The Fields of Britannia 0199645825, 9780199645824

Citation preview

T H E F I E L D S O F BR I T A N N I A

The Fields of Britannia Continuity and Change in the Late Roman and Early Medieval Landscape

STEPHEN RIPPON, CHRIS SMART, A N D BE N P E A R S

1

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries # Stephen Rippon, Chris Smart, and Ben Pears The moral rights of the authors have been asserted First Edition published in 2015 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2015931378 ISBN 978–0–19–964582–4 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

This book is dedicated to Mick Aston, a truly inspirational scholar and colleague

Preface and Acknowledgements In 1996, a young Stephen Rippon was interviewed for a lectureship at the Department of History and Archaeology at the University of Exeter. One future research project he talked about was the need to take a broad landscape-scale view of the transition from Roman to medieval Britain, and having got the job at Exeter, he set about exploring different aspects of that subject. An early paper discussed some initial ideas (Rippon 2000a), one of which—the need for more work on palaeoenvironmental sequences—was explored in South West England through a project titled ‘Landscapes in Transition’, generously funded by the Leverhulme Trust (award F/00144/D: Fyfe et al. 2003; 2004; Rippon et al. 2006). Several PhD students explored the Roman to medieval evidence from specific parts of the country, including Chris Smart (2008), who was awarded an AHRC studentship for his thesis ‘Continuity over crisis: the landscape of Southern Gloucestershire and South-East Somerset in the late Roman and early medieval periods’. There remained, however, the need for a larger-scale study with two particular characteristics: firstly, that covered the whole of Roman Britain but avoided simply repeating broadbrush narratives that obscured the possibilities of local/regional differences in the experience of native and immigrant communities in the fifth century; and secondly, a project that complimented the traditional approach towards this period—of writing a narrative—by using quantified data. And so was born the ‘Fields of Britannia Project’, generously supported by the Leverhulme Trust (award F/00 144/BI). In addition to the Principal Investigator, Stephen Rippon, it employed a team comprising Chris Smart and Ben Pears as the two Associate Research Fellows who worked on the archaeological and palaeoenvironmental material respectively, and Fiona Fleming as the Ph.D. student. Additional funding from the University of Exeter enabled research assistant Adam Wainwright to collect the palaeoeconomic data (animal bones and charred cereals). The fieldwork at Membury Court, in Devon, which contributed to Figure 7.5, was funded by Devon County Council and Natural England. Most of the illustrations in this book were produced by members of the project team, and in particular we thank Mike Rouillard, for whom working on this book was his last project in the Archaeology Department at the University of Exeter. In addition, we wish to thank the following for permission to reproduce their illustrations: Oliver Creighton (Figure 1.1, Caerwent), Northamptonshire County Council (Figure 1.2, Faxton), English Heritage (Figure 3.6, Knook Down East), Stewart Bryant (Figure 3.8, Cheshunt), Chris

viii

Preface and Acknowledgements

Gerrard and the late Mick Aston (Figures 3.9 and 3.10), Chris Green and the University of Oxford’s ‘English Landscape and Identities’ project (Figure 3.15, AIP analysis), Andy Mudd and Cotswold Archaeology (Figure 6.4, Hinckley Point photograph), Roger Leech (Figure 6.5B, Podimore photograph), and Cornwall County Council (Front Cover and Figure 7.8, Bosigran). Phil Pearce was the pilot during the flight from which photos in Figure 3.17 were taken. A key feature of the Fields of Britannia Project was the collection and analysis of unpublished grey literature, and we must thank the numerous archaeological units, Historic Environment Records (HERs), and specialists who helped with this. In particular, we are grateful to Context One Archaeological Services Ltd (the excavations at Cambria Farm, in Ruishton, near Taunton, in Somerset), John Blair (for information on the Shakenoak radiocarbon dates), Alan Hardy and Andy Mudd of Cotswold Archaeology (for information on Hayes Farm in Devon and Hinckley Point in Somerset), and Jim Stephenson of UCL (for information on Sadler’s Farm, in Essex). For their help with the palaeoenvironmental analysis we thank Mike Allen, Anthony Brown, Wendy Carruthers, Ken Crowe, Michael Grant, James Greig, Sarah Howard, Tim Mighall, Liz Pearson, Colin Pendleton, Nathan Pittam, Liz Propescue, Sue Stallibrass, Vanessa Straker, Robert Van de Noort, and Michael Fulford and Alex Smith of the University of Reading’s Roman Rural Settlement Project. Finally, we wish to thank the many friends and colleagues who have discussed various aspects of this project’s results at its Advisory Board and various conferences, symposia and research seminars, including Grenville Astill, Michael Fulford, Nick Higham, Neil Holbrook, and Bob Silvester.

Contents List of Figures List of Tables List of Appendices List of Abbreviations Note on Period and Other Terminology Presentational Conventions

1. Fields of Britannia: A Roman Legacy in the British Countryside

xi xvii xxi xxiii xxv xxv 1

2. A Regional Approach to Studying Landscape

19

3. A Landscape Approach to the Roman–Medieval Transition

57

4. The South East

124

5. East Anglia

169

6. The Central Zone

182

7. The South West

221

8. The Western Lowlands

247

9. The North East Lowlands

267

10. The Northern Uplands

280

11. Upland and Lowland Wales

295

12. Discussion and Conclusions

305

Bibliography Index

343 439

List of Figures Cover image: Aerial photograph of the heavily lynchetted field system at Bosigran, overlooking Porthmeor Cove in Zennor, Cornwall, looking east (# Historic Environment, Cornwall Council; F78-070 7 August 2007). Chapter 1 Fig. 1.1. The southern defences of the civitas capital of Venta Silurum (Caerwent), Monmouthshire.

2

Fig. 1.2. The unrelated cropmarks of an Iron Age/Romano-British landscape and medieval ridge and furrow within a former open field at Faxton in Northamptonshire.

8

Fig. 1.3. The late Roman and fifth-century occupation at Orton Hall Farm in Cambridgeshire, overlain unconformably by later medieval ridge and furrow.

14

Chapter 2 Fig. 2.1. Traditional binary divisions within Roman Britain, and its civitates.

20

Fig. 2.2. The distribution of Romano-British villas.

21

Fig. 2.3. The hierarchy of towns across Roman Britain.

23

Fig. 2.4. The distribution of Romano-Celtic temples, pottery production sites, mosaic pavements, and mosaic schools.

25

Fig. 2.5. Measures of prosperity in Roman Britain.

26

Fig. 2.6. The late Roman and early medieval phases at the Barnsley Park villa, Gloucestershire.

29

Fig. 2.7. The change in settlement morphology from rectilinear to curvilinear in the fifth century in western Britain at Poundbury, near Dorchester, Dorset, and Hayes Farm in Clyst Honiton, Devon.

32

Fig. 2.8. The distribution of early Anglo-Saxon burials.

33

Fig. 2.9. The results of fieldwalking in Raunds, Northamptonshire.

38

Fig. 2.10. Past mappings of medieval landscape character.

43

Fig. 2.11. The regions of Roman and early medieval Britain used within the Fields of Britannia Project.

48

Fig. 2.12. Simplified solid geology.

49

Fig. 2.13. Relief.

50

xii

List of Figures

Chapter 3 Fig. 3.1. The distribution of all pollen sites used within the Fields of Britannia Project.

59

Fig. 3.2. The distribution of pollen sites for the four periods of analysis in the Fields of Britannia Project.

60

Fig. 3.3. The changing proportions of pollen derived from the four major land-use categories—woodland, arable, improved pasture, and unimproved pasture—during the Roman and medieval periods.

64

Fig. 3.4. The changing proportions of woodland, arable; improved pasture, and unimproved pasture pollen in each region across the four periods.

66

Fig. 3.5. The study area for the analysis of animal bones and charred cereals.

68

Fig. 3.6. The exceptionally well preserved earthworks of the Romano-British settlement and field system at Knook Down East on the Salisbury Plain Training Area, Wiltshire.

86

Fig. 3.7. The late Roman and medieval features at Heathrow Terminal Five, Middlesex.

89

Fig. 3.8. The coaxial pattern preserved within today’s historic landscape at Cheshunt, Hertfordshire.

91

Fig. 3.9. Excavation of the southern boundary of Buddle Furlong in Shapwick, Somerset.

93

Fig. 3.10. Plan and section of the trench across the boundary of Buddle Furlong in Shapwick, Somerset.

94

Fig. 3.11. Excavations at Sherborne House in Lechlade, Gloucestershire.

96

Fig. 3.12. Schematic model showing the three relationships—aligned, oriented, and unrelated—between excavated Romano-British field systems and the historic landscape.

101

Fig. 3.13. Data collection for relevant excavations as part of the Fields of Britannia Project.

102

Fig. 3.14. The distribution of excavations used in the quantified analysis.

103

Fig. 3.15. Linear trend surfaces for commercial archaeological fieldwork that took place between 1990 and 2010 in England, based upon data collected by the Archaeological Investigations Project (AIP).

104

Fig. 3.16. Examples of the three main types of historic landscape character: former Open Field (enclosed by agreement, so retaining the shape of the strips and furlongs); Closes; and Late Enclosure.

106

Fig. 3.17. Aerial photographs of the historic landscape character types used in the Fields of Britannia Project.

108

Fig. 3.18. Excavations at Bishop Burton College in the East Riding of Yorkshire.

110

List of Figures

xiii

Fig. 3.19. The late Roman and fifth- to sixth-century phases at Barton Court Farm, Oxfordshire.

115

Fig. 3.20. A series of schematic models showing how the landscape may have developed in a series of regions across an east–west transect from the South West to East Anglia.

118

Chapter 4 Fig. 4.1. The pays in the South East region.

125

Fig. 4.2. Places referred to in the text for the South East region.

126

Fig. 4.3. The coaxial landscapes of southern Essex.

144

Fig. 4.4. The excavations at Dagenham Heathway in Dagenham, Essex.

147

Fig. 4.5. The excavations at Hunts Hill Farm in Upminster, Essex.

148

Fig. 4.6. The excavations at the Orsett Cock in Orsett, Essex.

150

Fig. 4.7. The excavations at the Stifford Clays in Stifford, Essex.

151

Fig. 4.8. The excavations at the High House in West Thurrock, Essex.

152

Fig. 4.9. The excavations at Ship Lane in Aveley, Essex.

155

Fig. 4.10. A newly discovered possible Roman road that runs unconformably across the coaxial landscape on the London Clay of southern Essex.

156

Fig. 4.11. The Hawthorn Path in Basildon, Essex, which follows the line of the possible Roman road.

157

Fig. 4.12. The excavations at Monument Borrow Pit in North Benfleet, Essex.

158

Fig. 4.13. The excavations at Washlands in Nevendon, Essex.

159

Fig. 4.14. Schematic model for how southern Essex coaxial landscapes could have developed.

162

Chapter 5 Fig. 5.1. The pays in the East Anglian region.

170

Fig. 5.2. Places referred to in the text for the East Anglian region.

170

Fig. 5.3. The excavations at Grange Farm in Snetterton, Norfolk.

178

Fig. 5.4. The excavations at Yarmouth Road in Broome, Norfolk.

180

Chapter 6 Fig. 6.1. The pays in the Central Zone region.

184

Fig. 6.2. Places referred to in the text for Central Zone region.

186

Fig. 6.3. The excavations at Cambria Farm in Ruishton near Taunton, Somerset.

200

Fig. 6.4. The excavations at Hinckley Point in Stogursey, Somerset.

201

xiv

List of Figures

Fig. 6.5. The cropmark complex, geophysical survey, and excavations at Podimore and Yeovilton, Somerset.

204

Fig. 6.6. The excavations at Gloucester Business Park on the Old Brockworth Airfield near Gloucester, Gloucestershire. 208 Fig. 6.7. The excavations at Weedon Hill in Aylesbury, Buckinghamshire.

212

Fig. 6.8. The excavations at Gossway Fields in Kirtlington, Oxfordshire.

213

Fig. 6.9. The cropmark complex at Dunstan’s Clump in Babworth, Nottinghamshire.

215

Fig. 6.10. The cropmark complex and small-scale excavations at Barnby Moor, Nottinghamshire.

216

Fig. 6.11. A wider view of the cropmark complex at Barnby Moor, Nottinghamshire.

217

Fig. 6.12. The small-scale evaluation excavations at Malton and Norton, Yorkshire.

218

Chapter 7 Fig. 7.1. The pays in the South West region.

222

Fig. 7.2. Places referred to in the text for the South West region.

223

Fig. 7.3. The excavations at Hill Barton in Clyst St Mary, Devon.

235

Fig. 7.4. The excavations around Hayes Farm in Clyst Honiton, Devon.

236

Fig. 7.5. The Romano-British enclosure complex at Membury Court in Membury, Devon.

239

Fig. 7.6. The geophysical survey and excavations adjacent to the cropmarks of the late Iron Age and Romano-British Penhale Round and its associated field system in Fraddon, Cornwall.

241

Fig. 7.7. Excavations at Scarcewater in Pennance, Cornwall.

242

Fig. 7.8. Aerial photograph of the courtyard house settlement and heavily lynchetted field system at Bosigran in Zennor, Cornwall.

244

Fig. 7.9. Ordnance Survey First Edn Six Inch map of the Bosigran field system in Zennor, Cornwall.

245

Chapter 8 Fig. 8.1. The pays in the Western Lowlands region.

248

Fig. 8.2. Places referred to in the text for the Western Lowlands region.

249

Fig. 8.3. The cropmark complex and excavations at Saxon’s Lode Farm in Ripple, Worcestershire.

261

Fig. 8.4. The cropmark complex and excavations at Linacres Farm in Claines, Worcestershire.

263

Fig. 8.5. Geophysical survey and excavations at Strensham, Worcestershire.

264

List of Figures

xv

Chapter 9 Fig. 9.1.

The pays in the North East region.

268

Fig. 9.2.

Places referred to in the text for the North East region.

269

Fig. 9.3.

Excavations at Hollow Banks in Scorton, Brompton-on-Swale, Yorkshire.

278

Chapter 10 Fig. 10.1. The pays in the Northern Uplands region.

281

Fig. 10.2. Places referred to in the text for the Northern Uplands region.

282

Fig. 10.3. Westwards view across the Tees Valley towards the Pennines.

291

Fig. 10.4. Royston Grange, Derbyshire.

293

Chapter 11 Fig. 11.1. The pays in the Upland and Lowland Wales regions.

296

Fig. 11.2. Places referred to in the text for the Upland and Lowland Wales regions.

297

Chapter 12 Fig. 12.1. Change in woodland pollen as percentage of TLP from the Roman to the early medieval period. 310 Fig. 12.2. Change in arable pollen as percentage of TLP from the Roman to the early medieval period.

310

Fig. 12.3. Change in improved pasture pollen as percentage of TLP from the Roman to the early medieval period. 311 Fig. 12.4. Change in unimproved pollen as percentage of TLP from the Roman to the early medieval period.

311

Fig. 12.5. Example of a laid hedge on the edge of Molland Common, Devon. 313 Fig. 12.6. The heavily lynchetted fields at Rosemergy, near the ruined Carn Galver tin mine in West Penwith, Cornwall.

321

Fig. 12.7. The extent to which excavated Romano-British field boundaries across the different regions share the same orientation and/or alignment as historic landscapes characterized by (A) former medieval Open Fields or Closes; and (B) former medieval Open Fields, Closes, and Indeterminate landscapes.

324

Fig. 12.8. An evaluation trench at Seighford, Staffordshire.

333

Fig. 12.9. The evocative ruins of Burgh Castle, Suffolk, which encapsulate the traditional view that Roman Britain ended and has little functional relationship to the landscape of today.

334

Fig. 12.10. A series of schematic models showing how the landscape may have developed in a series of regions across an east–west transect from the South West to East Anglia.

341

List of Tables Table 1.1. Average sizes of Grubenhäuser on selected early Anglo-Saxon settlements

15

Table 3.1. The proportions of pollen derived from the four major land-use types in the Roman and early medieval periods for each of the Fields of Britannia regions

62

Table 3.2. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for upland and lowland regions

63

Table 3.3. A statistical analysis of the percentages of cattle, sheep/goat, and pig from sites of different status in Roman Britain (King 1999, tab. 3)

75

Table 3.4. A comparison of the major domesticated animals on rural settlements, villas, and towns dating to the early and late Roman periods and fifth to seventh/eighth centuries

76

Table 3.5. A comparison of charred cereal grains from corn drier and non-corn drier contexts on rural settlements

81

Table 3.6. Changing crop regimes on rural settlements over the early and late Roman periods and fifth to seventh/eighth centuries using non-corn drier contexts (for sites/sources see Appendix IV)

82

Table 3.7. Summary of the relationships between excavated Romano-British field systems and the historic landscape for each of the Fields of Britannia regions

111

Table 4.1. Key character-defining features of pays within the South East region 127 Table 4.2. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the South East region 130 Table 4.3. The relationships between excavated Romano-British field systems and the historic landscape for each of the pays within the South East region

142

Table 5.1. Key character-defining features of pays within the East Anglia region 171 Table 5.2. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the East Anglia region

172

Table 5.3. The relationships between excavated Romano-British field systems and the historic landscape for each of the pays within the East Anglia region

176

xviii

List of Tables

Table 6.1.

Key character-defining features of pays within the Central Zone region

188

Table 6.2.

The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the Central Zone region

191

Table 6.3.

The relationships between excavated Romano-British field systems and the historic landscape for each of the pays within the Central Zone region

197

Table 7.1.

Key character-defining features of pays within the South West region

224

Table 7.2.

The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the South West region

225

Table 7.3.

The relationships between excavated Romano-British field systems and the historic landscape for each of the pays within the South West region

233

Table 8.1.

Key character-defining features of pays within the Western Lowlands region

250

Table 8.2.

The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the Western Lowlands region

252

Table 8.3.

The relationships between excavated Romano-British field systems and the historic landscape for each of the pays within the Western Lowlands region

259

Table 9.1.

Key character-defining features of pays within the North East Lowlands region

270

Table 9.2.

The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the North East Lowland region

271

The relationships between excavated Romano-British field systems and the historic landscape for each of the pays within the North East Lowlands region

274

Table 9.3.

Table 10.1. Key character-defining features of pays within the Northern Uplands region

283

Table 10.2. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the Northern Uplands region

285

List of Tables

xix

Table 11.1. Key character-defining features of pays within the Upland and Lowland Wales regions

298

Table 11.2. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, six to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the Lowland Wales region

299

Table 11.3. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for pays within the Upland Wales region

300

Table 12.1. The relationship between excavated Romano-British field systems and historic landscapes characterized by medieval Closes and former Open Fields in pays with different modern classifications of agricultural land quality

327

Table 12.2. The extent to which excavated Romano-British field boundaries share the same orientation and/or alignment as historic landscapes characterized by former medieval Open Fields or Closes in landscapes of predominant surface geology types

327

List of Appendices The following are available online through Open Research Exeter, at: . They are also available on the Oxford University Press website at . I List of Pollen Sequences Used in the Statistics (used in Tables 3.1, 3.2, 4.2, etc.) II Pollen Species by Land-Use Category III Animal Bone Data (used in Tables 3.3 and 3.4) IV Charred Cereal Data (used in Tables 3.5 and 3.5) V List of Excavations used in the Analysis (used in Tables 3.7, 4.3, etc.)

List of Abbreviations ADS

Archaeology Data Service

AIP

Archaeological Investigations Project

AOD

Above Ordnance Datum (i.e. above mean sea level)

BAR

British Archaeological Reports

BUFAU

Birmingham University Field Archaeology Unit

CBA

Council for British Archaeology

COMV

currently occupied medieval village

ECCFAG

Essex County County Field Archaeology Group

ECCFAU

Essex County County Field Archaeology Unit

EngLAID

English Landscape and Identities Project

HER

Historic Environment Record

HLC

Historic Landscape Characterization

MAFF

Ministry of Agriculture, Fisheries and Food

MNI

minimum number of individuals

NMP

National Mapping Project

NISP

number of identified specimens

OS 1st Edn Ordnance Survey First Edn Six Inch to the Mile maps drafted in the maps mid- to late nineteenth century SCCAS

Suffolk County Council Archaeological Service

S/G

sheep/goat

TLP

Total Land Pollen

TVAS

Thames Valley Archaeological Services

ULAS

University of Leicester Archaeological Services

Note on Period and Other Terminology The term ‘Saxon’ is not used in this study as a period term because the ethnic tag is inappropriate for those areas that did not see significant Anglo-Saxon immigration: where others have used terms such as ‘early Saxon period’ it is placed in inverted comas. The term Anglo-Saxon is used in this study to refer to the Germanic migrations into Britain and the resulting distinctive building tradition (Grubenhäuser), furnish burials, and other material culture. The term early medieval is used for the period between Britain ceasing to be part of the Roman Empire and the Norman Conquest, while later medieval is used for the period between the Norman Conquest and the end of the fifteenth century. The historic landscape is the term used for the pattern of fields, roads, settlements, and land-uses recorded on the nineteenth-century OS 1st Edn maps.

Presentational Conventions Site names in bold are illustrated.

1 Fields of Britannia A Roman Legacy in the British Countryside

BRITAIN’ S GREEN AND P LEASANT L AND In asking about ‘Our debt to Rome’, O. G. S. Crawford (1928, 173) asked: What do we inherit from our Roman Conquerors? To this question some reply, ‘little or nothing’, and some ‘the seeds of culture and religion’ . . . it is difficult to reach a decision because the decisive period, between 400 and 600 A.D. is one of the darkest in our history.

The traditional view of Britain in the fifth century—when it ceased to be part of the Roman Empire—is one of catastrophe and discontinuity: although it is well known that some of our major roads are Roman in origin, and that many of our cities lie on top of Roman predecessors, most archaeologists and historians have come to the conclusion that our countryside is essentially medieval in origin. Despite the majority of the British population now living in towns and cities there is still a great affection for this countryside and, in the summer of 2012, millions of people across the World were absorbed by Danny Boyle’s opening ceremony for the London Olympics that began with a portrayal of our rural landscape and culture, before showing the transformation of William Blake’s ‘green and pleasant Land’ by the ‘dark satanic Mills’ of the Industrial Revolution. One of the most distinctive features of the British countryside—as depicted in the Olympics—is its intricate pattern of small fields variously enclosed by hedges, banks, ditches, and walls, and it is often assumed that these fields were created when rapacious eighteenth-century landowners obtained an Act of Parliament to enclose open fields. For a broad swathe of countryside in central England this is generally true, but in the South East, the South West, and the West of England open fields never dominated the landscape, and the origins of this regional variation in the character of our ‘green and pleasant Land’ has received much scholarly attention. In the first half of the twentieth century, it

2

The Fields of Britannia

was assumed that villages and open fields—the so-called ‘champion countryside’ of England’s Central Zone—were introduced during the fifth and sixth centuries by the Anglo-Saxon immigrants that early medieval writers, such as Bede, tell us came to Britain from mainland Europe. In recent decades, however, both archaeologists and historians have agreed that champion countryside was the product of a transformation of the landscape around the eighth to twelfth centuries (that clearly post-dates the period of the Anglo-Saxon migrations by several centuries), which leads us to the central question to be addressed in this book: what happened to the landscape of Britannia between the period when it ceased to be part of the Roman Empire and the emergence of villages and open fields in some areas several centuries later? The evocative ruins of Roman towns such as Caerwent (Figure 1.1), Silchester, and Wroxeter, and a countryside that appears to be littered with longabandoned villas—such as Chedworth in Gloucestershire, whose sense of remoteness is increased by it being surrounded by woodland—reinforce the impression that the landscape of Britannia is largely unrelated to the countryside of today. The apparently contradictory evidence of some modern towns lying on top of Roman predecessors has generally been resolved through excavations showing that their urban character failed to survive much into the fifth century, with any occupation being of an essentially

Fig. 1.1. The southern defences of the civitas capital of Venta Silurum (Caerwent) in Monmouthshire. Evocative ruins such as these reflect a popular perception that Roman Britain ‘ended’ and has little functional relationship to the countryside of today (# Oliver Creighton).

Fields of Britannia: A Roman Legacy in the British Countryside

3

rural character: there may have been some continuity in these sites as locations for settlements of a largely rural character, but not in their function as administrative and market centres. A common view, therefore, is that Roman Britain ended quite suddenly and completely, and what came after was something very different. Costen (1992, 54), for example, suggests that the fifth century was one of: complete dislocation of the economy at all levels of society. No one was unaffected. Aristocrats in villas and their tenants and servants on their estates, merchants and the artisans who made goods for sale in the towns were all hit by the disappearance of the settled economy of the fourth century and the withdrawal of Roman administration.

Blair (1994, 3) paints an equally stark picture: The most obvious fact of the early fifth century is that towns and villas became irrelevancies, so that to look for ‘continuity’ in the modes of life which they served is to chase a will-o’-the-wisp. Complex government, bureaucracy, coinage and long distance trade simply could not survive the secession of Britain from the Empire.

Few would doubt that the fifth century saw a sharp discontinuity in the political history, the market-based economy, and the prosperity and power of the urban and villa-owning elite, but what about the majority of the population who farmed the countryside?

P E R S P E C T I V E S ON AN AG E OF TRANS I TI ON

The End of Roman Britain The perception of the fifth century as one of profound change is now deeply engrained in scholarship, with Romanists seeing a relatively sudden and complete end to the culture that they study: While parts of the east were rich, indeed wealthy, in the later Roman period, much of the region, materially at least, seems not to have been and at the end of the Roman era vanished almost with ease from the archaeological record (Going 1996, 104).

This view that Roman Britain ended is also reflected in Casey’s (1979) The End of Roman Britain, Esmonde Cleary’s (1989) The Ending of Roman Britain, and Faulkner’s (2000) The Decline and Fall of Roman Britain. That the fifth to seventh centuries were something different to what went before is also reflected in the scope of our learned societies (the Society for the Promotion of

4

The Fields of Britannia

Roman Studies and the Society for Medieval Archaeology) and their journals (e.g. Britannia; The Journal of Roman Studies, and Medieval Archaeology): the archaeology of Roman Britain is seen as quite separate from that of medieval Britain. Indeed, the distinctive archaeology of the fifth to seventh centuries— most notably the artefact-rich cemeteries and settlements characterized by Grubenhäuser—has led to a division between early medievalists who have focused on this ‘early Saxon period’,1 and those who have concentrated on the ‘middle and late Saxon’ periods (the late seventh to mid-ninth, and mid-ninth to mid-eleventh centuries, respectively) during which villages, open fields, the Church, urbanism, and many other aspects of the later medieval (i.e. post Norman Conquest) landscape appear to have originated. There has also been a division between scholars who have tended to focus on the south and east of Britain, where there were clearly profound cultural changes brought about by the Anglo-Saxon migrations (e.g Welch 1992; Lucy 2000; Tipper 2004), and the north and west, where this Anglo-Saxon influence came far later and was the result of conquest and political assimilation as opposed to folk migration (e.g. Rahtz 1983; Alcock 2003). The traditional view—based largely upon the extremely problematic documentary sources for this period—was that the remnants of what was only ever a relatively small Romano-British population were forced west by Germanic hoards, who therefore settled a largely depopulated and well-wooded landscape, with the result that the majority population of seventh-century England were of Anglo-Saxon descent, any of the remaining native Britons having been expelled or enslaved (e.g. Hoskins 1955, 38–60; Morris 1977). Early archaeological studies simply fitted the material remains of this period into that historical framework (Alcock 1971; Myres 1986). In the 1970s, however, this culture-historical approach, whereby changes in the archaeological record were seen as having been brought about by migrations, started to give way to post-processual ideas that cultural change results from human adaptation to a variety of environmental, social, economic, and technological factors (Chapman and Hamerow 1997; Hamerow 1997). Archaeological survey and excavation was also suggesting a far larger late Romano-British population, of perhaps 4 to 6 million (e.g. Salway 1981, 542–5) compared to estimates of c.0.5 million proposed in the early twentieth century (Collingwood 1929, 261), although most current estimates are around 3.7 to 4 million (Fowler 1983, 8; Millett 1990, tab. 8.5; Higham 1992, 20). In contrast, Williamson (2013, 12–13) has argued that ‘it is unlikely that there were many districts in which Romano-British population exceeded that at

The term ‘Saxon period’ is not used in this study, as the ethnic tag is inappropriate for those areas that did not see substantial Anglo-Saxon immigration: where others have used terms such as ‘early Saxon period’ it is placed in inverted commas. The term Anglo-Saxon is used in this study to refer to the Germanic migrations into Britain and the distinctive building tradition (Grubenhäuser), furnished burials, and other material culture that resulted. 1

Fields of Britannia: A Roman Legacy in the British Countryside

5

the time of Domesday Book’, because the density of Romano-British settlements revealed in fieldwalking surveys (assuming a population of 20 people per site) suggests a population that was less than Domesday records for those areas (allowing for a multiplier of 5.5). This argument can be challenged, however, as Romano-British settlements may well have contained a far higher population, and there are many examples of Romano-British settlements and field systems in areas that were unenclosed pasture in the medieval period (e.g. Chalton in Hampshire: Cunliffe 1973, cf. figs 5 and 7; Fenland: Phillips 1970; Palmer 1996; Rippon 2000b; Salisbury Plain: McOmish et al. 2002; Fulford et al. 2006b; Bradley and Fulford 2008). Another blow to the traditional view of a relatively low Romano-British population that fell even further in the post-Roman period was that palaeoenvironmental analysis had failed to find evidence for extensive woodland in the late Roman period or woodland regeneration in the early medieval period (Bell 1989; K. Dark 2000), and three seminal studies of the archaeology of this period—Arnold (1988), Hodges (1989), and Higham (1992)—all came to similar conclusions: that a substantial native RomanoBritish population survived into the early medieval period, and that there was only small-scale Anglo-Saxon immigration, which achieved social and political supremacy through elite dominance, not weight of numbers. A central argument of these revisionist studies was that Britain remained Roman into the fifth century: while the last quarter of the fourth century saw a marked recession in Roman Britain, ‘she was still a diocese of the empire with the administration, the economy and the society fashioned by three hundred years of Roman rule still firmly in place’ (Esmonde Cleary 1989, 121). White (2007, 195–214) has even argued that, despite the political upheavals, the province of Britannia Prima and its social elite, at least, remained essentially Romanized until perhaps as late as the seventh century. Others, however, have disagreed, suggesting that Roman Britain collapsed in the fourth century, such as Faulkner (2000, 143), who argued that ‘in the mid-late fourth century, agriculture was depressed and landlord surpluses were being squeezed hard. Moreover, this was no short-term crisis due to crop blight, transport failure or disrupted markets. It was a long-term structural collapse from which no escape was possible: full-blown systems failure.’ More recently, in contrast, Fulford et al. 2006a, 280) have observed that ‘we can see throughout the recent literature on late Roman Britain an urgency to end urban life early in the fifth century as a correlate of the demise of the most conspicuous and abundant forms of Roman material culture’, but they counter this with an argument that ‘it is hard to accept a rapid end to the use of surviving Roman material in the early fifth century’. At Silchester, for example, long seen as a classic example of an abandoned Roman town, there may have been occupation long into the fifth, and even sixth or seventh centuries (Fulford et al. 2006a; Fulford 2012): the debate about what happened at the end of Roman Britain clearly continues.

6

The Fields of Britannia

All of these studies focus upon cultural explanations for the decline of Roman Britain, although M. E. Jones (1996, 223, 237) provides an alternative model that explores environmental issues, echoing Postan’s (1972) approach to later medieval England. Jones argued that falling temperatures and increased rainfall, past over-exploitation of soils, a declining demand for food as the market economy collapsed, and an outbreak of disease led to a ‘severe crisis’ in the late Roman period, during which fields were abandoned. It is unusual today to see environmental constraints on human behaviour being expressed so clearly, not least because M. E. Jones’s (1996, figs 8 and 9) own maps show that lowland Britain cannot be classed as marginal in terms of its climate, although a number of other landscape studies have started to argue that the role of the natural environment in shaping human behaviour has been erroneously downplayed in recent years. Prominent amongst these studies is Williamson’s (2003; 2013; Williamson et al. 2013) exploration of the potential role of soil character in explaining why only some areas saw the development of open fields, although suggesting that the inherent properties of geologies, soils, and topographies made them particularly suited to certain types of agricultural practice is very different to arguing that slight changes in rainfall and temperature will have had a material effect on farming in temperate lowland Britain. Overall, while poor weather may have combined with changing socio-economic circumstances in the fifth century to make farming more difficult, Jones’s argument that environmental deterioration caused the decline of Roman Britain is not convincing.

A ‘Late Antique’ Landscape? Native Britons and Anglo-Saxons in Lowland Britain As Dark (2004, 279) has observed, ‘The development of the British landscape during the period c.AD 300–700 has often been depicted in terms of sharp discontinuities between the “Romano-British landscape” of partly stone-built villas, temples and “native settlements”, and the “Anglo-Saxon landscape” of timber-built small hamlets and farms.’ This is what a simple reading of the archaeological record tells us: Roman Britain ended, and from the chaos that ensued there emerged an ‘Anglo-Saxon’ landscape. The established view of the end of Roman Britain is that there is very little evidence for contact between the native British and immigrant Anglo-Saxon populations (e.g. Esmonde Cleary 1989, 131, 140, 153, 161, 197). There is, however, a paradox in that these same scholars who argue for a relatively abrupt collapse of Roman Britain also note that there appears to have been less change in the countryside and that ‘there is little evidence that the fifth century saw a significant decline in the global amount of land under cultivation in Britain’ (Esmonde Cleary

Fields of Britannia: A Roman Legacy in the British Countryside

7

1989, 158). If the amount of land under cultivation did indeed not decline significantly then at the level of rural peasant society there should have been a strong degree of continuity, yet the compartmentalization of research into the ‘Roman’, ‘Saxon’, and ‘medieval’ periods perpetuates the impression of discontinuities: Romanists see the cessation of an archaeologically very visible culture and so end their story, while Anglo-Saxonists see a new and distinctive suite of material culture that had no precedents in Roman Britain and so start their story in the fifth century. The extensive remodelling of the landscape several centuries later, that saw the creation of villages and open fields, is another context for discontinuity (Figures 1.2 and 1.3). In contrast to this fragmentary approach that traditional periodization brings about, this book will focus very explicitly on the transition from the Roman to the medieval periods. This takes us on to the debate about what happened to the native RomanoBritish population in the fifth century. Archaeologists (e.g. Arnold 1988; Hodges 1989; Henig 2002) and historians (e.g. Higham 1992; Blair 2013b, 2) have mostly rejected the traditional ideas that Anglo-Saxon immigrants wholly replaced the British population in lowland Britain, although few would go as far as the hyper-continuity hypotheses of K. Dark (2000) and particularly Pryor (2004, 96, 214), who argues that fourth- to sixth-century eastern England had ‘an essentially stable rural population existing in a political context that was changing quite rapidly’, with ‘no convincing archaeological evidence for “Dark Age” chaos, disruption or turmoil’. Pryor (2004, 214) even goes as far as to suggest that ‘Anglo-Saxon mass migrations into Britain never happened’. In sharp contrast, some linguists continue to argue that the native British population was largely displaced by Anglo-Saxon immigrants (e.g. Gelling 1993; Coates 2007), although the documents that survive from this period—written from the perspective of the ascendant Anglo-Saxon population, it must be remembered—simply show that the surviving British population was subservient to their Anglo-Saxon masters (e.g. the Law Code of King Ine of Wessex dated c.688–93: Grimmer 2007). That the starting point of a major conference in 2004 which examined Britons in Anglo-Saxon England started with the question of ‘whether or not there were many Britons within Anglo-Saxon England’ (Higham 2007b, 1), and that many of the papers— particularly by linguists—reject the notion of any significant British survival in lowland Britain, shows that many still stubbornly adhere to the traditional view. It is against this background of very divergent views that Nick Higham (2007b, 15) perceptively concluded ‘A significant British presence does, therefore, seem discernible right across the Anglo-Saxon period alongside evidence for large-scale discontinuity between Anglo-Saxon England and sub-Roman Britain, to the confusion of us all.’ The intention of this study, therefore, is to try and shed new light on this confusing period through an explicit focus on what happened to the rural landscape of Roman Britain.

8

The Fields of Britannia

Fig. 1.2. The unrelated cropmarks of an Iron Age/Romano-British landscape and medieval ridge and furrow within a former open field at Faxton in Northamptonshire. There is a clear discontinuity in this landscape’s history, but rather than at the end of the Roman period it may have been in the late first millennium AD when the open fields were laid out (NCC photograph SP7874/018 1 August 1986; # Northamptonshire County Council Historic Environment Record).

While views still differ as to whether there was a mass-folk migration from the Continent involving hundreds of thousands of people, or simply a political take-over by a small, male military elite, what is clear is that the late RomanoBritish population was so large that it cannot have been ethnically cleansed. Millett (1990, tab. 8.5) estimates that the population of late Roman Britain was around 3.7 million, of which the rural population amounted to c.3.3 million, the urban population c.240,000, and the army and its dependents c.125,000 (and see Higham 1992, 20; M. E. Jones 1996, 208). In contrast to the artefact-rich

Fields of Britannia: A Roman Legacy in the British Countryside

9

Anglo-Saxon communities, however, the native population in the fifth to sixth centuries appears virtually invisible: Roman pottery, coinage, and building techniques all appear to have gone out of use, and we are left with the impression that sub-Roman Britain was a period when a few dazed native Britons staggered around the ruins of once magnificent buildings in a virtually deserted landscape, while Anglo-Saxon immigrants took over their lands at will. A more positive view of this period, however, is presented by those who advocate a period of ‘Late Antiquity’ within which there was a strong degree of continuity from the preceding Roman period, and interaction between the native (often Christian) and immigrant (pagan) populations (e.g. K. Dark 2000; 2004; Collins and Gerrard 2004). There is, however, an almost total lack of rural settlements in eastern Britain that have been attributed to this native population, with all of Snyder’s (1996) ‘sub-Roman’ occupation in this area being within Roman towns or coastal fortresses. One possibility is that some Romano-British settlements have been continuously occupied ever since, and so lie buried under modern villages, reflected, for example, by proximity of some parish churches to Roman villas (Rodwell and Rodwell 1977; Leech 1982; Bell 1998; 2005). One example is Rivenhall, in Essex, where Rodwell and Rodwell (1986) found evidence for a phase of timber buildings associated with fifth-century pottery adjacent to a Roman villa and the overlying parish church, suggesting near-continuous, if not continuous occupation on the site.2 Rivenhall has become a much-cited ‘classic’ example of continuity, although it also illustrates the problem in studying the archaeology of this period that, due to the use of timber buildings, and the scarcity of material culture, is both ephemeral and poorly dated. Indeed, Millett’s (1987) review of the original report, and Clarke’s (2004) subsequent excavations on the site, have called into question some key elements of the Rodwells’ sequence. The evidence for such timber buildings being constructed during the latest phases of occupation within Roman settlements was traditionally seen as ‘squatter occupation’ (Brean Down in Somerset: ApSimon 1965; Silchester in Hampshire: Frend 1992, 126), although it is now thought to represent not the re-use of abandoned sites but the final phase of occupation by communities for whom the Roman aesthetic was becoming unsustainable or ‘socially irrelevant’ (Hamerow 2012, 13; and see K. Dark 1992; 2004; Rogers 2011, 158). The potential sophistication of timber buildings in this period have been postulated through Barker’s meticulous excavations at Wroxeter (Barker et al. 1997; but see Fulford 2002 and Lane 2014), and the sequence at the

2

Other late Roman sites with a stratigraphically late phase of timber buildings include Barnsley Park in Gloucestershire (Webster 1982), Brixworth in Northamptonshire (Brown and Foard 1998, 73), Frocester in Gloucestershire (Price 2000a, b), Gadebridge Park in Hertfordshire (Neal 1974), Latimer in Buckinghamshire (Branigan 1971), Orton Hall Farm in Cambridgeshire (Mackreth 1996), and Shakenoak in Oxfordshire (Brodribb et al. 2005).

10

The Fields of Britannia

nearby Whitely Grange villa shows very similar construction techniques in a rural context (a rubble platform beneath a mortar and clay floor); a remnant magnetic date for the last firing of the bath house is c.AD420–520 (White and Barker 1998, 126-8; Gaffney et al. 2007, 127). The latest timber phases on villa sites aside, however, there remain very few settlements that can be clearly associated with the native British population in the immediate post-Roman period: there are a small number of probably high-status sites in western Britain associated with pottery imported from the Mediterranean during the fifth to sixth centuries (e.g. Cadbury Congresbury and South Cadbury in Somerset: Rahtz et al. 1992; Alcock 1995; Tabor 2008), and a handful of lower-status rural sites (e.g. Poundbury in Dorset: Sparey Green 1987; Tatton in Cheshire: Higham and Cane 1996–7; Trethurgy in Cornwall: Quinnell 2004), but nothing equivalent in the South East of Britain. A number of linear earthworks in the West may also date to this period, and as they face north and or east they may have been built by native British communities under the leadership of social elites living in sites such as Cadbury Congresbury (e.g. Bokerley Dike in Dorset: Eagles 2004, 234; Wansdyke in Somerset and Wiltshire: Eagles and Allen 2011). There is more evidence for the native British population in the burial record. In addition to the recognition that a number of those buried in ‘Anglo-Saxon’ cemeteries in the east may have been of British descent (e.g. Lucy 2000), there is also a growing number of ‘sub-Roman’ cemeteries in the west of Britain (e.g. Cannington, in Somerset: Rahtz et al. 2000; Kenn, in Devon: Weddell 2000). Poundbury, just outside Dorchester in Dorset (Farwell and Molleston 1993) is another well-known example, and a review of the evidence for early medieval burial in Dorset has revealed a large number of other sites that are probably native British (Mees 2014, from which the following is taken). A small group of eleven or twelve graves on Hambledon Hill, for example, were associated with just two iron knives and a pin (Mercer and Healy 2008, 317), while a crouched inhumation within a stone cist at Shapwick was associated with just an early medieval bone comb (Woolls 1839, 105). One of the two undated burials near the Romano-Celtic temple in Maiden Castle is radiocarbon dated to 1315  80 BP (c. cal. AD 635: Brothwell 1971, 237). On Eggardon Hill, excavations uncovered three W–E unfurnished extended burials which have produced a radiocarbon date of cal. AD 640–980 (Putnam 1982; 1983; Cherryson 2005). Excavations at Tinney’s Lane in Sherborne have revealed four unaccompanied inhumations, one of which was a crouched inhumation radiocarbon dated to cal. AD 430–660 (McKinley 1999a). At Manor Farm in Portesham, eight W–E inhumations were excavated, two of which were radiocarbon dated to AD 650–780 and 660–890 (Valentin 2000). At Shepherd’s Farm, Ulwell, in Swanage, three unfurnished cist graves were found in 1949 (Farrar 1949), and in 1982 at least fifty-seven extended inhumations in N–S rows of W–E graves were

Fields of Britannia: A Roman Legacy in the British Countryside

11

discovered nearby, with radiocarbon dates suggesting that the cemetery was in use throughout the seventh century (Cox 1988). At Tolpuddle Ball a small cemetery of fifty W–E graves arranged in well-organized rows was established in the fourth or fifth century3 and continued in use until the seventh century4 (Hearne and Birbeck 1999, 55–63). These sites in Dorset are clearly part of the well-known sub-Roman cemetery tradition of Somerset and Devon in a region that lay beyond the Anglo-Saxon settlement, but a cursory examination of excavation reports to the north and east suggests that similar native British cemeteries are found in regions that did see fifth- to sixth-century Anglo-Saxon occupation, such as Bancroft in Buckinghamshire (Williams and Zeepvat 1994, 115–21) and Saffron Walden in Essex (Bassett 1982b). Recent radiocarbon dating in the Upper Thames valley is showing that a growing number of inhumation cemeteries with unfurnished graves date to the fifth and sixth centuries and are probably therefore probably sub-Roman (e.g. Queensford Farm outside Dorchester-on-Thames: Chambers 1987; Tubney Wood Quarry in Oxfordshire: Simmonds et al. 2011). Shakenoak Farm is a particularly good example, where a small inhumation cemetery was interpreted as ‘Saxon’ in the original published report despite there not being any accompanying grave goods (Brodribb et al. 2005): recent radiocarbon dates have confirmed that they are indeed fifth to sixth century,5 and in the absence of Anglo-Saxon grave goods it is surely most likely that they represent a native British population. Wasperton, in Warwickshire, is an unusual example of a Romano-British cemetery that continued in use into the post-Roman period, finally becoming the burial ground of a community with an Anglo-Saxon identity (Carver et al. 2009). It has traditionally been thought that there is very little material culture in the archaeological record that was produced and used by the native British population in the fifth and sixth centuries, which might be accounted for in three ways: firstly, that ‘absence of evidence is evidence for absence’ (i.e. there was no significant native British population in lowland Britain at this time); secondly, that the evidence has not survived (i.e. artefacts were made of organic materials); and thirdly, that the evidence survives but has just not been recognized (Härke 2007, 58). Certain aspects of the third proposition are well rehearsed, such as the possibility that early ‘Anglo-Saxon’ cemeteries may include the burials of native Britons, and that the post-built halls on early ‘Anglo-Saxon’ settlements incorporate elements of British design and so may 3

1660 +/ 35 BP (OxA-8299) cal. AD 250–450. 1470 +/ 35 (Ox-8320) cal. AD 530–660; 1450 +/ 30 (Ox-8300) cal. AD 550–650; 1440 +/ 35 (Ox-8298) cal. AD 550–660; 1380 +/ 35 (Ox-8297 cal. AD 600–690. 5 Six burials have yielded dates of 1531 +/ 24 BP (cal. AD 442–575); 1577 +/ 28 BP (cal. AD 390–527); 1580 +/ 25 BP (cal. AD 409–532); 1612 +/ 26 BP (cal. AD 433–533); 1616 +/ 26 BP (cal. AD 390–527); 1630 +/ 25 BP (cal. AD 434–535): John Blair pers. comm. Reynolds (2009, 41) discusses the weapon injuries. 4

12

The Fields of Britannia

have been built by British workers (but see Hamerow 2002 for a rebuttal of this view). While it appears that some categories of Romano-British material culture disappeared altogether, such as the shale industry (Cool 2000, 48), other craft production and consumption appears to have continued. There is, for example, growing evidence for the survival of British manufacture of metalwork such as ironworking (e.g. Silchester: Fulford et al. 2006a, 155), the ‘Quoit Brooch’ style, penannular brooches, and hanging bowls, and the technique of enamelling (Dickinson 1982; Geake 1999; Suzuki 2000, 109–10; Henig 2002, 10; Laing 2007; Youngs 2007; Green 2012, 69–74). Evidence of metalworking is relatively common on high-status fifth- to sixth-century sites in the west of Britain (e.g. Cadbury Congresbury: Rahtz et al. 1992). Briscoe (2011) has suggested that some fifth-century stamped pottery traditionally ascribed to Anglo-Saxon immigrants may, in fact, reflect a continuation of British designs, as appears to be the case with some woollen textiles found at early medieval settlements (e.g. Flixton in Suffolk: Boulter and Walton Rogers 2012, 117–21, and fig. 10.2). Owen-Crocker (2011) has suggested that a particular type of woollen twill found in some ‘Anglo-Saxon’ contexts has no precedent in the Germanic homeland, or the Frisian and Rhineland districts with which they traded, and so may also be of British manufacture. The chemistry of ‘Anglo-Saxon’ vessel glass is the same as in the Roman period, and the change from soda-lime-silica to potash glass only came in the late first millennium AD (Stiff 2003). Although it is widely thought that the late Romano-British pottery industries all collapsed by the early fifth century, a variant of South East Dorset BB1—Orange Wiped Ware—may have continued in production for several decades (Gerrard 2010). While it would appear logical that the void created by the collapse of the market-focused large-scale pottery industries of late Roman Britain was filled with local production, evidence for this remains limited. At Childerley Edge in Cambridgeshire, for example, the occupation of a late Romano-British farmstead clearly continued beyond the end of the fifth century and an artefact-rich ‘dark earth’ formed across the site (Abrams and Ingham 2008). There was a distinctive change in the animal bone assemblage from cattle to sheep/goat, and alongside the latest Romano-British massproduced colour-coated fine wares and shelly wares a small quantity of sherds with a grog-tempered fabric may represent sub-Roman production. Indeed, need the ‘hand-made coarse ware’ comprising ‘wide mouthed jars, very plain and undecorated’, recovered from a late Roman settlement at Boxworth in Cambridgeshire necessarily be ‘Saxon’, as is assumed (cf. Connor 2008, 115)? It is also very likely that late Romano-British material culture remained in use for at least the early part of the fifth century (Cool 2000), but without independent dating evidence it is impossible to say for how long. Although it is often suggested that wear on coins indicate that they remained in use for several decades on both urban (e.g. Cirencester, Gloucestershire: Simmonds and Smith

Fields of Britannia: A Roman Legacy in the British Countryside

13

2008; Holbrook 2013, 33) and rural sites (e.g. Barton Court Farm, in Berkshire: Miles 1986; Shakenoak villa, in Oxfordshire: Brodribb et al. 2005), Besly (2006, 83-4) has noted that in part this may reflect that they were poorly made objects in the first place. Overall, however, it is true that the native British population remains hard to identify in eastern Britain in the fifth and sixth centuries, but in part this is because the evidence has not been recognized. One reason why it remains difficult to identify the landscape of the native British population in the early medieval period is the tendency for any occupation in this period to be labelled ‘Anglo-Saxon’. At Latimer, in Buckinghamshire (Branigan 1971, 173, 187), for example, some simple post-built structures are interpreted as ‘Anglo-Saxon’ seemingly because it was assumed that occupation in this period had to be Anglo-Saxon (in the sense of immigrants from mainland Europe), while at Shakenoak a small inhumation cemetery was interpreted as ‘Saxon’ despite there not being any accompanying grave goods (see above). As a more detailed example of how post-Roman occupation potentially by a native population has been regarded as ‘AngloSaxon’, we can re-examine the sequence at Orton Hall Farm in Cambridgeshire (Figure 1.3). The General Introduction of the published report sets the agenda from the very start: ‘In Period 5 [c.375 to early sixth century], the Roman site became degraded with some buildings being reduced in size, although brewing on a large scale still continued. Anglo-Saxons occupied the east and west ends of the Roman main yard and gradually took over the whole plan, possibly retaining one of the barns in use all the time’ (Mackreth 1996, xv). The report clearly argues that Anglo-Saxons took over a working farm, but a closer examination of this site’s biography—as written in the archaeological record—reveals an alternative hypothesis. The structures that date to this phase were clearly built with reference to the late Romano-British farmstead: they are on the same orientation, and are positioned within the paddocks and in association with the buildings around the periphery of the existing farmyard, so respecting that open space. There is, however, not a single indisputable Anglo-Saxon building. At 22 m2, the single alleged ‘Sunken Featured Building’ (F.204) was considerably larger than the average sizes at other sites (see Table 1.1), had very shallow and irregular sides, and lacked roof posts (Mackreth 1996, figs 33 and 55). This is extremely rare—at West Stow, for example, just seven of the sixty-nine Grubenhäuser had no evidence for roof posts—and along with its shallow sides and irregular shape, raises the possibility that this was simply a working hollow. It is also unusual to find just a single Grubenhäuser on what was an extensively excavated site. Of the three other ‘Anglo-Saxon’ buildings, one is simply a rectangular post-built structure of which there is nothing distinctively Germanic; the second is a group of post holes, some of which do form short lines, but which are not a convincing rectangular building; and the third is a square setting of nine posts, interpreted as a granary. Granaries are, however, extremely rare on genuinely early

14

The Fields of Britannia Late Roman N

0

50 m

fifth century N

?grubenhäuser ?granary post-built structure

0

50 m

ridge and furrow N

medieval furrows 0

50 m

Fig. 1.3. The late Roman and fifth-century occupation at Orton Hall Farm in Cambridgeshire, overlain unconformably by later medieval ridge and furrow (after Mackreth 1996, plates VI and VII; drawn by Mike Rouillard).

Fields of Britannia: A Roman Legacy in the British Countryside

15

Table 1.1. Average sizes of Grubenhäuser on selected early Anglo-Saxon settlements Kilverstone, Norfolk Gamlingay, Cambridgeshire Barrow Hills, Radley, Oxfordshire West Heslerton, Yorkshire Mucking, Essex Yarnton, Oxfordshire West Stow, Suffolk Carlton Colville, Suffolk

8.96 m2 10.75 m2 10.96 m2

sample of 10 sample of 12 sample of 41

Garrow et al. 2006 Murray and McDonald 2005 Chambers and McAdam 2007

11.7 m2 12.9 m2 13.1 m2 13.5 m2 16.0 m2

sample of 140 sample of 212 sample of 7 sample of 69 sample of 35

Tipper 2004, 64 Hamerow 1993 Hey 2004 West 1985 Lucy et al. 2009

Anglo-Saxon settlements (i.e. those that have Grubenhäuser: Hamerow 2012, 61) and at Orton Hall Farm the only dating evidence came from a tenth post hole, to the south, which is not obviously part of the structure. Of the artefactual evidence at Orton Hall Farm, the ‘Anglo-Saxon’ pottery is dominated by handmade, undecorated, globular vessels whose simple, easyto-produce form suggests that they could equally have been made by the native Romano-British population. One sherd in this fabric is clearly of a mortarium (a distinctively Roman form), and while the excavator interprets this as ‘the only tangible evidence for a direct interaction between Roman and Anglo-Saxon’ (Mackreth 1996, 27), this view reflects the traditional ethnic interpretation of the archaeological record in this period as being one of two clashing cultures, of which the latter was dominant: this same sherd can equally be interpreted as the local sub-Roman population maintaining some of the trappings of their former life through locally producing handmade pottery in the established designs. Indeed, while several of the ‘Anglo-Saxon’ bowls have a bi-conical form, these are in a Romano-British fabric (Mackreth 1996, 205–6), and it is very noticeable that there are no sherds with AngloSaxon-style bosses, chevrons, or stehende bogen decoration, something that— although rare—is seen on domestic settlements elsewhere (e.g. Cambridge Backs: Dodwell et al. 2004, 117). There is one final aspect of Orton Hall Farm that is curious: the number of enclosure ditches that were re-dug during this fifth-century phase (being associated with this handmade pottery). The fifthcentury settlements that have been extensively excavated across South East Britain and which are clearly Anglo-Saxon (i.e. with large numbers of genuine Grubenhäuser) all noticeably lack boundary ditches (Hamerow 2012, 71),6 whereas it is common to find late Romano-British farmsteads dividing up their space in this way. The conclusion, surely, is that the continued practice of defining space within Orton Hall Farm through ditches reflects the site’s 6 Examples include Bishopstone in Sussex (Bell 1977); Foxholes Farm in Hertfordshire (Partridge 1989), Mucking in Essex (Hamerow 1993), West Stow in Suffolk (West 1985), and Yarnton in Oxfordshire (Hey 2004).

16

The Fields of Britannia

occupancy by a native British population. It could be argued, indeed, that there are just two securely ‘Anglo-Saxon’ aspects to Orton Hall Farm: a small number of vessels with stamped decorations and a bone comb of Frisian origin that can only be dated to the late fourth or fifth century (being a form that is common in the late Empire): these could all have reached the site during the late Roman period, or through exchange with a nearby immigrant Anglo-Saxon community in the fifth century. Orton Hall Farm provides an example of the complexities associated with individual site biographies which, when they reach this period, become fragmented, badly preserved, and poorly dated. It also shows, however, that the question of ethnicity has affected how this site was viewed by the excavator. The interpretation of F.204 as a Grubenhäuser is far from clear. The building(s) of posthole construction have nothing ‘Germanic’ about their character. Yes, some of the material culture from this phase shows Anglo-Saxon traits, but these items could have been brought to the site through exchange, while other elements of the ceramic assemblage are suggestive of a native community trying to carry on with life as they knew it. And finally, the continued definition of space within this settlement using ditches is a Romano-British tradition, not an Anglo-Saxon one. While this re-assessment of Orton Hall Farm suggests that it may have been the native British population who continued to manage this estate during the fifth century, we really need new ways of studying this period that do not rely solely upon the ephemeral, badly preserved, and poorly dated excavations of individual settlements. In this study, therefore, we explore two particular strands of evidence that provide a broader, landscape, perspective: firstly the palaeoenvironmental sequences that allow us to examine what happened to patterns of land-use from the late Roman through to the early medieval periods, and secondly, the related topic of what happened to the field systems of late Roman Britain and in particular their relationship to the medieval and modern landscape. Overall, the aim is to explore the legacy of Roman Britain in the fieldscape of today.

THE F IELDS OF BRITANNIA The fifth to seventh centuries are clearly a contested period, with very different views over whether it was one of continuity or change. Although most would agree that ‘the basis of the early Anglo-Saxon economy was the land’ (Arnold 1988, 17), one of the problems is that most fieldwork and discussion has focused on just two types of site—cemeteries and settlements—as opposed to the landscape as a whole. This study is an attempt to redress that imbalance, and represents the major outcome of the Leverhulme Trust-funded Fields of Britannia Project, carried out from 2010 to 2012. Its focus was the legacy of the

Fields of Britannia: A Roman Legacy in the British Countryside

17

Roman period in the rural landscape of today, and while informing the debate about the origins and development of regional variation in landscape character—such as why some areas saw the development of villages and open fields in the eighth to tenth centuries—the emphasis is on landscape evolution during the late Roman and earliest medieval periods (the fifth to seventh centuries). This is a study that tries to establish the big picture: while individual site biographies are explored, the primary aim was to assess what happened to the landscape as a whole at the end of Roman Britain. This is a study, therefore, that has quantified data as its starting point, based upon both archaeological and palaeoenvironmental evidence, and is free from the traditional historical framework (some might say straightjacket) that is based upon Gildas, Bede, and the Anglo-Saxon Chronicle. The Fields of Britannia Project therefore explores three specific topics: 1. Land-use: an analysis of palaeoenvironmental sequences in order to determine patterns of continuity or discontinuity in land management practices from the late Roman through to the early medieval periods. 2. Field systems: studying the extent of possible continuity or discontinuity in the physical fabric of the countryside by examining the relationship between late Romano-British field systems and their medieval successors. 3. Settlement patterns: an examination of the extent to which there was continuity or discontinuity in settlement patterns from the late Roman through to the early medieval period using three case studies that between them embrace areas with a continuous ceramic sequence (Norfolk), limited early medieval ceramics (Kent), and an aceramic early medieval period (Somerset). This book focuses on the first two topics (land-use and field systems), as the third (settlement patterns) was the subject of separate thesis by Fiona Fleming (2013). It is the final publication from the Fields of Britannia Project, and as such supersedes the interim reports that have previously been published (Rippon et al. 2010/11; 2011; 2012a, b; 2013). In Chapter 2 an approach to studying landscape evolution at a regional scale will be introduced as, in contrast to the traditional simplistic division of Roman Britain into ‘upland and lowland’, ‘military and civilian’, and ‘native and villa’ landscapes, this study has identified nine discrete regions (the South East, East Anglia, Central Zone, South West, Lowland Wales, Western Lowlands, North East Lowlands, Upland Wales, and Northern Uplands), each of which saw a different pattern of development throughout the late Roman and early medieval periods. Chapter 3 explores in greater depth the two particular facets of the late Roman and early medieval landscape that form the focus of this study: patterns of land-use and the evolution of fieldscapes. The methodologies

18

The Fields of Britannia

used to synthesize the data from palaeoenvironmental sequences and excavations of field systems are outlined, and this is followed by a general discussion of the problems of trying to determine whether there was continuity or discontinuity within the landscape. Chapters 4–11 then discuss the patterns of land-use and the evolution of fieldscapes in each of the major regions of first-millennium AD Britain (south of Hadrian’s Wall) with detailed discussion of selected sites and sequences placed within the broader context of how landscape development in that particular region compares to the other lowland or upland regions as a whole. In Chapter 12, the general themes that have emerged are discussed.

2 A Regional Approach to Studying Landscape LANDSCAPE CHARACTER IN ROMAN BRITAIN The character of the British landscape has been shaped by a combination of its natural topography, geology, and soils, its climate and changing sea levels, and a wide range of cultural factors, such as the response of communities to the emergence and decline of market economies and the influence of external societies. These processes were not, however, experienced uniformly across Britain. The South East, for example, saw the emergence of complex societies during the late Iron Age, in contrast to areas further west and north, where the pace of change was slower. The impact of Britain’s assimilation into the Roman world also varied from region to region, and between different social groups, and the combination of how these different native societies responded to Britain’s new social, economic, and political relationship with mainland Europe led to the development of landscapes of very different character, with highly urbanized and market-driven countryside in some areas, compared to others where engagement with the Empire was less. A feature of past scholarship on the landscape of Roman Britain has been a tendency to treat the lowlands as a single landscape. This simplistic view can be traced back to Haverfield’s (1912) division of Roman Britain into ‘civil’ and ‘military’ districts, and Fox’s (1932) identification of ‘lowland’ and ‘upland’ zones (Figure 2.1A), since when many scholars have continued to use such simplistic binary characterizations, the dividing line between them running roughly between the Blackdown Hills on the Devon/Dorset border and the Tees estuary in the North East (e.g. Salway 1981, 4–5). There is also a strong tradition of discussing the archaeology of Roman Britain in broadly thematic terms—the military establishment, towns, trade and industry, the countryside—which adds to the impression of homogeneity across large parts of the province (Figure 2.1B; e.g. Collingwood 1930; Frere 1967; Millett 1990). Dark and Dark’s (1997) substitution of the term ‘villa landscape’ for ‘civil zone’, and ‘native landscape’ for ‘military zone’ (Figure 2.1C),

20

The Fields of Britannia

A

C upland

upland

lowland

native

villa

upland 150 km

D CA RV ETI

I

TE IG

BR

G

LI

O

CO

I IC

I

V IC

I OV RN

O RD

N TA

E

ES A

civilian

D OB U N

D EM

ET

SI L U R E D

NI

TU CA

S

L VE

LA

EN

I UN OV AN

N

CO RI

E

A

I

IS

AN

R PA

G DE

military

I

S TE

TR IN

colonia municipium civitas capital legionary fortress

B

0

150 km

S

0

E

B E AT REB AT E CA S LG UR N T I A CI A O TR REGN EN SES IG

0

150 km

MN

I ON I

ES

DU

0

150 km

Fig. 2.1. Traditional binary divisions within Roman Britain, and its civitates: (A) the upland-lowland divide; (B) ‘military and civilian’ areas; (C) ‘villa and native’; (D) the civitates (after Fox 1932, map facing p. 28; Mattingly 2006, fig. 10; Dark and Dark 1997, p. 68; Frere 1967, fig. 1; drawn by Mike Rouillard).

not only reiterates this existing over-simplification but adds a new and misleading dimension in implying that all lowland areas were characterized by villas, when even a cursory examination of a distribution map shows that this is obviously not the case (e.g. Figure 2.2: East Anglia, for example, and in particular the east of that region, has few known villas, and those that have been found mostly lack mosaic pavements). In recent years, however, there has been a more sophisticated understanding of some local variations in the Romano-British landscape, with simplistic

A Regional Approach to Studying Landscape

21 land over 400 m land over 200 m villa

0

150 km

Fig. 2.2. Romano-British villas, showing how they are far from evenly distributed across lowland areas (after Jones and Mattingly 1990, map 7.6, and Taylor 2007, fig. 4.9; drawn by Chris Smart).

ideas that areas lacking villas were imperial estates having been challenged (e.g. Rodwell 1978; cf. Hingley 1989; Millett 1990, 120). In the most developed appreciation that the landscape of Roman Britain was far from uniform in its character, Mattingly (2006) rejects the traditional thematic approach to discussing Roman Britain (the conquest and garrisoning of Britain, its towns, the countryside, industry, etc.) in favour of exploring how three ‘communities’— military, civil (urban), and rural—interacted with each other in different areas. In his discussion of the development of regionally distinctive societies Mattingly prefers the term ‘discrepant experience’ to ‘Romanization’, which he defines as ‘the co-existence of very different perceptions of history, culture,

22

The Fields of Britannia

and relationships between colonizer and colonized’ (Mattingly 2006, 14, 17; and see Pitts 2008 for another recent discussion of ‘Romanization’, identity, and the effects of Roman imperialism). A similar philosophy is put forward by Sargent (2002), who discusses how there were different processes of acculturation in the South/East and West/North of Britain. The most recent discussion of rural settlement in Roman Britain, by Taylor (2007), also recognizes that there may have been significant differences in landscape character, although one problem with that study is its use of twenty-first-century units of regional government to divide up the landscape: the ‘The South East’, for example, stretches from Kent to the foot of the Cotswold Hills and therefore embraced regions of very different character in the Roman period. In contrast, it is argued here that we should seek to study sub-divisions of the landscape that would have been meaningful to the people living there in the past in order to achieve a more sophisticated understanding of the Romano-British landscape. Dark’s (2004, 281) assertion that ‘The eastern and southern parts of early fourth-century Roman Britain had a settlement pattern dominated by villas and towns’ is simply not true of all areas, as these different types of site were far from evenly distributed. Urbanism provides a good example of regional variation in landscape character (Figure 2.3). In many areas there was a welldeveloped hierarchy of towns, with large civitas capitals that provided administration, services, and market facilities for a territory, alongside a series of small towns, most of which appear to have been local market centres. Such a network of towns is seen across a broad swathe of central England (from Dorset and Somerset, through the East Midlands, and up into East Yorkshire), as well as northern Kent, Hertfordshire, and central/northern Essex, parts of the western Midlands, and South East Wales. There were, however, significant lowland areas that were either devoid of small towns (e.g. the Weald, the chalk downland of Wessex, and the South West), or where any potential local market centres were small and poorly developed (e.g. East Anglia). The more urbanized landscapes invariably correspond to areas with countryside characterized by villas, which is particularly clear in areas such as the western Midlands and South East Wales, where the few villas cluster around towns such as Viroconium Cornoviorum (Wroxeter), Pennocrucium (Water Eaton), Letocetum (Wall), and Venta Silurum (Caerwent). Elsewhere, areas that are noticeably devoid of small towns are also lacking in villas, such as southern Essex, the Weald, and the chalk downland of Wessex, even though the rich material culture from rural settlements in these areas shows that their populations were integrated with the market economy. The South West presents a rather different picture, with just a very small number of villas close to its civitas capital at Isca Dumnoniorum (Exeter) and rural settlements that yield relatively little pottery, coins, or other manufactured goods, suggesting a far less Romanized society, certainly before the late Roman period. East Anglia is another region whose relationship with Romanitas was different to the

A Regional Approach to Studying Landscape

23

legionary fortress colonia civitas capital ‘small town’, fortified ‘small town’, unfortified land over 400 m land over 200 m

0

150 km

Fig. 2.3. The hierarchy of towns that existed across parts of Roman Britain (after Jones and Mattingly 1990, map 5.12; drawn by Mike Rouillard).

majority of lowland Roman Britain. Although a series of possible ‘small towns’ have been recognized, relative to other parts of lowland Roman Britain they appear to have been small and poorly developed (J. Davies 2009, 179–86). The same is true of the villas in this region with only a few examples reaching the scale seen elsewhere in lowland Roman Britain, most of which lie in the far west along the Icknield Way on the margins of Fenland (J. Davies 2009, 187–93). The modest scale of even the most Romanized buildings in regions such as East Anglia and the South West is also seen in the distribution of mosaics, which

24

The Fields of Britannia

are extremely rare in these regions outside of their civitas capitals (Figure 2.4). The majority of Romano-British mosaics, particularly at villas, are fourth century and were produced by craftsmen who appear to have been based in the towns of Durnovaria (Dorchester, in Dorset), Corinium (Cirencester), Durobrivae (Water Newton), and Petuaria (Brough on Humber), with a fifth school in central southern Britain, perhaps based at Calleva Atrebatum (Silchester), Venta Belgarum (Winchester), or Noviomagus Regnorum (Chichester). It is very noticeable that there was no mosaic school in the South East or East Anglia, and indeed very few late Roman mosaics in these areas are known. Although this may reflect a society in which wealth and status was demonstrated in other ways, for example through portable artefacts or cattle, it could also indicate that these regions had undergone an economic decline—or at least did not share in the economic prosperity seen further west—during the fourth century. Investment in agricultural infrastructure such as corn-drying ovens, particularly the T-shaped forms that were most common in the late Roman period, was also far greater in regions such as the Central Zone, compared to the South East and East Anglia (Morris 1979), suggesting that there were complex processes at work that included receptiveness to innovation. Within Roman Britain we can therefore see differences in landscape character that varied in both time and space. Faulkner (2000, figs 10–11, 26, 33–4, and 60) has tried to quantify the prosperity and decline in settlements through a series of calculations: the number of rooms occupied in private houses within towns, the number of elite houses in towns, the value of construction work in towns, the number of rooms occupied in villas, the value of construction work in villas, and the number of native rural sites recorded through fieldwork in the journal Britannia for 1969–96 (Figure 2.5). There are many methodological problems with this approach—such as dating and identifying just which rooms within which settlements were used at specific times, and for what purposes—but this does provide at least a rough indication of broad trends. Stone-built villas were undoubtedly in decline in most regions during the second half of the fourth century—the Cotswolds were a notable exception— but a failure to maintain their opulence does not mean that they ceased to function as working farmsteads. This is clearly illustrated at Barnsley Park, in Gloucestershire, where the main house was demolished but a series of buildings of simpler construction were built to the east, showing that it was still a functioning agricultural unit (Figure 2.6; see below). Across Roman Britain there is also evidence for settlements being abandoned because field systems were extended and reorganized in the late Roman period, but again this suggests a still functioning landscape that was undergoing reorganization rather than decline. While some areas may have been abandoned altogether—a process known as agri deserti (Esmonde Cleary 1989, 29)—it is difficult to be sure that this was really the case in Britain, as the lack of

A Regional Approach to Studying Landscape A

25

C

0

0

150 km

rectangular Romano-Celtic temple polygonal Romano-Celtic temple

150 km

number of mosaics village or rural building town 1–3 1–3 4–9 4–9 10+ 10+ 30+ D

B

0 generalized distribution of Roman pottery kilns

150 km

0

150 km

Petuarian school Corininan schools Durnovarian school Durobrivan school South-Central school

Fig. 2.4. Aspects of landscape character in Roman Britain. (A) Romano-Celtic temples; (B) pottery production sites; (C) mosaic pavements; (D) mosaic schools (after Jones and Mattingly 1990, maps 8.23, 6.24, 6.39, and 6.41; drawn by Mike Rouillard).

A number of rooms occupied in villas

1,200 1,000 800 600 400 200

600 500 400 300 200

0 60

100

200

300

400 date

B

70 number of elite houses in towns

700

100

0

80

60 50 40 30 20 10 0 60 100

200

300

8,000

400 date

60 100

200

300

250

400 date

E

200

150

100

50

0 60 100

C

200

300

400 date

F

7,000 160

6,000

140 number of rural sites occupied

value of work done in towns (in 1000s of construction units)

D

800

value of work done in villas (in 1000s of construction units)

number of rooms occupied in private houses within towns

900 1,400

5,000 4,000 3,000 2,000 1,000

120 100 80 60 40 20

0

0 200

300

400 date

date

C5+ early C4 late C4 mid C4 early C3 late C3 mid C3 early C2 late C2 mid C2 early C1 late C1 mid

60 100

Fig. 2.5. Measures of prosperity in Roman Britain. (A) the number of rooms occupied in private houses within towns; (B) the number of elite houses in towns; (C) the value of construction work in towns; (D) the number of rooms occupied in villas; (E) the value of construction work in villas; (F) the number of native rural sites recorded through fieldwork in the journal Britannia 1969–96 (after Faulkner 2000, figs 10, 26, 11, 34, 33, and 60; drawn by Mike Rouillard).

A Regional Approach to Studying Landscape

27

ceramic evidence from field boundaries may simply reflect a shift from arable to pasture and hence a lack of manuring (e.g. Raunds, in Northamptonshire: Figure 2.9; Parry 2006, 81–3, figs 4.10–4.13). One example of these late Roman changes in the rural landscape can be seen on the Boulder Clay Plateau of northern Essex, where there has been a series of excavations in advance of the expansion of Stansted Airport, the new A120, and nearby urban expansion (Medlycott 1994; Lavender 1997; Humphrey 2002; Havis and Brooks 2004; Brooks and Holloway 2006; Timby et al. 2007a; Cooke et al. 2008, 153–78). Across this area a number of first- to third-century settlements were abandoned, and although occupation continued into the late Roman period at three sites, they were no longer associated with the small-scale field systems that had parcelled up parts of the landscape earlier in the Roman period. There was still arable cultivation, for example on the relatively loamy soils at the DCS/DFS site where the plant macrofossils associated with a corn-drying oven were suggestive of the largescale processing of spelt wheat, but overall the environmental evidence— including plant macrofossils, pollen, and soil micromorphology—suggest more limited arable cultivation than there had been in the earlier Roman period. There was, however, still very little woodland in this landscape, but abundant improved pasture and the animal bone assemblages show an increased emphasis on cattle husbandry (Havis and Brooks 2004, 269, 338; Cooke et al. 2008, 156–98). This pattern of late Roman reorganization is also seen at a series of sites along the nearby A120. At Strood Hall, while the early Roman settlement continued to be occupied in the late Roman period, the earlier field system and trackways were abandoned and, similarly, the density of field boundary ditches at the Parsonage Lane and Rayne Roundabout sites was much lower in the late Roman period (Timby et al. 2007a, 81–147). Environmental evidence, however, once again shows that arable cultivation continued, and that there was a growing specialization in the late Roman period on the growing of spelt wheat. At Strood Hall the cattle bones suggest that in the late Roman period the animals were living longer and were being put to heavy work, presumably ploughing fields. At nearby Great Notley a settlement enclosure went out of use in the late second century and was replaced by fields (Brooks and Holloway 2006), while at Woodside Industrial Park in Birchanger, Buildings Farm in Great Dunmow, and the Old Golf Course in Braintree, the earlier Roman pattern of small, rectangular fields and associated settlement were again replaced in the late third century with far larger fields (Medlycott 1994; Lavender 1997; Humphrey 2002). The picture that emerges from the Boulder Clay of northern Essex suggests that while some settlements were abandoned, this was within the context of a landscape reorganization, not landscape desertion. The replacement of early Roman fields with larger enclosures is seen quite widely in Essex, such as at Monument Borrow Pit, in North Benfleet, which is discussed and illustrated in

28

The Fields of Britannia

Chapter 4 (Figure 4.12; Dale et al. 2005, 35–9) and Woodham Walter (Buckley and Hedges 1987). Although all the settlements recently excavated around Stansted were of relatively low status, the landscape of northern Essex also contained a large numbers of villas. As a group they can be characterized as relatively modest early Roman structures that, although continuing to be occupied into the late Roman period, did not see the palatial embellishments that occurred elsewhere in Roman Britain (e.g. Chignall: C. Clarke 1998; Gestingthorpe: Draper 1985; Rivenhall: Rodwell and Rodwell 1986; Wendens Ambo: Hull 1963, 198; Hodder 1982). This can be contrasted with areas such as the Cotswolds, where many villas were constructed in the fourth century (Holbrook 2006, 103–8). One example is Barnsley Park, near Cirencester, where a small, stone-built bathhouse was added to a timber farmhouse in c.340–60, to which was added a stone-built, winged-corridor villa in c.360, which in turn was extended in c.375–80 through the addition of two rooms, each with a hypocaust (Figure 2.6A). By the end of the fourth century, however, the main house was re-floored with rough stone flagging, and other alterations—such as the main entrance being enlarged and a ramp being added—suggest that it had now assumed an agricultural function (these modifications being to enable carts to gain access). Around the early fifth century the villa was demolished and reduced to a carefully constructed stone platform with associated ramps, and as there was no evidence for a timber superstructure it has been interpreted as a stack-yard (Figure 2.6B). That the site was still functioning as a farmstead is shown by the construction nearby of a series of timber buildings, including K, whose internal span (7.5 m) was virtually the same as the former villa, raising the possibility that the roof timbers were re-used (Figure 2.6C; Webster 1981; 1982). When we come to consider what is meant by ‘continuity’ (see Chapter 3), Barnsley Park is an interesting example: while the demolition of the villa suggests a major discontinuity in the materiality of the community living there, in broader terms the site continued to function as part of an agricultural estate. It is also worth noting that if only the main stone villa complex had been excavated, then the later buildings to the east would not have been recognized which, along with the ephemeral features and artefact-poor nature of subRoman occupation, may explain why so many late Roman villas appear to have been abandoned so suddenly. That there was regional variation within the landscape of Roman Britain, and that this varied in both time and space, is therefore clear, although we know less about how this may have been linked to contemporary social and territorial structures. Roman Britain was divided up into a series of administrative territories known as civitates (Figure 2.1D) that were typically equivalent in size to between two and six medieval counties. The names of most civitates clearly refer back to the tribal groupings of the late Iron Age and so had a deep-rooted social identity, and in the case of less Romanized lowland

A Regional Approach to Studying Landscape A

29

C

N

0

N

0

15 m

15 m

B

K

rubble-filled platform buildings other features

N

0

15 m

Fig. 2.6. The late Roman and early medieval phases at the Barnsley Park villa, Gloucestershire. (A) fourth century; (B) ?early fifth century; (C) ?fifth century (after Webster 1982, figs 19–21; drawn by Mike Rouillard).

30

The Fields of Britannia

areas, such as East Anglia and the South West, a direct link with the preRoman tribes and Roman civitates seems clear (the Iceni and Dumnonii, respectively). One poorly understood aspect of the Roman landscape, however, is the extent to which these civitates were sub-divided into smaller districts, or pagi, of the kind that was found in mainland Europe (Millett 1990, 150; Eagles 2004, 234; Mattingly 2006, 286). One example is recorded on an early second-century wooden writing tablet found at Throgmorton Avenue in the Walbrook Valley, London, which refers to the disputed ownership of a wood called Verlucionun ‘quod est in civitates Cantiorum pago DIBVSSV . . . ’ [‘which is in the civitas of the Cantiaci in Dibussu . . . pagus’] (Tomlin 1996). A number of substantial small towns, such as Corbridge (in the civitas Brigantum), Ilchester (in the civitas Durotrigum),1 and Rochester (in the civitas Cantiaci), may represent pagi centres (Millett 1990, 77, 156; Mattingly 2006, 286–7, 359), but we otherwise have little indication of whether there were such sub-civitates territories in Roman Britain.

LANDSCAPE CHARACTER IN THE E ARLY MEDIEVAL PERIOD In turning to the extent of regional variation in landscape character during the early medieval period we are faced with a greater challenge than in the Roman period, such is the scarcity of data from some areas and some centuries. A broadly familiar three-fold division of the early medieval period is adopted here: the ‘earliest medieval period’ (the fifth to mid-seventh centuries) that equates to what many archaeologists refer to as the ‘Early Saxon period’ (despite the strong ethnic associations of this term that are inappropriate for large parts of Britain that never saw a mass folk migration, or even political conquest until the mid-seventh century); the ‘long eighth century’ (the late seventh to the mid-ninth centuries: see Rippon 2010) that broadly equates with the traditional ‘Middle Saxon period’, and the mid-ninth to the mideleventh centuries (that equates to the ‘Late Saxon period’).

The Earliest Medieval Period (Fifth to Mid-Seventh Centuries) From the discussion above, it is clear that we can only understand landscape evolution in the earliest medieval period within the context of developments 1 Fulford (2006) suggests that Ilchester may have been a civitas capital from at least the second century, although it has traditionally been regarded as representing a sub-division of the Durotrigian civitas.

A Regional Approach to Studying Landscape

31

that had started during the fourth century, and in particular the different trajectories of change that are apparent from region to region. As we move into the fifth century, however, any differences in landscape character across Britain become much less evident, as the highly stratified and materialistic society of the Roman period, with its market and taxation-driven economy and non-agriculturally productive military and urban populations, disappeared. In the west of Britain there is evidence for both high- and low-status settlement in the fifth and sixth centuries, but both appear to assume new forms, with the former often being on hilltops (e.g. Cadbury Congresbury and South Cadbury in Somerset: Rahtz et al. 1992; Alcock 1995), and the lowerstatus settlements becoming less Romanized, with curvilinear enclosures replacing rectilinear forms (Figure 2.7; e.g. Hayes Farm in Clyst Honiton, Devon: Simpson et al. 1989; Cotswold Archaeological Trust 2000; Hart et al. forthcoming; Poundbury in Dorset; Sparey Green 1987; and Wharram Percy in Yorkshire; Wrathmell 2012; possibly Sherborne House in Lechlade, Gloucestershire: Figure 3.11). In most areas, however, the native British landscape is difficult to detect, let alone characterize, and the settlements associated with Anglo-Saxon style buildings (Grubenhäuser), burials (Figure 2.8), and other material culture appear remarkably undifferentiated: although they vary in size (ranging from individual farmsteads to small hamlets) the buildings within them are remarkably similar, and there is a distinct lack of enclosure ditches and internal boundaries that might otherwise facilitate regional classification. The existence of large, communal cemeteries that probably served extensive folk territories (e.g. northern Lincolnshire: Green 2012, fig 11; Norfolk: Hills and Lucy 2013) in some areas but not others does suggest that there were regional variations in the social and territorial structures within which the landscape was managed, although we cannot at present map these in the same way as we can for the later medieval period. One aspect of landscape character that might show regional variation is the extent to which settlements continued to be occupied. Although Powlesland (1997) advocates an ‘Early Saxon Shuffle’, arguing that very few early Anglo-Saxon settlements lie on Romano-British sites, others have come to a different conclusion. Williamson (1986, 126), for example, notes that 35 per cent of the late Roman settlements found in his initial fieldwalking survey in North West Essex lay within 200 m of sites occupied in the eleventh century, with an unspecified number of other late Roman sites lying adjacent to later medieval settlements that may obscure an eleventh-century predecessor. Williamson’s careful re-examination of eighteen Romano-British sites revealed six with early medieval sherds, seven with possible early medieval sherds, and just five with none, suggesting a strong degree of continuity. In South East Somerset, Smart’s (2008) analysis of ninety settlements documented between the eleventh and the fourteenth centuries revealed that of the thirty-four that have seen archaeological investigation, thirteen (40 per cent)

32

The Fields of Britannia A Poundbury, Dorset Late Roman

Fifth–sixth century

N

0

N

50 m

0

50 m

B Hayes Farm, Devon Fifth–sixth century

Late Roman N

N

excavated features cropmarks

0

excavated features 50 m

cropmarks

0

50 m

Fig. 2.7. The change in settlement morphology from rectilinear to curvilinear in the fifth century in western Britain. (A) Poundbury near Dorchester, Dorset; (B) Hayes Farm in Clyst Honiton, Devon (after Sparey Green 1987, fig. 4 and Simpson et al. 1989; drawn by Mike Rouillard).

have produced Romano-British material. One element of the Fields of Britannia Project was therefore a study of the extent to which there was possible continuity in settlement patterns in three sample counties (Norfolk, Kent, and Somerset), the conclusions of which are discussed by Fiona Fleming (2013; forthcoming).

A Regional Approach to Studying Landscape

33

C

A

0 up to c.450 up to c.475

100 km

0

100 km

category A sites category B sites

B

0

100 km

Fig. 2.8. The distribution of Anglo-Saxon burials dating to (A) the mid-fifth; (B) early sixth; and (C) fifth to seventh centuries as a whole (after Hines 1990, figs 1–3). ‘Category A’ sites are ‘where combinations of material culture in closed contexts support a date for the creation of that context within that phase’ and ‘Category B’ sites are ‘where individual artefacts considered characteristic of these phases rather than any later phase are found without support of datable associations’ (Hines 1990, 26; drawn by Mike Rouillard).

34

The Fields of Britannia

The ‘Long Eighth Century’ (Late Seventh to Mid-Ninth Century) There is growing evidence that the ‘long eighth century’ saw a significant change in landscape exploitation across southern Britain (Rippon 2008a, 250–68; 2010). Palaeoenvironmental sequences show an intensification in land-use around the eighth century, with woodland clearance, an increase in arable cultivation, and perhaps the introduction of new systems of farming (e.g. in the South West: Rippon et al. 2006; in the southern part of the Central Zone: Rippon 2012b, 233–40). Another feature of this period was investment in infrastructure projects, including Offa’s Dyke (Hill 2002; Bapty 2011), and causeways at Street in the Somerset Levels (Brunning 2010), across the Thames in Oxford (Dodd 2003), and to Mersea Island in the Blackwater Estuary in Essex (Crummy 1982; Crummy et al. 1982). This period also saw the construction of large coastal fish traps, including those in the Blackwater Estuary near Mersea Island (Murphy 2010), and watermills (e.g. Ebbsfleet in Kent: Andrews et al. 2011; Old Windsor in Berkshire: Wilson and Hurst 1958; Fletcher 1977; Schove 1979; Astill 1997; Tamworth in Staffordshire: Rahtz and Meeson 1992). Such construction projects will have required considerable resources in terms of both labour and materials, and a centralized authority that controlled them. This is also the period when documentary evidence, in the form of charters, shows that the landscape was being divided up into discrete estates, which were granted by kings to the newly emerged Christian monasteries, and most of the major infrastructure projects we know of appear to have been on either royal (e.g. Old Windsor and Tamworth) or monastic estates (e.g Mersea in Essex, Oxford, and Street, near Glastonbury). The archaeological record also sees the emergence of a far more stratified settlement pattern during this period, and in particular high-status settlements across eastern England of a type that were not seen in the fifth to sixth centuries. The villa regalis at Yeavering, in Northumberland, is perhaps the best known, and Blair’s (2005, 275–85) argument that there were periodically occupied residences of peripatetic kings in this period may apply to that site. Elsewhere, however, a number of substantial settlements have been excavated that do appear to be permanently occupied estate centres. At Higham Ferrers, in Northamptonshire, for example, a substantial oval enclosure points to large-scale stock management, and a stone-built malting oven suggests centralized crop processing: although the ownership of Higham Ferrers in the seventh to ninth centuries is not known, the neighbouring parish of Irthlingborough was a royal manor, and Higham Ferrers is likely to have been its collection centre for tribute and food processing (Hardy et al. 2007). Other royal estate centres similarly became hubs for resource processing: Ramsbury, in Wiltshire, was involved in both iron smelting and smithing (Haslam 1980), while the unusual animal bone assemblage dominated by pig

A Regional Approach to Studying Landscape

35

heads at Wicken Bonhunt, in Essex, points to large-scale meat processing (Wade 1980; Rippon 1996, 121). The collection and processing of food and other resources seen at these royal sites was on a far larger scale than was seen in the fifth to sixth centuries, and was part of a far wider change within the landscape of lowland Britain, which also saw the emergence of coastal trading centres such as Ipswich in Suffolk, and Hamwic in Hampshire. The generation of a food surplus that would have been needed to support these settlements may well account for the intensification in landscape use seen in palaeoenvironmental sequences, but whether it was also associated with the physical restructuring of settlement patterns and field systems in England’s Central Zone is less clear. As late as the 1950s it was thought that our villages and open fields dated back to the early Anglo-Saxon migrations, although subsequent fieldwalking across England’s Central Zone and East Anglia has shown that during the fifth to seventh centuries settlement patterns were dispersed, with several small pottery scatters indicative of individual farmsteads or small hamlets spread across what was to become a parish.2 Excavations within deserted medieval villages have also generally failed to produce evidence for early Anglo-Saxon occupation. By the 1970s a revisionist view had emerged that villages and open fields originated in the mid-ninth to twelfth centuries in what Lewis et al. (1997, 191) referred to as ‘the village moment’. There was also a clear consensus that it was the East Midlands that saw the earliest transformation, and that the fashion for structuring the countryside around villages and open fields then spread out across other parts of the Central Zone, so leading to the major tripartite division in landscape character that has been mapped by the likes of Gray (1915), Rackham (1986), and Roberts and Wrathmell (2000a). Raunds, in Northamptonshire, is one of the most extensively investigated parishes and appears to illustrate this late origin for villages. Fieldwalking across the parish revealed around eighteen settlements dated to sometime between the mid-fifth and mid-ninth centuries, each with a manured ‘infield’, but all of which must have been abandoned by the mid-tenth century, as they lack later St Neots Ware (Figure 2.9D). Two such farmsteads lay under parts of the modern village of Raunds, but the nucleated village was not laid out until the later ninth or early tenth century (Audouy and Chapman 2009, 29). An extensive

2 Bedfordshire (e.g. Barton le Clay and Sundon: Hall 1991); Cambridgeshire (e.g. Maxey: Taylor 1983, fig. 39); Leicestershire (e.g. Langton Hundred: Bowman 2006; Medbourne: Liddle 1996; Knox 2004); Gloucestershire (e.g. Bishops Cleeve: Parry 1999; Holbrook 2000; Roel: Aldred and Dyer 1991); Lincolnshire (e.g. Billingborough, Pointon, and Sempringham; Dowsby; Morton, Bourne, and Thurby: Hayes and Lane 1992); Norfolk (e.g. Barton Bendish: Rogerson et al. 1997; Hales, Heckingham, and Loddon: Davison 1990; Witton: Lawson 1983); Northamptonshire (e.g. Brixworth: Taylor 1983, fig. 39; Ford 1995, fig. 1; Great Doddington: Foard 1978; Leckhampstead: Jones and Page 2006; Raunds: Parry 2006); and Somerset (e.g. Shapwick: Gerrard with Aston 2007).

36

The Fields of Britannia

fieldwalking survey in the Langton parishes, in South East Leicestershire, similarly revealed a dispersed scatter of fifth- to seventh-century settlements, each with an intensively manured infield (Bowman 2004). Another area of the East Midlands that has seen extensive archaeological survey is Whittlewood, where fieldwalking in lower-lying parishes such as Leckhampstead, in Buckinghamshire, revealed a dispersed settlement pattern similar to that seen in the Northamptonshire parishes of Brixworth and Raunds (Jones and Page 2006, 85). These settlements are once again associated with pottery that can be dated no more closely than the fifth to mid-ninth centuries, although later fabrics are absent. Although Partida et al. (2013, 38) have reiterated their view that there are several ‘Early to Middle Saxon’ settlements scattered across many East Midland parishes, and that these were nucleated into villages during the ‘Middle Saxon period’, Williamson et al. (2013, 58) have questioned this, noting that in 51 per cent of parishes there are no known sites associated with ‘Early to Middle Saxon’ pottery, in 28 per cent of parishes there is just one, in 10 per cent there are two, in 7 per cent there are 3, and in 4 per cent more. They therefore suggest that in most cases villages formed around what had been the single ‘Middle Saxon’ settlement within a parish, and that rather than a ‘village moment’ or a ‘great replanning’ there was a period of more gradual expansion from this single settlement over the course of several centuries. Parishes in the west of Whittlewood (on the Buckinghamshire/Northamptonshire border), for example, have not always produced a scatter of fifth- to eighth-century sites, but a counter-argument is that this reflects the fact that not all of the parishes could be fieldwalked (due to areas of permanent pasture, woodland, and modern housing). The higher elevation and heavier soils of this interfluvial plateau also suggests that this was a relatively marginal area of secondary settlement that may well have not been typical of lower-lying areas (Partida et al. 2013, 92–6). Although Williamson et al.’s (2013, 58) data would therefore appear to suggest that the widely accepted ‘Northamptonshire model’ for ‘Middle Saxon’ settlement nucleation only works in c.20 per cent of parishes, these statistics cannot be taken at face value, as the percentage of each parish that has actually been fieldwalked is not given. If, for example, only 25 per cent of the acreage in the parishes where just one site associated with ‘Early to Middle Saxon’ pottery has been found was fieldwalked, then statistically there are probably four sites associated with ‘Early to Middle Saxon’ pottery in each parish. Another problem with the Williamson et al. (2013) statistics is that they take no account of the technique of fieldwalking that was used: whereas grid-walking, for example, should have detected every site, widely spaced line walking is bound to have missed some. The data used by Partida et al. 2013 is ultimately derived from surveys by David Hall, and a previous account of the methodology he used makes it very clear that many parishes were only walked very rapidly as part of a survey that was primarily aimed at identifying the

A Regional Approach to Studying Landscape

37

headlands of open fields, and some Northamptonshire parishes were not walked at all (Foard et al. 2009, fig. 30). The Williamson et al. hypothesis— that there was no process of settlement nucleation—is therefore flawed because parishes were not completely walked, those areas that were walked were not done so systematically and in detail, and some of the parishes without a high density of settlements were in physically marginal areas. Away from the poorest soils, the evidence does suggest that there was a dispersed pattern of settlements associated with ‘Early to Middle Saxon’ pottery that was replaced by a single nucleated village (e.g. the intensively walked parish of Raunds: Figure 2.9). This debate between Partida et al. (2013) and Williamson et al. (2013)—over the outcomes of the same research project—also reminds us that in one of the most intensively investigated landscapes in the country controversy still rages over how villages and open fields came about! The date when villages started to form is also contested. Contrary to assertions such as ‘the excavation of more and more deserted and shrunken villages . . . showed very little evidence for occupation earlier than the eleventh or twelfth centuries’ (Lewis et al. 1997, 3), there is now growing evidence that the nucleation of settlement—although not necessarily the laying out of villages in their later form—may have begun somewhat earlier than the late ninth century. This has really only become apparent through recent development-led excavations within currently occupied settlements (e.g. Wright 2013), and there are various reasons why it was less apparent in earlier work. Some of the excavations within currently occupied villages that failed to find ‘Middle Saxon’ occupation were in peripheral areas some distance from the likely early settlement focus around the church (e.g. Great Linford in Buckinghamshire: Mynard and Zeepvat 1992, figs 11–12; but see below for earlier work at the church), while some ‘deserted medieval villages’ within which excavations failed to find ‘Middle Saxon’ occupation are actually secondary hamlets, and as such may have been founded later than their nearby parochial village (e.g. Grenstein in Tittleshall: Wade-Martins 1980; West Cotton in Raunds, Northamptonshire: Chapman 2010; and see Creighton and Rippon forthcoming). Other excavations have been in villages that had shifted away from their original locations by the parish church, and as such will inevitably not produce evidence for earlier occupation (e.g. Thruxton in Norfolk: Butler and Wade-Martins 1989). An example of a medieval village within which Middle Saxon occupation is now recognized is Wharram Percy (Wrathmell 2012, 118–79). This is also a good example of how the process of initial nucleation was separate from the creation of the physical layout of villages in the form which we are familiar with today, as the series of curvilinear ‘Middle Saxon’ enclosures clearly represent a nucleated settlement, but one whose layout was then replaced by the rectilinear layout of the later village. A re-examination of other old excavation reports reveal that deserted medieval settlements have in fact produced ‘Middle Saxon’ material

38

The Fields of Britannia

A

N

areas not fieldwalked

0

3 km

B

N

areas not fieldwalked

0

3 km

Fig. 2.9. The results of fieldwalking in Raunds, Northamptonshire. (A) middle Roman pottery; (B) late Roman pottery; (C) density per ha of Roman pottery; (D) ‘early to middle Saxon’ sites and manure scatters (after Parry 2006, 81–3, figs 4.11–4.14; drawn by Mike Rouillard).

A Regional Approach to Studying Landscape

C

39

N

areas not fieldwalked

settlement areas

0

D

3 km

N

manure scatters

Fig. 2.9. Continued.

areas not fieldwalked

settlement areas 0

3 km

40

The Fields of Britannia

(e.g. excavations adjacent to the churches in Great Linford and Milton Keynes, Buckinghamshire: Mynard and Zeepvat 1992, 109–10; Williams 1993). Fieldwalking in East Anglia also consistently shows that the scattered fifth- to seventh-century settlements lack ‘Middle Saxon’ Ipswich Ware, which is only found in a single scatter around the parish church (e.g. Barton Bendish: Rogerson et al. 1997; Hales, Heckingham, and Loddon: Davison 1990; parishes of the Launditch Hundred: Wade-Martins 1980; Witton: Lawson 1983), and excavations within villages is revealing Middle Saxon occupation (e.g. Whissonsett: Mellor 2004; Trimble 2006). This accumulation of evidence in East Anglia now means that ‘It is widely accepted that a great restructuring of the landscape occurred during the Middle Saxon period, not least the coalescence of the numerous, comparatively transitory early Saxon settlements into more permanent settlements’ (Hoggett 2010, 205). In East Anglia, the case for settlement nucleation around the eighth century is clear, although whether it was happening at the same time or slightly later in the East Midlands is disputed. Brown and Foard (1998; 2004) have argued that village formation was a two-stage process, with nucleation happening first, in the ‘Middle Saxon’ period, and a later restructuring of these settlements in the ‘Late Saxon’ period, which led to the creation of the village plans that have survived through to the present day (a model that also fits Wharram Percy: see above). Their assertion that ‘many Northamptonshire villages have produced evidence of Middle Saxon occupation’ (Brown and Foard 2004, 94) is a clear contradiction of Lewis et al.’s argument (1997) that the ‘village moment’ did not occur until the mid-ninth to eleventh centuries. In Whittlewood, Jones and Page (2006) also attribute the origins of villages to the late ninth and tenth centuries, suggesting that although all of the village cores that were subject to test-pitting produced pottery dating to somewhere between the fifth and the ninth centuries, the size of these ‘pre-village nuclei’ cannot account for the relocation of all the dispersed settlements that were abandoned. These pottery scatters of c.2–10ha could, however, have embraced several farmsteads, and the limitations of small-scale test-pitting also need to be acknowledged, particularly in a time when the use of pottery was not as extensive as in later periods, which leads to the distinct possibility that ‘Middle Saxon’ occupation could have been missed. In the first attempt to actually quantify the extent to which medieval villages have evidence for ‘Middle Saxon’ (i.e. late seventh- to early-ninth century) occupation, Wright (2010; 2013) has analysed all development-led archaeological investigation within currently occupied medieval villages in Norfolk, Cambridgeshire, Northamptonshire, Oxfordshire, and Wiltshire to reveal that a previously unsuspected proportion of these have evidence for ‘Middle Saxon’ occupation. In Northamptonshire, for example, of the twenty-three excavations within currently occupied medieval villages (COMVs), eight have produced stratified ‘Middle Saxon’ occupation, while a further five have

A Regional Approach to Studying Landscape

41

unstratified artefacts. Bearing in mind the limited extent of most of these excavations, the relatively ephemeral traces that ‘Middle Saxon’ occupation tends to leave due to the use of timber buildings, and the scarcity of distinctive material culture in use during this period, the figure of 57 per cent of COMVs having ‘Middle Saxon’ occupation must be regarded as very much a minimum figure. It is also remarkable how consistently high this figure is across other counties: in Cambridgeshire 45 per cent of the thirty-eight excavations within COMVs have ‘Middle Saxon’ occupation or artefacts, 49 per cent of thirtythree excavations in Norfolk, and 50 per cent of just twelve excavations in Wiltshire. In Oxfordshire, even the figure of 29 per cent of the forty-nine excavations having produced ‘Middle Saxon’ occupation is higher than one might have expected (Wright 2013, Table 8.1). Wright (2013, 37–39) also makes an important observation as to why the test-pitting within currently occupied medieval settlements carried out by the Higher Education Field Academy at the University of Cambridge has failed to find much ‘Middle Saxon’ material (e.g. Lewis 2010): when mapped on the First Edition Ordnance Survey Six Inch maps (OS 1st Edn maps), it can be seen that many of the test pits—although within the modern village—lay outside the historic settlement core. This discussion of settlement nucleation is important within the context of the Fields of Britannia study for two reasons. First, while it is well evidenced across England’s Central Zone and East Anglia, there is no indication that it was occurring in the South East, the South West, and the lowlands of western England: in these areas settlement patterns remained far more dispersed, and as such it is more likely that there was continuity within the wider physical fabric of the landscape. The second significance of this reorganization of landscapes in England’s Central Zone and East Anglia is that the numerous Romano-British settlements that lie abandoned beneath later open fields need not have been deserted in the fifth century: while it is clear that some sites were abandoned at this time, and certain new locations were occupied for the first time in the following two centuries, significant dislocation may not have occurred until as late as the eighth to tenth centuries.

The Mid-Ninth to the Mid-Eleventh Centuries Whether or not settlement nucleation began in the long eighth century or slightly later, a wide range of archaeological and documentary evidence shows that villages and open fields had spread across England’s Central Zone by the eleventh century, and that the tripartite division into the South Eastern, Central, and Northern & Western Provinces existed by that time (Rackham 1986; Williamson 1988; Lewis et al. 1997; Roberts and Wrathmell 2000a; 2002; Williamson 2003; Rippon 2008a). Since Gray’s (1915) seminal work on

42

The Fields of Britannia

English Field Systems (Figure 2.10A), scholars of the later medieval period have had a strong tradition of studying this regional variation in landscape character. Slater (1907, 47) has mapped Leland’s references to ‘champion’ and ‘enclosed’ countryside and these mirror very closely Rackham’s ‘Planned’ and ‘Ancient Countryside’ (Figure 2.10B–D), which in turn broadly correspond to the tripartite division within England mapped by Roberts and Wrathmell (2000a; 2000b; 2002) as the ‘South Eastern Province’, the ‘Central Province’, and the ‘Northern & Western Province’ (Figure 2.10D). In Rackham’s mind, the ‘Planned Countryside’ reflects the regular arrangement of fields and roads created through Parliamentary Enclosure of former open fields, although the term may arguably also apply to the process that created the open fields and their associated villages in the first place, which Brown and Foard (1998) have referred to as the ‘great replanning’ (a concept that is critically examined in Chapter 12). The mapping of settlement character by Roberts and Wrathmell is based upon nineteenth-century cartographic sources, as this is the first period for which we have maps of sufficient detail for the whole country, although the accounts of sixteenth-century and later topographical writers, such as John Leland (c.1542, see Toulmin Smith 1906), reveal how they were very aware of the differences in character between the open ‘champion’ countryside of central England, and the landscapes of enclosed fields and a more wooded (‘bosky’) appearance (due mostly to trees in hedgerows) in the South East, the South West, and West (as mapped by Slater 1907). It is also clear, however, that there was significant local variation in landscape character within these three broad ‘provinces’, and this was also recognized by early writers (e.g. for South West England see Rippon 2012b; for Suffolk and the Warwickshire ‘Felden’ and Arden’ see Palliser 1992, 189–90). It was French geographers who first referred to local areas with a distinctive character as pays, asserting that it was the effect of human activity over the passage of time which produced particular countrysides with their own personalities and distinctive ways of life (Muir 1999, 8). In Britain, the concept of pays has been employed by historians and historical geographers for several decades, with two distinct senses in which the term is used: particular locales with a unique character (such as the Black Country, New Forest, or Fenland), and generic types of landscape that occur in various different locations (such as downland, marshland, or moorland). Early research into pays in Britain was dominated by the work of the Leicester School of local historians (e.g. Everitt 1977; 1979; 1986; Fox 1989a; 1989b; Tranter et al. 1999) and the agrarian economist Joan Thirsk, who identified ‘farming regions’ in the countryside of sixteenth- and seventeenth-century England and Wales (Thirsk 1967; 1984; 2000; and see Dyer 1988, 12). Campbell’s (2000; 2008; Campbell and Bartley 2006) analysis of thirteenth- to fifteenth-century documents has also revealed the extent to which agricultural practices varied across later medieval England, while some diversity in landscape character can be mapped using data

A Regional Approach to Studying Landscape

43

C

A

0

0

150 km

150 km

Wales area with the two- and threefield system

ancient countryside planned countryside highland zone of England (including the South West) D

ce So u

0

150 km

cultivated and probably open champion

Leland’s route

0

nc vi

ra

ro

nt Ce

e

vi n lP ro

No

rth a nd West

er n Pr ov i nce

B

th

n er st Ea

P

150 km

Central Province

Fig. 2.10. Past mappings of medieval landscape character. (A) Gray’s (1915, frontispiece) extent of the Midland system; (B) areas that Leland described as ‘champion countryside’ in the mid-sixteenth century (mapped by Slater 1907, 47); (C) Rackham’s (1986, fig. 1) characterization; (D) Roberts and Wrathmell’s (2002, fig. 1.1) area of predominantly nucleated settlement, extended into Wales using Roberts 1987, fig. 1 (drawn by Mike Rouillard).

44

The Fields of Britannia

contained within Domesday Book, such as the density of population, ploughteams, meadow, and woodland (e.g. Darby 1952; 1962; 1977; Darby and Welldon Finn 1967). An analysis of the figures for the proportions of the domestic livestock (cattle, sheep, goats, and pigs) across the South West3 has also revealed some very stark differences in animal husbandry between individual districts (Rippon 2012b, 210–17). Before Domesday, however, there is very little documentary evidence for landscape character, although placenames do give an indication of very different amounts of woodland across England in the eleventh century (e.g. Roberts and Wrathmell 2000a, fig. 24).

MAPPING THE REGIONS AND PAYS OF FIRST-MILLENIUM A D BRITAIN

Regions The recognition that landscape character varies in both time and space is fundamental to this study, and the intention is to quantify how, in different regions, patterns of land-use may, or may not, have changed from the Roman through to the early medieval period, and how Romano-British field systems may, or may not, have survived over the same period. The quantification of data across the first millennium AD meant that a single series of regions had to be identified that were appropriate for both the Roman and early medieval periods. The underlying natural geology (both solid and drift), topography, and soils will have changed little over this time-span,4 with the exception of wetland areas. Although major wetlands form particularly distinctive pays in some parts of Britain, they are not included in this study as they were expensively flooded in the early medieval period, which accounts for the widespread abandonment of Romano-British settlements (Rippon 2000b; 2000c). Palaeoenvironmental sequences from wetlands were excluded for the same reason, as although some pollen would have blown in from the adjacent dryland areas, the majority of changes in the early medieval period reflect the increased local wetness. With regard to regional variation in the character of the cultural landscapes, simple mappings of settlement densities are highly misleading, as this can simply reflect the visibility of sites through aerial photography and the 3 Data on livestock is not recorded in the main ‘Exchequer’ Domesday, but was included in the ‘Exeter Domesday’, which was one of the regional returns used to compile the Exchequer version. 4 In the Fields of Britannia Project the characterization of solid and drift geology was based upon the British Geological Survey data available through Digimap, and published on the Geological Survey Ten Mile Map sheets (Scale 1:625,000). Soils data was derived from the Soil Survey of England and Wales Soils of England and Wales maps (Scale 1:250,000).

A Regional Approach to Studying Landscape

45

evaluations/excavations that are now required in advance of development (i.e. an archaeological bias). Attention here, therefore, focused on the character of the landscape reflected, for example, in whether there was a hierarchy of towns in addition to the civitas capitals (Figure 2.3), and the presence of villas (Figure 2.2) and Romano-Celtic temples (Figure 2.4A). An indication of the extent to which communities were integrated into the Roman economy is where pottery was produced locally and used extensively on domestic sites (Figure 2.4B) and the degree to which buildings were furnished with mosaics that, although some date to the first and second centuries AD, are mostly late Roman (Figure 2.4C). The distribution of mosaic schools, for example, is a reflection of how wealth and the desire to appear ‘Roman’ in third- and fourth-century Britain was greatest in central southern Britain and the East Midlands (Figure 2.4D). For the fifth to seventh centuries we can map the extent of Anglo-Saxon influence based on the presence of new forms of settlement, notably Grubenhäuser (Tipper 2004) and early Anglo-Saxon burials (i.e. those furnished with Germanic grave goods) of the fifth and sixth centuries (Figure 2.8). Another aspect of early medieval landscape character that can be mapped is the relative density of woodland based upon Roberts and Wrathmell’s (2000a, fig. 24) analysis of place-names and the Domesday Book evidence. There have been various characterizations of the later medieval landscape based on contemporary sources, such as Gray’s (1915) plotting of the documentary evidence for areas with open fields, as well as the provinces and sub-provinces identified by Roberts and Wrathmell (2000a) based upon nineteenth-century settlement patterns (Figure 2.10).

Pays In addition to identifying nine major regions, a series of smaller pays were defined both in terms of their cultural characteristics (based upon the data-sets outlined above) and their geology and topography. Occasionally the boundaries of these regions were easy to define, due to sharp divisions in the geology, topography, and soils, but often this was not the case, and academic judgement had to be used: that the character of two areas is different will often be clear, but where to draw a line between them is not (particularly in areas of flatter topography). It is also important to appreciate that a pays can include localized diversity in physical character: the Chalk Downland of Wessex, for example, includes both rolling plateaus and river valleys. There are also areas with relatively uniform topography—such as the claylands of East Anglia—but very different cultural landscapes reflected, for example, in the distributions of material culture/artefact styles from the Iron Age through to the early medieval period, including the extent of Romanization and the scale of Anglo-Saxon settlements and burials in the fifth century, that are very different

46

The Fields of Britannia

either side of the Gipping and Lark valleys in Suffolk (e.g. Rippon 2007; 2008a). To summarize the character-defining features of each pays, a table has been provided at the start of each regional chapter that summarizes the principal geologies that affect landscape character, whether it be the solid geology, overlying recent drift deposits, or a combination of both. These tables also include three broad characterizations that have been developed to allow the potential continuity of Romano-British field systems to be compared across broadly similar landscape types, such as chalk downland and the Boulder Clay (Tables 4.1, 5.1, 6.1, 7.1, 8.1, 9.1, 10.1, and 11.1), and which have been used in subsequent comparative analysis across the whole of Britain (see Chapter 3, Tables 3.3–3.6 and Chapter 12, Tables 12.1 and 12.2): • ‘predominant surface geology’: the topography and soils of an area are determined by a combination of the solid and drift geology. In some areas, the solid geology outcrops on the surface (which is particularly the case in southern Britain), whereas elsewhere it is masked by a capping of recent ‘drift’ deposits, such as glacial Boulder Clay and estuarine alluvium. All pays contain a variety of geologies, topographies, and soils, but in many cases they are characterized by one particular formation (hence the term used here, ‘predominant surface geology’). In other cases, a pays may in fact be characterized by its geological diversity, such as the limestone hills and clay vales of northern Somerset and southern Gloucestershire, and in such cases the ‘predominant surface geology’ is described as ‘mixed’. In many cases the ‘predominant surface geology’ is fairly obvious, such as the various areas of Boulder Clay that give rise to relatively heavy soils, but elsewhere the situation is more complex, such as the Oolitic limestone which underlies the North Yorkshire Moors, which might be expected to give rise to light, free-draining calcareous soils (as on the Cotswolds), but which are actually capped with heavy, acidic, peaty soils. The various pays are grouped into one of the following ‘predominant surface geologies’: Boulder Clay, Jurassic Clay, London Clay, other clays, chalk, limestone, sandstone, heathland, valley terraces, marshland, the highly mixed geologies of the South West (Dumnonia), other mixed geologies, and uplands. • relief: pays are categorized as lowland, intermediate, or upland: lowland pays are all below 180 m above ordnance datum (AOD); intermediate pays are mostly below 180 m AOD but have limited higher areas between 180 m and 600 m (e.g. the highest parts of the Cotswolds and Salisbury Plain, etc.); upland pays are wholly above 180 m, except for occasional valleys cutting through the high ground. • agricultural land classification, based upon The Ministry of Agriculture, Fisheries and Food’s (MAFF 1979) Agricultural Land Classification mapping at 1:625,000, and the Soil Survey of England and Wales’ Land Use Capability mapping at a scale of 1:1,000,000 (Mackney 1979). The five

A Regional Approach to Studying Landscape

47

categories of ‘good’, ‘good/intermediate’, ‘intermediate’, ‘intermediate/ poor’, and ‘poor’ in Tables 4.1, 5.1, 6.1, 7.1, 8.1, 9.1, 10.1, and 11.1 are directly related to the MAFF Grades 1–5 (1: excellent, 2: very good, 3: good/moderate, 4: poor, 5: very poor), and the Soil Survey’s seven classes of land capability, with its Class 4 and 5 equivalent to ‘intermediate/poor’ and Classes 6 and 7 to ‘poor’.

THE F IELDS O F BRITANNIA REGIONS OF LATE R O M A N AN D E A R L Y ME D I E V A L B R I T A I N From this review of regional variation in landscape character, it will be clear that between the Roman period and the eleventh century we can map a series of distinct regions. The extent to which some of the boundaries between areas of different character appear to have been stable is very striking (Lambourne 2010). Examples include the division between the ‘champion’ (i.e. open field) and the ‘woodland’ landscape in the West Midlands—between the Felden and the Arden—that can also be seen in the Roman period, as has been highlighted by Roberts and Wrathmell (2000b), while the south-western boundary of the ‘Central Province’, through the Blackdown Hills, can be traced back into prehistory (Rippon 2008a; 2012b). That there were long-standing divisions in landscape character within some of these three provinces is also now well established, for example along the Gipping and Lark valleys, which cut a diagonal line across Suffolk (Williamson 2006; Rippon 2007; Martin and Satchell 2008). Whilst acknowledging that some boundaries between landscape character areas will have changed over time, a study such as this—which is comparing change and continuity over the course of a millennium—has to use one set of regions throughout, and so the following have been identified as the major regions in Britain south of Hadrian’s Wall, based upon the character of the natural and cultural landscape, such as the extent of Romanization (reflected in whether there was an urban hierarchy, Romano-Celtic temples, mosaics, and pottery production), the scale of Anglo-Saxon immigrations, and the creation of villages and open fields in the later first millennium AD (Figure 2.11): • • • • • • • • •

South East East Anglia Central Zone South West Western Lowlands North East Lowlands Northern Uplands Lowland Wales Upland Wales

The Fields of Britannia

Northern Uplands

East Anglia

Western Lowlands

Ce n

tra

lZ

on

Upland Wales

North East Lowlands Lowlands

e

48

Lowland Wales South East South West

0

150 km

Fig. 2.11. The regions of Roman and early medieval Britain used within the Fields of Britannia Project (drawn by Chris Smart).

South East England The South East England region extends from Kent as far north as Essex and south western Suffolk, and down to the south-west as far as Dorset. Geologically (Figure 2.12) the region is dominated by a series of chalk uplands and clay lowlands, but also with extensive areas of sand and gravel. The northern edge of this region is defined by the chalk escarpment that passes along the northern edge of the Wiltshire Downs and the Chilterns, and it is separated

A Regional Approach to Studying Landscape

49

0

150 km

alluvium

Jurassic-Oolitic limestones

Ordovician, Silurian and Cambrian

sand and clays

Liassic, Triassic and Permiansandstones and clays

Intrusive

Cretaceous-chalk

Carboniferous

Jurassic and Cretaceous clays and sands

Old Red Sandstone and Devonian

Fig. 2.12. Simplified solid geology (drawn by Mike Rouillard).

50

The Fields of Britannia above 400 m 200–400 m 100–200 m 0–100 m

0

150 km

Fig. 2.13. Relief (drawn by Mike Rouillard).

from the East Anglia region by the Rivers Gipping and Lark. The region is mainly below 200 m AOD, the exceptions being parts of the North and South Downs, and the Wessex chalkland and Chiltern Hills (Figure 2.13). The region is drained by a series of rivers with their associated floodplains, although the Thames, with its associated gravel terraces, is the only one to be sufficiently extensive to form a pays in its own right. Culturally, the South East was a highly Romanized landscape, with a hierarchy of large and small towns associated with a network of roads and a large numbers of villas and Romano-Celtic temples. These were distributed across most areas, with the exception of the Weald, London Clay Basin (in

A Regional Approach to Studying Landscape

51

southern Essex and southern Hertfordshire), and the heathlands of South East Dorset and southern Hampshire, which are all characterized by their poor soils. Pottery use and production was widespread, and although there are mosaics at some sites, they are not as common as in the Central Zone. Although there is considerable evidence for Anglo-Saxon settlement from the fifth century, the distribution of furnished burials and Grubenhäuser is not evenly spread, with particular concentrations around the Lower Thames Estuary in northern Kent and southern Essex, and parts of Sussex. Culturally, the affinities of the immigrant communities appear to have been Saxon rather than Anglian (Lucy 2000, fig. 5.6). Some parts of this region, however, appear to be devoid of Anglo-Saxon settlement, including most of Hertfordshire, central and northern Essex, and southern Suffolk, suggesting this may have been a British enclave. The evidence of early medieval place-names and Domesday Book indicates that there was significant woodland cover across some of this region when compared to the adjacent Central Zone and East Anglia. During the later medieval period, the rural landscape was characterized by dispersed settlement and predominantly enclosed field systems, with open fields being very limited in extent. In terms of previous characterizations of the historic landscape, there are broad similarities with Roberts and Wrathmell’s ‘South East Province’ and Rackham’s ‘Ancient Countryside’, although the boundaries of the region as defined here do not correlate precisely, as factors such as the character of the Romano-British landscape have been taken into account.

The Central Zone This large region extends from the Blackdown Hills in Somerset, north through the Midland Plain, as far as the Yorkshire Wolds. Its southern edge follows the chalk escarpment of the Wiltshire Downs and the Chilterns (which divides it from the South East region), while to the north east lies Fenland and the North Sea. Its western boundary is marked by the Warwickshire Avon (that divides the open-field landscape of the Feldon to the south from the wellwooded Arden to the north), with the Peak District and Pennines to the north west, and the wetlands around the head of the Humber Estuary and scarp slope of Yorkshire Wolds to the north. This is a predominantly lowland region, the majority of which lies below 200 m AOD, with a mixed geology of limestones, clays, mudstones, and sandstones, and large areas capped by Boulder Clay; the uplands are mostly limestone and chalk (Mendip, the Cotswolds, and the Yorkshire Wolds). The region is drained by a number of large river systems including the Nene, Trent, and Warwickshire Avon, which are associated with extensive floodplain deposits.

52

The Fields of Britannia

This region was heavily Romanized and contained a hierarchy of large and small towns, associated with a network of roads, a series of Romano-Celtic temples, and the greatest density of villas in Roman Britain. Pottery use and production was widespread, as was the occurrence of mosaic pavements, which is indicative of the considerable economic prosperity that this region enjoyed in the late Roman period. The distribution of early ‘Anglo-Saxon’ burials (of both Anglian and Saxon traditions: Lucy 2000, figs 5.5 and 5.6) and Grubenhäuser suggests that there was Germanic influence across the east of the region (from Humberside to Warwickshire and Oxfordshire) from as early as the mid-fifth century, although much of Gloucestershire, Somerset, and Dorset remained British until the seventh century. Domesday and place-name evidence suggests that across most of this region woodland was relatively scarce, the exception being Gloucestershire, Somerset, and Dorset. In the later medieval period, this vast swathe of countryside was dominated by nucleated villages and open fields, and it broadly corresponds to Roberts and Wrathmell’s ‘Central Province’ and Rackham’s ‘Planned Countryside’.

East Anglia The East Anglian region identified here covers the whole of the historic county of Norfolk and the northern part of Suffolk, and as such corresponds very closely to the early medieval kingdom of the same name. The region is bordered by Fenland to the west and the South East England region to the south (the boundary being the Gipping and Lark valleys). Physically, the region is characterized by a relatively flat terrain, with the underlying bedrock mostly capped by Boulder Clay and sandy drift deposits, all under 200 m. Most of the region is drained by the rivers Bure, Wensum, Yare, and Waveney, which flow east into the wetlands of the Norfolk Broads. The area was only moderately Romanized, with just a small number of modest villas, mostly in the west of the region, of which only Gayton Thorpe has produced evidence for tessellated pavements (Bowden 2013, 54). In addition to the civitas capital of Venta Icenorum (Caistor-by-Norwich), there was just one walled small town (Brampton), and while there were a number of other roadside settlements that probably served as market centres, this region lacked the large numbers of small towns seen in the South East and Central Zone. Pottery production and use was, however, widespread. The distribution of early ‘Anglo-Saxon’ burials and Grubenhäuser indicate extensive Germanic influence from as early as the mid-fifth century. The associated artefacts are broadly Anglian in affinity, and in contrast to the South East region, East Anglia is characterized by large, communal cremation cemeteries that would have served large folk territories (Williamson 1993, 53–4, 65–7; Lucy 2000, fig. 5.5; Hills and Lucy 2013). In the later medieval period, settlement was

A Regional Approach to Studying Landscape

53

relatively dispersed, reflected in this being included in Roberts and Wrathmell’s ‘South Eastern Province’, although this hides the region’s greater similarity with the Central Zone in the early medieval period: during the eighth century there is evidence for settlement nucleation which, unlike in the Midlands, did not last long, as from around the tenth century settlement tended to drift away from the early village foci towards greens and commons. This complex history and its impact on the later landscape reflects Rackham’s (1986, fig. 1.3) division of East Anglia between his Planned and Ancient Countryside. The density of woodland at the time of Domesday appears to have been low.

The South West The South West region includes Cornwall, Devon, and the far west of Somerset. The boundary between the South West and the Central Zone runs through the Quantock and Blackdown Hills (Rippon 2008a; 2012b). The region has a varied geology, comprising sandstones, limestones, and mudstones, but is probably better known for its series of granite uplands (e.g. Dartmoor, Bodmin Moor, and West Penwith). The presence of these uplands, as well as Exmoor, the Quantocks, and the Blackdown Hills (each all extend beyond 180 m to a maximum of 621 m AOD on Dartmoor), have led to a popular view that the whole of the South West is an upland region (e.g. Fox 1932; Rackham 1986, fig. 1.3), although roughly two-thirds of it actually consists of rolling countryside below 180 m AOD, which comprises fertile agricultural land that was as densely settled in Domesday as the neighbouring parts of the ‘Central Zone’. This was, however, a relatively unRomanized region, with just a single town—Isca Dumnoniorum (Exeter), the civitas capital—and a very small number of villas, all of which lie in the far east of the region (the site at Camborne, in Cornwall, probably being associated with the working of tin, rather than being a true villa in the sense of the head of an agricultural estate). Apart from a small group of villas on the Blackdown Hills (on the boundary between the South West and the Central Zone), the only mosaics are from Exeter and Camborne, and pottery production and use was on a far more limited scale compared to the adjacent Central Zone (some sites, such as Littlehempston in Devon, even appear to have been aceramic: Reed and Turton 2005). There is no evidence in the South West for early ‘AngloSaxon’ burials or Grubenhäuser. Domesday and place-names suggest that this was a landscape with a moderate amount of woodland. During the later medieval period, the landscape was characterized by a predominantly dispersed settlement pattern associated with a mixture of small open fields and extensive areas of closes held in severalty. The South West region matches closely Roberts and Wrathmell’s ‘South West Peninsula’ sub-Province within

54

The Fields of Britannia

their ‘Northern & Western Province’, and while Rackham (1986, fig. 1.3) clearly recognizes that it is different in character to areas further east, he is wrong to regard it as part of the Highland Zone.

Western Lowlands The Western Lowlands are bounded by the Welsh Uplands to the west, the Lake District to the north, the Pennines to the north east, and the Warwickshire Avon to the south east. The gently undulating topography mostly lies below 100 m AOD, but is punctuated by discrete upland areas such as Cannock Chase and the Wrekin, which rise to 240 m and 400 m, respectively. The geology across this region is dominated by mudstones and sandstones with overlying glacial sand, gravel, and till. During the Roman period this landscape was not as Romanized as the Central Zone. Although there was a well-developed urban hierarchy, including a series of small towns associated with the small number of villas, relatively few buildings were furnished with mosaics, and pottery production and use was relatively limited. Like the Dumnonii in the South West peninsula (see above), the Cornovii appear to have been unreceptive to imported material culture, or were otherwise not strongly engaged with the Roman way of life, and so may have had other ways of demonstrating wealth and status (Sargent 2002, 225; Gaffney et al. 2007, 279–80). There is evidence for fifth- to seventh-century Anglo-Saxon immigration in just the far eastern fringes of the zone (e.g. Wasperton). Domesday and place-name evidence suggests a relatively well-wooded landscape. The later medieval countryside was characterized by a predominantly dispersed settlement pattern associated with mostly enclosed fields. This region falls within Rackham’s ‘Ancient Countryside’ and Roberts and Wrathmell’s ‘Northern & Western Province’.

North East Lowlands The North East Lowlands lie between the Pennines and the East Coast, and are bounded by the Humberhead wetlands to the south, and the Tyne Estuary/ Hadrian’s Wall to the north. The solid geology of this region is varied and includes sandstone, limestone, mudstone, and clays, extensively capped with Boulder Clay. Much of this region lies below 60 m AOD, although it extends to 180 m AOD along the eastern fringe of the Pennines, and over 200 m in the North Yorkshire Moors (a relatively isolated upland block, whose character is very different to the rest of this region). This was a moderately Romanized and militarized landscape, with a well-developed urban hierarchy, including a series of small towns closely associated with forts and the road network, and

A Regional Approach to Studying Landscape

55

a series of villas in the lower-lying areas. There was extensive pottery production and use, and mosaics on a number of sites. The distribution of early medieval burials (of an Anglian tradition: Lucy 2000, fig. 5.5) and Grubenhäuser shows that Germanic immigration appears to have begun in the south during the fifth century, becoming more extensive by the seventh century. Domesday and place-name evidence indicate that in common with the Central Zone to the south, this was a relatively open landscape with relatively little woodland. In the later medieval period the region was characterized by nucleated villages and open fields. The North East Lowlands corresponds very closely to the northern part of Roberts and Wrathmell’s ‘Central Province’, although Rackham (1986, fig. 1.3) wrongly regards this area as part of the Highland Zone.

Northern Uplands The Northern Uplands extend from the Derbyshire Peak District northwards through the Pennines and the Lake District. The defining character of this region is of rugged, expansive uplands, which form the most mountainous parts of England. Geologically, the region is composed of sandstones and limestones, alongside igneous rocks. The overlying superficial deposits are dominated by extensive peat deposits. This was a relatively unRomanized landscape, with the only evidence for fifth- to seventh-century Anglo-Saxon settlement being a cluster of burials in Derbyshire. It was relatively well wooded in the early medieval period, and during the later medieval period was characterized by a dispersed settlement pattern with enclosed fields, although large areas of the higher ground were unenclosed moorland. The region forms part of Roberts and Wrathmell’s ‘Northern & Western Province’, and Rackham’s (1986, fig. 1.3) ‘Highland Zone’.

Lowland Wales Lowland Wales is physically and culturally very distinct from the upland landscapes to the north. This region consists of a narrow belt of land, which stretches from the River Wye in the east to Carmarthen in the west. Physically, it lies below 180 m AOD, with geology consisting of sandstone, limestone, and coal measures. Although the Roman military extended their reach across the region, including having a legionary fortress at Caerleon, it became moderately Romanized, with a small number of villas and small towns alongside the civitas capitals at Caerwent and Carmarthen (Arnold and Davies 2000, figs 7.1 and 7.3). There is no evidence for fifth- to seventh-century Anglo-Saxon colonization, and the later medieval settlement pattern is characterized by a

56

The Fields of Britannia

mixture of small hamlets with enclosed field systems alongside villages and open fields created during the Anglo-Norman colonization in the late eleventh and twelfth centuries (Sylvester 1969; Rippon 2008a). Wales fell outside of the area considered by Roberts and Wrathmell (2000a) in their analysis of the rural landscape of England, but is included in Roberts’ (1987, fig. 1.1) area of nucleated settlement.

Upland Wales The Welsh Uplands cover most of the principality and are characterized by their rugged coastline, mountains, and moors, and would have been one of the most inhospitable landscapes in Roman Britain. It does, however, also include some lower-lying areas such as Anglesey and the north west coastal plain. In the Roman period settlement was focused in the river valleys, with the uplands only sparsely occupied (Arnold and Davies 2000, 72–3), and although native forms of settlement predominated, the recent discovery of a late-third- to early-fourth-century villa near Aberystwyth suggests that there was greater Romanization amongst the social elite than was previously suspected (Driver and Davies 2012). During the later medieval period the landscape was characterized by dispersed farms and hamlets associated with predominantly enclosed fields, although large areas of upland were open moor. This area lies outside Roberts and Wrathmell’s study, but it lies within Rackham’s (1986, fig. 1.3) ‘Highland Zone’.

3 A Landscape Approach to the Roman–Medieval Transition This chapter explores three landscape-based approaches to studying what happened to the countryside of late Roman Britain: two are related to studying regional variations in the patterns of land-use—pollen sequences and palaeoeconomic data (animal bones and charred cereals)—and the other is concerned with the relationship between Roman and medieval field systems. The sources and methods are discussed, and the results described on a national scale by way of context for the subsequent region-by-region reviews in Chapters 4–11.

RECONSTRUCTING P ATTERNS O F LAND-USE

Palaeoenvironmental Sequences The reconstruction of past patterns of land-use through examining pollen preserved within sediment sequences is now well established within archaeology, although more so for prehistory than for the Roman and medieval periods, and in upland areas more than in the lowlands (Dark, P. 2000).1 The limited attention that is paid to palaeoenvironmental evidence in some studies of the historic period (e.g. Harrington and Welch 2014, 56) suggests that there is a long way to go before its use is regarded as mainstream in this period. The problem with the upland sequences is that they lay beyond core settlement areas, but in recent years a combination of development-led archaeology and research projects has led to an increased number of sequences from lowland areas that cover the Roman and medieval periods (for example, through the 1

Key early study areas included the Lake District (Dickinson 1975), the North York Moors (Bartley, 1975; Atherden, 1976), the Pennines (Conway 1954; Tallis, 1964), the Brecon Beacons (Chambers 1982), Snowdonia (Hibbert and Switsur 1976; Bostock 1980), Exmoor (Merryfield and Moore 1974; Beckett and Hibbert 1979; Francis and Slater, 1990), Dartmoor, (Caseldine and Hatton 1993) and Bodmin Moor (Brown, 1977).

58

The Fields of Britannia

sampling of small valley mires and palaeochannels).2 The aim of the Fields of Britannia Project was to analyse all these pollen sequences as far north as the pays through which Hadrian’s Wall runs, which embrace all or part of the Roman and early medieval periods, in order to compare the proportions of pollen from the four major categories of land-use (the species indicative of each land-use type are listed in Appendix II): 1. ‘Woodland’, which includes shrubs and hedge species; 2. ‘Arable’, which includes both cereal crops themselves and also weed species indicative of cultivated ground; 3. ‘Improved pasture’; 4. ‘Unimproved pasture’. For the purposes of this discussion, the early medieval period was subdivided into three shorter periods: the fifth century, the sixth to mid-ninth century, and the mid-ninth to mid-eleventh century. Sequences from wetlands were excluded unless they were within 1 km of the fen-edge. In preparing averaged figures for pays, regions, and the uplands and lowlands (Tables 3.1–3.2, 4.2, 5.2, etc.), the individual figures from each pollen sequence were averaged. Pollen gives little indication of the types of animal husbandry being practiced, or the cereals being grown, and so a separate analysis of excavated faunal and charred cereal assemblages in the Roman and earliest medieval periods was also undertaken (see below, and Rippon et al. 2014). Data was collected through a search of the published literature and the Environmental Archaeology Bibliography. Pollen sequences were only included in the statistical analysis if securely dated to the Roman and/or the early medieval periods through radiocarbon dating or stratified artefact assemblages (for on-site pollen sampling from features such as ditches and wells). Undated pollen sequences that appear to cover all or part of the Roman and/or early medieval periods on the basis of interpolated radiocarbon dates are sometimes referred to in the discussion but were not used in the quantitative analysis. A total of 194 dated sequences were identified for this quantitative analysis (Figures 3.1 and 3.2, and Appendices I and II): there is still a bias towards upland areas, and although the lowlands are now far better represented than was the case when Petra Dark (2000) last reviewed the evidence, the sites are still very unevenly spread, and some pays have little or no material. 2 Key study areas include the valleys of the Great Ouse (French and Heathcote, 2003), Nene (Brown and Keough 1992a; 1992b; French et al. 1992; Brown 2009), Soar (Brown and Keough 1992b; Smith et al. 2005), Thames (Robinson and Lambrick 1984; Parker and Robinson 2003), Trent (van de Noort and Ellis 1998; Brown 1999a), Severn (Shotton 1978; Taylor and Lewin 1996), Exe (Fyfe et al. 2003) and low-lying fenland, including the Humber Levels (van de Noort and Ellis 1999), the Wash, (Waller 1994; Hall and Coles 1994) and the Somerset Levels (Rippon 2000b; 2006).

A Landscape Approach to the Roman–Medieval Transition

59

pollen sequence included in statistics other pollen sequence

0

150 km

Fig. 3.1. The distribution of all pollen sites used within the Fields of Britannia Project, showing how they are spread across all regions (drawn by Chris Smart).

It is important to appreciate that the percentages of major land-use types represent the ‘Total Land Pollen’ (TLP) of species indicative of that type, and that these percentages do not equate directly with the proportion of the landscape that these land-uses covered, as different plant species produce very different amounts of pollen (Sugita et al. 1999; Fyfe 2006; Broström et al. 2008). Grasses, for example, produce larger amounts of pollen than many herbaceous species, and as such will be over-represented in pollen sequences. In contrast, some arable crops (notably barley and wheat) will be under-represented as they produce so little pollen (barley and wheat are autogamous, i.e. self-fertile, in contrast to rye, which is wind pollinated),

60

The Fields of Britannia AD 410–500

AD 43–410

0

150 km

0

AD 500–850

0

150 km

150 km

AD 850–1066

0

150 km

Fig. 3.2. The distribution of pollen sites for the four periods of analysis in the Fields of Britannia Project (drawn by Chris Smart).

although the categorization of plants to the ‘arable’ land-use group also includes some species of weeds that only grow in disturbed (i.e. ploughed) ground. One indication of this under-representation of arable-indicative pollen is to compare the data with the land-use recorded in Domesday Book. There have been various attempts to calculate the area of England that was arable in 1086 (Wales was not covered in that survey), which is far from a simple task, and although Lennard (1959, 393) suggested 7.2 million acres (22 per cent of England’s 32.3 million acres), the most recent estimate is c.5.9 million acres (18 per cent of England’s 32.3 million acres; Broadberry et al.

A Landscape Approach to the Roman–Medieval Transition

61

2011, 18). In contrast, the TLP in lowland pollen sequences dated to 850–1066 is just 5 per cent (Table 3.2). Another issue in interpreting the data is that although many tree species are prolific pollen producers, this is only when they reach maturity (Hicks 2006), with coppiced or pollarded woodland and regularly laid hedges therefore being invisible in the pollen record. Another potential problem is that any one local pollen sequence may be untypical because of its particular location and localized pollen catchment area, but this is overcome by aggregating the data from a series of separate sites from each pays and region. Overall, it is impossible at present to reconstruct the proportions of the landscape covered by different types of land-use based upon the percentages of pollen of different plant species in palaeoenvironmental sequences, but we can at least look at broad regional variations in the proportions of pollen from these four land-use groups in time and space and so demonstrate that woodland, for example, was more common in one region or one period than another.3 The data are presented in three ways. First, for each region, Chapters 4–11 contain a narrative account that discusses the main characteristics of the landuse relative to the other upland and lowland regions in the Roman and early medieval periods. In addition to the quantifiable data from pollen sequences (summarized for each region in Table 3.1, and aggregated across all the uplands and lowlands in Table 3.2), use is made in these narratives of other off-site environmental indicators such as land mollusca (snails) from chalkland valley sediment sequences. This narrative is informed by statistical analysis presented in two ways: first a series of pie charts that compare the percentage of pollen from each of the four major land-use categories in the Roman and early medieval periods (Figure 3.3, based upon the data in Table 3.1), and secondly a set of histograms that show how these percentages changed in each region for the Roman period, the fifth century, the sixth to mid-ninth century, and the mid-ninth to mid-eleventh century (Figure 3.4).

Palaeoeconomic Data Whilst the systematic analysis of pollen sequences provides insights into regional variations in the possible relative significance of four major types of 3 Some variation over time might represent changes in land-use unrelated to a particular land-use group. For example, a shift from heather-dominated rough grazing to improved grassland will lead to an increase in the percentage of tree pollen, even though the actual number of trees in the landscape remains constant: as heather produces large amounts of pollen, when it is replaced by grass, that produces less pollen, there is less pollen in the air and so the tree pollen increases as a proportion of the TLP even though the absolute amount remains unchanged (Fyfe 2006).

62 Table 3.1. The proportions of pollen derived from the four major land-use types in the Roman and early medieval periods for each of the Fields of Britannia regions Roman and early medieval land-use expressed as % key pollen types Region

Period

Woodland

n %

Lowland Wales Northern Uplands South West North East Lowlands Western Lowlands East Anglia Central Zone South East

Roman early medieval Roman early medieval Roman early medieval Roman early medieval Roman early medieval Roman early medieval Roman early medieval Roman early medieval Roman early medieval

31 31 2 1 27 32 15 23 6 9 26 25 3 4 26 17 16 20

32 36 36 40 39 40 27 32 24 25 33 34 16 16 14 16 31 37

+4 +4 +1 +5 +1 +1 0 +2 +6

% 0.5 0.6 3 0 1 1 1 3 2 3 3 4 2 9 7 6 6 5

Improved pasture

change + 0.1 3 0 +2 +1 +1 +7 1 1

% 44 39 39 46 36 32 53 42 48 42 41 37 68 61 58 51 45 37

change 5 +7 4 11 6 4 7 7 8

Unimproved pasture % 23 24 22 14 24 27 19 23 27 29 24 25 14 14 21 27 18 22

change +1 8 +3 +4 +2 +1 0 +6 +4

The Fields of Britannia

Upland Wales

change

Arable

A Landscape Approach to the Roman–Medieval Transition

63

Table 3.2. The proportions of pollen derived from the four major land-use types in the Roman period, transition (fifth-century) period, sixth to mid-ninth, and mid-ninth to mid-eleventh centuries for upland and lowland regions Roman AD

n Woodland Lowland Upland

94 58

Arable Lowland Upland

94 58

Early medieval

43–410 % TLP 26 35 4 0.8

AD

n 43 41 43 41

411–499 % TLP 33 35 4 0.6

AD

n 70 51 70 51

500–849 % TLP 29 38 4 0.9

AD

n 42 20 42 20

850–1066 % TLP 28 42 5 0.9

Improved pasture Lowland 94 Upland 58

49 40

43 41

40 37

70 51

43 36

42 20

43 37

Unimproved pasture Lowland 94 Upland 58

21 24

43 41

23 28

70 51

24 26

42 20

24 21

land-use—arable, improved pasture, unimproved pasture, and woodland— they tell us nothing about the types of crops that were growing in the fields, or the nature of the animals grazing the pasture. For this we must turn to the evidence of plant macrofossils and animal bones that are found on many excavated settlements. In the past, however, comparative analyses of animal bones and charred cereals from the Roman and medieval periods have focused on variations between sites of different social and economic status, for example towns, rural settlements, and military sites (e.g. Grant 1988; King 1989; 1991; 1999; Albarella and Davis 1994; Albarella 2005; Van der Veen 2008; Van der Veen et al. 2007; 2008; 2013), with almost no attention paid to potential variations in farming practice between different regions. A recent edited volume on Food in Medieval England (Woolgar et al. 2006a) provides a good example of this with a series of papers studying how patterns in the consumption of cereals, meat, and other foodstuffs varied on sites of different social status, and how these patterns changed over time, but with only passing reference to the possibility that there may have been regional variation. Indeed, the comment ‘There are some indications of variations in diet on a regional basis, but it is less marked in these studies [in Food in Medieval England] than might have been expected’ (Woolgar et al. 2006b, 274) is a reflection of how the data has not been systematically analysed in this way. The failure of archaeologists to explore regional differences in agricultural practice is all the more surprising as historians have been doing it for decades, for example through Thirsk’s (1967; 1984) identification of

64

The Fields of Britannia Northern Uplands

North East Lowlands

Western Lowlands Roman n = 27

Roman n = 26

early medieval n = 32

Roman n=6

early medieval n = 25

East Anglia

Upland Wales

Roman n = 31

Roman n=3

early medieval n = 31

Lowland Wales

Roman n=2

early medieval n=9

early medieval n=4

South East

early medieval n=1

South West

Roman n = 16

Central Zone

early medieval n = 20

woodland arable improved pasture

Roman n = 15

early medieval n = 23

Roman n = 26

early medieval n = 17

unimproved pasture

Fig. 3.3. The changing proportions of pollen derived from the four major land-use categories—woodland, arable, improved pasture, and unimproved pasture—during the Roman and medieval periods (drawn by Chris Smart).

distinctive ‘farming regions’ in the sixteenth and seventeenth centuries, and Campbell’s (2000; 2008; and Campbell and Bartley 2006) demonstration of marked differences in farming practice on thirteenth- to fifteenth-century demesnes. An initial review of the Romano-British and medieval animal bones and charred cereals from five historic counties in the south west of Britain (Cornwall, Devon, Dorset, Somerset, and Wiltshire), however, has clearly shown that there were significant differences in arable cultivation and animal husbandry on different geologies (Rippon 2012b), while subsequent work also revealed variations within Essex (Rippon 2013). The pays used in these two studies were defined by both their natural (geology, topography, and soils) and

A Landscape Approach to the Roman–Medieval Transition

65

cultural landscapes (e.g. the degree of Romanization), and the conclusions were that both environmental and cultural factors were significant in shaping framing practice. Hesse (2011), who recently reviewed the animal bones from the Thames Valley in the context of Britain and mainland North West Europe, came to a similar conclusion that ‘while environmental factors and personal/ cultural tastes set the parameters for what forms of animal husbandry were feasible, economic concerns played a critical role in determining how this translated into reality’. It was beyond the scope of the Fields of Britannia Project to study all of the Romano-British and early medieval animal bone and charred cereal assemblages from south of Hadrian’s Wall, and so first Rippon’s (2012b; 2013) existing data (from Cornwall, Devon, Dorset, Somerset, Wiltshire in the far South West, and Essex in the South East) was updated in the light of more recent work and disaggregated into early and late Roman. Secondly, these areas were linked through further data collection in Gloucestershire, Oxfordshire, Northamptonshire, Buckinghamshire, Middlesex, Hertfordshire, Bedfordshire, Huntingdonshire, Cambridgeshire, Norfolk, and Suffolk, so forming a transect from the South West, through the South and East Midlands, and across into East Anglia and the South East. The transect includes a wide variety of geology and soil types, including chalk downland, limestone hills, clay vales, Boulder Clay Plateaus, and sandy heathland (Figure 3.5). This data are presented in Tables 3.4–3.6 and Appendices III and IV.4 A comparison of the early and later medieval periods is published elsewhere (Rippon et al. 2014).

Methodological Considerations For animal bones, analysis was restricted to cattle, sheep/goat (S/G, as their bones cannot normally be distinguished), and pig, referred to below as ‘major domesticates’, that together formed the mainstay of farming economies and contributed the vast majority of meat in the diet: while some wild animals, such as deer and domesticated fowl, were consumed during the Roman period, the numbers of bones are so small that they clearly comprised an insignificant part of the diet (e.g. on the sites excavated in and around Stansted Airport and the A120 on the Boulder Clay of North West Essex, 3,654 bones were cattle, 3,005 sheep/goat, and 2,132 pigs, but just 39 deer and 20 domestic fowl: Havis and Brooks 2004; Timby et al. 2007a; Cooke et al. 2008). As the vast majority of animal bone reports only provide a simple fragment count or Number of Identified Specimens (NISP) that data is used here (as opposed to the 4 In all the counties studied, quantified animal bone and charred cereal data were collated from published reports (in county journals and in monographs), and grey literature on the ADS. The authors are extremely grateful to the Roman Rural Settlement Project at the University of Reading for access to their database.

66

The Fields of Britannia WOODLAND

70%

70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

10%

0

0 South East

East Anglia

Central Zone

AD 43–410

South Western North Lowland Northern Upland West Lowlands East Wales Uplands Wales Lowlands AD 410–500

AD 500–850

AD 850–1066

ARABLE

70%

70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

10%

0

0 South East

East Anglia

Central Zone

AD 43–410

South Western North Lowland Northern Upland West Lowlands East Wales Uplands Wales Lowlands AD 410–500

AD 500–850

AD 850–1066

Fig. 3.4. The changing proportions of woodland, arable; improved pasture, and unimproved pasture pollen in each region across the four periods of analysis (drawn by Chris Smart).

A Landscape Approach to the Roman–Medieval Transition IMPROVED PASTURE

70%

67 70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

10% 0

0 South East

East Anglia

Central Zone

AD 43–410

South Western North Lowland Northern Upland West Lowlands East Wales Uplands Wales Lowlands AD 410–500

AD 500–850

AD 850–1066

UNIMPROVED PASTURE

70%

70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

10% 0

0 South East

East Anglia

Central Zone

AD 43–410

Fig. 3.4. Continued.

South Western North Lowland Northern Upland West Lowlands East Wales Uplands Wales Lowlands AD 410–500

AD 500–850

AD 850–1066

68

The Fields of Britannia

A

area of data collection

Norfolk

Gloucestershire

e hir ms ha ing ck Bu

Bedfordshire

ire sh on Cambridgeshire pt am rth No

Oxfordshire

r He

ire sh rd tfo

Suffolk

Essex

Wiltshire Somerset Devon Dorset

Cornwall

0

150 km

Fig. 3.5. The study area used for the analysis of animal bones and charred cereals. (A) the counties studied; (B) the geological formations studied, drawn by Stephen Rippon).

Minimum Number of Individuals (MNI) method), although it should be borne in mind that a fragment count tends to produce a slight bias towards cattle (as larger animals have larger bones that tend to fragment into more identifiable pieces). For reasons such as this it must be stressed that the percentages of cattle, sheep/goat, and pig bones do not simply equate to the different proportions of animals kept, although as these taphonomic factors are constant on all sites, we can still compare the relative proportions of cattle, sheep/goat, and pig bones across different areas. Another factor to bear in mind is that animals kept for their meat, and which are therefore killed young,

A Landscape Approach to the Roman–Medieval Transition

69

Boulder Clay

B

London Clay Jurassic Clay Chalk Limestone valley terraces and heathland marshland lowland Dumnonia other

0

150 km

Fig. 3.5. Continued.

will be over-represented compared to livestock that was kept into maturity for products such as milk and wool, and further work is needed on age-at-death to determine how far this varied across different geologies. The research presented in this chapter is, therefore, very much a ‘big picture’ study which aims to explore whether there are broad differences between different surface geologies, rather than a detailed investigation of animal husbandry at individual sites. Of the arable crops, wheat, barley, oats, and rye have been included, as these represent the vast majority of cereal crops grown. In the Roman and early medieval periods legumes were not significant: on the seven Romano-British

70

The Fields of Britannia

settlements excavated at the Cambourne New Settlement, on the Cambridgeshire Boulder Clay, for example, 933 cereal grains were recovered but just 13 celtic beans (Vicia fabia) and 22 garden peas (Pisum sativum) (Wright et al. 2009). Similarly, the late Roman farmstead at Kenn Moor, in Somerset, produced 570 cereal grains from non-corn-drier contexts and just 16 bean seeds (Rippon 2000b), while at fifth- to early eighth-century5 West Stow, for example, 1,056 charred cereal grains were identified but just a single pea-sized leguminous seed (West 1985, 104). Some variability in crop husbandry is to be expected, as the major cereals have somewhat different growing preferences (Hillman 1981; Jones 1981; Kerridge 1992; Van der Veen 1992; Campbell 2000). Spelt wheat (Triticum spelta), which was by far the dominant variety grown in the Roman period, is demanding in terms of nutrients but can grow on light or heavy soils. In contrast, bread wheat (Triticum aestivum), which became common in the early medieval period, is the most demanding cereal in terms of its nutrient requirements, and prefers moist loamy/silty/clayey soils (emmer wheat, which prefers light soils and drier conditions, was not extensively grown in these periods). Barley (Hordeum sativum) grows across a wide range of soils, and does not require as much nitrogen as wheat, and so was often grown on lighter soils. Rye (Secale cereale) can grow in various conditions except on heavy soils but tolerates high acidity, some drought, and low fertility (it requires less nitrogen than wheat) and so was also particularly suited to light or poor soils. Oats (Avena sativa) can also tolerate difficult conditions including high acidity and low fertility, and in particular can cope with heavy soils and high rainfall. Although it is difficult to determine whether charred oat grains were wild or cultivated (as the grains are indistinguishable), this problem is the same in all regions and so if more oats are present in one area compared to another then the assumption has to be that more oats were being grown (it being unlikely such marked differences are due to the fields in one area being more weed infested than in others). There is also good documentary evidence for the greater significance of oats in areas such as the South West, where they are well represented in the archaeological record (Fox 1991). Charred grains from corn driers have been excluded, as these will often reflect the processing of a single crop, which as such may not be representative of cropping patterns over several years. On the Boulder Clays of the South East Midlands in the early Roman period, for example, of the 1,860 grains recovered from three corn driers6 99.7 per cent were wheat, compared to the average of 81 per cent wheat from non-corn-drier contexts (Tables 3.5 and 3.6). This bias towards wheat is borne out when the grains recorded in all corn driers are compared with those from non-corn-drier contexts (Table 3.5), 5

Based on the revised dating of Ipswich Ware: Blinkhorn 2012. Haynes Park, Beds: Luke and Shotliff 2004; Renhold Water End, Beds: Timby et al. 2007b; Wavendon Gate, Walton, Beds: Williams et al. 1996. 6

A Landscape Approach to the Roman–Medieval Transition

71

although occasionally other crops are preserved, such as at Figheldean, on the chalk downland of Wiltshire, where barley comprised 88 per cent of the 280 grains (Graham and Newman 1993). Comparing the relative significance of cereals in the Roman and early medieval periods is problematic as glume varieties (e.g. spelt wheat) will be better represented due to the lengthy processes that are required to extract the grain from the glume. These processes, such as parching and pounding, will mostly have occurred within settlements, in contrast to free-threshing varieties (such as bread wheat) that may have been processed in the field, which will lead to them being under-represented on settlement excavations. Another factor to consider when interpreting cereal assemblages is that the survival of grains in the archaeological record will in part be determined by what they were used for, for example whether they were being grown for human and/or animal consumption (as cereals are more likely to have been charred if they were being threshed and cleaned for human use, although small amounts of animal bedding and fodder may also have got burnt). A final factor in the interpretation of charred cereal assemblages is the possibility that some of the crops may not have been raised locally, but have been purchased at a nearby market town whose hinterland contained a greater diversity of surface geologies. Once again, these are factors that can hopefully be considered in more detail in future work, although the purchase at market of cereals grown outside the locality is likely to reduce, not increase, the diversity within on-site assemblages, and so the differences in the charred cereal assemblages from particular geologies presented here are likely to reflect the minimum extent of variation in farming practice. Some important temporal and regional patterns emerged in the dataset assembled for this study. It was only in the 1970s that the systematic collection of palaeoenvironmental data began, for example through wet sieving, although practices have been far from consistent. Delays in publishing old excavations mean that even reports published in the 1980s and 1990s can be devoid of charred cereals in particular, due to the lack of sampling procedures on the original excavations. Until the late 1980s it was often only larger-scale developments that were preceded by archaeological work, although from the early 1990s implementation of the then Planning Policy Guidance Note 16: Archaeology and Planning (PPG16) and its successors has ensured that the vast majority of developments that threaten archaeological deposits are subject to proper investigation. This has resulted in far more archaeological work being carried out, with around £7m funding per year in 1990 and over £100m a year today (Thomas 2013, 12). Although a large amount of work remains unpublished, a growing body of grey literature is now available through the ADS and the websites of the major archaeological units.7 This development-led 7

ADS that includes reports digitized as part of the Roman Rural Settlement Project at Cotswold Archaeology/Reading University (); Albarella

72

The Fields of Britannia

archaeology is not, however, evenly spread across Britain, with certain areas having seen a large amount of published work (e.g. North West Essex and central Bedfordshire) while other areas have seen relatively little. Increasingly, however, development-led archaeology is leading to fieldwork in areas that until recently had produced little or no palaeoenvironmental data, such as the lowlands of Devon (Mudd and Joyce 2014). The South West also demonstrates another factor that produces biases in the available data: the fact that animal bones are not preserved in acidic soils. The assemblages included in this study have been attributed to one of five site types based upon their social status:     

rural settlements; villas; major towns; small towns; elite sites (specifically in the early medieval period).

Temple/ritual sites were excluded, as the pattern of animal husbandry was so different to domestic sites: on the nine sites within the study area, of the 94,170 bones cattle comprised just 15 per cent, sheep/goat 82 per cent, and pig 3 per cent (with sheep/goat 85 per cent in the early Roman period and 75 per cent in the late Roman period).8 The forty-two villa sites in the study area conform to earlier studies (e.g. King 1989; 1991; 1999) in showing a higher proportion of pig and a lower proportion of sheep/goat than on lower-status sites (Table 3.3), and so this chapter focuses solely upon rural settlements, for which there is a far larger sample size, making it possible to compare animal husbandry practices across different regions. The earlier analysis of South West Britain (Rippon 2012b) simply compared the Roman and early medieval periods, but in subsequent work in Essex (Rippon 2013) it became clear that there were significant differences between the early and late Roman periods, and so wherever possible they are distinguished here. Of the 120,031 cattle, sheep/goat, and pig bones from rural settlements studied here, 69 per cent could be phased as early (AD 43–250) or late Roman (AD 250–410) (Table 3.4 and Appendix III). It is frustrating that so

and Pirnie 2008 (); Oxford Archaeology (); Wessex Archaeology (); TVAS (); and University of Leicester Research Archive (). 8 Bancroft in Wolverton, Bucks (Williams and Zeepvat 1994), Brigstock in Northants (Albarella and Pirnie 2008, using Greenfield 1963), Chelmsford in Essex (Wickenden 1992), Great Chesterford in Essex (Medlycott 2011), Great Dunmow in Essex (Wickenden 1988), Ivy Chimneys in Witham, Essex (Turner 1999), Coxwell Road in Faringdon, Oxon (Weaver and Ford 2004), Stanegrove Hill in Harlow, Essex (France and Gobel 1985), and West Hill in Uley, Gloucs (Woodward and Leach 1993).

A Landscape Approach to the Roman–Medieval Transition

73

many bone assemblages from multi-phase sites are simply published as ‘Roman’, even when there are assemblages that could be phased (e.g. the 1,511 bones from Rayne, in Essex, that are published as one assemblage, even though the settlement has three distinct phases: late first to early second, late second to early third, and late third to early fourth centuries: Smoothy 1989). A total of 84,155 Romano-British charred grains were recorded in this study, of which 68,148 came from non-corn-drier contexts (see above): of these just 10,455 could be phased as early Roman and 9,510 as late Roman, leaving 48,183 as unphased ‘Roman’. The early medieval data included sites published as ‘Early Saxon’ (i.e. fifth to seventh century) and ‘Early to Middle Saxon’ (i.e. fifth to eighth/early ninth century) as in some regions a single ceramic style covers this whole period. Sites that are clearly dated ‘Middle Saxon’ are treated separately. The sites used in this analysis are grouped by predominant surface geology, and for certain geologies in some periods it was possible to sub-divide the sites into different regions (listed in brackets below):  Boulder Clay (South East Midlands [Buckinghamshire, Bedfordshire,

            

Huntingdonshire, and Cambridgeshire]; Essex and Hertfordshire; East Anglia; The Felden; The Midland Plateau) Jurassic Clay (Wessex [Dorset, Somerset, and Wiltshire]; South Midlands [Bucks and Oxon]; Gloucestershire) London Clay Clay-with-flints Chalk (Wessex [Dorset and Wiltshire]; Berkshire/Marlborough Downs; Chilterns; Cambridgeshire/East Anglian hills) Limestone (South West Jurassic Hills [Somerset and Dorset]; Cotswolds; Midlands) heathland (Breckland; South East Dorset) valley terraces (Vale of Gloucester; Upper Thames Valley; Great Ouse Valley; Cam Valley; East Anglia; Essex/Lower Thames Valley; other) unreclaimed marshland reclaimed marshland bedrock islands within wetlands Fenland-edge sites that join Boulder Clay Fenland-edge sites that join Chalk ‘mixed’ (where there is no predominant surface geology).

The term ‘valley terraces’ has been used instead of the actual geology recorded in excavation reports—sand and gravel—as the latter implies light soils, whereas these low-lying locations actually had high water tables and

74

The Fields of Britannia

heavy soils (including a capping of alluvium that is sometimes stripped off before excavation). The proximity to rivers and streams also means that there was an ample supply of water for livestock, which is particularly important for cattle that require large amounts (e.g. Rippon et al. 2014, fig. 6). In Tables 3.3–3.6 percentages have been rounded up or down to avoid giving an unwarranted impression of precision. The data given is for the total cattle, sheep/goat, and pig bones, or oats, barley, rye, and wheat grains, from the relevant phases of each site. Sites/phases with less than 100 bones are included, which is not normally the case in archaeobotanical and zooarchaeological studies. Were the focus of this study to be the interpretation of individual sites, then using such small sample sizes would indeed be inappropriate, but in this study the level of analysis is on a different scale—the landscape as a whole, as opposed to an individual site—and it is argued here that some of these individually small assemblages can be combined where they were found on the same geology. One welcome result of PPG16 and its successors is that in many parts of the country there has been a dramatic increase in archaeological excavation and therefore palaeoenvironmental sampling, although many of the resulting plant macrofossil and faunal assemblages are small. If the minimum sample size that is used for analysing individual assemblages—say 100 specimens—had been adopted here, then many of these smaller assemblages would have been rejected. Rather than consigning these data to the spoil heap, however, the contention is that these assemblages can be useful if data from several sites on the same predominant surface geology are aggregated. An example is rural settlements at the margins of the South East Midlands Boulder Clay and Fenland, where six sites have produced 37, 72, 76, 86, 90, and 105 cattle, sheep/goat, and pig bones. Individually these are all small assemblages, so should such data be disregarded? The contention here is no, as when these data are aggregated we get a far larger sample size (466 bones) across which the pattern of animal husbandry is consistent: across all the sites cattle are 58 per cent, sheep/goat 35 per cent, and pig 8 per cent, and if the largest assemblage of 105 bones is removed then the remaining sites still show that cattle are 56 per cent, sheep/goat 35 per cent, and pig 9 per cent.

Animal Husbandry King (1989; 1991; 1999) has shown how the proportions of cattle, sheep/goat, and pig bones varies on sites of different status (military, urban, villa, and lower-status rural settlements) across the province as a whole, which evidently reveals the dietary preferences of these communities (Table 3.3). King examined ninety non-villa rural settlements and found that cattle comprised 47.1 per cent, sheep/goat 41.2 per cent, and pig 11.7 per cent, but what he did not

King 1999, tab. 3 Site type

No. sites Cattle

Sheep/goat

This study (see Table 3.4 and Appendix III) Pig

Early Roman No. sites Cattle

Late Iron Age Legionary fortress Vici Towns Villas Rural settlements

10 16 69 50 58 90

39.0  14.9 63.5  16.8 56.3  17.6 53.5  18.5 55.6  15.8 47.1  15.7

31.3  8.5 14.9  11.2 31.9  15.5 27.0  14.0 29.8  12.2 41.2  15.7

29.7  14.5 21.6  11.0 11.8  7.7 19.5  9.8 14.6  11.7 11.7  9.0

Not studied Not studied Not studied 33 51% 16 48% 146 49%

Sheep/goat

31% 32% 41%

Late Roman Pig

No.

Cattle

18% 20% 10%

Not studied Not studied Not studied 23 57% 14 53% 122 58%

Sheep/goat Pig

29% 30% 36%

14% 18% 6%

A Landscape Approach to the Roman–Medieval Transition

Table 3.3. A statistical analysis of the percentages of cattle, sheep/goat, and pig from sites of different status in Roman Britain (King 1999, tab. 3)

75

Table 3.4. A comparison of the major domesticated animals on rural settlements, villas, and towns dating to the early and late Roman periods and fifth to seventh/eighth centuries Early Roman No. sites

Cattle

Sheep

Nos

Nos

Villas (sites listed below) Boulder Clay Chalk Limestone Valley terraces Mixed Totals

4 912 3 423 2 1,394 4 1,194 3 442 16 4,365

Pigs

No. sites

% Nos % Total

Cattle

Early to mid-Saxon

Sheep

Pigs

Nos

%

Nos

% Nos %

8,434 2,038 34 860 2,347 2,281 393 2,467 2,163 388 188 813 1,112

63 74 49 75 42 51 47 62 56 72 49 38 65

4,233 637 34 191 2,848 1,729 297 1,264 1,451 115 158 1,172 524

32 23 49 17 51 39 35 32 38 21 41 55 31

49 16,902 41 4,224 10 41,630 122 23,518

58

60 447 30 715 54 638 43 813 50 338 48 2,951

77 39 54 54 46 53

55 4,638 40 661 6 11,689 57 707 36 133 7 1976 24 42 3 1,121 73 1,526 47 176 40 58 13 445 30 2,799 64 240 6 4,347 55 726 38 131 7 1,911 44 417 39 183 17 1,072 56 4,745 34 1,330 10 13,963 45 567 44 140 11 1,282 6 450 93 7 1 486 33 1,083 59 44 552 51

158 9 1,845 62 6 1,088

29 172 51 256 25 533 29 789 38 100 32 1,850

11 18 21 28 11 20

30 7 1 2 16 6 4 24 11 3 2 2 8

1,531 3 1,624 1,394 3 1,822 2,565 3 2,229 2,796 3 4,216 880 2 161 9,166 14 10,052

656 83 1 96 404 451 148 249 253 39 35 153 79

5 3 1 8 7 10 18 6 7 7 9 7 5

Total No. sites

13,323 19 2,758 8 69 1,147 5,599 11 4,461 2 838 1 3,980 21 3,867 6 542 381 2,138 1 1,715

Cattle

Sheep

Nos

Nos

%

%

Pigs Nos

Total %

6,560 56 3,849 33 1,269 11 11,678 1,117 57 476 24 384 19 1,977

2,873 474 6,814 7,225 1,166

48 2,816 42 539 9 5,928 37 540 42 270 21 1,284 38 8,187 45 3,125 17 18,126 62 2,235 19 2,134 18 11,594 46 817 32 566 22 2,549

325 40

410 51

72 9

807

3 6,324 58 976 9 3,600 33 10,900 14,653 36 2,647 6 40,818 72 32,878 51 20,006 31 11,959 18 64,843

379 1,243 1,200 2,732 172 5,726

18 100 5 2,103 26 1,631 35 4,696 29 698 17 4,127 35 903 12 7,851 49 16 5 349 30 3,348 18 19,126

The Fields of Britannia

Rural settlements Boulder Clay 37 6,390 Jurassic Clay: totals 8 1,136 London Clay 2 363 Clay-with-flints 3 211 Chalk 15 1,308 Limestone 7 1,054 Heathland 5 472 Valley terraces 46 7,888 Mixed 10 575 Unreclaimed marshland 3 29 Reclaimed marshland Fenland islands 4 604 Fenland edge—Boulder 6 474 Clay Early medieval elite sites Totals 146 20,504 Sources: see Appendix III

%

Late Roman

76

Area

Small towns

13 5,091 57 2,753 31 1,106 12 8,950

12 8,372 62 4,018 30 1,034 8 13,424

Major towns

20 13,385 49 8,298 31 5,448 20 27,131

9 13,251 54 6,810 28 4,372 18 24,433

Small and major towns Boulder Clay Chalk Limestone Valley terraces Fenland island (Stonea) Totals

10 2,093 3 2,826 5 3,387 14 9,751 1 419 33 18,476

47 35 60 57 40 51

1,830 3,214 1,403 4,108 496 11,051

41 522 12 4,445 40 1,995 25 8,035 25 810 14 5,600 24 3,100 18 16,959 48 127 12 1,042 31 6,554 18 36,081

8 3 4 7 1 23

6,495 3,822 5,355 4,702 1,249 21,623

62 43 70 61 44 57

3,010 3,142 1,339 2,066 1,271 10,828

29 944 9 10,449 35 2,150 24 9,114 17 972 13 7,666 27 1,000 13 7,768 44 340 12 2,860 29 5,406 14 37,857

Sources: Asthall, Oxon (Booth 1997); Aylesbury, Bucks (Albarella and Pirnie 2008 using Allen 1982); Braintree, Essex (Drury 1976; Havis 1993; Garwood and Lavender 2000); Chelmsford, Essex (Drury 1988; Wickenden 1992; Nicholson and Roberts 2007); Chesterton, Oxon (Booth et al. 2001); Cirencester, Gloucs (Walker 1991; Maltby 1998; Holbrook 2008); Colchester, Essex (Luff 1993; Brooks 2004; Pooley et al. 2011); Colne, Cambs (Regan et al. 2004); Crandon Bridge, Somerset (Rippon 2008b); Dorchester, Dorset (Woodward et al. 1993); Dorchester on Thames, Oxon (Torrance and Durden 1998); Ducklington/South Leigh, Oxon (Booth and Simmonds 2011); Dunstable, Beds (Mudd 2004a); Godmanchester, Cambs (Albarella and Pirnie 2008, using A. Jones 2003); Great Chesterford, Essex (Draper 1986; Medlycott 2011); Great Dunmow, Essex (Wickenden 1988; O’Brien 2005); Hacheston, Suffolk (Blagg et al 2004); Heybridge, Essex (Johnstone and Albarella 2002); Kempston, Beds (Dawson 2004); Long Melford, Suffolk (Craven 2012); Scole, Norfolk (Albarella and Pirnie 2008, using Baker 1998); Standon, Herts (Potter and Trow 1988); Stonea, Cambs (Jackson and Potter 1996); Towcester, Northants (Lambrick 1980); Wellingborough, Northants (Foster et al. 1977); Wycomb, Whittington, Gloucs (Timby 1998)

A Landscape Approach to the Roman–Medieval Transition

Sources: Great Holts Farm, Boreham, Essex (Germany 2003); School Lane, Welwyn, Herts (Grassam 2005b; Brogan and Unger 2007); Shapwick, Somerset (Gerrard with Aston 2007); Bury Close, Fawler, Oxon (Allen 1988); Wendens Ambo, Essex (Hodder 1982); Brigstock Road, Stanion, Northants (Tingle 2008); Hockwold-cum-Wilton, Norfolk (Salway 1967); East End, Combe, Oxon (Speake 2012); Bancroft, Wolverton, Bucks (Williams and Zeepvat 1994); Shakenoak Farm, North Leigh, Oxon (Brodribb et al. 2005); Chignal St James, Essex (Clarke, C. 1998); Brigstock Road, Stanion, Northants (Walker 2012); Frocester, Gloucs (Gracie and Price 1979); Ditches, North Cerney, Gloucs (Trow et al. 2009); Gorhambury, Herts (Neal et al. 1990); Castle Hill, Ipswich, Suffolk (Wessex Archaeology 2003); Rivenhall, Essex (Rodwell and Rodwell 1993); Great Wilbraham, Cambs (Ette and Hands 1993); Box, Wilts (Hurst et al. 1987); Coberley, Gloucs (Wessex Archaeology 2008a); Withington, Gloucs (Wessex Archaeology 2006b); Boxmoor, Herts (Neal 1974–76); Chew Park, Chew Magna, Somerset (Rahtz and Greenfield 1977); Redlands Farm, Stanwick, Northants (Albarella and Pirnie 2008, using Davis 1997); Roughground Farm, Lechlade, Gloucs (Allen et al. 1993); Little Oakley, Essex (Barford 2002); Itter Crescent, Peterborough, Cambs (Henley et al. 2012); Latimer Park Farm, Latimer, Bucks (Branigan 1971); Atworth villa, Wilts (Erskine and Ellis 2008); Dicket Mead, Welwyn, Herts (Rook 1983–86); Halstock, Dorset (Lucas 1993); Bucknowle, Dorset (Light and Ellis 2009); Bancroft, Wolverton, Bucks (Zeepvat et al. 1987); Wooton Fields, Wootton, Northants (Chapman and Thorne 2004); Barnsley Park, Barnsley, Gloucs (Webster et al. 1985); Castle Copse, Great Bedwyn, Wilts (Hosteller & Noble Howe 1997); Claydon Pike, Lechlade/Fairford, Gloucs (Miles et al. 2007); Cosgrove, Northants (Quinnell 1991); Church Langley, Harlow, Essex (Medlycott 2000); Bulls Lodge Farm, Boreham, Essex (Lavender 1993); Stantonbury, Great Linford, Bucks (Zeepvat et al. 1987); Clapthorn Road, Oundle, Northants (Maull and Masters 2005).

77

78

The Fields of Britannia

consider is the extent to which animal husbandry may have varied over the course of the Roman period, and from region to region, or how it changed in the early medieval period. As part of this study, therefore, animal bone data from 319 non-villa Romano-British and fifth- to seventh/eighth-century rural settlements was collected, which is presented in Appendix III and summarized in Table 3.4 (note that some sites have seen excavations on several occasions, and the assemblages from these separate excavations have been amalgamated here, for example at Figheldean in Wiltshire: McKinley and Heaton 1996; McKinley 1999b). In Appendix III, the first column lists the areas from which data has been recorded, with individual excavations listed below. The next set of columns is for assemblages that are phased as ‘Roman’, followed by those phased as ‘early Roman’ and ‘late Roman’. The next set of columns combines these three sets of data and so includes all ‘Roman’ material. The final set of columns relates to fifth- to seventh/eighth-century material (which is often referred to as ‘Early Saxon’ or ‘Early to Middle Saxon’). Early Roman assemblages have been recovered from 146 sites, late Roman from 122, and there are 72 fifth- to seventh/eighth-century sites; there is unphased material from 117 Roman sites (some sites have produced material from several phases, or that is both phased and unphased). For the Roman period as a whole, cattle comprised 52 per cent, sheep/goat 40 per cent, and pig 8 per cent (Appendix III and Tables 3.3 and 3.4), but these overall figures hide some marked variations in both time and space. For the early and late Roman periods, cattle rose in significance from 49 to 58 per cent, sheep/goat fell from 41 to 36 per cent, and pig fell from 10 to 6 per cent. This trend is not uniform, however, across all geologies, with the shift towards cattle being particularly marked on heavy clay soils and valley terraces, but also the drier soils found on the chalkland and heathlands. On limestone (of which the vast majority of the data comes from the Cotswolds), in contrast, there was a shift from cattle to sheep/goat. The differences between geologies are very marked: in the late Roman period cattle were 62 per cent on the valley terraces (compared to a late Roman average of 58 per cent), 63 per cent on the Boulder Clay, and 74 per cent on the Jurassic Clays, but just 42 per cent on the chalk, 47 per cent on heathland, and 51 per cent on limestone. Sheep/goat, in contrast, were particularly significant on chalk (51 per cent, compared to a late Roman average of 36 per cent) and the Cotswolds (40 per cent). Although unphased, the very high proportion of sheep/goat in Dumnonia (73 per cent) is also very striking. All of the regions used in this analysis of faunal remains contained a diversity of geologies and soils, but with the exception of those categorized as ‘mixed’ there is a ‘predominant surface geology’ that provides it with a distinctive character. Some pays, however, were more distinctive than others, and when they occur in close proximity we get additional insights into animal husbandry. One example is around the margins of Fenland, where intertidal

A Landscape Approach to the Roman–Medieval Transition

79

saltmarshes to the north, around the Wash, merged with vast expanses of freshwater peatland to the south, that in turn lay adjacent to two very different geologies: Boulder Clay to the south west (in Cambridgeshire) and light soils to the south east (in the far north east of Cambridgeshire and East Anglia). The Boulder Clay that extended across much of southern Cambridgeshire supported animal husbandry regimes within which cattle (55 per cent) were particularly common in the early Roman period (cf. 49 per cent in the study area as a whole) rising to 63 per cent in the late Roman period (cf. 58 per cent in the study area as a whole). Communities living on the fen-edge—on the margins of Fenland and the Boulder Clay—initially had animal husbandry that was less strongly based upon cattle that were 44 per cent in the early Roman period (cf. 55 per cent on the nearby Boulder Clay), although the significance of cattle rose to 65 per cent in the late Roman period (cf. 63 per cent on the nearby Boulder Clay). The unphased assemblage of Romano-British bones from the Fenland site at Parson’s Drove also shows this dominance of cattle (58 per cent), and so it appears that the wetland-edge communities were making use of the rich pastures and meadows around the fringes of Fenland. This is seen even more clearly on sites around the south east margins of Fenland (where it joins light soils derived from chalk) with cattle on the unphased Romano-British sites at 82 per cent. In contrast, communities living on the Fenland islands—which lay closer to the margins of the freshwater backfens and the coastal saltmarshes—had animal husbandry that was based far more on sheep/ goat (59 per cent in the early Roman period, and 55 per cent in the late Roman period), with cattle just 33 per cent in the early Roman period and 38 per cent in the late Roman period). Isotope analysis from faunal assemblages in Belgium has also shown how sheep were taken to graze on coastal saltmarshes (Müldner et al. 2014). This dominance of sheep is not surprising as they are particularly suited to grazing on the extensive unreclaimed marshland that would have lain within easy reach of Stonea, as the salty conditions help to prevent foot rot and liver fluke, two of the major ailments that sheep suffer from on heavy freshwater soils: the latter is transmitted to sheep via a snail (Lymnaea truncatula) that lives on damp freshwater grassland, but cannot tolerate salt and so is absent on saltmarshes (C. French 1996, 653; Stallibrass 1996, 591, 606). Although the sample sizes for villas are far smaller (Table 3.4 and Appendix III), pig bones are clearly more common than on lower-status rural settlements. It is also noticeable that on Boulder Clay cattle are more significant than the average for villas (cattle are also dominant on lowerstatus rural settlements on the Boulder Clay). This might suggest that the nature of the soils exerted an influence on the choice of animal husbandry regime within villa estates, although it must be remembered that the sample sizes are small. The same may, however, be seen with the hinterlands of

80

The Fields of Britannia

towns. As non-agriculturally productive sites, towns will have drawn their food from a wide hinterland whose geology will have been far more varied than for a rural settlement. It is noticeable, however, that for the geologies for which we have a significant amount of data, cattle are again relatively common in towns whose hinterland is predominantly Boulder Clay and limestone, while sheep/goat predominate on chalkland towns and the Fenland island of Stonea (Table 3.3). When considering animal husbandry in the early medieval period there are far fewer sites (just seventy-two). There is a general decline in cattle from 58 per cent in the late Roman period to 51 per cent in the fifth to eighth centuries, which is to be expected as the market-based economy of Roman Britain collapsed. Some areas, however, buck this trend, most notably all of the chalkland areas (where cattle increased from 42 per cent to 48 per cent) and the Fenland islands, where cattle increased very slightly from 38 per cent to 40 per cent. The consistent pattern that emerges, however, is that the increased specialization seen in the late Roman period disappears as animal husbandry regimes revert to their early Roman and potentially pre-Roman character. Interestingly, scientific analysis of the Romano-British and early medieval cattle bones from Portchester (Hampshire) shows a difference in the mean carbon isotope value that may indicate a change in how cattle were fed between the late Roman and the ‘Early Saxon’ periods, with a decline in foddering and increase in foraging (Allen 2013). This fits with the analysis of the faunal assemblages presented here that suggests a move away from market-based production. On the Boulder Clay, for example, cattle increased from 55 per cent in the early Roman period to 63 per cent in the late Roman period, but fell back to 56 per cent in the early medieval period. On the Jurassic Clays, the late Roman specialization in cattle (74 per cent) is even more marked—in the early Roman period they were 57 per cent, virtually the same as on the Boulder Clay (55 per cent)—and in the early medieval period they fell back to 57 per cent. On the mixed geologies of East Anglia (Appendix III), cattle increased from 62 per cent in the early Roman period to 92 per cent in the late Roman period, falling back to 62 per cent in the early medieval period, and a similar picture is seen in Breckland (cattle increased from 45 per cent in the early Roman period to 60 per cent in the late Roman period, falling back to 38 per cent in the early medieval period) and the Nene Valley (where cattle increased from 35 per cent in the early Roman period to 82 per cent in the late Roman period, falling back to 42 per cent in the early medieval period). Against this background of what appears to be a less market-based animal husbandry, there are some broad continuities with the late Roman period: on the Boulder Clays, valley terraces, and areas of mixed geologies within which cattle predominated in the Roman period (such as the Nene Valley and East Anglia) cattle were above average

A Landscape Approach to the Roman–Medieval Transition

81

in significance, while on chalk and heathland sheep/goat were particularly significant.

Cereal Cultivation The data used for this discussion is contained within Tables 3.5 and 3.6, and Appendix IV, which is laid out in the same way as Appendix III (animal bone: see above), and only includes non-corn-drier contexts (see above). Data was collected from a total of 175 Romano-British and fifth- to seventh/eighthcentury sites, with fifty-seven sites producing early Roman material, thirtynine producing late Roman, and fifty producing fifth- to seventh/eighth-century material (Appendix IV). In contrast to the animal bones, which suggest pronounced changes in animal husbandry over the course of the Roman period, a comparison of the cereal grain assemblages suggests little change in crop regimes over the study area as a whole, with wheat dominant (66 per cent in both the early and late Roman periods), barley very much a secondary crop (24 per cent in the early Roman period and 29 per cent in the late Roman period), and oats a very minor component (10 per cent in the early Roman period and 5 per cent in the late Roman period): rye was hardly grown at all (0.4 per cent in the early Roman period and 0.1 per cent in the late Roman period). This dominance of wheat in the Roman period is entirely expected and would have helped support the agriculturally non-productive parts of society (in towns and military establishments) through the money-based market economy. These overall figures, however, hide some very marked differences between geologies, on most of which wheat increased in significance: on chalk from 35 per cent to 78 per cent, on the Jurassic Clay from 51 per cent to 77 per cent, on heathland from 51 per cent to 69 per cent, on limestone from 61 per cent to 86 Table 3.5. A comparison of charred cereal grains from corn drier and non-corn drier contexts on rural settlements Corn driers (%)

Non-corn driers (%)

Early Roman Oats Barley Rye Wheat

0.3 5 0.1 95

10 24 0.4 66

Late Roman Oats Barley Rye Wheat

4 21 0 76

5 29 0.1 66

82

The Fields of Britannia

Table 3.6. Changing crop regimes on rural settlements over the early and late Roman periods and fifth to seventh/eighth centuries using non-corn drier contexts (for sites/ sources see Appendix IV) Early Roman (%)

Late Roman (%)

Fifth to seventh/eighth centuries (%)

Oats Barley Rye Wheat

10 24 0.4 66

5 29 0.1 66

19 39 9 33

Wheat Boulder Clay Jurassic Clay Chalk Heathland Valley terraces Dumnonia Fenland islands

86 51 35 51 48 67 83

97 77 78 69 49

60 68 56 8 24 27 48

Oats Boulder Clay Jurassic Clay Chalk Heathland Valley terraces Dumnonia Fenland islands

7 5 3 30 13 27 0

0 5 2 6 2

Barley Boulder Clay Jurassic Clay Chalk Heathland Valley terraces Dumnonia Fenland islands

7 43 62 19 39 6 17

3 18 19 20 49

Rye Boulder Clay Jurassic Clay Chalk Heathland Valley terraces Dumnonia Fenland islands

1 1 0 0