Fig. 6.26), Lynch Farm (Fig. 6.27), Radley Barrow Hills, Stanton Harcourt, Wasperton, and Winterbourne Down, can now be complemented by many more examples from across

southern and central England, including the various Boscombe Down cemeteries, Bletsoe, Brettenham, Chignall, Kempston, Foxton, Friary Fields Dunstable, Godmanchester (Parks) Lynch Farm, Ruxox, Tolpuddle, Uffington and Watersmeet, as well as several examples in the middle Thames valley (Booth 2001; Booth et al. 2007: 226-7). With some exceptions (for example Queensford Farm, Cannington) their size is similar to those of the earlier period. The difference over time should not be exaggerated, since rural burials as a whole are generally better documented in the late Roman period (2.5). ‘Cemeteries’ or burial groupings of this larger size seem in general to have been commonly in use for limited time periods, rarely more than a century and often less in settlements often occupied for several centuries. This can be illustrated on some extensively excavated sites. At the Keston villa (Fig. 6.24-25) formal burial areas are only documented for part of the occupation of the site, with the change in burial location in the later second century being associated with a significant change in the layout of buildings on the site, though the dating evidence is too imprecise to assess how closely contemporaneous were the changes. Though not yet fully published, and with burial visibility partly compromised by poor preservation of inhumed bone, at Mucking the Roman period burials within enclosures mainly date to the early-mid imperial periods, though the settlement was in use into the fifth century AD and beyond. The individual enclosures are also used for periods of a century or so, with use recurring at a later stage in some cases. A cemetery was only documented for the late phase of occupation at Claydon Pike (Fig. 6.29). The punctuated visibility (or clustering) of burial over time at several other longoccupied sites, for example from Great Barford bypass and Stansted excavations from the 1990s and 2000s (Fig. 6.20), Tollgate and Wavendon Gate (Fig. 6.20) further 98

DEATH IN THE COUNTRY

illustrate this point. The agglomeration and dispersal of burial suggested for Owslebury, with a ‘cemetery’ documented from only one phase seems not uncommon.

third / fourth centuries AD, with occasional exceptions such as the Babraham Institute, Guilden Morden, Ely Prickwillow Road and Ruxox. The late fourth and early fifth centuries is also a common hiatus, though radiocarbon dates obtained for individual samples remind us of the possible length of time over which ‘late Roman’ cemeteries might be in use (2.2). Only in a few instances are very large rural cemeteries of several hundred burials analogous to early medieval cemeteries documented as serving as communal burial spaces for several settlements over several centuries. The use of some, such as Cannington, Cassington or Wasperton, continued into the post-Roman period (Booth 2001; Rahtz 1977; Rahtz et al. 2000: 422-5; O’Brien 1999).

Where the same space is used for burial over a longer period, sometimes several centuries, this is often punctuated by long periods of disuse, as already discussed in relation to the Hampshire and Hertfordshire samples (e.g. Snell’s Corner, Thruxton, Welwyn Hall). It is also documented at Westhampnett, perhaps Mill Hill Deal, and Overton Down where sites were used for burial in the Bronze Age, Roman period and in the latter two case also the Anglo-Saxon period. Over a shorter period of time at Bancroft, despite continuity of occupation of the site from the Iron Age to the post-Roman period, the use of individual cemeteries on the slope above the villa of late Iron Age / early Roman, mid-Roman and postRoman is short-lived and interrupted by what seem to be substantial ‘gaps’ in the burial record between the late first and late second centuries and between the mid third and the mid fourth centuries (Fig. 6.22).

In summary burials are not at all times and places grouped in clusters which might even represent the dead of the smallest social units, though preservation affects our ability directly to compare numbers between sites. Larger clusters of burials come into view at different periods regionally, with a qualified emphasis on the later Roman period as being more widely significant for their emergence. In general burial spaces are in use for relatively short periods, typically a century or less, often a briefer period than the farms and other buildings with which they are associated. They are closely associated with other characteristics, especially of landscapes enclosed by settlement and field boundaries, trackways and ditches, an association which is now further explored.

Some similarity of date in episodes of burial aggregation on a regional basis can be glimpsed, as far as can be assessed from unevenly available and imprecise dating evidence. For example continuity of use from preconquest cemeteries into the second half of the first century AD is rarely documented in south-east Britain, as illustrated for example at Stansted. Few cemeteries appear to continue in use from the second century to the

Figure 6.22 Settlement, cemetery and shrine at Bancroft, (A) 1st century BC to 1st century AD, (B) 1st to 2nd century AD, (C) mid 4thearly 5th century AD (after Williams and Zeepvat 1994: Fig. 5, facing p.6, Fig. 6, facing p.8)

99

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 6.23 Iron Age, Roman and Anglo-Saxon burials at Area 2, Westhampnett (after Fitzpatrick 1997a: Fig. 6)

(3.4) with animal skulls (e.g. Prittlewell Marshall’s Farm; 6.2.2. for Herts examples), sometimes as fragments where provenance is more difficult to assess (e.g. Kempston Marsh Leys, Newport Pagnell). In general little attention has been paid to fragmentary human bone from Roman period sites which is often recovered in post-excavation work on the faunal remains. Publication of such material is usually inadequate for further analysis, for instance to ascertain its possible derivation from disturbed graves or structured deposits (3.4). For example at Dragonby scattered human bone was derived from all (Iron Age and Roman) contexts from a minimum of fourteen individuals but no information other than phase was provided. Deposition of body parts and skeletal fragments on settlement is not as common as in the pre-Roman period, but deserves more systematic analysis (see also Esmonde Cleary 2000: 134-5).

6.4.2 Infants and other burials on settlement ‘interiors’ The age-based distinction for burials within or closer to occupied areas and those beyond these is well established though requires some nuance. The commonest burials excavated on site interiors were those of infants, half of those documented in Struck’s (1993c: 315) survey being within structures, close to corners and walls, and a high proportion within yards. Infant burials are also often associated with other deposits or features, especially animal burials and corn driers (E. Scott 1991). As at Owslebury other clusters of infant burials have been identified (e.g. Bradley Hill, Barton Court Farm, Frocester Court, Hambleden, Shiptonthorpe). Occasionally they are also to be found on settlement margins but they are generally absent from the more formally separated burial areas. The infant :: interior as adult :: exterior distinction can sometimes be demonstrated on the same site (Alington Avenue, Catsgore, Frocester Court, the first centuries BC and AD at Owslebury and Poundbury).

Complete burials of adults have occasionally been documented within buildings, for example an early second century AD cremation burial at Chignall. Wells too were the occasional repositories for whole cremated or inhumed individuals (Fishbourne, Hambleden), though rarely on as spectacular a scale as at Friary Field Dunstable. Often burial was the final deposit in a fill sequence (Ewell, Rudston; see also Baldock) when the nature of contemporary settlement is often unclear. This is one manifestation of the re-use of the interior of abandoned sites for burial. At Curbridge late Roman inhumation burials post-dated a Romano-British settlement of which they overlie the features. Fourth to seventh century burials also overlay the villa buildings at Kempston. At Figheldean scattered late Roman inhumations burials cut through Iron Age and earlier Roman period features. Inhumation burials at Syreford Mill overlay an area earlier used for quarry pits and

Adult burials have also occasionally been recorded on site interiors in several forms. Fragmentary skeletal material has been recovered from Roman as well as Iron Age non-funerary contexts. Most investigation of this phenomenon in the Roman period has been directed to obvious ‘ritual’ deposits such as the Walbrook skulls (Cotton 1996) as well as other well / pit contexts, mostly on urban sites, where circumstances of deposition and bone condition bear evidence of execution or postmortem ritual, for example the scalped heads at Folly Lane (Fulford 2001; Isserlin 1997). Instances of skull deposition can also be exemplified on rural sites, for example at Odell and Herriott’s Bridge, as well as other instances possibly in ‘structured’ or ‘placed deposits’

100

DEATH IN THE COUNTRY

rubbish dumping. However since burials could also be placed in areas from which habitation had shifted to another part of the site, for example at Gatcombe,

Gravelly Guy and Whitton, such evidence need not in isolation indicate abandonment of a settlement.

Figure 6.24 Burials and shafts (stippled) at Keston in relation to Period VI settlement, AD 200-300. WC = earlier west cemetery; NWC and NC are north-west and north cemeteries (adapted from Philp et al. 1999: 184, 191-5, Figs. 67, 69-70)

Figure 6.25 North cemetery at Keston AD 200-300 (Period VI) (black = monument foundations; grey = burials) (after Philp et al. 1999: 46, Fig. 21)

101

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

In the absence of other dating evidence a post-Roman date is also possible for burials on settlement interiors, as the re-use of Roman sites for post-Roman burial has been attested on several occasions. Percival’s list (1976: 217, n. 1) of burials on villa sites can be supplemented by further examples, including Bancroft, Barton Court Farm, Beadlam, Claydon Pike, Great Staughton, Kingscote, Kingsweston, Llantwit Major, Roughground Farm, Southwell, and Winterton. In the absence of radio-carbon dating the post-Roman date of many deposits is assumed rather than securely established. The mid- to late Saxon dating of burials at Frocester Court and Claydon Pike exemplifies the sometimes considerable time which had elapsed between settlement abandonment and burial. A further association to have emerged since Percival’s discussion is that between burials and contemporary or earlier temple sites in the south-west (Rahtz 1977; Watts and Leach 1996), although Esmonde Cleary (1989: 185) has challenged the particular significance of this within the general re-use of Roman period sites in later periods. Neither the number of examples nor the size of burial groups is yet equivalent to that from continental provinces (Marzano 2007: 218-22; Lewit 2003: 262-3; Percival 1992).

Woodcuts; Woodyates. External boundaries were more frequent locations for burial. Occasionally fragmentary human and animal skeletal material (Prittlewell Marshall’s Farm; North Shoebury) and more often burials were deposited in or along the inside of settlement enclosure boundaries: examples include Alington Avenue (second to fourth century); Eastrea; Frocester Court; Market Deeping; Poundbury; Wakerley; Woodcuts. In other cases field boundaries were used in similar fashion (e.g. Asham; Elm’s Farm). Sometimes the alignment of burials suggests the existence of archaeologically otherwise invisible boundaries, for instance at Eyewell Farm or Maiden Castle Road, Dorchester. Other groups of burials seem to have been deposited immediately beyond a settlement enclosure boundary (Alington Avenue (first century AD); Bower Road (Sneeth); Frocester Court; Hayton; Lynch Farm; Marshfield (late Roman); Odell (first century AD). Burials were usually aligned on these adjacent features, though study of the orientation of burials in late Roman Hampshire has been shown to be conditioned but not determined by them (6.3.3). Other examples can also be given, for instance Market Deeping where burials were aligned north-south or east-west rather than on the settlement boundary ditch, and at Frocester Court where the head was preferentially placed to the south-west in both adult and infant inhumation burials regardless of location.

The hypothesis of widespread violent slaughter which used to account for these burials is long discredited (Webster 1969: 233). Webster attributed the change to a decrease in the concern to bury at a distance from settlements as part of a breakdown of Roman values. However the practice need not be a post-Roman innovation but a continuity of the practice of placing the in dead in abandoned areas of settlement. Whether the occasional striking choice of location for such burials, for example on the Orpheus mosaic at Winterton, is the most visible manifestation of a more generally structured choice of burial place would repay further examination; continental examples show the practice often to exist side by side with continued occupation of the settlements concerned, a phenomenon as yet rare in Britain.

In many cases cemeteries themselves were partly or wholly bounded, often within enclosures that were part of wider landscape division. Boundaries sometimes coopted existing features (e.g. Ely Prickwillow Road, Kempsford Quarry, Wavendon Gate, Wasperton, Huntingdon Watersmeet), which sometimes dated from considerably earlier periods (see below). In other cases the cemetery was enclosed with a new ditch and probably a bank too, though direct evidence rarely survives; examples are widespread including Boscombe Down, Great Barford site 4, Each End Ash, Manston, Rectory Farm Godmanchester, Mucking, Stanton Low (Fig. 628), Strood Hall, Tollgate, and Woodyates. A very small number were enclosed within walls (see 6.4.4 below). Individual burials could also be surrounded by an enclosure ditch both within cemeteries and in isolation, for example at Claydon Pike, Dorchester Maiden Castle Road, Kempston Church End and Tollgate.

6.4.3 Burial and boundaries Although some adult as well as infant burials seem to have been placed within the inhabited spaces, to separate and / or distance the dead from the living is much more common. Distance from housing and areas of activity varies substantially, from several tens of metres (Bancroft mausoleum, Keston (circular tomb and associated burials), Laxton, Litlington, Tockington, Winterton) to sometimes several hundred metres (Boscombe Down, Radley Barrow Hills, Claydon Pike, Foxton). The separation was also achieved by boundary features, with which burials were commonly associated, though this association took various forms. Burials were occasionally placed to the immediate exterior of structures (Newhaven, Stonea) or in or by the internal boundary ditches between different parts of a settlement or the entrances that led from one area to another or in ditch terminals, as is exemplified at the following sites: Cambourne Jeavons Lane, Catsgore; Dragonby; Godmanchester A14/604 junction; Gravelly Guy, Old Sleaford; Parlington Hollins; Thurnscoe Billingley Drive;

The associations between burials and other features on site peripheries identified in the Hampshire sample (6.3.4) can be substantiated by further examples. The close juxtaposition of burial and corn drier may often owe more to a shared general location rather than deliberate association (e.g. Dorchester Fordington Bottom; Roughground Farm, perhaps Great Barford sites 4 and 8). In other cases it can be attributed to the spread of burial over abandoned areas when the corn drier had gone out of use. However in cases of stratigraphic connection or immediate proximity it is difficult to explain this as a product of chance. At Kemp Farm a well furnished cremation burial was deposited into a corn drier / oven and at Lambs Lea a possible cremation was found on the floor of the main flue. The most direct association is the 102

DEATH IN THE COUNTRY

charred skeleton placed head first into the stoke hole of a corn drier at Welton Wold. Elsewhere inhumation burials were cut into or deposited adjacent to corn driers when (Biglis; Broadwell Spring; Eyewell Farm; Figheldean, Plant’s Farm Maxey; Swaythorpe Farm; Wavendon Gate; Wollaston; Woodcuts). The association with corn driers may be subsumed within a wider association of burials with craft / industrial activities and taking various forms, including the deposition of an adult inhumation head first into a kiln flue (Little Chester) and of a child within a furnace chamber at Godmanchester (the Parks), the juxtaposition of burials and kiln sites (Friary Fields), and the re-use of disused kilns for burial (Crambeck, Wakerley). Other instances of the association between burials and kilns are noted by Swan (1984: 50) and (Jones 2003: 87), although examination of the microfiche to Swan’s (1984) report suggests that a very direct association characterises only a very few cases and most instances may reflect the placing of both kilns and burials in extra-mural areas.

Shiptonthorpe). Others have noted this association with reference to infant burials and in well infill deposits but it can be extended to adult cremation and inhumation burials (see above, 6.4.2). The most striking instance is the juxtaposition between stone coffin and massive shaft (F) containing the carcasses and articulated limbs of c. 70 animals at Keston. However the limited dating evidence for the burial, an inhumation packed around with plaster, makes it difficult to assess the possible contemporaneity of burial and shaft. An association with midden deposits can also sometimes be proposed, though truncation will usually have removed evidence. At Scole Dickleburgh seven cremation and one inhumation burials were deposited within and close to an 8m x 4m midden deposit on the site boundary. The midden sealed the timber lined inhumation burial of a juvenile and was contemporary with the cremations. Burials could be an integral part of boundary formation. Three LPRIA / early Roman cremation burials were deposited at the base of a lynchet prior to or during its formation at Asham. Late Roman animal and human burials and cremation debris pits were spread across a site on Roman period field boundaries at Maddington Farm, Shrewton. The human burials and pyre debris pits interleaved with phases of lynchet formation. In occasional cases burial seems to be closely associated spatially with other ritual activity, most clearly on the shrine site at Duxford, less certainly when it took place in adjacent enclosures, for example perhaps at Bancroft, Kempston Church End, Orton Longueville, Swaffham Prior, or Westhampnett.

It is beyond the scope of this chapter to explore wider depositional associations with burials, in particular ‘structured’ or ‘placed’ deposits in similar boundary positions on rural sites (Chadwick 2012) but this aspect would repay further attention. There is, for example, a recurring association in the placing of human and animal burials / part burials (e.g. Berwick Down; Friary Field Dunstable; Kempsford Quarry; London Stratford Market, Maddington Farm; North Shoebury; Prittlewell Marshall’s Farm; Roughground Farm; Syreford Mill;

Figure 6.26 Settlement and burials at Bradley Hill (infant cemetery in building 3) (adapted from Leech 1981: Fig. 2, 179)

103

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 6.27 Enclosures with individual burials and cemetery at Lynch Farm (adapted from Jones 1975: Figs 3 and 4, 97 and 99)

The preference for ‘backlands’ burial, i.e. burial adjacent to the rear and side boundaries of houses and yards is a common characteristic of burials on minor centre sites, well documented for example at Ilchester, Fenny Stratford or Hibaldstow (Burnham and Wacher 1990: 279-81; Esmonde Cleary 2000: 129). It is a commonplace that rural Roman burials were located in similar positions (Leech 1982; Philpott and Reece 1993; R. F. Smith 1987). However the ‘front’ and ‘rear’ of enclosures is not always easily established, even in large scale excavations, where it can be confused by multi-period occupation and the presence of contemporaneous multiple entrances, as for example at Owslebury. While an arrangement of this type is documented at some extensively excavated rural settlements (Alington Avenue, Frocester Court, Herriott’s Bridge) it is not universal and ‘front’ and ‘rear’ are in any case difficult to establish in settlements with multiple enclosures and lacking a monumentalised architectural layout (6.3.4). The location of burials at entrances, either to sites or between the constituent parts of larger settlements has been noted in Hampshire and occasionally elsewhere. At Longthorpe a cremation burial was recovered from the northern antenna ditch leading into Yard I, although other burials were located at the rear of the enclosure. At Woodcuts inhumation burials were deposited in the ditches in the south-eastern approach to the settlement and a mound immediately outside a possible entrance to the west had a secondary cremation of Roman date. At Worksop Raymoth Lane too adults were buried in a ditch terminal at the entrance to the site. A relationship between the placing of burial and trackways has been identified by others (e.g. Millett 1995b: 121-31; Philpott and Reece 1993: 421) and can be illustrated in many instances, for example Boscombe Down, Bourton Grounds, Claydon Pike, Each End Ash, Great Barford, Kempsford Quarry, Manston, Radley Barrow Hills (probable), Roden Downs, Saltwood Tunnel, Stanton Harcourt, Totterdown Lane Horcott, West Thurrock, Winterbourne Down, and Wyke Regis. The linear arrangement of inhumation burials at Radley Barrow Hills suggests the existence of a trackway of which not trace had otherwise survived. The siting of

funerary monuments in relation to roads is discussed below. At times more isolated locations were chosen. Cave sites sometimes served as burial places, although given evidence for the other uses to which they were put they should not necessarily be classified as ‘isolated’ (Branigan and Dearne 1992; Philpott 2006: 79). Other examples of burials found away from permanent settlement include a late Roman double inhumation burial recovered on the edge of salt-water tidal flats at Hartlepool (although the charcoal deposits in the fill could be a product of ritual or of nearby agricultural or industrial activity), an isolated burial in a stone coffin recovered on a fen island at Stuntney, a single burial in a timber coffin associated with Bronze and Iron Age enclosure boundaries at Eastrea, and the two burials found within an area of seasonal pasture at Nash on the Gwent levels. Isolated burials in such positions of possible Roman date but lacking direct dating evidence, especially in areas of Britain with long traditions of unaccompanied inhumation burial (2.3) may well be under-reported. Other features were also selected for burial within the wider landscape. The association between Roman burials and prehistoric sites noted in the Hampshire and Hertfordshire samples is widely attested. Since Williams’ survey (1998a) of the re-use of prehistoric barrows further examples have accumulated (e.g. Ardleigh, Radley Barrow Hills, Saltwood; Tutt Hill; Uffington). Dating evidence for many instances is poor and a postRoman date often cannot be excluded when the use of prehistoric sites is much more widely attested (Williams 1998b). Other earlier sites were also used. For example at West Thurrock inhumation and cremation burials of 1st century AD date were buried along a length of Bronze Age / Early Iron ditch, itself adjacent and parallel to trackway ditches contemporary with the burials. Similarly at Enderby six inhumation burials were placed near or in long parallel ditches created as boundaries in the mid-late Iron Age and at Beechbrook Wood, Hothfield, five furnished cremation graves of mid first century AD date 104

DEATH IN THE COUNTRY

were placed close to the entrance of an enclosure created in the Middle Iron Age and probably fallen into disuse by the time of burial. At Ferrybridge a small number of burials of probable Roman date were buried in the upper fills of pits in a Late Iron Age pit alignment. In a very unusual sequence at Duxford a shrine site originating in the middle Iron Age on a knoll overlooking the Granta was also used as a cemetery from c. 100 BC to AD 200.

The choice of such sites may prompt an association between contemporary occupation and an ancestral past (Williams 1998b); the practical influence as surviving and substantial landscape features, visually prominent and conditioning the distribution of activity, may be equally important.

Figure 6.28 Late Roman settlement and burials (circles = cremations) at Stanton Low (adapted from Woodfield and Johnson 1989: Figs. 4-5, 140-41)

Figure. 6.29 Late Roman cemetery at Claydon Pike, setting (left) and cemetery by trackway (right, burials in grey) (adapted from Miles 2007: 170, 185, Figs 6.1 and 6.13)

105

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

6.4.4 The context of rural funerary monuments Through their association with various contemporary and relict features of the settled landscape the place of burials was indirectly marked. Of the likely modest markers specifically associated with many burials only occasional evidence survives for post holes (e.g. Great Barford Bypass site 8, Beechbrook Wood, Hothfield; Radley). A very small number were marked by more substantial monuments. In comparison to other provinces the range and quantity of funerary monuments from central and southern Britain is limited (cf. Graen 2008), and in many cases only the foundations have survived. The robbing out of masonry-built monuments may inflate the seeming popularity of the barrow as commemorative form, though these too have suffered from being robbed and ploughed out and in some cases only indirect evidence indicates their original existence, for example the disposition of possible satellite burials of 2nd-3rd century date at the Babraham Institute site, or the lack of intercutting of the richly furnished graves at the Duckend Car Park (DCS) site, Stansted by later Roman settlement features (Fig. 6.20).

line between these and the enclosure ditches which sometimes demarcate smaller and larger groups of burials within cemeteries (see above). Other features with which burials were associated, whether settlement boundaries or prehistoric monuments, could also have served to mark burials. Poverty of evidence makes the duration of monument use difficult to assess, but it seems to be similar to that of other cemeteries discussed above, i.e. rarely for more than a century. Where several monuments are sited together they seem to have been created over relatively short periods, in the best dated case at Bartlow Hills over several decades. The sequence excavated on the slope above the Bancroft villa shows the mausoleum to have stood for perhaps a maximum of 150 years. The burial ground around the Keston barrow was used for a similar period. Some sequences of subsequent burial or other activity are longer however. The continued performance of commemorative ritual in the enclosure associated with burials at Brisley Farm for up to two centuries after their creation in the early first century AD seems atypical in its duration.

Funerary inscriptions from a rural context are rare and are only occasionally associated with their monumental context (e.g. Keston, Stanwick). Funerary sculpture too is uncommon, the decontextualised fragments from Stanwick and Shakenoak (two animal sculptures, perhaps lions, plus other stonework, possibly derived from a mausoleum) or the Welwyn Hall sarcophagus being among the few exceptions (see chapter 7). With the exception of tumuli, monument form must usually be inferred from foundations alone. The lack of immediately associated burials renders the funerary purpose of several buildings uncertain; the difficulties of identifying the possible late Roman mausolea at Chedworth or Wells illustrates the problems well, albeit in particular circumstances. The handful of ‘temple-tombs’ represent the most architecturally complex monuments but with the exception of Lullingstone none has survived higher than foundation level; foundations at Langley hint at the existence of rural monuments analogous to the tower or pillar tombs of parts of Gaul and the Rhineland (Landes 2002; Moretti and Tardy 2006).

Some monuments were set within larger cemeteries, although lack of fieldwork in their environs frustrates assessment of this. For example individual or small numbers of secondary burials have sometimes been noted in and around some barrows (e.g. Keston, Holborough, Riseholme). The probable barrow at the Babraham Institute site seems to have been at the beginning of a longer sequence of burial enduring from the first to the fourth centuries AD. Antiquarian and more recent reports of skeletons excavated around barrows at Bartlow Hills and Limlow Hill may suggest that monuments could sometimes be the focus for larger cemeteries, but in both cases too small an area was excavated to assess this fully and the date of these burials is unknown and need not be Roman. At Swaffham Gallows Hill, for example, inhumation burials of Anglo-Saxon date were associated with the enclosure. In other cases monumentalised burials seem to have been separate from others. Two very large second century burial assemblages (25 and 26) buried separately from the rest on the Duckend Car Park (DCS) site at Stansted were perhaps marked by barrows (Fig. 6.20). The Southfleet cemetery, near Springhead, rediscovered by geophysical survey, and the rich conquest period burials of Tollgate (sites D and L) are also separated within large enclosures from other burials. In the former case the contrast is striking with the nearby cemetery at Pepper Hill, Springhead, crammed with modestly furnished burials which intensively re-use the same space.

Other masonry-built mausolea are few in number and of relatively simple form (Jessup 1959; Rodwell 2001: 4547). Barrows, generally of turf and timber constructions occasionally with a masonry-built central chamber, are more common. With occasional exceptions (Keston, Mersea Island, perhaps the Bartlow Hills) they lack the monumental retaining walls, entrance passages and sculptural embellishment of some examples from Belgium and the Rhineland (e.g. Soupart et al. 2008; Wigg 1993a; 1993c). The most frequently attested, albeit somewhat neglected marker is the enclosure, occasionally taking the form of a stone wall but more commonly that of an enclosing ditch, square, rectangular, round and penannular. Examples include the following: Ben Bridge; Bokerley Dyke; Boscombe Down; Kempston Church End, associated with a masonry base for a possible tomb; Maiden Castle Road; Ruxox. There is no clear dividing

Burial monuments were occasionally located in close proximity to buildings, as is best attested at Lullingstone. Roman Farm, Pitney, may be a similar instance but the context recorded in RIB for epitaphs RIB 182 and 183 in the courtyard of the villa is unlikely to be correct. Others were located at a greater distance, tens or hundreds of metres away from the settlement focus, for example 30m 106

DEATH IN THE COUNTRY

at Keston, 250m at Roughground Farm and a kilometre or more in the case of Swaffham Gallows Hill and Overton Down. Like rural burials in general, boundary features sometimes separated monuments from the space of the living (e.g. Marshfield Farm). Only occasionally, as is also attested in other provinces, can the siting of monumental tombs be suggested to have been deliberately integrated within the layout of villas (e.g. Bancroft; Lullingstone; cf. Graen 2008: 191-3). Rodwell and Rodwell (1985: 32-3, 48) have argued that a funerary monument was set up within the ‘prospect’ of the Rivenhall villa on an axis bisecting the villa compound, but evidence of this monument’s existence is suggestive rather than conclusive.

However in all the cases cited above the assessment is impressionistic; the visibility of funerary and other elements of contemporary landscapes needs fuller testing; the principal example of intervisibility so far tested by GIS viewshed analysis reveals a surprising lack of prominence in a wider landscape for the Bartlow Hills, their scale only being potentially appreciated by those passing close to them (Eckardt et al. 2009a). 6.4.5 An alternative model for the relationship of burial to settlement space The previous sections (6.4.2-6.4.4) have identified recurring characteristics in the placing of burials which are similar to and expand upon those noted for the smaller case study areas (6.2.4. 6.3.4). Evidence of burials regularly occurs set within in a palimpsest of archaeological features related to the subdivision of rural landscapes into fields, paddocks, trackways and settlements. Specific chronological relationships between burial and associated settlements may be difficult to establish in every case, especially where burials lack dating evidence. Nonetheless associations between burials and settlement features can be established common to ‘cemeteries’, dispersed single or small groups of burials and to some extent also to funerary monuments. Given the general separation of the living and (the adult) dead in a Roman setting, it is not surprising that the boundary feature has emerged as the recurring context of burial. This location is a classic illustration of liminality (van Gennep 1960), as has been previously suggested (Hingley 1990a). The ambiguous object, the corpse or cremated bone, was deposited at interfaces between spaces of different character, the enclosure ditches between different parts of settlements or between the inhabited area and fields, at entrances and routeways on the point of transition to the outside world, and in some cases in association with abandoned and prehistoric sites. A front / rear distinction in the placing of burial does not appear to characterise this distribution pattern so well as a concentric ordering (cf. Hingley 1990a: 143). The innermost ‘ring’ comprised primarily infant burials, within or close to buildings and the interior of settlement space, the outer ‘rings’ child and adult burials on the settlement periphery and at more distant locations, quite often associated with trackways in the case of both graves and monuments. Burials were also separated from the living by boundaries, by distance, and perhaps conceptually also in time as well as in space by the location of burial on abandoned sites of Roman or prehistoric date. Burials occasionally appear to have been the final act in a sequence of deposits like other ‘rites of termination’ identified by Merrifield (1987: 48-50), for example in the tops of wells or ditch fills (see above).

Monuments were sometimes sited with their visibility in the wider landscape in mind, as discussion of the Hertfordshire sample indicated and has been stressed in previous discussion (Jessup 1959). It is a commonplace of literature on burial monuments in Britain, Gaul and Germany that close relationships to roads determined their location (Graen 2008; Krier and Henrich 2011; Martin-Kilcher 1993a; Roth-Congès 1993), although recent re-examination of barrows in north-eastern Gallia Belgica has suggested that the significance of this relationship may have been exaggerated (Wigg 1993a: 379). There are undoubtedly examples in Britain of monuments within clear view of roads, for example the walled cemetery at Southfleet above Watling Street, or directly by them, for example a burial enclosure directly adjacent to Ermin Street at Field’s Farm, Duntisbourne Abbot’s, the barrows at Emmanuel Knoll (Godmanchester), or the Overton Down palisaded mounds close to the Bath to Mildenhall road, but such a relationship has not been directly established in many cases. Other monuments were situated several hundred metres away but still visible from routes; for example the mausoleum at Bancroft has been claimed as visible from Watling Street 2.5 km to the south-east. The location of the latter just beneath the brow of the hill may be to take account of the earlier enclosure, as the excavators suggest, and is also a common topographical position for Romano-Celtic temples (Lewis 1965: 130-31). The low hill on which the temple-mausoleum complex was sited at Swaffham Prior made it visible not only from the villa at Reach, to which it was linked by a trackway, but also in the wider landscape. Similarly the Holborough barrow was situated on the final crest of a spur overlooking the Medway valley, ‘commanding a fine view of the surrounding countryside’ (Jessup 1954: 2) and the Keston tombs also have an extensive view to the south over the villa with which they are associated and beyond. The position of some of the prehistoric monuments re-used for burial in the Roman period also lent itself to visibility, the Neolithic long barrow at Uffington being a striking example. Other aspects of visibility also require further consideration; Rodwell (1978: 19) for example considers only the landward relations of villa, mausoleum and barrow on Mersea Island, but the visibility of the monuments from the sea may have been equally important (cf. Lafon 2002).

However, as ethnographic case studies collected by Bloch and Parry (1982b) show in other contexts, the relationship of burials to the spaces of the living in space and time contained ambivalence; as well as separating the dead and the living, the associations to other features in the landscape also brought them within the daily experience of the communities. In this regard the general association of the dead with boundary features also has various 107

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

potential specific significances in a Roman context. Roman rural burials have in general been interpreted as markers of property or estate boundaries, since the texts of the Roman land surveyors sometimes discuss the relationship between tomb and boundaries. According to these the placing of tombs on the edges of properties caused their frequent confusion with other property boundary markers and the purpose of advertising ownership is explicitly acknowledged in these references, with burials able to serve as evidence of ownership in legal disputes (Keppie 1983: 126-7; Meffre 1993). This has sometimes been applied to British examples (e.g. Fasham 1980; Miles 1985: 40). However this does not mean that all tombs should be so specifically interpreted. Leaving aside issues of the applicability of Roman law in provincial settings, given the high degree of uncertainty over the nature of land and property ownership in the north-western provinces and the near impossibility of identifying estates, the hypothesis is difficult to pursue (cf. Gregson 1987). The context of burials identified here is in any case primarily settlement rather than estate boundaries, although burials at a greater distance from settlements are almost certainly under-represented.

different placing of human remains within Iron Age settlements (Cunliffe 1992; Hill 1995). The cutting and periodic clearance and re-cutting of these ditches that bound settlements, field and trackways can also be argued to have a symbolic as well as a practical dimension, ‘inscribing’ a group’s identity onto a landscape (Chadwick 2012: 301-2). The placing of burials is also clearly connected to this longer process, even if in the absence of stratigraphic connections the precise chronological relationships between burial and boundary development can be difficult to reconstruct. The discussion of the wider corpus has also established that in general, where evidence is available, the burial groupings and monuments are generally short-lived, even in the case of the larger cemeteries of several tens of burials (6.4.1-2). This is, to some extent, in contrast with the relative permanence of houses, though this impression may depend too closely on villa sites to be representative. To inhabitants of a settlement the presence of ancestral graves was both a reminder of and assertion of group identity in settlement space. The commonly occurring linear arrangement of burials had the potential to embody the shared descent group with reference to which the living are typically hypothesised to define their position (e.g. as articulated by Hingley 1989). A longer connection to a landscape was perhaps activated and reiterated by the placing of burials in features related to a near or more distant past, as strikingly at Thruxton where as at other places a memory must have persisted of features several hundred years old. In other contexts the re-use of much older monuments has been interpreted as a physical manifestation of a fictive genealogy, but the trend in Roman Britain is more diffuse and less frequently attested than the seemingly more systematic and recurring appropriation of prehistoric monuments in other contexts in Britain and beyond (e.g. Alcock 1991; Hingley 1996b).

Textual evidence also suggests another and perhaps more fruitful approach, which may be more profitably exploited than relationships to landholding and is likely to be relevant below the elite level. The designation of a grave as a locus religiosus protected the grave from destruction and was related to the right of descendants to access to ancestral graves in land that might no longer be their own (Robinson 1975: 177-79; de Visscher 1963). Literary evidence for elite self-definition through graves as a repository of ancestors on family estates has been documented by Bodel (1997). Whatever the legal setting, the placing of tombs in the contexts described above potentially guaranteed ready access to ancestors. The relationship was through, rather than to, land or property. It is impossible to discover the specific significance of these places for contemporaries, but the location of burial asserted itself on the daily visual experience of settlement inhabitants not only by the grave markers but also and perhaps more so, by the obviousness of the larger features, like boundary ditches with which burials were associated. Ethnographic evidence suggests a possible context for burials in similar contexts, an informal focus point of social or working activity, usually taken for granted and periodically remembered in a more ritualised fashion (e.g. Goody 1961). The locations chosen for burial as described above also potentially integrated burial into the recurring rhythms of farming communities. The association of burials with other characteristics of activity in boundary settings, processing and production (corn driers, kilns etc), other local rituals, such as the burial of animals within placed deposits, perhaps inserted the burial of the dead within the cycles of transformation of the agricultural year and its associated calendar. Placing of burials in proximity to ditches, at least part of the time filled with water, may have evoked metaphors of mortality and moisture, decay and regeneration. Such associations at least may echo those of the somewhat

Overall an alternative pragmatism can be proposed that casts the burials discussed here not as the placing of low status individuals in ‘convenient’ features to save the labour of excavation or the loss of working land; rather their location was associated with the dynamic form taken by settlements with which the lives and identities of the dead and the burying community were closely associated. This is a generalised account of the placing of the dead in a rural setting but it does not apply equally in all times and places. Its historically specific dimensions require discussion, even if it is difficult confidently to identify associated phenomena. To some extent the dead as placed on settlement margins become more archaeologically visible in the later Roman period, within a very long-term process from the late Iron Age onwards (2.4). The greater frequency of their clustering in larger groups in this period has also been noted, with the caveat that this is regionally variable and in part represents the greater abundance of excavated graves of third to early fifth century date. What this replaces is not clear. It may in 108

DEATH IN THE COUNTRY

part represent a change in burial ritual, but while evidence exists for the continued deposition of fragmentary skeletal material on Roman period settlements, it is much less commonly attested than in the Iron Age. The ‘emergence’ of burial seems unlikely to have a direct association with abandonment of the excarnation process which may account for some of this fragmentary material (Carr and Knüsel 1997; Madgwick 2008). It may also indicate a shift in the preferred context of burial closer to settlement sites, with a consequently greater likelihood that they might be archaeologically visible and liable to excavation in association with settlement complexes. Demographic change has been argued not to be a significant contributory factor (2.4-5).

well beyond the small minority resident at villas. If burial in the contexts described here can be said to manifest the claimed connections between burying group, its history and a particular landscape setting, then the greater frequency of visible burial in the fourth century AD manifests an intensification of such claims where rural society responded to these external pressures. However in view of the regional diversity of visible burial horizons, the weak chronological resolution of this process and the continued general short duration of burial clusters this can be proposed only tentatively. 6.5 Conclusion The first parts of this chapter demonstrated the localised variation in the availability of burial evidence from Hampshire and Hertfordshire, to complement the regional diversity outlined in chapter 2. The study of both sample areas revealed that although rural burial practices were generally similar to those of towns and minor centres in the same region, the variability in the early Roman period appeared to be wider within relatively small data samples (remembering too that these may not represent all, or even the majority of, the population). Taking into account too aspects of context, for example the availability of the pottery deposited as grave goods, burial can be argued to serve as a more significant arena for display and as a medium for differentiation than in the towns or minor centres in the early Roman period. Late Roman burials continued to show significant differentiation, but its significance is perhaps diminished when the diversity of contemporary settlement architecture of the late Roman period is allowed for. While burial practice may connect to wider socio-economic changes it is difficult to find in it the increased social stratification which housing indicates.

In other archaeological contexts the placing of burials closer to settlements and within formal bounded cemeteries has been interpreted as indicating a strategy to legitimise access to critical resources (Chapman 1981; Goldstein 1981). This connection has however been more often asserted than demonstrated, and often leaves unclear the question of what constitute critical resources (Morris 1991), although agricultural land is often implied, and population pressure with accompanying economic change considered the main motor. Contextual evidence shows burials almost always to be associated with landscapes extensively and closely subdivided by settlement and field boundaries, droves and trackways, a characteristic of much of southern, eastern and central Britain of the later Iron Age and Roman periods (Taylor 2007). However while in individual cases the shifts in spaces of burial and of occupation seem closely associated (e.g. Bancroft villa), visible burial practice does not in general develop closely in tandem with enclosure of this type; rather burial gradually becomes more archaeologically visible over time in a landscape which was long ‘plotted and pieced’ (Fowler 2000). Nor do the major changes in settlement landscapes now increasingly identified at a regional level correlate closely with changes in burial practice; for example in the Upper Thames Valley the early second century sees a substantial discontinuity in settlement occupation but a significant expansion in the numbers of documented burials and cemeteries is not seen for a further two centuries (Booth et al. 2007).

Some caveats to this characterisation are needed. This differentiation is currently documented on a regional rather than site basis; the next chapter will show that the accentuation of difference through burial ritual is likely to be dependent on other factors, in particular political geography and communications. It is also geographically limited, since outside south-east England markers of difference such as larger assemblages or monuments are rarer (cf. Struck 2000; see 6.4.4). In particular the sitespecific models offered by Martin-Kilcher and others in this sample too do not convincingly account for the variability within the evidence, especially the claims for status difference based on the immediate setting of burial. The reductive character of these simple models of social differentiation means that the information of setting has not been fully exploited for its insights into the organisation and conceptualisation of space. Many if not the majority of excavated burials derive not from formal cemeteries, but from groups or individual burials dispersed across settlement sites. The previous characterisation of burials in close association with, often ‘re-using’ settlement features as those of ‘low status’ individuals was considered within this broader reevaluation of current hypotheses for the expression of social hierarchies in rural burial sites. It has been rejected as difficult to substantiate in any given case and

This later Roman horizon of visible burial for some of central and southern Britain coincides in very broad terms with significant socio-economic changes. In Millett’s influential formulation (1990a) the late Roman countryside saw increased investment on the part of elites in the fabric of their villas, which may well have taken on part of the role of urban public spaces, and on the extraction of a greater agricultural surplus, as evidence for example in a new phase of agricultural innovation. With this came a new emphasis on social hierarchy as greater demands were made on the rural workforce (see also Hingley 1989; Scott 1990b; Scott 1994; 1995). Changes in the rural economy and in use or control of land had likely consequences for the organisation of agricultural production for households and communities 109

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

reductionist from the point of view of both theory and data. From an examination of the relationship of burial to settlement in detail in these two counties and in a wider sample, the definition of the ‘rural cemetery’ was problematized; to a not inconsequential degree the ‘cemetery’ is the exception rather than the rule, many burials being placed in close association with settlement features rather than in discrete cemeteries. A number of recurring locations have been identified in the distribution of individual, small groups of burials and larger cemeteries and monuments in relation to settlement features, primarily in association with, and beyond, settlement boundary features. The relatively rapid shift in the space of burial, with few locations seemingly used for more than a century, whether or not the graves were marked by a monument, has also been noted. It seems that the placing and use of cemeteries relates closely to the changing configuration of settlement space, as new spaces are enclosed, built over or abandoned so the placing of the dead also seems to shift. It is difficult to evaluate the strength of the different associations, but previous emphasis on the rear of settlement enclosures, in situations analogous to those documented in minor centres, has been exaggerated. An alternative characterisation has been outlined which interprets burial ritual and location as part of the dynamic creation of social relations through space; this relates to an expansion in the social archaeology of rural settlements which argues that not only houses but also other spaces with which burials are particularly associated, especially enclosure ditches and associated characteristics, need to be built into understanding of how space constitutes social relations (Chadwick 2008; Taylor 2001). In this context a suggestion for contextualizing the increased visibility of burials on settlement margins to broader socio-economic processes in the late Roman countryside has been very tentatively offered. Quantitative examination of rural burial practice from the sample areas will be used in the next chapter together with the urban and small town evidence discussed in chapters 4 and 5 to explore the relative importance of different contexts, urban, minor centre and rural as settings for display in burial rituals and monumental tombs.

110

CHAPTER 7: THE GEOGRAPHY OF FUNERARY DISPLAY 7.1 Introduction Previous chapters explored the character of burial display through its form and setting in individual urban, minor centre (chapters 4 and 5) and rural contexts (chapter 6). This chapter builds on these to explore the distribution of selected indicators of burial display, monumental and depositional across these different contexts, in order to establish the space in which display took place.

Belgae, and St Albans and environs, extending out to the wider civitas of the Catuvellauni, drawing on the evidence assessed in previous chapters; it also puts the results in a wider context from Britain. The significance of the results obtained from British evidence is also assessed through brief comparison with neighbouring provinces. 7.2 Burials and the urban periphery

Current orthodoxy holds that rural cemeteries were the focus of funerary activity on the part of elites, expressed either through abundant grave furnishing or through the raising of funerary monuments (Esmonde Cleary 1992: 38; Collis 1990; Haselgrove 1989: 11; Jones 1983: 81; Mattingly 2006: 478; Millett 1987; 1990a: 23; Philpott 1991; Rodwell 1978; Struck 2000). There is some divergence of view over minor centres, Philpott (1991: 218-19) arguing that their cemeteries, especially smaller burial plots, are also privileged contexts of burial display, while Struck (2000: 93-94) has argued that monuments and other ‘status’ indicators, including dining equipment, especially metal vessels, were more likely to be found in the cemeteries of villas and forts than those of towns or minor centres.

7.2.1 The creation and monumentalisation of urban cemeteries in Roman Britain The placing of burials distant from the settlement area and beyond the pomerium was an explicit aspect of Roman town foundation. As seen at Verulamium and Venta Belgarum and beyond the separation of burial and settlement areas was a ubiquitous element of town planning in Britain (Esmonde Cleary 1987; Wacher 1995), seemingly reflecting the requirement, as explicitly ordained in urban foundation charters, for burial and cremation to take place at a minimum distance from the pomerium (Robinson 1975; Rykwert 1976). This also converges with pre-conquest trends in Britain. The spatial distinction between the deposition of the dead and the inhabited area at oppida has already been noted at St Albans and Baldock (4.2) and can be seen in other instances, for example at Colchester (Crummy 1997).

The context for this view is a more general model for Romano-British towns which envisages their lesser importance as spaces for ceremonial and socio-political display than the countryside, especially the villa. A combination of epigraphic and archaeological evidence indicate that civitates in Britain had central places with public buildings which were the seats of local ordines. The same evidence suggests however only a very limited adoption of euergetism characteristic of Mediterranean towns (Blagg 1990; Mattingly 2006; Millett 1990). By the late Roman period on some accounts even this had significantly diminished, although the vitality of the late Roman civitas capital in Britain is debated; Reece (1980; 1992; also Faulkner 2000) casts them as no more than administrative villages; others infer greater administrative and social importance from their extensive walls, substantial stone-built and mosaic-decorated housing and the clustering of villas around them (Esmonde Cleary 1989: 71-81; Millett 1990a: 142; Rogers 2011).

The distribution of burials around Romano-British cities has mainly been considered in relation to urban topography. A linear arrangement along the roads radiating from towns was common in Britain, though the distance of burial clusters from the occupied area varied. It has been demonstrated at a small number of sites in detail and at a larger number more generally that early cemeteries lie at a greater distance from the inhabited areas of Roman towns than later (e.g. Crummy et al. 1993: 263; Hurst 1985: 62-4; Esmonde Cleary 1987).. This is not however universal, as Winchester shows (5.1). In the large inhumation cemeteries of the late Roman period the distribution of burial areas seems to have expanded away from the road frontage. This is sometimes considered as marking the advent of the ‘managed’ cemetery, though it is argued above that this exaggerates the distinction between early and late Roman cemeteries (5.3.6).

This chapter will argue that this underestimates the significance of an urban setting, ignoring the evidence of display in urban cemeteries and urban hinterlands and over-emphasising direct site-type association to the neglect of wider political and social geography. The question is approached in two stages. The first assesses the relationship between burial display and urban centres, extending the analyses of St Albans (chapter 4) and Winchester (chapter 5) to a broader body of mainly monumental evidence from the towns of Roman Britain. The second stage explores the relationship of burial display to the wider settlement landscape. This is based primarily on two regional case study areas, Winchester and environs, extending out to the wider civitas of the

The relationship between cemetery space and display has received much less attention, beyond Esmonde Cleary’s (1987) consideration of monument setting and Creighton’s (2006: 126-48) discussion of temples associated with possible ‘founder burials’ on urban margins. Chapters 4 and 5 reviewed two case study cities, Winchester and St Alban’s (chapters 4 and 5), with respect to the relationship between monuments, grave furnishing and context, especially the relationship to the road frontage. In both cases the corpus of burial monuments is limited, with no certain funerary inscriptions known and limited evidence for monuments,

111

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

were adopted to only a limited degree in Britain beyond the frontier zones (Biró 1975; Blagg 1990; Hope 1997; Mann 1985). The funerary inscriptions recorded from Britain’s various cities illustrate this trend (Fig. 7.1). Given the small numbers, new discoveries change the ‘ranking’ of individual towns; for example the discovery of three inscriptions in the last two decades from Gloucester’s northern cemeteries has increased the surviving corpus from the city by 150% from the total recorded in RIB. Mann’s (1985) observation that the small corpus of urban monuments is dominated by soldiers and individuals of extra-insular origin continues to stand (Mattingly 2008: 60-1, table 2). The distribution is also very uneven. Inscriptions are best represented at London, York and Lincoln, which together account for 59% of the 167 instances documented. This mirrors the general distribution of monumental inscriptions among towns (Mattingly 2008: 58, Table 1). In the samples from York, London, Lincoln and Colchester epitaphs also derive from a wider range of monument types; the ‘other’ category includes thin limestone and marble plaques from mausolea and enclosure walls as well as coffins, at least one statue base and larger mausolea of uncertain form. Since CSIR is not yet fully published the distribution of funerary sculpture cannot yet be quantified in an equivalent form. However the corpus from York (RCHME 1962; Tufi 1983), London (Merrifield 1977; RCHME 1928; Henig 2000), Lincoln and to a lesser extent Colchester (Huskinson 1994) is similarly much more abundant and varied than the limited evidence so far noted in other towns, with the exception of Bath (Cunliffe and Fulford 1982). The map of provincial sculpture compiled by Stewart (2010; Map 2) reveals this clearly, though many of the pieces from urban contexts outside the towns listed above are not or not likely to be funerary in character.

some of it decontextualised. The latter primarily comprises small timber post-settings and masonry tombs as well as enclosures. Excavation of road frontages outside Winchester (Victoria Road) and St Albans (St Stephens) indicates monuments to be placed intermittently rather than as a continuous band; in neither case was the frontage preferred for larger deposits of grave goods; in the late Roman period at Winchester evidence for monuments was distributed across the interior of cemetery areas. The major exception is the Folly Lane complex at St Albans, the location of spectacular funerary rituals in the mid-first century AD and later enhanced by the construction of a temple over the pyre site (4.2.3-4). In the following paragraphs this evidence is put in the context of other Romano-British towns. Although the largest sample of burial evidence derives from an urban setting, its distribution is very uneven (2.3), with few early Roman urban cemeteries having been excavated on a large scale. Excavations of the interface between burial and road frontage, a key context for assessing display, are generally lacking. We depend instead on reading the relationship between monument and its setting from small-scale excavations and the general distribution on urban margins of funerary monuments and burial assemblages. The ability to assess spatial setting is constrained too by the recovery of most fragments of funerary monuments, whether epitaph or sculpture, away from their original context. The first evidence to be discussed is the general distribution of inscriptions and funerary sculpture. Recent discoveries have not changed the general characterisation that funerary monuments of the 1st to 3rd centuries AD

Figure 7.1 Numbers of funerary inscriptions from the towns of Roman Britain (RIB I and III, Britannia to 2011)

Caerwent Silchester Leicester Dorchester Aldborough Chichester Canterbury Gloucester Carlisle Wroxeter Colchester Cirencester Lincoln London York 0

10 Stele

20 Other / uncertain

112

30

40

THE GEOGRAPHY OF FUNERARY DISPLAY

its higher position may explain the greater concentration of military and civilian monuments around the Mount, in particular close to the junction of the Tadcaster and Aldborough roads.

A very similar discrepancy emerges when one considers the evidence from individual cities. The majority of surviving elements of monuments from London derives from poorly known contexts or from re-use in the bastions of the town wall, most famously the Classicianus tomb, of which fragments were found at Tower Hill (Blagg 1983; Grasby and Tomlin 2002). Indirectly this indicates the presence of monuments in all of the city’s cemeteries north of the river but removal from primary context obstructs understanding of monument type and position. Recent excavations in the east London cemetery suggest the general clustering of monuments along roads: in-situ monuments were documented in the western part of the cemetery, nearer the town, although redeposited fragments concentrated in the excavated areas most distant from the walled circuit (Barber and Bowsher 2000: 301). The clearest example of a monumentalised street of tombs has been excavated in Southwark (Great Dover Street). Four funerary monuments were constructed adjacent to one another and parallel to Watling Street in the late first to late second centuries AD, two enclosures around the bases of masonry structures and two possible mausolea (Fig. 7.2). Architectural fragments included a pine cone and head of a river god. Siting of monuments may also have taken the river as well as roads into account. The Shadwell ‘tower’, if rightly claimed as a mausoleum, associated with a second century AD cremation cemetery, would have been prominent for those arriving in London up the Thames. For 1st century AD London, like Colchester, Hayward’s petrological analyses (2009) have shown that diverse continental and insular sources of stone were drawn on from funerary monuments.

At Lincoln the sculptural evidence, inscriptions and antiquarian observations also imply a diversity of monument types, including likely mausolea and barrows (Esmonde Cleary 1987: 109-13; Jessup 1959: 1; Richmond 1946; Jones et al. 2003: 40-42, 108-13). The findspots, again usually of pieces no longer in situ, suggest monumentalised cemeteries on all sides of the town with two concentrations, one on Ermine Street south of the river and the other on the road east from the upper town. The majority of inscriptions from the former commemorate soldiers and the origins of this cemetery area seem to lie in the military phase. Excavations in Monson Street revealed the wall of a large mausoleum or burial enclosure of later 1st century AD date and two likely slots for stele as well as four cremation burials. This confirms the location of a 1st century AD cemetery in the area assumed from the findspots of several soldiers’ tombstones (RIB 253, 258 and 260).

Currently available evidence suggests some difference in numbers and types of grave goods deposited in different cemeteries, favouring that to the north (Barber and Hall 2000; Hall 1996: 66-73), but further assessment must await the gathering of larger samples of reliable excavated data to compare to the eastern cemetery. Shepherd (1988: 11) noted the highly visible position to those approaching from the west of an early Roman burial plot from Warwick Square which, to judge from containers including a lead and porphyry urn, may have been those of higher status individuals, but no evidence survives for how these tombs were marked. In any case however evidence so far accumulated from London does not suggest that burial rituals included deposition of large quantities of grave goods, though cremation could sometimes be a significant focus of destruction (3.5).

At Colchester the Lexden barrow, the massive ditched enclosures at Stanway and the Gosbecks enclosure, if a funerary function lies at its origins, indicate a preconquest tradition of monument building (Creighton 2006: 130-5; Crummy et al. 2007). In the Roman period inscriptions, sculpture and tomb foundations are primarily restricted to the western cemetery area but their distribution suggests a dense cluster of monuments of diverse form around the cross-roads south-west of the Balkerne Gate (Crummy et al. 1993: 263; Hull 1958: 252-54). Of those commemorated on Colchester inscriptions (RIB 200-206) four are soldiers and one a Roman eques. Pre-Roman burials are not known within this area (Hawkes and Crummy 1995) so the cemetery may owe its origins to the legionary fortress burial ground. In the 19th century Joslin collection from Colchester’s western cemetery there are several large first and second century grave assemblages of ceramics, glass, and personal ornament, as well as the extra-ordinary collection of figurines in the ‘child’s grave’, but the reliability of grave groups must be questionable and information on the relationship of burials to the road frontage is lacking. This was not however the only zone in which monuments were constructed. Excavations in the environs of the circus have shown sporadically occurring monuments, including mausolea and barrows.

Most funerary inscriptions and sculpture from Roman York derive from cemeteries south of the colony centred on the Railway station and the Mount (RCHME 1962; Esmonde Cleary 1987: 162-163). Roman period re-use of monuments and limited documentation from their recovery in the nineteenth century impede detailed reconstruction of monument context. Nevertheless Jones (1984a) identified a greater concentration of monuments closer to the road within the Railway station cemetery, although this may have been only a road serving only the cemetery. A preference for greater visibility afforded by

Sample sizes from other towns are smaller. Little work has taken place since the 19th century on the cemeteries of Roman Wroxeter (White and Barker 1998). The best documented cemetery is to the east of the city, where the recovery of apparently in-situ monuments allows some reconstruction of cemetery monumentalisation (Wright 1872: 341-362). The principal excavations reported by Wright took place outside the East Gate. While trenches close to the town wall recovered no burials, a concentration of cremation burials and four inscriptions and perhaps parts of other funerary monuments was noted 113

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

likely first century inscriptions (based on epigraphic formulae) were found close to the road, one close to the Silchester gate (RIB 111) and others (RIB 108-110) at Watermoor approximately one hundred metres further south, just south of the fort rampart. It is likely that military tombstones were raised as the first memorials along this road, probably in association with the two first century AD military phases identified at Cirencester (McWhirr 1988). Three tombstones (dated possibly to the early third century) recovered from the rampart and probably deriving from the same cemetery suggest the continuing erection of stone monuments on this road (Hassall in McWhirr 1973: 212-13). Tombstones from Oakley Cottage and Phoenix Way (the latter re-used in a later Roman building), as well as a walled enclosure with a central platform and an urned cremation burial (monument 1103; McWhirr et al. 1982: mf. 5/5, B13) indicate the existence of monuments on the Fosse Way west of the town. Aerial photographic and geophysical evidence however strongly indicate the existence of a many probable burial monuments, including substantial mausolea, in the environs of the Tar Barrows c. 600m east of the Verulamium Gate, close to the shift northward of the line of the Fosse Way.

200 metres to the east and to the south of ‘The Lane’. RIB 293-295 were excavated close to one another, and to their north-west RIB 292 was recovered close to the foundations of a large building possibly associated with a cremation burial, though its funerary identification is questionable. In the field opposite, to the north of ‘The Lane’, Atkinson excavated a small group of poorly preserved cremations, some dated to the second century AD. One lay within a stone foundation six metres square, ‘possibly the base of a very considerable funerary monument’ (Collingwood and Taylor 1922: 253). The precise relationship of burial to Watling Street in this area is uncertain as the course of the road has not been confidently reconstructed. The tombstones suggest that at least part of this cemetery originated as that of the legionary fortress, although soldiers did not have exclusive use (RIB 295) and it continued in use after the final military departure in the 90s AD. This was not the only cemetery with monumental buildings; a stone platform with a glass urn containing ashes and other items was discovered in the 18th century at the confluence of the Tern and Severn a quarter of a mile north-west of the city. Perhaps the fragments decorated with imbrications re-used in Atcham church may derive from this or associated tombs (White and Barker 1998: 98).

In other cities very few funerary inscriptions or sculpture survive, monuments found in excavation are few and the evidence for some structures identified as Roman monumental tombs, for example Canterbury’s Dane John barrows, is inconclusive. Information on burial practice is also limited through lack of excavation in many cases. Where evidence is available burial display is more frequently documented as the deposition of large assemblages of grave goods. The lack of association between abundant grave furnishing and street frontage seen at Winchester and St Albans can also be seen at Chichester. In the excavation of the St Pancras cemetery, the excavated zone lay adjacent to Stane Street, but the areas of the cemetery closest to the road were either not excavated or badly damaged by later Roman gravel digging. The better preserved area adjacent to the road was devoid of burials, even though earlier features survived, and the road does not seem to have had a strong influence on the organisation of the cemetery, neither in terms of density of burial nor of the placing of larger assemblages.

At Carlisle epitaph findspots concentrate on Gallows Hill (RIB 954, 955, 959), as well as on Murrell Hill to the west (RIB 958 and CSIR I.6. No. 497, Coulston and Phillips 1988: 167-68). In the major cemetery area along the London Road/Botchergate, which extended for at least a mile south-east of the town, no direct evidence for funerary monuments was detected in excavations in the 1990s (Charlesworth 1978; Esmonde Cleary 1987: 29; Patten 1974; Zant et al. 2011: 103-4). At Gloucester the major early cemetery areas were north and east of the city (Esmonde Cleary 1987: 83; Heighway 1980). A concentration of inscriptions is noted at Wotton Pitch, including first century soldiers’ tombstones and others of similar date (RIB 121-2, 3072-3; Simmonds et al. 2008: 5). This space is c. one kilometre north-east of the north gate of the city, at the junction of the roads between Cirencester and Kingsholm and to the east gate at Gloucester. This point was on a ridge higher than the Kingsholm fortress or the city (Heighway 1980: 60). However this was not the only monumentalised area; to the west on the likely course of the road to the east gate were traces of probable burial monuments at St Margaret’s Hospital (Heighway 1980: 64), and a late second or early third century tombstone of a veteran of the twentieth legion has been recovered from the Kingsholm cemetery (RIB 3074).

The later Roman period is more difficult to assess and is considered only briefly here. With occasional exceptions the setting up of funerary inscriptions was no longer practised after the mid third century AD, at least not often in towns (Handley 2001). Little funerary sculpture can also be dated to the fourth century AD. However consideration of Winchester’s cemeteries has shown that timber and masonry markers and monuments continued to be constructed in some numbers (5.3); the Verulam Hills mausoleum with painted ceiling has also been noted (4.2.2). The evidence for late Roman monuments has never been systematically collected, but examples are attested in the cemeteries of several towns, though as at Winchester many are heavily robbed and are represented by foundations or construction trenches. The most

The number of funerary inscriptions from Cirencester is the largest of any civitas capital but contextual information is poor and, with the possible exception of a tombstone at Oakley Cottage, shows that the monuments, almost all stelae, were almost never found in situ (McWhirr 1973). Findspots of inscriptions are distributed along the road south of the Silchester gate. A group of 114

THE GEOGRAPHY OF FUNERARY DISPLAY

RCHME 1962), but there is little evidence for the privileging of particular spaces within the cemetery for their deposition. There is so far from Britain little evidence for burial ad sanctos, i.e. the creation of martyria within cemeteries and the favouring of their vicinity as a prestigious burial place; it is suspected but not yet documented at St Albans’ Abbey and at Butt Road the claimed cemetery church lacks evidence in its immediate environs for burials privileged in ritual or marker than other burials south of late Roman Colchester.

impressive and best-preserved examples derive from Poundbury, where rectangular masonry built burial chambers were distributed across the cemetery, some with evidence for painted plaster decoration on their interior; other monuments at Poundbury and Little Keep comprise ditched enclosures of a similar type to Lankhills. A wide variety of monuments is also visible at Colchester, built in the same area west of the Balkerne Gate where early Roman tombs had been raised; the most striking is a hexagonal tower tomb on the Royal Grammar School site as well as graves in large timberlined pits closer to the gate at Balkerne Heights, all of third to fourth century AD date. Excavations on the former Colchester garrison site show monuments also to have punctuated the landscape south of the town in the environs of the circus, including ring-ditched barrows and masonry mausolea. In the cemetery zone east of the basilical structure at Colchester Butt Road were traces of various markers, including one possible masonry-built mausoleum and graves marked by spreads of building material, postholes and, in six cases, evidence of very substantial timber vaults, though the above ground manifestation of these is not clear. Individual masonry structures are attested elsewhere, for example at the Mount, York, and Gloucester Kingsholm, and possible timber built structures at Leicester Newarke Street. London’s northern (Bishopsgate, Spitalfields) and eastern cemeteries preserve evidence of timber and stone built structures, the former not yet published. Though widespread, as the Winchester sample shows such monuments are not ubiquitous and some late Roman cemeteries in other cities also lack any evidence for monuments marking tombs. Little or no indication of funerary monuments has been noted in excavation of cemeteries, such as Cirencester Bathgate or York Trentholme Drive which on other grounds, especially the frequency of burial intercutting, can be identified as those of lower status groups.

7.2.2 Streets of tombs in Roman Britain? The evidence presented above documents significant variation in the degree to which urban burials were monumentalised by archaeologically visible markers of a greater scale and complexity than a mound of upcast or timber post. It is dangerous to place too much emphasis on this, as the survival of inscriptions on stone makes monuments from a small number of sites much more visible than the majority of towns where an extensive epigraphic culture was not adopted and where early Roman cemeteries have not been investigated in a way likely to recovery more fragmentary or ephemeral traces of funerary monuments. The examples of Folly Lane and, perhaps, the Tar Barrow complex at Cirencester, serve as a reminder of the potential funerary monuments in towns from which few epitaphs have yet been recorded. Nevertheless on the basis of this evidence two different trends in the use of burial space can be proposed. One mode is arguably exemplified by Winchester, St Albans and other civitas capital cemeteries in southern and central England such as Canterbury, Silchester, Chichester, Dorchester, Caistor-by-Norwich and Leicester. Here tombs bearing an epitaph or sculptural decoration are rarely found. Monuments are not absent but where excavated occur, as at St Stephens and Victoria Road, as small individual masonry or timber monuments or enclosures, more rarely in more substantial forms such as tumuli. This situation can be paralleled in examples from Gaul and Germany, where single or small clusters of monuments and enclosures are distributed along several hundred metres on roads traversing cemeteries. Examples include Kempten (Cambodunum) (Mackensen 1978), the port cemetery at Avenches (Aventicum) (Castella 1987) and the south-west cemetery at Tongeren (Atuatuca Tungrorum) (Vanvinckenroye 1984). This may represent the more common form of Gräberstraße in the north-western provinces, to contrast with the more extensive Gräberstrassen documented in Lyons and Gallia Narbonensis, where roadsides monumentalised by tomb facade sometimes extend for several hundred metres (Tranoy et al., in Blaizot et al. 2009: 263-6).

There is little direct evidence for the relationship between burial and road frontage in late Roman Britain, but indirect evidence, including the almost ubiquitous prevalence of west-east burial orientation, and the distribution of monuments across cemeteries suggests the diminished importance of the road frontage as interface between living and dead. The example of Poundbury, where Green (1982: 64) noted that the cemetery commands a fine view of Dorchester to the east, suggests that placement away from road frontage need not be confused with a disinterest in visibility. As at Winchester, in other late Roman cemeteries some burials are differentiated by containers or treatment of the body (e.g. a lead liner or stone coffin), or, more occasionally by grave goods (e.g. in antiquarian discoveries at York;

115

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 7.2 Funerary monuments and associated burials, Great Dover St, London late 1st-2nd century AD (adapted from Mackinder 2000: 11, 15, figs 8 and 12)

Figure 7.3 Funerary monuments at the Derby Racecourse cemetery (adapted from Wheeler 1985: Fig. 98, 223)

In this connection a further observation may be made concerning the Victoria Road cemetery at Winchester. In towns in Italy and Spain, and occasionally beyond, epigraphic evidence identifies the provision of a funeral and/or burial place at public expense as a mark of honour to decurions and others (Antico Gallina 1997; WeschKlein 1993). The site of such burials is most clearly revealed at Pompeii where of seventeen inscriptions, recording that either land was donated or monument cost

supported for tombs of magistrates by the ordo (typically with the formula locus datus decreto decurionum), thirteen lay c. 30m from the city gates, i.e. the probable distance of the pomerium from the town walls (Kockel 1983: 12-14). Such inscriptions are almost entirely absent from the north-west provinces, though for individual monuments a comparable circumstance has been proposed, for example the tomb of the Julii at Glanum, the Tour de l’Horloge at Aix-en-Provence and the 116

THE GEOGRAPHY OF FUNERARY DISPLAY

the location of many of the cemeteries from which such monuments derive has frustrated a characterisation of the cemetery environments (Petts 2003). However a combination of accumulating excavation and prospection evidence is allowing the character of a much bigger sample to be more securely established. Rather than continuous Gräberstrassen this suggests the existence of clusters of monuments on the margins of extra-mural settlements and along through routes, for example at Maryport, Vindolanda, Binchester, Lancaster, Newton Kyme, or Derby (Fig. 7.3), sometimes at a greater distance, with scale and / or setting rendering them prominent in relation to wider movement by land or water, for example at Shorden Brae (Corbridge), Brougham or Heronbridge. These larger monuments have had less attention than the stelae erected to commemorate lower ranking soldiers and their (probable) dependents (Hope 1997; 2001).

massive Flavian masonry-revetted barrow hard by the east gate of Augusta Raurica (Schaub 1992; Tranoy et al. in Blaizot. 2009: 263)). In this single case from Winchester there is no evidence for a similarly privileged zone. Within this group of towns, the Folly Lane monument (4.2) is exceptional, though at other sites a funerary association with an extra-mural sanctuary is proposed, for instance Gosbecks site outside Colchester, where the complex also includes a theatre (Creighton 2006: 135-48; Forcey 1998). Abroad, perhaps the most direct parallel is at Avenches in Switzerland (Castella 1993), where a Romano-Celtic type temple was built in association with an Augustan burial and remodelled in c. AD 25-30; a second enclosure, perhaps also with a combined commemorative and cult role, was excavated immediately to its south. This sanctuary lies on the road that links Avenches to its port on Lake Morat, though it competes with other very substantial funerary monuments in the ‘En Chaplix’ zone (Bossert 2002; Flutsch 1993). Sites of this type potentially blur the distinction between civic monuments and the tombs of a family or other group. The practice of endowment to support commemorative rituals by civic communities is occasionally documented elsewhere in the west (e.g. Gasperini 1996) and more commonly in the eastern empire; the ‘heroization’ extensively attested in Hellenistic and Roman Asia Minor, where the celebration of the dead of a particular family was made public, even a part of the civic calendar by endowments for mass feasting at subsequent commemoration, might offer a possible analogy for Folly Lane and related sites (cf. Cormack 1997: 141-43; Schnitt-Pantel 1982).

Another aspect of the greater monumentalisation of the cemeteries of York and London may lie in the character of their urban societies, both of which developed in the context of cities which were new foundations and did not develop from a pre-conquest focus. A possible though unproven contribution to London’s foundation may have come from a conventus civium Romanorum, a self governing body of Roman citizens and traders, often freedmen (Creighton 2006: 97-102; Morris 1982: 92; Millett 1996; Wilkes 1996). This community was sufficiently organised to plan settlement layout from the outset and, on analogy with such settlements elsewhere, would have had its own magistrates. In these circumstances, the more abundant legacy of inscriptions, funerary art and monuments in comparison to other cities in southern Britain seems to fit what might be anticipated from a population coming to Britain from areas with a more intense epigraphic habit, an initially fluid set of social relationships and, potentially, a large freedman element in the population with wealth derived from the movement of commodities on a key communication route; however no negotiatores are however as yet explicitly attested among Londinium’s epitaphs and only one, perhaps, in York (RIB 678). The quantity of funerary monuments from London is also small compared with those from other cities with known conventus, and there is so far little sign of monuments analogous to those of Roman merchants from the Magdalensberg in Noricum (Alföldy 1974: 70-72), or of the freedmen and magistrates attested at cities on key communication routes in Gaul or at Narona and elsewhere on the Dalmatian coast, for example (Wilkes 1969: 297-98; Woolf 1998; van Andringa 1998). Nor does London’s monumental record compare well with that from sites in neighbouring provinces equivalent to London’s later status as ‘provincial capital’. The number of funeral monuments recorded, for example, from Mainz (Boppert 1992a; 1992b; Hope 2001), Cologne (Galsterer and Galsterer 1975), Trier and Reims (Rheinisches Landesmuseum 1984; Wightman 1985: 166-67) and Lyon (Tranoy 2000; Tranoy et al. in Blaizot et al. 2009: 265-73) is several times greater than that from London. Recording of in-situ elements in these same cities reveals

A second mode of use of urban burial space can be identified at a group of cities with shared background, including colonial foundation and a connection to provincial administration, where more monumentalised cemeteries lined characterised urban approaches. These sites include York, London, Lincoln and Colchester and to a lesser extent Carlisle, Wroxeter, and Gloucester, in the latter two cases primarily in the first century AD only. Esmonde Cleary (1987: 168) identified a preference for location of funerary inscriptions on routes to London for several towns; road junctions and topographically prominent features can also be included amongst the preferred locations. In all cases there is a clear connection to the army, given the importance of soldiers as commemorator or commemorated among epitaphs from towns. This seems likely in part to reflect the much greater frequency of commemoration in this form by soldiers in general, as attested in the substantial corpus of epitaphs from garrison sites, both legionary and auxiliary (Biró 1975; Hope 1997). Occasionally findspots for such evidence, along with antiquarian discoveries of burials, fragments of sculpture and occasional in-situ monuments can be plotted to reconstruct extensive Gräberstrassen, for example the ‘Via Appia’ extending south of Chester for up to a kilometre towards the Dee (Mason 1987: 16465; Henig 2004; Pollock 2006: 24-6; cf. Caerleon, Pollock 2006). More often the limited knowledge of even 117

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

surviving evidence for decoration suggests a focus on the interior of the tomb, though in some cemeteries these monuments may have been more widely visible in the landscape.

extensive and densely built up monumental street frontages not yet seen in London, for example at MainzWeisenau (Fasold and Witteyer 1995) or Bonner Straße, Aachener Straße and Luxemburger Straße in Cologne (Gabelmann 1987) and at Lyons, especially the ‘majestueuse allée monumentale’ of the Trion mausolea (Tranoy et al. in Blaizot et al. 2009: 265). At some sites considered above, such as the Magdalensberg, limited subsequent occupation assisted in-situ preservation, but at others re-use in bastions accounts for a good number of known monuments (Blagg 1983: 130). It is therefore difficult to evaluate the differential impact of taphonomic processes in relation to the relative poverty of inscriptions and sculpture from London; Cologne, also in a region with relatively little stone suitable for building, dependent on imported stone and susceptible to the intensive recycling of stonework has a much greater corpus of monuments (e.g. Horn 1987: 157-60). Little survives of the Roman period fabric of London’s walls, but it is difficult to assess how far this distinguishes the city from its continental counterparts.

7.3 Funerary display beyond the city 7.3.1 Introduction: a framework for analysis The following section shifts focus to consider the distribution in funerary display across different types of settlement context, using the evidence assembled in previous chapters (4-6). The sample areas chosen, St Albans and Winchester and their environs, are particularly appropriate for re-evaluation here as they provide the basis for the general positions on the location of burial display outlined in the introductory chapter (1.4). Using the information assembled on burial assemblages and monuments in each area from urban, minor centre and rural sites (chapters 4-6), regional indicators of ‘burial display’ are defined and the distribution of these indicators is examined. An alternative approach, applied by Struck (2000; 2001) is to examine the proportion of predetermined ‘status’ indicators among the burials from different types of site. However whereas the much larger urban sample is often derived from large-scale cemetery excavations which recover the full range of surviving evidence, rural burials have often been noted as isolated monuments or assemblages without study of their context. An analysis that only explores patterns in relation to site categories such as ‘town’, ‘villa’ etc decontextualises the site from its location and its spatial relations to others in its environs. This section therefore assesses the type of context with which these indicators of burial display were associated and examines them in relation to broader political and social geography. As will be seen below this has a significant impact on our understanding of burial practice. Hodder and Millett’s (1980) study of villa location in relation to towns offers an analogy for the procedure adopted, although visual examination was preferred to statistical analysis as the sample size and data quality did not satisfy the demands of the methodology used by Hodder and Millett.

The record of funerary monuments from London therefore has two faces depending on the perspective from which it is viewed. In comparison to the rest of Britain the monumental character of London’s cemeteries is much more substantial, but from an inter-provincial perspective the corpus of epigraphic and architectural evidence is relatively slender. Additionally unlike many northern Gallic cities epigraphic evidence for peregrinae from London in other parts of the empire is lacking (cf. Wierschowski 1995). In fact the monuments from York are both more numerous (Fig. 7.1) and, very unusually for Britain, include a higher proportion of civilians of high rank, for example members of the ordo as commemorated (RIB 674; 3203), or commemorator (RIB 683; 3201), though many of these are closely connected to the army through marriage or parentage. Most of the inscriptions mentioning this group are on stone sarcophagi. Since the constraints on excavation within York’s historic centre means that it will always be difficult to recover a sense of the city’s monumental public buildings, the funerary monuments remain a key resource for understanding urban society here and hint at a phase of display through funerary monuments in the early third century AD among civic elites. Nevertheless from a continental perspective the number of inscriptions from York too is small. In no case for Britain can a culture of commemoration comparable to that in cities on the Rhône-Rhine yet be documented.

The location of burial display is examined from the LPRIA to the late Roman period, though the constraints of the evidence dictate a primary focus on the early first to mid-third centuries AD. The geographical limits are more difficult to set. In earlier chapters (2, 4 and 6) modern counties (Hertfordshire and Hampshire) have been used as convenient units for data collection. Within their boundaries lies a range of site types which can be potentially compared with practice in towns. The counties bear no relation to Roman period political boundaries, in particular those of the civitates. Using civitas boundaries in analysis raises several methodological problems, the principle one being to establish them with any degree of confidence. LPRIA coinage zones have long been used as the bases for Roman period civitates, with the support of some literary evidence (Rivet 1968). The relationship of the coinage distributions to the pre-Roman social groupings has been modelled by Haselgrove (1987b), but

In the late Roman period the distinction argued above between two groups of cities can no longer be seen, largely because a key source of evidence (inscriptions) is no longer available. However in almost every city where cemeteries have been extensively examined the construction of substantial monuments persists, as does a pattern of distinguishing some burials, sometimes associated spatially with these tombs, by the preparation of the body, container for it or grave goods. The limited 118

THE GEOGRAPHY OF FUNERARY DISPLAY

For the Roman period Hertfordshire is only part of the area within the civitas of the Catuvellauni as conventionally understood (Rivet 1968: 145-8). Before extending analysis to the remainder of the civitas it is necessary briefly to note the limited evidence for its territorial extent. Iron Age burial evidence has occasionally been employed as evidence for tribal boundaries but does not coincide strongly with other evidence (e.g. Peacock 1971: 175; Rodwell 1976a: 219). For the Roman period Ptolemy’s Geography lists the tribal affiliation of too small a number of settlements apart from civitas capitals and coloniae to be of substantial help. His inclusion of Salinae amongst the Catuvellauni, if the latter is to be identified with Droitwich, does not enhance his credibility as a source for civitas affiliations (Rivet and Smith 1979: 120-21). Iron Age coin distribution provides the major basis for its reconstruction, along with some topographical evidence. The Roman period civitas is said to lie within the Eastern coinage area, bounded by the Thames to south and west, the Ouse to north-west and the Stour to north-east. The Thames estuary and south-east Essex lie beyond the boundary of the Eastern series. The major problem for the purposes of this analysis is to use the different issues of the Eastern series to delineate a boundary between the civitates of the Catuvellauni and Trinovantes. The distributions of different issues have been interpreted as evidence for the changing political and military fortunes of these tribes (e.g. Allen 1944; Peacock 1971; Rodwell 1976a), or more recently as evidence for individual polities within the eastern group and the fluctuating spheres of influence of paramount chieftains or ‘kings’ (Haselgrove 1982; 1984: 25; Creighton 2000: 55-79). Whilst several issues of the Eastern series have a restricted distribution, they do not conform to any of the boundaries which have been hypothesised below. The other ambiguous area is that on the Fen edge north from the confluence of the Cam and Ouse to Godmanchester and Water Newton, which lies between three coinage zones.

we know little of the precise basis on which pre-Roman social formations were organised into Roman period civitates. Various administrative and territorial expedients may be postulated in the immediate post-conquest period, based on the better documented situation in Gaul (Haselgrove 1984) and the boundaries that have been established are best envisaged at most as zones than strict lines of demarcation (Millett 1990a: 67, Fig. 16). Mattingly (2006: 354-7) also emphasises the possibility of very extensive territories being appropriated from indigenous communities for the imperial estate and the army. As well as the discussion of boundaries, the use of the civitas as a unit of analysis also requires an assessment of the degree to which burial practices were sufficiently similar across it to justify comparison, as well as of the underlying distribution of burial evidence within it. 7.3.2 Display and the dead I. The civitas of the Catuvellauni The first analysis is focused on the Catuvellauni from the late Iron Age to the early 3rd century AD. From the presence of recurring artefact types in Late Iron Age burials in south-east England, in particular hearth furniture and amphorae, Stead (1967; 1976a) differentiated a ‘Welwyn’ group, divided into Welwyn and Lexden phases, from other Aylesford burials north of the Thames. Whimster (1981) and Rodwell (1976a) also adopted this category and divided other Aylesford burials into two further groups based on the size and content of assemblages, producing a tripartite distinction argued to reflect degrees of social status. Less subjective methods of differentiating assemblages, based on numbers of artefact types, have isolated a similar group of burials to Stead’s Welwyn group (e.g. Haselgrove 1982), to which new examples continue to be added (e.g. Baldock California, Stanway and Dorton). Previous characterisation has suggested an association with rural sites rather than ‘oppida’ (1.3.1) but an association with LIA central places was argued above in relation to Baldock and St Albans (4.7) (see also Bryant and Niblett 1997). Welwyn burials are also associated with other LPRIA centres elsewhere, for example the Lexden and Stanway burials with the dispersed complex at Colchester (Hawkes and Crummy 1995) and perhaps Cambridge, although most instances near Cambridge are poorly recorded and the ‘oppidum’ designation has not been fully argued (Burnham and Wacher 1990: 246). Some examples (e.g. Dorton, Snailwell, and Hertford Heath) are not close to known LPRIA centres. Nonetheless current evidence suggests that may be a stronger association between funerary display and oppida than previously acknowledged. The contested status of both the term oppidum and the character of later Iron Age central places must of course be borne in mind (Haselgrove 1986; Moore 2012; Woolf 1993). The British examples referenced above comprise dispersed residential, craft and religious funerary / areas over very many hectares and the distribution of high status metalwork and imported ceramics also suggests elite residence in smaller rural settlements at this time (Millett 1990a: 23).

For the purposes of this analysis the borders have been delineated as follows. The western and southern boundary with the Dobunni and the Atrebates, probably constituted by the Thames and Cherwell, is the least controversial. The watershed between the Nene and Welland valleys in Northamptonshire may mark the northern part of this western border, although analysis of Iron Age coinage may extend the Catuvellaunian boundary slightly to the north (J. Taylor pers. comm.). The towns of Dorchester-on-Thames and Alchester, and Towcester, Irchester and Ashton therefore lay in the zone where the civitates of the Catuvellauni, Dobunni and Corieltauvi met. Water Newton, at the northern edge of the civitas, may have become the centre of a civitas in its own right by the late Roman period (Burnham and Wacher 1990: 90). The border with the Iceni and Catuvellauni, following the coin distribution, may lie east of Godmanchester and Cambridge, the latter being close to the junction of the civitates of the Iceni, Trinovantes and Catuvellauni. The course of the boundary as it runs south-east through Essex between the Catuvellauni and 119

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

that the thresholds defined from the Hertfordshire sample (see above) are appropriate for wider analysis. Using other thresholds did not significantly modify the patterns noted.

Trinovantes is probably the most difficult to establish. Dunnett (1975) argues that it follows the river Lea, while Rivet (1968: Fig. 9, 161) places it from Godmanchester / Cambridge to Heybridge, allowing a very small area to the Trinovantes. Millett’s boundary lies to the west and ends instead at Southend. London probably lies on the boundaries of Catuvellauni, Trinovantes and the Cantiaci, though it may have been for a time in the territory of the latter (Millett 1996: 35). London’s burial practice is not considered in the present discussion, though some aspects were discussed in the previous section.

For the analysis of monument distribution the following categories have been used to summarise the diversity of types in this region, one of the richest for funerary monuments outside zones where the Roman army was garrisoned, comprising barrows, inscriptions, sculpture and other monumental evidence, including stone and timber structures and ditched and walled enclosures. The difficulty sometimes met in attributing monuments to the Roman period has already been discussed (6.2.2) and only those where more secure evidence of Roman date has been obtained are used in analysis (see Appendix 6.5). Even where date is confidently attributed there are significant uncertainties over form or even, in some cases, over the funerary role of some structures (e.g. Gallows Hill, Cambridge). Their fragmentary state and de-contextualisation also make the attribution of a funerary character to some finds of sculpture and inscriptions uncertain.

Burial evidence is not equally available across this area, being more abundant in Hertfordshire, northern Essex and southern Cambridgeshire than in modern counties in the northern and western part of the civitas (2.3). The absence of burials and monuments from northern Greater London for example may reflect the general scarcity of burial data, although there is a scatter of burials, mostly cremations, from settlements here (Celoria and MacDonald 1973; Merrifield 1983: Fig. 20). The longstanding archaeological tradition around Cambridge may be in part responsible for the concentration of evidence around the city; much of the evidence from development work on the periphery of Cambridge in the first decade of the 20th century is only partly published (Timberlake et al. 2007: 6-8), while by contrast a very abundant sample is now available from intensive development around Stansted airport (Medlycott 2011a: 42-4). Evidence from the central and south-eastern part of the area is biased to the early Roman period and that from the north-western part to the late Roman period. As the early Roman period is better represented in the bulk of the civitas, analysis is restricted to it. Despite the differential degree of attention, the attributes considered, i.e. monuments and large burial assemblages, are be more likely to be reported than simpler burial type sites and thus may mitigate the bias in the availability of evidence.

Mapping large assemblages characterised by large numbers of ceramics and diversity of grave goods produce similar distribution maps. Figures 7.4 and 7.5 show clusters of assemblages in the cemeteries of St Albans and Baldock (see 4.2-3) and along the northeastern boundary region of the civitas between Godmanchester, Cambridge, Braughing and Great Dunmow. Similar zones were also revealed when burials with individual attributes were plotted, such as the metal and glass vessels, very widely associated with larger burial assemblages or burial containers such as lion-head caskets and others with metal fittings (Philpott 1991: 230). From the Cambridge to Great Dunmow zone the graves with the largest and most diverse assemblages are particularly concentrated in the area around Stansted between Braughing and Great Dunmow, an impression accentuated by the recent concentration of fieldwork associated with Stansted airport but already substantially attested by earlier finds. A further group of burials with large assemblages but which fall outside the specific criteria used to create figures 7.4 and 7.5 also derive from a similar area, for example at Arrington, Hauxton, Girton and Stebbing. In contrast only a small number of assemblages, including Radnage, Bancroft and Dorchester on Thames, are documented in the western part of the civitas. The immediate context of burials varies, including urban, minor centres (burials from Great Dunmow and Dorchester-on-Thames supplementing those already documented at Baldock and Braughing; see 4.3-4)) and rural settings. In most cases the latter are difficult to specify; they include examples of villa (e.g. Turnershall) and non-villa sites (e.g. Stansted DCS.

While evidence occasionally exists for other aspects of ceremonies, comparison of burials across this zone depends primarily on two characteristics, grave good assemblages and monuments. Analysis of neighbouring urban, minor centre (4.2-5) and rural cemeteries (6.2.2) of early Roman date from Hertfordshire showed that average provision of ceramics on all sites varied between two and three vessels. Doubling the latter to six accessory vessels provides a threshold used here as one of the indicators of burial display. Form composition, recently demonstrated by Biddulph (2005) to vary significantly across this same area, is considered further below. Nonceramic artefacts were also common in burial assemblages. To treat the evidence of more rarely occurring artefact types together, a minimum NAT of 3 has been used as an alternative indicator of display, again doubling the highest urban or minor centre average in the area, which was that at Skeleton Green of 1.61 (4.4.2). The similarity of grave furnishing in quantity and type in excavated cemeteries from neighbouring counties (e.g. Bancroft, Fenny Stratford, Great Barford Bypass, Great Dunmow, Kempston Marsh Leys, Mucking, Rectory Farm (Godmanchester), Strood Hall, Stansted) suggest

The distribution of monuments is broadly similar. Figure 7.6 shows a cluster of monuments in and around St Albans, from a zone within seven to eight kilometres of the town, an association of individual monuments with some minor centres, for example Welwyn, and a second 120

THE GEOGRAPHY OF FUNERARY DISPLAY

concentration within five to eight kilometres of Great Chesterford, Cambridge and Godmanchester, in the boundary zone of three civitates. Scattered individual monuments and fragments of funerary sculpture are broadly associated with the boundary between the Catuvellauni and the Corieltauvi; all, with the exception of Hampstead Norris, situated between Thornborough and Bancroft in Buckinghamshire north to Stowe Nine Churches in Northamptonshire and Water Newton in Cambridgeshire. Some of these are associated with minor centres, including Towcester, Alchester (its military phase) Irchester, and Water Newton in the north-western part of the civitas, and Cambridge and Godmanchester in the north-east. Within the centre of Roman Cambridge the site of Ridgeon’s Gardens has been argued to be analogous to the sanctuary at Folly Lane in having a role in mortuary ritual (Taylor in Alexander and Pullinger

2000: 79-80) but the similarities are not close; as well as its intra-mural location, the Cambridge site lacks any obvious central burial, fragments of adult human bone are not unknown in non-burial contexts and the focus of late Roman rituals lies on infant burial. Recent work at Water Newton has confirmed that cemeteries to the south of the town include many more stone coffins than are documented in others, but the existence of a Gräberstraße proper on the road to Irchester and London, as proposed on the basis of aerial photographs, awaits further examination through excavation (Casa Hatton and Wall 2006). Otherwise the distribution does not allow a more precise classification than broadly rural: an association with villas applies at Bancroft, probably at Arbury, Wansford and Borough Hills, and possibly at Bartlow Hills and Litlington.

Figure 7.4 Burial assemblages from the civitas of the Catuvellauni and neighbouring areas with 6+ accessory ceramic vessels (K. Robbins)

121

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Figure 7.5 Burial assemblages from the civitas of the Catuvellauni and neighbouring areas with three or more artefact types as grave furniture (K. Robbins)

Figure 7.6 The distribution of first to third century funerary monuments within the civitas of the Catuvellauni (K. Robbins)

122

THE GEOGRAPHY OF FUNERARY DISPLAY

or ‘central southern English’ burial tradition and this has been retained for the purposes of this analysis. Varying this threshold expanded the number of burials which would be plotted but did not alter the character of the distribution.

There is some difference in the distribution of individual monument types. The small quantity of funerary sculpture and epigraphy was recorded mostly in the western, especially the north-western part of the civitas. The few temple-mausolea cluster around St Albans, with outliers at Welwyn and Bancroft, while barrows mostly derive from the north-eastern part of the civitas, especially in the zone south of Cambridge. If barrows sometimes argued to be of Roman date are also included then their distribution favours the eastern part of this civitas (see also Foster 1986: 191).

Few such burials were excavated in early Roman cemeteries at Winchester (5.2.3). Although the only sample of urban burials available for study, the evidence of the Victoria Road East cemetery is particularly significant in this respect because of its location where one might expect a concentration of urban burial display (7.2.2). The immediate context of the East Hampshire burials was rural, as noted above (6.3.2), but further examination shows that the majority of the burials cluster in three groups. One, of predominantly first century AD burials, is located between a few hundred metres and a maximum distance of five to six kilometres from Winchester, probably in association with extra-mural settlements but in no cases has the scale of excavation been sufficient to establish this (Fig. 7.7). A second cluster of first to third century AD burials lies within three kilometres of Neatham. To the small cluster in the Basingstoke area, only one of which is reliably documented, should perhaps be added a second century AD burial discovered in metal-detecting at Baughurst, with an assemblage including a bronze jug and pan as well as a silver brooch, although large numbers of ceramic vessels were not reported.

Taking the evidence of assemblages and monuments together reveals an overlap in two principal areas, the first within and around St Albans, and the second in the boundary zone between the civitates of the Catuvellauni, Iceni and Trinovantes, centring on the area between Cambridge and Great Chesterford, though there is little evidence of this type from the latter’s own cemeteries. There were slight differences in emphasis within the latter area; the distribution of monumental evidence extended to the northern part of this area, while the concentration of burial assemblages between Braughing and Great Dunmow is not matched by monumental evidence. Along the western boundary of the civitas the distribution of exceptional burial assemblages and monuments was sparser. The concentration of sculpture and epigraphy towards the northern civitas boundary may reflect greater availability of suitable stone from Northamptonshire and south Lincolnshire.

Neatham lies in what is regarded as the eastern boundary zone between the civitates of the Belgae, Atrebates and Regni, as well as at the boundary of Thiessen polygons drawn around the civitas capitals of these three entities (Millett and Graham 1986: 156). The ‘Basingstoke’ group may lie within the civitas of the Atrebates. In the LPRIA the area of Hampshire may have comprised part of the kingdom of Cogidubnus which was perhaps divided by the Roman administrative framework between the Belgae and Regni (Cunliffe 1993: 211; Millett 1990a: 68: Rivet 1968). To the west the boundary between the Southern and South-western zones in Iron Age coinage distribution between the Stour and Test (Kimes et al. 1982: 123) may indicate the Roman period civitas boundary between Belgae and Durotriges. Doubt has been expressed as to the validity of Ptolemy’s inclusion of Bath within the civitas of the Belgae (Rivet and Smith 1979: 120-21) which would extend its north-western boundary along a strip of territory through central Wiltshire to the Bristol Channel, thus including part of Avon and southern Gloucestershire, and cutting across the distribution zones of the South-western and Western coinages. If this area was included within the civitas, the chef-lieu at Winchester must have been so distant as to have possessed very limited significance as an administrative, social or economic centre. The different burial practices of this western strip of the civitas impedes direct comparison. Although similar in the Iron Age (Whimster 1981), they developed on a differing trajectory in the early Roman period. Very few early Roman burials are known from Wiltshire, northern Somerset and Avon and inhumation rather than cremation

7.3.3 Display and the dead II. Hampshire and the civitas of the Belgae A burial tradition comparable to the Welwyn type cannot be easily identified in LPRIA central southern England. Individual exceptional burials drew on a wide range of traditions of artefact deposition and were occasionally marked by monuments (6.3.2). The spatial associations of these few LPRIA examples do not allow clear generalisations. The Viables Farm and Owslebury burials were associated with smaller enclosed sites, while the Hurstbourne Tarrant barrow lies 30m south-east of the entrance to a banjo enclosure within a wider dyke system which is a possible candidate for a late Iron Age central place (Corney 1989: 112). The location may be analogous to that of the robbed barrow with in-situ cremation at Handley in Dorset, immediately to the north-east of the northern passage through the multiple ditches of the Gussage Hill complex (Corney 1989: 120-21; White 1970). From the conquest period a more consistent tradition of burial display can be identified from the recurring inclusion of large numbers of ceramic accessory vessels. The average burial assemblage in the Victoria Road East cemetery at Winchester contained just over one accessory vessel (5.2.1) but was higher for rural burials in Hampshire, over seven vessels at Owslebury and several vessels in the small number of other assemblages, admittedly an average distorted by small sample size (6.3.2). Millett (1987) used a minimum of ten vessels to define the burials, classified within the ‘East Hampshire’ 123

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

observation on the deposition of large sets of ornaments characterising town rather than country. Elaborate grave types, characterised by depth, size or construction, and monuments / gullies concentrated in Winchester’s cemeteries. There is a similar small grouping around the minor centre at East Anton, in the boundary zone between the Belgae and the Atrebates (Fig. 7.8). Stone and lead coffins have a wider distribution within Hampshire, some associated with villas (6.3.4), as well as a small number in Winchester’s cemeteries (5.3). Elaborate wooden coffins, identified by the numbers or size of nails or brackets, occurred in both the Winchester cemeteries and in some rural contexts, for example Burntwood Farm or Brook.

predominated (2.3). With occasional exceptions (e.g. Fowler 1964; Smith and Simpson 1964) this region rich in prehistoric barrows lacks a reliably documented Roman period barrow burial tradition. The most significant evidence of burial display comprises first to third century AD funerary inscriptions and sculpture from minor centres and rural settlements (RIB 108-18, 120-3, 133-4, 136-7, 184, 3045; Cunliffe and Fulford 1982; Henig 1993). These are thinly scattered with the major exception of Bath, where many derive from a likely Gräberstraße in the Walcot area on the Fosse Way, some 600-700m to the north of the town. This phenomenon has been explained with reference to the town’s unusually cosmopolitan population (Millett 1990a: 110); the commemoration of many soldiers here is also likely to be a product of the reasons discussed above for colonies and garrison sites (7.2.2).

Comparison with the western part of the civitas in the late Roman period is facilitated by the growing similarity of rituals across the southern part of the province, though complicated by the favouring of the latter area, in comparison to Hampshire, with supplies of lead and of stone. The distribution of lead coffins shows a concentration in the cemeteries of Bath and in rural sites to the north between Bath and Gloucester (Toller 1977: 3, Map 1). Stone coffins as well as mausolea have been documented at minor centres at Nettleton Scrub and Shepton Mallett and excavations at Boscombe Down have also shown how some burials were differentiated by the use of stone coffins and substantial ditched enclosures.

In the late Roman period different criteria for identifying burial display apply. Certain assemblage types and grave forms were shown to differentiate groups within the burial population in both urban (5.3.7) and rural Hampshire (6.3), although their co-association could be better demonstrated in the urban sample. The provision of ornaments with the corpse, such as crossbow brooches, belt sets, precious metal jewellery and large quantities of bracelets and beads, was characteristic of burials in Winchester’s urban cemeteries alone and not beyond. This pattern fits Philpott’s (1993a) more general

Figure 7.7 The distribution of burials in the East Hampshire Tradition (K. Robbins)

124

THE GEOGRAPHY OF FUNERARY DISPLAY

Figure 7.8 The distribution of stone and lead coffins and monumental burials in late Roman Hampshire (K. Robbins)

Contemporary evidence elsewhere can be cited in support of this observation. Manuela Struck (2000: 89-93) has mapped the distribution of some ‘elite’ monuments and assemblage types, though her study does not include inscriptions or sculpture. She emphasised their rural context, but when her maps are examined, it is clear that in southern and eastern Britain there is a clustering around some urban centres, primarily the public towns. This ‘halo’ effect can therefore be documented beyond St Albans and Winchester at other cities, both those with more monumentalised Gräberstrassen and those where such evidence is currently lacking. To the diverse burial monuments around Colchester already cited (7.2) can be added in its hinterland the two very substantial barrows on Mersea Island on the Colne estuary, the main entrance by sea and river to the colony; it remains to be formally tested but a deliberate siting for visibility by water and land routes seems likely. At Lincoln a scatter of monuments is attested within a few kilometres of the city, including a late first century AD barrow at Riseholme an epitaph found near Greetwell in the Foss Dyke (RIB 275) and a 0.9 x 0.5m fragment of an ansate panel from a substantial funerary monument at Branston. On the latter the formula in his praediis (indicating burial on an estate) is not otherwise recorded in Britain but is paralleled in Italy and North Africa (RIB 3179; Lengrand 1996: 123). Similarly there is also a cluster of possible Roman barrows on the periphery and in the hinterland of Canterbury (Wacher 1995: 200-1; Weekes 2011).

The most obvious change in the late Roman period is in the relationship of burial display to Winchester and its immediate hinterland. Some of the indicators of display have been recovered from the urban cemeteries alone rather than extra-mural settlement, especially those related to certain modes of dressing the dead. To the east the early Roman cluster of burials at Neatham has no late Roman equivalent, but this area is generally lacking in late Roman burial evidence of any sort. However the association of more elaborate graves as well as stone and lead coffins with the minor centre at East Anton and minor centres and rural sites in the west of the civitas cited in the previous paragraph show that Winchester enjoyed no monopoly of burial display. 7.4 The context of burial display - settlement type and civitas location In the Roman period display through burial is therefore polyfocal, a series of zones rather than attached to specific settlement types. In the previous section St Albans and Winchester were shown to be foci for burial display both in the form of grave assemblages in the larger urban cemeteries and as assemblages and, in the case of St Albans, monuments too, in ‘haloes’ around them spreading from several hundred metres to several kilometres away among small groups of burials and isolated examples.

125

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

peripheries argued to be significant for the early Roman period.

This pattern does not apply to every city. The distribution of large burial assemblages in the south of the civitas of Regni has been a key example for the model of rural display (Philpott 1991: 218-19), Beyond the city rural assemblages distinguished by their large varied artefactual assemblages were recovered across the Sussex coastal plain (Avisford, North Lancing, Marquis of Granby, Westhampnett 20820 or Westergate, Worthing) and an early Roman monumental tomb at Pulborough. The distribution of these shows little clustering around Chichester, although in the St Pancras several first and second century burials are furnished with assemblages equivalent in size to examples from the East Hampshire Tradition and are in practice part of the same grouping (Fig. 7.7). In other cases the emphasis lay more on the urban cemeteries rather than a broader peri-urban area. At York there is little evidence of monuments constructed beyond the urban cemeteries, though the inscribed coffins from sites north of the city (Hovingham and Suttonunder-Whitestonecliffe) may well relate to the frequency of this practice in the city which is not otherwise seen in Roman towns in Britain (RIB 720 and 3209; 7.2.1). Like villas large burial assemblages or funerary monuments are generally absent from the environs of London, with the exception of a cluster of monuments south-east of the city, including the barrow and mausolea at Keston, Otford and (of later date) the temple-mausoleum at Lullingstone. Millett (1990a: 88) attributed the lack of villas around London, in contrast to their clustering around many civitas capitals, to the difference in the character of social elites, those from London being based on the city itself rather than land-holding in its hinterland. The patterning in the funerary culture may perhaps be attributed to the same reasons. The distribution for funerary display may depend in part on integration between settlements and their hinterlands.

The distribution pattern identified for the early Roman period is similar to that of villas, the concentration of which in towns’ immediate environs has long been appreciated and used as a marker of the socio-political importance of urban centres (Millett 1990a: 191-95). Since the tombs discussed generally predate the phase in which villas reach their most extensive and monumental forms, we might use them as an indicator of the attachment to urban centres as a space of display throughout the Roman period. It was argued with reference to Folly Lane that its placing served to extend monumental urban space and associated ceremonial to the periphery of the city. Folly Lane and possible analogous sanctuaries focused on ‘founder burials’ (Creighton 2006: 124-48) may be part of a broader phenomenon, in that around many cities was a zone distinctive in its burial culture, expressed either in the richer character of the burial ceremonies that were carried out, evidenced through numbers and types of grave good and in the studding of the landscape with burial monuments. In such a zone the ceremonial space of the town was extended, in the more casual experience of monuments and tombs by passers-by, as well as during funerary and commemorative rituals. Some of the monuments in these locations were highly visible to the traveller, either by their roadside location or siting in prominent positions. Comment has already been made on the intervisibility of monuments around in the environs of St Albans has been noted above, though this aspect remains to be closely tested; the only study so far of landscape position through viewshed analysis emphasised local rather than distant visibility (Bartlow Hills; 6.2.4). In occasional cases, in this hinterland the size of some funerary enclosures, for example at Southfleet, Tollgate or Limlow Hill suggest the possible participation of large numbers in funerals or commemorative rituals, though they may alternatively have demarcated horti associated with the burial. At Wood Lane End near St Albans, possible facilities for commemorative ritual by large groups have been identified including a bath house.

For the late Roman period the case studies do not allow such clear characterisations for towns. Examination of Winchester and its environs showed the enduring importance of late Roman urban cemeteries as a focus of display. This is no longer the case for the peri-urban zone, though burial evidence is generally absent from the several villas documented known in this area. Late Roman evidence is more limited for Hertfordshire but the distribution of late Roman stone and lead coffins and monuments shows some limited association with St Albans and environs. Late Roman funerary monuments at other cities have already been discussed (7.2.1). 55% of the lead coffins discussed by Toller (1977: 2-4) were associated with walled urban centres and a substantial further number with other minor centres; the further 23 examples reported by Taylor (1993: 209-12) do not alter the distribution pattern. The distribution of plaster burials also clearly clusters around cities, with the exception of York, in central and southern England, especially Leicester, London and Dorchester (Philpott 1991: 90-1, 438-9). In general these indicators demonstrate an enduring focus on towns, but the evidence is too slight to assess the relative importance of towns versus their immediate hinterlands or the more distant civitas

This characteristic, i.e. the distribution of rich funerary monuments or grave assemblages for several kilometres outside the urban perimeter, can also be documented in other provinces (von Hesberg and Zanker 1987a: passim; Tranoy et al. in Blaizot et al. 2009: 273-75). Such ‘haloes’ have been identified, for example, around Cologne, Aquileia and Salona (Cambi 1987: 269; Gabelmann 1987; Noelke 1984; Reusser 1987: 244-47). They can also be seen elsewhere, for example in Aquitaine; despite Sillières and Soukassian’s (1993) assertion of the rural setting of tower tombs, their distribution is clearly focused on Auch and St Bertrandde-Comminges. The distances from the town in these situations do not preclude part of the burial ceremony taking place within it, and the subsequent transport of the corpse or cremated bone a few hundred metres or few kilometres to the rural burial place; in Italy the bodies of aristocrats could be taken from the town to rural estates 126

THE GEOGRAPHY OF FUNERARY DISPLAY

towards Irchester and ultimately London being the area where stone coffins have most frequently been found (Casa Hatton and Wall 2006). Other anecdotal evidence can also be given to support this association, for example Kelvedon’s late Roman cemeteries included a barrow with ring ditch, possible foundation for a mausoleum and six timber-lined massive burial pits of a similar form to those at Butt Road; the less well documented cemeteries of Great Chesterford include examples of stone and lead coffins and a possible mausoleum. Beyond the case study areas too late Roman burial monuments can be found in a small town setting, for example Fosse Lane Shepton Mallett. In summary there are therefore some indicators in both case studies for minor centres and civitas peripheries enduring into late Antiquity as a focus of display.

for burial (Dyson 1992: 144). A preference for rural estates is the likely explanation for the low numbers of monuments to decurions at Pompeii and other Italian towns (Mouritsen 2005: 50-1). The polyfocal character of these zones of burial display has been emphasised above. In the early Roman period minor centres and their hinterlands and civitas peripheries also emerged from the case studies zones of burial display. Concentrations of display indicators were documented in and near Braughing, Baldock and Welwyn and on the civitas boundaries, especially in the tract of territory where the civitates of the Catuvellauni, Trinovantes and Iceni met near Cambridge and Great Chesterford. A similar cluster at Neatham hints at the importance of the boundary of the Belgae with their neighbours. The type of burial display varies in form and, as at the civitas capitals, the context of association also varies, though the quality of evidence makes it difficult to assess location precisely in many cases. Some burials are located within communal cemeteries on settlement peripheries (e.g. Baldock Walls Field), some in smaller plots in similar locations (e.g. Skeleton Green, Neatham) and others in association with rural sites in the environs of small towns, for example at Neatham or Cambridge. When set in this context the funerary culture of Bath, though exceptional in the number of inscribed monuments, becomes an instance of a broader phenomenon. The conquest period burials from Tollgate and early third century inhumations from Southfleet join a cluster of monuments and spectacular burial assemblages around minor centres including Rochester and Springhead in western Kent (6.4.1) (Struck 2000: 9091, figs 9.5-7). Several of the small corpus of non-urban tombstones from in southern Britain have been documented in and around small towns (e.g. RIB 133, 134, and 136 in the hinterland of Kingscote, Glos.) In general however a clustering of burials and monuments is not so apparent in the immediate periphery of these settlements as around the civitas capitals; rather, in the case of the Catuvellaunian study area, they associate with zones at the boundaries between civitates.

The analysis above confirms and extends the previously observed relationship between burial display and minor centres (1.4). The implications of this association have not been fully considered, neither by students of burial practice nor of ‘small towns’. A brief reminder of their character is necessary in order to put the burial evidence in context. Minor centres pose persistent problems of classification and function. Hierarchies based on administrative status and function have been proposed (Burnham 1995; Burnham and Wacher 1990; Millett 2001; Mattingly 2006: 286-90). Both criteria differentiate a few settlements morphologically closer to civitas capitals, a larger group characterised by defences, street grid or occasional monumental building and a large ‘rump’ of agglomerated settlement areas, many known only from survey. The civitas of the Catuvellauni provides examples of all types. The size, layout and monumental character of Water Newton distinguish it from most other minor centres (Burnham and Wacher 1990: 90-91). Evidence for monumental buildings is otherwise more occasional, for example large masonry buildings at Braughing (Burnham and Wacher 1990: 1089; Partridge 1977: 65-67), and possible mansiones at Chelmsford (Wacher 1995: 209) and Godmanchester (Burnham and Wacher 1990: 126-27). The most frequently attested monuments are the walled circuits built for small towns on the civitas boundary zone (e.g. Godmanchester, Cambridge, Great Chesterford). In the east of the civitas is a very dense network of minor centres of which the degree of archaeological knowledge is extremely variable (Burnham and Wacher 1990; Millett 1990a: 143; R. F. Smith 1987). Their spacing broadly fits a suggested role as local centres for rural dwellers to exchange produce or acquire goods or services, although that they served this function often remains to be demonstrated (Burnham 1995; Condron 1995). A social as well as economic role, in particular as cult centres, has been flagged for British and continental European examples (Hiddink 1991: 218; Hingley 1989: 92; Millett 1995a: 36; Slofstra 1991; Vermeulen 1995). Within the sample from the Catuvellauni, there are many individual examples of shrines, although it is not clear that there are any sites which have more than local significance for a particular community, let alone the supra-regional importance of a sanctuary like Bath. On certain

In the late Roman period, as for towns, the picture is less clear. Display in the form of markers or elaborate coffins can be documented at minor centres. Within Hampshire, for example, there is some congruence in the grouping of both burials of this type and villas around East Anton. Late Roman evidence is too limited in the Catuvellauni case study, especially St Albans, for detailed comparison across contexts but the occurrence of monuments as well as stone and lead coffins shows individual examples in or around Braughing, Baldock, Welwyn, and Ware. Minor centres in the north of the civitas, especially Cambridge and Water Newton are especially rich in evidence for stone coffins; the availability of suitable stone, especially Barnack Rag, is likely to have had a significant influence here (Taylor 1984; 1993: 212).Less frequently lead coffin liners and plaster burials are also known from small towns in the Thames Estuary area (Philpott 1991: 438-9; Toller 1977: 3, Map 1). Perhaps the most abundant evidence comes from Water Newton, the road south 127

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

emphasis on the road frontage. Comparison with provinces immediately neighbouring Britain shows the limitations of this phenomenon, in particular for the case of London (7.2.2). In the second group, comprising many of the civitas capitals, though space for burial developed according to the same principles, less evidence has so far been documented for the construction of monuments. Instead a greater emphasis on display through variation in burial assemblages can be noted, for example in Winchester, St Albans and Chichester. At Colchester both large burial assemblages as well as monuments have been recorded.

indicators, for example density of occupation, evidence for craft production, especially ceramics, construction of walled circuits, and association with individual or clusters of villas, the floruit of minor centres has been dated to the third and fourth centuries. Their development has been explained by the patronage of local elites as alternative foci to civitas capitals, within devolution of economic importance and associated power to civitas peripheries in the late Roman period (Millett 1990a: 147-51). The burial evidence brings minor centres and civitas peripheries to prominence at an earlier date. Several factors argued to explain their late Roman development are also not inapplicable to the early Roman period. These include their roles as political, religious and social foci on the edge of very large civitates by local elites perhaps lacking a place on the ordo. The presence of monuments and large burial assemblages on, for example, the eastern boundary of the civitas of the Catuvellauni implies the existence of alternative spheres of elite competition to the civitas centre, perhaps of greater significance in these zones than civic public buildings (see also 4.6). By the criterion of funerary ceremonial, expressed in grave assemblages or monuments, there is less to differentiate core and peripheral areas, at least in this case. An analogous phenomenon is also visible on the near continent, albeit one represented by a much richer corpus of monuments. Hiddink (1991: 220) associated the substantial second and third century funerary monuments from minor centres in Gallia Belgica with the ‘middlemen’ between distant patrons, based elsewhere in or outside the civitas, and local communities; this position does not quite do justice to the character or numbers of monuments found in towns of the civitas of the Treveri or Tungri, such as Neumagen or Maastricht (Crowley 2011; Krier and Henrich 2011; Numrich 1997; Panhuysen 1996).

In sections 7.3 and 7.4 analysis was extended beyond the urban cemeteries to plot the distribution of display indicators in the same period. This indicated that while the immediate context of some burials or monuments might be ‘rural’, when plotted these showed clear clustering. Clusters included zones surrounding several cities. This included examples where the urban cemeteries proper were used to varying degrees as contexts for display. This clustering has analogies with that of late Roman villas focused on towns and in the significance that can be derived from it for the social and political life of early Roman towns. The same can be proposed for cities in other provinces. In both the context of the individual funeral (or commemorative ceremony) and in the enduring impact of monumental tombs the ceremonial space of a town could be significantly extended. For the later Roman period it is more difficult to offer characterisations, partly because of the disappearance of inscriptions as an indicator of display. Millett (1990a: 142) and Woodward (in Farwell and Molleson 1993: 239) have proposed that the size of late Roman urban cemeteries provides evidence for the continued role of towns as symbolic or social centres. The rituals and monuments also documented in many, though not all, supports Esmonde-Cleary’s (1992:38) view of the late Roman urban cemeteries as ‘an acceptable (though not mandatory) place for the making of social statements through burial ritual’.

7.5 Conclusion It was noted at the beginning of this chapter and in chapter 1 that burial evidence had been argued to support a characterisation of the limited development of oppida and the Roman period public towns, as social and political centres in Britain. Consideration of the case studies together with other evidence from the province allows us to revisit and challenge this characterisation.

The burial evidence also offers a different perspective on the minor centres and civitas peripheries (7.3-4). Millett (1990a) argued that in the late Roman period settlements of this type represent significant alternative social and economic poles to the civitas capitals, but there had been no examination of funerary evidence in relation to this. The evidence assembled here implies an earlier floruit for these centres than previously inferred. On the evidence of burial, as well as other criteria, they can be characterised as significant arenas for a competitive negotiation of social identities. It was not possible to identify in the main sample area of the Catuvellauni a correlation between the burial evidence and specific characteristics of minor centres, but location in the civitas boundary zones was in some cases associated with display indicators.

The degree to which burial display is associated with central places in the LPRIA has been under-estimated, although current information is too limited to permit a confident new generalisation and this is not the main focus of this chapter. For the early Roman period the evidence presented here, both from urban peripheries and their wider hinterlands, shows that understanding ceremonial life of public towns must not omit consideration of burial. In section 7.2 it was argued that the major public towns could, in the early Roman period, be broadly differentiated into two groups by the way in which burial space was used. In the first, comprising primarily York, London Lincoln and Colchester, a regular tradition developed of setting up inscribed and sculpted masonry tombs in places of high visibility with an

The quality of evidence from Britain will not allow direct comparison with the nuances of changing spatial relationships between burial practice and its environs 128

THE GEOGRAPHY OF FUNERARY DISPLAY

which can be reconstructed for central Italy where built monuments survive (1.3.2). Nevertheless aspects of this relationship can be reconstructed. On the basis of the characterisation offered above, a public / private distinction for classifying burial, or an identification of urban burial as ‘public’ and rural as ‘private’ do not seem useful (cf. 1.3). Previous emphasis on assessing individual site types, urban, ‘small town’, rural etc as contexts of burial also emerges as unhelpful: attention to broader distribution patterns is more fruitful in revealing structure in burial data.

129

CHAPTER 8: THE FORM OF BURIAL DISPLAY: A WIDER CONTEXT 8.1 Introduction Discussion of the location of burial display in chapter 7 primarily used the criteria of numbers of ceramics and other artefact types as well as monuments to define display. Attention is now turned to the composition of these burial assemblages. This chapter considers these objects placed with the dead from the perspective outlined in the introduction, where it was argued that they can allude to and evoke central areas of behaviour through which identities were created (1.3.2). It identifies their recurring artefactual contents and considers how burial ceremonies may have served to assert claims to power and knowledge through the mise en scène of artefacts placed with the dead and the potential readings which might be attached to them. The social groups with which these burial assemblages are associated may include landed aristocrats, the urban curial class or similar, but in the absence of epigraphic evidence or, in almost all cases, other specific indicators of rank, this cannot be definitively established. The emphasis lies on characterising burial display through analysis of its media, rather than speculative attribution to specific social statuses.

available containers, i.e. ceramic or glass vessels, and sometimes, north of the Thames, in wooden caskets with metal fittings in lion head form (chapter 4). The possible porphyry urn from London shows the occasional use of more exotic materials (Pritchard 1986: 175). Occasionally cremated bone was deposited in an organic container for which direct evidence no longer survives (Owslebury 1, Grange Road 2, or Neatham 2). In individual instances, for example Alton, the weight deposited in the undisturbed burials (over 1000g in three of four instances) was much higher than in contemporary urban cemeteries (cf. Fig. 4.4), but this was not noted in equivalent burials north of the Thames where such information was available (e.g. Baldock Clothall Road, Braughing A, Stansted). The presence of a higher than average quantity of cremated bone has sometimes been documented in comparable assemblages in neighbouring provinces (e.g. Roosens and Lux 1970: 19; 1973: 56). The artefacts placed with the dead were often in wide circulation amongst the living. Unlike some colonial contact situations, the material culture of burial here does not represent an inversion of everyday habits (Barley 1994: 73). The most salient characteristic of these assemblages is the emphasis on dining, expressed through the deposit of large numbers of ceramics, as well as glass and metal vessels, usually in the grave but occasionally also on the pyre (3.5). There is some regional variability in practice. For example the number of ceramics with the East Hampshire Tradition burials, especially at Alton, is much larger than elsewhere. As in burials in general, tablewares rather than storage or preparation vessels predominated (Philpott 1991: 36). Amphorae are a frequent component of Welwyn burials, but after the conquest period (Stanway, Folly Lane, Stanfordbury, Snailwell), they were only occasionally deposited (e.g. Stansted 25, Thornborough). The typical ceramic assemblage in these graves comprises a single or small number of flagons, beakers and jars, often but not always in local fabrics, and many cups and dishes, commonly in imported fabrics or in local wares imitating imported forms. Numbers of serving vessels such as flagons are roughly equivalent between burials but the greatest difference lies in the numbers of dishes and cups. ‘Services’ with smaller numbers of flagons or jars and larger number of cups and bowls are anticipated in some pre-conquest burial assemblages, for example Welwyn Garden City and Hertford Heath. In occasional cases the underfired fabric of some burial vessels rendered them useless for practical purposes (e.g. Alton 8, Daneshill, Neatham 1 and 2, Owslebury 1). This might be attributable to poverty but is better considered in the light of widespread evidence for the use of useless, broken or damaged (‘killed’) objects which can symbolise appropriate qualities but also mark the separation of the dead from the living (Hodder 1982c: 199). Deliberate damage is widely documented for ceramics placed in Roman burials (Willis 2004: 9.6).

The burials considered here contain almost as wide a variety of artefact types as those recorded in RomanoBritish burials as a whole but primary attention is given here to their common and recurring characteristics. The assemblages studied were derived from the sample areas examined in chapters 4 to 7, and to a lesser extent from elsewhere in Britain, especially the south-east. Individual assemblages are described and referenced in appendices 6.4 and 6.6. The latter do not exhaustively document graves from Britain outside the sample area which meet the criteria used in chapter 7, but contain examples to support arguments developed from the case studies. Kent is probably under-represented among the examples discussed. Many assemblages of the type considered have probably been excavated but older excavation records often hint at rather than fully document their existence, for example at Canterbury Ramsgate Road and Chattenden (Haverfield et al. 1932: 76, 150). Excavations in advance of construction of the Channel Tunnel Rail Link have provided better documented assemblages here, especially of the conquest period, most spectacularly from Brisley Farm and Tollgate (Millett 2007; Booth et al. 2011). The discussion examines change over time in the character of burial display from the Late Iron Age to the late Roman period but the major focus lies on the early Roman period for which the evidence in the case studies has been consistently richest. 8.2.1 The composition of early Roman burial assemblages This first section describes the principal characteristics of early Roman assemblages and their differences from the pre-conquest burials, especially those of the ‘Welwyn’ type (Ccf. Stead 1967; Stary 1991; Fitzpatrick 2007). The cremated bone was usually deposited within commonly 130

THE FORM OF BURIAL DISPLAY

and 36 (Tyers 1996: 106; Webster 1996: 18). Yet while only one burial assemblage corresponds exactly to the twelve piece service identified by Drexel (St Albans ‘tripod’), several more include a very similar range of combinations of samian or imitation samian (Table 8.1). Despite recent discoveries elsewhere, there is a notable cluster of such burials around first century AD Winchester (Crab Wood, Grange Road 1 and 2, Highcliffe, Milland, Winchester Hyde Street 566). In many of the burials listed in appendices 6.3-6.5 there are combinations of smaller numbers of samian vessels. Decorated samian, which accounts for between one in three and one in four vessels in first century settlement assemblages, very rarely occurs in the assemblages considered here or in neighbouring provinces (e.g. Berger and Martin-Kilcher 1976: 161; Ebel 1989: 40; Willis 2011: 226-7).

Some change over time in the imported fabrics in burial assemblages can be observed. In the pre-conquest period samian (terra sigillata) is largely absent from burial assemblages, although it is admittedly only present in settlement assemblages in very small quantities. The scarcity of pre-Flavian samian from burials, despite its presence on contemporary settlement sites, and the favouring of Gallo-Belgic wares, especially terra nigra and terra rubra, has been noted (Drury 1978; Millett 1993; Willis 1997: 46) and can be supported by some further examples (Colchester-St Clare Drive, Stanwaywarrior burial; Tollgate). However several pre-Flavian assemblages can also be noted where samian accounted for a substantial proportion of the ceramic assemblage, often in combination with Gallo-Belgic imports (Baldock Clothall Road 6 and 7, St Albans Folly Lane, Sheepen). From the Flavian period onwards samian comprised most of the imported dishes and cups in these assemblages. With the exception of Bayford 2, the largest samian assemblages, i.e. with ten or more vessels, all date to the later first century, but there are as many assemblages with over six samian vessels dated to the second as to the first centuries AD (e.g. Neatham 3, Stansted 25 and 26, Rochford, Arbury Road, Cambridge, Baldock Walls Field, Baldock Convent of Providence, Girton, Sompting (Marquis of Granby), Wotton Hillfield Lodge and Bayford 2). Large deposits of samian as grave goods are also known from second century Gaul and Germany. The remnants of funerary feasting, sometimes comprising the broken remnant of several hundred vessels are not yet paralleled in Britain (e.g. Abegg 1989a; van den Hurk 1984; Massart 2007; Schucany 2006:113-30, 375-81; Wigg 2000).

Burial assemblages could also include cup and dish forms in other fabrics. The combination of samian with other vessels at Folly Lane to produce an overall assemblage of ten cups, ten dishes and ten platters illustrates this well. ‘Services’ with little or no samian are also visible in other fabrics, for example in terra rubra (Colchester St Clare Drive, Stanway, Colchester Taylor Collection No. 4), in an unrecorded fabric at Jordan Hill and in a soft fabric probably made for burial in Owslebury 1, with seven or eight examples of cups imitating samian forms, larger cups and dishes. In Neatham 1 and 2 the number of Farnham Ware bowls is almost exactly half that of the dishes, comprising 19/20 (burial 1) and 13/14 (burial 2) services of one bowl and two dishes. While the many combinations cannot be accommodated within a restricted group of services, nevertheless burial ceramics clearly therefore do not represent a random accumulation of vessels. The ‘services’ are sometimes assumed to represent the dining equipment of a single individual, with vessels for several courses in which food was presented with different sauces in separate bowls, as well as drinking vessels (Balsdon 1969: 41-42; Nierhaus 1959: 42). However evidence for this is limited even in a central Italian context and the consumption of particular foods or use of particular dining equipment is however unlikely to survive a translation between contexts unchanged (Goody 1982; Walvin 1997). The probable use of the Greek crater at the Roman convivium as a container for wine rather than for mixing it with water exemplifies this and it is likely that use of vessels in the provinces differed from the ‘Roman’ norms (Dunbabin 1993; Meadows 1995: 137). Others have considered that the repeated sets of two-three vessels equate to multiple place-settings, typically three or four (Cool 2006: 195-99) though other dining-related artefacts occur singly. In some graves, (e.g. Alton and Neatham) the much greater number of ceramics is certainly too great to be accommodated within this single service. Burial assemblages are not necessarily a reliable guide to the tableware of the living, which may be indicated pars pro toto rather than in its entirety. Some vessels in use among the living, for example decorated samian, are not included. The safest characterisation which can be

Open forms are typically favoured in these burials (Willis 2004: 9.5.5; 2011: 222-3; also Biddulph 2005), but recurring combinations of vessels can also be identified. Analysts of cemeteries in other north-western provinces have used the ‘services’ proposed by Drexel (1927) as a framework for the interpretation of samian burial assemblages (e.g. Abegg 1989a: 216-17; Bayard 1993; van den Hurk 1984; Mackensen 1978: 170-72; MartinKilcher 1976: 84-7; Nierhaus 1959: 40-48; Schucany 2006: 377-81; Struck 1996: 103). From the correspondence between the description of silver table ware in a first century AD papyrus and samian burial assemblages in Germany Drexel advocated a samian ‘service’ comprising a larger dish or bowl with four dishes, four larger and four smaller cups or bowls, although variants have been recognised. The number of variants may be due to the ‘loss’ of some vessels on the pyre and the vagaries of collection of pyre debris by mourners and archaeologists, but a tendency to coin a new service for each new combination of vessels and to neglect the forms of non-samian vessels can be noted in some of the discussions referenced above. Others have questioned the validity of this concept (e.g. Ettlinger et al. 1990: 46-47) and the identification of ‘services’ has not been a pre-occupation of students of samian in Britain (S. Willis pers. comm.; 2011: 224). ‘Sets’ in a British context refers to pairs of vessels, usually a bowl and cup / dish of similar form but different size, for example Dr 35 131

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

combination of glass jug and bowl (e.g. Hallaton, Radnage) occasionally mirrors the pairing of metal vessels. Instances of fine imported vessels are rare: the bowl with Nilotic scenes from the Girton cemetery and the pillar-moulded mosaic bowl from Radnage comprise the principal exceptions.

offered is that that these burials include the dining gear of a single individual or small group. Larger vessels related to storage, preparation or consumption are only occasionally represented. There is some evidence for actual consumption, rather than its symbolic representation, although it is difficult to distinguish pyre goods from residues of funerary feasting, for example in the backfills of funerary chambers at Folly Lane and Stanway, or in grave fills from Alton. Amphorae smashed around the burial at Holborough, for example, may have derived from drinking of their contents at the funeral, but the origin of burnt ceramics recovered in non-burial deposits is less clear. This aspect of funerary rituals is better attested in Aschengruben associated with similar burials in neighbouring provinces (3.4). For example Abegg (1989a) estimated from an approximate seven piece set of samian per person that between eighteen and twenty people participated in the funerary feast at Siesbach. Other deposits suggest a similar scale of participation. For example at WadernOberlöstern, just south of Trier, an pit associated with a second century barrow burial contained the burnt remains of 50-60 ceramic vessels as well as glass and metal containers on pyre and most spectacularly at BiberistSpitalhof, north-west Switzerland, where the remnants of more than 300 vessels featured among the pyre and associated debris from a late second century funeral (Schucany 2006: 375-81; Wigg 2000).

The anomalous Welshpool deposit aside, the hearth furniture, i.e. cauldrons, firedogs, and suspension chains which are the striking feature of many Welwyn-type assemblages are not documented after the conquest, nor are the large wooden buckets whose copper alloy fittings survive in many of the same graves (e.g. Baldock Tene). In the pre-Roman period silver drinking cups and copper alloy jugs and pans were deposited in graves (Stead 1967). Save for their possible deposition on the pyre at Folly Lane, the former have not hitherto been recorded in post-conquest period burials in Britain. The deposition of copper alloy vessels continued. These occur most often as the jug and pan combination (sometimes referred to as ‘patera and ewer: see appendix 6.2 for discussion of terminology). Copper alloy dishes (Baldock Convent of Providence), bowls (Bartlow Hills, Godmanchester Nun’s Bridge), cups (Elsenham, Bayford 1) and strainers (Stanway ‘doctor’) are also reported; the cup with leopard handle, in a cremation burial near Abergavenny is so far unparalleled in other Romano-British graves. A small number were gilded, silvered or enamelled and most handles are decorated with animal heads or mythological figures or scenes (Toynbee 1964: 317-27). Nuber’s examination (1972: 153) initially suggested that in Britain as elsewhere the floruit of depositing the jug and pan pair lay in the late first and second centuries AD, but a larger sample collected more recently suggests little change in the frequency of deposition of metal vessels later in the second century gathered by Philpott (1991: 124). Although scholarship continues to relate this pair of vessels to the mixing of wine and water and the pouring of wine libations (e.g. Henig 1984: 131, 193; 1995: 70), their function as represented in artistic depiction is for hand washing, especially as purification before a sacrifice, but also before and between courses of a meal and as part of toilet routines (Nuber 1972: 96-112, 11725).

There are occasional instances of glass vessels in LPRIA (e.g. Hertford Heath) and pre-Flavian graves (e.g. Baldock Clothall Road 5) but glass becomes more common in the later first and especially the second centuries AD. As in burials in general the most popular forms in the assemblages considered here are bottles and jugs (Philpott 1991: 115-16). Their high representation is paralleled in the vessel forms in settlement assemblages (e.g. Cool and Price 1995: 235-36). In occasional cases multiple glass vessels were deposited. Some of these consisted of bottles alone (Arbury Road, St Albans St Stephens 11, Weston Turville); in similar instances in other provinces bottles were deposited boxed (Koster 1993; Roosens and Lux 1974) but this was not the case at Arbury Road and grave layout is not known in the other two examples. Bottles are usually considered as storage rather than table vessels, although different types probably fall into different categories (van Lith and Randsborg 1985: 424; Cool and Price 1995: 222). Totenmahl tombstones from lower Germany sometimes show very similar forms to large glass bottles Isings 50 and 51 beneath or beside the three-legged tables on which food is served (e.g. Noelke 1974: Nos. 10 and 11). Similar forms as well as drinking cups are also represented on the Simpelveld sarcophagus (Zinn 1997). Other assemblages comprised a combination of storage, serving and drinking vessels (Braughing Skeleton Green 33, Girton and Stansted 25, Turnershall Farm) although there are no known equivalents from Britain to the glass ‘services’ of multiple bowls and plates from the second and third centuries AD in the Low Countries (Massart 2001; de Groot 2006). As Philpott (1991: 116) notes, the

As well as dining equipment there is some evidence for foodstuffs. The presence of several different animal species, cremated or uncremated, distinguishes graves considered here from normal funerary faunal assemblages (Baldock Clothall Road 6 and 7, Grange Road 2, Little Waltham, North Marston, Milland, Owslebury 1, St Albans Folly Lane, Winnall A). The predominance of pig in funerary assemblages in Gaul and Germany is not replicated in this sample (Lauwerier 1983; Lepetz 1993; Méniel 1993). Instead the principal domesticates are evenly represented, although chicken is more frequently attested than in settlement assemblages. Wild species, including deer, hare and woodcock, rare in settlements and otherwise rare in graves have been recovered from a small number of burials of the type under discussion (Baldock Clothall Road 7, Milland, Winnall, Youngsbury) (Grant 1989: 144; King 1999; Philpott 132

THE FORM OF BURIAL DISPLAY

these examples. Paired spoons have been recovered from a small number of Late Iron Age burials but given their different form, much larger and of different shape, their function remains enigmatic and is unlikely to be the same as that of the cochleare (Stead, in Parfitt 1995: 105-07).

1991: 197). This rarity and diversity (even in contemporary burials from the same cemetery e.g. Baldock Clothall Road 6 and 7) are the most striking characteristics, rather than a consistency of species (or body parts) which might indicate a consistent sacrificial character to the food presented (cf. Blaizot et al. 2009).

Knives are also frequent in the East Hampshire graves (Alton 1, 2 and 5, Grange Road 2) and are associated with larger cremation burial assemblages (Philpott 1991: 281). Longer knives with a straight blade, often found in association with animal remains in burials (Philpott 1991: 176), probably served for food preparation and consumption, but their depiction with other sacrificial implements on altars suggests that they also were used for sacrifice. Triangular knives have been interpreted as craft tools or razors; a small number of other knife forms (e.g. Birdlip) may have been toilet-related (Boon 1991; Stead and Rigby 1989: 105).

Among other dining related items, bone and tinned bronze spoons were placed in three graves of the East Hampshire Tradition, all of the cochleare type with small round bowl and handle tapering to a point. Other examples derive from graves with substantial assemblages (Philpott 1991: 282). Silver examples are documented in burials in Belgica (e.g. de Groot 2006). In contrast to late Roman graves the significance of spoons, even of bone, in the early Roman period has received little comment (Böhme 1970; Sherlock 1973). The nuances of status distinction identified by Swift (2009: 122-3) for sets of late Roman spoons cannot be seen in

Table 8.1 Samian services in early Roman burial assemblages from Britain The table describes assemblages with a minimum of six samian vessels. The form descriptions (i.e. cup / dish etc) are taken from Webster (1996). The descriptions from older (Bartlow Hills 4, Hallaton, Bayford 1 and 2, Astwick) or unpublished excavations (St Albans Lindum Place, Wotton Hillfield Lodge) suggest the existence of similar assemblages in other contexts. Large sets of imitation samian vessels have also been recorded, for example at Chichester St Pancras 199 (eight imitation Drag. 27 cups, four of each size) and Winchester Grange Road 1 (one Drag. 36 and five imitation Drag. 18 dishes and six imitation cups (four Drag. 27, two Drag. 35/36). Grave Arbury Road Colchester Joslin 8/68 Grange Road 2 Crab Wood Highcliffe Milland Victoria Road 438 Victoria Road 566 Neatham 3 Rochford St Albans William Old Stansted 25 Stansted 26 Sompting York The Mount Baldock Clothall Road 6

Samian assemblages Three Drag.36 and one Curle 15 dishes, two Drag.35 and one Drag.46 cups Two Drag.18 and two Drag.15 (i.e. 15/17) dishes, four large Drag.27 cups and four Drag.24/25 cups A larger Drag.18R dish, four Drag.18 dishes, four larger and four smaller Drag.27 cups Four Drag.18 dishes, three Drag.35/36 and one Drag.42 larger cups and five Drag.27 smaller cups One larger Drag.18R dish, four Drag.36 and three Drag.35/36 small dishes / bowls, one Drag.46 and eight Drag.35/36 bowls, four Drag.27 cups 1 larger Curle 11 bowl, five Drag.15/17 and one Drag.36 dishes, four Drag.33 larger cups, four Drag.35 and Drag.27 smaller cups/bowls Two Drag.18 dishes, four Drag.15/17 dishes and two Drag.24/25 cups Five Drag.18 dishes and five Drag.35 and Drag.27 cups Two small Drag.42 dishes, two larger Drag.42 dishes and 2 Drag.42 cups, Drag.67 beaker Two Dr18/31 dishes, two Drag.33 and one Drag.35 cups, and Drag.67 decorated beaker One large Drag.18 dish, four Drag.18 dishes, four larger Drag.27 cups, four smaller Drag.27 cups One Drag.18 and two Drag.42 dishes, four Drag.35 and one Drag.33 cups One decorated bowl, one Drag.18/31 dish, two Drag.35 and one Drag.27 cups Two Walters 79 dishes, one Drag.31 dish, two Drag.33 and two Drag.39 cups One small Drag.18 / 31 and two larger Drag.15/17 dishes, four Drag.27 cups (two large and two small), four Drag.35 cups (one large and three small) One Drag.29 decorated bowl, five Drag.18 dishes, two Drag.27 cups

133

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

It has been suggested that table layout or ‘place settings’ may be inferred from grave layout (e.g. Cool 2006: 19397; Philpott 1991: 237) but while plausible in individual cases no consistent layout of dining material in the grave could be identified in the sample considered here and the nature of the meal, for a single or multiple diners remains uncertain. The absence of some key elements of the dining repertoire, in particular decorated samian, must not be forgotten. Vessels were sometimes stacked (e.g. Arbury Road, Welshpool) but more frequently were laid on the floor of the grave, sometimes differentiated by form or function; figures 8.1 and 8.2 illustrate some variants. At Grange Road 2 joints of meat were placed with a dish, cup, two knives and a spoon on a shale board. Bronze and glass jugs were placed to the south with three samian cups, and to the west a line of samian dishes and cups ended in a jar and flagon. In Baldock Clothall Road burial 6 joints of meat had also been placed on a wooden tray between two rows of pots. In the neighbouring grave (Clothall Road 7) these joints had been placed in separate pots. Closed forms, especially cups, were placed around the sides and dishes in the centre of the St Albans ‘tripod’ grave pit. In Winchester Hyde Street 566 grave contents were arranged concentrically, the cremation vessel at the centre and a ring of flagons and then of bowls around it (Ottaway 1992: 79). Vessels in graves at Neatham (2, 3 and 5) were arranged in a crescent. In the Stanway ‘doctor’ and ‘warrior’ graves amphorae, serving vessels and dishes and cups were placed separately. In some very richly furnished grave fills occasional cases furniture or ‘steps’ within the grave pit were used to allow the accumulation of massive quantities of objects on different levels, for example in these Stanway burials while at Tollgate the burial is argued to have been progressively backfilled as the objects were placed within the tomb in order to maintain them in place. Therefore whilst careful arrangement is clear in many cases, no consistent layout could be identified beyond local preferences, as at Neatham and Chichester St Pancras; grave dispositions cannot be considered an unambiguous clear guide to

spatial organisation in dining amongst the living, but are perhaps an allusion to their use in some cases. Ornament and items related to care of the body were more frequently deposited in early Roman burials than in the pre-conquest period, while some items, for example copper alloy mirrors decorated with La Tène artwork disappear from the toilet repertoire (Farley 1983; Fitzpatrick 1996; Joy 2011). A handful of burials of midfirst century AD date contain many brooches (e.g. Bancroft 4, Colchester St Clare Drive and Joslin collection 9 (May 1930) and St Albans King Harry Lane 86 and 370), but from the later first century AD brooches were an infrequent element among the dress items which occasionally form a substantial component of grave furniture (Linton, Southfleet, Winchester Hyde Street 466). The deposition of large numbers of ornaments and items of personal care in burials from the western cemetery at Colchester in the Joslin collection, for example in groups 7, 69 and 81b (May 1930), is rarely paralleled in other urban samples. Although the integrity of some of these grave groups is not certain, the cumulative effect is nevertheless impressive. Occasionally precious metal items of this type were buried with the dead, for example silver brooches or other items (e.g. Baughurst, Stanfordbury B, Stebbing, Turnershall Farm, Winchester Victoria Road East 466) and, at Southfleet, a gold necklace, ring and snake-headed bracelets with a third century AD child burial. The latter provides a British example of a wider phenomenon in the Roman west, the occasional burial of precious metal jewellery with juveniles or young adults (Rottloff 1995). Wearing a gold ring (Alton 2, Bartlow Hills II, Southfleet) in the early empire was an equestrian privilege (Henig in Millett 1986: 57) but the privilege was transgressed at Rome (Griesbach 2001; Wallace Hadrill 1994: 144) and is unlikely to indicate specific status in Britain. The association with a child at Southfleet throws further doubt on the association.

Figure 8.1 Cremation burial at Avisford (from Roach Smith 1848: Pl. XLIV)

134

THE FORM OF BURIAL DISPLAY

Figure 8.2 Plans of 1st-2nd century AD cremation burials approximately to same scale (for descriptions see Appendix 6.2 and 6.6) (clockwise from top left) Burial 2, Winchester Grange Road (after Biddle 1967: 232, Fig. 5); St Albans ‘Tripod burial’ (after Niblett and Reeves 1990: 445, Fig. 3); Burial 7 Clothall Rd Baldock (after Stead and Rigby 1986: 72, fig. 32); Burial 566 Winchester Victoria Road (after Ottaway et al. 2012: : 261, fig. 102)

Cosmetic and toilet sets, razors, toilet knives and shears, mirrors and glass and occasionally metal unguent or perfume flasks were also frequently recovered within these assemblages (Philpott 1991: 182). The deposit of strigils, in one case tripled (Bayford 2) elsewhere paired (Bartlow Hills IV) or single (Bayford 1, Prittlewell and St Albans Folly Lane 27), or bath flasks (Bayford 1, Sheepen) evoked the more specific sphere of bathing: in some cases these occur in the same assemblage (e.g. St Albans ‘tripod’ burial). Deposition of a similar range of

bathing equipment is documented in graves in neighbouring provinces (Mariën 1980: 245-48; Swift 2012). Furniture items have occasionally been documented, fragments of bone inlay from a couch from St Albans Folly Lane (as well as the Colchester ‘child’s’ burial), a tripod for serving food from St Albans William Old and folding chairs in Bartlow Hills IV and Holborough, and a probable table from burial 6260 at Tollgate. The chairs 135

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

being rare instances of a wider provincial Roman phenomenon (see below).

have been interpreted as the seats of magistrates (sella curulis) or army commanders (sella castrensis) (Struck 2000) but both the latter have a common origin in a domestic form from which these and other archaeologically known examples are impossible to distinguish (Jessup 1954: 28; Nuber 1972: 171). Their association with female burials in continental examples also undermines such an association (de Groot 2006: 11213). At most they may evoke such authority through recall of the performance of legal, political and ceremonial roles with which there were associated. Sceptres found in a burial at Brough-on-Humber may be symbols of a specific religious authority but their precise significance is unknown. Lamps, lamp holders and occasionally candelabras were a much more frequent component of the assemblages considered here than in burials more widely or in rural ceramic assemblages in general (Philpott 1991: 191; Eckardt 2002). Usually they had been placed in the burial but at Rougham one projected from the wall of the tile cist and at Avisford they were apparently placed on ‘corbels’ in the cist’s corners.

8.2.2 ‘porticus et balnea et conviviorum elegantia’: the interpretation of early Roman burial display The reading of these assemblages takes its cue from Richmond’s impression of cultural change derived from early Roman burial assemblages in Essex: ‘A quiet revolution in manners can be seen to have taken place, providing a commentary upon the Roman faith in ‘conviviorum elegantia’ as a civilising medium. (Richmond 1963: 19). His comment explicitly references the passage reporting the enthusiasm of British elites for assimilating Roman values and practices (Tacitus Agricola 21.3) and reflects a now problematic view which endowed Roman material culture with inherent ‘civilised’ qualities (1.2). Nonetheless this observation is worth developing as it identifies a key property of these assemblages which may be integrated into the broader debate regarding cultural change in a provincial Roman setting. The general lack of attention to the burial sphere has already been indicated (1.3), but additional factors may explain the lack of attention paid to the assemblages under discussion. It can be argued that the LPRIA media for burial display, in particular imported amphorae and table wares, were devalued as prestige goods by their ubiquity in the postconquest period. What were once scarce commodities, amphora-transported products such as wine and olive oil (Fulford 1991; Sealey 1985) and imported pottery, especially samian (Willis 2004; 2011) were available in much greater quantity. They could also be acquired from a wider variety of sources than allowed by pre-conquest exchange networks (Dannell 1979; Trow 1990). It has therefore been widely argued that burial display reached its acme in the conquest phase after which conspicuous consumption shifted to public and private architecture (Millett 1987; 1990a; 1995b: 123-24; Philpott 1991: 31; Struck 1995: 147; Trow 1990: 108).

Sets of gaming pieces, occasionally with dice, were deposited in burials from the first century BC (Welwyn Garden City) through the conquest period (Alton 2; Winchester Grange Road 2, Stanway warrior and doctor; Stanfordbury; Cobham), to the second century AD (e.g. Elsenham; Old Newton). They are generally associated with larger grave good assemblages (Cotton 2001: 3031). Their placing for play at Stanway, and perhaps also in the otherwise more modestly furnished grave at Ewell, is not paralleled elsewhere; usually they occur as clusters, perhaps sometimes buried in a bag. The ‘healer’s’ grave at Stanway suggests that traces of wooden boards in other tombs may be of gaming boards rather than trays. Writing equipment was also recovered from this group of burials, for example two styli and a seal box from Grange Road 2, and a likely inkwell from Elsenham, though not all burial assemblages with writing-related artefacts are so lavish (Philpott 1991: 185; Griffith 1912). Again similar burials in neighbouring provinces contain writing gear (e.g. Ambs and Faber 1998; von Boeselager 1993).

To some extent however this mis-characterises trends in burial practice. The decline in numbers of ceramics is undoubted in some burial traditions, for example in central southern England (Millett 1987), but is not universal. For example at Chichester St Pancras the number of large burial assemblages increases in the second century (Jones 1993b: 251), and most of the well furnished burials on the Sussex coastal plain date to the second century AD (7.4). The persistence of samian as a substantial component of burial assemblages has been identified above, and in the second century glass and metal vessels were more frequently deposited. The number of funerary monuments, especially barrows, also increases in the second century (Struck 2000: 88-89); some of the most substantial examples, for instance at Bartlow Hills, Mersea Island or Keston date to this period. Where resources which had previously been the media for status differentiation became more widely available, in particular amphorae and their contents and imported ceramic finewares, the deposition of possibly

The small number of weapons noted in Roman graves primarily date to burials of the conquest period, including shields (Stanfordbury A, Stanway warrior, Ashford Brisley Farm), spears (Little Walden, Old Newton, Stanway ‘warrior’, Ashford Brisley Farm) and in one instance chain mail (Folly Lane) and a sword (Toppesfield). This seems to represent the final phase of a pre-conquest mode of burial furnishing that is found widely across southern Britain (Parfitt 1995; Hunter 2005; Sealey 2007; Stevenson 2012). The arrows deposited in a second century burial at Wheathampstead Turnershall Farm are exceptional and probably comprise hunting equipment rather than weapons. Tools are very occasional finds, the medical instruments in one of the Stanway burials, planes from Wheathampstead Turnershall Farm and the adze hammer from Wendover 136

THE FORM OF BURIAL DISPLAY

1979; Mariën 1982; De Grooth and Mater 1997). According to Toynbee (1964: 317-27) the iconography of animals, deities and mythical creatures on copper alloy vessels found in graves indicates a specific ritual use. This is by no means impossible, a widely attested function of the jug and pan being for hand washing in the context of sacrifice as well as the meal, but this seems unnecessarily specific and restrictive. The popularity of such motifs may have been derived from the display of paideia that visual representation of episodes from myth such as the madness of Ajax (Bayford jug) or the wider occurrence of single motifs, for example the Triton and Medusa on one of the Turner’s Hall Farm burials.

now more exclusive items, in this instance glass and especially metal vessels, and perhaps also precious metal ornament, complemented the existing repertoire. The process envisaged by Miller (1982), in which the symbolic behaviour of elites continually modifies in response to its emulation by subordinate groups, has potential relevance for the evolution of the funerary sphere identified here. However this attempt to reclaim grave assemblages as instances of conspicuous consumption needs further nuance. First, precious metal vessels, in the highest place in the hierarchy of value, were not deposited in burials (Fulford 1985). In early imperial Italy literary evidence and finds from Pompeii show silver tableware to be widely available but it is difficult to know how far to extrapolate this picture to the north-western provinces. After the conquest period there is little evidence for the use of silver table vessels from Britain until third and fourth century hoards render them more archaeologically visible. The contents of the first century AD hoard from Hildesheim, and the earliest pieces from third century hoards from Gaul, for example at Berthouville, Chaourse or Rethel may suggest what was available as domestic plate in Britain (Hobbs 2006). Bronze vessels have been identified as a recurring component of these burial assemblages, although assessing their availability is complicated by recycling and by the dependency of their archaeological visibility on structured deposits and tombs (Eggers 1966; Cool 2006: 47-50). In the absence of an equivalent survey for Britain to that of van Lith and Randsborg (1985) the relationship of glass burials from grave and other contexts is difficult to assess, but the quality of grave glass, as defined by the proportion of storage vessels to table wares or presence of imports, is at least comparable to that from settlements (see above). The difference in form preferences among the ceramics between burial and other contexts has already been discussed in this chapter and elsewhere (4.2; 4.6; 5.2). Typically the proportion of samian vessels is higher in these burials than on contemporary settlement sites. Forms chosen for burial do not exactly match those in general circulation, decorated samian in particular being almost entirely absent. The condition of objects deposited with the dead varies but samian vessels in particular are not infrequently worn or repaired (e.g. St Albans ‘tripod’; Willis 2004: 9.6; 2011: 222). The safest current characterisation is that such burials are not impoverished as depositional contexts in comparison to settlement assemblages, but are drawn from material in general circulation with an occasional preference sometimes exercised for materials whose accessibility may have been constrained through cost; the associations of individual objects may have been significant (ch. 9). A similar absence of ‘mobilier de luxe’ is noted in the tombs of Roman Gaul, though this is regionally variable (Blaizot 2009: 32-33). Exceptional pieces of the type sometimes documented in similar tombs in Gallia Belgica, for example rock crystal and amber objects from Cortil-Noirmont or Heerlen, or the iconographically complex copper alloy vessels are not commonly documented in Britain, (cf. Braun 2001; Faider-Feytmans

The implications of the dining-related items for a more elaborate etiquette of consumption may be more significant than connotations of prestige or luxury associated with the objects themselves. The presence of jug and pan may signify an adoption of hand washing practices that typified the Roman meal and sacrifice (Feugère 1993: 153) and the use of spoons the adoption of a Roman-style form of eating not known before the conquest. Even if the reading of ‘services’ is uncertain (see above), the vessel assemblages imply increasingly elaborate modes of serving and consuming food, separating elements of the meal into different vessels and perhaps different courses. The aesthetics of such services, including colours, iconography, and the effect of the diffuse light shed by the lamps and candelabras deserve further consideration (cf. Brown et al. 1997). The importance of colour in particular has probably been underestimated. Colour images of burials at Berlingen and Helshoven, for example, showing a separation within the burial assemblage by colour, with red table wares, white flagons and grey / black storage jars in different corners of the grave, illustrate the potential significance of this aspect (Roosens and Lux 1973; 1974). The disappearance of certain items from the burial repertoire, especially amphorae and hearth furniture, may reflect change in the organisation and context of dining as much as their devaluation as a currency of prestige in the post-conquest period. Although access to wine and olive oil was increasingly less exclusive in the post conquest period, this does not by itself explain the absence of amphorae, as imported ceramics continued to be deposited. However it was perhaps the manner of serving and consumption of the varied commodities transported in amphorae may have taken precedence over access to them. The disappearance of hearth furniture, ubiquitous in pre-conquest assemblages (Fitzpatrick 2007), may reflect the separation of the sphere of cooking and preparation from that of serving and consumption. The presence of glass bottles in burials need not contradict this argument (see above). Blaizot et al. (2009: 334) suggest that changes of this type, which also characterise burials in Gaul, signal the shifting representation of the deceased, becoming dinner guest rather than, as before, ‘sacrificateur et chef du symposium’. However the changing spatial arrangements for dining and its increasing complexity seem as if not equally relevant to interpretation. 137

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

been lost to decay from these burial assemblages. Elaborate dress could also be burnt on the pyre: an example from Munigua hints at the potential splendours otherwise lost (Alfaro Giner 2005).

The same elaboration can also be identified in the items that relate to the modification of personal appearance. The jugs, pans, strigils, unguent and oil flasks, cosmetic and toilet sets, razors and shears may have been used to groom the corpse, as described in the literary sources (Scheid 1984; Toynbee 1971: 43-8). Structures associated with burials, for example the chambers at St Albans Folly Lane and Stanway, or, later, the post-built structures at Roughground Farm or Holborough may indicate spaces dedicated to this. These instruments may also have been used by participants in mortuary rituals. However they also evoke the increasing complexity of techniques for care of the body and its presentation that were generally practised. Hill (1997) has drawn attention to the increasing numbers of toilet sets in the late Iron Age and early Roman period, a development mirrored in cosmetic sets (Jackson 1985). In a funerary context they occur more commonly in these larger assemblages, perhaps suggesting that the grooming and toilette routines which they enabled were not universally practised, although in settlement contexts they are widely distributed (Eckardt and Crummy 2008). The strigils and oil flasks evoke bathing, exercise and forms of sociability associated with the bath house (Toner 1995: 54-55; Yegül 1992: 31-42). The strigil is symbolically potent, if bearing differing connotations, in Greek and Etruscan art and grave furniture (Kotera-Feyer 1993), but has received less attention within a Roman or provincial Roman context, although elaborately decorated examples show that it was not a simply functional artefact (e.g. Boon 1980; Mallet 2009).

References were also made in burial to activities such as bathing to be classified within the general realm of otium, or appropriate leisure. The presence of wild animals and occasionally of weapons perhaps used in hunting, may recall a privileged access of elites to wild game or evoke demonstrations of hunting prowess. The burial evidence perhaps anticipates the emphasis on hunt related scenes in elite self–representation seen in art from the Antonine period onwards, albeit not much in Britain (Anderson 1985; Aymard 1951). The presence of writing equipment has a potentially wide range of allusion, as examination of its sculptural depiction and the contexts in which documents were written and created suggest (von Boeselager 1989; Bowman and Woolf 1994; Freigang 1996: 216). The frequency of representation of the parchment roll or other writing equipment on funerary monuments has been interpreted as an expression of pride in entering an elite community united by education in a common literary tradition (Freigang 1997; Goudineau 1980: 362). The exchange of letters was also a medium of patronage and a mechanism of amicitia (Bowman 1994: 123; Stowers 1986). Other possible spheres of allusion include the production of ‘bureaucratic’ documents, inventories, accounts, censuses etc. Possible magico-religious connotations should also not be ignored, not only the specific context of letters to the dead said to be sent by the Gauls (Diodorus Siculus 5.28, 6) or curse tablets inserted in Roman tombs (Gager 1992), but also that writing, from altars to curse tablets, was in Britain a new and socially restricted mode of communication between the human and divine.

Several of these items have been used to sex grave occupants where the cremated bone was not examined or did not allow diagnosis of sex, but attributions made on this basis can be challenged. Millett, for example, identifies burials 3 to 5 at Alton as female from the presence of cosmetic sets and a pyxis, but an exclusively female association for cosmetic sets is not sustained by burials at King Harry Lane (Foster 1993). Grooming routines of male members of the Roman elite, including depilation with tweezers, provide further evidence that these were not exclusively female practices (Macmullen 1982b). This should not surprise given the long standing association of such items with both sexes in European prehistory (Treherne 1995).

Gaming pieces evoke an activity which, according to literary sources, took place both at the dinner party and in the porticoes of fora and temples (Balsdon 1969: 49, 15457). Archaeological evidence also indicates that game playing took place in public as well as private space; at Rome and other cities gaming tables were scratched into city pavements (Toner 1995: 90) while study of the distribution of gaming pieces at Vindonissa showed a concentration in one of the forum porticoes (Holliger and Holliger 1984). The presence of multiple gaming pieces is likely to indicate games of strategy, but since no grave contains the same number of pieces the specific games indicated are difficult to identify; Roman and other games can be proposed (Philpott 1991: 185, Schädler, in Crummy et al. 2007: 359-75). Other connotations are possible. The association in the Stanway ‘healer’s’ burial with the rods placed above the game board, perhaps for divination, may suggest a ritualised context for gaming (Crummy et al. 2007). More widely in the northern European Iron Age gaming pieces have been argued to be emblematic of martial skill (Whittaker 2006).

The dress and jewellery of the living or dead is less commonly evoked through what survives in these assemblages. From the later first century AD brooches were much less frequently deposited in Romano-British graves than in an earlier period: their presence is also less frequent than in cemeteries in other north-western provinces (e.g. Martin-Kilcher 1993b). The occasional large numbers of individual items, for example of brooches (e.g. Bancroft grave 4, Colchester St Clare Drive), suggest that they might be better characterised as offerings, as documented in votive deposits, than as grave goods (Woodward and Leach 1993). Objects preserved by exceptional conditions, such as the shoes with goldthread from the central burial in the funerary enclosure at Southfleet, occasionally remind us of what may have 138

THE FORM OF BURIAL DISPLAY

cultivated discrimination. Nevertheless just as the passage, if read appropriately, remains informative on the interaction between Roman administrators and native elites, so the tricolon balnea et porticus et elegantia conviviorum has implications which can be considered in relation to the burial evidence. In this context the most significant aspect of burial assemblages is that they symbolise the broad adoption of a new lifestyle. If they enable an afterlife of pleasure, then that pleasure had taken on a largely Roman form, centred on the convivium in particular but also drawing on other aspects of social and civic life (cf. Revell 2009). Freigang (1996: 218) explains the popularity of the scenes ‘de repas en famille’ on Moselle funerary sculpture as a manifestation of love of family life ‘retiré dans une sphère privée idyllique’. This emphasis seems too modernist. The convivium was a locale for the reproduction of hierarchy within patronclient relationships, simultaneously including, discriminating and offering opportunity (d’Arms 1990; Dunbabin 2003). Beneath the ideology of equal treatment of dinner party guests, distinctions operated through the quality of food, the promptness or demeanour of servers, but the convivium also offered access to potential friends and patrons. The importance of the convivium and rituals of presentation and reception to the creation of hierarchies in the western provinces is now appreciated to a much greater extent for the late Roman period, based on both the subject matter of mosaic pavements and the interrelationship of villa art and architecture (S. Scott 1994; 1995; Slofstra 1995). The burial evidence however suggests that its importance for the early Roman period is underestimated.

Roymans (1990: 245) and Haffner (in Rheinisches Landesmuseum Trier 1984, 286-87) have argued that the disappearance of weapons in early Roman graves in northern Gaul indicate a decreasing emphasis on status attainment through warfare in favour of one achieved by craft specialisation symbolised in the appearance of sets of medical instruments or other tools in burials, but the latter burials, though documented on the continent, are scarcely known in Britain (e.g. Bachmann and Czysz 1977; Boyer 1990; Henning 1991; Künzl 1982). The recurring associations identified in preceding paragraphs, with dining, toilet routines and leisure suggests the central areas of social reproduction which this chapter set itself the task of identifying. Treherne (1995) has drawn attention to the persistence from the Bronze Age to the early medieval period of presenting the ‘beautiful death’ of the adult male warrior through burial with weapons and toilet equipment. This topos is reflected in these Roman burial assemblages but in combination with another theme with a long pedigree in prehistory, the deposition of items related to the consumption of food. Williams (2004) has argued that as in other periods and places the deposition of ceramics in Roman graves has a mnemonic effect, prompting recall of other times and places in which feasting, and the particular objects used for it, served as a setting for the negotiation of social relations. The presence of gaming equipment is also widely documented in prehistoric and early medieval burials (Whittaker 2006). In some sense therefore the spheres of activity evoked in Roman burials are those in which claims about identity are negotiated cross-culturally: they are certainly ubiquitous in the European Iron Age. There is however a more culturally specific perspective from which these assemblages can be interpreted. Reference was made above to Richmond’s seeing in these assemblages the reflection of a ‘quiet revolution in manners’ and his linkage of them to Tacitus’ description of the power of dining. This connection can be developed further, as citation of the famous passage in the Agricola shows:

As Webster (1997a: 328) points out, different social practices may obstruct the negotiation of relationships, especially given the value judgements which are put upon barbarian behaviour. The burial assemblages imply styles of behaviour and self-presentation that were compatible between, or tolerable to, the different parties in the networks of provincial and metropolitan elites that the empire engendered. The cumulative effect of the constellation of symbols present in burial was to communicate a common culture (cf. Wiessner 1990). Burial makes visible a currency of manners in the areas of dining, of hygiene and cosmetic practices and of education and appropriate leisure perhaps analogous to that defined by Elias (1978) for late mediaeval and early modern Europe, where behaviour proper to a person of standing demanded accomplishment in self-presentation in parallel areas of social practice.

‘little by little the Britons went astray into alluring vices: to the promenade, the bath, the wellappointed dinner table (porticus et balnea et conviviorum elegantiam). The natives gave the name of ‘culture’ (humanitas) to this factor of their slavery’ (Tacitus Agricola 21.3) There is a close correlation between the associations evoked by burial of the different aspects of a ‘cultivated’ lifestyle and the values adopted by the Britons in their mistaken reading of humanitas. Tacitus’ description cannot be taken at face value but represents a particular reworking of the discourse on humanitas, in which he combines the justification of Roman imperialism as the propagation of humanitas with another long-lived theme, that of Roman decadence (Woolf 1998: 68-71). Tacitus perhaps also reacts to the reflection offered on humanitas by another culture’s selection of what it was to be Roman, which he may perceive as stereotyped values won through imitation rather than the development of a

Roman funerary portraiture reminds us that the values presented in death may not be universal and be contested: for example depictions of manual work or commerce through figured scenes or metonymically by tools would have been read differently by elites and freedmen (e.g. Kampen 1981). Likewise the qualities argued above to be evoked by the objects presented with the dead also possessed the potential for divergent readings. The perspective of the Roman literary sources shows the balance to be achieved, for example, between sophistication and excess in the presentation of food and 139

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

connotations as much as negative may well have been attributed to practices read as traditional or archaic; their straightforward linking to resistance is not convincing. The evidence of burial monuments shows a recurring echo of local pre-Roman forms across the empire (Jiménez 2008; 1.3). Second, the rituals described above are not isolated phenomena but can be closely paralleled in continental provinces. The deposition of quantities of ceramic and glass tablewares in similar combinations characterises burial assemblages across the north-west provinces, in France, Belgium, the Netherlands, Germany, Switzerland and Hungary (Bayard 1993; Blaizot et al. 2009; den Boesterd 1973; Castella 1993; Crowley 2011, with references; Ebel 1989; Wigg 1993c; Feugère 1993; Fitz 1980; van den Hurk 1984; Massart 2001; Mocsy 1974: 148; Roosens 1976; Schucany 2006: 113-30; Simon-Hiernard 1993; Struck 1993d; Walke 1963). The other recurrent elements identified, personal ornaments, cosmetic sets, styli and other writing equipment, gaming pieces, tripods and folding stools, also occur from burials from Britain to Bulgaria, often beneath barrows (e.g. Ambs and Faber 1998; Becker 1993: 370; von Boeselager 1993; de Groot 2006; de Grooth and Mater 1997; Mariën 1980). The distribution of the paired set of bronze jug and pan illustrates the distribution of such assemblages especially clearly (Nuber 1972: 162-71). While there are significance similarities in detail in what is present or absent, for example the avoidance of decorated samian, there is also diversity. Burial of horse gear and vehicles, for instance, was common in Hungary and to a lesser extent northern Gaul in the LPRIA and Roman periods (Becker 1993; Mariën 1979; Roymans 1990: 244-45) but almost absent from Britain. Weapons from Britain are also less common than in some parts of Gaul (Feugère 1996; Nicolay 2007: 199-206). Nevertheless this style of burial characterised large areas of northern Europe within the empire. The question arises then of what is a ‘Roman’ burial practice? With what, for example, would Julius Classicianus, probably of Treveran origin (his wife certainly), but in Britain as Roman procurator, have been buried beneath his altar tomb? In view of this wide distribution the burial evidence is not as inconsistent with other aspects of cultural change as some commentators have suggested. It alerts us to common patterns of social practice across the empire that need not radiate from Rome as an explicit package but which develop within the interaction of provincial elites, from participation in the military or Roman bureaucracy, and through the persistence of networks of economic and social connections already in existence in the late Iron Age.

dining (e.g. Gowers 1993). The adoption of certain cosmetic practices also risked characterisation as effete and ‘Greek’ (MacMullen 1982b: 178). Bathing was connected with a proper hygienic regime but was also physically and morally debilitating (Toner 1995). The context of gaming and gambling, the occasion and the social level of the participants, were also crucial to its valuation (Balsdon 1969: 154-57; Toner 1995: 94-95). Otium was differently defined and valued by elites for themselves and for the masses: ‘it was seen as entirely proper to treat leisure with due regard for the status of its participants.’ (Toner 1995:25) The cluster of connotations identified in burial also partially corresponds to the trope in funerary epigraphy of balnea vina Venus, a response to death by the appeal to sensual pleasure less constrained by refinement (Kajanto 1969). The Roman perspective should not be the only one from which burial assemblages should be judged. As noted above, many of these trends noted, for example the deposition of dining services or items related to personal care are already visible in the late Iron Age. Burial assemblages show an evolutionary development, less radical than the change visible, for example, in settlement architecture. Burial assemblages allude to long-lived themes in (elite) social practice further shaped through pre-conquest contact with Gallic elites. Even the type of contact between Roman governors and provincial elites implied by Tacitus’ description of interaction between Agricola and British principes need not imply a great difference in some areas of culture, especially when Roman officials could be derived from a northern Gallic background, for example Classicianus. Is it paradoxical that although the burial assemblage evokes a ‘Roman’ lifestyle, the style of burial seems not to be ‘Roman’? In other words it does not correspond to the burial practice of central Italy in the first century BC and AD, which are generally characterised by deposition of very few grave goods (Fasold and Witteyer 1998; Heinzelmann 2001)? On this basis the large grave assemblages in early Roman north-western Europe have therefore been considered as evidence for traditionalism in burial practice, Struck further developing this to argue for the potentially resistant character of such rituals, in contrast to other spheres of social life more engaged with Roman norms (1995; 1.3). Therefore against the argument developed above of an emphasis on shared culture, the different burial practice might have seemed alien or off-putting to a Roman observer. However there are various arguments to be made against this.

8.3.1 The composition assemblages

of

late

Roman

burial

A briefer examination is offered here of trends in late Roman burial practice in Britain. The changes in the Winchester sample (5.3) are used as illustrative for the purposes of this discussion but the degree to which they are representative of broader changes in Britain is also considered. These changes are also briefly put in the wider setting of the north-west provinces. As before the

First, while a mos Romanus in burial was occasionally identified by Roman authors (Morris 1992: 31) and contrast made with the curious or even repulsive burial practices of others (e.g. Montserrat 1997: 38-39), it is not clear how important specific ways of buryng the dead were to defining humanitas. For a Roman viewer positive 140

THE FORM OF BURIAL DISPLAY

‘temple-mausoleum’ at Lullingstone, along with a set of glass gaming pieces, or substantial glass assemblage (eight vessels) from a child’s grave at London Spitalfields exemplify the persistence of the dining-related grave furnishing of the early Roman period. Glass or ceramic assemblages related to grave-side ceremonies were also deposited elsewhere, for example broken fragments of up to six conical beakers were deposited in the lead-lined coffin and the grave fill at Welford-on-Avon. However in general numbers of vessels related to eating and drinking are much reduced; ceramics are usually represented as individual vessels, typically a flagon or beaker, rather than as large grave good assemblages (Philpott 1991: 109-113).

change in burial practice is summarised before considering the significance of the themes which can be identified. The emphasis on personal ornament in burial assemblages at Lankhills in particular illustrates a key change in late Roman grave good assemblages. This can also be identified, albeit usually in smaller percentages of graves, in many late Roman cemeteries in Britain. Hair pins, ear rings, beads, rings and especially bracelets, worn or unworn, have a wide but predominantly urban distribution, being recovered with a varying proportion of female and child graves in almost all late Romano-British urban cemeteries, although with the possible exception of York, not with the same frequency as at Lankhills (Philpott 1993a; RCHME 1962; Booth et al. 2010: 534-7; Cool, in Hartley 2006: 154-7). Within this sphere toilet instruments became a much diminished component in comparison to dress items and ornaments, with occasional exceptions, for example the exceptional jet and glass objects from London Spitalfields. Items such as belt sets and / or cross-bow brooches have a much more restricted distribution in graves. Outside Lankhills most other cases individual examples occur in larger cemeteries, for example at Brougham, Gloucester Kingsholm, East London or as isolated burials, for example Shorden Brae (Corbridge) and Water Newton (Normangate Field). These male-associated items in particular, and less often some associated with women, have been primarily interpreted in terms of ethnicity and military / civilian status, although the results of isotope study at Winchester have undermined any simple association with geographical origin (5.3.2).

This is a phenomenon that extends beyond Britain. The deposition of large numbers of ornaments in female graves and of cross-bow brooches and belt sets with male burials spans much of the northern Roman world and beyond (Böhme 1974; 1986; Feugère 1993: 155; Lányi 1972; Swift 2000). In occasional instances of exceptional preservation textiles have also been preserved; examples from Naintré (Vienne) or Milan, bear witness to the splendour of the dressed fourth century corpse (Bedat et al. 2005; Roche-Bernard and Ferdière 1993: 15-16; Rossignani et al. 2005). The British evidence is not however representative in every particular. Some dining-related objects which occur in continental graves are more usually placed in ‘structured deposits’ in Britain (3.4), for example silver spoons (Böhme 1970; Hobbs 2006), glass bowls with figure cut and engraved decoration (Harden et al. 1987; Price 1995) or pewter vessels (Beagrie 1989; Lee 2009). Nonetheless the number of dining-related items, in particular ceramics, the key characteristic of earlier assemblages, declined very substantially from the early Roman period, although large ceramic or glass assemblages in late Roman graves in northern Gaul and Germany are documented (e.g. Follmann-Schulz 1989; Noelke 1984; Rheinisches Landesmuseum Trier 1984: 210-19; Pirling 1993; Young 1977: 37). Weapon burial is so far little documented in Britain but much more frequently attested in late Antique northern Gaul (Theuws and Alkemade 2000).

Chance preservation of organic material, usually in ‘plaster’ burials or sealed coffins, occasionally allows a more detailed examination of the appearance of the individual and reinforces the emphasis in grave goods on dress. Textile fragments or ‘plaster’ impressions are usually interpreted as the remains of shrouds, although only in exceptional instances is this proven (e.g. RCHME 1962: 108-109). These may also derive from artefact wrappings as well as clothing (Crummy et al. 1993: 129; Wild 1970a: 91-94). Fabric fragments, for example of silk at Butt Road (Crummy et al. 1993: 128), of textile remnants dyed with imperial purple with a young child in lead-lined wooden coffin excavated at Alington Avenue (Dorset), or of gold thread (Winchester St Martin’s Close, Poundbury, Verulam Hills Field, London Spitalfields) illustrate a form of burial investment normally not preserved. Similarly the organic components of footwear, for example the leather shoes from Southfleet or the pairs with a child and adult burial in a stone coffin at Amesbury, one of calfskin and the other cork-soled shoes, perhaps lined with deerskin, or the heads of hair preserved from Dorchester (Crown Buildings) and York (RCHME 1962: 79) illustrate further facets of the cadaver’s careful presentation.

The differences between the early and late imperial periods in the importance of dress should not be overemphasised. Differences in the source materials for reconstructing dress, principally tombstones from the earlier period and mosaics and wall painting from the later influence our perceptions of dress from the different period. The creation of impact through fabric or colour of cloth, the latter suggested by some literary evidence, will have a very different archaeological visibility to that achieved through dress accessories and personal ornaments made of metal (Wild 1985: 408-9). Evidence for dress is also much more likely to have been consumed on the pyre in the early Roman period (see above). The significance of the presentation of the body was also noted as a recurring element in late Iron Age and in the early Roman assemblages discussed above. What has

The change in emphasis from the earlier period is not absolute. The deposit of a paired set of glass, ceramic and metal vessels and crossed knives and spoons in the 141

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

of burial. Objects of this type have been found with child burials, for example the two belts with chip-carved buckles placed by the head of nine to eleven year old child at Butt Road (Crummy et al. 1993: 145) or the belt fittings and, in one case, a crossbow brooch with a child and adolescent burials at Lankhills. Cool (in Booth et al. 2010: 308) explains the latter as the likely male equivalent of burying females of a similar age with abundant jewellery, perhaps an allusion to the expectations for their prime which they did not live to see. Alternatively they represent the presentation of junior family members with objects connoting an authority those burying them had hoped they would later assume.

changed, especially in the British material, is the shift of emphasis in burial towards the clothed and adorned body of the deceased, even if the most elaborate examples of male and female ornaments and dress items, especially in gold, have been recorded from hoards rather than burials (Johns 1996). 8.3.2 ‘cingulum ponere’: the interpretation of late Roman burial display A significant strand in the interpretation of late Roman burial has focused on male and female personal ornament as indicators of migration into the late Roman empire, in particular groups settled as laeti and foederati (1.3.1). Although belt sets were originally argued to have a particular military association (Hawkes and Dunning 1961), subsequent scholarship has identified a less exclusive group of belt wearers. Literary and artistic evidence suggests that belts were worn by both soldiers and late Roman government officials, the association being sufficiently strong for ‘cingulum sumere’ and ‘cingulum ponere’ to become metaphors for entry to and exit from imperial service (Bishop and Coulston 1993: 178, 192; Jones 1964: 566; Tomlin 1976). Individual burials make a striking impression here, for example the individual in burial 1846 at Lankhills, exceptional in his combination of silver belt fittings, gilded crossbow brooch and spurs (5.3.2), or burial 538 in East London, similarly with crossbow brooch and chip-carved fittings from a wide belt, plausibly to be read as burials of individuals of high rank.

Though much more abundant, the significance of female ornament has received less attention, perhaps because it is difficult to integrate it in such a specifically historical interpretation. Gowland’s (2001) and Cool’s (2010; also in Booth et al. 2010: 292-307) analyses of Lankhills have shown a clear association, in the context of burial at least, between the wearing of ornaments and age. These associations are also likely to reveal regional variation in appearance and costume, but the general use of large assemblages of ornaments with girls and younger women is widely distributed. The common emphasis on personal ornament may be put in a wider context as a shifting mode of funerary display. As in earlier periods the characteristics of burials of this type in Britain, along with more commonly adduced evidence, such as domestic space and decoration, indicate the development of display culture common to much of the Roman world, both in the form of funerary rituals and the connotations of the objects deposited during them. Status was asserted within the late Roman world through more elaborate and dramatic costume, jewellery and insignia (MacMullen 1964; Wild 1968: 176; 1985: 413).

However the widespread distribution of belt fittings on all types of site may even suggest that their wearing was not confined to imperial service. Similarly it seems difficult to prove that cross-bow brooches were exclusively worn by late Roman soldiers and officials in imperial service (Ager 1987; Clarke 1979: 263, 289; Swift 2000: 3-4). Certain sub-types of belt or brooches may have been particular to certain groups, although this is only made explicit by (some) inscriptions (Noll 1974). Halsall (1992) has argued that in late Roman Gaul the wearing of such symbols reflects a more overt display of power in burials, related to the assumption of power by local leaders, as the power of central Roman authorities retreated. This perhaps seeks too quickly to replace one specific group to whom this metalwork was attached with another, but acknowledges that the symbolic power of these artefacts was widely recognised and exploited. How to differentiate between the metalwork of office-holders and others is not clear and the spatial and social distribution is likely to have widened over time. It is sufficient for the purposes of this discussion, that both crossbow brooches and belt sets may have evoked the credentials of late Roman authority without necessarily identifying office holders. This is perhaps part of the wider ‘militarisation’ of the late Roman costume of authority, especially in a more elevated social milieu (Von Rummel 2005).

‘Had Cicero...stepped into Ammianus’ world, what certainly would have made him stare the most would have been people’s clothing: children in red tunics, riders on horses decorated with silver plates and fancy harness in black and red, magistrates in long jewelled chlamydes, some wearing the red leather belt of imperial service’ (Macmullen 1964: 450). Late Roman elites would have manifested themselves thus on a spectrum of occasions from the imperial adventus to meetings of the town curia, court sessions, visits to the baths or amphitheatre, temple, or church to the reception of guests and clients at home (Brown 1978; 1987; Macmullen 1981; 1984). A rare indication of such activity from Britain is the turn-out of the Pelagians ‘resplendent in dress’ (veste fulgentes) to confront St Germanus in the early fifth century (Constantius Vita S. Germani: 14: 261, 9); other processional urban rituals can be reasonably conjectured (Esmonde Cleary 2005). Representations of toilet and dressing scenes, for example on the Proiecta casket or the Silistra funerary paintings

Even if the status for the living which this metalwork designated could be identified with greater confidence, it is possible that its associations may change in the context 142

THE FORM OF BURIAL DISPLAY

particular on the objects deposited with the body, developing the argument advanced in chapter 1 that the forms by which graves are differentiated are not arbitrary but can provide cultural and social insights in their own right.

suggest that these in themselves could form a spectacle in their own right (Dimitrov 1962; Shelton 1981). The burial evidence allows no certain conclusion as to when and for whose benefit the ‘beautified’ corpse would have been visible. Although earlier Roman literary sources refer to the display of the body of the deceased, especially of the elite, prior to the funeral, at what stage of the ceremony the body was placed in the coffin, or when coffins were closed to view is unknowable; it is also very difficult in most cases to establish whether bodies were shrouded or dressed burial (Booth et al. 2010: 473-4). Some of the splendour of the funerals of emperors or of saints, for example of Germanus of Auxerre or Martin of Tours, the latter attended by the whole town and likened by the latter’s hagiographer Sulpicius Severus to a triumphal procession, may have extended to members of the late Roman elite (Février 1980: 436; Price 1987; Srejovic and Vasic 1994). Perhaps in late fourth century Winchester, London and elsewhere the bodies of important individuals and those connected to them would have been accompanied by their peers amongst decurions, churchmen, soldiers and officials, local patrons and landowners, as well as clients, household members, and slaves and other townsfolk.

In the absence of documentary sources it is impossible to identify the status groups to which these burials belong. What is more significant is the enduring concern to differentiate a privileged category of the dead, signifying a more dynamic character to the society of Roman Britain than the model which envisages that, for the civilian zone at least, a group of Iron Age elites consolidated their power within the Roman period. It is important not to exaggerate the degree to which the dead were privileged, since the abundant contextual information indicates that there were many other and potentially more significant spheres than burial for articulating social differences. A glance at the votive deposits of the late Roman period containing precious metal or pewter silver ware clearly reminds us, for example, that elaborate dining rituals play a social role in the later Roman period which we would not imagine from the burial evidence in isolation. The advantage of the abundant contextual information is that it not only allows a quantitative or qualitative comparison of burial with other spheres of life, but it also allows us to pursue the connotations of burial treatment, the allusions prompted in the minds of the mourners by both the individual items presented with the dead and their combinations in the grave.

Archaeologists and historians have widely drawn attention to the increasingly ceremonious nature of domestic space in late Antiquity, usurping functions previously in the public arena (Brown 1978; Brown 1980; Ellis 1988; S. Scott 1994; Thébert 1987). The material culture of the household, as exemplified in the great silver plate hoards of late Antiquity, reflects a similar process (Leader-Newby 2004). The potential for burial evidence to contribute to understanding this process has hitherto been much less exploited. Cameron (1993: 109) has argued that the lack of social mobility within the later Roman empire is exaggerated. Instead various overlapping mechanisms for transforming the status of an individual can be postulated, the army, imperial bureaucracy, church, land-ownership or control and wealth, all of which engendered increasingly complex patronage networks (Garnsey and Woolf 1989; Garnsey and Whittaker 1997; MacMullen 1963, 1984; Whittaker 1993). Patrons and clients still needed each other, perhaps to an even greater degree than before because of the proliferation of such networks. In this context therefore the material evidence for resources expended in burial on the container, the tomb and on displaty of the body may signify not only a social distance between honestiores and humiliores but also the precarious nature of that difference and of sustained efforts to assert it.

In Roman Britain from the first to third centuries AD the dominant emphasis lay on the table and on the broader sphere of otium. These burial assemblages have sometimes been read as expressing continuity with Iron Age practice, an indication of conservatism or resistance but it has been argued here that while this may have been one inference made by contemporaries, this is too simplistic a reading of the multivocal properties of these assemblages, fails to take account of post-conquest changes in burial ritual (in particular towards an emphasis on consumption) and does not acknowledge its broader Roman provincial context. Instead the reading proposed here emphasises the importance of the convivium as much in the early Roman period as in the later, when its significance as a location for the negotiation of power relationships has been inferred from the grand reception spaces of villas. It has also been argued these burials serve to demonstrate the acquisition of a particular savoir faire in the consumption of food and drink, in presentation of the body and in behaviour, which can be plausibly linked to the theme of humanitas. The burial assemblages may give a clue to refashioned lifestyles within the networks of provincial and metropolitan elites which converged on a common model of living, though with the possibility for diverse views of and judgements upon efforts to share in and shape that model. It was also argued that the traditional and ‘non-Roman’ character of the burial ritual has been exaggerated, since very similar practices have been documented across much of northern Roman Europe. These funerary rituals themselves here could be read as expressing part of that commonality of

8. 4 Conclusion The study of two sample areas in chapters 4 and 5 as well as the broader-focused analysis in chapter 6 demonstrated that across different types of site and throughout the Roman period a proportion of burials were differentiated from others, as seen in the treatment of the body, the marking of the grave and the objects with which they were deposited. Chapter 7 explored the spatial distribution of such graves. This chapter has focused in 143

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

culture. In the terms used by the analysts of prehistoric European burials (1.3.2), the prestige of these developing ways of looking and acting were given further sanction by being the symbols of the ‘beautiful death’ which were asserted against the encroachment of mortality. By the fourth century AD there was a change in signifiers and arguably of emphasis in signification. The repertoire of grave goods in Britain and beyond remained broadly the same, albeit much less abundant, but with a shift in emphasis towards dress items and ornaments; ceramics and other table vessels became much less common. Specific types of male dress items have traditionally been interpreted in terms of specific status or ethnicity or office-holding, to the exclusion of broader consideration of this phenomenon. This shift in emphasis has been interpreted here within a broader context of the marking of social difference through dress and ornament. In contrast to previous periods the ‘beautiful dead’ of late Roman cemeteries were no longer presented as sophisticated consumers (or dispensers) of dining largesse. Rather their presentation attempted to render them as distant as their living counterparts, though often perhaps by allusion rather than through deposition of the most precious items.

144

CHAPTER 9: CONCLUSION 9.1 Introduction The burial practices of the communities examined here were a complex and dynamic composite, framed by a number of contexts. These include the particular circumstances of the dead individual and the burying group at the time of death and norms concerning burial customs incorporating locally and externally derived practices (‘Roman’ in all its possible variety), continually ‘re-invented’ on the occasion of death. These intersect with an evolving suite of material culture for use in rituals and commemoration, and particular social and spatial settings. This final chapter draws on the results of the individual analyses to assess the character of burial data and consider the place of burial in the negotiation of social relationships. Some implications for future research to extend the approaches applied here are also considered.

This is not, of course, the last word, on the distribution of burial evidence. The biases identified here in its availability are susceptible to further analysis. There is scope to model taphonomy with much greater sophistication and nuance; the relationship of burial distribution to the burial environment was only characterised at the coarsest level here. More systematic and extensive radio-carbon dating for unaccompanied burials, especially inhumations, may modify the chronological distribution suggested above, especially as its application so far has varied by regional (2.3-4). This survey has also been based on only a sample of the evidence, although this sample was argued to be effective for characterising relative trends in data availability (2.2). The most significant lacuna is the absence of systematic survey of the grey literature, which will undoubtedly bring to light a larger corpus of evidence, especially for rural sites (Fulford and Holbrook 2011), although it is unlikely to affect significantly the overall distribution of available data (2.6).

9.2 Burial evidence – distribution and character A highly uneven distribution of archaeologically visible burial evidence characterises the Roman period in England and Wales by region, period and site type (Chapter 2). Burials have been much more abundantly documented from southern and central England than from other areas, but even here the distribution of burials varies substantially in space and time. The varying distribution at a local level was demonstrated by the case studies of Hertfordshire and Hampshire (6.2, 6.3). In general, especially for many counties of southern and central Britain from Somerset, Wiltshire and Hampshire to Lincolnshire and East Yorkshire the bulk of available burial evidence dates to the late Roman period, with an important caveat. Burial in the form of extended inhumation, sometimes in a coffin, can be shown to last from the early Roman period, where it accounts for a significant number of burials across the province, to the mid first millennium AD. Attribution of late Roman date on the strength of ritual alone is not very secure (2.3).

Taphonomic factors do not however wholly explain why the bulk of the evidence available for rural burial dates to the late Roman period, especially for the sample from central and southern England, where issues of preservation or the scale of fieldwork are less pertinent (see further below). The continuity from the Iron Age into the Roman period of an ‘invisible’ burial ritual for a proportion of the population has been advocated (2.5). The rituals by which some of the ‘missing’ dead may have been treated are, by definition, difficult to characterise. Although to some extent the burial of skeletal fragments and individual body parts may continue into the Roman period (6.3), this fragmentary material is less abundant than in the Iron Age. Even for the latter period we may doubt how far the deposition of fragmented human skeletal remains is a majority ritual (Hill 1995; Madgwick 2008). Instead we must contemplate funerary rituals which have left no trace, such as excarnation without effort to recuperate skeletal residue, cremation without subsequent burial (or burial in a context where archaeological recovery is unlikely, such as water or shallow burial without container or furnishing) or perhaps also the placing of burials in ‘off-site’ areas away from the foci of settlement which attract development-led fieldwork. The horizon at which archaeologically visible burial practice emerges is seemingly very extended, being drawn out over a longer time period than other aspects of change in visible burial practice. Emulation may help explain the spreading use, for example of cremation or inhumation, or of particular objects to accompany the dead, but this connects to a longer process where visible burial, directed at the interment of the body as separated entity rather than fragment intersects with slowly changing notions of personhood (Fowler 2004; Hill 1995; 1997).

The available burial evidence is biased both absolutely and in relation to the Roman period distribution of the population to urban sites, especially the public towns and to a lesser extent the minor centres (2.3). Low representation of military burials is due to an enduring lack of excavation of cemeteries as well as often poor bone preservation in the areas under lasting garrison. Elsewhere military burials are difficult to distinguish from others at sites which develop as urban centres alongside or following a military presence. Although the sample has increased significantly in the last two decades, rural burials are scarce in absolute terms and relative to the proportion of the population. Given that on any reconstruction the majority of the province’s inhabitants lived in the countryside, demographic, social or cultural inferences, based on the distribution of characteristics such as grave good furnishing or osteological indicators, may be compromised by the limitations of the rural sample; to what degree will vary to a significant extent between regions.

Publication of some major pre-PPG16 projects will contribute significant new data on rural burials, for example Mucking (Essex) or Stanwick (Northants), but new information will be produced by a slow accretion,

145

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Mizoguchi 1993), and for the Roman period via textual evidence. Scheid (1984; 2005: 161-88; see also Maurin 1984) has mapped the changing identities constructed for the dead and the living through these rituals. Casting funerary and commemorative rituals as sacrifices formally analogous to those practiced for the gods, Scheid identifies the roles of the living, the dead and divinities at the funeral; differentiated by the materials, spaces or modes of consumption, the dead were assimilated to the Manes, while the living rejoined their civic communities. Even if the model perhaps overdetermines the interpretation of the evidence, its application to Roman period cremation burials from central and southern France shows what may be achieved with high quality data from different phases of the ceremony where the recovery of ritual process is an aim from the inception of fieldwork (Blaizot et al. 2009).

since only small numbers of burials are often excavated from large rural sites. The derivation of most new burial data from developer excavations means that examination of rural sites are less subject to the biases of earlier excavations towards architecturally more impressive residential buildings, and ‘peripheral’ areas of farms and villas have received greater scrutiny. The same dependence on developer-excavation also means however that existing regional biases are very likely to be perpetuated. For the burials attached to upland frontier sites opportunities to excavate are likely, in the main, to be limited, although prospection allied to small-scale excavation may produce significant results for studying some aspects of cemeteries, in particular their setting and context (see below). As well as context, surviving data is also biased by the phase of ritual to which it relates. In order to analyse variation in burial practice, account should be taken of as much as possible of the burial process as a whole, not simply the final interment and what survives of it, the containers for the dead and grave goods as well as skeletal remains. Well-preserved Roman cemeteries, especially cremation cemeteries can potentially contain a variety of deposits, including pyre sites, associated debris deposits, the residues from other funerary and commemorative activity as well as burials. There is now a much greater sensitivity to this diversity and to the potential confusion between burial and other features, but formation processes are not fully understood and it is difficult retrospectively to re-interpret features that have not been extensively sampled and fully reported (3.3-4).

In a contextualising approach data pertaining to the archaeological study of the dead is not bounded by the limits of the grave. Even if not as frequent as in the preceding period, further attention is needed to fragmentary disarticulated human skeletal material noted from Roman period non-funerary contexts, in particular to differentiate between human skeletal remains in ‘structured deposits (3. 4) and chance disturbance, a likely occurrence where spaces of the living and the dead are closely interleaved (2.5, 6.2-3). More significant, perhaps, are the possibilities for studying the spatial and chronological relationships between living and dead. This includes characterising the immediate environs of burials, explored in considering the relationship between burials and settlement space on rural sites (6.4) and the relationship of monuments and burials to broader landscape setting and communications, especially the road frontage, the signature interface for living and dead in a Roman context (6.4, 7.2-4).

To assess the potential of this evidence pyre sites were selected as a focus for study using documentary and archaeological evidence. Literary and epigraphic evidence was shown to be of limited relevance for understanding the form the pyre took although inscriptions are essential to understanding the organisation of the cremation process (3.3); they suggest that the burning of the corpse like other aspects of provision for the dead fitted within the sphere of patronage of the familia, the collegium (at least as specifically documented in Rome and environs) and other small-scale networks and groups. The notion of a ‘public’ or ‘municipal’ cremation facility was rejected on the basis of both documentary and archaeological evidence (3.3). When a fuller range of deposit types and the behaviours that can be inferred from them are compared, a more nuanced comparison of graves and cemeteries is made possible, even if this has only so far been realised for individual deposits (3.1); the gaps in the current evidence impede wider conclusions. The various phases of burial ritual are not only alternative contexts for display. Where evidence for a ritual sequence can be established, there are further possibilities for exploring the structure of burial rituals through time within the rites de passage framework of separation, transition, and incorporation (van Gennep 1960; Huntington and Metcalf 1992). This has been influential on analysis of Neolithic and Bronze Age burial practice in Britain (e.g. Barrett 1988; 1993; Thomas 1991;

There is substantial scope for further exploration of the spatial context of cemeteries through non-invasive fieldwork. In the urban cemeteries discussed above much of the surviving remains of monuments has been found out of context, disturbed by the intensive continued use of the margins of historic centres. Re-examination of the context of older discoveries of inscriptions from Britain has produced worthwhile information on funerary monumentalisation (7.2), but will be more profitable when applied to the larger samples available from other provinces (see below). This is supplemented so far by only a handful of excavations which enable burials to be related directly to their environs (7.2-3). Where lack of modern development permits, in Britain at now unoccupied urban sites (or their peripheries) and at military garrisons at northern and western Britain, prospection, especially geophysics offers significant potential for reconstructing burials and monuments in their setting. Urban examples where this potential is already demonstrated include Silchester (J. Creighton, pers. comm.) and Cirencester (P. Guest pers. comm.; Winton 2009). The windfall of new data from recent geophysical survey of the environs of military sites in northern England and Wales has yet to be fully exploited 146

CONCLUSION

urban and rural burials but this could be extended into the late Roman period.

in this regard, but the examples of Birdoswald and Binchester show the potential for identifying burial spaces when mapping extra-mural areas around forts. The focus of cemetery excavations typically lies on the acquisition of samples of burials but an alternative approach combining prospection and small-scale excavation would assist in understanding such cemeteries as (monumentalised) locales of activity over time.

The change over time in Romano-British burial rituals is sometimes presented as a process of homogenisation. This draws on a wider discourse about the nature of Roman burial practice and is argued to be manifested in Britain in a long-term diminishment of grave good assemblages from their apogee in the burials of the Welwyn tradition (mid first century BC to mid first century AD), incorporating the most exotic and rarest of objects, to sober unaccompanied inhumation burials in regular rows by the fourth century AD. This impression may account in part for burial’s relative neglect in syntheses of the province’s social and cultural history (1.3.1). This alleged homogenisation is nuanced and contested here, situating study of grave goods in the context of other aspects of burial ritual and in relation to the wider circulation of objects placed with the dead.

GIS also offers under-utilised possibilities in this regard. This can assess systematically arguments previously offered for the use for burial of ‘poorer’ land as classified by soil type, slope, drainage etc) (e.g. van Doorsaeler 1967; Lintz 1993). More importantly it can enhance understanding of the audience of funerary display through assessing how far burials, especially monumental burials, were placed with a concern for their visibility in a wider landscape. In comparison to other periods in Britain this has been little practised for the Roman era, the study of the Bartlow Hills being the key exception (Eckardt et al. 2009a); significant potential exists for assessing the spatial relationships proposed in a variety of settings (6 and 7).

First, the observation that burials are few for the early Roman period bears on differentiation among the dead; if correct then a distinction is drawn between those individuals afforded an archaeologically visible burial and those not. From the mid first century to early third century AD cremations comprised the majority of visible burials in the study areas, with inhumation accounting for a minority of adult as well as most infant burials. The significance of the difference is not clear; while crouched inhumation might be attributed to continuity with Iron Age burial practices, extended inhumation is more difficult to interpret. Niblett (2000: 101) suggests for St Albans that conquest period inhumations may represent burials for which the full ritual sequence was not completed by burning, though cremation burials have not yet produced evidence for the burning of ‘dry’ rather than ‘wet’ bone. Since space for burial does not seem restricted and as some early inhumation burials are limited in their furnishing or lack coffins (St Albans Folly Lane and St Stephens, Winchester Victoria Road; 4.2.1, 5.2.1) the difference may also relate to the resources available to the buriers, but this is difficult to insist on as a global explanation. At Baldock, by contrast, and in cemeteries outside the sample area (e.g. East London; Gloucester London Road) contemporary cremation and inhumation burials were both furnished with few if any objects. The variety of forms taken by inhumation and of their relationships to associated cremations suggests that by itself the ritual lacks a single specific social or cultural association and instead could be read as a marker of group identity at multiple scales.

9.3 Burial rituals and Romano-British society 9.3.1 Differentiating the dead These difficulties notwithstanding, chapters 4 to 6 examined the differentiation in burial rituals in regional case studies with a view to assessing the role and significance of burial within social dynamics. Qualitative and quantitative differences between cemeteries have been previously noted in discussion of Romano-British burial practice and broadly attributed to status differences, especially for south-east Britain where the largest assemblages and the majority of monuments are documented. The emphasis in the analyses was to assess earlier conclusions using as a broad a range of evidence for ritual as possible, seeking to explore how social identities, especially status, were negotiated rather than attempting to map particular burial types onto social categories (1.4). A pragmatic approach was adopted to selecting case studies, Winchester and St Albans, where good samples of urban burial data were complemented by minor centres (for St Albans) and rural sites in their environs. Comparable burial practices generally existed across these different contexts, at least in terms of archaeologically visible elements, though Winchester lies in a transitional zone between cremation and inhumation as majority traditions in the early Roman period (6.3). In order to establish the rural sample county boundaries were partly used but the broader civitas context was also considered in both cases for the analysis of ‘display’ in the form of larger and more diverse grave good assemblages and monuments (6.2-3, 7); the imprecision and uncertainty over the location and nature of civitas boundaries is acknowledged. For St Albans and its environs, it was possible to assess early Roman urban, minor centre and rural assemblages in relation to the burial rituals of the preceding period. Winchester and its environs allowed only assessment of

Following the choice of to bury or to burn, the ritual could be mapped in greatest detail for cremation burials of this early period. With occasional exceptions, some spectacular such as Folly Lane (4.2.1), others more modest such as Brandschuttgräber from Winchester (5.2.1), the data on the cremation process itself is limited to the human bone collected from the pyre. Unlike neighbouring provinces few cremation burials contain pyre debris, though the number is higher in more recent excavations (3.5). This, coupled with previous neglect of this residue where noted usually obscures its significance. However sufficient 147

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

The shift to inhumation reduces the evidence available for the ceremonies preceding or succeeding interment and the burial itself gives the main opportunity for analysis. In the transition to inhumation the significance of the coffin as a marker of difference, either by its presence or material (stone of varying types, timber, or lead-lined) has been neglected. It is near impossible to calculate the relative cost of a stone, lead-lined and / or massive timber coffin, but the extraction of raw materials, the creation and transport of objects of this type would seem likely to incur very much greater expense, than the use of commonly occurring ceramics in early Roman rituals. Lead liners or stone coffins were relatively rare in Winchester, which was quite distant from the sources of raw materials; elsewhere they were more common (5.5; 7.4). Evaluation of this aspect would be helped by analysis of the provenance, production and distribution of stone coffins, a rarely studied topic (cf. Taylor 1984). Hayward’s (2009) provenancing of commemorative stonework of the early Roman period shows the potential of such an approach (cf. Savay-Guerraz 1990). Some burials were also elaborated through the use of ‘plaster’ within the wrapping of the corpse and as packing around it in the coffin, but the implications for use of resources differ significantly according to materials (chalk, gypsum etc).

examples occur to show that the pyre is a place for the spectacular and / or habitual destruction of objects across the province; inter-regional differences previously proposed are not supported (3.5). The vast majority of buriers were able to muster sufficient resources to fuel a pyre which burnt the corpse fully, to judge from the colour and condition of cremated bone across the samples of cremated human bone, though it has been noted that Roman period cremations in general show a less even or complete degree of burning in relation to other periods (4.6, 6.2). In general analyses presented here have depended on ‘traditional’ evidence, meaning that burial container, number and type of grave goods provided the principal basis for comparison between assemblages. Containers for the dead varied in all the contexts studied, with an occasional association between burials with larger grave good assemblages and non-ceramic containers for the cremated bone; only the metal fittings for wooden casket burials hint at the sophistication of containers in perishable materials (4.2-5, 5.2, 8.2). The clearest differentiation between burials lay in the number and variety of grave goods, with a small number of graves being characterised by larger numbers of ceramics, especially samian dishes and bowls and sometimes also other items, typically glass and occasionally metal vessels.

Grave goods are much less common than in the previous period; the Winchester, especially the Lankhills sample, is atypical in their high frequency. However objects which bear connotations of luxury or are possible emblems of rank occur sporadically within this more limited sample, for example crossbow brooches and belts. Other manifestations include female ornament in rarer raw materials (gold, ivory, silver) or object forms which are rarely documented in settlement contexts, for example some glass of the glass vessels from Lankhills (5.3.2). The presentation of the corpse (5.3-5, 8.3.1) during funerary rituals is poorly understood and almost certainly underplayed, hinted at in tantalising scraps of fabric or leather in plaster burials or as mineralised impressions on dress fittings and ornaments; the occasional cases of exceptional preservation conditions for organic remains are important for better understanding of this aspect of late Roman ritual (8.3).

Very limited direct evidence for burial monuments survived, though limited intercutting between graves in most cemeteries suggests that most were marked. A small number of timber and stone-built tombs were documented, though with partial exceptions southern and central Britain is poorly represented in terms of textual commemoration. Among monuments greater numbers and a wider diversity of types were noted in the hinterland of St Albans (6.2). The limited dating evidence suggests that the second century saw the apogee of commemorative monument building, for the countryside at least. In occasional instances monuments had the potential to accommodate participation in ritual by large numbers or people (6.2.2; 6.4.4). For the late Roman period inhumation is not quite the universal ritual in the sample area (the significance of the proposed archaeologically ‘invisible’ mode of burial is likely to be reduced). The occasional occurrence of cremation in the case study area is fairly typical for the province, though at times it accounts for a higher percentage of burials elsewhere. It is near impossible to rank cremation in terms of investment of resources and energy in relation to contemporary inhumations; diversity within cremation burials may be more significant (as seen in Winchester; 5.3). Like early Roman inhumations, it too served to differentiate within the burying community at scales varying from subgroups within mixed cemeteries, for example Alington Avenue, Frocester Court, or Lankhills, to the majority rite in single cemeteries (for example some burial plots south of the circus at Colchester or at Brougham).

For the late Roman period too the general absence of intercutting suggests that most burials were marked above ground but again little direct evidence survives. The view that layouts of the type illustrated by most of Winchester’s late Roman cemeteries manifest the development of the socalled ‘managed cemetery’ was rejected and with it the view that this bears on increased intervention by, and therefore the capacities of, late Roman civic authorities (5.3.5). The disappearance of the epigraphic habit should not be confused with a decline in monument building. Evidence in the form of surviving fragments of superstructure or foundation cuts shows the continued marking out of some burials through substantial and prominent monuments into the fourth century AD, especially on urban sites. The types include barrows, masonry-built mausolea in various forms including temple 148

CONCLUSION

over time within the broad periods into which analysis was divided (cf. Bradley 1988). Glass and bronze vessels are much more commonly buried with these ‘privileged burials’ than in graves in general, although other objects, for example samian pottery are not. By contrast some related objects to which prestige is traditionally attached were not present in ‘privileged burials’, for example decorated samian. Instead, it was argued that through their evocation of different areas of social practice these assemblages gave insights into social and cultural change (8.3-8.4). The reading of individual objects is argued to be conditioned by others in the burial assemblage, the cumulative allusions being to a constellation of behaviours which corresponded quite closely to a Tacitean formula of humanitas, in particular dining, leisure (in specific forms such as bathing, perhaps also hunting), and care for and presentation of the body. This differentiation is based less on conspicuous destruction of scarce or exotic items and more on the possession of cultural knowledge, of methods of consuming food and drink, of presenting the body and in behaviour, of lifestyles refashioned around Roman modes. From the study of public spaces in towns of the early-mid Roman periods Revell (2009) has argued that their architecture functioned not only as general symbols of euergetism but also to shape certain behaviours characteristic of the adult elite male in particular. To a degree the activities evoked (often pars pro toto) in burial practice overlap with these behaviours (for example the emphasis on bathing), but reveal complementary expectations for behaviour in private space. In the very few cases in which sex could be attributed to cremated remains these assemblages did not generally reveal significant differences in the treatment of men and women or individuals of different ages. Women were therefore associated in burial with metaphors for the same types of cultural savoir faire as men. While chronological change deserves fuller investigation, this evidence must be taken into account in the characterisations offered of women as less involved in the public sphere and therefore engaged less intensively with wider cultural change (e.g. MacMullen 1990: 63; Terrenato 1997).

tombs and fenced or hedged enclosures (6.4.4, 7.2.1). In a rural setting the association of some burials with substantial elements of settlements or fields endowed some burials with an alternative form of monumentality. The description of changing rituals over time shows that burial retained to the end of the Roman period its potential to articulate difference between the mourners through the medium of the bodies of the dead and associated rituals related to the resources which could be used; this difference was sometimes sustained through different stages of funerary rites, allowing for the varying visibility of activity preceding interment. As was noted in the introduction (1.3.1), these differences have been previously associated with specific social groups, albeit broadly defined; the equation of variability in burial treatment, especially grave goods, with varying social status was disputed on methodological and theoretical grounds. Some help is afforded in other provinces of the empire where epigraphy identifies the commemorator or commemorated, but funerary inscriptions from the sample areas are rare and none of those known here or elsewhere in Britain are directly associated with the dead in the primary context in which they were raised. Spatial setting, for example the same peri-urban context as the villa, or object symbolism, for example the cross-bow brooch, with its possible connotation of office-holding, may indicate a more specific identity for the buried or buriers. The difficulty of translating the burials with the most elaborated treatment into those of elites is however less important than the continued dynamic use of burial to differentiate by the continuous introduction and dissemination of ritual innovation. The broad date ranges for most burials make it difficult to assess how far a mechanism of emulation accounts for wider dissemination, or precise assessment of the relationship between burial and other spheres of social practice in this respect (cf. Cannon 1989; Miller 1982). In the absence of a widespread commemorative portraiture tradition, at least away from Britain’s frontiers and legionary fortresses, means that funerary sculpture cannot be exploited for insights into Selbstdarstellung (1.3.2). However the funerary display most accessible to us comes from the transient presentation of the corpse or its cremated residue, its ephemeral manifestation to participants before and during its placing in the grave. Chapter 8 discussed certain ‘privileged burials’ from this perspective. The term, preferred in recent scholarship on Roman Gaul (‘sépulture privilegiée’; Crowley 2011: 1956) has fewer problematic connotations than ‘elite’ or ‘aristocratic’ tombs and emphasises the action of burial in generating relationships as well as reflecting them.

In important respects, in particular the deposition of large quantities of grave goods, the early Roman burial ceremonies described from southeast Britain did not follow contemporary practice in Rome (8.4) and show similarities to burial practices of the pre-Roman period. This apparent tension between objects and depositional mode might, at least in the formative period of provincial cultures, reveal the Janus-like property required of elite provincial actors, nodding towards different cultural traditions in the same practice, or an adaptation of imported objects according to local uses (Terrenato 2005). However the degree to which such a burial tradition would be judged as alien is not certain; the perspective and experience of participants needs some consideration. The similarity of burial practice (as show by grave good assemblages) across a very extensive area of the northern provinces has been attributed to the interaction of provincial elites (8.4) and begs the question of how a ‘Roman’ burial rite is defined. What is perceived as such is likely to vary both chronologically

For the early-mid Roman periods the examination of larger burial assemblages from within and beyond the sample areas showed the recurring presence of certain artefact types. Comparison with non-burial contexts suggested that a straightforward reading of these as prestige goods was unsatisfactory, since relatively few can be argued as rare or exclusive or of restricted accessibility, though for some their availability varied 149

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Garrow (2012: 196) propose that many metal objects bearing decoration in late Iron Age graves, including weapons, mirrors and horse gear, had accumulated a substantial ‘biography’ among the living, as evidenced by their age at time of burial, or traces of use, repair or refashioning. They thus embodied and evoked personal and group histories and relationships. This seems a productive approach to Roman burials, even if this property further complicates the assignation of status to a specific individual where the object stands for relationships rather than personal attributes. Inscriptions on a small number of objects in burials also bear on this (Biddulph 2006a); as well as names some record gift giving, for example the vessel dedicated by Regillinus of the dendrophori of Verulamium from a burial at Dunstable Friary Fields (RIB II.8 2503, 114); others, such as the brooch from Lankhills (5.3.2) or bone plaque from York (RIB II.3 2441, 11) bearing exhortations, are also likely to derive from gifts. The deposition of objects, including ceramics, especially samian, metal vessels and personal ornaments (including cross-bow brooches) in graves in a much worn and sometimes broken state, many decades or centuries later than their manufacture is a more common phenomenon and may also indicate a similarly complex history (e.g. Cool et al. 2004; Cool in Booth et al. 2010: 282-3; Wallace 2006; Willis 2011: 222).

and spatially along the trajectory of assimilation into different provincial societies of different aspects of Roman practices. Thus the interpretation of these burials as ‘conservative’ (1.3) does not acknowledge the polyvalency of these deposits and the range of connotations which could be drawn from them, especially with regard to changed lifestyles (see above). For the late Roman period attention was focused on general trends in the mise-en-scène of the body more than specific questions of religious or ethnic affiliation which have been the centre of many studies of this period. Although the general range of objects deposited was similar to earlier periods, increasing emphasis lay on the cultivation of physical appearance more than dining, hospitality, literacy or other areas symbolised by the objects deposited in earlier assemblages. This now, occasionally, included objects which may symbolise office-holding on the part of the deceased; the association of some examples with children suggest that they do not always have a straightforward rank-denoting character for the dead individual (8.3). Occasional evidence for costume or footwear, or of implements related to cosmetic preparation or application, indicates that these ornaments were part of a presentation that emphasised the beautiful corpse. Significantly greater emphasis was also placed on the framing and packaging of the body, whether in lead, stone or timber coffin distinguished by its thickness of wood or elaboration of fittings, and rituals such as plaster burial. The not yet fully published burial of the ‘Spitalfields princess’ epitomises some of these characteristics.

9.3.2 Space and Society: burial context Contextual information also allows this differentiation in burial practice to be mapped onto its spatial setting, and to identify multiple theatres for burial display. The case studies showed that the (archaeologically visible) burial practices developed in tandem across different contexts. As others have noted in other contexts, so here with occasional exceptions it was difficult to find evidence for distinctive rural and urban ‘faciès’ of burial practice (Blaizot et al. 2009; Esmonde-Cleary 1992; Ferdière 1993b). In this respect burial does not sit comfortably in a scheme of structured oppositions which situates it with the areas of provincial life less articulated with Roman culture (1.2), although the broad dating of many burials may mean that adoption and dissemination of innovations cannot yet be traced closely enough to state this too confidently.

As before burials of this kind in Britain show common characteristics with equivalent examples from neighbouring provinces, although there are also some significant differences of emphasis; objects associated with dress and presentation of the body are complemented by others which are (much) less significant in Britain. Not only are the likely symbols of authority such as crossbow brooches and belt sets more common in northern Gaul, as well as dress ornaments from distant places including the transfrontier zone; the inclusion of weapons is also a recurring characteristic of burials, especially in northern Gaul; the ‘rhetoric’ of landscapes won by force of arms and clearance of forest and wild animals symbolised by the presence of swords, axes and spears (Swift 2000; Theuws and Alkemade 2000). The complex accoutrements of late Antique dining also occur more commonly beyond Britain, where they are more likely to be met in structured deposits (8.3.2).

As well as sharing similar rituals an equivalent variability in those rituals was also documented across different settlement types, for the early Roman period in both samples, and also for urban and rural settings in consideration of late Roman Winchester and environs, as established from containers for the body, objects and monuments (chapters 4 and 5, 6.2-3). On this basis it is difficult to see differences in the hierarchies presented by the burying communities in different contexts. This allows some further comment on the character of urban burial. From similar kinds of evidence Martin Millett and Martin Pitts have drawn opposite conclusions concerning the ‘ordinary’ burials of urban centres furnished with small numbers of pots and occasionally other items, typical of the Winchester and St Albans case studies. Millett (1993) interprets modestly furnished burials in conquest period St Albans as those of deracinated urban migrants lacking

There is further scope for exploration of the range of allusion and metaphor offered by grave goods. Some objects with significant allusive potential have not been considered. For example Gottschalk’s (1996) discussion of spindle whorls in graves, especially those in rarer materials, emphasises the gender and status-related symbolism of items related to textile production. The possession by an artefact of specific rather than generic resonance for the mourners, based on its ‘biography’, also deserves further consideration. For example Gosden and 150

CONCLUSION

burial evidence, assemblages and monuments, supplements characterisation from other evidence of the role of these places as social and ritual as well as economic centres and makes it evident from the early Roman period.

networks to take care of more than minimal burial ritual (4.2.4). Martin Pitts (2007: 707-8) attributes the modest furnishing of urban and small town burials in early Roman south-east Britain to the exercising of responsibility for burial ritual by collegia or similar corporate groups, characteristic of better documented Roman cities. Arguably both these models place too much emphasis on urban burials in isolation: from the case studies presented here, revealing similar variability across contexts, urban society as presented in burial, does not appear as more competitive, highly stratified or structurally more complex than other settings.

Examination of the more local context of monuments, especially for towns (7.2) also gave insights into the use of monuments and space. As in other provinces the cemeteries of Romano-British cities spread out along the routes on the urban periphery but they do not in general take the form of monumentalised Gräberstrassen (cf. 1.3.2). The area close to the town gates, probably on the pomerium, was not a focus of burial display, either as expressed through monuments or rituals, either in Britain or more widely in the Roman north-west, with some exceptions (5.2.2, 7.2.2) and caveats; the epigraphic evidence which might clarify the analysis of any such preference is lacking and this area has rarely seen excavation. As for the exploitation of the road frontage, the deposition of larger burial assemblages does not seem to have favoured this zone (4.2.3, 5.2.3). The placing of monuments was sporadic rather than a continuous facade, occasionally punctuating the roadside sometimes out into the wider hinterland (see above). Monuments sometimes clustered in positions made more visible by the combined effects of roads and road junction and their situation wider landscape, for example at Colchester or Gloucester (7.2.2). A much greater monumentality than average, reflected in the abundance and variety of commemorative markers, characterised the cemeteries of towns associated with the army and colonial foundation, provincial administration and a nodal position on key routes for moving commodities, especially York, London and Lincoln, to a lesser extent Colchester. This might suggest urban life more akin to the competitive dynamics in some Roman cities beyond the province where there were multiple routes to social mobility (military service, commerce etc.). However any interpretation of London or York from this perspective must also note their limited funerary monumentalisation in comparison to the hundreds of inscriptions as well as architectural fragments and in-situ monuments from the major riverine cities of Gaul and Germany.

The regional case studies, put into a wider context (chapter 7) also allowed further analysis of the distribution of burial ‘display’, as defined by qualitative and quantitative criteria. Context-related differences have been widely repeated in earlier scholarship, principally to assert that such display was more characteristic of the Romano-British countryside than other spaces (1.3.1). To a limited extent the case studies support this, especially for the early Roman period. More burials contained large quantities and diverse types of grave goods in a rural than an urban setting. In the case of the Catuvellauni, large-scale monuments were also more frequently attested in the country (6.2-3). To a lesser degree the same could be said of small towns in relation to St Albans (Ch. 4). Consideration of the non-burial context helps to develop this observation. While, as noted, items deposited in burials were generally in widespread circulation, the degree of availability varied significantly between contexts. Samian in particular was a more rarely available commodity in minor centres and in the countryside but was deposited with greater frequency in burials, not only those with the largest assemblages but also larger groups (4.2, 4.6, and 5.2). However classification of burial display by site type alone obscures significant spatial patterning (7.3 and 7.4). What arguably emerged more strongly from these analyses was the existence of several zones of display. The immediate hinterland of civitas capitals has been demonstrated to be one of these in the early Roman period, the clustering of relevant assemblages being similar to that of later villas around towns. Close by early Roman Verulamium this applied not only to the residues of the transient display of the funeral itself; commemoration took more lasting form through substantial and architecturally varied monuments, placed with a concern for wider visibility (6.2.4). Comparison of rural and urban contexts from late Roman Winchester and environs (admittedly of very different sample sizes) showed the city also at the centre of a cluster of tokens of burial display, especially the containers for the dead, but with greater emphasis on urban cemeteries proper over the peri-urban hinterland. Here the potential emblems identified of rank and luxury, rare in settlement contexts, are not yet paralleled in rural burials. However overall the evidence base is too limited to assert this conclusion too strongly (7.4).

For the later Roman period a less clear picture is visible of this roadside interface between living and dead on the urban margins. Evidence from Winchester and beyond shows funerary monuments of varying size and sophistication continuing to be erected in late Roman cities in Britain (5.6, 7.2); this is more widely characteristic, as the plans for cemeteries across the north-west provinces assembled by Schmidt (2000) show. In the absence of excavations for this period where the roadside spaces of cemeteries have been directly investigated it is difficult to test its continued attraction but some monuments seem to be placed with a concern for visual prominence (7.2.1); the differentiation of the dead by these monuments continued to be signalled to a wide audience.

Another zone of display identified (Hertfordshire / Catuvellauni sample) is that of the small town and its environs, especially on civitas boundaries (7.3.3). The 151

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

further assessment where a sample of burials can be established based on dating criteria independent of grave goods.

There is considerable potential for extending the scope of analysis of Roman urban cemeteries as spaces of display through assemblages and monuments to other provinces at the various scales suggested. This can partly be achieved by analysis of older discoveries, where contextual data allow some general replacing of monuments in the setting in which they were raised; the examples of Cologne (Gabelmann 1987; Spiegel 1994) and Mainz (Boppert 1992a; 1992b; Hope 2001) show what can be achieved. As in Britain where urban environs are not overlain by modern development prospection can reveal significance further evidence of monuments in their setting, as the various prospection campaigns on the towns of the Tiber valley and Portus have recurrently demonstrated (e.g. Germoni et al. 2011: 237-8; Hay et al. 2010; Johnson et al. 2004: 96-7; Keay et al. 2005: 28890). For studies of the regional dimension to funerary display the approach applied here can be extended to the much larger samples of funerary epigraphy, architecture and sculpture. A glance at the results of past studies from Gaul, for example, gives contrasting impressions. Though argued to have a rural distribution, the tower tomb type characteristic of south-west France, the ‘pile funéraire’, also clusters around towns, especially Auch and St Bertrand-de-Comminges, even if immediate settings are rural (Sillières and Soukassian 1993: Fig. 1, 300). By contrast mausolea in south-eastern France are closely tied to the ‘trunk’ road network, especially the via Domitiana, but with no obvious clustering in the immediate environment of the chartered towns (Roth-Congès 1993: Fig. 1, 391). Wightman’s comparison of urban and rural epigraphic, sculptural and architectural assemblages from Gallia Belgica suggests that neighbouring civitates are characterised by very varied use of urban and rural space as locales of burial display (1985: 163-8). For furthering all of these analyses at different scales, landscape distribution, site association and visibility GIS offers a so far little exploited tool in comparison to other periods.

The spatial associations of the dead can also be explored from an alternative perspective, in particular through considering burial within settlement space characterised as a medium through which experience and social relationships are continuously negotiated. Different locations on site boundaries were identified as consistently favoured for burial, even if the precise chronological relationship between burial and associated feature is often difficult to establish. These included the peripheries of settlement enclosures and fields, not only their ‘rear’ (in the few cases where ‘front’ and ‘back’ could be established), but also entrances, the ditches of roads and trackways, and with features surviving from abandoned settlements and prehistoric sites. Other features such as ‘corn driers’ and kilns were also sometimes chosen as burial places (6.4.3). The placing of the dead is argued not to be expedient, pragmatic or fortuitous. Rather the choices integrated the space of the dead with the daily routines of the living. This close association with other spaces, ditches, wells, work installations, enclosure features and so on may in part explain why the duration of burial in particular locations was usually fairly short, only a few decades. This contrasts with the longer use of cemeteries from some earlier or later periods, for example, the Bronze Age or the Anglo-Saxon period, although in some cases a sense of space as particularly adapted to the dead seems to have endured through use over several centuries, punctuated by long intervals (6.4.1). The sample considered demonstrated the well-known differentiation of burial location according to age, infants being buried much closer to or within inhabited spaces and buildings, adults in peripheral positions. Not considered in this study, but undoubtedly with significant potential, is the relationship between burial space and gender; in the modest sample of rural burials from Hampshire, for example one gender dominated most of the small groups of burials, but this remains to be more widely tested (Pearce 1999). This may be the rural counterpart of the gender imbalance in late Roman urban cemeteries though, as Davison (2000) shows, the latter requires considerable source evaluation and a much larger sample of data has subsequently accumulated to test this characterisation. Apart from occasional observations of this type burials and their spatial setting are as yet unexploited in the limited literature on the engendering of rural settlement space (cf. Hingley 1989; Taylor 2001).

In previous scholarship near invisibility (to contemporaries) had been used as a spatial property to attribute a particular category of burials, those outside formal cemeteries, on farm peripheries, often in or adjacent to enclosure ditches, to low status groups, slaves, workers, outcasts and so on (1.3.1). This group has been near invisible to archaeologists too, the occurrence of burials mainly in small groups having perhaps deterred previous analysis; but although some statistical approaches may be impeded by sample size, this does not mean that such burials do not reward closer scrutiny, especially in connection to their relationship to the settlement landscape. From the analysis of funerary rituals and re-examination of spatial setting this classification as individuals of the lowest social status is challenged (6.4). As already stressed, such burials often represent the few known from a rural context (2.5) and they are as likely as ‘typical’ burials on urban and or small town sites to be buried within containers and, probably, to be accompanied as often by grave goods. It should however be noted that the Hampshire sample’s composition on which this is based depends in part on the presence of grave goods for its compilation (6.3); this argument risks circularity and needs

The historical context for the evolution of visible burial practice as applied to an increasing number and proportion of individuals is difficult to specify. This placing of the dead was one facet of the increased subdivision of space through physical and symbolic boundaries on and around rural settlements. This association can only be claimed in very general terms; the subdivision of the rural landscape into farms, paddocks, fields and tracks had been underway for a much longer period, and the placing of increasing numbers of burials in the interstices of these features is a gradual process not occurring everywhere (e.g. Chadwick 152

CONCLUSION

Kent or southern Dorset, to a lesser extent the CotswoldsSevern region (2.3); such a study would continue to be impossible for the frontier zone, even if the sample character is slowly growing. Outside Britain there are rich datasets where it would be possible more consistently to compare evidence from rituals and monuments. The Treveran area is a very good example of a region which offers the opportunity to compare epigraphic, sculptural and architectural evidence with evidence of rituals over the long term, although an increasing corpus of high quality data of varied types is also available in other areas of Gaul (e.g. Blaizot et al. 2009; Crowley 2011; papers in Ferdière 1993a).

2009: 129-33). Groups of burials in such locations became a little more common through the Roman period (6.4.5), though this is regionally very diverse and not true, seemingly of south-east England where late Iron Age and early Roman rural burials are more common (2.3). One effect of this emphasis on boundaries was increasingly to associate settlement and group identities via the places with recurrent use for burial over three to four generations, in a context where rural populations came under greater social and economic pressures as elites aimed to extract a greater surplus. Explanation of this trend must take note also of the wider dimension to this phenomenon. Very similar trends can be identified in rural Roman Gaul, i.e. that ‘cemeteries’ are small and of short duration, with a partial exception for the late Roman period where some ‘pérennisation’ of the burial space is observed in a small number of cases (Blaizot et al. 2009: 255-62; see also Ferdière 1993a: passim).

There is also significant scope for greater methodological sophistication. The patterns proposed in the analyses presented here are sometimes impressionistic and the quantitative methods simple, partly to accommodate data of varying quality. In future analyses more statistically sophisticated techniques should be applied to test the strength of individual patterns, to map burial variability in a more complex and nuanced way and to explore the associations documented (Cool and Baxter 2005). The possibilities for GIS to assess spatial relationships have already been noted.

9.4. Last words This study has proposed that burial evidence can both complement and modify interpretations of provincial Roman society based on other forms of archaeological evidence. With a focus on specific characteristics it has aimed to show that burial can shed significant light on wider social and cultural dynamics and the construction of identities across time for the Roman period, neither limited to the formative period of provincial culture nor the last days of religious conversion and population change. The study has aimed to draw on some non-funerary aspects of context but it is essential to emphasise that only some of the rich burial data from a Roman province potentially have been exploited here. There is significant scope to develop and extend a context-based exploration of the relationships between the living and the dead in Roman Britain and elsewhere, even if, despite attempts to persuade (Morris 1992), this promise is not yet acknowledged; the most influential and wide-ranging recent study of mortuary archaeology scarcely refers to classical examples (Parker Pearson 1999a).

Part of the argument made here was based on observations concerning the occurrence of objects placed as grave goods in other archaeological contexts. This contextualising data has been drawn on to only a limited degree. Recent work on ceramics has shown how the choice of grave goods can be informed by comparison across contexts (8.2-3), and other classes of material offer further scope here (glass, small finds etc.). This will help strengthen arguments over the identification of indicators of ‘display’ and the the connotations of objects in burial in broader systems of meaning (8.2-3). Approaches have been suggested by studies of prehistoric contexts (e.g. Fitzpatrick 1997a; 2000; Parker Pearson 1999b), but the Roman period possesses a much fuller range of contextualising data. As well as quantitative comparison, visual and textual evidence from beyond the grave gives opportunity to assess the context-specific meanings attached to objects and practices which can be exploited more fully. This applies strongly to the dimensions of identity which have been overlooked in this study, in particular age and gender.

Much of the argument presented here has been based on two case studies, selected for the availability of data across different contexts, especially site types, but with significant lacunae. The lack of evidence from key phases of ritual has been emphasised enough not to need repetition here. The key rural samples were heavily outweighed by those from minor centres and towns. Even the characterisation of the latter is founded on a slim evidence base; without the excavation of the Folly Lane tomb-sanctuary for example, funerary monumentality would have looked very different for St Albans, a city from which a single epitaph has yet to be recorded. The location of the case studies from southern Britain in regions where cremation is the majority tradition and deposition of grave goods frequent means they are not fully representative of the province, in particular for the early Roman period. Only limited consideration was given here to the late Roman period. Fieldwork of the last decade means that other regions would now satisfy the criterion of availability of data from different contexts over time, for example Colchester and environs, north-west

At first view age and gender appear to be differentiated to a much lesser degree in the early Roman burial rituals than in late Roman urban cemeteries where recent study has shown strong associations with these dimensions of identity in the surviving evidence for rituals (1.3.2, 5.3-4). This is in part due to the limitations on osteological sexing inflicted by cremation, which has reduced the sample size considerably. However study of burials from the Iron Age to the fourth century in Dorset show patterning related to age and gender applying throughout the Roman period and analyses of the third century cremation burials at Brougham reveals the unexpected associations with age and gender of objects which might not be anticipated to have such associations (Cool et al. 2004; Hamlin 2007). 153

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Integrating analysis of these dimensions in the study of future samples of first and second century AD cremation burials should be a priority of future work. The quality and quantity of skeletal data will also allow much more rigorous assessment of the relationship between gender and spatial setting for burial. The observations on gender and age are also a reminder of the increasingly rich human skeletal dataset which allow ‘representations’ of identity through ritual to be compared to the variations in lived experience as interpreted in the skeleton (1.3.1; also Pitts and Griffin 2012). The flagging of alternative approaches, datasets left out and potential methodological improvements perhaps risks ending on an anticlimactic note. However to emphasise the wealth of future possibilities is a reminder of the richness of the resource of Roman burial data and its context in Britain and beyond.

154

CONCORDANCE Concordance to cemeteries, burials and pyre sites (Britain) referenced in text and appendices Site Abingdon Vineyard Abergavenny Aldwincle Alton Ancaster

County Oxon.

Further detail App. 5.2

Northants. Hants. Lincs.

App. 6.6

Andover S. Distributor Rd Appleford Arbury Rd, Cambridge Ardleigh Arkesden Arrington Asham Ashford Brisley Farm Ashton Aston Astwick Avisford Ayot St Lawrence, nr church Babraham Institute Baldock Baldock Three Valleys Pipeline diversion Baldock Icknield Way East Baldock California

Hants. Oxon. Cambs. Essex Essex Cambs. E. Sussex Kent Northants. Herts. Beds. W. Sussex Herts. Cambs. Herts. Herts.

App. 6.7 App. 5.2 App. 5.2, 6.3 App. 5.2 App. 6.3 App. 6.3 App. 5.2 App. 5.2

Herts. Herts.

App.3.1

Baldock Clothall Rd Baldock Convent of Providence Baldock Icknield Way East & environs Baldock Mercia Road Baldock Royston Rd (inc. Stane St, Sale Drive, Yeomanry Drive, Icknield Way Roadside) Baldock ‘The Tene’ Baldock Walls Field Baldock Wallington Rd

Herts. Herts. Herts.

App. 6.3 App. 6.3 App. 2.1.3

Herts. Herts.

App. 2.1.3 App. 2.1.3

Herts. Herts. Herts.

App. 6.3

App. 6.3 App. 6.6 App. 5.2, 6.5 App. 5.2

Balksbury Bancroft Barnack Barley Homestall Farm Bartlow Hills

Hants. Bucks. Cambs. Herts. Essex

App. 5.2 App. 5.2, 6.3 App. 5.2

Barton Court Farm Barton Deadman’s Hill Barton, Lord’s Bridge Basingstoke Basingstoke Worting Baughurst Bayford, Sittingbourne Beadlam Beckfoot Bedford Beechbrook Wood, Hothfield Ben Bridge, Chew Valley Berwick Down Biglis Billericay Binchester

Oxon. Cambs. Cambs. Hants. Hants. Hants. Kent N. Yorks. Cumbria Beds. Kent Avon Dorset S. Glam. Essex Durham

App. 5.2 App. 6.5 App. 6.5 App. 6.6 App. 6.6

App. 6.3, 6.5

App. 2.1.3

App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2

155

Reference Frere 1990: 334 Worrell 2004: 321-3 Jackson & Ambrose 1976 Millett 1986 Cox 1989; Burnham & Wacher 1990: 235-40; Wilson 1968b Jennings 2000 Hinchcliffe & Thomas 1980 Fell 1956; Frend 1956 Brown 1999 Hull 1963b: 39 Taylor 1993 Curwen & Curwen 1930 Stevenson 2012 Burnham & Wacher 1990: 279-81 Rook et al. 1982 Page & Keate 1908: 4 Haverfield 1935: 49; Roach Smith 1848: 123-4 Archaeologia 53, 1892: 253 Timberlake et al. 2007 Phillips 2009 Thorpe et al. 2004 Burleigh 1993; Frere 1989: 298 Burleigh 1982; Burleigh & Fitzpatrick-Matthews 2007: 47-51; Murphy 1990 Stead & Rigby 1986: 63-73 Westaway 1976 Burleigh & Fitzpatrick-Matthews 2007 Burleigh & Fitzpatrick-Matthews 2007: 69-73 Burleigh & Fitzpatrick-Matthews 2007; North Herts Archaeology Service archive Stead & Rigby 1986 Westell 1930 Burleigh 1993: 44; site archive; Burleigh & FitzpatrickMatthews 2010 Wainwright & Davies 1995 Williams & Zeepvat 1994 Simpson 1993a Taylor 1956: 138 Gage 1834; 1836; 1840; 1842; Hull 1963b: 39-45; Eckardt et al. 2009 Miles 1985 Dunning & Jessup 1936: 48; Taylor 1993: 224 Dunning & Jessup 1936: 49; Taylor 1993: 224 Franks 1852: 9 Ellaway & Willis 1934: 88 Anon. 1993 Payne 1877; 1886 Neal 1996 Bellhouse 1954 Page & Keate 1908: 5 Barclay et al. 2006: 38 Rahtz & Greenfield 1977: 82-91 Wainwright 1968: 105 Robinson 1988 Rudling 1990 Burnham et al 2008: 283-5; Wessex Archaeology 2008:

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Binsted Birchanger Lane Birdlip Birdoswald Bishopstoke Bishop’s Stortford Dunmow Rd Bishops Stortford Cannon’s Close, Seymour Estate Blackhorse site, Letchworth Black Notley, Braintree Bletsoe Bokerly Dyke / Woodyates Borden Borough Hill, Daventry Boscombe Down Bourn, Moulton Hills (or Arms Hills) Bourton Grounds Bower Rd, Sneeth Boxfield Farm, Chells Manor Boxmoor Bradley Hill Braughing ‘A’ Braughing ‘B’ Braughing Lark’s Hill Braughing Skeleton Green Braughing Station Rd Bray Brettenham – Melford Meadows Brighton Springfield Rd Broadwell Spring Brook Brough Brougham Burntwood Farm Caerleon Lodge Hill Cambourne Jeavons Lane Cambridge, Ridgeon’s Gardens Camelon Cannington Canterbury Castle Area II, Rosemary Lane Car Park Canterbury Cranmer House Canterbury Dane John barrows Canterbury Stour Street Carlisle Botchergate Cassington Catsgore Chalton Charlton Mackrell Chedworth Chichester St. Pancras Chichester Theological College Chignall Chilmark - Eyewell Farm Choseley Farm Cirencester Bathgate Cirencester Oakley Cottage Cirencester Old Tetbury Rd Cirencester Phoenix Way Cirencester Tar Barrows Claydon Pike

Hants. Essex Gloucs. Cumbria Hants. Herts. Herts. Herts. Essex Beds. Dorset / Wilts. Kent Northants. Wilts.

App. 6.7 App. 5.3; 6.3 App. 6.4 App. 6.7

Moss Eccardt 1988 Drury 1976: 113; Hull 1963b: 163 Dawson 1994 Hawkes & Piggott 1947; Rahtz 1961

App. 6.3 App. 5.2 App. 5.2

Jessup 1959: 23-24 Brown 1977 Anon 1997; Burnham et al. 2003: 350; Burnham et al. 2005: 439-441 ; Pearce et al. 2008 Taylor 1993: 224; Liversidge 1977: 23

App. 6.5 App. 5.2

Cambs.

App. 6.2

Bucks. Kent Herts. Herts. Somerset Herts. Herts. Herts. Herts. Herts. Berks. Suffolk W. Sussex Somerset Hants. Cumbria Cumbria Hants. Gwent Cambs. Cambs. Stirling Somerset Kent

App. 5.2 App. 5.2

9-11 Millett 1974b Medlycott 1994 Staelens 1982 Wilmott 1993; Wilmott & Hirst 2009 Toller 1977: 34; Haverfield 1900: 309-10 CBA Group 10 Newsletter 10: 4 Taylor 1957: 219

App. 5.2 App. 6.3 App. 2.1.3 App. 6.2 App. 6.3 App. 5.2 App. 5.2 App. 6.4 App. 5.2

App. 5.2; 6.7 App. 5.2 App. 2.1.3 App. 5.2

Kent Kent Kent Cumbria Oxon. Somerset Hants. Somerset Gloucs. W. Sussex W. Sussex Essex Wilts. Hants. Gloucs. Gloucs. Gloucs. Gloucs. Gloucs.

App. 2.1.3

App. 2.1.3

Gloucs.

App. 5.2

App. 5.2

App. 5.2 App. 5.2 App. 5.2 App. 2.1.3 App. 2.1.3

156

Johnson 1975 Booth & Diez 2006: 19-20 Going & Hunn 1999 Neal 1978 Leech 1981 Partridge 1977 Partridge 1977 Page 1914: 150-51 Partridge 1981 Partridge 1977 Wilson 1971: 284 Mudd 2002 Kelly & Dudley 1981 Leech 1980: 338 Brown 2009 Jones 1977 Cool et al. 2004 Fasham 1980 Evans & Maynard 1997 Wright 2009: 49-50, 89-90 Alexander & Pullinger 2000: 45-7, 53-7, 79-80 Breeze & Rich-Gray 1980 Rahtz 1977 Bennett et al. 1982 Frere et. al 1987 Wacher 1995: 200-1 Bennett 1980 Zant et al. 2011 Harman et al. 1981; Taylor 1937: 237 Leech 1982 Frere 1957 Leech 1980: 358 Goodburn 1979b Down & Rule 1971 Down & Magilton 1993 Clarke 1998 Fitzpatrick & Crockett 1998 Morris 1986a McWhirr et al. 1982 Reece 1962 Holbrook 2008a: 109-31 Wright & Hassall 1974: 461 Holbrook 1994: 82-83; Holbrook 2008a: 308-10; Winton 2009 Miles 2007

CONCORDANCE AND APPENDICES

Coggeshall Colchester Balkerne Height Colchester Barnhall Colchester Butt Rd Colchester ‘Child’s grave’ Colchester Circus / Garrison ‘alienated land’ Colchester Gurney Benham House Colchester Handford House Colchester Joslin Collection Colchester Lexden / Lexden tumulus Colchester 6 Lexden Road Colchester St Clare Drive Colchester St John's Abbey & Maldon Rd Colchester Stanway Colchester Turner Rise Coleford Corbridge Bypass Cowdery’s Down Crabtree Lane, North Lancing Crambeck Crendon Cross Farm, Harpenden

Essex Essex Essex Essex Essex Essex

Curbridge Dalton Parlours Daneshill Dellfield, Berkhamsted Densworth, Funtington Derby Racecourse Dorchester Alington Avenue Dorchestr Little Keep Dorchester Maiden Castle Dorchester Poundbury Dorchester Maumbury Rings Dorchester Crown Buildings Dorchester Fordington Old Vicarage Dorchester Maiden Castle Rd Dorchester-on-Thames Vicarage Dorton Dragonby Duxford (Hinxton Rd) Dunstable Friary Field Each End, Ash Easton Easton Lane Eastrea, Whittlesey Elm’s Farm, Heybridge Elsenham Ely Prickwillow Road Emmanuel Knoll, Godmanchester Enderby Enfield Ewell Fenny Stratford Ferrybridge Field’s Farm, Duntisbourne Abbot’s Figheldean Finkley Fishbourne Fordington Bottom, Dorchester Fox Covert Farm, Market Deeping

Clarke 1988 Birbeck 2009 Hawkes & Crummy 1995: 164, 170 Crummyet al. 1993 Eckardt 1999 Crummy 2005; Pooley et al. 2011

Essex Essex Essex Essex

App. 2.1.1 App. 2.2 App. 6.4

Hull 1958: 258-9 Orr 2010 May 1930 Foster 1986; Hawkes & Crummy 1995: 164, 170

Essex Essex Essex

App. 2.1.3 App. 6.4

Brooks 2006 Hull 1942 Crummy, Crummy & Crossan 1993

Essex Essex Gloucs. Northumbs. Hants. W. Sussex Humbs. Bucks. Herts.

App. 6.4

Oxon. W. Yorks. Hants. Herts. Kent Derbys. Dorset Dorset Dorset Dorset Dorset Dorset Dorset

App. 5.2 App. 5.2 App. 5.2; 6.6

Crummy et al. 2007 Burnham et al. 1997: 434 Webster 1990 Casey & Hoffmann 1995; Breeze 2006: 426-7 Millett & James 1983 Kelly & Dudley 1981 Corder 1989 Smith 1908: 6 Burnham et al. 1994: 276; unpublished information from Simon West. Chambers 1976a Wrathmell & Nicholson 1990, Millett & Schadla Hall 1992 Thompson & Holland 1977 Smith 1858 Wheeler 1985 Davies et al. 2002 McKinley and Egging Dinwiddy 2009 Wheeler 1943 Farwell & Molleson 1993 Bradley 1976 Green et al. 1981 Startin 1981

Dorset Oxon. Bucks. Lincs. Cambs. Beds. Kent Hants. Hants. Cambs. Essex Essex Cambs. Cambs. Leics. G. London Surrey Bucks. W. Yorks. Gloucs. Wilts. Hants. W. Sussex Dorset Lincs.

App. 2.1.1 App. 5.2 App. 6.4 App. 5.2

App. 2.1.1 App. 2.1.3 App. 5.2 App. 5.2 App. 5.2

App. 5.2 App. 6.3

App. 5.2 App. 5.2

Smith et al. 1997 Taylor & Harden 1939: 293 Farley 1983 May 1996 Lyons 2011 Matthews 1981; Matthews & Hutchings 1972 Hicks 1998 Anon 1998; Burnham et al. 1998: 426 Fasham et al. 1987 Taylor 2011 Atkinson & Preston 1998: 94-98 Johns 1993 Atkins & Mudd 2003 Green 1973 Harvey 2011 Smith 1903 Cotton 2001 Neal 1987 Roberts 2005 Mudd et al. 1999

App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2

Graham & Newman 1993 Stevens 1872 Cunliffe et al. 1996 Smith et al. 1997 Unpublished: information from D. Trimble

App. 5.2 App. 5.2 App. 5.2 App. 5.2

App. 5.2 App. 5.2 App. 6.2 App. 5.2 App. 6.5 App. 5.2 App. 6.2 App. 5.2

157

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Foxholes Farm Foxholes, Hitchin Foxley Hill, Ashwell Foxton Frilford Frocester Court Gadebridge Park Gatcombe Girton Gloucester Hillfield Lodge, Denmark Rd Gloucester Kingsholm Gloucester 35 & 82 Kingsholm Rd. Gloucester Kingsholm, Gambier Parry Lodge Gloucester 76 Kingsholm Rd. Gloucester London Rd Godmanchester Godmanchester: A14/604 junction Godmanchester Nun’s Bridge, Hinchingbrooke Godmanchester Rectory Farm Grateley Gravelly Guy Great Barford Bypass Great Chesterford Great Dunmow Great Staughton Great Witcombe Great Wymondley Guilden Morden Gussage All Saints Hales Hallaton Hambleden Ham Hill Hampstead Norris Hamworthy Handley Hardwick Hartlepool Haslemere Hauxton Hayton Herd Hill Heronbridge Herriotts Bridge Hertford Heath Hibaldstow Hildersham, Maypole Hill High Torrs Hinxworth Hinxworth Place Hinxworth New Inn Hoddesdon Holborough, Snodland Holybourne London Rd Houghton Down Hucclecote Huntingdon, Watersmeet Hurstbourne Tarrant Icklingham Ilchester Little Spittle Ilchester Northover House Ilchester Townsend Close

Herts. Herts. Herts. Cambs. Oxon. Gloucs. Herts. Avon Cambs. Gloucs.

App. 5.2

App. 5.2 App. 5.2 App. 6.2 App. 5.2 App. 6.3, 6.5 App. 6.4

Partridge 1989 TEHAS 8, 22-25 Wilson 1969: 221; Page 1914: 148-49 Price et al. 1997 Burnham & Wacher 1990: 178-83 Price 2000; unpublished information from E. Price Neal 1974: 39-41 Branigan 1977 Liversidge 1977 Frere 1992: 345

Gloucs. Gloucs.

Hurst 1985; Hurst & Hills 1989 Frere 1984a: 314-315

Gloucs.

Frere 1984a: 315; Frere 1985: 300-302; archive

Gloucs. Gloucs. Cambs. Cambs. Cambs.

Frere 1988: 469 Simmonds et al. 2008 Mays, S. A. 1993b; Taylor 1997 Wait 1991 Harden 1968; Liversidge 1977: 24-25; Wilson 1968a: 191 Frere 1991: 256; F. McAvoy n.d. Cunliffe & Poole 2008c Lambrick & Allen 2004 Timby et al. 2007a Medlycott 2011b: 94-103 Wickenden 1988 Greenfield et al. 1995 Leach 1998 Westell 1938 Lethbridge 1934; Liversidge 1977 Wainwright 1979 Goodyear 1974 Page & Keate 1907: 212 Cocks 1921; Eyers 2011 Whimster 1981: 388 Page & Calthrop 1906: 209-10 Jarvis 1993 White 1970 Chambers & Williams 1976 Daniels et al. 1987 Holmes 1949 Liversidge 1958; Liversidge 1977: 28-29 Burnham et al. 1997: 417-19; Halkon & Millett 1997 Bellhouse 1954 Burnham et al. 2005: 422-23 Rahtz & Greenfield 1977 Holmes & Frend 1959; Hüssen 1983; Stead 1967 Smith 1987: 189-198 Dunning & Jessup 1936: 49; Taylor 1993: 224 Breeze & Graham Ritchie 1980

Cambs. Hants. Oxon. Beds. Essex Essex Cambs. Gloucs. Herts. Cambs. Dorset Staffs. Leics. Bucks. Somerset Berks. Dorset Dorset Oxon. Cleveland Hants. Cambs. Humbs. Cumbria Cheshire Somerset Herts. Lincs. Cambs. Dumfries & Galloway Herts. Herts. Herts. Kent Hants. Hants. Gloucs. Cambs. Hants. Suffolk Somerset Somerset Somerset

App. 5.2, 6.3 App. 5.2 App. 6.3 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 6.3

App. 5.2 App. 5.2 App. 5.2 App. 6.3 App. 5.2 App. 6.5 App. 5.2

App. 5.2 App. 6.6 App. 5.2

App. 5.2 App. 5.2 App. 6.5

App. 6.5 App. 5.3, 6.4 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2

158

TEHAS 4, 1908: 160 TSAAAS 3, 1932: 141-50 Dunning & Jessup 1936: 50 Jessup 1954 Graham 1991 Cunliffe & Poole 2000c; Cunliffe & Poole 2008b Thomas et al. 2003 Nicholson 2006 Dewar 1929; Hawkes & Dunning 1930 West & Plouviez 1976 Leach 1982 Frere 1983a: 319-20 Leach 1982

CONCORDANCE AND APPENDICES

Irchester Itchen Abbas Jordan Hill, Weymouth Juliberrie’s Grave, Chilham Kelsall The Crown Kemp Farm Kempsford Quarry Kempston Kempston Marsh Leys Farm Keston Ketton Garley’s Field Kibworth Harcourt Kingscote King’s Hedges, Milton Kingsweston King’s Worthy Knob’s Crook, Woodland Lambs Lea Lanchester Langley (aka Joy Wood, Lockham) Laxton Leicester Clarence St. Leicester Newark St. Lincoln Monson Street Linton Litlington Litlington, Limlow Hill Little Chester, Derby Little Shelford, Foulness Little Walden Little Waltham Little Wymondley Litton Cheyney Llantwit Major London Atlantic House London Bishopsgate London Classicianus tomb London Eastern cemetery

Northants. Hants. Dorset Kent Herts. E. Sussex Gloucs. Beds. Beds. Kent Leics. Leics. Gloucs. Cambs. Avon Hants. Dorset E. Sussex Durham Kent Northants. Leics. Leics. Lincs. Cambs. Cambs. Cambs. Derbys. Essex Essex Essex Herts. Dorset South Glam. G. London G. London G. London G. London

London Giltspur Rd London Great Dover Street London St Bartholomew’s London Shadwell London Spitalfields

G. London G. London G. London G. London G. London

London Stratford Market London Warwick Square Long Crendon Longthorpe Low Borrowbridge Lullingstone Lynch Farm M1 Junction 8 Maddington Farm, Shrewton Marquis of Granby, Sompting Marshall’s Farm, Prittlewell Marshfield Martin’s Down Maryport Maxey West Field Melandra Castle Meldreth, Mettle Hill Mersea Island Micheldever Wood Middle Wallop

G. London G. London Bucks. Cambs. Cumbria Kent Northants. Herts. Wilts. W. Sussex Essex Gloucs. Hants. Cumbria Cambs. Derbys. Cambs. Essex Hants. Hants.

App. 6.4

App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 6.3 App. 6.3 App. 6.6 App. 6.4

App. 5.2 App. 5.2

App. 6.5 App. 2.1.3, 5.2 App. 6.5

App. 6.2 App. 6.2 App. 5.2 App. 5.2

App. 2.1.3

App. 5.2

App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 6.4 App. 5.2 App. 5.2 App. 2.1.3 App. 5.2 App. 2.1.3 App. 6.5 App. 5.2

159

Taylor & Collingwood 1926: 223; 1927: 201 Archaeology in Hants 1987; 1991 Biddle 1967; RCHME 1970: II.3. 617 Jessup 1937 Page 1914: 158 Gilkes 1989 Booth & Stansbie 2007 Dawson 2004 Luke & Preece 2011 Philp et al. 1991; 1999; Hassall & Tomlin 2006: 467 Carlyle 2008 Page & Keate 1907: 213 Timby 1998: 35, 275-76 Frere 1991: 256 Boon 1993 Burnham et al. 1998: 427; P. McCulloch pers. comm. Fowler 1964 Gilkes 1990 Turner 1990 Haverfield et al. 1932: 158-60; Jessup 1959: 26-27 Jackson & Tylecote 1988 Gardner 2005 Cooper 1996a; Derrick 2009 Steane 2001: 19-21 Lethbridge 1937 Kempe 1836; Liversidge 1977: 29-30 Clark 1938; Taylor 1993: 225 Brassington 1971 Hull 1963b: 132-33 Hull 1963b: 195 Drury 1978 Went & Burleigh 1992; archive information Bailey 1967; Whimster 1981: 256-57 Hogg 1974 Watson 2003 Swift 2003 Grasby & Tomlin 2002 Barber et al. 1990; Siddell & Rielley 1998; Whytehead 1986; archive information Barber & Hall 2000 Mackinder 2000 Bentley & Pritchard 1982 Lakin 2002: 10-11, 26 Swain & Roberts 2001; Thomas 2004; Montgomery et al. 2010: 217-9. Hiller & Wilkinson 2005 Shepherd 1988 Smith 1908: 6 Dannell & Wild 1987: 75 Lambert 1996 Meates 1979; 1987 Jones 1975 Stansbie 2012: 33-37, 211. McKinley & Heaton 1996 Ainsworth & Ratcliffe-Densham 1974 Wymer & Brown 1995: 40, 161 Blockley 1985 Pitt Rivers 1898 Biggins & Taylor 2004; Breeze 2006: 405-7 Pryor & French 1985 Webster 1971 Taylor 1993: 225 Hull 1963b; Rodwell 1978: Fasham 1987 Piggott 1949

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Mill Hill Mucking

Kent Essex

Nash, Gwent levels Nazeing Neatham Nettleton Scrubb Newhaven Newton Kyme Newport Pagnell Northchurch Northfield Farm, Long Wittenham North Marston North Shoebury Oakley Down, Wimborne St Giles Oakridge Odell Odiham Old Down Farm (Andover) Old Down Farm (E. Meon) Old Newton Old Sleaford Old Winchester Hill- Giant’s Grave Old Winteringham Orton Longueville (Monument 97) Ospringe Otford Overton Down Owslebury Park Street Parlington Hollins Petersfield Petty Knowes, High Rochester Pickford Hill, Harpenden Pitney, Roman Farm Plant’s Farm, Maxey Popley Portesham Portway West Ind. Est, Andover Poxwell Priory Park, Prittlewell Radley Barrow Hills 1 Radley Barrow Hills 2 Radnage Rayne /Riseholme, Lincoln Rochford Cherry Orchard Brickfield, Eastwood Rockbourne Roden Downs Rothamsted Rotherley Rougham Roughground Farm Roxton Royston Grange Ructstalls Hill Rudston Ruxox St Albans Abbey St Albans Folly Lane St Albans King Harry Lane St Albans Kingsbury Manor St Albans Lindum Place St Albans St Michael’s

Gwent Essex Hants. Wilts. E. Sussex N. Yorks. Bucks. Herts. Oxon. Bucks. Essex Dorset Hants. Beds. Hants. Hants. Hants. Suffolk Lincs. Hants. Humberside Cambs. Kent Kent Wilts. Hants. Herts. W. Yorks. Hants. Northumb. Herts. Somerset Cambs. Hants. Dorset Hants. Dorset Essex Oxon. Oxon. Bucks. Essex Lincs. Essex Hants. Berks. Herts. Wilts. Suffolk Gloucs. Beds. Derbys. Hants. Humberside Beds. Herts. Herts. Herts. Herts. Herts. Herts.

App. 5.2 App. 2.1.3, 5.2, 6.2 App. 5.2 App. 6.6 App. 5.2 App. 5.2 App. 5.2 App. 6.3 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 6.4 App. 5.2

App. 5.2 App. 2.1.3 App. 5.2 App. 6.6 App. 5.2 App. 6.7 App. 2.2 App. 6.5 App. 5.2 App. 5.2 App. 5.2 App. 6.4 App. 6.7 App. 5.2 App. 6.3 App. 5.2 App 5.2 App. 6.3

App. 6.3 App 2.1.3, 5.2 App. 6.5 App. 5.2 App. 6.3, 6.5 App. 5.2

App. 5.2 App. 5.2 App. 5.2 App. 6.3, 6.5 App. 6.3 App. 6.2

160

Parfitt 1995 Cambridge Archaeological Unit 2008 Meddens & Beasley 2001 Hull 1945 Millett & Graham 1986 Wedlake 1982 Bell 1976 Boutwood 1996 Morris & Carlyle 2011: 73-5. Neal 1977 Gray 1977 Farley 1971; Wilson 1974: 436 Wymer & Brown 1995: 40, 161 Brown et al. 1995 Mays 1988; Maltby 1993; Oliver 1992 Dix 1979; Goodburn 1978 Jenkins 1990 Davies 1981 Whinney & Walker 1980 Low 1909 Elsdon 1997 RCHME 1979: 18 Stead 1976b Mackreth 2001 Whiting et al. 1931 Ward 2011 Fowler 2000: 51-4; Smith & Simpson 1964 Collis 1968; 1970; 1977b Saunders 1961 Roberts et al. 2001 Moray-Williams 1908; 1909 Charlton & Micheson 1984; Wilson 2004 Page 1914: 153; Smith 1922 Dewar 1949; Leech 1980 Gurney et al. 1993 Wright et al. 2009: 28-31 Fitzpatrick 1996 Hughes 1980 Hurst & Wacher 1986 Hull 1963b: 167 Atkinson 1952 Chambers & Macadam 2007 Skilbeck 1923; Harden et al. 1987: 51 Smoothy 1989 Thompson 1954 Taylor 1954: 98; Hull 1963b: 127 Sumner 1914 Hood & Walton 1948 Lowther 1937 Pitt Rivers 1888; Hawkes & Piggott 1947: 36-42 Dunning & Jessup 1936: 51; Fox 1911: 294-95, 315-16 Allen et al. 1993: 102 Taylor & Woodward 1985 Hodges 1991: 76; Marsden 1982 Oliver & Applin 1979 Stead 1980 Dawson 2004 Biddle & Kjølby-Biddle 2001 Niblett 1999 Stead & Rigby 1989 Wheeler & Wheeler 1936: 135 Freeman 1971; Wilson 1972: 329 Frere 1983b: 273

CONCORDANCE AND APPENDICES

St Albans St Stephens

Herts.

St Albans Verulam Hills Field St Albans ‘William Old’ St Peter’s School, Coggeshall Sandy Saltwood Tunnel Sandy Tower Hill Sawbridgeworth Scole Dickleburgh Shakenoak Sheepen Shefford

Herts. Herts. Essex Beds. Kent Beds. Herts. Norfolk Oxon. Essex Beds.

Shepton Mallett Fosse Lane Shiptonthorpe Shorden Brae Silchester Silchester Latchmere Green Six Hills, Stevenage Slip End, Ashwell

Somerset E. Yorks. Northumb. Hants. Hants. Herts. Herts.

Snailwell Snell’s Corner, Horndean Snettisham bypass Soberton South Cadbury Southfleet

Cambs. Hants. Norfolk Hants. Somerset Kent

South Shields Southwell South Wonston Sparsholt Springhead (settlement) Springhead Pepper Hill Stanfordbury Staniwells Farm, Hibaldstow Stansted Stansted (Framework Archaeology) Stanton Harcourt Stanton Low Stanwick Stebbing Stonea Stotfold Strood Hall (A120) Stuntney Suddern Farm Sutton Valence Swaffham Prior, Gallows Hill Swaythorpe Farm, Kilham Syreford Mill Takeley Takeley Street Temple Farm, Prittlewell Thorley Thornborough Thornborough Thornhill Thruxton Thurnscoe Billingley Drive Tilford Tockington Tollgate Toppesfield

T. & W. Notts. Hants. Hants. Kent Kent Beds. Lincs. Essex Essex Oxon. Bucks. Northants. Essex Cambs. Beds. Essex Cambs. Hants. Kent Cambs. E. Yorks. Gloucs. Essex Essex Essex Herts. Bucks. Bucks. Hants. S. Yorks. Hants. Glos. Kent Essex

App. 2.1.1, 2.1.3, 6.3 App. 6.3

App. 5.2

App. 5.2 App. 6.4 App. 6.3 App. 5.2

App. 6.2 App. 6.5 App. 6.3 App. 5.2 App. 6.7 App. 2.1.3 App. 6.5 App. 2.1.3 App. 5.2 App. 6.6 App. 2.1.1 App 2.1.3, 2.2 App. 6.3 App. 5.2 App. 5.2, 6.3 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 6.4 App. 5.2 App. 6.3 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 5.2 App. 6.3 App. 6.3

App. 6.3 App. 6.3, 6.4 App. App. 5.2 App. 6.6 App. 5.2 App.5.2, 6.4 App. 6.3

161

Davey 1935; Frere 1990: 338-40; Niblett & Thompson 2005: 141-5; archive Anthony 1968; Page 1914: 137-8; Corder 1941 Niblett & Reeves 1990 Clarke 1988: 51, 58 Dawson 1995 Mckinley et al. 2006: 23 Johnston 1975 Page 1914: 163 Burnham et al. 1994: 278; T. Ashwin information Brodribb et al. 2005: 188-91; CSIR I. 7, 163-4. Hull 1963a: 193-5; Niblett 1985 Fox 1923: 213; Page & Keate 1908: 11-12; Kennett 1971 Leach et al. 2001 Millett 2006 Gillam & Daniels 1961; Bidwell 2010 Boon 1974 Fulford & Creighton 1999 Dunning & Jessup 1936: 50 Burleigh 1976; Burleigh & Fitzpatrick-Matthews 2007: 14-15 Stead 1967: 57 Knocker 1956 Flitcroft 2001 Haverfield 1900: 266 Barrett et al. 2000 Rashleigh 1803a; 1803b; Jessup 1959: 29-30; Davies 2000; Andrews et al. 2011: 132-4 Snape 1994 Daniels 1966 Whinney 1987 Collis 1977a Andrews et al. 2011 Biddulph 2006a Stead 1967; Page & Keate 1908: 13-14 Smith 1987: 189-98 Havis & Brooks 2004 Cooke et al. 2008 McGavin 1980 Woodfield & Johnson 1989 Dix 1987; Frere 1987a Going 1978; Goodburn 1978: 452; Rodwell 1976b Jackson & Potter 1996 Webley et al. 2007: 4 Timby et al. 2007b: 118-38 Taylor 1984 Cunliffe & Poole 2000b Jessup 1959 Malim 2006 Mackey 2001 Timby 1998 Hull 1963b: 185 Hull 1963b: 185 Wymer & Brown 1995: 161 McDonald 1997; unpublished Johnson 1975 Liversidge 1960 Jennings et al. 2004: 65, 108-9 Henig & Soffe 1993; Cunliffe & Poole 2008d Neal & Fraser 2004 Millett 1974a Masser & McGill 2004 Allen et al. 2012 Hull 1963b: 192-93

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Tutt Hill, Westwell Twyford Uffington Viables Farm, Basingstoke Vindolanda Wakerley Walkington Wold Ware Allen & Hanbury’s Ware Buryfields Warren Farm Wasperton Water Newton Normangate Field Water Newton (A1) Watford Wavendon Gate Welford-on-Avon Wells Welton Wold Welshpool Welwyn Garden City Welwyn Hall Welwyn Grange Wendover Wellwick Farm Westergate, Worthing West Field, Maxey Westhampnett West Lane, Kemble Weston Turville West Park West Thurrock West Wood, nr Somerton Wheathampstead, Turnershall Farm Whitcombe Whitton Wilbury Hill Willersey Willington Winchester Cattle Market Winchester Chester Rd Winchester Andover Road (Eagle Hotel) Winchester Grange Rd Winchester Highcliffe Winchester Hyde Street (Evans Halshaw) Winchester Hyde Street Late Winchester Lankhills

Kent Hants. Oxon. Hants. Northumb. Northants. Humberside Herts. Herts. Beds. Warws. Cambs. Cambs. Herts. Bucks. Warws. Somerset Humberside Powys Herts. Herts. Herts. Bucks. W. Sussex Cambs. W. Sussex Gloucs. Bucks. Hants. Essex Somerset Herts. Dorset S. Glam. Herts. Gloucs. Beds. Hants. Hants. Hants.

Winchester Milland Winchester Nun’s Walk Winchester Oram’s Arbour

Hants. Hants. Hants.

Winchester 45 Romsey Rd, West Hill Winchester St John’s Street Winchester St Martin’s Close Winchester Victoria Road East

Hants. Hants. Hants. Hants.

Winchester Victoria Rd West Winchester Winnall Winchester Winnall Winklebury Winnall Down Winterbourne Down Winterton villa

Hants. Hants. Hants. Hants. Hants. Wilts. Humberside

Hants. Hants. Hants. Hants. Hants.

App. 5.2 App. 6.7 App. 5.2

App. 5.2 App. 5.2

App. 5.2 App. 5.2

App. 6.5 App. 5.2

App. 5.2 App. 6.4 App. 6.5 App. 6.3 App. 6.3 App. 5.2 App. 2.1.3, 5.2 App. 5.2 App. 6.3 App. 6.3 App. 5.2 App. 6.3 App. 5.2

App. 6.7 App. 6.7 App. 6.6 App. 6.6

Barclay & Brady 2006: 21 Biddle unpublished note for Hants. County Council Miles et al. 2003 Millett & Russell 1982; 1984; Gibson 2004 Birley & Blake 2005; 2007 Jackson & Ambrose 1978 Bartlett & Mackey 1972 Frere 1977: 401 Gentlemen’s Magazine 1802 Dawson & Slowikowski 1988 Carver et al. 2009 Burnham & Wacher 1990: 87, 91; Dannell & Wild 1969 Casa Hatton & Wall 2006 Dunning & Jessup 1936: 50; Page 1914: 165 Williams et al. 1995 Booth 1994 Rodwell 2001: 40-4 Wilson 1973: 282 Boon 1961 Stead 1967 Rook et al. 1984; McDonald forthcoming Kindersley 1922; Westell 1930 Zeepvat 2004 Haverfield 1935: 67 Pryor & French 1985 Fitzpatrick 1997a King et al. 1996 Waugh 1961: Smith 1908: 15-16 RCHME 1983 200 Hull 1963b: 189; Philpott 1991: 252, 257 Gater et al. 1993: 44-7 West 2004 Aitken & Aitken 1990 Jarrett & Wrathmell 1981 Applebaum 1933; 1949; Whimster 1981: 213 O’Neil 1971 Pinder 1976 Collis 1978: 142-49; Taylor 1937: 243 Grew 1980: 363; Ottaway et al. 2012: 179-89 Richards 1999: 84-107; Ottaway et al. 2012: 120-7 Biddle 1967 Collis 1978: 103-08 Birbeck & Moore 2004 WMS archive; Ottaway et al. 2012: 118-20. Clarke 1979; Booth et al. 2010. Chapman et al. 2009, 273 Collis 1978: 93-103 Collis 1978: 149-55 WMS archive; Esmonde Cleary 1987: 155; Whinney 1994 WMS archive; Qualmann 1981

App. 6.7 App. 6.6 App. 6.6

App. 6.7 App. 6.7 App. 6.6

Haverfield 1900: 290 Morris 1986b; Ottaway et al. 2012: 189-193 WMS archive; Ottaway 1992: 76-81; Ottaway et al. 2012: 42-57, 77-91 WMS archive; Ottaway et al. 2012: 97-118 Biddle 1975 Collis 1978: 65-74; OS card, Hants SMR Fasham 1985 Algar 1963 Stead 1976b; Goodburn 1976: 327; Grew 1981; Frere 1983a: 296; Rankov 1982: 350; Wilson 1974: 424

App. 6.2 App. 6.6

App. 5.2 App. 5.2

162

CONCORDANCE AND APPENDICES

Wollaston Bypass Woolbury / Stockbridge Down Woodcuts Common Wood Lane End, Hemel Hempstead Worksop Raymoth Lane Wyke Regis York Blossom St York The Mount York The Mount (vault) York Trentholme Drive Youngsbury, Ware

Northants. Hants. Dorset Herts.

App. 5.2 App. 5.2 App. 5.2 App. 6.5

Chapman & Jackson 1992 Cunliffe & Poole 2000a Hawkes & Piggott 1947 Neal 1984

Derbys. Dorset N. Yorks. N. Yorks. N. Yorks. N. Yorks. Herts.

App. 5.2 App. 5.2

Palmer-Brown & Munford 2004 Leonard 2008 Frere 1990: 326 Dickinson & Wenham 1957: 287, 314-16 RCHME 1962: 95-96 Wenham 1968 Evans 1890

App. 2.1.3 App. 6.5

163

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

APPENDICES Figures 2.5 and 2.6 Cemetery excavations recorded from 1920 to 2010 in JRS and Britannia

APPENDIX ONE DATA FOR FIGURES IN CHAPTER 2 Figure 2.1 The number of cemeteries from five counties in different sources (SMR, JRS/ Britannia, Philpott)

Cornwall Cambs Devon Hants Herts

SMR 16 160 9 115 161

JRS/ Brit. 1 38 4 44 58

1921-25 1926-30 1931-35 1936-40 1941-45 1946-50 1951-55 1956-60 1961-65 1966-70 1971-75 1976-80 1981-85 1986-90 1991-95 96-2000 2001-05 2006-10

Philp. 0 33 1 39 49

Figure 2.2 Dated and undated cemeteries from three counties in different data sources (Herts, Hants and Cambs)

Cambs (SMR) Cambs (JRS/Br.) Cambs (Philp.) Hants (SMR) Hants (JRS/Br.) Hants (Philp.) Herts (SMR) Herts (JRS/Br.) Herts (Ph.)

Early 27

Inter 10

Late 20

U 113

8

6

9

15

9 20 9 10 53 22 13

11 7 1 4 16 15 9

8 31 24 14 26 8 177

5 57 10 11 60 13 10

Early 27

Inter 10

Late 20

8

6

9

9 20 9 10 53 22 13

11 7 1 4 16 15 9

8 31 24 14 26 8 17

JRS Britannia Philpott

Crem. 47 4 10 35 10 12 79 18 18

Inhum. 115 25 16 66 29 25 32 10 26

C&I 8 9 7 11 5 2 27 28 5

C & I. 6 2 5 4 0 0 3 3 1 19 12 11 18 20 13 6 33 16

U 7 5 4 2 0 2 1 2 2 0 1 3 3 10 8 1 4 8

Total 39 35 36 27 9 15 26 23 15 56 55 106 87 113 111 106 143 110

Early 63 156 208

Inter. 19 67 121

Late 53 229 249

U 138 387 345

Figure 2.8 Dated burials in different data sources (JRS / Britannia)

JRS Britannia Combined

Early 978 1216 2194

Inter. 1032 2365 3397

Late 486 4169 4655

U 901 2100 3001

Figure 2.9 Cemeteries of different types in different data sources (JRS, Britannia, Philpott)

Figure 2.4 Cemetery types from three counties (Herts, Hants and Cambs) in different data sources

Cambs (SMR) Cambs (JRS/Br.) Cambs (Phil) Hants (SMR) Hants (JRS/Br.) Hants (Phil) Herts (SMR) Herts (JRS/Br.) Herts (Phil)

Inhum. 13 13 15 15 7 6 14 15 6 26 29 72 49 56 61 71 65 53

Figure 2.7 Dated cemeteries in different data sources (JRS, Britannia, Philpott)

Figure 2.3 Dated cemeteries from three counties (Herts, Hants and Cambs) in different data sources

Cambs (SMR) Cambs (JRS/Br.) Cambs (Phil) Hants (SMR) Hants (JRS/Br.) Hants (Phil) Herts (SMR) Herts (JRS/Br.) Herts (Phil)

Crem. 13 15 12 6 2 7 8 3 6 11 13 20 17 27 29 28 41 33

JRS Britannia Philpott

U 0 0 0 3 0 0 15 2 0

Crem. 79 212 320

Inhum. 128 458 548

C&I 41 131 55

U 25 38 0

Figure 2.10 The number of burials in different types of cemeteries (JRS / Britannia)

JRS Britannia Combined

164

Crem. 539 998 1537

Inhum. 920 4592 5512

C&I 1910 4179 6089

U 28 81 109

CONCORDANCE AND APPENDICES

Figure 2.18 Burial types from different settlements (JRS / Britannia) Crem. Inhum. U

Figure 2.11 The number of different cemetery types through time (Britannia)

Early Inter. Late U

Crem. 94 15 17 89

Inhum. 42 12 191 213

C&I 14 39 14 61

U Urban Mil/vicus Min. centre Rural Villa

6 1 7 24

Figure 2.12 The percentage through time of different burial types (Britannia simplified)

Early Inter. Late U

Crem. 880 1293 131 508

Inhum. 287 930 4035 1421

Burials 5819 847 3050 3252 279

U 38 153 153 167

Cemeteries 325 76 167 491 53

Fig. 2.15 Dated cemeteries at different settlement sites (JRS / Britannia)

Urban Mil/vicus Minor centre Rural Villa

Early 52 18

Inter. 27 10

Late 97 14

U 150 34

25 119 5

16 33 0

57 93 21

69 244 28

Figure 2.16 Dated burials at different settlement sites (JRS / Britannia)

Urban Mil/vicus Min. centre Rural Villa

Early 1018 342 321 468 45

Inter. 1109 296 1554 438 0

Late 2462 118 855 1081 139

U 1230 91 320 1265 95

Figure 2.17 Cemetery types from different settlements (JRS/ Britannia)

Urban Mil/vicus Min. centre Rural Villa

Crem. 69 31 32 153 6

Inhum. 170 27 97 250 42

C&I 64 13 26 66 3

3945 190 1558 2313 229

151 1 143 111 2

Figure 2.19 The distribution of Roman period cemeteries by county (JRS / Britannia and Philpott samples) JRS/ County Brit. Philpott Avon 25 23 Bedfordshire 26 8 Berkshire 10 19 Buckinghamshire 19 12 Cambridgeshire 38 33 Cheshire 10 14 Cleveland 1 1 Cornwall 1 0 Cumbria 18 18 Derbyshire 2 17 Devon 4 1 Dorset 27 64 Durham 2 1 Essex 99 78 E. Sussex 1 9 Gloucestershire 63 40 G. London 97 54 G. Manchester 1 1 Hampshire 44 39 Hereford & Worcs 10 0 Hertfordshire 58 49 Humberside 27 11 Kent 115 90 Lancashire 1 2 Leicestershire 29 17 Lincolnshire 43 19 Merseyside 0 0 Norfolk 23 15 Northamptonshire 31 11 Northumberland 9 5 N. Yorkshire 52 60 Nottinghamshire 4 5 Oxfordshire 42 25 Shropshire 3 9 Somerset 24 25 S. Yorkshire 6 2 Staffordshire 17 6 Suffolk 16 16 Surrey 3 9 Tyne & Wear 2 4 Wales 37 22 Warwickshire 25 14 W. Midlands 0 0 W. Sussex 16 20 W. Yorkshire 11 15 Wiltshire 18 43

Figure 2.13 – 2.14 Burials and cemeteries from different settlement types (JRS/Britannia)

Urban Mil/Vicus Min. Centre Rural Villa

1723 656 1349 828 48

U 21 5 12 23 2

165

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Fig. 2.20 and 21 The distribution of Roman period cremation, inhumation (+ data for early and late cemeteries) (JRS / Britannia)

APPENDIX TWO DESCRIPTIONS OF FEATURES DISCUSSED IN CHAPTER 3

2.1 Pyre sites from Britain and other provinces Avon Bedfordshire Berkshire Buckinghamshire Cambridgeshire Cheshire Cleveland Cornwall Cumbria Derbyshire Devon Dorset Durham Essex E. Sussex Gloucestershire G. London G. Manchester Hampshire Hereford & Worcs Hertfordshire Humberside Kent Lancashire Leicestershire Lincolnshire Merseyside Norfolk Northamptonshire Northumberland Nottinghamshire N. Yorkshire Oxfordshire Shropshire Somerset S. Yorkshire Staffordshire Suffolk Surrey Tyne & Wear Wales Warwickshire W. Midlands W. Sussex W. Yorkshire Wiltshire

Crem. 3 5 5 10 4 4 0 0 13 1 2 1 1 53 1 3 16 1 10 2 18 1 48 0 2 8 0 8 4 3 1 8 6 2 1 3 10 2 2 0 17 1 0 6 1 0

Inhum. 20 16 5 7 25 3 1 1 1 0 1 20 1 22 0 46 51 0 29 6 10 24 34 0 25 28 0 10 21 5 3 35 26 1 20 2 6 13 1 0 19 22 0 4 7 12

Early 5 5 3 5 8 4 0 1 4 0 1 6 0 38 1 4 14 1 9 1 22 3 27 0 5 6 0 6 3 2 0 2 3 1 3 1 4 1 1 0 9 2 0 5 0 1

Late 4 7 5 4 9 3 1 0 5 0 0 8 1 9 0 17 21 0 24 1 8 7 23 0 11 14 0 4 9 1 3 15 17 1 15 0 3 1 1 0 3 12 0 4 2 7

2.1.1 ‘Permanent’ pyre sites from Roman Britain Colchester Gurney Benham House (Essex) (Hull 1958: 2589) Three ‘flues or furnaces’ set into a ditch fill, in the western cemetery area. The base of the flues was constructed of tile with a surviving wall of ‘conglomerate’ i.e. ‘gravel indurated by iron’, underlying loam reddened by heat; each ended in a round chamber. The plan shows that the flues were a minimum of 2.5m long and 0.95m wide, although these do not agree with Hull’s measurements of 12ft by 1ft. There is no mention of charcoal, pyre debris, or cremated bone. The only associated deposit reported was a ‘grey powder’. As Hull suggests, both the flue type and the quantity of ironstone may suggest that the features were connected with ironworking rather than cremation. Only the cemetery location supports an interpretation as a burial related feature. Corbridge (Northumberland) (Casey and Hoffmann 1995) A 1.5m x 1.5m square cobble platform capped by a layer of orange red burnt clay (Contexts 12), with a spread of charcoal immediately to the south, perhaps from cleaning out the structure. To the north-east an area of localised burning almost 2m long centred on three charcoal-filled post settings (Context 11). In the easternmost of these, a 1.25m long x 0.5m arc, a screen of timbers may have stood. The feature was recovered within a small portion of a 1st and 2nd century cemetery excavated on either side of Dere Street: the pyre feature was east of the road and the cremations to the west. Densworth, Funtington (Sussex) (Smith 1858: 175) A 1.3m long burnt flint setting, adjacent to an extensive charcoal spread 2.8m x 0.6m x 0.4m deep, next to one of the burial enclosures in this rural cremation and inhumation cemetery. St Stephens, St Albans (Herts.) (Davey 1935: 245; Hood and Walton 1948: 13-14)) Within a 1st to 3rd century urban cremation cemetery ‘burning chamber’ 1 consisted of a ‘brick-lined chamber’, about which no further information could be recovered before it was destroyed. ‘Chambers’ 2 and 3 were rectangular earth cut pits, respectively 1.9m x 0.8m x 1.3m deep and 2.5m x 1.4m x 1.7m deep, filled with a mixture of charcoal, cremated human bone and hundreds of nails. Chamber 2 seems to have been earlier in date than 3. Springhead (F8) (Kent) (Penn 1965; Penn 1968: 180) A reddened thick clay rectangular base 3.2m x 1.5m which had been subjected to great heat. The feature was beside a 40ft x 40ft chalk base (F8) into which a 1st century urned cremation burial had been made. The ensemble was in an isolated position west of the settlement. Springhead (F30) (Kent) (Penn 1965: 116-117). ‘a platform composed of square tiles; ...upon it were fragments of calcined human bones and masses of charcoal. The tiles were much damaged by the action of fire’. The feature, approximately 2.2m x 0.8m and rectangular was found in association with several cremation burials east of Springhead in road widening in 1845.

166

CONCORDANCE AND APPENDICES

Linz (Austria) (Ruprechtsberger 1983: 21-22) The feature was recovered from one cemetery attached to this minor centre. It comprised a platform of gneiss slabs 1.8m x 0.6m and 0.8m high, immediately to south of which lay a c. 3m square and 0.1m thick spread of charcoal and sherds covering an area of burnt earth. No date for the feature is given. Ruprechtsberger suggests that the spread represented the site of cremation and that the corpse was displayed on the platform.

2.1.2 ‘Permanent’ pyre sites in other Roman provinces Briord (Ain, France) (de Klijn 1987: 141) Two parallel walls in dry stone 2.2m long and 0.6m wide with traces of burning, with a layer of pyre debris between them. To the south was a single Brandschüttungsgrab of the second half of the 2nd century. Beska (Yugoslavia) (Marijanski-Mariojlovic 1987: 82) Three intercutting ‘ovens’ on the western edge of a large cremation and inhumation cemetery of the 1st to 4th centuries. Their bases were circular and varied in diameter from 0.95m to 1.43m. That of III was paved with bricks. The collapsed dome was preserved to a depth of 0.15m. Traces of a stokehole were preserved with I. A cremation burial had also been placed within it. Around the feature ‘very hard fired earth...with a vast quantity of soot, charred round logs and heaps of ashes support the fact that this space must have been used for cremation of the corpses in this necropolis.’ In the absence of analysis of this ‘ash’ the features cannot be accepted as definite pyre sites. The chambers are too small to have contained bodies for cremation. The features may relate to industrial processes rather than burial.

Louroux (Creuse, France) (Dussot 1987: 6-7) An almost square ditched enclosure 7m x 6m, in centre of which lay 2 large pits filled with pyre debris. It was set against the boundary wall of a rural 1st and 2nd century AD cremation cemetery. The enclosure seems to have separated the area of cremation or deposition of cremation debris from that of burial rather than functioning in the cremation process. Mackwiller (Haut Rhin, France) (Hatt 1964: 78-82) Two similar features within the cemetery area of a minor centre of which the identification as pyre sites must be doubted. The smaller consisted of a rectangular chamber 2m by 3m, constructed from tile set into clay, the walls and base of which were lightly burnt. The larger, which overlay the smaller, consisted of a ‘chambre de combustion’ 3m by 3.75m, of seven sections separated by corbelled supports built of tiles which had been heavily burnt. A 1.3m x 3.7m long flue led into this chamber, the base of which was overlain by a thick layer of charcoal. At the other end of this flue was a stokehole roughly 3m square, with lines of tiles demarcating its north-west and south-eastern edges. No evidence is given for the presence of pyre debris other than charcoal. The feature resembles a tile kiln in every respect, as Hatt himself says. No supporting evidence for the identification of these features as a pyre sites was presented, for example the presence of pyre debris or cremated bone. Location seems to be the primary evidence for identification as a pyre feature.

Carnuntum (Austria) (von Groller 1900: Col. 113, Tafeln V and XIII) The feature was situated close to other monuments and graves on the Gräberstraße to the east of the town. It comprised a circular pit 2.4m in diameter and 0.95m deep lined with a 0.6m thick rubble wall, the foundation to support a wall surviving to a height of 0.3-0.4m. According to the excavator this may have supported a platform. On the eastern side of pit was a curved opening, outside which lay a 0.1m thick gravel layer covered by 0.2m of burnt material, the ‘praefurnium’ according to von Groller. This feature may however be better explained as a means of improving ventilation or allowing pyre residue to be raked out, rather than as some sort of flue or stokehole. Perhaps it was even related to the urned cremation burial which overlay it. The pit itself was half filled with pyre debris, including cremated bone and molten fragments of bronze and glass.

Ostia (Floriani Squarciapino et al. 1958) The presence of ustrina is identified in association with several collective tombs by the Porta Romana and Porta Laurentina. The very brief descriptions indicate brick built platforms, square or rectangular of a few square metres in extent. In no case do the few associated inscriptions reference the pyre site. Associated materials are not recorded, although burnt fragments of carved ivory from funerary biers of late Republican date are documented from Ostia’s cemeteries (Letta 1984).

Cologne Luxemburger Straße (Naumann-Steckner 1997: 147; Riedel 1980) Excavated with 1st and 2nd century graves within an extensive urban cemetery on road south-west of Cologne was a walled enclosure 13m by 12m, at the centre of which lay a platform with a pit set into it, over which the pyre may have been constructed, but the feature has yet to be described in any detail. The enclosure separated the area of cremation from that of burial rather than functioning in the cremation process.

Reichenhall (Bavaria, Germany) (Hell 1957: 46) Located by a river on the periphery of a large 1st and 2nd century cremation cemetery was a 5m x 5m paved platform surrounded by a low parapet with two entrances. The description resembles that of other features from Austrian cemeteries (e.g. Salzburg, Wels) but insufficient evidence is available to be confident of the identification.

Gravelotte (Moselle, France) (Faye et al. 1993: 90-1) A stone walled rectangular structure enclosing an area 3.7m x 3.1m (external measurements). Pyre debris was deposited in nearby pits and ditch. The feature was situated within a small rural cremation cemetery of the late 1st to early 2nd century AD. Innsbruck (Austria) (Ruprechtsberger 1983: 23) The feature was recovered from a cemetery attached to this minor centre. It comprised a double square stone wall enclosing an area 8.6m x 8.9m (external measurements), with walls varying between 0.45m-1.15m thick. Three cremation graves of the first half of the 3rd century AD were found within the centre of the feature.

Rheinzabern (Rheinland-Pfalz, Germany) (Ludovici 1908: 185, 201-2) Within a minor centre cemetery was a structure comprising two concentric walls constructed from brick / tile enclosing an area 2.6m square (external dimensions) with a tile pillar in the centre. Ludovici suggests that this was originally topped by a platform on which the pyre was erected. The feature possibly dates to the 2nd century. Supporting evidence for the identification of this feature as a pyre site, for example

167

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

debris, was not reported and the feature can alternatively be interpreted as the base of a monument.

difficult to distinguish pyre bases from graves as pyre debris was also deposited in the latter in this cemetery and in the fill of Iron Age quarry pits on the north-eastern side of the cemetery alongside material from disturbed cremations. The pyre features lay predominantly between the ditched boundary of the settlement and a late Iron Age to late Roman cemetery.

Saint-Paul-Trois-Châteaux (Drôme, France) (Bel 2002: 778) ‘Fosse B’, identified as an ustrinum, from the cremation cemetery to the south-west of the city, was immediately west of massive round masonry built tomb with internal buttresses. At the base of the masonry-lined pit, 1.5m x 1.1m x 0.4m, was charcoal, carbonised branches and some cremated bone, as well as burnt ceramics and glass dated to AD 60-80. The small amount of cremated bone suggests it was used for a single cremation. After the end of its use the structure was deliberately demolished and sealed with clay. Nearby was a damaged feature of similar form.

Baldock Wallington Road (Burleigh and FitzpatrickMatthews 2010: 184-5) A near oval hollow 3m x 1.5m, containing extensive carbonised wood, a small quantity of cremated bone, mostly animal, many fragments of copper alloy, iron nails and very small quantities of coarsewares, c. 40m to the north of the nearest burials in the Wallington Road cemetery.

St Lambert, Fréjus (Var, France) (Béraud and Gébara 1987: 25) Within a 1st and 2nd century cemetery east of the colony was excavated a rectangular feature of dry stone wall construction identified as a pyre site in the provisional report (measurements not given). Another pyre site of unknown type was filled with ash, charcoal, glass fragments and sherds and fragments of bone inlay from funeral biers. It was initially separate from burials and served by a path from the main road but was cut by later cremations.

Bayford, Sittingbourne (Kent) (Payne 1886) Several burials and possible pyre features were excavated during the extraction of brickearth. Next to burial 2 was a pyre feature identified from the presence of burnt earth, tiles, antler offcuts, a pig and ox jaw bones; another similar feature nearby included sherds, some from a mortarium, horn cores and an ox skull. A further larger feature, argued by Payne to be a communal pyre comprised an area 36ft x 30ft of burnt earth, charred wood, bones of deer, ox, horse, pig, dog, sherds of amphorae, samian, Upchurch, and colour coated wares, tiles, bones and tines of antler. Three features close by included further burnt material, tiles and pottery, animal bone, oyster shell, burnt pins, a coin, glass stud, jet bracelet, iron nails, a bronze pendant, the bases of two samian cups and a mortarium. The lack of explicit evidence for cremated human bone, as well as the presence of the tile and of the antler offcuts, throw doubt on the interpretation as specifically pyre features rather than some other part of the burial process or even as deposits of occupation debris.

Bürglstein, Salzburg (Austria) (Hell 1957) Located on the river bank on the southern edge of the minor centre/ urban cemetery of this minor centre in use from the 1st to 4th century. The cremation burials, one within the enclosure and five others in its immediate vicinity, were of the mid 1st to 2nd century. A double wall of undressed stone enclosing an area 6m x 6m (external measurements) with a thick layer of charcoal, sherds, ash and some cremated bone, as well as a single cremation burial against the north wall of the feature. The outer wall (0.6m thick and surviving to a height of 0.2m) was of much better quality than the inner (0.8m thick). The latter was possible the remnant of the floor of the feature.

Braughing ‘B’ (Herts.) (Partridge 1977: 74) The site of a possible pyre was indicated by a patch of heavy burning and charcoal roughly 2m long and 1m wide at the northern end of the excavated area of a 1st and 2nd century cremation cemetery. It was not recorded in detail.

‘Les Vernes’, Faverdines (Cher, France) (Fourteau-Badarji et al. 1993) Situated within a rural cemetery was a rectangular stone wall construction enclosing area of c. 4.4m x 3.9m (external measurement). Pyre debris deposited in nearby pits and ditches but there were no contemporary burials, only later inhumation burials.

Camelon (Stirlingshire) (Breeze and Rich-Gray 1980) Six pits were recovered a few metres from the earlier discovery of two weapon burials, possibly connected to the fort at Camelon. The four documented pits were mostly circular, with a diameter varying from 1.45m to 1.90m and a depth from 0.25m to 0.75m and varying profiles, V-, U-, and dish-shaped and V-shaped with a flat bottom. The tops of the pits were sometimes burnt and the fills comprised layers of charcoal, burnt sand and soil. One contained a small amount of possibly human cremated bone. Radio-carbon dating of the charcoal allowed a 1st-century BC or AD date. Although the excavators draw attention to the parallels of these pits with busta, the minute amount of bone and the layered deposits of charcoal and burnt soil suggest that these may be pyre sites from which cremated bone had been collected for burial elsewhere and which had been later re-used.

Wels (Oberösterreich, Austria) (Rieß 1974: 156; Ruprechtsberger 1983: 23) Within a minor centre cremation cemetery, although some distance from graves was a double ring or ellipse of rough hewn dry stone walling 0.45m thick, enclosing an area with a diameter of a maximum of 5.5m (external measurement), and showing signs of burning. The interior was of trampled earth overlain by pyre debris. A possible cremation grave was excavated in the centre of feature. When the cemetery was abandoned it was used as a rubbish dump.

Canterbury Cranmer House (Kent) (Frere et al. 1987: 68) A possible cremation related feature within a mid-1st to early 3rd century cemetery, investigated only in house foundation trenches to the west of Canterbury, was a large shallow depression containing carbon and burnt clay and burnt nails. Evidence for burning extended around the feature which was flanked by two large post-holes. These might have braced the pyre but there is insufficient information to assess this feature properly. It may have been a bustum, or even unconnected with burial.

2.1.3 ‘Temporary’ pyre sites from Iron Age and Roman Britain Baldock Royston Road (Herts.) (Burleigh 1993: 44-5; Burleigh and Fitzpatrick-Mathews 2007: 98-99) Pyre sites were represented by discrete patches of pyre debris including cremated human and animal bone, fragmentary ceramics and iron nails as well as burnt soil and chalk. The largest of these measured 15m by 5m. All had been reduced to a depth of a few centimetres by plough damage. It is

168

CONCORDANCE AND APPENDICES

Chichester St Pancras (Sussex) (Down and Rule 1971: 58-59, Figs 5.3 - 5.4) Cremation areas were identified from patches of burnt soil and large quantities of residual charcoal in overlying medieval deposits. Information is too scant to classify these features definitively as pyre sites or residues of other activities. The features were dispersed across the mid 1st to early 3rd century civitas capital cemetery.

Enclosure 4 (pp. 90-97): BF 32, a ditched enclosure c. 10.5m square, charcoal and heat-affected metalwork documented in a ditch fill and one pit with scorched base in the centre of the feature (BF42). Ten pits across the central and southern parts of the enclosure may be post-holes from timber supports for one or more pyres, broken (but not burnt) ceramics in fills. Site D (pp. 410-413): pyre features associated with 1st century AD cremation burials, including (i) five shallow pits of a maximum diameter 1.2m but generally much smaller, up to 0.15m deep with charcoal, cremated bone and, in two cases, heat-affected artefacts in fills and (ii) 14 features with charcoal in fill but containing no cremated bone or artefacts, interpreted by excavators as burnt remains of roots from tree clearance.

Cirencester Bathgate (Glos.) (McWhirr, Viner and Wells 1982: 72, 99-100) Undated cremation-related features were recovered from a 4th century civitas capital inhumation cemetery. 293 was a shallow rectangular east-west oriented scoop of which the sides had been scorched and the fill consisted of a thick layer of charcoal flecked with limestone, including a small amount (unquantified) of poorly fired bone from an adult male. There were two other cremations in the same trench, 180 and 294 (CS 72-3), both of which were also of adult males. The cremated bone in both was well fired. The evidence of burning and the poorly burnt state of the cremated bone suggests that the feature is a ‘one-off’ pyre site, perhaps connected to one of the nearby cremations.

Derby Racecourse (Derbyshire) (Wheeler 1985: 234) Two features situated within a walled enclosure where 1st to 4th century cremation and inhumation burials had been interred, serving a fort / vicus, were possible pyre sites. Features 163 and 171 were both roughly inhumation grave shaped features of which the sides and base had been scorched. These were possible pyre or bustum sites but are too heavily damaged by subsequent graves to give a more precise interpretation. The incorporation of sometimes large amounts of pyre debris in most of the graves within the enclosure suggests that cremation took place with the enclosure, although this might rather have derived from Brandschuttgräber.

Cirencester Oakley Cottage (Glos.) (Reece 1962: 53) A number of possible pyre-related features were excavated within a 1st and 2nd century civitas capital cemetery. A 2.5m x 0.3m x 0.2m trench, packed with charcoal and ash, was interpreted as a possible draught flue for a pyre and several features, 0.25m-0.4m square, ‘neatly cut with vertical sides’ and filled with ash and charcoal and cremated human bone were identified as pits of pyre debris. It is equally possible that they were Brandgrubengräber.

East London (Greater London) (Barber and Bowsher 2000) Pyre-related features were recovered from 2 ‘plots’ of only one of the areas excavated within the eastern cemetery, that at Hooper Street. Conditions here were more favourable to the preservation of pyre sites as strata on this site was less truncated than others. It was also excavated relatively late in the excavation programme when pyre material had been recognised as such.

Colchester 6 Lexden Road (Essex) (Brooks 2006, 3-4) Three irregular burnt orange surfaces associated with but predating a late Roman monumental tomb, perhaps associated with cremation burials disturbed by the same tomb. The largest feature was sub-circular in shape with a diameter of c. 2m. Evidence for similar features had previously been observed in the site’s vicinity.

The cremation debris from plot 21 dated from 100 to 160 AD. It lay along the southern edge of the enclosure, separated from the cremation burials and consisted of a truncated dump distributed across a series of shallow features. At its deepest it was 0.21m thick; approximately 1800 litres of material were recovered. McKinley estimates that the surviving pyre debris represents one third of the original deposit. It contained cremated bone from a minimum of 19 individuals.

Colchester Stanway (Essex) (Crummy et al. 2007) Several pyre related features were excavated in association with the late Iron Age and conquest period burial enclosures. Enclosure 3 (pp. 157-159) BF17: a tub or barrel placed in a pit east of chamber BF6, the fill of which contains several layers of charcoal with copper alloy and iron fragments though no evidence for cremated bone.

In plot 28 there were six dumps of cremation debris of a later date than in plot 21. Three were the products of single cremation episodes while the others were larger, similar to the dump from plot 21. Cremated bone from a minimum of 18 and a maximum of 29 individuals was recovered.

CF7 (pp. 160-164): a dump of pyre debris dated 50-10 BC predates enclosure 5 c. 25m to the north. The circular feature 0.5m in diameter contains charcoal and a small amount of cremated human bone scattered through the fill, as well as sherds from a single vessel, an iron needle and other iron fragments.

Litlington (Cambs.) (Kempe 1836: 370) Two heaps of wood ash ‘as much as would have loaded five carts’ were taken away from the south-eastern and southwestern corners of this 2nd to 4th century walled burial enclosure containing approximately eighty cremations and two hundred and fifty inhumations. It is possible but unprovable that this derived from a pyre site.

Enclosure 3 (pp. 85-90): west of chamber BF6 two pits surrounded by amorphous patches of scorched material and thin charcoal layers, with a scatter of cremated bone and iron and heat-affected copper alloy which may derive from military equipment. Two cut features with fills containing pyre debris possibly served as flues for pyres: BF1 2.3 x 13m shallow-bottomed pit with charcoal-rich layers; BF16 pit 5.0 m long and 0.2 m deep, charcoal and cremated human bone in fill. The latter was adjacent to and perhaps associated with BF6.

Melandra Castle (Derbys.) (Webster 1971: 79) About 650 feet south of the fort were two areas of burning, each approximately 8ft x 3ft, and close to Road A, from which fragmentary Roman period artefactual material was recovered. They lie within a general cemetery area but the degree of confidence in their identification must be low.

169

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Maryport (Cumbria) (Collingwood 1936: 91, Fig. 3) A 0.35m deep layer of charcoal to the side of a burial monument within the northern cemetery at Maryport may have derived from a pyre site but alternatively came from a burial or Aschengrube.

into a coherent plan. The two hearths contained charcoal, animal bone, hundreds of nails and Hearth B a little cremated human bone. The features are too small to be pits in which bodies were cremated, and no evidence for in-situ burning is reported. These are more likely to have been pits of pyre debris. Contemporary cremation burials were not recovered.

Mucking (Essex) (Cambridge Archaeological Unit 2008: RB Cemetery II Gazetteer) A near square feature, 0.97 x 0.84 m and 0.15m deep, its fill comprising a mass of burnt material including oak, poplar, and hazel charcoal, carbonised pine kernels, date stones and hazel nut shell fragments, fragments of fired clay, and a large assemblage of burnt and broken ceramics (eight platters, three tazze, eleven similar lamps, five beakers) and (probably burnt) coins (nine sestertii, latest datable issue giving a tpq of AD 175). A small quantity of bone fragments was recorded, but no certain human remains have yet been identified.

A slightly larger ‘hearth’ (1.3m by 0.8m) containing charcoal, pyre debris and some cremated human bone, was dismissed by the excavators as a cremation-related feature because of its late Roman date, an argument that would now be rejected, the late date being no reason to dismiss the feature as cremation-related. Its status as a pyre related feature or Brandschuttgrab. There were no contemporary cremation burials and ten inhumations St Albans St Stephens (Herts.) (St Albans Museums Archive) Ten possible pyre related features were excavated in this mid 1st to early 3rd century civitas capital cremation cemetery. The ten features were all roughly rectangular, ranging from 1.1 x 0.6m to 1.75 x 0.7m (undisturbed 1.5 x 0.8m), and vary in depth between 0.07 and 0.5m (undisturbed 0.25 to 0.4m). They were much larger than most certain cremation burials. In-situ burning was visible in all features save one, usually along the upper edges of the pit. The fills comprised large deposits of charcoal, nails, pot sherds, burnt animal bone and small poorly burnt amounts of cremated bone (98.2-410.3g), never derived from more than one individual. Grave goods were not recovered. There was no evidence of any monument over these pits. There were two groups of such pits, located behind the burials that lay on Watling Street.

Ospringe (Kent) (Whiting et al. 1931: 20) A patch of clay burnt red and a ‘great deal of wood ash’ from a 1st to 4th century cremation and inhumation cemetery associated with a minor centre. Closer identification as a pyre site or other cremation feature is impossible. Pepper Hill, Springhead (Kent) (Biddulph 2006a: 11-14; Boston and Witkin 2006: 45-47, 58-59) Several cremation-related features were documented in this excavation of a substantial cremation and inhumation cemetery, its main period of use lying in the mid 1st to early 2nd centuries AD. Pyre sites, almost all of 1st century date were recorded in the centre of the cemetery, associated with a small cluster of busta of similar date. The pyre sites comprised generally rectangular cuts with burnt soil on their bases and at their edges, with abundant charcoal in their fills, including oak and with smaller quantities from other taxa, including ash and cherry. All had been truncated by subsequent burials. The cuts measured on average 0.49 m by 0.94 m and were up to 0.22m deep. Two (11182, 11823) had extensions on their sides. They were distinguished from busta by the small quantities of cremated bone, with two exceptions from 1-282g. The remains of a minimum of a single individual were present in each case, with one exception (10687) where fragments from at least three individuals were noted. The excavator interprets this as meaning only the remains from the final cremation were present, but it might also mean that some pyres were only used once.

McKinley interprets these features on the basis of the lack of the expected amount of bone as pyre sites rather than busta, but pyre sites used for a single cremation only. The arguments for dismissing other hypotheses (3.3.2) support her conclusions. South Cadbury (Somerset) (Barrett et al. 2000: 105-117) A complex deposit in the south-west gate including burnt and unburnt fragmentary human skeletal material from a minimum of 22 individuals, as well as metalwork including brooches (some burnt) and fragments of arms and armour. The deposit, the product of several episodes spread across the middle decades of the 1st century AD, formed from multiple possible processes, including the violent destruction of the gate, the deposition of pyre debris, structured deposition of objects and body parts from secondary burial rituals.

Pyre debris in the form of small deposits of burnt soil, charcoal and cremated bone was found in other features, including graves and other small pits. The oval feature 10613, a large pit in the centre of the cemetery, near the pyre sites, was much larger (4.4 m long by 3.4 m wide and 1.1 m deep). It was dug and filled during the 1st century AD. Its filling comprised twenty three separate deposits of pyre debris mixed in varying proportions with soil, interleaved with ‘clean’ deposits, perhaps laid to seal earlier fills. The cremated bone included the remains of at least 10 individuals, as well as some burnt animal remains.

South Shields (Tyne and Wear) (Snape 1994) In the southern corner of a small-scale excavation of the cemetery south of the fort the fill of gully 54 included a large amount of charcoal and coal with fragments of cremated bone, glass, hobnails and molten lead. Ceramics were recovered from the same feature but not reported in detail. The feature is pyre-related but more precise identification is not possible. Two other patches of charcoal were noted nearby. To the east were two Brandschüttungsgräber of 2nd century date.

Pyre sites for later cremations were not recorded, though a cobbled surface west of the cemetery is tentatively identified as a possible place of burning.

York Trentholme Drive (N. Yorks.) (Wenham 1968) A portion of a pyre related feature was excavated within this 2nd to 4th century cremation and inhumation cemetery. It extended over 7.5m in length, taken from a spread estimated to be roughly circular with a radius of 9.5m, although no evidence is presented for this. The depth of the material, composed of layers of pyre debris, was 0.4m at its thickest. No contemporary cremation burials were interred within the feature, but later inhumation graves cut through it. As

Roden Downs (Berks.) (Hood and Walton 1948: 13-4) The earliest phase of the site, probably of late 1st or early 2nd century date, consists of spreads of pyre debris across an area 15m by 7.5m, with the heaviest concentrations around two features interpreted as ‘cremation hearths’. Associated postholes were also documented though these cannot be resolved

170

CONCORDANCE AND APPENDICES

cremated bone from the feature was not examined it is not absolutely certain that the feature was related to cremation rather than another part of funerary ritual.

abundant, with fragments of brooches and burnt human and animal bone also present. Avenches (Switzerland) (Castella 1987: 21, 91) ‘Tomb 36’ was a 1.75m x 0.85m x 0.15m rectangular pit with burnt walls and base, filled with charcoal, a small amount of cremated bone (5g) and a large number of ceramics which had been burnt and broken but not dispersed by sorting. Castella therefore suggested that the human bone had been removed from the fill of this ‘bustum’ and that a second ceremony had taken place within the pit which accounted for the spread of ceramics in the pit. It seems more straightforward to argue that this feature was a pyre site. It was situated to the east of over thirty cremation burials of the mid 1st to 2nd centuries AD which lay to the immediate east of the road connecting the port and town at Avenches (Fig. 7.8a and 7.8b).

Westhampnett (W. Sussex) (Fitzpatrick 1997a) Twenty one Iron Age and Roman period pyre sites and pyre related features were identified at Westhampnett in association with 1st century BC and 2nd century AD cemeteries. There was a clear spatial division between the Iron Age pyre features and burials; the former were located on the edge of the burial area, especially to its north-east (Fig. 6.81). The clearest examples of pyre sites consisted of T (e.g. 20645) or X (e.g. 20283) shaped cuts. The length of cuts varied between 0.7m and 2.6m, their width between 0.2m and 0.5m, and depth between 0.1 and 0.2m. The cuts are interpreted as ventilation flues for the base of the pyre, the X or T shapes allowing for different wind directions. The debris in the fill comprised cremated human bone, remnants of artefacts and animal bone, nails, charcoal, burnt soil, burnt flint and gravel. Evidence for burning of the ground surface was rare. Charcoal and charred logs may have lain in situ in some of the features, for example 20121 and 20823, but it is difficult to extrapolate from them to the structure above. They do not contradict the normal reconstruction of layers of timbers at right angles to one another. Gale (in Fitzpatrick 1997: 82) argues that the relatively abundance of ash and oak timbers may be due to their position on the cooler exterior of the pyre. Traces of burnt daub and the presence of many nails and other ironwork suggest re-use of building materials as cremation fuel.

Blicquy (Hainaut, Belgium) (de Laet 1972: 24) A single feature in Trench 63/31, immediately outside the zone of burial, consisted of a rectangular pit 1.6 x 0.75m, 0.9m at its deepest. A 0.1-0.15m layer of charcoal, sherd and tile fragments and very small pieces of cremated bone lay on the curving base of this feature. There were no traces of burning on the soil. ‘Les Bolards’, Nuits-St-Georges (Côte d’Or, France) (Planson 1982) Pyre features were derived from the late 1st and 2nd century cemetery to the south-east of the minor centre. For several metres the fills of the north-eastern and south-western cemetery ditches included large but unquantified deposits of fragmentary ceramics, pipeclay figurines and animal bone including horse and dog. The base of the north-eastern ditch showed evidence of burning. Planson argues that the ditch fills relate to the ustrinum but in the absence of full analysis of cremated bone it is impossible to assess this.

The mixing of pyre debris and later damage impede an estimate of how often any of these pyre sites were used, but unequivocal evidence for the presence of more than one individual in the cremated bone assemblages from pyres and pyre related sites was not recovered. Other features of less regular form may represent pyre sites or dumps of pyre debris. Their fills were similar to those of pyre features and contained small amounts of cremated bone.

Destelbergen (East Flanders, Belgium) (de Laet et al. 1970; van Doorsaeler and Rogge 1985: 159-65; Polfer 1996: 21). The pyre-related feature, dated to the second half of the 2nd century, lay within a rural cemetery of unknown extent. It comprised a rectangular ditch (13.3m x 1.4m x 1m) filled with debris comprising charcoal, ceramics, brooches, coins, molten glass, bronze and worked bone objects, nails and cremated bone from a minimum of fifteen individuals. The feature was initially interpreted as a mass grave related to a catastrophe but is now a ‘type site’ for dumped deposits of cremation debris.

The weights of cremated bone from graves vary between 30.9 to 999.2g, with an average of 301.1g. The average of 72.9g from pyres and related features is considerably lower; ranging from 0g to 422.7g. However some of the features may have been graves rather than dumps of pyre debris. Pyre debris was after all present in many of the graves, albeit only in small amounts with the exception of grave 20252. Pyrerelated feature 20300 for example contained a substantial cremated human bone assemblage (264.3g), and fragments of a single brooch and pot. The irregular shaped feature 20770 contained a cremated human bone assemblage of 359.9g; some of this bone may have been collected before deposition. 20258, 2.2m long, 0.9m wide, subrectangular, with evidence for in situ burning and a large cremated bone assemblage (369.8g), is not dissimilar to a bustum. Against an interpretation of these features as Brandschuttgräber or busta is their spatial association with pyre sites on the margin of the burial area.

Dillingen-Pachten (Saarland, Germany) (Glansdorp 2005: 54, 482-3) Four areas of burnt debris, with limited recording. ‘Brandschicht 1’ comprised a layer of charcoal 15-25 cm thick over an area of at least 19m by 7m otherwise generally clear of cremation burials. The presence of cremated human bone, coins, sherds, nails and other artifacts was noted. ‘Brandschicht 2’ was also extensive (10 x 6m) but again little information is recorded about it and other burnt areas. Dreiborn (Luxembourg) (Metzler 1976) A partially destroyed walled rural burial enclosure included a small number of 3rd century cremations and 4th century inhumations and a pyre-related feature, a 2m x 2.5m subrectangular spread of charcoal and burnt ceramics, which is impossible to assign to a particular category of feature.

2.1.4 Temporary pyre sites from other provinces Altforweiler (Saarland, Germany) (Ames-Adler 2004) A c. 10 x 8m irregular area of charcoal up to 0.3 m thick, part of an originally larger pyre area heavily truncated by ploughing and other damage, at the northern end of a small rural cemetery (52 burials) of the 1st to 3rd century AD. Pits containing pyre debris had been dug in the same area. Artefactual material, in particular ceramics and nails, was

171

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Haltern (Nordrhein-Westfalen, Germany) (Berke 1991: 15152) Pyre sites comprised many large interconnected irregular concentrations of charcoal, sherds and small finds, scattered among the burials and monuments. These features are very shallow and lack evidence for heating through discoloration.

sieved fills contained small amounts of cremated bone, varying from a few splinters to several hundred grams. The pits were on the south-western side of the road and behind enclosures XXIX to XXXII which lay on the road frontage. On this combination of evidence Witteyer interprets these features as one-off pyre sites.

Histria (Romania) (Ota 2007: 77, 96, Pl. III) T. XXX, a sub-rectangular spread of burnt material 3.15 x 2.5m in extent, crossed by two narrow trenches at right angles to one another, one 3m and the other 2.1m long, both 0.4m wide, with a smaller sub-circular pits (0.4-0.6m in diameter and 0.2-0.3m deep) at each corner of the area. The cut features were filled with burnt debris including cremated bone, charcoal, ash, nails, and burnt artifacts.

Overpelt (Limburg, Belgium) (Vanderhoeven 1973) Ustrinum X. 6.25m x 2.1m x 0.5m. Ustrinum Y 5.75m x 1.25m x 0.4-0.6m Both features were elongated elliptical pits, with sides and base burnt and overlain by a 0.1-0.2m thick layer of charcoal containing a little cremated bone and burnt artefactual material. A layer of unspecified pit fill was sandwiched between this and an upper layer 0.1-0.2m thick of very clean cremated bone free from charcoal or artefactual material which lay beneath the topsoil. No good reason is offered for the layering of the different fills of pyre debris and cremated bone; it is not easy to envisage a post-depositional process to account for this sorting. Nor is it possible that these features are busta; the size and shape of the features is unlike other busta and the dense mass of cremated bone cannot be accounted for by a single cremation. The features were ten metres from one another within a 2nd century rural cremation cemetery.

Kempten (Bavaria, Germany) (Mackensen 1978: Beil. 1; Faber 1998: 160-70) Four features were indicated as pyre sites (nos. 51, 60, 74, 102) by Mackensen and published in full in the publication of non-burial features from the cemetery (Faber). All were shallow depressions or cut features 3.2-4m long, 1.75-3m wide and 0.30-0.70m deep. Evidence in varying quantities of pyre debris was recovered from all of them, including charcoal, ash, fragments of ceramics, glass and metal and cremated bone. No evidence for the form taken by the pyre survived, except for groupings of cobbles at the base of some features which perhaps facilitated supply of air to the pyre base. Scattered across the cemetery, the features were in association with individual groups of burials and enclosures. Faber (1998: 170) plausibly proposed that many other cremation areas had originally been scattered across the cemetery, all trace of which had been removed by truncation of its ground surface.

Rusovce (Gerulata, Slovakia) (Kraskovská 1976: 4-5) A pyre related feature was recovered within a cremation cemetery related to a minor centre. Grave 94 was a 3m x 1.8m x 0.9m rectangular pit. Its walls were burned to a height of 0.3m and strips along the base of the pit were also burnt and on both walls and base were traces of timbers 0.15-0.25m in diameter. The pit fill comprised burnt pottery, nails, glass and charcoal and a small amount of cremated bone distributed around pit edges. The feature was interpreted as a pyre site on the basis of its size, form and the sorting of the fill.

Landscheid (Rheinland-Pfalz, Germany) (Schindler 1973) A pyre feature was located just to the south-east of the centre of an almost 6m square walled rural burial enclosure of the late 1st to mid 2nd centuries, while burials had been placed around the edge of the enclosure. It comprised a 2.4m x 1.2m rectangular pit, possibly wood lined, with a charcoal fill containing cremated bone and burnt artefacts which seems likely to have been a pyre site. The lack of analysis of cremated bone makes it impossible to assess the number of times that it was used.

Saint-Paul-Trois-Châteaux (Drôme, France) (Bel 2002: 778) Five fire-reddened depressions were identified as possible cremation sites. In one case joining artifacts allowed connection with an adjacent tomb (features 70 and 71). Sampont (Luxembourg, Belgium) (Noel 1968: 18) Nine circular pits, varying from 4m² to 16m², 0.3m deep with a layer of pyre debris 0.1m deep with fragments of timber, cremated bone, burnt ceramics and nails. The sand beneath was burnt to a depth of several centimetres. The features were argued by the excavators to be concentrated on the north-eastern edge of the 1st and 2nd century rural cemetery but the full extent of the cemetery was not defined.

Lazenay, Bourges (Cher, France) (Troadec 1993) A spread of cremation residue in the probable pyre area 1520m in diameter and 0.2m at its thickest with many burnt ceramics and cremated bone, from a large rural cremation and inhumation cemetery in use throughout the Roman period.

Schankweiler (Rheinland Pfalz, Germany) (Ludwig 1988: 61) A 2m x 1.6m subrectangular area of charcoal 0.2m thick with small amounts of ceramics and cremated bone. A strip of ash lay to the north of this with a setting of stones that may have part defined the cremation area. The pyre area lay on the northern periphery of a large 1st century AD rural cemetery of over eighty cremation burials, but its small size suggests that it cannot be the only pyre area.

Lellig (Luxemburg) (Thill 1970) A 2.1m x 1 x 0.4m sub-oval layer of charcoal, burnt ceramics, glass, bronze and iron, located in the south-eastern half of a rural walled cemetery (9.2 x 8m) of the second half of the 1st century AD. Cremation burials were located in the north-western half. Lack of analysis of any cremated bone makes full assessment of the feature impossible. Mainz-Weisenau (Germany) (Fasold and Witteyer 1995: 20; Witteyer 1993) At least thirty intercutting rectangular pits, on average 2m x 1.5m and 0.2-0.5m deep beneath Roman ground level in the 1st and 2nd century cemetery that lay alongside the road from Mainz to the fort at Weisenau. The features were burnt around the sides and contained pyre debris, charcoal, burnt coins, ceramics and glass distributed around edges of pit. The

Seebruck (Bavaria, Germany) (Fasold 1993a: 89, Beil. 2) Two features c. 20m apart c. 4.5m long, 1.3m-1.8m wide and up to 0.25m filled with charcoal, large pieces of charred wood, cremated bone and burnt ceramics which are not reported in detail. Around the west of Ustrinum 1 lay a setting of boulders which may have partly enclosed the feature. No evidence for in-situ burning is recorded.

172

CONCORDANCE AND APPENDICES

Ustrinum 2 was less well preserved as it had been cut by several later burials. Large pits (Abfallgrüben) between the two features contained further pyre debris. The features were in the centre of a late 1st to early 3rd century cemetery of a minor centre. In the north-western part of the cemetery burials and monuments were close to the road whilst the pyre features were located away from the road frontage and closer to the ditch which formed the other boundary of the cemetery.

former from the Claudio-Neronian period to the early 2nd century, and in the latter during the 2nd century. Wederath (Rheinland Pfalz, Germany) (Kaiser 2006) Extensive excavation of this Iron Age and Roman period cemetery associated with the small town of Belginum has documented hundreds of pyre-related features, 434 Aschengruben, 120 Aschenflächen and 38 features for which there was insufficient evidence to classify them. These are scattered throughout the cemetery, generally in individual instances in association with particular enclosures and groups of burials, with some clustering close to the route through the cemetery. At the eastern edge of the cemetery is the largest of the Aschenflächen, measuring c. 20m x 26m, as well as a cluster of pits and spreads of debris. The form taken by the Aschengruben varies considerably, c. 35% being near circular (with a diameter between 0.3 and 1.2m) and 22% approximately orthogonal in plan (measuring between 0.2m x 0.2m and 1.5m x 1.1m). Most were between 0.11m and 0.25m deep though some were deeper to a maximum of 0.75m. Generally they took the form of hollows, though some had steeper sides and a flatter base. In 98 instances (23%) there was evidence for scorching to the sides or base, mostly to the former. Single postholes were associated with four examples. The fills included charcoal, ash and sometimes cremated human bone (84 instances) as well as burnt and fragmentary artefactual material including ceramics, nails and glass and metal objects. The Aschenflächen comprised spreads of similar material. They varied from 0.2m x 0.45m to 20 x 26m. Larger examples of Aschengruben may in some cases have served as ventilation pits over which the pyre was set up and later collapsed: this was most plausible in the case of AG 361, a 1.5 x 1m rectangular pit. The fill at the base of the 0.58m deep feature included a layer of burnt stones, above which was compacted charcoal and abundant remains of burnt and broken ceramics. Further charcoal, ceramics and cremated human bone were documented in the upper fills. In general the charcoal rich fills of the Aschengruben and Aschenflächen layers are likely to represent redeposited pyre debris: the significant heating indicated by the evidence for burning and fragmentation of artefacts suggests that these were placed on the cremation pyre, either in association with the deceased or after use in rituals contemporaneous with cremation.

Septfontaines-Dëckt (Luxembourg) (Polfer 1993; 1996: 1618; 2000) A sub-rectangular spread 15.75m x 11.25m of burnt material, of which the depth varied generally varied between 0.1m and 0.2m, at its greatest 0.53m, although it had been extensively eroded. No stratigraphy was observed within this spread which contained a very large ceramic assemblage. It was impossible to relate the possible posthole features beneath it to a coherent plan. Whether or to what extent the soil beneath was burnt is not recorded. Several pits of debris with similar fills were excavated on the margin of the feature. The pyre site was approximately 20m north-west of a large rural cremation cemetery of the late 1st to early 3rd centuries. Tongeren (south-west cemetery) (Belgium) (Vanvinckenroye 1963: 37, Afb. 3) Small-scale excavations in the town’s south-west cemetery revealed a 6 x 4.5m sub oval pit filled with charcoal, cremated bone, burnt earth, and burnt and broken artefacts, pottery, coins, nails, glass, bronze. It was cut by later cremation and inhumation burials. The feature was either a pyre site or debris pit. Other pyre related features were excavated but not reported in detail Tongeren (south-west cemetery) (Belgium) (Vanvinckenroye 1984) Larger scale excavations of a 1st and 2nd century cemetery south-west of the town revealed several pyre features; ustrina 1 and possibly 4 appear to have served for a single burning, the others perhaps for repeated cremation. Ustrinum 1. 2.1 x 1.2m rectangular pit, with burnt sides and base and filled with debris. Parts of the pyre structure, including timbers nailed together appeared to be lying in situ, and the cremated human bone, derived from one individual only, still lay in approximate anatomical relationships. Ustrina 2 and 3. 5m x 2.5m x 0.7m subrectangular pits. Burnt sides, filled with charcoal, cremated bone and nails. Ustrina 4, c. 4m x 2.5m, subrectangular pits, also with timbers nailed together from pyre or bier. Associated with grave 289, probably lay beneath barrow. Vatteville-la-Rue (Seine-Maritime, France) (Lequoy 1987) A subrectangular spread of pyre debris including charcoal, burnt ceramics, other artefactual material and cremated bone, roughly 20m by 10m, 0.6m at its deepest, was excavated on the southern periphery of this 1st to 3rd century AD rural cremation cemetery of 440 burials. Cremation may have taken place in a depression on the northern side of this area, from which the debris was then discarded across the surrounding area. A wall had possibly acted as a part enclosure for the pyre area. Velzeke (Oost-Vlaanderen, Belgium) (Van Doorsaeler and Rogge 1985) Two rectangular features, one 6.8m x 2.2m and 1m deep and the other 6.2-6.4m x 5.2-5.4m and 2.4m deep both contained layers of pyre debris, cremated bone and burnt clay, the gradual accumulation of debris from a series of cremations in a rural cremation cemetery. Material was deposited in the

173

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

2.2 A sample of busta Note: ‘Verziegelung’ has been translated by baking; ‘(P)’: measurements taken from plan; U: Unknown Description

Quantity of cremated bone

Depth of pyre debris

Length

Width

La Calade (Bérard 1961) Rectangular grave cuts with walls and / or base reddened by heat. Parts of framework of timbers at right angles to one another preserved (grave 15). Colchester Handford House (Orr 2010) Two busta, F47 (a) and F134 (b), the former rectangular, the latter oval, both grave cuts narrowing to their bases and having two fills, the lower with a much higher concentration of cremated bone and charcoal. In F47 the bone was distributed in anatomical order, in F134 it was concentrated at one end. Erbericher Hof (Gaitzch and Werner 1993: 56-57, Abb. 2) Baking of edges and sides. Ergolding (Struck 1996) (Graves 18 and 76) Oval and rectangular grave cuts with baked sides and edges. Kempsford Quarry (Booth and Stansbie 2007) Graves 4857, an irregular linear pit with cremated bone and charcoal mixed in upper fill, with evidence of in-situ burning. Krefeld-Gellep (Pirling 2002) 215 examples identified, mainly of 1st and 2nd century date. The figures given are for the commonly occurring examples, which are mainly rectangular and straight-sided, with occasional insitu evidence of charred logs. In some cases cremated bone had been collected and redeposited, in most cases in a probable organic container, sometimes unaccompanied by unburnt objects. L’Hospitalet du Larzac (Vernhet 1987) Oval and rectangular grave cuts. London Eastern cemetery (Barber & Bowsher 2000, 62). A single example, a rectangular cut reddened at top, containing lengths of charred wood with human skeletal elements and other burnt objects of which the arrangement was not recorded, damaged by a later inhumation burial, surmised not to be in correct anatomical order. Lisbon. Encosta de Sant'Ana (Angelucci 2008; Gonçalves et al. 2010) Burial 1. An oval feature excavated in the bedrock with its rim built up in a mud-brick technique, with heat-affected zone below. The fill contained burnt and unburnt human and animal remains, charcoal, burnt sediment, and burnt remains of an unguentarium, ceramics and nails. Gonçalves et al. (2010: 126) designate this as a bustum on the site of a feature previously used as an ustrinum, but evidence for previous use is not convincing. La Favorite, Lyon (Type 1A) (Tranoy 1987: 44) Rectangular or square grave pits, walls reddened. Mainz-Weisenau (Witteyer 1993) Rectangular, square and oval grave cuts. On the edges of one grave cut were traces of timbers parallel to the grave sides, in base of others traces of parallel lengths of timber. Pyres must have been established over and within the pits.

U

0.1-0.15m

1.6-2.2m

0.8-1.3m

(a) 1368g (b) 1818g

0.05 0.2

2.1m 1.35m

0.7m 0.9m

U

Very little 108g and 495g

0.2-0.5m (P) U

0.4-1.25m (P) 0.68m

0.05-0.15m (P) U

665g

c.0.1m (P)

0.9-3.5m (P) 1.86m (diameter) 2.2m length 2.5m

(b) 0.6m

U

6-1200g

U

1.5-1.8m

0.5-1.8m

0.02-0.04m

U

U

1-3m

0.5-2.5m

U

835g

0.3m

1.8m

1m

0.02m

529.85g

Up to 0.15m

2m

1.3m

c. 0.04m of reddening, heat effects documented to 0.15m below base.

Little

0.1 to 0.2m

2.2m

1.5m

0.02-0.03m

U

U

2.1-3m

1.5 x 2.3m

U

174

c.

Depth of burning to sides / base U

CONCORDANCE AND APPENDICES

Noviodunum, Galaţi-Barboşi and Tomis (Ota 2007) 176 examples from Moesia, primarily from these three sites. Most are covered by earth mounds, some with tiles. Timbers from pyre structure occasionally survive. Two sub-types of pit can be identified from difference in form, (a) simple rectangular and (b) with stepped sides. Moers-Asciburgium (Bechert 1979: 490) Rectangular grave cuts, baked edges and bases. Timbers preserved in base of some grave cuts. Nave (Passi Pitcher 1987) A strip of baked clay around the edge of the burial cut. Preserved timbers at right angles to one another in grave 57. Neuss (Müller 1977: 14) Rectangular and oval grave cuts, baked bases and sides. Neuss Münsterplatz (Härke 1980: 534, 542) Rectangular grave cuts, sandy subsoil of sides reddened but not baked. Petty Knowes (Charlton and Micheson 1984) Eight busta. Rectangular grave cuts, stake holes around edge of cut of 15. Pepper Hill Springhead (Biddulph 2006a; Boston and Witkin 2006) Eight busta, rectangular cuts, sides and bases burnt, with charcoal fill at base, higher concentrations of cremated bone in upper layers of fill. Winchester Lankhills((Booth et al . 2010: 502-4) Seven likely examples, rectangular cuts with in-situ burning, spread of cremated human bone on charcoal layers in limited quantities, carbonised plant material from grave environs. Worms (Grünewald 1990) Rectangular grave pits, sides and bases burnt, frequent traces of pyre timbers at right angles to one another on base.

U

Variable

a. 1.8-2.5m b. 1.92.85m

a.0.4-2.3m b. 1-1.9m

0.02-0.15m

U

0.15m

1.25-2.5m

0.9-2.2m

0.05-0.07m

U

0.03m

Little

?

1.25-4.2m

0.65-3m

Little

0.01-0.05m 0.04-0.05m

Very little

U

1-1.8m

0.5-1m

U

Undisturbed burials, 730g, 1211g, 1424g. Disturbed 34886g 171-1641g (in six of seven cases > 1000g)

0.2m max

1.5m (avg.)

0.6m (avg.)

0.06-0.13m

-

1-2.44

0.39-1.14

U

1.7-2

0.6-1.1

Several hundred grams (urned), few grams (unurned)

Experimental (Gaitzch and Werner 1993) Baking of sides.

0.1-0.3m

175

0.05-0.10m (P)

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

APPENDIX THREE DATA FOR FIGURES IN CHAPTER 4 Appendix 3.1 Note on unpublished cemeteries from Baldock and St Albans Unpublished data were made available by Ros Niblett (St Albans) and Gil Burleigh and Mark Stevenson (Baldock) and I am grateful to them for help and discussion related to the archives in their respective care. St Albans St Stephens. Data on the 1930s excavations are derived from Davey’s published report (1935). Data from the unpublished excavations of the 1980s at St Albans St Stephens by Hertfordshire Archaeological Trust were compiled from the site archive, Niblett’s preliminary catalogue of burials and McKinley’s catalogue of cremated bone assemblages. Niblett’s spot-dating of ceramics was used as the basis for burial date. Folly Lane. Data were initially derived from the archive catalogue and Niblett’s interim reports (1992; 1995a). They were subsequently updated with reference to the site publication (Niblett 1999). Baldock The relatively small scale examination by Stead of Baldock’s cemeteries has been complemented by much more extensive work from the 1980s onwards. Different site nomenclature has been offered by different workers and the publication of the cemeteries by Burleigh and Fitzpatrick-Matthews (2007) supplies a concordance of site names. Draft versions of the catalogue and discussion on the Wallington Road cemetery and several small LPRIA cemeteries were made available to me, but only information from Wallington Road is used in detail in this thesis. This cemetery has now been published (Fitzpatrick-Matthews and Burleigh 2010). Their published discussion contains some revisions, meaning that there are some minor differences in statistics reported, for example in quantities of cremated bone per burial or of ceramics deposited. These do not affect the general burial patterns established here on which the argument is based. Brief selective reference is also made in chapter 5 to the online publication of other cemetery data from the town (Burleigh and Fitzpatrick-Matthews 2007). Concordance of site names for Baldock cemeteries referred to in this study Stead and Rigby 1986 Burleigh 1993 / 1995 N. Herts Arch. Service Archive Site A Upper Walls Common

Burleigh & FitzpatrickMatthews 2007 BAL 23

Name used

Area 15

BAL-15

Area 30A / 31

BAL-30

The Tene

Site D Walls Field Site P Walls Field Site S Upper Walls Common

Royston Road (4) (plus adjacent areas south of the Icknield Way) The Tene (6) (Iron Age) and Roman (see also Burleigh 1980) Clothall Road (8) Walls Field (1) Wallington Road (2)

Downlands Enclosure A Royston Road

Area 36 Area 25 Area 11

BAL-36 BAL-24 BAL-11

Site R Convent of Providence

Convent of Providence

Area 28

BAL-28

Icknield Way East (5)

Area 45

BAL-45

California (3a and b) Iron Age shrine with burials

Area 1 Areas 2-4

BAL-1 BAL-2

Clothall Road (8 Walls Field Wallington Road Convent of Providence Icknield Way East California Mercia Road

Site E, Upper Walls Common (Fig. 8, mislabelled at base) Site J The Tene (Fig. 9 and 36)

Appendix 3.2 Data for figures in chapter 4 (4.6-4.8, 4.11-4.13, 4.17-4.18, 4.20-4.21, 4.23-4.28) Figure 4.6 The quantity of cremated bone in burials from a sample of Romano-British cemeteries Weight (g) St.A KHL St.A SS Br. Sk.G Br. 'B' Bal. WR 0-249 27 33 8.5 12 27 250-499 19 15 19 24 21 500-749 19 18 30 24 21 750-999 17 15 13 32 13 1000-1249 8 8 15 0 5 1250-1499 6 6 8.5 4 9 1500+ 4 5 6 4 4

Win. VR 31 12 16 11 13 9 8

St A. KHL St Albans King Harry Lane; St A. SS St Albans St Stephens; Br Sk. G. Braughing Skeleton Green; Br. ‘B’; Braughing ‘B’; Bal WR Baldock Wallington Road; Win.VR Winchester Victoria Road Figure 4.7 Numbers of different ceramic forms, St Albans St Stephens (all excavations) Flagon Beaker Bowl Jar Dish Cup No. of vessels 91 74 18 22 29

176

Lid 14

Vessel 52

85

CONCORDANCE AND APPENDICES

Figure 4.8 The number of burials with different numbers of ceramics, St Albans St Stephens (SS1 = Davey excavation; SS2= HAT excavations) No. of vessels 0 1 2 3 4 5 6 7 10 No. of burials SS1 40 26 13 7 3 2 1 2 0 SS2 222 57 35 13 5 1 4 1 1 Figure 4.11 Numbers of different ceramic forms, Baldock Wallington Road Jar Flagon Beaker Dish Cup No. of vessels 24 35 29 42

Other 9

27

Figures 4.12 The number of burials with different numbers of ceramics, Baldock Wallington Road No. of vessels 0 1 2 3 4 5 No. of burials 52 43 31 14 0 1 Figure 4.13 The number of burials with different numbers of ceramics, Baldock Walls Field No. of vessels 0 1 2 3 4 5 6 7 No. of burials 7 57 52 35 17 2 5 1 Figure 4.18 Numbers of different ceramic forms, Braughing Skeleton Green Flagon Beaker Bowl Jar Dish Cup No. of vessels 39 31 4 1 31

8 1

Lid 4

9 2

Vessel 0

2

Figure 4.19 The number of burials with different numbers of ceramics, Braughing Skeleton Green No. of vessels 0 1 2 3 4 6 No. of burials 0 11 19 14 3 1 Figure 4.20 Numbers of different ceramic forms, Braughing B Flagon Beaker Bowl Jar No.,of vessels 79 48 9 0

Dish

Cup 31

Vessel 6

1

Figure 4.21 The number of burials with different numbers of ceramics, Braughing B 6 No. of vessels 0 1 2 3 4 5 1 No. of burials 9 13 14 9 1 0 Figure 4.23 Numbers of different ceramic forms, Welwyn Flagon Beaker Bowl Jar No.,of vessels 57 47 0

Dish 0

Cup 47

Vessel 10

6

Figure 4.24 The number of burials with different numbers of ceramics, Welwyn No. of vessels 0 1 2 3 4 5 No. of burials 15 21 22 20 2 1 Figure 4.25 Percentage of burials with ceramic accessory vessels from urban & minor centre cemeteries Figure 4.26 Average number of ceramic accessory vessels in burials from urban & minor centre cemeteries Figure 4.27 Samian as a percentage of ceramic accessory vessels from urban & minor centre cemeteries Figure 4.28 Percentage of burials with glass accessory vessels from urban &minor centre cemeteries

Baldock Wallington Road Baldock Walls Field Braughing Skeleton Green Braughing B Welwyn Grange St Albans St Stephens (Davey) St Albans St Stephens (HAT)

WLR WLF SKG BRB WLN SS1 SS2

% with acc. vessels 67 93 100 87 83 35 58

Av. no. of acc. vessels 1.28 2.27 2.28 2.05 1.65 1.14 1.07

177

% of samian vessels 28 29 28 17 32 13 16

% with glass. vessels 0.5 5 10 2 1 3.2 2.1

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

century AD, though it did not investigate areas on the eastern margins of the cemetery where Clarke identified the latest burials in his sequence (Booth et al. 2010: 453-62). These reservations are significant for the phases established and dated at other cemeteries in Winchester during the 4th century AD by reference to Lankhills.

APPENDIX FOUR DATA FOR FIGURES IN CHAPTER 5 Appendix 4.1 Note on cemeteries from Winchester, with particular reference to cemetery chronology Data on unpublished cemeteries from Winchester were collected in February 1995. They were made available by Ken Qualmann and I am grateful to him and Malcom Gomersall for help and discussion related to the archive. Data for Victoria Road East were derived from a catalogue established by Patrick Ottaway, and his draft discussion, for Victoria Road West, Hyde Street, Oram’s Arbour, Chester Road and St Martin’s Close cemeteries from a catalogue and draft discussion by Qualmann and Gomersall. These have now been published (Ottaway et al. 2012). The published analyses differ little from those of the archive and I have therefore made explicit reference to the publication only where new information was available. I have not corrected the minor differences based on quantification from the archive evidence. Subsequently further excavations have been undertaken by Winchester Museums Service on the Andover Road – Eagle Hotel site close to the site of the Victoria Road excavations; the results are summarised in chapter 5. For characterising burial practice at the Lankhills cemetery, quantified results unless otherwise stated relate only to burials published by Clarke (1979). For the major excavations were undertaken by Oxford Archaeology from 2000 to 2005, the excavation publication (Booth et al. 2010) is referred to for individual aspects of analysis. The excavations of a further 56 graves by Wessex Archaeology in 2008 are unpublished (Chapman et al. 2009: 273).

Victoria Road (West) The first phase is dated to 270-320 from the continuing occurrence of cremation and from burial orientation. It is given a terminus post quem by colour coated beakers with a date range of c.270-400. The transition from predominantly cremation to predominantly inhumation is dated to c. AD 300 but this date lacks a large sample of 3rd century burials at Winchester in its support, although the inhumations excavated on the Cattle Market site might be of 3rd century date (Clarke 1979: 6). Occasional cremation continues to characterise all phases at Victoria Road and Lankhills. The shift to west-east orientation in Phase 2 is dated by the predominance of the same orientation at Lankhills at 300-320 (a slight over-simplification of Clarke’s scheme). However changing orientation at Lankhills is at least partly influenced by the changing relationships of graves to non-burial features and the degree of change is much slighter than the change between phases 1 and 2 at Victoria Road. The characteristics of the third phase, shallower graves, more frequent intercutting, less regular orientation and body position and rare use of coffins are easier to accept as consistent parallels to practice at Lankhills which are there dated to 390 and later. There are small differences in the number of burials attributed to different phases in the archive and in eventual publication (Ottaway et al. 2012: 116-17). The figures from the archive are retained here; the further final phase suggested in the archive notes, for which there is very limited evidence, was included in phase 3 in analysis.

Victoria Road (East) There are two sources for the date of individual graves, stratigraphic relationships and grave goods, usually ceramics. The stratigraphic phasing is not based on burial intercutting, which was limited, but on the level from which graves were cut. More precise dates for of a small number of graves are derived from the ceramics analysed by Matthews, which were generally consistent with the stratigraphic phasing. Together this allows three phases of burial to be assigned, the first two to the late 1st to mid 2nd century AD and the third to the mid-2nd to early 3rd century AD. Very few burials dated to the latter. In order to retain as large a sample of graves as possible a slightly modified version of Ottaway’s archive phasing has been used. Occasional inconsistencies between pottery and stratigraphic dating are resolved by following the ceramic dating where grave goods consistently indicate a different date to the stratigraphy. This yielded 136 phase 1 and 49 phase 2 burials, slightly different to Ottaway’s final published figures (2012: 80-2. 88-9).

Hyde Street The cemetery area is dated to the second half of the 4th century AD on the basis of the similarity to the post-350 phases at Lankhills and the second and third phases at Victoria Road. The coin of the House of Theodosius (Victoria Augg, 388-402 AD) deposited as a grave good in grave 16 supports this date. Chester Road The frequent stratigraphic relationships between graves allow burial phases to be more easily established, although their absolute dating is more problematic. Absolute dates for the cemetery are provided by coins deposited as grave goods, the five coins dated to the 280s recovered from a cremation grave of the second phase (579) and a Vict. Augg. coin of Theodosius dated from 388-395 from a grave of the fifth phase (553). The final phase can perhaps therefore be dated to the early 5th century. Qualmann amalgamated the many phases into six main periods (with a seventh added in publication). The absolute dating was transferred from comparable dated changes in practice at Lankhills. The earliest period (2) at Chester Road was dated by artefacts, north-south orientation and the presence of cremations to c.270 to 350. 3 and 4 were characterised by a shift to broadly west-east orientation and are therefore dated to 350-370. 5 and 6 were dated by coins and increasing irregularity of burial practice to 370-390; the latest graves may date to the early 5th century.

Lankhills Phasing for the Clarke excavations at Lankhills was established initially on the basis of coin evidence. This suggested a horizontal stratigraphy for the site progressing from west to east, although the use of different areas overlapped (Clarke 1979: 114-118). Ceramic evidence supported this chronology (Clarke 1979: 120). Individual graves within area W could also be dated by stratigraphic relationships. Comparison of different areas of the cemetery was then used by Clarke as a proxy for the comparison of trends in burial practice over time. As Wilson (1983) noted, the caution of initial statement of phasing is replaced by increasing confidence in the method used to establish chronology during the course of the report, although only 14% of graves are closely datable by these methods. The Oxford Archaeology publication is based on a simpler comparison of burial practice before and after the mid fourth

Again dating by comparison with Lankhills can be challenged. Many of the arguments already advanced with reference to Victoria Road are relevant here. The changes in orientation to predominantly west-east are of a different kind

178

CONCORDANCE AND APPENDICES

and degree. The increasing variety in the later phases at Lankhills of body position and orientation have not been established at Chester Road, and there is only slight evidence for a decrease in regularity of burial.

Figure 5.11 The number of burials with different artefact types, Lankhills (see figure 5.5) Figure 5.12 Average NAT of burials with different attributes, Lankhills

Oram’s Arbour Abraded mid-3rd century deposits from the earliest ditch fills with burials, 3rd century vessels in two burials early in the sequence and the cutting of the latest burials through a layer associated with a possible hoard with coins dated from 388402 suggests the use of the ditch for burial from the late 3rd to the early 5th century.

Artefact Overall avg Hobnails Animal bone Coin Vessel Comb 2+ vessels Spindle whorl Bracelet Belt fittings Glass Ivory bracelet Pewter Vessel Knife Brooch Rings Pins Beads Silver artefact

4.2 Data for figures 5.4-5.6, 5.8-5.9, 5.11-5.13, 5.15-5.18 NB For all graphs representing percentage values the original numerical values are recorded here. Figure 5.4 Cremated and inhumed populations, Victoria Road East (Juv. 2-18, most not closely definable within that age; Inf. 24m. Barrow, built over a ‘saucer-shaped depression’ contained Roman material. These may represent Roman deposits in an earlier barrow.

207

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Kibworth Harcourt, Leicestershire (SP 6894) (Page and Keate 1907: 213) Linton, Cambs. (Taylor 1993: 222) (TL 5847) Litlington, Limlow Hill, Cambs. (Clark 1938; (Frere and St Joseph 1983: 226-227; Taylor 1993: 225) (TL 32254172)

Roman ceramics recovered in barrow fill. At centre a stone pavement, possibly a cist, from which a bone ‘bodkin’ and iron object were recovered. Primary Roman cremation and secondary Anglo-Saxon inhumations in barrow. Near-square enclosure defined by ditch surrounds barrow in hilltop position, barrow flattened in 1888 and reported as 12.8m in diameter and 5.5m high, with pit beneath. 2nd century AD material reported in enclosure ditch, c. 7m wide, 1.9m deep, enclosing an area c. 64m wide. Skeletons with 1st and 2nd century coins reported near the mound. Barrow, Lead coffin with five unguent bottles, copper alloy armlet and bone pin found when mound destroyed Six barrows possibly of Roman date, no recent excavation, average 15-20m diameter and 5m in height when recorded in 1930s Barrow on bank of river Colne levelled in 1860 with a ‘Roman interment’ in a probable tile cist with ‘‘several gold things and copper coins and a lot of pickle jars with burnt bones in them.’ Radlett kilns half a mile to east.

Meldreth, Mettle Hill, Cambs. (Taylor 1993: 225) (TL 32254172) Six Hills, Stevenage, Herts. (Dunning and Jessup 1936: 50) (TL 2323) Watford, Herts. (Dunning an Jessup 1936: 50; Page 1914: 165) (TQ 1395)

4. Sculpture and inscriptions Site / County / NGR Reference Alchester (found in Sauer 2005; context of reuse in 3121 & 3122 excavation of 3rd century Wall) (SP570203)

Bedford Purlieus, two miles west of Wansford bridge, Thornhaugh and Wansford parishes, Northants. (c. TF 038005) Earith Campground (TL 37757825) Girton, Cambridge, Cambs. (TL 425608) Girton, Cambridge, Cambs. (TL 425608)

RIB

Haverfield 1902; CSIR GB 1.8.35

Henig in Regan et al. 2004, 128-30; Evans and Regan in Malim 2005: 164-6. CSIR GB 1.8.41; Liversidge 1977: 19 CSIR GB 1.8.66; Liversidge 1977: 19

Irchester, Northants. (SP 9265)

RIB 233

St Albans, bed of river Ver to north of north gate (TL 1408)

Blagg 1984

Stanwick, Northants. (SP 972717)

Dix et al. 1987; Frere 1990: 253; 1991: 285 RIB 3136

Stanwick, Northants. (SP 972717) Stowe Nine Churches Northants. (SP624562) Towcester, Northants. (SP 6948) Thornhaugh, Wansford, Northants., 4.5 miles SE of Stamford (TL 075985) Water Newton, Cambs. (TL 1097)

and

Hunn

CSIR GB 1.8.90 CSIR GB 1.8.67 RIB 234 RIB 232

Description Two fragmentary stelae, one near complete, the other preserved only as a fragment. The stele of Lucius Valerius Geminus, a legionary veteran whose monument dates to the mid-1st century AD (and to the site’s military phase), was found reused in the late Roman fortification of the small town. Texts: RIB 3121 Dis Manibus / L(ucius) Val(erius) L(uci) (filius) Pol(lia) Gemi/nus For(o) Germ(anorum) / vet(eranus) leg(ionis) [I]I Aug(ustae) / an(norum) L h(ic) s(itus) e(st) / he(res) c(uravit) / e(x) t(estamento) RIB 3122 D(is) [M(anibus)] / [ Two headless and legless torsos, one 0.59m high, other 0.71m high, Barnack rag, youths in short tunics, each with whip, pair of hunters or charioteers, found in association with large ceramic urn (32 inches in height) containing human bones, glass, a minimum of three samian vessels and large fragments of a hunt cup. Architectural fragment c. 0.43 m in height showing bearded deity, with cornice above carrying remains of feline(?) legs, the latter indicating probable origin from funerary monument. The piece was found re-used in associated with an inhumation burial. Torso of draped male figure, limestone, 0.35m high, found in rubbish pit in Roman and Anglo-Saxon cemetery. Head, paws and tail of lion, fragments of prey and base, limestone, probably from funerary monument, found in rubbish pit in Roman and Anglo-Saxon cemetery. 43 x 20 inches slab from monumental tomb re-used over robbed out later burial on intra-mural site: D(is) M(anibus) S(acrum) | Anicius Satur(ninus) | strator con(n)s(ularis) m(onumentum) s(ibi) f(ecit) A stone with imbricated decoration, possibly derived from a tower tomb. A stone of similar dimensions but with all decoration eroded was recovered from same area. Similar tombs in Gallia Belgica (Hatt 1986) date from the 1st to first half of the 3rd centuries AD. Re-use of fragmentary material probably from funerary monuments in hypocaust flues of 4th century building. Fragments from statues of Minerva, river god, slave girl, barbarian head trampled by horse feet. A fragment of a possible funerary inscription marble, 0.145 x 0.157m, 0.09m thick, found with sculptural fragments described in previous entry: ] T(iti?) f(ilius?) [. Part of architectural relief with nude figure of youth, built into fabric of St Michael’s church, probably of funerary origin. Female head 0.53m in height, limestone, possibly originally set directly on a base, portrait or deity, probably from funerary monument 12 x 11 inches portion from right hand side of tombstone ‘found near Roman buildings’: ...]no | ...] astru |…] XXV | ...] Flau(…) | … Part of a monumental inscription, possible from a tomb, found in context of re-use in Roman period wall: ...] MARTO [...

208

CONCORDANCE AND APPENDICES

Water Newton, Cambs. (TL 1097)

CSIR GB 1.8.65

Lion 0.6m high, 0.77m long, limestone, standing to right with head bent down, probably from funerary monument

Appendix 6.5 Burial assemblages of the ‘East Hampshire Tradition’ Figure 7.7 The distribution of burials in the East Hampshire Tradition Appendix 6.5a Burial assemblages of the ‘East Hampshire Tradition’ – certain examples The descriptions follow the criteria given in appendix 6.3. Site / NGR/Reference No. NAT Assemblage vessels Alton Kemps Yard, Hants. 16 3 1 Claudian ; cremation burial; (SU 716390) (Millett Two flagons, four bowls, two jars, six dishes and two lids; 1986) Shears (basis for excavator’s male sexing), knife. 31* 5* 2 Claudio-Neronian; adult cremation burial in casket; Three flagons, one bowl, three jars, twenty one dishes (including single Drag. 18R) and three lids; Two square glass bottles, gold ring with intaglio (basis for excavator’s male sexing), gaming board and counters, dice, two copper alloy spoons, knife, bead (incomplete). 22 2 3 Claudio-Neronian; adult cremation burial; Two flagons, two bowls, four jars, twelve dishes (Drag. 18) and two lids; Cosmetic set (basis for excavator’s female sexing), pyxis. 14* 2* 4 E. Flavian; adult cremation burial; Two flagons, one bowl, four jars, three dishes and four lids; Cosmetic set (basis for excavator’s female sexing), finger ring (incomplete). 13 + 40 4 5 E. Flavian; adult (25+) cremation burial; Upper fill: one flagon, one bowl, nine jars, one dish and one lid, two sheep/goat lower molars, cow scapula frag. and scatter of cremated bone; Lower fill (separated from upper by plank): three flagons, one beaker, two bowls, four jars, twenty four dishes and six lids; Horse skull, two brooches, two finger rings, iron knife, cosmetic set (basis for excavator’s female sexing). 10 1 6 Flavian; probable cremation burial; Three jars and seven dishes. 5 + 22 2 7 Flavian; adult (30+) female cremation burial in casket; Upper fill: three jars and two dishes; Lower fill: one jar, eighteen dishes and three lids; Brooch. 2+22 1 8 Flavian-Trajanic; adult female (?)cremation burial in casket; Upper fill: One jar and one dish; Lower fill: three flagons, two jars, fifteen dishes (eight of which are very poorly fired) and two lids. 32* 1* 9 Late Flavian; cremation burial; Two flagons, one bowl, five jars, twenty two dishes and two lids (incomplete). Avisford, W. Sussex (c. 24 4 Cist and wooden coffin. In centre a cremation burial in square glass bottle; SU 9705) (Haverfield Three larger and one smaller flagons, twenty larger and smaller dishes of 1935: 49; Roach Smith ‘coarse red ware’, globular glass vessel and handled oval dish with agate at 1848: 123-4) centre, hobnailed footwear, two lamps, two candelabras, and lamps at each corner on some sort of corbel. Chichester St Pancras, W. 11 2 14. Late 1st to early 3rd century AD; cremation burial in two jars; Sussex (SU 8604) (Down Four flagons, beaker, three shallow bowls / cups with frilled rims, plate, two and Rule 1971) bowls, lampholder. 11? 1 90. Late 1st to early 3rd century AD; cremation burial ; Flagon, two jars (possibly cremation containers), jar, two beakers, five dishes. 12 2 127. Late 1st to early 3rd century AD; cremation burial in jar; Flagon, jar, beaker, four dishes, four imitation Drag. 27 cups and other cup; Oyster shells in cup, frags of rib and mussel shell on one dish. 16 2 199. Late 1st to early 3rd century AD; cremation burial in wooden box; Flagon, jar, two beakers, four plates (one including food bones), eight imitation Drag. 27 cups, in sets of two sizes; ceramic lamp. 13 4 201. Late 1st to early 3rd century AD; cremation burial in two jars in wooden box; Two flagons, beaker, two bowls, six dishes including two samian forms (Drag. 35 and Curle 15), two cups including imitation Drag. 27; Ceramic lamp, iron object (poss. lock of box); Animal bones in dishes, oyster and mussel shells strewn around.

209

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

Daneshill, Hants. 657541) (Millett Schadla Hall 1992)

(SU and

Haslemere, Surrey (SU 907335) (Holmes 1949: 29)

Neatham, Hants. 744413) (Millett Graham 1986)

(SU and

Neatham, Hants. (SU 742410) (HCMS Acc. No. 304.1/21-32) Owslebury, Hants. (SU 522247) (Collis 1977b)

15

1

11

4

16

3

13

5

20

1

11

1

15*

1*

15*

1*

14*

1*

65*

1*

48

1

14*

2*

11

3

33

1

13*

U

>36*

2*

10

1*

11

1

>12

207. Late 1st to early 3rd century AD; cremation burial in jar beneath wooden box; Two flagons, jar, nine dishes including seven of same size, three small beakers. 213. Late 1st to early 3rd century AD; cremation burial in wooden box; Two jars, six bowls including two pairs of bowls, two dishes including Drag. 35/36, lid; Slate tablet, copper alloy object, iron object. 214. Late 1st to early 3rd century AD; cremation burial in wooden box; Two flagons, jar, four beakers, two bowls, two dishes, and five cups (four Drag. 27, one Drag. 33); Lead lamp holder, hobnailed footwear. 217. Late 1st to early 3rd century AD; cremation burial in jar in wooden box; Two flagons, three beakers, four bowls, four dishes including one Drag. 35 and one Drag. 36; Glass bottle, lead lampholder, iron knife, copper alloy coin. 1. Late 2nd to 3rd century AD; cremation burial in jar; Three flagons, two beakers, three bowls, three jars, eight dishes and one vessel. 2. Late 2nd to 3rd century AD; adult female (?) cremation burial in jar; One flagon, one beaker, one bowl, four jars, one dish and three vessels. 1903.1 60-80 AD; Urned cremation burial; One flagon, three bowls, two jars, two dishes, three cups and four vessels (incomplete). 1905. 1 AD 70-80; cremation burial in jar; One flagon, three beakers, one bowl, four jars, two dishes, one cup and three vessels; Burnt brooch (incomplete). 1905. 2 ; cremation burial in jar; Two beakers, three bowls, four jars, two dishes, one cup and two vessels (incomplete). 1. AD 50-75; cremation burial in jar; Two flagons, nineteen bowls, one jar, forty dishes, one lid and two vessels (many vessels underfired) (incomplete). 2. AD 50-75; cremation burial in jar; Two flagons, fourteen bowls, two jars, twenty six dishes, two lids and two vessels (many vessels underfired). 3. AD 100-120; probable cremation burial?; Two flagons, two jars (including one Drag. 67), two small Drag. 42 dishes, two larger Drag. 42 dishes and two Drag. 42 cups and four other vessels; Oyster and mussel shells. Arranged in crescent (incomplete). 4. Mid 1st century AD; cremation burial in wooden box; Two flagons, two bowls, two jars, three dishes, and two lids; Hobnailed footwear, brooch, bracelet. 5. AD 130-165; double cremation burial in three deposits (urned cremation burials in A and B); Pit A. Six dishes (including single Drag. 15/31 and Drag. 36), five bowls, one jar, two lids; Pit B. One flagon, one beaker, one jar, two bowls, six dishes (including single Drag. 18/31R); Main pit; Four dishes (including single Drag. 18/31 and Drag. 31R), two bowls, one jar and one lid. 2nd century AD; cremation burial; Two flagons, one jar, three bowls, two dishes, five lids. 1. 2nd century AD; adult male and female cremation burial in jar (damaged) ; A full grave inventory was not available. Ceramics included flagons, bowls, dishes and a set of cup forms imitating Drag. 27 in a very crumbly buff fabric; Ceramic lamp, ‘frying pan’, Bone pin, Worked bone inlay carved in diamond, triangle and crescent shapes with bone pegs; Burnt sheep, pig and bird bone. 10. 1st century AD; young adult cremation burial in jar (damaged); One jar, four bowls, one cup and four other vessels. 11. 1st century AD; young adult female (?) unurned cremation burial; Three beakers, two bowls, three dishes and three cups. 41. 1st century AD ; adult male unurned cremation burial (damaged); One flagon, three beakers, two bowls, six dishes and other sherds;

210

CONCORDANCE AND APPENDICES

Sparsholt, Hants. (SU 4239) (Collis 1977a)

20*

1*

Tilford, Surrey (SU 879442) (Millett 1974a)

22*

1*

Winchester Grange Road, Hants. (SU 473273) (Biddle 1967)

12*

5*

15

16

Winchester Highcliffe, Hants. (SU 494287) (Collis 1978: 103-5)

21*

1*

Winchester Victoria Road Street, Hants. (SU 479130) (Winchester Museums VR Tr. VIIIXV)

9

4

22

4

10

3

40*

3*

Winchester Milland, Hants. (SU 489288) (Collis 1978: 93-103)

Copper alloy strips and frags, iron razor/knife, one whetstone and a further whetstone possibly associated, pig's jaw. Late 1st century AD; probable cremation burial; Four Drag. 18 dishes, seven other dishes, three Drag. 35/36 and one Drag. 42 larger cups and five Drag. 27 smaller cups (incomplete). Late 1st century AD; probable cremation burial; Four flagons, one beaker, two bowls, three jars, eight dishes and four cups (incomplete). 1. Flavian; adult male (?) cremation burial in flagon; One samian dish (Drag. 36) and five imitation Drag. 18, six imitation cups (four Drag. 27, two Drag. 35/36); Unburnt bone frags of dog(?), copper alloy disc brooch, part of shale bracelet, five nails and iron and copper alloy box fittings (incomplete) 2. Flavian; unurned immature (?) female (?) cremation burial; Flagon, beaker, larger Drag. 18R dish, four Drag. 18 dishes, four larger and four smaller Drag. 27 cups; Glass jug, metal jug, shale tray, copper alloy spoon, two iron knives, copper alloy pin, copper alloy finger ring, iron finger ring, seal box lid, two iron styli, twelve white, four black and two blue glass gaming pieces, iron tool, bell, eight melon beads, fossil, five iron nails, pig r. hind limb, l. skull and maxilla, bird ulna. Late 1st century AD; adult (19-30) cremation burial in jar; One larger Drag. 18R dish, four Drag. 36 and three Drag. 35/36 small dishes / bowls, one Drag. 46 and eight Drag. 35/36 bowls, four Drag. 27 cups (incomplete). 438 later C1 AD; child / subadult cremation burial in beaker (Brandschüttungsgrab); Pyre goods burnt flagon, two Drag. 18 dishes, four Drag. 15/17 dishes and two Drag. 24/25 cups; melted unguent jar, molten frags, charcoal; Grave goods Nail fragments, copper alloy con of Claudius. 566 c. AD 70; adult cremation burial in jar; Four flagons, one beaker, three bowls, two jars, five Drag. 18 dishes, five Drag. 35 and Drag. 27 cups and two lids; Two dragonesque brooches, cockerel furculum, sheep/goat radius and ulna, two burnt bone discs. 622 LC1/EC2 AD; cremation burial in jar; One beaker, two dishes, seven cups including Drag. 27 imitations; Eight iron nails, iron plate and frag. 1st century AD; urned cremation burial; Thirteen jars, seven bowls, one larger Curle 11 bowl, five Drag. 15/17 and one Drag. 36 dishes, two other dishes including one in terra nigra, four Drag. 33 larger cups, four Drag. 35 and 1 Drag. 27 smaller cups/bowls, a lid and other vessel; One glass unguent bottle; pig left forelimb, whole chicken and woodcock on dish, possible whetstone (incomplete).

Appendix 6.5b Assemblages possibly of the East Hampshire Tradition Site / County / NGR / No. NAT Assemblage Reference vessels Alton Westbrooke House, U U 1st century AD; cremation burial. Hants. (SU 716390) (HCMS acc. no. 304.1/3650) Aylesfield, Hants. (SU 16 U 3rd century AD; cremation burial. 715432) (HCMS Acc. No. A1938.196) Basingstoke, Hants. (SU U U 1st? century AD; cremation burial. 646527) (Franks 1852: 9) Basingstoke Worting, U U 1st? century AD; cremation burial. Hants. (SU 6151) (Ellaway & Willis 1934: 88) Baughurst Hants. (SU U 4 2nd century AD (?);Cremation burial? Ceramic vessel; glass vessel; copper 5860) (Anon. 1993) alloy jug (face mask on handle) and pan (ram’s head handle); silver brooch. Worthy Down Camp, Kingsworthy (SU 476350)

U

4

Mid 1st century AD; Cremation burial? (no human bone recovered); Fragments of several samian dishes and bowls and other ceramics;

211

CONTEXTUAL ARCHAEOLOGY OF BURIAL PRACTICE. CASE STUDIES FROM ROMAN BRITAIN

(Burnham et al. 1998: 427; McCulloch pers. comm.) Winchester Nun’s Walk, Hants (SU 448305) (Collis 1978: 149-55) Winchester Winnall, Hants (SU 491301) (Collis 1978: 65-74)

Sherd from base of glass vessel; Colchester brooch; copper alloy spoon. 10

2

43

2

Flavian-Hadrianic; adult male (