New Visions of the Countryside of Roman Britain: The Rural Settlement of Roman Britain 0907764436, 9780907764434

It has often been stated that Roman Britain was quintessentially a rural society, with the vast majority of the populati

204 35 47MB

English Pages 469 [497] Year 2016

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

New Visions of the Countryside of Roman Britain: The Rural Settlement of Roman Britain
 0907764436, 9780907764434

Citation preview

SMITH ET AL.

THE RURAL SETTLEMENT OF ROMAN BRITAIN Alexander Smith, Martyn Allen, Tom Brindle and Michael Fulford

THE RURAL SETTLEMENT OF ROMAN BRITAIN

29

BRITANNIA MONOGRAPH SERIES No. 29 Published by the Society for the Promotion of Roman Studies 2016

Romanindx.indd 470

15/09/2016 17:26:24

NEW VISIONS OF THE COUNTRYSIDE OF ROMAN BRITAIN

VOLUME 1: THE RURAL SETTLEMENT OF ROMAN BRITAIN

Romanprelims.indd 1

15/09/2016 15:33:50

Romanprelims.indd 2

15/09/2016 15:33:50

NEW VISIONS OF THE COUNTRYSIDE OF ROMAN BRITAIN VOLUME 1

THE RURAL SETTLEMENT OF ROMAN BRITAIN BY

Alexander Smith, Martyn Allen, Tom Brindle and Michael Fulford

Series Editors Michael Fulford and Neil Holbrook

Britannia Monograph Series No. 29 Published by the Society for the Promotion of Roman Studies Senate House Malet Street London WC1E 7HU 2016

Romanprelims.indd 3

15/09/2016 15:33:50

BRITANNIA MONOGRAPH SERIES NO. 29 Published by the Society for the Promotion of Roman Studies Senate House, Malet Street, London WC1E 7HU

This monograph was published with the aid of a grant from Historic England

© Copyright Society for the Promotion of Roman Studies 2016

British Library Catalogue in Publication Data A catalogue record of this book is available from the British Library ISBN 978 0 907764 43 4

Front Cover illustration: Artist’s reconstruction of Roman farmstead at Haddon, Cambs. (By Jon Cane, after Hinman 2003, 60, fig. 23) Back Cover illustration: Iron Age and Romano-British farmsteads at Fingland, Solway Plain, Cumbria; photograph taken 7 July 1984. (© Robert Bewley)

Printed by 4Word Ltd, Bristol BS13 7TT Printed in Great Britain

Romanprelims.indd 4

15/09/2016 15:33:50

CONTENTS

List of Figures List of Tables Preface Acknowledgements Summary CHAPTER 1: INTRODUCTION by Michael Fulford and Tom Brindle Introduction History of research into the Romano-British countryside The nature of the dataset Developing a methodology A regional analysis of rural Roman Britain The research structure

Case studies: Middle Thames Valley and the Hampshire Downs Region summary

vii xvii xviii xxii xxiii

129 139

CHAPTER 5: THE CENTRAL BELT by Alexander Smith 141 The nature of the landscape The Central Belt dataset 142 Roman rural settlement patterns 145 Buildings 167 Landscape context and infrastructure 175 Settlement hierarchies: the social and 183 economic basis of settlements Case study: the Cambridgeshire Fen edge 192 Region summary 206

1 4 8 9 15 16

CHAPTER 6: THE EAST by Alexander Smith 208 The nature of the landscape The East dataset 209 Roman rural settlement patterns 212 Buildings 225 232 Landscape context and infrastructure Settlement hierarchies: the social and 234 economic basis of settlements Region summary 240

CHAPTER 2: RURAL SETTLEMENT IN ROMAN BRITAIN: MORPHOLOGICAL CLASSIFICATION AND OVERVIEW by Martyn Allen and Alexander Smith Introduction 17 17 Classification of rural settlement Farmsteads 20 Villas 33 Nucleated settlements: roadside settlements, military vici and villages 37 Summary 43

CHAPTER 7: THE NORTH-EAST by Martyn Allen The nature of the landscape 242 The North-East dataset 243 Roman rural settlement patterns 245 Buildings 262 270 Landscape context and infrastructure Settlement hierarchies: the social and 273 economic basis of settlements Region summary 280

CHAPTER 3: BUILDINGS IN THE COUNTRYSIDE by Alexander Smith Introduction 44 45 The architectural dataset Building form and material 47 Building function 54 An architectural continuum: building types 64 From national overviews to 74 regional syntheses

CHAPTER 8: THE CENTRAL WEST by Tom Brindle 282 The nature of the landscape The Central West dataset 283 Roman rural settlement patterns 286 Buildings 294 298 Landscape context and infrastructure Settlement hierarchies: the social and economic basis of settlements. A case study from the Chester 300 and Wroxeter hinterlands Region summary 306

CHAPTER 4: THE SOUTH by Martyn Allen The nature of the landscape 75 The South dataset 76 Roman rural settlement patterns 78 Buildings 102 114 Landscape context and infrastructure Settlement hierarchies: the social and 121 economic basis of settlements v

Romanprelims.indd 5

15/09/2016 15:33:50

vi

THE RURAL SETTLEMENT OF ROMAN BRITAIN

CHAPTER 9: THE NORTH by Tom Brindle The nature of the landscape 308 The North dataset 309 Roman rural settlement patterns 311 Buildings 320 Landscape context and infrastructure 322 Settlement hierarchies: the social and economic basis of settlements 324 Region summary 329 CHAPTER 10: THE SOUTH-WEST by Tom Brindle The nature of the landscape 331 The South-West dataset 332 Roman rural settlement patterns 334 Buildings 346 Landscape context and infrastructure 349 Settlement hierarchies: the social and economic basis of settlements 353 Region summary 357 CHAPTER 11: UPLAND WALES AND THE MARCHES by Tom Brindle The nature of the landscape The Upland Wales and Marches dataset Roman rural settlement patterns

Romanprelims.indd 6

360 360 363

Buildings 373 Landscape context and infrastructure 378 Settlement hierarchies: the social and economic basis of settlements 380 Region summary 383 CHAPTER 12: CONCLUSIONS: THE RURAL SETTLEMENT OF ROMAN BRITAIN by Alexander Smith and Michael Fulford Introduction 385 The regions 385 Chronological patterns 404 Rural population 416 Town and country 418 Roman rural settlement: reflections 419 and future research APPENDICES 1. I ntroduction and guide to the digital resource by Tim Evans 2. Kernel density 3. Finds categories

421 423 425

BIBLIOGRAPHY

426

INDEX (Peter Ellis)

455

15/09/2016 15:33:50

LIST OF FIGURES

Chapter 1 1.1 1.2

1.3

1.4 1.5

Quantity and distribution of excavated Roman rural sites over time Proportion of each major type of intervention with only imprecise dating evidence available Average number of pottery sherds recovered from excavations, evaluations and watching briefs Distribution of sites with radiocarbon dates Map showing Natural England’s ‘Natural Areas’ within the Roman Rural Settlement Project regions

2.14 Plan of early Roman complex farmstead at Wavendon Gate, Buckinghamshire 30 2.15 Plans of complex farmsteads at (a) Haddon, Cambridgeshire, and (b) southern site at NIAB Huntington Road, Cambridge 31 2.16 Chronology of different types of 32 complex farmsteads in use 2.17 Distribution of complex farmsteads by type 32 33 2.18 The origins of complex farmsteads 2.19 Distribution of excavated villas 34 in relation to NMR ‘villas’ 2.20 The origins of villas 35 2.21 Kernel density distribution of villas from late first to fourth 35 century a.d. 2.22 Number of villas able to be classified 36 2.23 Plans of late Roman ‘complex’ villa at Barton Court Farm, Oxon (a) and mid-Roman ‘enclosed’ villa at Chilgrove 2, West Sussex (b) 36 2.24 Distribution of excavated nucleated settlements 37 2.25 Geophysical survey plan of the roadside settlement at Westhawk Farm, Kent 38 2.26 Geophysical survey plan of the fort and vicus at Caer Gai, Gwynedd 39 2.27 Excavation plan of the roadside settlement at Moor Lane, Stamford Bridge, Yorks 39 2.28 Interpretative plan of the roadside settlement at Scole 40 2.29 Earthwork survey plans of village settlements at Chisenbury Warren, Wiltshire, and Chalton, Hampshire 41 2.30 Plan of Roman village settlement at Mucking, Essex 42

7

11

12 13

16

Chapter 2 2.1

2.2 2.3 2.4 2.5

2.6

2.7 2.8 2.9 2.10

2.11

2.12 2.13

Frequency of farmsteads and/or villas with and without the attribution of morphological classifications by area of excavation 19 Frequency of classification of farmstead and/or villas by area of excavation 19 Site plan of Melton wastewater works 20 near Brough, East Riding Distribution of all excavated sites 21 recorded as farmsteads Distribution of excavated open farmsteads and plan of mid-Roman open farmstead at Strood Hall, Essex 22 Distribution of excavated enclosed farmsteads and plan of enclosed farmstead at Bishopstone, 23 East Sussex Relative frequencies of different enclosed farmstead types by region 24 Distribution of enclosed farmsteads by type 24 Plans of ‘Banjo’-type enclosures 25 Distribution of enclosed farmsteads with identified masonry- or earthen26 walled enclosures Development of masonry-walled enclosed farmstead at 27 Cefn Graenog II Plans of multi-ditched enclosed farmsteads 28 Distribution of excavated complex farmsteads and plan of complex farmstead at Cotswold Community, Wilts/Glos 29

Chapter 3 3.1

3.2 3.3 3.4

Distribution of all excavated late Iron Age and Roman rural buildings 45 The context of buildings 45 Average number of buildings per site by region and settlement type 46 Use of circular and rectangular buildings over time 47

vii

Romanprelims.indd 7

15/09/2016 15:33:50

viii

3.5

3.6

3.7

3.8 3.9 3.10

3.11

3.12

3.13

3.14

3.15 3.16 3.17 3.18

3.19 3.20

3.21

3.22 3.23 3.24 3.25

Romanprelims.indd 8

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Distribution of all excavated circular and rectangular buildings, c. late Iron Age–late Roman 48 Distribution of sites with different architectural forms from the late 49 Iron Age to the 4th century a.d. Relative frequency of circular and rectangular buildings according to major settlement type 50 Distribution of all excavated timber/ mass-walled and masonry buildings 52 Distribution of circular masonry buildings 53 Internal functional areas within Structure 6 at Langdale Hale (a) and functional designation of buildings within the Romano-British port settlement at Camp Ground, 55 Colne Fen (b) Distribution of excavated sites with corndryers and plan of Yewden villa, Bucks 58 Distribution of excavated sites with four-post structures and buildings interpreted as granaries and plans 59 of selected ‘granaries’ Distribution of excavated sites with buildings interpreted as workshops and plan of late Roman pottery 61 workshop at Stibbington, Cambs Distribution of excavated sites with structures interpreted as religious in nature and plan of shrine at Rutland 62 Water, Rutland Plan of religious enclosure at Higham Ferrers, Northants 63 Plans of ‘single-roomed’ buildings 64 Distribution of excavated sites with 65 Roman cellared buildings Distribution of excavated sites with aisled buildings and selected aisled building plans 66 Number of aisled buildings in use 67 over time Distribution and relationship of excavated aisled buildings associated with villas 68 Distribution of excavated farmsteads with multi-room buildings and plans of buildings in Gloucestershire 70 Types of villa within the dataset by 71 total number and by final form Plans of selected villa types 72 Chronology of villa types 73 Distribution of different types of excavated villas 74

Chapter 4 4.1 4.2 4.3

4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14

4.15 4.16 4.17 4.18 4.19 4.20

4.21 4.22

The South region in relation to modern county boundaries 75 Constituent landscape zones of 76 the South region Kernel density of South region records and all excavation records from National Monument Records 77 (NMR) Index Excavated late Iron Age/Roman rural settlement in the South region 79 Relative frequency of main settlement types by landscape 79 zones in the South region Number of settlements in use over 81 time in the South region Variations in settlement chronology by selected landscape zone within the South region 82 ‘New’ and ‘abandoned’ settlements 83 over time in the South region ‘New’ and ‘abandoned’ settlements over time in selected landscape zones 84 of the South region Number of open, enclosed and complex farmsteads in selected landscape zones of the South region 85 Relative frequency of farmstead types in use over time in the South region 85 Distribution of farmstead types in 86 use over time in the South region Distribution of farmstead types in relation to geology in the South region 87 Plans of complex farmsteads at Pingewood, Burghfield, Berkshire (a), Beam Washlands, Dagenham, Greater London (b), and Chigborough Farm, 88 Essex (c) Site plans of complex farmsteads 88 on the Hampshire Downs Site plans of larger complex 89 farmsteads in Hampshire Phased site plan for the open Romanperiod farmstead at Foxholes Farm, Hertford 90 Chronology of occupation on villa 91 sites in the South region Distribution of villas over time in 92 the South region Chronological variation in date of villa construction between eastern and western areas of the South region 93 Phased plans of the site at 93 Beddington, Surrey Villas as a percentage of the total number of settlements in different landscape zones of the South region 94

15/09/2016 15:33:51

LIST OF FIGURES

4.23 Geophysical survey results at Brading villa, Isle of Wight 95 4.24 Plans of villas at Sparsholt, 95 Hampshire (a) and Keston, Kent 4.25 Excavation and interpretative plans of late Roman villas in Hampshire 96 4.26 Plan of the roadside settlement at Shapwick, Dorset 98 4.27 Variation in dating evidence from different excavations in Staines 99 4.28 Site plans of settlements on the Salisbury Plain 100 4.29 Aerial survey plan of the village settlements at Charlton Down, Upavon Down, and Compton Down, Wiltshire 101 4.30 Plan of first-century a.d. enclosures at Ower, Poole Harbour, Dorset 101 4.31 Distribution and frequency of rural buildings on excavated sites in the South region 102 4.32 Frequency of settlements with evidence for buildings by different landscape zones in the South region 103 4.33 Number of sites with circular and/or rectangular buildings over time in 104 the South region 4.34 Distribution of excavated masonry buildings over time in the South region 105 4.35 Relative frequency of masonry and timber buildings found at farmsteads within selected landscape zones in the South region 105 4.36 Plans of internally partitioned posthole structures at Winnall Down, Hampshire (a), and late first-century a.d. timber-framed structure at Chichester Harbour, Sussex (b) 106 4.37 Site plan of the late Roman building complex at Woodhouse Hill, Dorset 107 4.38 Plans of aisled buildings with later modifications 107 4.39 Distribution of aisled buildings across the South region 108 4.40 Plans of timber aisled buildings at Furfield Quarry, Boughton Monchelsea, Kent (a), and Bower 108 Road near Smeeth, Kent (b) 4.41 Plans of cottage/strip-house villas at Houghton Down, Hants (a) and Shillingstone, Dorset (b) 109 4.42 Plans of winged-corridor villas at Barcombe, Sussex (a), Lullingstone, Kent (b), and Sedgebrook Field, Plaxtol, Kent (c) 109

Romanprelims.indd 9

ix

4.43 Plans of courtyard/‘palatial’ villas at Eccles, Kent (a), Fishbourne, Sussex (b), and Darenth, Kent (c) 110 4.44 Plans showing the development of the villas at Bignor, West Sussex (a), and Rockbourne, Hampshire (b) 111 4.45 Plan of the ‘villa’ complex at 112 Chiddingfold, Surrey 4.46 Distribution of excavated villas according to their final ‘type’ in the South region 113 4.47 Spot height analysis for villas, complex farmsteads and enclosed farmsteads in the South region 114 4.48 Distances of villas and farmstead types from the major road network 115 in the South region 4.49 Distances of villas, enclosed farmsteads and complex farmsteads from a major walled town in the South region 115 4.50 Riverside distribution of villas in north Kent 116 4.51 Plan of North Bersted, West Sussex 117 4.52 Plan of Lea Farm, Hurst, Berkshire, showing late Iron Age enclosed farmstead (a) and an early Roman complex farmstead with trackway (b) 118 4.53 Distribution of sites with excavated late Iron Age/Roman field systems in the South region 119 4.54 Number of excavated field systems in the London Basin and the Wessex landscape zones over time 120 121 4.55 Plan of Syon Park, Brentford 4.56 Frequency of major artefact categories on all sites in the South region 122 4.57 Frequency of major artefact categories recovered across selected landscape zones in the South region 123 4.58 Frequency of selected categories of artefact on sites in selected landscape zones in the South region 123 4.59 Frequency of major artefact categories on different types of rural settlement 124 in the South region 4.60 Frequency of major artefact categories on villas and farmstead types in the 124 South region 4.61 Relative frequency of major livestock taxa from farmsteads in the London Basin, Hampshire Downs and South 126 Wessex Downs landscape zones 4.62 Inter-site type comparison of mean relative frequencies of major livestock taxa in the South region 127 4.63 Percentage presence of cultivated plant taxa on sites in the South region 127

15/09/2016 15:33:51

x

THE RURAL SETTLEMENT OF ROMAN BRITAIN

4.64 Percentage presence of cultivated plant taxa on sites in selected landscape zones in the South region 128 4.65 Percentage presence of cultivated plant taxa on different site types in the South region 129 4.66 Distribution map of sites in the Middle Thames Valley case study area 130 4.67 Plans of three late Iron Age open farmsteads in the Middle Thames Valley 131 4.68 Plans of three Roman-period complex farmsteads in the Middle Thames Valley 131 4.69 Plan of Heathrow Terminal 5 showing middle Iron Age open settlement and LIA/ER complex settlement and field system 132 4.70 Plan of Ashford Prison, Spelthorne 132 4.71 Plans of later Roman droveway complexes in the Middle Thames Valley 133 4.72 Distribution of sites in the Hampshire Downs case study area 136 4.73 Plans of late Iron Age enclosed farmsteads in the Hampshire Downs region 136 4.74 Plans of Dunkirt Barn, Abbots Ann 137 4.75 Plan of East Anton, Hampshire 138 4.76 Plan of the villa and field systems 139 at Fullerton, Hampshire Chapter 5 5.1

The Central Belt region in relation to modern county boundaries 141 5.2 Constituent landscape zones of the Central Belt region 142 5.3 Kernel density of Central Belt region records and all excavation records from National Monument Records (NMR) Index 143 5.4 Excavated late Iron Age/Roman rural settlement in the Central Belt region 145 5.5 Relative frequency of main settlement types by landscape zones in the Central Belt region 147 5.6 Number of settlements in use over time in the Central Belt region 148 5.7 Variations in settlement chronology by selected landscape zone within the Central Belt region 148 5.8 ‘New’ and ‘abandoned’ settlements over time in the Central Belt region 149 5.9 ‘New’ and ‘abandoned’ settlements over time in selected landscape zones 149 of the Central Belt region 5.10 All excavated farmsteads in the 150 Central Belt region

Romanprelims.indd 10

5.11 Percentage of farmsteads classified by selected landscape zone in the Central Belt and breakdown of classification 151 5.12 Site plans of selected open, enclosed and complex farmsteads in the Central Belt region 152 5.13 Relative frequency of farmstead types in use over time in the Central Belt region 153 5.14 Chronological distribution of farmstead types in use in selected landscape zones in the Central Belt 153 5.15 Distribution of farmstead types in use over time in the Central Belt 154 5.16 Enclosed farmsteads at Claydon Pike, Glos and Old Shifford Farm, Oxon 155 5.17 Distribution of complex farmstead types in the Central Belt and site plans of farmsteads at Rudgeway Lane, Glos and NIAB Huntington Road, Cambridge 156 5.18 Site plan of Banbury Flood Alleviation Scheme, Banbury, Oxfordshire 157 5.19 Distribution of complex, enclosed and unclassified villas in the Central Belt 158 5.20 Number of villas in use over time in the Central Belt with indications of development 158 5.21 Plans of (a) ‘Complex villa’ at Roughground Farm, Glos. and (b) ‘Enclosed villa’ at Whitton Lodge, South Glamorgan 159 5.22 Distribution of all excavated Roman nucleated settlement in the Central Belt region, in relation to walled towns 161 5.23 Cropmarks of the ‘village’ settlement at Lockington, Leicestershire 162 5.24 Site plan of extensive excavations of the nucleated Roman ‘village’ settlement at Gill Mill, Ducklington, Oxfordshire 163 5.25 Magnetometer survey of regular ditched enclosures probably associated with the Roman roadside settlement at Fleet Marston, Aylesbury, Bucks 164 5.26 Excavation site plan of roadside settlement at Higham Ferrers, Northants and geophysics plot of the roadside settlement at Somerdale, Keynsham, Bath and North-East Somerset 165 5.27 Plan of mid-second to early thirdcentury a.d. phase of the roadside settlement at Stanwick, Northants 167 5.28 Distribution and frequency of rural buildings on excavated sites in the Central Belt region 168

15/09/2016 15:33:51

LIST OF FIGURES

5.29 Number of sites with circular and/or rectangular buildings over time in the Central Belt region as a whole, and in the Cotswolds and West Anglian 169 Plain landscape zones 5.30 Relative frequency of masonry and timber buildings found at Roman-period farmsteads within selected landscape zones in the Central Belt region 170 5.31 Plans of Roman multi-room buildings at (a) Newhouse Park, Chepstow, South Wales and (b) Horcott 171 Quarry, Glos 5.32 Plans of Roman buildings at (a) Alfreds Castle, Oxon, (b) Chilton Fields, Oxon, and (c) Croughton, Northants 172 5.33 Distribution of excavated villa types within the Central Belt 174 5.34 Plan of parchmarks and geophysical survey results showing outline of multi-courtyard villa complex at Turkdean, Glos 174 5.35 Spot height analysis for villas, complex farmsteads and enclosed farmsteads in the Central Belt region 175 5.36 Distances of Roman farmsteads and villas from major roads in the Central Belt region 176 5.37 Plan of Roman settlement and trackway at NIAB, Huntingdon Road, Cambs 177 5.38 Cropmarks showing Romano-British settlements and linking trackways, south of Cirencester in the Upper Thames Valley 178 5.39 Distribution of sites with excavated and dated field systems in the 180 Central Belt 5.40 Plan of second century a.d. farmstead and associated field system at Tubney Wood Quarry, Oxon 181 5.41 Plan of field system at Eye Quarry, Cambs 181 5.42 Distribution and chronology of sites with evidence for lazybeds in the Central Belt region 182 5.43 Frequency of major artefact categories on all sites in the Central Belt region 183 5.44 Frequency of major artefact categories recovered across selected landscape zones in the Central Belt region 184 5.45 Frequency of samian, amphora and mortaria recovery within settlement types in the Central Belt region 185 5.46 Frequency of major artefact categories on different types of rural settlement in the Central Belt region 186

Romanprelims.indd 11

xi

5.47 Frequency of major artefact categories on villas and farmstead types in the Central Belt region 186 5.48 Occurrence of selected high status objects on ‘rich’ farmsteads and villas compared to other farmsteads in the Central Belt 186 5.49 Relative frequency of major livestock taxa across all Central Belt landscape zones 188 5.50 Relative frequency of major livestock taxa over time in the Central Belt, Cotswolds and Thames & Avons Vales 189 5.51 Relative frequency of major livestock taxa across different settlement types in the Central Belt 190 5.52 Percentage presence of cultivated plant taxa on sites in the Central Belt region 191 5.53 Percentage presence of cultivated plant taxa on sites in selected landscape zones in the Central Belt region 191 5.54 Percentage presence of cultivated plant taxa in (a) major settlement types and (b) farmstead types within the Central Belt Region 192 5.55 The Cambridgeshire Fen edge case study area 193 5.56 Huge area excavation of RomanoBritish rural settlement in North-West Cambridge 194 5.57 Chronological patterns of case study settlements 195 5.58 Settlement development in the case study area in the late Iron Age 196 5.59 Late Iron Age farmsteads on the Cambridgeshire claylands 196 5.60 Settlement development in the case study area in the early Roman period 197 5.61 Comparison of selected finds assemblages from ‘rich’ complex farmsteads at Langdale Hale and Vicar’s Farm 198 5.62 Settlement development in the case study area in the late Roman period with selected sites named in the text 199 5.63 Roman complex farmsteads at (a) Vicar’s farm, Cambridge, (b) Knobbs farm, Earith, and (c) Langdale Hale, Earith 200 5.64 Mid–late Roman farmstead at Fenstanton in relation to surrounding cropmarks in the lower Ouse Valley 201 5.65 National Mapping Programme ‘Iron Age and Roman’ cropmark data in relation to excavated sites to the north and west of Cambridge 202

15/09/2016 15:33:52

xii

THE RURAL SETTLEMENT OF ROMAN BRITAIN

5.66 Plans of excavated Roman agricultural landscapes at (a) Low Fen, Fen Drayton and (b) The Fields, along with (c) NMP cropmark of Iron Age/Roman field system in the Lower Ouse Valley 203 Chapter 6 6.1 6.2 6.3

6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 6.15

6.16 6.17 6.18 6.19 6.20 6.21 6.22

Romanprelims.indd 12

The East region in relation to modern county boundaries 208 Constituent landscape zones of the East region 209 Kernel density of East region records and all excavation records from National Monument Records (NMR) Index 210 ‘Roman’ NMP cropmark data in the Norfolk Broads/coast 211 Excavated late Iron Age/Roman rural settlement in the East region 213 Number of settlements in use over time in the East region 214 Variations in settlement chronology by selected landscape zone within the East region 214 ‘New’ and ‘abandoned’ settlements 215 over time in the East region All excavated farmsteads in the East region 216 Number of excavated farmsteads in use over time in the East region 216 Relative frequency of farmstead types in use over time in the East region 217 Development of the farmstead at Kilverstone, Thetford, Norfolk 217 Plans of complex farmsteads in Essex 218 Plans of enclosed farmsteads from 219 the East region Distribution of excavated villas in the East region, compared with ‘villas’ recorded in the NMR index 220 Chronological patterns of excavated villas in the East region 221 Plan of excavations south of the villa at Chignall, Essex 222 Distribution of nucleated settlement 223 in the East region Plan of Roman roadside settlement at Baldock, Herts 224 Plan of nucleated settlement to the east of the coastal fort at Brancaster, Norfolk 225 Distribution and frequency of rural buildings on excavated sites in the East region 226 Distribution and frequency of (a) circular and (b) rectangular architecture in the East region 226

6.23 Number of sites with circular and/or rectangular buildings over time in 227 the East region 6.24 Plans of late Roman (a) workshop and (b) granary/storehouse at Frogs Hall Borrow Pit, Takeley, Essex 227 6.25 Plans of farmstead buildings at (a) Strood Hall, Essex, (b) Melford Meadows, Norfolk and (c) Brandon Road, Norfolk 228 6.26 Plans of villa buildings from (a) Exning, Cambs and (b) Gayton Thorpe, Norfolk 230 6.27 Plan of villa complex south of Caistor, Norfolk 231 6.28 Plan of field system ditches radiating out from late Roman settlement at MTCP site, Stansted 233 6.29 ‘Roman’ field systems in the Norfolk Broads as revealed through NMP cropmarks 234 6.30 Frequency of major artefact categories on all sites in the East region 235 6.31 Frequency of major artefact categories on different types of rural settlement in the East region 236 6.32 Relative frequency of major livestock taxa across in the East region and in selected landscape zones 238 6.33 Relative frequency of major livestock taxa over time in the East region 238 6.34 Relative frequency of major livestock taxa across different settlement types in the East region 239 6.35 Percentage presence of cultivated plant taxa in farmsteads and roadside settlements within the East region 240 Chapter 7 7.1 7.2 7.3

7.4 7.5 7.6 7.7 7.8

The North-East region in relation to modern county boundaries 242 Constituent landscape zones of the North-East region 243 Kernel density of North-East region records and all excavation records from National Monument Records (NMR) Index 244 Excavated late Iron Age/Roman rural settlement in the North-East region 246 Number of settlements in use over time in the North-East region 247 Variations in settlement chronology by selected landscape zone within the North-East region 247 ‘New’ and ‘abandoned’ settlements over time in the North-East region 248 Site plans of Roman Ridge, West Yorkshire, showing (a) the late Iron Age/

15/09/2016 15:33:52

LIST OF FIGURES

7.9

7.10 7.11 7.12 7.13 7.14 7.15 7.16

7.17

7.18 7.19

7.20 7.21 7.22 7.23 7.24 7.25

7.26

7.27 7.28

7.29 7.30 7.31

Romanprelims.indd 13

early first-century field system, trackway and enclosures, and (b) the line of the Castleford to 248 Tadcaster Roman road Plan of the twin enclosure complex at Allerton Park Quarry, North Yorkshire 249 Distribution of all farmsteads in the North-East 250 Numbers of farmsteads in use in the North-East over time 251 Relative frequency of farmstead types over time in the North-East 251 Phase plans of High Wold, Bempton Lane, Bridlington 251 Plans of two types of complex 252 farmstead in the North-East Plan of Holmfield Interchange, Site Q 252 Plans of two late Iron Age/early Roman enclosed farmsteads integrated with a rectilinear field system 253 Site plans of Wattle Syke, West Yorkshire, showing (a) the late Iron Age–middle Roman enclosure complex (village), and (b) the late Roman (fourth century a.d.) open settlement 254 Phase plans of the settlement at Parlington Hollings, West Yorkshire 254 Distribution of excavated villas in the North-East compared to ‘villas’ recorded in the NMR index 255 Chronological patterns of excavated villas in the North-East 256 Plans of the villas at (a) Welton Wold and (b) Ingleby Barwick 257 Distribution of nucleated settlement and military sites in the North-East 258 Interpretive plan of the geophysical survey results at Hayton 258 Chronological development of nucleated settlement in the North-East 259 Plan of cropmarks showing the fort and vicus at Newton Kyme, North Yorkshire 260 Plan of cropmarks showing the late Iron Age/Roman ladder complex at Burton Fleming 261 Distribution of all buildings in the North-East 262 Distribution of circular and rectangular buildings in the North-East 263 Use of circular and rectangular buildings in the North-East over time 264 Use of circular and rectilinear buildings on farmsteads of different type 264 Plans of drystone masonry structures on the North York Moors 265

xiii

7.32 Post-built ‘structures at (a) Melton A63, (b) Cedar Ridge, Garforth, and (c) Stile Hill, Colton 266 7.33 Villa building forms in the North-East 266 7.34 Comparative plans of six villa houses 267 7.35 Plans of buildings at roadside settlements 269 7.36 Spot height analysis for major settlement types in the North-East 270 7.37 Plans of long distance trackways with associated field systems and 271 settlements 7.38 Plan of Romano-British ‘brickwork’ fields at Dunston’s Clump, Babworth 272 7.39 Recovery of samian and mortaria from different types of site in the North-East 274 7.40 Frequency of major artefact categories on different types of rural settlement 275 in the North-East region 7.41 Relative frequency of cattle, 277 sheep/goat and pig 7.42 Religious monuments at Ferry Fryston 278 7.43 Presence of arable crops and other plant taxa from roadside settlements/ vici and farmsteads in the North-East 279 Chapter 8 8.1

The Central West region in relation to modern county boundaries 282 8.2 Constituent landscape zones of the Central West region 283 8.3 Distribution of excavated Roman rural sites and all excavation records from National Monument Records (NMR) Index (in the Central West region 284 8.4 Distribution of excavated Roman rural sites in relation to builtup areas 285 8.5 Distribution of sites in the Central West by six sub-regions 286 8.6 Number of farming settlements in use over time in the Central West region 288 8.7 Number of farming settlements in use over time in selected sub-regions of the Central West 289 8.8 Establishment dates for farming settlements in the Coal Measures and the Mosses and Meres/Shropshire Hills/Wrexham area 289 8.9 Plans of enclosed farmsteads of rectilinear, irregular, D-shaped and curvilinear form 291 8.10 Plan of farmstead at Sharpstones Hill Site E, Shropshire 292

15/09/2016 15:33:52

xiv

8.11 Site plan of a mid-second to midthird century a.d. complex farmstead, north of Langley Mill, West Midlands 293 8.12 Sequence of development at 294 Bullerthorpe Lane, Swillington 8.13 Plan of buildings at Wilderspool, Warrington 295 8.14 Use of circular and rectangular buildings in the Central West over time 296 8.15 Use of circular and rectangular buildings on farmsteads in the Central West over time 296 8.16 Plan of building at Pentre Farm, Flint 297 8.17 Plan of a corridor villa at Hales, Staffordshire 297 8.18 Plan of a winged corridor villa at 297 Eaton-by-Tarporley, Cheshire 8.19 Spot height analysis of farmsteads and villas in the Central West region 298 8.20 Plan of a trackway leading to a midto late Roman farmstead at Billingley Drive, Thurnscoe, South Yorkshire 299 8.21 The Chester and Wroxeter case 300 study area 8.22 Chronology of farming settlements in the Chester and Wroxeter hinterlands 301 8.23 Frequency of major artefact categories on different types of settlement in the Chester and Wroxeter hinterlands 303 8.24 Frequency of major artefact categories on farmsteads in the Chester and Wroxeter hinterlands 303 8.25 Proportion of coins and brooches among all finds recorded by PAS in the Chester and Wroxeter hinterlands 304

316 North region 9.10 Distribution of all excavated 316 farmsteads in the North region 9.11 Plan of complex farmstead at Blagdon Park 2, Northumberland 317 9.12 Plan of complex farmstead at Poulton-le-Fylde, Lancashire 317 9.13 Plan of a single-ditched curvilinear enclosed farmstead at Crosshill, Penrith, Cumbria 318 9.14 Plan of a double-ditched rectilinear enclosed farmstead at Burradon, Northumberland 318 9.15 Distribution of farmsteads with single and double-ditched enclosures 318 in the North region 9.16 Plan of an irregularly shaped enclosed farmstead at Milking Gap, Northumberland 319 9.17 Distribution of farmsteads with rectilinear, curvilinear and irregular shaped enclosures in the North region 319 9.18 Use of circular and rectangular buildings on farmsteads and other rural sites in the North region over time 321 9.19 Spot height analysis on major settlement types within the North region 323 9.20 Distribution of excavated field 326 systems in the North region 9.21 Distribution of sites with agricultural tools in the North region 327 9.22 Distribution of excavated sites with features interpreted as paddocks/ stock enclosures 328

Chapter 9

Chapter 10

9.1 9.2 9.3

9.4 9.5 9.6 9.7 9.8 9.9

Romanprelims.indd 14

THE RURAL SETTLEMENT OF ROMAN BRITAIN

The North region in relation to modern county boundaries Constituent landscape zones of the North region Distribution of excavated Roman rural sites and all excavation records from National Monument Records (NMR) Index in the North region Distribution of excavated late Iron Age/Roman rural settlements in the North region Number of settlements in use over time in the North region Number of vici in use over time in the North region Number of farmsteads in use over time in the North region Number of farmsteads in use over time in the North region, north and south of Hadrian’s Wall Farmstead morphology in the

308 309

310 312 313 314 314 315

10.1 The South-West region in relation to modern county boundaries 331 10.2 Constituent landscape zones of the South-West region 332 10.3 Distribution of excavated Roman rural sites and all excavation records from National Monument Records (NMR) Index in the South-West region 333 10.4 Distribution of excavated late Iron Age/Roman rural settlement in the South-West region 336 10.5 Number of farmsteads in use over time in the South-West region 337 10.6 ‘New’ and ‘abandoned’ farmsteads over time in the South-West region 338 10.7 Number of farmsteads in use over time in Cornwall and Devon 339 10.8 Continuity, abandonment and the establishment of new farmsteads over 339 time in Cornwall and Devon

15/09/2016 15:33:52

LIST OF FIGURES

10.9 Map showing the geographical shift in focus of settlement over time 340 10.10 Distribution of farmsteads over time in the eastern part of the South-West region 341 10.11 Plans of four Cornish ‘rounds’ showing the variation in their form 343 10.12 Plans of Courtyard House settlements from Cornwall 344 10.13 Plans of enclosed farmsteads from Devon 345 10.14 Plan of multi-roomed, timber, domestic building from the possible roadside settlement at Topsham 346 10.15 Plans of villas at (a) Magor Farm, Illogan, Cornwall and (b) Crediton, Devon 347 10.16 Plan of a timber-built circular building 348 from East Worlington, Devon 10.17 Plan of circular and oval buildings within the round at Threemilestone, Kenwyn, Cornwall 348 10.18 Spot height analysis on farmsteads/ villages in Devon and Cornwall 350 10.19 Distribution of farms and villages in the South-West region, identifying those situated at under 50 m elevation 351 10.20 Distribution of late Iron Age and Roman farmsteads/villages and excavated sites of other periods from the South-West region shown in relation to modern soil geochemistry 352 10.21 Plan of earthwork remains of a probable field system associated with an enclosed farmstead at Stoke Gabriel, Devon 353 10.22 Frequency of major artefact categories on farmsteads and villages in Devon and Cornwall 355 10.23 Average frequency of major classes of finds at hillforts and farmsteads/ villages in Cornwall 356 10.24 Frequency of major artefact categories on hillforts and farmsteads/villages in Cornwall 357 Chapter 11 11.1 Upland Wales and the Marches in relation to the England/Wales border, English counties and the preserved counties of Wales 359 11.2 Regional sub-divisions used for analysis in Upland Wales and the Marches 360 11.3 Distribution of excavated Roman rural sites in Upland Wales and the Marches 361

Romanprelims.indd 15

xv

11.4 Plan of the nucleated settlement at 362 Tai Cochion, Anglesey 11.5 Plan of the ‘village’ at Gateholm, Pembrokeshire 363 11.6 Number of farmsteads in use over time in Upland Wales and the Marches 365 11.7 Number of farmsteads in use over time in the three sub-regions of Upland Wales and the Marches 366 11.8 Establishment dates for farmsteads in Upland Wales and the Marches 366 11.9 Distribution of farmsteads in Upland Wales and the Marches established in the late Iron Age, later first century a.d. and 368 second century a.d. 11.10 Sub-regional variation in the form of rural settlement enclosures at farmsteads in Upland Wales and the Marches 369 11.11 Sub-regional chronology of enclosure forms 369 11.12 Plans of settlement enclosures from South-West Wales 370 11.13 Plans of settlement enclosures in North Wales 371 11.14 Plans of settlement enclosures in East Wales and Marches 372 11.15 Use of circular and rectangular buildings in Upland Wales and the Marches over time 373 11.16 Distribution of sites in Upland Wales and the Marches with (a) timber circular buildings, and (b) stone circular buildings 374 11.17 Distribution of site types in Upland Wales and the Marches with (a) timber rectangular buildings, and (b) masonry rectangular buildings 375 11.18 Plans of villa buildings at Abermagwr and Cwmbrwyn, Dyfed 377 11.19 Spot height analysis on farmsteads in Upland Wales and the Marches 378 11.20 Distribution of excavated sites in Upland Wales and the Marches with associated field systems 379 11.21 Frequency of major artefact categories on farmsteads in North Wales and South-West Wales 381 11.22 Frequency of all pottery, samian, amphora and mortaria recovery within farmsteads in North Wales and South-West Wales 382 Chapter 12 12.1 Density of records across the different regions of the project 387

15/09/2016 15:33:53

xvi

THE RURAL SETTLEMENT OF ROMAN BRITAIN

12.2 Kernel density of records from England and Wales and all excavation records from National Monument 388 Records (NMR) Index) 12.3 Distribution map of ‘Iron Age/Roman’ settlements in England recorded as earthworks, and cropmarks 389 and soilmarks 12.4 Kernel density of late Iron Age and Roman artefacts recorded by the Portable Antiquities Scheme (PAS) across England and Wales 390 12.5 Modern soil pH values for England and Wales 391 12.6 Proportion of different settlement types within the project regions 392 12.7 Relative frequencies of farmstead types by project region 393 12.8 Distribution of farmstead types over time 395 12.9 Percentages of excavated farmsteads with indication of pottery present by project region 396 12.10 Density of quantified late Iron Age and Roman pottery on farmsteads across England and Wales 397 12.11 Frequency of selected artefact categories on farmsteads in different 398 project regions 12.12 Distribution of animal bone assemblages 399 across England and Wales 12.13 Relative frequency of cattle, sheep/ 399 goat and pig by project region 12.14 Distribution of sites with corndryers, shown at their most expansive period in the later third century a.d. 400 12.15 Distribution of excavated late Iron Age and Roman field systems and trackways, across England and Wales 401

Romanprelims.indd 16

12.16 Comparison of settlement chronology by southern and eastern regions 12.17 Comparison of settlement chronology by northern and western regions 12.18 Kernel density plot of excavated settlements dating to the late Iron Age and early Roman period 12.19 Kernel density plot of excavated settlements dating to the middle and late Roman period 12.20 Settlement ‘establishment’ and ‘abandonment’ during the early post-Claudian conquest period 12.21 Settlement ‘establishment’ and ‘abandonment’ during the second century a.d. 12.22 Settlement ‘establishment’ and ‘abandonment’ during the third century a.d. 12.23 Settlement ‘establishment’ and ‘abandonment’ during the fourth century a.d. 12.24 Excavated late Roman settlements with indications of continued activity into at least the early fifth century 12.25 Percentage of site types located within 20 km of a major RomanoBritish town 12.26 Principal types of settlement in use over time within 20 km of the Roman colonia at Gloucester

405 405

406

407

409

411

412

413

415

418

419

Appendix 2 A2.1 Distribution of excavated late Iron Age and Roman rural sites, shown as points (a) and kernel density of excavated late Iron Age and Roman rural sites 424

15/09/2016 15:33:53

LIST OF TABLES Chapter 1 1.1

Classification of site types on Roman Rural Settlement Project database

7.2 10

7.3

Chapter 2 2.1 2.2

Chapter 8

Quantification of excavated rural 18 settlements by type Numbers of single, double and tripleditched enclosures by region 27

8.1

8.2

Chapter 3 3.1 3.2

Recorded characteristics of buildings Buildings interpreted as workshops: associated site types and industries

4.2 4.3 4.4

8.3

61

Chapter 9 9.1

Number of settlement types by 80 landscape zone in the South region Proportion of farmsteads classified over time in the South region 87 Buildings in the South region 103 Proportions of the three main domesticates from sites in the Middle 134 Thames Valley case study area

9.2 9.3 9.4 9.5

Chapter 5 5.1 5.2 5.3 5.4 5.5

6.2 6.3 6.4

311 312 320 321

326

10.1 Number of excavated sites and density (per km²) by landscape zone in the South-West region 334 10.2 Number of settlement types in each landscape zone in the South-West region 335 10.3 Numbers and proportions of objects of different functional types recorded by the Portable Antiquities Scheme from Devon and Cornwall 355 Chapter 11

Number of excavated sites and density (per km²) by landscape zone in the 209 East region Number of settlement types by landscape zone in the East region 212 Buildings in the East region 228 Distribution of selected object categories across landscape zones of the East region 235

11.1 Number of excavated sites and density (per km²) by sub-region in Upland Wales and the Marches 11.2 Number of settlement types by sub-region in Upland Wales and the Marches

361

362

Chapter 12 12.1 Selected sample of radiocarbon dates from sites with late Roman activity suggesting possible activity in the post-Roman period

Chapter 7 7.1

Number of excavated sites and density (per km²) by landscape zone in the North region Number of settlement types in each landscape zone in the North region Buildings in the North region Building material at farmsteads in the North region Presence of archaeobotanical evidence at rural sites in the North region

Chapter 10

Number of excavated sites and density (per km²) by landscape zone in the Central Belt region 144 Number of settlement types by landscape zone in the Central Belt region 146 Buildings in the Central Belt region 169 Villa architecture within the Central 173 Belt region Quantity of object types by settlement category 187

Chapter 6 6.1

Number and density of sites in the Central West arranged by sub-regions 286 Major site types in the Central West region displayed by sub-region 287 Buildings in the Central West region 295

46

Chapter 4 4.1

245 zone in North-East region Number of settlement types by landscape 246 zone in the North-East region Buildings in the North-East region 265

Number of excavated sites and density (per km²) by landscape

415

xvii

Romanprelims.indd 17

15/09/2016 15:33:53

PREFACE By Michael Fulford and Neil Holbrook This is the first of three volumes in a series entitled New Visions of the Countryside of Roman Britain. These books represent one of the main products of a collaboration that dates back to 2006. In that year we agreed to work together to create a project which sought to define and evaluate the contribution that investigations undertaken as part of the planning process made to academic knowledge of Roman Britain. Our hope was that the project would play a major part in addressing concerns that were being increasingly aired about the dislocation between archaeologists charged with the investigation of landscapes in advance of development and those involved in teaching and research (who are largely based in universities). The dramatic increase in the volume of archaeological fieldwork as a consequence of the introduction of new planning guidance in England in 1990, and Wales in 1991, has been well rehearsed (for instance Darvill and Russell 2002, 52, and below in Ch. 1). This guidance articulated a clear presumption in favour of the physical preservation of archaeological remains that would be affected by a proposed development, but, where it was decided that this was not an appropriate outcome, it put responsibility on developers, rather than the state, to fund prior archaeological recording and the subsequent analysis and dissemination of the results. As the amount of archaeological work undertaken in the UK increased in the 1990s, so did concerns about the fragmentation of the archaeological discipline and, in particular, the realisation that the enormous quantity of new data being generated on an almost daily basis by archaeological contracting organisations was having very little impact on how British archaeology was being taught and researched in Universities (for instance, Olivier 1996; James and Millett 2001, 1; and for the fact that these concerns had not gone away, Southport Group 2011, 14–17). This lack of cohesion was due to a number of factors, not least the unfamiliarity of the products of commercial investigations, which can be divided into two categories: fieldwork investigations, which normally occur prior to the determination of a planning application (‘evaluations’), and those secured as a condition of consent (‘postdetermination’ work, so-called because it occurs after the determination of planning applications). The former aim to characterise the archaeological deposits present and inform decisions on their

management; they use techniques such as surface collection, geophysical survey and trial trenching. The latter are designed to make a record of deposits prior to destruction and involve methods such as open-area excavation, strip-and-record sample excavation and watching brief. The outcomes of evaluations and small-scale postdetermination works such as watching briefs are unpublished reports produced in very small numbers that are normally deposited in the local Historic Environment Record (HER). It is generally envisaged that the results of more substantial pieces of post-determination work to mitigate the effects of development will be published in conventional ways, although sometimes little more than a note may appear in a county journal. For larger excavations it is commonplace to compile a post-excavation assessment report that seeks to summarise what has been found (features, finds and environmental evidence), assess the potential of these data to address specific research questions, and propose an appropriate level of further analysis to be contained in the final published report. The reports that document the results of the various investigations described above have been termed grey literature, which we can characterise as unpublished reports produced in small numbers and with very limited distribution. Until comparatively recently grey literature reports could only be consulted by visiting an individual county or district HER in person, although the development of the on-line library of unpublished fieldwork reports hosted by the Archaeology Data Service (ADS) and Archwilio in Wales has had a profound impact on increasing the accessibility of grey literature (over 34,000 reports were available for download from the ADS in December 2015). For the two decades following 1990, however, many researchers decided, not unsurprisingly, that they would get a better return on their limited research time by restricting themselves to conventionally published, and thus more generally available, works (Fulford 2011). A number of major national syntheses published in that period therefore made little use of the new evidence generated by commercial archaeology (for instance, Cunliffe 2005; Mattingly 2006). Some academics did, however, rise to the challenge presented by the myriad of new evidence, and research students began to engage with this source on a more regular basis (early examples of

xviii

Romanprelims.indd 18

15/09/2016 15:33:53

PREFACE

published works that made extensive use of grey literature include Moore 2006 and Yates 2007). Richard Bradley was one of the leading early pioneers in this area (Phillips and Bradley 2004) and he published an influential paper in 2006 which concluded that syntheses concerned with prehistoric Britain and based purely on conventionally published data contain serious lacunae in a number of important areas (Bradley 2006). It was against this backdrop that we developed a proposal to English Heritage (now Historic England) for a project entitled Assessing the Research Potential of Grey Literature in the Study of Roman England. The concept was to examine the research dividend that could be gained from a study of grey literature relating to investigations that have discovered Roman period remains in England, and investigate ways of bridging the gap between individual typescript reports held in the HERs and overarching regional or national syntheses. The project was conceived from the outset as a partnership between academia and the commercial sector, and was designed in three stages. Stage 1 was concerned with a rapid national overview of how much work had been done between 1990 and 2004, where it was located, and an assessment of what proportion of grey literature had reached conventional publication (Holbrook and Morton 2011). Stage 2 targeted four pilot areas (Essex, Somerset, South and West Yorkshire combined and Warwickshire) for a more detailed assessment of the research potential of the grey literature (Holbrook 2010; 2011; Hodgson 2011; 2012). This work suggested that about 9000 separate interventions had encountered Roman archaeology in England in the period 1990–2010 and that, while gains in knowledge had been uneven, the evidence pertaining to the Romano-British countryside had the highest research potential both in terms of its quantity and its potential for hitherto unattempted synthesis (Fulford and Holbrook 2011). Stage 3 was always envisaged as a national survey of England (and hopefully Wales as well), but the results of Stages 1 and 2 had highlighted the considerable level of funding necessary to achieve this, even if analysis was restricted solely to the Roman countryside. The costs of the national survey were beyond those that could be met by the resources available to English Heritage alone, and so a grant proposal was developed jointly to both English Heritage and the Leverhulme Trust in 2011. The proposal was accepted and work commenced in March 2012 on a study entitled The Evaluation of PPG16, ‘Grey’ Literature and the Rural Settlement of Roman Britain. Initially the area of study was restricted to England,

Romanprelims.indd 19

xix

but, thanks to a second grant from the Leverhulme Trust in 2015, we were able to expand the area under consideration to include Wales. Even with this level of funding the scale of the project was daunting, and a number of pragmatic decisions were made. The major towns of Roman Britain always lay outside the remit of this study, although an evaluation of the contribution of commercial archaeology to knowledge of these important urban centres has been published elsewhere (Fulford and Holbrook 2015). The evidence from the so-called ‘small towns’ was a challenge to accommodate within the resources available, and a purely pragmatic decision was taken to exclude the defended small towns. The arbitrariness of this division was always apparent to us, and we were able to rectify the situation in 2015 thanks to a generous donation from Mr Paul Chadwick that allowed us to collect data from the walled small towns. This evidence was not available in time to inform the analyses presented in this volume, although it will be included in the studies contained in the subsequent ones. The methods adopted by the project are set out in Chapter 1, but, from the outset, one of our major objectives was to actively engage with regional audiences who had an interest in the project and its outcomes (such as those responsible for specifying and undertaking commercially funded investigations, academics and local interest groups). We were always clear that we did not wish to collect data in a vacuum without an opportunity to feedback and discuss our emerging ideas. We therefore held seven regional meetings between March 2013 and December 2014 in Cambridge, Leicester, London, Birmingham, Exeter, York and Durham, each attended by between 50–100 people, with a national meeting in Reading in April 2015 to launch the project database and a meeting in Cardiff to discuss the findings in Wales in November 2015. These meetings were dynamic events that proved extremely valuable in raising awareness of the project, winning local support and generating engagement between the project team and those with local and regional expertise. They also provided a forum to critically examine aspects of contemporary professional practice in the investigation, analysis and publication of work on Romano-British rural settlements. The value of such events is a lesson from this project that future national surveys might usefully consider. This series of volumes is a timely survey of the present state of knowledge of the Romano-British countryside. There is already a pressing need to demonstrate to the wider community the extent and quality of the knowledge gain stimulated by the implementation of the new planning policies and the public benefit that accrues from the

15/09/2016 15:33:54

xx

THE RURAL SETTLEMENT OF ROMAN BRITAIN

considerable sums invested by developers in the investigation of late Iron Age and Roman rural settlements impacted by development. Why should national and local government require developers to continue to fund this work if its public benefit cannot be appreciated? Indeed the true potential of work done since 1990 cannot be fully realised without a synthetic project of this kind. At just over £1 million, the cost of this project represents less than half of one per cent of the estimated £200 million or so spent by developers on investigations relating to the rural settlement of Roman Britain in the period 1990–2010. In those terms expenditure of this order to make sense of much of what has gone on surely represents value for money. This project tests a number of hypotheses relating to the Romano-British countryside through an examination of the excavated evidence. Our approach has been an integrative one, which gives equal weight to investigations reported solely in grey literature compared to those disseminated in more conventional formats. Through a consideration of settlement archaeology, finds and environmental data our overarching objective is to produce a new characterisation of the RomanoBritish countryside, and in this respect this project can be considered as one of a clutch of so-called ‘Big Data’ projects started, and in some cases completed, over the last few years. These include the Fields of Britannia project at the University of Exeter (Rippon et al. 2015); English Landscapes and Identities (‘EngLaid’) at the University of Oxford and Hoarding in Iron Age and Roman Britain at the University of Leicester and the British Museum. In all these projects the hope has been that, through the examination of particularly large datasets, new patterning will be identified that might not be apparent with smaller samples. Such projects pose their own particular challenges, not least that the size of the datasets does not permit a critical evaluation of individual records. For example, in this project it has not been feasible to re-evaluate the reliability of the site chronologies presented in the original reports. Undoubtedly in some cases the dating suggested by the excavators might at best be questionable or at worst plain wrong. But to have done this for all of the c. 2500 sites would have been a truly Herculean task. The expectation is, however, that, through amalgamating numerous interventions and looking to the bigger picture, individual biases will fade into the background and broader trends will shine through. Another major challenge with Big Data projects is to recognise factors that might bias the data recovered, and to evaluate the degree to which patterning is a true reflection of past activities rather than a product of modern ones.

Romanprelims.indd 20

As this project is concerned with excavated evidence, a fundamental consideration is the degree to which spatial patterns apparent in the data are a product of the complex set of factors that dictate where archaeologists excavate. And these have not been constant over time. Whereas in the nineteenth and most of the twentieth centuries much attention was focused on villas, due both to their high visibility in the landscape and the prevailing research interests of the time, from the second half of the twentieth century, and especially from 1990 onwards, the predominant factor in determining the location of archaeological work has related to where development was to take place. The project team is acutely aware of this issue and in each regional chapter they have sought to critically evaluate the causes behind spatial patterning. While the number of sites investigated varies considerably between regions, in every case the authors have been able to break away from previous characterisations based upon a small number of ‘type-sites’ and are able to present a much more subtle and nuanced analysis that draws on the richness of the data now available. As stated above, this is the first in a series of three books. This volume considers the settlement evidence, principally the morphology of rural settlements and the architecture of their constituent buildings. Conclusions are drawn on a variety of topics including the influence of existing patterns of late Iron Age settlement on later arrangements; regional diversity in settlement form and location; variation in the chronologies of rural settlements in different parts of the country; the interplay between landscape and cultural influences on settlement, and the effects that proximity to urban centres and roads had on settlement location. In many respects this volume sets the framework for the further studies that will be provided in volumes 2 and 3, and, for that reason, sizeable parts of this book are necessarily descriptive. What this volume does not do is to offer any firm conclusions on overall settlement densities in Roman Britain as the excavated evidence is but one source for this analysis (which is better accomplished through extrapolation from extensive regional surveys and estimations from all currently known late Iron Age or Romano-British settlements contained in sources such as Jeremy Taylor’s 2007 study or Historic England’s AMIE (Archives and Monuments Information England) database. There are a few areas, however, such as the environs of Cambridge and the Upper Thames Valley east of Cirencester, where development has been preceded by so many archaeological interventions that we reach the nearest approximation to a total excavated landscape.

15/09/2016 15:33:54

PREFACE

The second volume in the series will be concerned with the rural economy of Roman Britain and examine themes associated with agriculture, industry, transport and markets, drawing heavily on the environmental and artefactual evidence. Volume 3 will examine the rural population of the countryside, and the evidence for their rituals and religion. It will look at the visibility in the record for the multifarious identities that made up the province, and the considerable evidence for religious expression, including regional and chronological variations in burial practice. All three books are underpinned by the project database hosted by the Archaeology Data Service, which is described in Chapter 1 and Appendix 1. Launched in April 2015 this has already proved to be a valuable and popular resource: the website received over 6000 visits and 7000 downloads of reports and site plans in the first seven months following its release. The great beauty of the database is that it permits individual researchers to undertake their own analyses and follow their own research directions. Given the time-limited funding of this project there is no mechanism to update the database with new discoveries made since 2012/13. However, given

Romanprelims.indd 21

xxi

the linkages implicit in its design it should be possible for individual HERs to build upon our work and they will remain the primary source of information on the results of work undertaken within the planning system. While we have spent much time in this project reviewing what has been learnt, it is equally important that we also look ahead as the archaeological and land-use planning communities need to learn from what has been achieved in the first 25 years of commercial archaeology, and how it has been achieved, in order to influence best practice in the future. This relates not only to the sampling strategies determined by curatorial archaeologists for the areas and volumes to be excavated, but also the way material culture and environmental data are recovered, quantified and reported. This is a key moment to evaluate past practice and set down benchmarks for the future. We will present our thoughts on the effectiveness of methodologies used to examine and report on rural settlements in a separate paper in 2017, which will draw heavily on the thought-provoking discussions that formed such a vital part of the regional meetings.

15/09/2016 15:33:54

ACKNOWLEDGEMENTS The expansion of the project into Wales was facilitated by the support of Kate Roberts of Cadw and Chris Martin, and the inclusion of the walled small towns was made possible by the generosity of Paul Chadwick. The project was managed for Cotswold Archaeology by Nathan Blick with exemplary efficiency, and Rob Skinner played a vital role in liaising with the many HER officers. He also dealt with numerous organisations responsible for producing the grey literature, and the willingness of these contractors to allow their reports to be disseminated through the project website was a major boost to us.The data collected was assimilated and entered into the project database by Martyn Allen, Tom Brindle and Alex Smith, and more latterly by Lisa Lodwick. The on-line project database was developed by the Archaeology Data Service at the University of York, and we are grateful to Julian Richards, Catherine Hardman and, especially, Tim Evans who was crucial to making the website such a powerful and valuable resource. Tim’s presentations on the database were an important part of the regional meetings. Our main debt, however, is to the three principal authors of this volume who were employed at the University of Reading to undertake the Leverhulme project: Senior Research Fellow Alex Smith and Research Fellows Martyn Allen and Tom Brindle. Their expertise, enthusiasm and sheer hard work over the last three years has made the project what it is. Earlier drafts of sections of this volume were commented upon by the regional experts named above, and we are also grateful to Paul Bidwell, editor of the Britannia Monograph series, for his assistance in bringing it to publication. The illustrations were produced by Sarah LambertGates and Daniel Wheeler, with the large number of maps produced by Tom Brindle. Copy editing was undertaken by Val Kinsler, 100% Proof, and the volume was guided through the press for the Roman Society by Lynn Pitts with her normal exemplary efficiency.

The research that underpins this report would not have been possible without the generous grant awarded by Historic England to Neil Holbrook at Cotswold Archaeology and by the Leverhulme Trust (RPG-417) to Professor Michael Fulford at the Dept of Archaeology, University of Reading and to Professor Julian Richards at the Archaeology Data Service, University of York. Given the collaborative nature of this project there are many people and institutions who have made significant contributions to our work. Barney Sloane, Head of Strategic Planning and Management Division at Historic England, supported the project from the outset and was instrumental in obtaining the necessary resources. His colleagues Kath Buxton, Roger Thomas and Pete Wilson also assisted the project in various ways. At a time when local authority historic environment services were subject to severe funding restrictions we were acutely aware that the success of the project would rest in no small measure on the ability of HER staff to supply digital copies of grey literature reports. We therefore developed a close relationship with various leading members of the Association of Local Government Archaeological Officers (ALGAO), principally, Dave Barrett, Stewart Bryant, Fiona Gale and Fiona MacDonald. Their enthusiastic support from the outset, coupled with that of their colleagues in individual HERs, enabled us to interact successfully with all 92 HERs in England and Wales, a truly excellent outcome. Stewart also represented ALGAO on the project steering committee. We benefited greatly from the involvement of a series of regionally based experts who were influential participants at the open meetings and also reviewed an early draft of the chapters in this volume that covered their areas of knowledge: Paul Booth (London and the South East); Nick Cooper (East Midlands); Christopher Evans (Eastern); Peter Guest (Wales); Nick Hodgson (North East and North West); Ian Roberts (Yorkshire and Humber), and Andy Wigley (West Midlands). We thank them for putting their knowledge at our disposal. The project benefited greatly from the wise counsel afforded by the project steering committee under the skilful chairmanship of Stephen Rippon.

Michael Fulford, University of Reading Neil Holbrook, Cotswold Archaeology December 2015

xxii

Romanprelims.indd 22

15/09/2016 15:33:54

SUMMARY It has often been stated that Roman Britain was fundamentally a rural society, with the vast majority of the population living and working in the countryside. Yet there was clearly a large degree of regional variation, and with the mass of new data produced since the onset of developerfunded archaeology in 1990, the diversity of Roman-period rural settlement across the landscape can now be demonstrated. Drawing on the evidence from c. 2500 excavated settlements, this volume presents a new framework of eight regions for the study of rural Roman Britain, in which has been developed a characterisation of the mosaic of communities that inhabited the province and the way that they changed over time. Centre stage is the farmstead, rather than the villa that has for so long dominated discourse in the study of Roman Britain. These farmsteads exhibit substantial regional variation, from the enclosed ‘rounds’ evident in parts of Cornwall to the rectangular, dry-walled enclosures of north-west Wales and large enclosure complexes of the central English river valleys. Farmsteads have been classified through site morphology (using site plans), with further exploration of their associated buildings, surrounding landscape context and wealth of material culture and environmental evidence. The physical, social and economic relationships between farmstead types and other forms of settlement (villas, military vici, villages and roadside settlements) have been explored, allowing the demonstration of distinct variation across space and time. Many of the regional differences would appear to relate to the variable settlement patterns and social structures of pre-

Roman Britain, but there is little evidence to corroborate existing understanding of the territories of Iron Age ‘tribes’ and subsequent Roman administrative regions, or civitates. The increased emphasis on farmsteads, as the most numerous type of rural settlement, has had a major impact on our understanding of chronological patterning in the Roman countryside. In particular the second century emerges as the period when most settlements were in use, and is therefore arguably the point at which the population was at its height, drawing us away from previous assertions of a ‘golden age’ of rural expansion in the late Roman period. There is, nevertheless, a notable shift in the overall focus of settlement over time away from the south-east, with parts of central and western England maintaining relatively high numbers of settlements into the third and fourth centuries a.d. These included many large farmsteads and sites that developed architectural characteristics which would define them as villas – all part of the increasing wealth that appears concentrated in this region, which could readily be described as the ‘bread basket’ of Roman Britain. The analysis presented in this volume has been primarily devoted to gaining a better understanding of the nature, form and development of Roman rural settlements, while a further two volumes will focus on the economy and the peoples and rituals of rural Roman Britain. All volumes take an integrated approach, utilising many different strands of evidence in order to breathe new life into our understanding of the Romano-British countryside.

xxiii

Romanprelims.indd 23

15/09/2016 15:33:54

xxiv

THE RURAL SETTLEMENT OF ROMAN BRITAIN

RÉSUMÉ Il a souvent été déclaré que la Bretagne romaine était une société fondamentalement rurale, dont la vaste majorité de la population vivait et travaillait à la campagne. Pourtant, il existait manifestement d’importantes variations régionales et, grâce à la masse de nouvelles données générées depuis le début de la privatisation de l’archéologie en 1990, il est à présent possible de démontrer la diversité de l’habitat rural de l’époque romaine à travers le paysage. En puisant parmi l’ensemble des témoignages issus de la fouille d’environ 2500 habitats, ce présent volume permet de présenter un nouveau cadre de 8 régions pour l’étude de la Bretagne rurale romaine, un cadre au sein duquel s’est développée une caractérisation de la mosaïque des communautés qui ont peuplé la province et la manière dont ces dernières ont changé au fil du temps. La ferme est au devant de la scène, plutôt que la villa qui, elle, a dominé le discours de l’étude de la Bretagne romaine depuis fort longtemps. Ces fermes présentent des variations régionales notables, allant des enclos circulaires évidents dans certaines parties de la Cornouaille jusqu’aux enclos rectangulaires en pierre sèche du nord-ouest du Pays de Galles et aux complexes de grands enclos des vallées fluviales centrales britanniques. Les fermes ont fait l’objet d’une classification selon la morphologie du site (utilisation de plans de site), à laquelle s’est ajoutée l’exploration plus exhaustive des bâtiments qui leur sont associés, du contexte du paysage environnant et de la richesse de la culture matérielle et des restes environnementaux. Il a ainsi été possible d’explorer les liens physiques, sociaux et économiques existant entre les types de fermes et autres formes d’habitats (villas, vici militaires, villages et habitats en bordure de route), ce qui a permis de démontrer les degrés de changement divers à travers l’espace et dans le temps. Nombre des différences régionales notées

Romanprelims.indd 24

sembleraient à la fois relever de l’agencement des habitats et des structures sociales variables de la Bretagne préromaine, mais il existe peu de preuves permettant de corroborer ce que nous comprenons des territoires des « tribus » de l’âge du fer avec les régions administratives romaines qui ont suivi, ou civitates. L’accent croissant mis sur les fermes, comme type d’habitat rural le plus représenté, a eu un impact majeur pour comprendre la structuration chronologique au sein de la campagne romaine. En particulier, le IIème siècle émerge comme le moment auquel la plupart des habitats étaient en usage, et donc sans doute une période durant laquelle la population était à son maximum, ce qui contredit les précédentes affirmations d’un « âge d’or » de l’expansion rurale à l’antiquité tardive. Il faut noter, néanmoins, un déplacement notable de l’objectif d’ensemble de l’habitat au fil du temps marqué par un éloignement du sud-est, avec des parties de l’Angleterre centrale et ouest maintenant un nombre d’habitats relativement élevés durant les IIIème et IVème siècles après J.-C. Il s’agissait de nombreuses grandes fermes et sites dont les traits architecturaux les auraient communément définies comme des villas – tout ceci faisant partie intégrante de la richesse qui semblait être concentrée dans cette région et que l’on pourrait décrire comme le « grenier à blé » de la Bretagne romaine. L’analyse présentée dans ce volume a pour vocation première de mieux appréhender la nature, la forme et l’évolution des habitats ruraux romains, tandis que les deux suivants se focaliseront sur l’économie puis les peuples et rituels de la Bretagne rurale romaine. Les trois ouvrages ont une approche intégrée et utilisent de nombreux éléments distincts de témoignages afin d’améliorer notre compréhension de la campagne romanobritannique.

15/09/2016 15:33:55

SUMMARY

xxv

ZUSAMMENFASSUNG Es wird oft behauptet, dass das römische Britannien grundsätzlich eine bäuerliche Gesellschaft war, in welcher der Großteil der Bevölkerung auf dem Land lebte und arbeitete. Doch bleibt festzuhalten, dass es sicherlich große regionale Unterschiede gab, und mit der Datenmasse die seit 1990, mit der Einführung der vom Bauträger finanzierten Archäologie gewonnen wurde, kann nun die gesamte Vielfalt römischer, ländlicher Siedlungsweisen demonstriert werden. Auf dem Hintergrund von ca. 2500 ausgegrabenen Siedlungen präsentiert der vorliegende Band ein Gerüst von acht Regionen für die Forschung des ländlichen römischen Britanniens, in welcher eine Charakterisierung der Gemeinden, die wie Mosaiksteine die Provinz ausmachten, entwickelt wurde und es wird aufgezeigt, wie diese sich mit der Zeit verändert haben. Im Mittelpunkt steht der Bauernhof und nicht etwa die Villa, die so lange den Diskurs der Forschung des römischen Britanniens dominiert hat. Diese Höfe weisen wesentliche regionale Variationen auf, von den umschlossenen „runden“ in Teilen Cornwalls über die rechteckigen Trockenmaueranlagen in Nordwest Wales zu den großen eingefriedeten Anlagen der zentral englischen Flusstäler. Die Bauernhöfe wurden durch Morphologie von Grabungen (Untersuchung der Grabungspläne) klassifiziert. In die Klassifizierung flossen Informationen assoziierter Gebäude, der Kontext umgebender Landschaften, der durch Funde abzuleitende Wohlstand und Umweltbedingungen ein. Die physischen, sozialen und ökonomischen Beziehungen zwischen den Hoftypen und anderen Arten von Siedlungen (Villas, militärische vici, Dörfer und Straßensiedlungen) wurden untersucht. Dies ermöglichte eine Darstellung der Variationen durch Raum und Zeit. Viele regionale Unterschiede scheinen in Zusammenhang mit den variablen Siedlungsmustern und sozialen

Romanprelims.indd 25

Strukturen des vorrömischen Britanniens zu stehen, doch gibt es nur wenige Nachweise die unser Verständnis eisenzeitlicher Stammesterritorien und nachfolgender römischer Verwaltungsgebiete oder civitates bestätigt. Dass der Schwerpunkt auf Bauernhöfe, die häufigste ländliche Siedlungsart, gelegt wurde, hatte einen erheblichen Einfluss auf unser Verständnis der chronologischen Musterung der römischen Landschaft. Insbesondere das 2. Jahrhundert sticht als die Periode, in der die meisten Siedlungen in Betrieb waren, heraus und ist somit vermutlich die Zeit zu welcher die Bevölkerungsdichte am höchsten war. Dies steht im Gegensatz zu einer früheren Annahme, die in der spätrömischen Periode das „goldene Zeitalter“ der ländlichen Ausdehnung sah. Es gibt dennoch eine merkbare Verlagerung des allgemeinen Schwerpunktes durch die Zeiten, hinweg vom Südosten, währendTeile Zentral- undWestenglands relativ hohe Siedlungszahlen im 3. und 4. Jahrhundert aufrechterhalten. Dies beinhaltet mehrere große Höfe und Standorte, welche architektonische Charakteristiken entwickelten, die sie als Villen definieren würden – alle Teil des zunehmenden Wohlstandes der sich in dieser Region zu konzentrieren scheint, welche ohne weiteres als die „Kornkammer“ des römischen Britanniens bezeichnet werden kann. Die in diesem Band vorgelegten Untersuchungen konzentrieren sich im Wesentlichen darauf ein besseres Verständnis der Natur, Form und Entwicklung römischer ländlicher Siedlungen zu erhalten, während zwei weitere Bände sich mit der Ökonomie und den Menschen und den Ritualen des römischen Britanniens beschäftigen werden. Alle Bände legen eine ganzheitliche Betrachtung vor und nutzen unterschiedliche Beweisführungen um neues Leben in unser Verständnis der römischbritischen Landschaft zu bringen.

15/09/2016 15:33:55

Romanprelims.indd 26

15/09/2016 15:33:55

Chapter 1

Introduction By Michael Fulford and Tom Brindle

INTRODUCTION

past human activity (Darvill and Fulton 1998, 63–5). With the publication of A Matter of Time (RCHME 1960), the Royal Commission on the Historical Monuments of England had taken the lead in the systematic mapping of the aerial photographic evidence (National Mapping Programme (NMP)). Also, the national organisation for heritage, the Ancient Monuments Directorate of the Dept for the Environment, which became English Heritage in 1984, now, since April 2015, Historic England, began to support projects to map systematically the evidence of aerial photography across regions such as the Thames Valley, which were seeing a rapid expansion of development and mineral exploitation, particularly of the gravels (Benson and Miles 1974; Gates 1975; Leech 1977). Meanwhile the national (English) approach (NMP) continued under the auspices of Historic England following the merger of RCHME with English Heritage in 1999. Similar initiatives to map the historic landscapes of Wales continue to be undertaken by the Royal Commission on the Ancient and Historical Monuments of Wales (RCAHMW), Cadw and the Welsh archaeological trusts. Building on the pioneer work of Phillips in the Fenland (1970) new fieldwalking surveys, involving the systematic collection of pottery and building materials from the surface of ploughed fields and mapped against the evidence of crop and soilmarks plotted from aerial photography, were commissioned to improve our knowledge of wider landscapes, rather than the individual settlement. These approaches drew attention to a more diverse and populous settlement and exploitation of the countryside than had hitherto been perceived. In the years since 1990 techniques of geophysical survey have improved immeasurably so that they, too, have been incorporated into site and landscape survey as a more efficient and costeffective way of covering greater areas. In parallel with the increase in the measures to manage the destruction of Britain’s archaeological heritage through preservation by record, i.e. through archaeological investigation, there were also parallel and significant developments in what can be collectively termed as ‘environmental archaeology’ during the last 30 years of the twentieth century. In order to understand better

The last quarter of the twentieth century saw the beginning of a revolution in our knowledge and understanding of the countryside of late Iron Age and Roman Britain, its regional diversity, settlements and settlement hierarchy, agricultural exploitation, field systems and communications, rural industry and the wider environment of the landscape. This intensified in the 1990s after the introduction in England in 1990 of a change in planning policy, PPG 16 (Planning Policy Guidance 16), and the beginning of a radically more regulated approach to the potential loss of archaeological heritage through development and mineral exploitation. Although this policy has been modified in the years since and its essence since 2012 now enshrined within the National Planning Policy Framework (NPPF), 25 years of implementation have seen huge investment in the excavation, recording and publication of sites and settlements in the countryside of late Iron Age and Roman Britain (fig. 1.1). Similar policies were introduced in Wales in 1991 and Scotland in 1994. While some of this work has been slow to come into the wider public domain, with reports residing as unpublished ‘grey literature’ in local authority Historic Environment Records (HERs, formerly Sites and Monuments Records (SMRs)), its contribution in general terms has been to shift our gaze from the elite of Roman Britain and their villa residences to the population of the countryside at large. The context for the introduction of PPG 16 in 1990 was the growing pace of development of all kinds and the associated increase in mineral exploitation, notably sand and gravel, from the 1960s onwards, whose archaeological impact could not be responsibly accommodated either through voluntary funding agreements with developers and quarry owners or with the increasing gap between the available public funds required to excavate, record and publish the threatened sites and the growing scale of the threat. In parallel with the increasing development pressure, local authorities were encouraged from the mid-1970s to establish systematic records of their archaeological heritage (Site and Monuments Records (SMRs), now the HERs), trawling the literature and aerial photography for indications of 1

Romanch1.indd 1

06/09/2016 11:58:16

2

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 1.1.   Quantity and distribution of excavated Roman rural sites over time (by date of report)

Romanch1.indd 2

06/09/2016 11:58:55

INTRODUCTION

how the countryside of Roman Britain was exploited it was necessary to gain knowledge of the crops that were grown and the animals that were raised for consumption, and this was made possible by the development of the twin disciplines of archaeobotany and archaeozoology. These rapidly took shape to be effective as nationally available resources through the agency of English Heritage until the implementation of PPG 16. For the Roman period, the first synthetic fruits of these new directions began to appear from the late 1970s and early 1980s (e.g. King 1978; Jones and Dimbleby 1981; Luff 1982). The quantitative approaches that are integral to those disciplines also began to be applied to the study of material culture, particularly ceramics and coins (e.g. Reece 1973), which allowed for the possibility of characterising rural settlements from chronological, economic and social perspectives, as well as to the people themselves through developments in human osteology (cf. Roberts and Cox 2003, 107–63; 2004). Thus the two decades or so before the implementation of PPG 16 saw a very significant increase in the range of approaches that could be applied to the archaeological heritage and their potential for the advancement of knowledge was very evident. However, in order for PPG 16 to have the support of developers, it was essential to offer them the possibility of choice in whom they commissioned to undertake the archaeological work required by the planning authorities, and so competitive tendering between archaeological organisations, which previously had relied heavily on state funding for their projects, became embedded in the process. Notwithstanding the very real risk that a drive on cost could lead to a decline in standards, it was recognised that development-led archaeological projects should embrace, with appropriate justification, the full and developing toolkit of methodologies that could be applied. Therefore, the outcome of the integration of archaeology within the planning process over the last 25 years has meant not only a rapid increase in the number of reported investigations, but also in their overall quality. For example, information on settlement plan, change over time and material culture, which was perhaps the best that could be expected of post-WWII archaeological investigations over the third quarter of the twentieth century, was complemented by reports which, in the rural context, would give a more systematic insight into the agricultural economy, the people and the diversity of rural society in Roman Britain. The application of quantitative approaches provided potential for comparative analysis across both regions and the province as a whole. Nevertheless, while we should

Romanch1.indd 3

3

stress that, overall, the quality of information has improved over the last 25 years, there is still much variability, and the research that has underpinned this volume, and its planned successors, has identified where improvements could and should be made in the recovery of archaeological evidence, its recording and its publication. These methodological considerations will be the subject of a separate study to be published in 2017. At the same time, it is a tribute to what has been achieved since 1990 that research to capture and analyse the data that relate to the rural settlement of Roman Britain is necessary. While a good proportion (approximately 60 per cent) of developer-funded work has reached publication over this 25-year period a significant percentage has not and, particularly in the period before the wider dissemination of digital reporting, unpublished reports, the ‘grey literature’, languished within individual HERs without wider dissemination. Earlier pilot projects undertaken by Cotswold Archaeology, with funding from English Heritage, had identified both the volume and significance of the unpublished work (Cotswold Archaeology 2008; 2009; Fulford and Holbrook 2011) and this in turn stimulated the approach taken by this, Leverhulme Trust-funded, project to realise the potential of both the published and the unpublished work and to address a wide range of questions about the society, economy and people of the countryside of late Iron Age and Roman Britain. The sheer volume of information initially led to a focus entirely on England, but the additional resource has allowed the scope to include Wales and thus embrace a Roman Britain south of Hadrian’s Wall, but with comparative data from England to the north of that frontier. Therefore we have included those parts of Britain that were behind a frontier, defined around the turn of the first and second centuries a.d., initially as the Stanegate and then, after c. 120, as Hadrian’s Wall, which persisted through the remaining period of Roman administration into the early fifth century. Clearly there is scope to include Scotland, but in a context where territory within Scotland under Roman rule fluctuated between the campaigns of Agricola of the late first and those of Septimius Severus in the early third century, a contextualised approach exploring diversity and regionality, but which embraces Scotland as a whole in the late Iron Age/Roman period, might be more appropriate. Where this project departs significantly from previous attempts to synthesise the growing body of evidence from the countryside of Roman Britain is in its focus on the excavated evidence and the exploration of its potential in some detail. In his survey of Roman England, Taylor (2007) drew

06/09/2016 11:58:56

4

THE RURAL SETTLEMENT OF ROMAN BRITAIN

upon the evidence (to 2003) from the SMRs which, altogether, recorded a total of 27,902 rural settlements from across the country. This count was based on a variety of records, including cropmarks/soilmarks, earthworks, pottery scatters/ finds, and excavated evidence, although the dating of the great majority of the cropmarks and earthworks remained unconfirmed. Our study, however, is based on a much more selective sample of sites and only covers those with excavated evidence.With the inclusion of the Welsh settlement data in 2015 to complement that from England, we now have some 3600 records of rural sites (accounting for c. 2500 individual settlements; see Ch. 2) drawing on over 5000 published and grey literature reports, and equivalent to about 10 per cent of the settlements included in Taylor’s survey (2007). Allowing for upward adjustment in respect of settlements yet to be discovered and for some downward adjustment to figures following evaluation of undated sites, this 10 per cent nevertheless probably represents a minimum figure for our sample of the late Iron Age and Roman rural settlements of England and Wales, though there is undoubtedly variability between regions. Our prerequisites for inclusion are that the data can contribute to our understanding of settlement morphology, field systems, architecture, industry, people, ritual and systems of belief, and to the broader questions of social and economic status. In the case of the latter this requires the material or ecological data to be sufficiently robust in terms of characterisation and quantification to address such questions. It was found that small-scale evaluations and excavations, pre- or post-1990, generally did not provide data with suitable potential. However, even in a situation where a significant number of reports did not contribute enough morphological information to allow a settlement to be classified (see Ch. 2), the information that could be derived from other categories of evidence justified inclusion in the project. An important caveat is that in the years since the introduction of the first Ancient Monuments Act in 1913, there has been a protection policy that has favoured, through the process of ‘scheduling’, the ‘Roman’ aspect of the rural landscape. Owing to the large amount of research on villas into the third quarter of the twentieth century, and the corresponding lack of interest in non-villa settlements, along with the associated difficulties of designation given their irregular morphology and often unclear relationship with the wider landscape, the greater proportion of investigation since 1990 has been directed towards sites with no legal protection as Scheduled Ancient Monuments (SAMs) – the non-villa settlements. In practice this means that there are fewer

Romanch1.indd 4

investigations of villas that have benefited from modern approaches towards their excavation and the treatment of their finds, particularly the ecofacts. In the first instance the data analysed here were captured in terms of the modern political geography of regions and counties. As collection for each English region, and then for Wales, was completed, summary analyses were presented to regional seminars. This not only provided feedback to the team but it also began to provide an evidential basis for defining regions that reflected more closely the situation in the countryside of late Iron Age and Roman Britain. As the following section, which explores the development of research into the countryside of Roman Britain shows, previous approaches to synthesis had worked to the simple binaries of upland: lowland, military: civil, villa: non-villa, Roman: native, or a combination of arbitrary geographic regions, or modern counties, linked to the political geography of the tribes of Roman Britain insofar as it could be defined. Exceptions to this approach have been the detailed multi-period surveys of distinctive regions defined on the basis of geology and environmental considerations, drawing primarily on the evidence of aerial photography, earthwork survey and surface collection, such as the Fenland (Phillips 1970), Salisbury Plain (McOmish et al. 2002), Thames Valley (Booth et al. 2007) and the Solway Plain (Bewley 1994). The approach to regionality adopted here, and in the planned monographs to follow, also represents a compromise which, on the one hand aims to respect the distinctive pays, such as the Weald or the Fenland, within the newly defined regions, but, on the other, also allows for the assembling of data sufficient for comparative analysis across those regions, and that takes the study of the countryside of Roman Britain beyond the binary approach that has dominated research and synthesis up to now. A notable innovation of the analysis presented here is that the characterisation of the ‘new’ regions takes account of the material and environmental evidence as well as the morphology and architecture of settlements and their distribution through time and space. The underpinning data for this volume have been published online by the Archaeology Data Service of the University of York (M. Allen et al. 2015). HISTORY OF RESEARCH INTO THE ROMANO-BRITISH COUNTRYSIDE If one takes the beginning of the modern study of Roman Britain as Haverfield’s The Romanization of Roman Britain (first published in 1912, but based

06/09/2016 11:58:56

INTRODUCTION

on a lecture given to the British Academy and published in its Proceedings in 1905), the twentiethcentury study of rural settlement, and the countryside of Roman Britain at large, was dominated by considerations of what was perceived to be the character and extent of ‘Roman’ influence. A clear distinction was made between the uplands of the north and west, occupied by troops, and the lowlands of the south and east occupied by civilians (Haverfield 1912, 24). Appreciation of the latter had been built up by successive generations of curiosity-driven research from the early nineteenth century onwards on villas, with the expectation of discovering mosaic pavements and bath suites. Such buildings, or complexes of buildings, were defined as farming establishments in the countryside, and differentiated from ‘native’ settlements or ‘villages’ on architectural grounds, because their ‘dominant element is the Roman provincial fashion which is borrowed from Italy’ (Haverfield 1915, 46; Collingwood and Myres 1936, 208–25; Rivet 1958, 103–5; cf. Reece 1988, 59–76). By the mid-1930s some 500 villas were known and mapped, the vast majority noted as lying in the ‘lowland zone’ to the south-east of a line joining the rivers Trent and Severn (Fox 1932; Collingwood and Myres 1936, 217). Thus the emphasis of research was very much on the establishments of the elite, rather than on the ‘unromanized peasantry of the villages, which in fact must have comprised the great majority of the inhabitants of Britain’ (ibid., 221). The study of the ‘villa’ in isolation absorbed the majority of amateur and academic attention until the late 1970s (e.g. Rivet 1969; Todd 1978), seeing the development of a typology based on plan: cottage house, winged corridor house, courtyard house, aisled house, outbuildings and other types (Collingwood and Richmond 1969, 133–53), but only latterly considering its possible social implications (Smith 1963; 1978; 1998; Hingley 1989; Cunliffe 2013a; see Ch. 3). As we have seen, there was less enthusiasm for the investigation of ‘native’ settlements and PittRivers’ excavations and detailed publications of such settlements in Cranborne Chase, Dorset, remained a model of their kind until the late twentieth century (1887; 1888; 1892). The plan of the successive phases of Woodcutts, the evidence re-worked by Hawkes (1947), was reproduced as the typesite of a ‘homestead’ (as opposed to a ‘village’) in Richmond’s extensively revised edition of Collingwood’s The Archaeology of Roman Britain (Collingwood and Richmond 1969, 177) and in Applebaum’s agrarian history of Roman Britain (Applebaum 1972, 152). Besides the ‘homestead’, Richmond defined four other types of ‘native settlements’: ranches, villages, settlements in the

Romanch1.indd 5

5

military zone (or ‘upland areas’) and fortified community settlements, also seen to be characteristic of the fringes of the province in North Wales, the Pennines and the Lowlands of Scotland (Collingwood and Richmond 1969, 175–92). Other than the ‘homestead’ category, little could be said of the others, Richmond noting, for example, that ‘clearly defined Lowland villages have mostly been neglected’ (ibid., 180). Nevertheless, despite the lack of information about ‘native’ settlements or ‘villages’, Collingwood had previously teased out certain patterns of change, notably a perceived decline in rural settlement on the chalklands of southern England considered to be due to ‘a deliberate transplantation of village-dwellers to serve the capitalistic landlords’ (Collingwood and Myres 1936, 223–4). He also discussed the economic and legal status of the peasants and their plight in the later Roman period by saying that ‘As the years went by, there was no general increase in their prosperity… they became more and more liable to arbitrary exactions and oppression of every kind... they became increasingly servile in their position’ (ibid., 225). Comparing the situation in Britain with that in Gaul, Collingwood concluded ‘A survey of the state of the people in fourth-century Britain shows at least that, of the material conditions which might lead to such [peasant] revolts, none was lacking’ (idem). Richmond’s premature death meant that he was unable to include the outcome of a landmark conference held in 1965 on ‘Rural Settlement in Roman Britain’ (Thomas 1966) in the revised edition of The Archaeology of Roman Britain (Collingwood and Richmond 1969). This conference broke away from the tradition of treating rural settlement according to its constituent parts, which in effect, would have meant focusing on the ‘villa’, with papers considering the regional diversity of settlement right across the Roman province. Profusely illustrated with distribution maps and settlement plans, very many deriving from the ongoing work of the Royal Commissions on the Historical Monuments of England and Wales, and published in advance of the completion of the relevant county inventories, it provided a rich agenda for future research which was sensitive to regionality. The first detailed study of a distinct landscape of Roman Britain followed in 1970 with the publication of Phillip’s The Fenland in Roman Times, subtitled as ‘Studies of a major area of peasant colonization’. This drew heavily on the analysis of aerial photography, combined with surface collection of material from potential sites sufficient to date them and to offer some assessment of their economic base. This research on Fenland

06/09/2016 11:58:56

6

THE RURAL SETTLEMENT OF ROMAN BRITAIN

settlement was contextualised in relation to a rapidly developing knowledge of the Holocene sedimentary sequence and the incidence and extent of marine sedimentation as opposed to freshwater flooding. Although synthetic studies of Roman Britain during the post-WWII period looked afresh at rural settlement and economy, the evidential basis had changed little over the third quarter of the century. So Richmond (1955), Rivet (1958), Frere (1967), Salway (1981) and Todd (1981) essentially drew on much the same information as Collingwood had done in the 1930s and treated it in much the same way. In a major departure, Martin Millett broke new ground by promoting the archaeological evidence, at the expense of a historically driven narrative, in his The Romanization of Britain (1990). The first work dedicated specifically to an agrarian history of Roman Britain (Applebaum 1972) also had to work largely with material gathered before 1950. Almost a third of his study is based around analyses of the buildings found in the countryside with successive and complementary chapters on houses, byres and stables and farms and their uses, the latter section divided between the villa and the highland farm. The problems relating to this material were legion: dating was poor and the interpretation of the function of individual buildings as byre, barn, stable, granary, pigsty or slave quarters, was based solely on plan and conjecture, rather than on the combination of plan and supporting environmental evidence. It did not, however, prevent further conjecture about the workings of individual farms or villas, with estimations of the size of estates or herds, the acreage of arable or crop yields, as for example at Bignor villa, West Sussex (ibid., 212–14). Nevertheless, thanks to Helbaek’s research on cereals (1952) and Godwin’s The History of the British Flora (1956), Applebaum was able to say more about crops and plants than had been possible before (1972, 108–21). Chapters also surveyed the evidence of agricultural tools and ploughs, the latter distinguished from their Iron Age predecessors by the addition of the coulter. With a greater knowledge of, and insight into, field systems and agricultural technology, and with better excavated settlements and farms to draw on, Fowler (2002) took forward the agrarian theme in the broader context of the first millennium a.d. as a whole.  Applebaum’s approach to the reconstruction of villa estates was also adopted in individual excavation publications, for example Shakenoak villa (Applebaum 1978), Barton Court Farm, Oxfordshire (Jones 1986) and Gatcombe, Somerset (Branigan 1977a), and was also reflected in Branigan and Miles (1989), a collection of essays

Romanch1.indd 6

that contained the first quantitatively informed considerations by Reece and King respectively of coins and faunal data in relation to villas. As Collingwood had done in the 1930s, writers continued to interpret regionality and change over time against what could be derived from written sources, or linked to wider understandings of land and estate management across the Roman Empire. Though Frere appreciated the difficulties of disentangling rural organisation on the basis of archaeological evidence alone (1987, 259), the absence of written sources did not prevent continued speculation. For example, the lack of villas on Salisbury Plain and in the Fenland, where there was also evidence of drainage (the Car Dyke) and reclamation, was seen as an indication of the presence of imperial estates (ibid., 266–9; cf. Collingwood and Myres 1936, 223–4). The narrative continued to be framed around the written sources. For example, despite considerable source criticism on the part of archaeologists, Applebaum continued to be persuaded that wool production was a significant element of the agrarian economy, particularly across the chalklands of southern England (1972, 232–4; cf. Collingwood and Myres, 239; Frere 1987, 290–1). Without the listing of a gynaecium (weaving works) in the Notitia Dignitatum and of two British wool products in Diocletian’s Price Edict, it is doubtful whether such conclusions would have been reached on the basis of the archaeological evidence then available. Likewise the incidence of corndrying furnaces was linked to the written sources that implied the regular production of corn surpluses, including its export to the Rhineland, in the fourth century (Applebaum 1972, 229–31). It was also appreciated that cattle rearing played an increasingly important role, particularly in the later Roman period, though without perceiving its role, as a provider of manure, in cereal production (ibid., 208–14, 232–4). The evidential base began to change significantly with the rise of rescue archaeology in the 1970s and the increasing application of systematic methods to recover both botanical and faunal remains, alongside improved excavation methodologies to provide sequencing and chronology. Studies by King (1978; 1984; 1988; 1991), Maltby (1984) and Grant (1989) synthesised the increasing availability of faunal assemblages, demonstrating the growing role that cattle played in the economy, at the expense of sheep, compared with the Iron Age. As well as a source of meat and leather, cattle provided traction for ploughing and transportation, and in addition there was recognition of the part that animals played in ritual and religion. Complementary research on macroscopic plant remains, seeds and pollen also

06/09/2016 11:58:56

INTRODUCTION

shed more light on the agrarian regime of Roman Britain, including evidence for the expansion of areas under cultivation and the introduction of hay meadows, as well as the consumption of plantbased foods (Jones 1981; 1989; 1991; Lambrick and Robinson 1988; Lambrick 1992, 97–105; Robinson 1992, 56–9). By the end of the twentieth century, there were sufficient pollen data to allow the assembly of a province-wide review of environmental change between the late Iron Age and the Anglo-Saxon period, which considered inter alia the relationship between cultivated or open land and woodland and the impact of the establishment of the northern frontier systems on the local environment (Dark 2000, 81–129). The intensification in survey, from the 1980s onwards, both from the systematic plotting of available aerial photography, but also by ground level, ‘walk-over’ survey with systematic collection of artefacts revealed by cultivation, also led to enhanced understanding of the density and distribution of settlement. One project, the Maddle Farm survey on the Berkshire Downs, interpreted the significance of the pottery scatters in the ploughsoil as evidence for the rearing of cattle on the chalk downland, their manure helping to fertilise the intensive cultivation of cereals (Gaffney and Tingle 1989). Collectively, these surveys encouraged the further modelling of population numbers leading to the projection of some high numbers for the later Roman period (Millett 1990, 181–6; Salway 1981, 542–52). Further important regional surveys, necessarily multiperiod in scope, followed, including the Solway Plain (Bewley 1994); Salisbury Plain (McOmish et al. 2002); the Thames Valley (Booth et al. 2007), the Lincolnshire Wolds (Jones 1998; Winton 1998), the Yorkshire Wolds (Stoertz 1997) and further survey of the Fenland (Hall and Coles 1994, 105–21). Nevertheless, even by the end of the twentieth century the narrative of the Roman countryside continued to be dominated by the duality of villa and non-villa, for example in Dark and Dark (1997) and in Hingley, although the relationship of the built environment and rural social organisation is a major theme of his Rural Settlement in Roman Britain (1989). There were only sufficient data to offer reliable distribution maps of villas, as illustrated in successive editions of the Ordnance Survey’s A Map of Roman Britain, the fourth edition of which appeared in 1978, or in Jones and Mattingly’s Atlas of Roman Britain (1990), and formed the basis of quantitative studies such as Millett’s further reflections on the relationship of villas with towns (1990, 190–7). In contrast, nonvilla settlements continued to be characterised and discussed on an individual basis.

Romanch1.indd 7

7

Malcolm Todd, acknowledging the increase in knowledge since his Research on Roman Britain 1960–89 (1989), produced a new overview only 15 years later, his A Companion to Roman Britain (2004). He saw that the treatment of rural settlement required separate chapters, dividing Britain into ‘north’ and ‘south’ (Hingley 2004; King 2004a) with a further chapter by A. Grant (2004) on ‘Domestic animals and their uses’, which embraced urban as well as rural developments. Following Millett’s explicitly archaeological approach, David Mattingly set out his concern over a Roman-centric approach to the study of Roman Britain in the introduction to his An Imperial Possession: Britain in the Roman Empire (2006, 3–20). More specifically in relation to the ‘Rural communities’ and in a symbolically titled chapter ‘The villa and the roundhouse’, he drew attention to the bias in the information available for the study of rural settlement towards the villas and the elite society of Roman Britain (2006, 356–8). Nevertheless, whereas previous academic syntheses of Roman Britain had treated the countryside in a single chapter, Mattingly’s consideration extended over four chapters, representing about one-fifth of the whole book. There was an extensive treatment of both ‘provincial landscapes’, similar to the approach taken in Todd (2004), which reviewed the character and diversity of settlement according to broad geographical regions of England, Scotland and Wales (while also taking account of the political, tribal geography), and of ‘rural culture and identity’ whose scope was able to embrace both elites and non-elites. Together this represented a distinctive break with the synthetic treatments of the second half of the twentieth century whose structure and content could still be firmly linked to Collingwood’s Roman Britain of the 1930s. The first attempt to assimilate the large datasets relating to all rural settlement in Roman Britain was published the following year (Taylor 2007). This study drew on the evidence of local authority SMRs across England supplemented by case studies on certain areas where the evidence from more intensive aerial and ground survey, as well as excavation, was more plentiful and provided a new classification of non-villa rural settlement based on their morphology: enclosed, unenclosed and linear system settlements. Regionality was a major area for consideration but it was addressed through the modern political English regions, rather than through any construct of the civitates. At a simpler level Taylor saw a Roman Britain of two (if not more) worlds: eastern and central Britain, characterised by villas and linear system settlements associated with ‘continuously bounded landscapes’

06/09/2016 11:58:56

8

THE RURAL SETTLEMENT OF ROMAN BRITAIN

linked by long-distance trackways, as well as rural industries, and a northern and western Britain with continuity of a roundhouse tradition and ‘settlement-focused patterns of enclosure with significant areas of unbounded land in the wider landscape’ (ibid., 113–14). The first world was one that saw the development of agricultural production strategies designed to create a significant surplus, while the second world demonstrated no evidence for such a strategy (ibid., 115, and figs 7.2–4). Taylor’s analysis of SMR data (up to 2003) logged a total of 27,902 settlements, the majority of which were identified either as earthwork, cropmark and soilmark sites, without independent evidence of date, or as finds scatters. The inclusion of types of settlement on morphological grounds, rather than in combination with excavated evidence, reduces the reliance that can be placed on the recorded numbers, though there is no reason to doubt the overall trends in space and time (2007, 11–22). Nevertheless, it is Taylor’s study that has provided the foundation for this book and the planned successor volumes, which draw on the excavated data. It was published, as we have seen, when the volume of excavation of non-villa settlement was gathering pace in the boom years at the very beginning of the twenty-first century. In this introduction we make the case for dividing Roman Britain south of Hadrian’s Wall into eight regions, each with a distinctive range of settlements and their associated material and environmental characteristics. In areas where there has been a large amount of recent, developmentled excavations we can begin to identify yet more locally defined and distinctive patterns of rural settlement; the two approaches draw us away from identifying any connection with the administrative arrangement of Roman Britain based on the tribal civitates, in so far as it can be reconstructed. Inevitably we look across the Channel to see to what extent similar patterns of complexity and regionality are emerging from the study of late Iron Age and Roman settlement in the neighbouring provinces of Gaul and Germany, but note the continued dominance of the villa in current discourse there (e.g. Roymans and Derks 2011a; Jeneson 2013). THE NATURE OF THE DATASET In order to synthesise the vast amount of excavated evidence available for a new study of the RomanoBritish countryside, one of the most important aspects was the development of a project database to facilitate management and analysis of the substantial body of data generated. At the outset it was the intention to make the database available as

Romanch1.indd 8

an online resource, which is available at http:// dx.doi.org/10.5284/1030449 (see Appendix 1). This section introduces the key geographical, chronological and typological parameters of the dataset. While the title of this volume is the Rural Settlement of Roman Britain, in modern geographical terms it would be more accurate to describe this work as the Rural Settlement of Roman England and Wales. Although the initial phase of the project focused on England only, a second phase, also funded by the Leverhulme Trust, has facilitated the inclusion of data from Wales, meaning that the geographical scope of this work includes the Roman province of Britannia, south of Hadrian’s Wall, as well as an area up to nearly 90 km north of the line of the Wall in North-East England (for rationale, see above, p. 3). Although the Roman period forms the chronological focus of this work, a fundamental research question centres upon the impact the Roman Conquest had on late pre-Roman Iron Age societies in Britain, and for this reason the chronological parameters have been extended to include sites occupied prior to the conquest. The late pre-Roman Iron Age in Britain tends to be defined as the approximate century and a half prior to the Roman conquest of Britain, from around the start of the first century b.c. to the mid-first century a.d, a period characterised by profound changes such as the introduction of coinage and wheelmade pottery, and the establishment of nucleated settlements (e.g. Millett 1990, 10; Creighton 2000, 4–21; Cunliffe 2005, 402–6). These changes did not occur uniformly across Britain, however, being confined largely to the south and east, meaning that it is arguably inappropriate to regard the late Iron Age as having a single start date that is applicable to the whole island. Nevertheless, for reasons of pragmatism and consistency, all sites with evidence for occupation during the first century b.c. and early first century a.d. have been included in the project database as late Iron Age sites. Sites that were wholly occupied and abandoned prior to the first century b.c. (i.e. those generally described as middle Iron Age) have not been included in the dataset. However, late Iron Age sites with earlier origins have been included, and where possible these have been distinguished from those sites actually established during the late Iron Age (i.e. mid–late Iron Age sites and late Iron Age sites). At the opposite end of the chronological spectrum, in order to explore potential evidence for post-Roman continuity at Romano-British sites, settlements with activity in the fifth century a.d. and beyond have been included where there was evidence for Roman-period occupation,

06/09/2016 11:58:57

INTRODUCTION

although sites newly established after the early fifth century a.d. fall outside the study’s scope. Coins of the House of Theodosius and radiocarbon dates are the principal sources for determining occupation after a.d. 400, the latter especially for settlements in the north and west of Britain. As Fulford and Holbrook (2011) have previously described, the project is concerned with all sites occupied within the above chronological parameters in England and Wales, and not just those that produced evidence of Romano-British culture. The remit of the project therefore includes sites within areas such as the south-western peninsula and parts of Wales and Northern England that include evidence for occupation during the late Iron Age or Roman periods, even where they display relatively little evidence of integration with the Roman provincial administration and economy (ibid., 326). This study originated as an assessment of the value of commercial archaeology to the study of the Romano-British countryside, with a focus on the range of excavated evidence available. Excavations conducted in the commercial environment are undertaken at a variety of scales, employing a range of methods. These include exploratory pre-planning-determination ‘trial excavations’ or ‘evaluations’, as well as postdetermination watching briefs, strip-map-andrecord sample excavations, and open-area excavations (Fulford and Holbrook 2011). The archaeological reports produced for all of these types of intervention represent the primary sources of data gathered by the project. While one of the initial aims of the project was to consider the impact of commercial archaeology on our understanding of the Romano-British countryside, it was decided to extend the scope of the project to include data from all excavations concerning Romano-British rural settlement, whether they were conducted in response to development or not, and sites subject to excavation under other circumstances (e.g. research-driven or community-focused excavations) have been included on the project database as long as the reports contain sufficient data to contribute to the overarching aims of the study (see above, p. 4). The major increase in development-led archaeology since the introduction of PPG16 in England in 1990 means that the majority of records on the project database are represented by excavation reports produced during the late twentieth and early twenty-first centuries, but the database includes reports from a wide range of dates; several antiquarian reports dating to the nineteenth century are included, for example, with the earliest dating from 1808. Sources of data include traditional published excavation reports as well as

Romanch1.indd 9

9

archaeological ‘grey literature’ – unpublished reports generated during the planning process and deposited with Historic Environment Records, which have traditionally been little utilised by archaeological researchers. Sites represented by this body of unpublished material form a substantial component of the database; 46 per cent of database records include information retrieved from grey literature reports, and 35 per cent are represented by unpublished grey literature reports alone. As the focus has been on evidence generated through excavation, Romano-British sites identified exclusively through other forms of archaeological prospection such as aerial photography, geophysics, fieldwalking or metal-detector survey have not been included in the project database. However, where these methods have been employed alongside excavation, the results and associated site plans have generally been incorporated together with the excavated material. It is important to stress here that the emphasis placed upon excavated data results in an inevitable geographical bias towards areas that have seen most excavation, and the initial phase of the project identified a clear emphasis on the south and east in terms of the distribution of development-led archaeological interventions (Fulford and Holbrook 2011, 330). As Taylor has shown, however, some parts of England and Wales are substantially better represented by evidence from (undated) cropmarks and earthworks than they are by evidence from excavations (Taylor 2007, 11–17), and consideration of the excavated evidence without regard to information from other archaeological sources is unlikely to allow meaningful discussion of issues such as regional variation in Romano-British settlement density. While excavated sites represent the study’s primary dataset, in order to mitigate the effects of geographically uneven levels of excavation, other sources of archaeological evidence (e.g. cropmark data, earthwork survey, fieldwalked data, Portable Antiquities Scheme records) have been consulted and incorporated, where, for instance, they can contribute to a better understanding of regional patterns of settlement density or the circulation of material culture in a given area. The potential impact geographical bias has on our understanding of rural settlement density is considered on a regional basis in Chapters 4 to 11, and in overview in Chapter 12. DEVELOPING A METHODOLOGY Archaeological excavation reports were collected in two ways. For unpublished material an English Heritage-funded team from Cotswold Archaeology was dedicated to sourcing developer-funded grey literature concerned with Roman rural sites by

06/09/2016 11:58:57

10

THE RURAL SETTLEMENT OF ROMAN BRITAIN

requesting reports from individual Historic Environment Records (HERs) in England and Wales. The response was overwhelmingly positive. Of the 88 HERs in England and the four in Wales, all engaged with the project, and over 2575 grey literature reports were collected from the two countries. Where digital versions of reports existed these were sent electronically, and where reports existed in paper format only, members of the Cotswold Archaeology team negotiated with the HERs (and in some cases the originating archaeological contractors) to produce electronic versions using Optical Character Recognition technology, allowing the reports to be searched electronically. The scanned, electronic reports were then fed back to the HERs (and uploaded to the Archaeology Data Service’s grey literature library), meaning that in many cases HERs benefited directly from providing access to the data. The unpublished reports gathered by Cotswold Archaeology were then sent to the research team at the University of Reading who mined them for data. Excavation reports published in national and regional archaeological journals and monographs were located directly by the Reading research team. Data collection was divided between the three University of Reading researchers geographically, using the modern European Parliamentary Regions for England and areas covered by the four HERs in Wales. Reports were scrutinised on an individual basis and any site, whether published or unpublished, with potential evidence for Romano-British rural settlement was considered for inclusion on the project database. However, the research team was selective in their choice of sites to include, and only reports of investigations with sufficient evidence to contribute data to at least one of the research questions were included in the database. Reports associated with very small-scale interventions that produced only slight and poorly characterised evidence for Romano-British activity, or those entirely lacking information about finds and environmental evidence, were generally excluded. Given the ambitious scope of the project it was necessary from the outset to establish strict criteria for what was to be included as a ‘rural site’ for the purposes of inclusion in the project database, as it was impractical to consider including all Roman sites within England and Wales. Sites with clear non-rural characteristics, such as Roman military sites and major walled urban centres, including civitas capitals and coloniae, were therefore omitted from the database. For pragmatic reasons related to what could be achieved within the project budget it was also deemed unfeasible to collect data from all ‘small towns’, and so only undefended nucleated sites (i.e. those without walls or major

Romanch1.indd 10

table

1.1: classification

of site types on rural

roman settlement project database

Major site type

Minor site type

Rural settlement farmstead, villa, agricultural building, isolated building, hillfort, cave Rural landscape

field system

Industry pottery production, tile production, quarry, iron production, other metal production, mill, other industry Nucleated settlement roadside settlement, village, oppidum Religious, ritual  and funerary

Romano-Celtic temple, shrine, funerary site

Military fort, vicus, other military Communications

road, jetty/bridge, mansio, port

ditches) were selected for inclusion. Although canabae associated with the three long-lived legionary fortresses of Caerleon, Chester and York have been excluded, vici attached to other military sites (with the exception of those on Hadrian’s Wall and the Stanegate, once again for reasons of pragmatism) have been included in order to provide evidence for nucleated settlements in the north comparable to that for the villages and ‘small towns’ (here termed ‘roadside settlements’; see Ch. 2), which occur much more widely in the south and east of the province. Aside from the above restrictions, all other Romano-British sites were considered for inclusion, and were entered into the database using a two-tier classification system, with ‘major’ and ‘minor’ site-type headings (table 1.1). Where appropriate, sites entered onto the database were assigned multiple site-types. For example, a site characterised as a farmstead, but which also produced evidence for pottery production, was classified under ‘Rural settlement and industry’ in the major site type category, and under ‘Farmstead’ and ‘Pottery production’ in the minor site type category. Similarly, a site that originated as a farmstead in the late Iron Age or early Roman period but which later developed into a villa would be classified as both ‘Farmstead’ and ‘Villa’, and the project database includes phase date fields within which the chronological details of major changes to settlements could be recorded. As explained above, Roman military sites were generally excluded from this study, and the major site-type category ‘Military’ includes ‘Fort’ and ‘Other military’ principally to allow recognition of sites that may have had military phases but which were ‘rural’ during other periods of their occupation within the Roman period. The

06/09/2016 11:58:57

INTRODUCTION

Romanch1.indd 11

of occupation for sites varying dramatically between different reports. Some reports present meticulously phased site chronologies, with start dates, end dates and periods of hiatus or transformation provided to within 50-year blocks or better (e.g. occupied between ‘50 b.c. and a.d. 300, with a hiatus during the second half of the second century’). Some sites, however, are presented in reports by very broad date ranges, with chronological descriptions limited to statements such as ‘occupied during the Iron Age and Roman periods’. There are several reasons for the discrepancies between the levels of dating information provided in different reports, but as pottery represents the principal dating tool for most sites occupied during the Roman period, in most cases the precision and confidence of a site’s chronology is directly associated with the quantity and type of pottery recovered during excavation. There is a tendency, for example, for sites examined in evaluations and watching briefs to be less precisely dated than sites subject to open-area excavation; fig. 1.2, for example, shows how evaluations are twice as likely to have only very broad dating information available compared with excavations. This partly reflects the situation that these types of interventions tend to yield far smaller pottery assemblages, presenting pottery specialists with more limited collections on which to draw chronological interpretations (fig. 1.3), but also that ceramic assemblages recovered during evaluations and watching briefs, because they are often of relatively limited size, are perhaps less likely to progress to specialist analysis than larger assemblages from excavations. There is also considerable regional variation in the extent to which pottery was used in Britain during the Iron Age and Roman periods, meaning

15%

% of sites

specific criteria for inclusion under the settlement categories presented in table 1.1 are discussed in detail in Chapter 2. From the outset, the project was designed to take an integrative approach towards the excavated material, and the database was developed with fields to record detailed information for a range of archaeological information for each site. The facility exists to record approximately 500 specific fields of data for each database entry. These include core data fields detailing bibliographical and geographical information, the form of archaeological intervention (i.e. evaluation, watching brief, area excavation) and the extent of the area excavated, as well as more detailed information for sites, including the form and character of the settlement, the chronology of activity and the presence, frequency and form of features such as domestic and non-domestic structures. Where reports contained site plans that allowed sites to be classified based upon their morphological characteristics (see Ch. 2), these were scanned by the Reading team and linked to the database. The detailed information collected on settlement form, chronology and building types was fundamental for the analyses presented in this volume, allowing the exploration of regional and intra-regional patterns of settlement character, temporality, and architecture. Specialist data fields also allowed collection of detailed information on burials, pottery, small finds and environmental evidence, including faunal and botanical remains. Summary details were recorded for finds and environmental assemblages, and, where possible, finds were also quantified in order to enable statistical comparison between sites. These categories of data represent the basis for the discussion of rural settlement economies and social hierarchies that form a component of each of the regional chapters in this volume, but will be dealt with more fully in volumes 2 and 3 of this series. Amalgamating and attempting to record data in a standardised format from several thousand archaeological reports, which were produced by different organisations, using different methods, operating under different circumstances at different times, has not been straightforward, and there have been a number of resultant methodological issues to overcome. For various reasons it was not always easy to classify sites according to the system outlined above, and this particular methodological problem is considered in Chapter 2, where the project’s method for classifying sites is described in more detail. The ability of the team to assign confident date ranges to individual sites was also immensely variable, and this was influenced by a number of factors. One such is the level of detail included in the literature, with the presentation of the chronology

11

10%

5%

0% Evalua ns (736 sites)

Watching Briefs Excava ns (196 sites) (2650 sites)

fig. 1.2.  Proportion of each major type of intervention with only imprecise dating evidence available

06/09/2016 11:58:58

THE RURAL SETTLEMENT OF ROMAN BRITAIN

average number of po ery sherds per site

12 2500 2000 1500 1000 500 0 Excava ons (2650 sites)

Evalua ons Watching Briefs (736 sites) (196 sites)

fig. 1.3.  Average number of pottery sherds recovered from excavations, evaluations and watching briefs

that the availability of pottery as a dating tool is geographically very variable. Many rural Iron Age sites in northern England and elsewhere are essentially aceramic (Bewley 1994), meaning that even those producing pottery of Roman date may well have been occupied earlier, in phases that produced little dating evidence. In some areas even sites definitely occupied in the Roman period produce very little pottery evidence. Many Roman rural sites in the hinterland of Wroxeter, for example, appear to have used little pottery (Fulford and Holbrook 2011, 326; Gaffney et al. 2007), apparently representing a continuation of midand late Iron Age patterns of pottery use (Wigley pers. comm.). Even in regions where pottery occurs as a common site-find, differences in our understanding of the chronology of diverse types of Romano-British pottery in various parts of the province have a major influence on the precision pottery has as a dating tool. Pottery of Iron Age and early Roman date is particularly problematic. In the south, for example, there are clear and reasonably well-understood differences between pottery of mid- and late Iron Age date, and wellstratified sherds can be exceptionally good indicators of the date of origin and sequences of activity at sites. At Copse Farm, Oving, West Sussex, for example, it was possible to establish from the pottery assemblage that two adjacent enclosures, initially identified through aerial photography, were not contemporary, with the earlier enclosure established in the second century b.c. and abandoned by the beginning of the first century a.d., when the second enclosure became the focus for activity (Bedwin and Holgate 1985). Such resolution is often unattainable in other parts of the country. In parts of the East Midlands, for example, ‘Middle Iron Age’ pottery traditions such as ‘Scored Ware’ continue in some areas until the mid-first century a.d., perhaps even into the early Roman period (Chadwick 2010, 334; Elsdon

Romanch1.indd 12

1992, 86; Willis 2006, 112–13), meaning that some regional types of pottery are unable to provide anything other than a very broad suggestion as to the period of occupation for sites, such as ‘Iron Age or Roman’. At some sites in such areas, where well-dated and diagnostic Romano-British or imported Roman pottery such as samian is found, it may be possible to establish with confidence that a site was occupied during the Roman period, but the imprecise chronology of many regional coarse-ware pottery traditions means that it may be impossible to establish with any meaningful degree of precision when a site actually originated or, indeed, when activity ceased. Even in regions where ceramic traditions are well dated, the closest resolution provided in a report is normally to within 25–50 years at best, and it is common for reports to describe dateranges for activity at sites as ‘from the later first or second century to the earlier fourth century’, as at, for example, Burton Wold Farm, Burton Latimer (Edgeworth 2008, 41). Chronological information has been recorded on the database numerically, and so such a statement would generally result in a site being provided with a start date of a.d. 100 and an end date of a.d. 325, and it is important to be aware that the start dates, end dates and phase dates for major changes at sites are only intended to be very approximate, as pottery rarely affords a greater degree of precision. Coins form an additional and important strand of dating evidence at some late Iron Age and Roman period sites, and 52 per cent of the settlements recorded on the database produced Iron Age or Roman coins. However, coins are far less commonly recovered from low-status, rural settlements than from villas or nucleated sites (present at 42 per cent of farmsteads, for instance, compared with 86 per cent of villas, villages and roadside settlements), and they are therefore unavailable from many rural sites. What is more, whereas villas and nucleated settlements typically produce coins regardless of where they are located, coins are geographically distributed far less evenly across other classes of rural sites. Of 38 farmsteads in Northumberland, for instance, only three (8 per cent) produced coins, all single examples. In Gloucestershire, by contrast, coins have proven to be considerably more likely to be found, and here 55 of 91 farmsteads (60 per cent) produced coins, in some instances numbering in their hundreds. The starkly uneven distribution of coins at farmsteads therefore compounds the above dating issues associated with pottery, as the areas in which coins appear not to have circulated widely in the countryside typically correspond with those that saw low levels of pottery use or conservatism in pottery traditions.

06/09/2016 11:58:59

INTRODUCTION

The other major source of dating evidence, particularly in regions where ceramics and coins tend to be poorly represented, is radiocarbon dating, although radiocarbon dates were available for just 278 sites (7 per cent of sites) recorded on the project database. While the use of such a dating method is clearly geographically widespread, it is particularly well represented in areas such as the South-West and the North-East (fig. 1.4), where little closely dated pottery is available at rural sites. Radiocarbon dating was, in the decades following its widespread introduction as a dating method in the 1950s, a very imprecise technique, meaning that sites dated by radiocarbon dating in its formative years were often provided with very broad date ranges, often spanning centuries (Barker 1991, 243), and there has been a perception that, for the Roman period, artefactbased dating provides far greater precision (Bayliss 2009, 140). However, subsequent developments in the technique, in particular the introduction of calibration, dating by accelerator mass spectrometry (which increased the range of material available for sampling) and the widespread

fig.

Romanch1.indd 13

13

adoption of Bayesian chronological modelling, means that considerably more precision can now be achieved (e.g. Bayliss 2009, 141). It would therefore be advantageous to see the approach adopted more widely for Romano-British rural sites. A further methodological problem has been a lack of consistency in the way information is presented in excavation reports. This issue is common to all types of material recovered from archaeological sites, although it is particularly acute for ceramics, and this is especially problematical because pottery is the single most common type of artefact recovered from most sites of Roman date. While there has been undoubted improvement in recent years, pottery reports still suffer from a lack of standardisation in the recording of ceramics, both regarding terminology and quantification. Pottery (and other finds) reports from the earliest excavations are generally the least useful, where the presentation of information on the pottery is often limited to catalogues and descriptions of complete and unusual vessel forms, with little attempt at quantification, making

1.4.  Distribution of sites with radiocarbon dates

06/09/2016 11:59:12

14

THE RURAL SETTLEMENT OF ROMAN BRITAIN

meaningful comparison with pottery assemblages from other sites impossible. Modern excavation reports generally do quantify pottery, although different pottery specialists employ a range of methods. Total number of sherds and total weight are the two most commonly used quantitative measures for pottery, and so these units were selected for the purposes of recording pottery assemblages in the database. However, two other common methods include presentation by Minimum Number of Vessels, or by Estimated Vessel Equivalents (EVEs) (Orton et al. 1993, 172, 210). Presentation of pottery data using these methods is sometimes used alongside sherd count/weight, which does not represent a problem, but all too often they are used exclusively, meaning that the reports are incompatible with those produced using alternative methods. Although the creation of a National Roman Pottery Fabric Reference collection (Tomber and Dore 1998) provided terminology that has facilitated consistent descriptions of Roman imports and selected major British pottery products, the volume did not include all regional wares, meaning that these continue to be described using local terminology by pottery specialists operating in different geographical areas (Booth pers. comm.). This has made it problematic to attempt consistent recording of anything other than very basic information on the types of pottery recovered from Romano-British rural sites, and this was limited to noting the presence/absence and quantification of major types; samian, amphorae and mortaria. Currently, all artefacts and ecofacts suffer from similar issues to the pottery. In particular, a lack of consistency in the methods used by specialists for presentation of the data in reports means that it was not always possible to quantify and analyse material on the database, even where the report provided some information. In these instances summary fields provided in the database prove useful, as the data may still be interrogated at some level using text searches. There is insufficient space here to provide more detail on the methodological problems associated with the way site finds are excavated and recorded. These issues will be addressed further in a forthcoming study, which will present a critique of the ways our rural settlements in England and Wales are currently excavated and reported on, providing some recommendations for best practice in the future. While pottery represents the principal dating evidence available for most sites, the study of site finds otherwise forms a fairly minor component of the work presented in this volume, and more detailed analysis of the artefactual and environmental data will form the focus of

Romanch1.indd 14

subsequent volumes. However, artefact and environmental assemblages have been incorporated in each of the regional chapters to explore broad spatial, social and economic patterns in major classes of settlement. Furthermore, in the regions where environmental evidence is scarce because of highly acidic soils (chiefly in the north and west), slightly greater emphasis has been placed on the artefactual evidence. For artefacts, the principal method for comparing the use of objects across site-types has been a simple evaluation in terms of presence/absence of a range of 20 broad classes of objects (see Appendix 3 for details of object classes), with the number of sites producing objects presented as percentages. Where faunal remains were available in sufficient quantities, different classes of site were compared using relative frequencies of the major species of domesticated livestock (cattle, sheep/goat and pig). In order to avoid bias caused by small sample sizes, the relative frequencies were calculated for each site-type using a mean value of all percentages for all sites that had a minimum of 100 identified animal bones (NISP – Number of Individual Specimens identified to species). The variability in the availability of quantified archaeobotanical assemblages prevented a similar approach to the consideration of plant remains, and for these data site-types were compared using presence/absence of different taxa. The availability of environmental evidence varies considerably from region to region, chiefly owing to the effects that variations in soil acidity have on the preservation of organic remains, with human burials, animal bone and plant remains all affected. Sites with well-preserved bone (human and animal) assemblages, for example, are overwhelmingly biased towards the south and east of Britain, where soils are, in some areas, much less acidic than in the north and west. This has a profound effect on our ability to use burial evidence, faunal remains and botanical evidence to help characterise sites in some regions, and our understanding of various aspects of the rural agricultural economy is therefore almost entirely reliant on well-preserved evidence from the south and east. A further issue that may impact upon the availability of various types of archaeological evidence is the considerable variation that exists between different archaeological contractors, both in terms of approach and levels of competency and commitment. There is also regional variation in curatorial practice, including differing policies concerning the way sites should be investigated by contractors, as well as differences regarding the enforcement of these policies (Holbrook and Morton 2011, 7).

06/09/2016 11:59:13

INTRODUCTION

This section has sought to present some of the principal methodological problems associated with the collection of such a vast body of disparate data. While these issues are often difficult to overcome at the level of the individual site, the large number of sites in the database and the broad geographical scope mean that analyses can often be undertaken once problematic sites with known biases (poorly quantified or incomplete assemblages, for instance) have been identified and removed, without having too detrimental an effect on the sample size available for study. A REGIONAL ANALYSIS OF RURAL ROMAN BRITAIN During the data-collection phase of the project, data were collected and entered into the database using modern political boundaries, principally at the level of the European Parliamentary regions for England, and of the areas covered by the four Welsh HERs, which represent Clywd-Powys, Glamorgan-Gwent, Gwynedd and Dyfed. This was to facilitate ease of systematic data collection and entry into the database. A regional approach towards the analysis of the data is necessary to explore geographical variation in the RomanoBritish settlement pattern, but it is clear that the modern regions used during data collection are meaningless entities in terms of the ancient countryside, and so using them as geographical units for comparative analysis (as has been done by some recent studies of the Romano-British countryside, e.g. Taylor 2007) would be problematic. As discussed earlier in this chapter, the regional variation in the Romano-British settlement pattern has previously been explored in various ways, and it became clear during the data collection phase that few previous studies of the Romano-British countryside had established satisfactory units for analysis, with most adopting overly simplistic models such as ‘civilian’ and ‘military’ (Haverfield 1912), ‘upland’ and ‘lowland’ (Fox 1932), or ‘military’ and ‘villa’ (Dark and Dark 1997) zones, which failed to take account of the regional diversity in the character of the landscape within these bipartite categorisations. An attempt to use ancient Roman administrative units, civitates, was discounted, as these were Roman political constructs and our understanding of the territorial extent of these units is very imprecise and debated (Jones and Mattingly 1990, 154; Millett 1990, 67); in any case the boundaries of these territories are likely to have shifted over time (Frere 1987, 194) (see Ch. 12 for discussion of ‘tribes and civitates’). The approach recently adopted by the Fields of Britannia project (Rippon et al. 2015) (itself influenced by Roberts and Wrathmell’s 2000 and

Romanch1.indd 15

15

2002 studies of medieval England), which investigated continuity and change in the agricultural landscape between the late Roman and early medieval periods, offered a good way forward. In that study, a new series of nine regions was defined, based on a combination of distinctive natural and anthropogenic characteristics. These included geology, soils and topography, as well as Roman and post-Roman cultural distinctions such as the extent of ‘Romanisation’, the scale of AngloSaxon immigration and the creation of villages and open fields in the early medieval period (ibid.). For our purposes eight distinct regions were established to provide a best-fit between the results of an analysis of the archaeological dataset, and the character of the physical landscape. These regions comprised the North, the North-East, the Central West, Upland Wales and the Marches, the Central Belt, the East, the South, and the South-West. Whereas the North-East and South-West project regions share their names with European Parliamentary regions, they do not coincide with these modern administrative areas (fig. 1.5). As the regions were partially defined based upon physical characteristics of the landscape, there are some broad similarities between the project regions and those utilised by Rippon et al. (2015), and, indeed, the ‘English provinces’ recognised by Roberts and Wrathmell in their study of the medieval rural landscape (2000; 2002), although there are some subtle and important differences. Each of the project regions was initially defined by the project team by mapping various distinctive aspects of the Romano-British settlement pattern, including the distribution of villas, nucleated settlements and different classes of farmstead (see Ch. 2). Broad patterns in the distribution of these different types of site were identified and the eight regions were coarsely drawn around them. In order to avoid entirely arbitrary regional boundaries, the precise extent of each region was then formed by amalgamating several of Natural England’s ‘Natural Areas’ in England (these have since been succeeded by ‘National Character Areas’, which differ slightly and are more refined than the original Natural Areas; Natural England 2014). Natural Areas were designed by Natural England to inform decision-making concerning the natural environment, and are intended to describe geographical, ecological and historical variations in landscape character. Their boundaries follow physical lines in the landscape rather than administrative boundaries, and these were used in order to marry the physical characteristics of the landscape with the archaeological evidence as closely as possible. In the absence of an equivalent set of character areas for Wales, topography was used, though

06/09/2016 11:59:13

16

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

1.5.  Map showing Natural England’s ‘Natural Areas’ within the Roman Rural Settlement Project regions

more distinctive landscape zones were identified for the area of South Wales included in the Central Belt region (e.g. Gwent Levels, Vale of Glamorgan etc.; see Ch. 5). It is important to recognise that each of the Roman Rural Settlement Project regions incorporates considerable intra-regional diversity, as clearly demonstrated by the range of different Natural Areas, which often display quite markedly diverse settlement characteristics (fig. 1.5). The eight regions defined for this project must not, therefore, be regarded as homogeneous entities, but rather as convenient units for purposes of inter-regional comparison. Each displays some common physical and cultural characteristics which make them preferable to using either entirely arbitrary modern administrative units, or adopting the numerous Natural Areas defined by Natural England as the sole units of analysis. However, individual Natural Areas, termed ‘landscape zones’ for our purposes, have formed the principal units for exploring sub-regional variation, allowing intra-regional comparison of rural settlement chronology, form and building styles. In regions with relatively small numbers of excavated sites, however, such micro-scale

Romanch1.indd 16

sub-division has not always been practical. In these cases other methods of sub-division have been employed on a case-by-case basis. THE RESEARCH STRUCTURE The establishment of our eight regions in this chapter, the definition of our settlement types in Chapter 2 and their constituent buildings in Chapter 3, provide a framework for the characterisation of the rural settlement of each region as set out in Chapters 4 to 11. The chronological data allow analysis of change over time: of the different classes of settlement, their buildings and their distributions across the landscape zones, and in relation to the road network, the major towns and military establishments of each region between the first century b.c. and the fifth century a.d. The characterisation of the rural settlement of each region is further enriched through the analysis of the material culture and ecofactual data associated with the settlement hierarchy. The concluding Chapter 12 draws all the regional approaches together to give a province-wide perspective.

06/09/2016 11:59:32

Chapter 2

RURAL SETTLEMENT IN ROMAN BRITAIN: MORPHOLOGICAL CLASSIFICATION AND OVERVIEW By Martyn Allen and Alexander Smith INTRODUCTION

straightforward. Although some excavations are of sufficient scale to reveal ‘whole’ settlement plans, many investigations only reveal parts of the occupation site, providing little evidence for overall layout and form. These ‘fragmentary’ sites are further explored here to show why they present an important contribution to the study, and how they might become classified within the proposed model. Once established, this framework provides a more systematic basis for examining regional diversity, continuity and change in Roman rural settlement patterns over a considerable proportion of England and Wales. A further point to make here is that many rural settlements were, of course, immersed in a landscape of field systems, which represent the proportion of land directly or indirectly exploited for food production, either through the cultivation of cereals and other crops or the husbanding of domestic livestock animals (Fowler 2002, 127–60; Chadwick 2007; 2008). Field systems are not directly dealt with in this chapter, but are instead considered in more detail in their regional contexts in Chapters 4 to 11.

It has long been accepted that the rural settlements of Roman Britain were profoundly diverse in their size, function and form, and that there is a great deal of regional variation (Thomas 1966; Hanley 1987; Hingley 1989; Millett 1990, 181–211; Mattingly 2006, 379–427; Taylor 2007). Although a range of terms used to describe rural settlements persists in the archaeological literature, there exists no standard consensus for classifying them in a systematic manner. Rather than detailing objective criteria for distinguishing between different forms of rural settlement, many surveys have tended to present the evidence from exemplary examples of excavated sites (‘typesites’), in order to highlight regional diversity (Miles 1989; Hingley 2004; King 2004a; Mattingly 2006). Taylor (2007, 8) has recently commented that ‘there has been little agreement on common terms to describe or analyse the nature and form of distinctive rural settlements, apart from the recognition of certain architectural traditions, such as villas’. An added problem is encountered in the array of vernacular terminology used to describe regionally based settlements, such as ‘ladder settlements’, ‘scooped settlements’, ‘hut groups’, and ‘rounds’ (Hogg 1966; Jobey 1966; Quinnell 1986; Stoertz 1997). These have the effect of creating caricatures of regionality, and although they may be useful in local contexts, they are less helpful for inter-regional, comparative analysis of known Roman rural settlement evidence across England and Wales. This chapter examines the full range of settlement sites recorded in the database. It is intended to provide an overview of the main types of settlement encountered, summarising their principal characteristics, distribution, and the level of variation inherent within each group. By classifying rural settlements on the basis of their scale, context and morphological complexity, this framework provides a model by which settlements can be cross-examined independent of their location. While much of the regional detail of the Romano-British rural settlement pattern will be presented in Chapters 4–11, broad trends across England and Wales outlined here will provide a wider context. As we shall see, however, the process of classifying sites is less than

CLASSIFICATION OF RURAL SETTLEMENT As discussed in Chapter 1, traditionally, studies of Roman rural settlement have focused almost exclusively on villas, in effect isolating them from the rest of the rural settlement landscape (Percival 1976; Scott 1993; Smith 1998). Even in more detailed regional surveys the emphasis upon villa settlement is clear, though most archaeologists now accept that this is a product of the greater visibility of their surviving remains (Henig and Booth 2000; Bird 2004; King 2004a; Russell 2006). While the significance of villas is less in question, it is the precise definition of what constitutes a villa that is problematic, since this can vary quite dramatically between different studies. It is for the most part a definition based upon certain architectural elements (masonry building, tiled roof, painted plaster walls, hypocaust, bathhouse, etc.), though even here the interpretations are wide-ranging, with Smith (1998, 10), for example, including any building that provided evidence of a ‘Roman form of 17

Romanch2.indd 17

14/09/2016 12:44:43

18

THE RURAL SETTLEMENT OF ROMAN BRITAIN

planning’ (see Ch. 3 for discussion of how villas have been classified in this project). Hingley (1989, 3, 75–8) was one of the first Roman archaeologists to move away from simply looking at architecture to define rural site types, and to begin to explore the character and development of wider settlement morphology, placing an emphasis upon whether farmsteads (or ‘compounds’) were open or enclosed. The morphology of Roman rural settlement was developed by Taylor (2007, 19–21), who distinguished between ‘enclosed settlement’, ‘open settlement’ and ‘linear system settlement’. Further distinctions were made within these settlement categories using size and period data, against which additional evidence for architecture, craft activity, burial/ceremonial practice and the presence of military activity, could be explored (ibid.). Taylor’s (ibid., 12) decision to classify sites by settlement form, rather than traditional site classes such as ‘farmstead’, ‘villa’, ‘small town’, etc., was prompted by his inclusion of large quantities of site data gathered from cropmark and earthwork surveys, which populated many of the Historic Environment Record (HER) archives. The approach to rural settlement classification taken here essentially follows on from the work of Hingley and Taylor, taking into account settlement size, morphological complexity, architectural distinctions, local context and aspects of the material culture. Rather than simply accepting the interpretations found in excavation reports, each site entered into the database has been re-evaluated, based upon these factors. While most of our classifications have followed the original interpretations of the excavators, re-assessments

table

Settlement type

Romanch2.indd 18

2.1: quantification

South Central East Belt

have been necessary on occasion to standardise the classification of all sites recorded across the country. Romano-British rural settlements were clearly diverse, and are perhaps best understood as a morphological continuum, though it still remains essential to attempt to classify sites in order to better understand their variety. As outlined in Chapter 1 (see table 1.1), the main site types used in this study are domestic settlements, specialist industrial sites, specialised religious/funerary sites, and sites that contain elements of rural landscape but for which the associated domestic settlement is not known, such as field systems, trackways, and livestock enclosures. Domestic settlements have been divided into larger nucleated sites and smaller rural settlements, the majority of which are simply defined as ‘farmsteads’, lacking any of the architectural characteristics that would suggest that they should be assigned as villas (see Ch. 3) (table 2.1). Both farmsteads and villas have been further classified on morphological grounds (their physical layout), following the tripartite divisions defined by Taylor (2007) outlined above, with ‘open’, ‘enclosed’ and ‘complex’ (a development of Taylor’s ‘linearsystem’) forms (see details below, pp. 21–33). It must be acknowledged here that, although many farmsteads have been classified based upon their settlement morphology, a significant number have not.This does not mean that these ‘unclassified farmsteads’ are no longer of use within the study. On the contrary, many provide significant and clearly identifiable settlement features and finds assemblages. They were simply not useful for the analysis, which focused upon their morphological characteristics. Unfortunately, those settlements

of excavated settlements by type

NorthEast

SouthUpland Wales West and the Marches

Central West

North

Total

%

Open farmstead

 19

  31

  5

 17

 4

 5

  2

  3

  86

3.3

Enclosed farmstead

111

 138

 28

 51

37

45

 38

 53

 501

19.1

Complex farmstead

 43

 135

 21

 34

 1

 0

  3

  8

 245

9.3

Unclassified farmstead

255

 583

 91

126

32

10

 58

 20

1175

44.7 12.4

Villa

113

 157

 14

 20

 2

 6

 14

  0

 326

Total farmsteads/villas

522

1012

154

231

72

61

113

  81

2333

Roadside settlement

 34

  84

 30

 14

 4

 2

 12

  2

 182

6.9

Village

 21

  23

  4

  7

 2

 1

  0

  1

  59

2.2

Vicus

  2

   2

  3

  5

 0

17

  8

 16

  53

2.0

Total nucleated settlements

 57

 109

 37

 26

 6

20

 20

  19

  294

Total

579

1121

191

257

78

81

133

100

2627

14/09/2016 12:44:44

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

defined architecturally as villas are rarely able to be classified morphologically, whether as ‘complex villas’, ‘enclosed villas’ or ‘open villas’. This is primarily a consequence of the large proportion of villa excavations having been undertaken prior to 1990 and the widespread adoption of developerfunded archaeology. In this context, the emphasis of the excavations typically focused upon the villa buildings rather than the wider form of the settlement. For this reason, the ‘villa’ category in table 2.1 has not been further sub-divided on morphological grounds in the same way as farmsteads, although such divisions are discussed further below, and in the regional chapters for the South (Ch. 4), the Central Belt (Ch. 5) and the East (Ch. 6). PROBLEMS AND LIMITATIONS OF SITE CLASSIFICATION While our classification criteria are firmly set out and adhered to, as with any taxonomic system, problems and limitations exist. Most commonly these pertain to excavations that only minimally expose settlements, but they also arise from the

% of sites

diverse excavation strategies utilised on different sites, and the variation in recording and reporting. Obviously, the larger the area of excavation, the more we can understand about the form and layout of a settlement. The frequency of farmstead/ villa settlements that can be assigned to a morphological type – either open, enclosed or complex – clearly reflects the total area of excavation (fig. 2.1). Sites with relatively small areas of excavation (below 0.5 ha) remained most frequently unclassified, whereas above 0.5 ha they are more likely to be placed within a morphological class. Furthermore, the area of excavation can have an effect upon the classification of specific settlement types, with smaller investigations less likely to identify open and complex forms (fig. 2.2). This may contribute to a bias towards enclosed settlements, which must be taken into account in any analysis of settlement distribution. In some cases the use of non-intrusive techniques, such as geophysics or aerial survey, alongside small-scale excavation, allows sites to be classified. At Melton wastewater works near Brough, East Riding, for example, a complex

classi ed (n=740)

50

19

not classi ed (n=1194)

40 30 20 10 0 up to 0.1ha >0.1-0.5ha >0.5-1.0ha >1.0-2.0ha >2.0-5.0ha

>5.0ha

Area of excava on fig. 2.1.   Frequency of farmsteads and/or villas with and without the attribution of morphological classifications (open, enclosed or complex) by area of excavation

enclosed se lement (n=430) 35

open se lement (n=77)

complex se lement (n=233)

% of sites

30 25 20 15 10 5 0 up to 0.1ha >0.1-0.5ha

fig.

Romanch2.indd 19

>0.5-1.0ha >1.0-2.0ha Area of excava on

>2.0-5.0ha

>5.0ha

2.2.   Frequency of classification of farmstead and/or villa settlements by area of excavation

14/09/2016 12:44:46

20

THE RURAL SETTLEMENT OF ROMAN BRITAIN

N

Geophysics plot Excavated feature 0

50 m 1:1000

fig. 2.3.  Site plan of Melton wastewater works near Brough, East Riding (Neal 2002), showing complex farmstead identified through geophysical survey and trial trench evaluation

farmstead could be identified from a combination of evaluation trenches and geophysical survey (fig. 2.3). Non-intrusive techniques are of course inhibited by the fact that, in isolation, they cannot disentangle different phases of development of a site, with a palimpsest of features likely to obscure the identification of specific phases of a settlement’s development. In the case of the Melton wastewater works site, most of the features were identified as being broadly contemporary, indicating the presence of an enclosure complex with areas of domestic activity and internal trackways (Neal 2002). It is notable that the results of this evaluation were significant enough to persuade the developers to move the construction of the wastewater facility 200 m to the east, preserving the settlement in situ. This entailed the excavation of a second area of land, revealing further features that most likely related to peripheral areas of the now preserved farmstead, expanding our understanding of its size and function (Bishop and Westwood 2004). Another issue associated with fragmentary evidence is how to differentiate a farmstead (a single-unit, domestic settlement) from a nucleated settlement, such as a village or a roadside settlement. For some ‘fragmented’ sites, we have been able to identify a number of different excavated elements that are likely to have been part of a larger nucleated settlement (for example, a number of excavations around Bishop’s Cleeve, Gloucestershire, appear to reveal parts of a possible village; e.g. Joyce 2010a; Lovell et al. 2007). In other cases, we must rely more on contextual and

Romanch2.indd 20

artefactual evidence, such as Lincoln Road, Enfield, in Greater London (Gentry et al. 1977). Here, a very small excavation provided little indication of the nature of the site, but an assessment of its position next to Ermine Street just north of London, coupled with the fact that it produced a large number of Roman coins, suggested that the site was actually part of a larger roadside settlement. Sites in Greater London perhaps provide the greatest difficulties in this regard. The scale of twentieth-century urban expansion around the capital has provided many opportunities for investigating settlement evidence in its hinterland, but the relatively small-scale excavations that have been undertaken mean that there are real difficulties in identifying and understanding the form and character of the settlements (Bird 2004, 60–67). Other constraints on the classification of settlements include the truncation of the upper levels of settlement remains, a lack of understanding or definition of settlement phasing, and excavation strategies that are too focused upon targeting structural evidence rather than wider settlement features (particularly relevant for many villa excavations). Pre-excavation decisions could place greater emphasis upon the classification of settlements and, although largescale excavations are not always possible, investigation techniques that seek to characterise/ classify the settlement as a whole, rather than focus on one small part of it, are desirable if we are to continue to develop our understanding of Romano-British rural settlements. FARMSTEADS One of the most important developments of the past 25 years, in terms of the study of Roman rural settlement, has been the greatly increased number of farmsteads (i.e. small rural settlements without ‘villa’ architecture) that have been excavated and reported (fig. 2.4). Of the 1866 farmsteads recorded on the database, over 80 per cent have been reported on since 1990, enabling a complete transformation of our understanding of economy and society in the Romano-British countryside, unshackled from the previous dominance by villas. As discussed above, for various reasons not all farmsteads are able to be classified morphologically, and even where this is possible, there exists a considerable degree of regional, local and individual variation across England and Wales. A discussion now follows of the three broad types of farmstead defined in this project, along with some account of the extent of this variation. More detailed regional patterns will be examined in Chapters 4–11, where attempts are made to move

14/09/2016 12:44:47

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

fig.

2.4.   Distribution of all excavated sites recorded as farmsteads

beyond morphological classification, to explore associations with landscape context, chronology, economy and social status. OPEN FARMSTEADS Unenclosed or ‘open’ farmsteads are defined as settlements where there does not appear to be any traceable boundary enclosing the main domestic core, as illustrated by the mid- to late Roman phase of settlement at Strood Hall, Essex (fig. 2.5). Here, a series of roundhouses were set alongside a number of pits, an enclosure, and a track/droveway that led directly to the settlement, which otherwise appears to be unbounded at this time (Timby et al. 2007a). Open farmsteads have a wide distribution across England and Wales, though they are relatively few in number, possibly because, as discussed above, they are difficult to identify from smaller excavations. It is also likely that the lack of ditches on open farms reduces their visibility as cropmark sites, which would have an impact on excavation strategies. They are most densely concentrated in and around the Central Belt region, with a particular cluster in the Middle Thames Valley. Further north, they appear almost

Romanch2.indd 21

21

completely absent from large areas of the Midlands, reappearing again in Yorkshire and County Durham. Although comparatively sparse, the distribution of excavated open farmsteads generally follows that identified from cropmarks and other survey techniques by Taylor (2007, 27), but there remains a lack of them in the northern half of the Pennines. In this region, Taylor was able to locate from survey evidence numerous open settlements (ibid.), but the dating and function of many of these has yet to be corroborated by excavation. Within the current dataset, open farmsteads are almost exclusively a later Iron Age phenomenon, their existence perhaps reflecting a declining midto late Iron Age settlement pattern (cf. Knight 2007). Analysis of the subsequent histories of open farmsteads demonstrates considerable variability, with 28 being abandoned, 16 becoming enclosed farmsteads, 17 developing into complex farmsteads and 8 being replaced by field systems. Where open farmsteads developed into new forms of settlement, there is evidence for both continuity in domestic activity from an unenclosed phase, and for complete reorganisation of their land use. In the Middle Thames Valley there is good evidence

14/09/2016 12:44:59

22

THE RURAL SETTLEMENT OF ROMAN BRITAIN Roundhouse gullies

Pits

N

Trackway

0

100 m 1:2000

fig. 2.5.   Distribution of excavated open farmsteads and plan of mid-Roman open farmstead at Strood Hall, Essex (Timby et al. 2007a)

of late Iron Age open farmsteads becoming overlain by new settlement types and their associated field systems, indicating a complete change in local land management during the first century a.d. (see case study, Ch. 4). This evidence forms part of a wider pattern of development at

Romanch2.indd 22

this time, when the vast majority of changes to open settlement occur. There is no evidence for open farmsteads developing into other forms of settlement after the second century a.d., and the remainder were gradually abandoned between then and the end of the fourth century.

14/09/2016 12:45:11

23

N

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

Corndryer

Projected boundary 0

100 m 1:2000

2.6.  Distribution map of excavated enclosed farmsteads and plan of enclosed farmstead at Bishopstone, East Sussex (Bell 1977)

fig.

ENCLOSED FARMSTEADS Enclosed farmsteads are defined as settlements where all, or the majority, of domestic and associated activity was contained within one or two enclosures and where internal space was not

Romanch2.indd 23

further sub-divided to a significant degree (e.g. the ‘round’ at Trethurgy, Cornwall: see Ch. 11). They are the most widespread and frequently identified farmstead type, found in most areas of England and Wales, but particularly common across northern parts of England, around the

14/09/2016 12:45:23

24

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Welsh coast, in Cornwall, and on the southern chalk downland (e.g. Bell 1977; fig. 2.6). In a number of regions their distributions form small clusters, such as on the Yorkshire Coal Measures and on the Hampshire Downs. In contrast, they appear quite rare in areas of Lincolnshire and Nottinghamshire, and Gloucestershire and Somerset, despite the fact that relatively high numbers of settlements are recorded in these rec linear

areas. A number of unexcavated examples are seen as cropmarks in the Cotswolds, which highlights the problem of relying on excavated evidence alone (RCHME 1976). Despite being considered here as a homogeneous group, enclosed farmsteads vary considerably in form and construction technique, and have been further divided into four morphological types (rectilinear, curvilinear, irregular, and D-shaped).

curvilinear

irregular

D-shaped

% enclosed farm type

70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 East

Romanch2.indd 24

Central Belt

South

North-East

Central West

North

fig.

2.7.   Relative frequencies of different enclosed farmstead types by region

fig.

2.8.   Distribution of enclosed farmsteads by type

Upland South-West Wales and the Marches

14/09/2016 12:45:37

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

Rectilinear enclosures are those with only straightsided circuits. They are usually rectangular in shape (e.g. fig. 2.6), though a few are square or trapezoidal. Curvilinear enclosures have a boundary that curves along its full length, and are either circular or oval in shape (e.g. fig. 2.9). Importantly, they do not have corners. D-shaped enclosures have, as their name implies, one straight side with the remainder of the circuit curving around the settlement. Irregular enclosures do not necessarily conform to any particular shape, though the vast majority are of a form that fall half-way between a rectilinear enclosure and curvilinear enclosure (e.g. Ch. 8, fig. 8.9, Whitwood Common). They may be approximately square or rectangular in outline (i.e. with corners), but they do not have any straight sides. These enclosure types show distinct geographic patterning (figs 2.7 and 2.8). Enclosed farmsteads in the South, Central Belt and East regions are predominantly of rectilinear types, with curvilinear and D-shaped types in the minority. In the NorthEast and Central West rectilinear types continue to dominate, but higher frequencies of irregular types occur (c. 30–35 per cent). In Upland Wales and the Marches and in the North, rectilinear, enclosed farmsteads no longer dominate the settlement pattern but feature among a more mixed group, including higher proportions of both irregular and curvilinear enclosed types. In the South-West a

25

region, the pattern is completely reversed by a dominance of curvilinear enclosed farmsteads, exemplified by the type commonly known as ‘rounds’ (e.g. Quinnell 1986). Even in regions dominated by a particular type of enclosed farmstead, small clusters of other types can be identified, such as the group of curvilinear, enclosed farmsteads present on the Hampshire Downs in the South region (see Ch. 4). These are predominantly of the well-known, ‘banjo’ form, characterised by their broadly circular enclosure and funnelled entrance (fig. 2.9). Banjo enclosures are generally Iron Age in date, though some are known to have been respected or even maintained into the Roman period (Cunliffe and Poole 2008c). They are thought to be mainly a chalkland phenomenon of southern England, although the form has been revealed elsewhere, as for example a coastal enclosure at Ewanrigg on the Solway Plain, Cumbria (Bewley 1992). Further work in this region is required to indicate whether or not this is an isolated example. In Pembrokeshire and Carmarthen, banjo-type enclosures have been excavated at Drim, Dan-y-Coed, Woodside Camp and Penycoed, Dyfed, each broadly dating between the first century b.c. and second century a.d., with the latter possibly continuing into the fourth century (Murphy 1985; Williams and Mytum 1998). A group of about 20 enclosures, not b

N

c

0

100 m 1:2000

fig. 2.9.  Plans of ‘Banjo’-type enclosures at (a) Ewanrigg, Cumbria (Bewley 1992), (b) Dan-y-Coed, Dyfed (Williams and Mytum 1998) and (c) Grateley South, Hants (Cunliffe and Poole 2008c)

Romanch2.indd 25

14/09/2016 12:45:38

26

THE RURAL SETTLEMENT OF ROMAN BRITAIN

dissimilar to banjos, were identified in West and South Yorkshire, with a strong focus around South Kirby, occurring on both the Magnesian Limestone and Coal Measures (Roberts 2010, 30, 33). Recently, it was argued that banjo enclosures in Gloucestershire were high-status sites that were inextricably linked to the late Iron Age oppidum at Bagendon (Moore 2012). Further analysis of the finds and environmental assemblages from these types of settlement may provide useful crossregional comparisons to see how far they reflected similar lifestyle and farming practices. In some regions, enclosed settlement boundaries were defined with earthen embankments or masonry walling, though earthen embankments are likely to be under-represented, owing to destruction by later ploughing, especially in central and southern England. Most examples with surviving evidence for an embankment or walling occur in Cornwall, west Wales, and in northern England, particularly in the Hadrian’s Wall area (fig. 2.10). In Cornwall and Dyfed, earthenwalled, enclosed farmsteads are common, associated with a predominance of curvilinear enclosed types in these areas. The settlements in

fig.

Romanch2.indd 26

Gwynedd in the north-west of Wales differ by having numerous farmsteads surrounded by drystone-walled enclosures. The drystoneenclosed settlements of this region are underrepresented in our dataset, since many are known through aerial and field surveys, but very few have been systematically excavated (Waddington 2013). There is evidence for the use of stone in revetments on sites on the Magnesian Limestone, west of the Vale of York, which tend to date to the early Roman period (e.g. O’Neill 1999; Brown et al. 2007, 66; Waddington 2012, 46–7; Martin et al. 2013). However, the small numbers of sites that have been investigated in Wales suggest that the settlements there had long histories, commonly extending from well before the first century b.c. up to the fourth century a.d. and beyond (Waddington 2013). A notable aspect of these types of enclosed farmstead is that many of the internal structures are contiguous with the outer walls, and, although many produce evidence for different construction phases, they appear to be largely homogeneous entities (Fasham et al. 1998; fig. 2.11). These distinctive forms of settlement will be further examined in Chapter 10.

2.10.   Distribution of enclosed farmsteads with identified masonry- or earthen-walled enclosures

14/09/2016 12:45:51

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

Evidence for both double- and, more rarely, triple-ditched enclosures is present in variable numbers across the country (table 2.2). Multiditched enclosures are most frequently encountered in the South-West region, associated with the curvilinear earthwork enclosures found in Cornwall. The higher proportion of such multiple ditch and embankment boundaries here may be due to greater table

2.2: numbers

Region

27

perceived defensive needs, or as an expression of social status. They are quite different from the earthwork farmsteads found in Dyfed which, as has been shown, tend to be of the ‘banjo’-type. Triple-ditched enclosures are generally rare across England and Wales, but are mostly found within the South region. However, even here there is rarely much conformity in the form and size of

of single, double and triple-ditched enclosures by region

Single ditch no. %

Double ditch no. %

South  93 89.4  5  4.8 Central Belt 101 84.9 16 13.4 East  20 83.3  4 16.7 South-West  22 71.0  8 25.8 North-East  44 95.7  2  4.3 Central West  28 77.8  8 22.2 Upland Wales and the Marches   38 90.5   4   9.5 North  40 81.6  9 18.4 Total 386 56

Triple ditch no. %

Total

6 5.8 104 2 1.7 119 0 0.0 24 1 3.2 31 0 0.0 46 0 0.0 36 0 0.0 42 0 0.0 49 9 451

Phase I

Phase II

Phase III

Phase IV

Phase V

Phase VI

Bank

Cropmark Projected building New building Remaining from earlier phase (in use) Remaining from earlier phase (not in use) fig.

Romanch2.indd 27

N

0

50 m 1:1000

2.11.   Development of drystone masonry-walled enclosed settlement at Cefn Graenog II (Fasham et al. 1998)

14/09/2016 12:45:51

28

THE RURAL SETTLEMENT OF ROMAN BRITAIN a

b

c

N

0

100 m 1:2000

fig. 2.12.   Plans of multi-ditched enclosed farmsteads at (a) Waylands Nursery, Wraysbury, Berkshire (Preston 2003), (b) Dairy Lane, Nursling, Hants (Adam et al. 1997), and (c) Orsett Cock, Essex (Carter 1998)

these enclosure types. Some may be seen as symbolically defensive, such as the late Roman settlement at Waylands Nursery, Berkshire, where three concentric rectilinear ditches enclose a number of four-post structures located towards the corner of the enclosure (Preston 2003; fig. 2.12(a)). At other sites, an elaboration of the enclosure boundary may have been designed to control movements of livestock into and out of the bounded areas, such as the two discontinuous, corner ditches within the enclosure at Dairy Lane, Nursling, Hampshire (Adam et al. 1997; fig. 2.12(b)). The use of annexes is also a feature of some enclosed farmsteads and may relate to the penning of livestock away from the main habitation areas (e.g. Carter 1998; fig. 2.12(c)). While enclosed farmsteads have been categorised as a distinctive group of settlements, differentiated from open and complex farmsteads, there is clearly great variation within the group. The shape of enclosed farmsteads, the manner of their enclosure, and the arrangement of entrances reflect a substantial range of sub-types. In many

Romanch2.indd 28

areas there are numerous enclosed farmsteads that are irregular in form, and in certain cases it is difficult to see why such shapes were chosen rather than more regular rectilinear and curvilinear layouts. It is possible that some reflect topographic elements that are not visible in the archaeological record, such as tree or hedge lines, although much of the variance may simply reflect local preference. COMPLEX FARMSTEADS Complex farmsteads are defined as settlements where there appears to be significant differentiation of space, either as a system of conjoined enclosures or as a principal outer enclosure with many internal sub-divisions (e.g. Cotswold Community, Wilts/ Glos.; Powell et al. 2010; fig. 2.13). The differentiation of space tends to reflect different activity areas (e.g. domestic, storage, agricultural processing, industrial, livestock enclosures etc.), though excavation is not always extensive enough to enable such zones to be defined. In certain instances, the enclosure of these areas might be seen as discrete and progressive developmental stages.

14/09/2016 12:45:52

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

29

N

Industrial/working area

Storage and transit?

Agricultural/ working area

Stock management? Domestic

Trackway

0

100 m 1:2000

fig. 2.13.   Distribution of excavated complex farmsteads and plan of complex farmstead at Cotswold Community, Wilts/Glos. (Powell et al. 2010)

Romanch2.indd 29

14/09/2016 12:46:05

30

THE RURAL SETTLEMENT OF ROMAN BRITAIN

The growth in numbers of complex farmsteads appears to be a phenomenon of the early Roman period, although their distribution is regionally varied. They occur most frequently in the Central Belt region and in lesser numbers in the NorthEast, East and South regions, but are largely absent from the South-West, Upland Wales and the Marches, and appear infrequently in the North. This pattern is illustrated in fig. 2.13, which shows their concentration around the Fens, with clusters of sites located along the river valleys of the Ouse, the Nene, and, to a lesser extent, the Middle and Upper Thames (see Ch. 5). They also occur widely, if not as densely, across the low-lying areas of Yorkshire, north of the Humber and on the Magnesian Limestone. Compared with enclosed farmsteads, complex farmsteads are very rare on the chalk downland, and appear to be predominantly situated in the river valleys. In general, the distribution pattern of complex farmsteads follows the distribution of ‘linear system settlements’ identified by Taylor (2007, 26).

Owing to the fact that complex farmsteads consist of multiple-bounded areas, there is a wide range of forms, varying far more than that of open and enclosed farmsteads. However, despite this variation, two main systems of development in complex farmsteads have been identified, subdivided enclosures and linear complexes. Sub-divided enclosures are best described as large enclosed farmsteads, with internal areas large enough for the space to be significantly sub-divided (e.g. Wavendon Gate, Bucks; Williams et al. 1996; fig. 2.14). They often appear to involve a degree of planning, with large external boundary ditches constructed initially, followed by divisions formed by shallower ditches or gullies. In contrast, linear complexes often involve a series of connected enclosures, usually extending out from a domestic focus (e.g. Haddon, Cambs (Hinman 2003); NIAB Huntington Road, Cambridge (Luke 2014); fig. 2.15). They often incorporate significant land boundaries, usually either trackways or field systems, and appear to have developed more

N

0

100 m 1:2000

fig. 2.14.   Plan of early Roman complex farmstead (sub-divided enclosure type) at Wavendon Gate, Buckinghamshire (Williams et al. 1996)

Romanch2.indd 30

14/09/2016 12:46:05

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

31

a N

Barn?

Barn?

Structure 3 Structure 2 b

Cistern

Cistern

0

100 m 1:2000

fig. 2.15.   Plans of complex farmsteads (linear complex type) at (a) Haddon, Cambridgeshire (Hinman 2003), and (b) southern site at NIAB Huntington Road, Cambridge (Luke 2014)

Romanch2.indd 31

14/09/2016 12:46:06

32

THE RURAL SETTLEMENT OF ROMAN BRITAIN

linear complexes

sub-divided enclosures

complex farmsteads (uncertain)

120

no. of sites

100 80 60 40 20 0 Late Iron Age

Late 1stC A.D.

3rdC A.D.

fig.

2.16.   Chronology of different types of complex farmsteads in use

fig.

2.17.   Distribution of complex farmsteads by type

organically compared with sub-divided enclosures. The distinguishing factor between these two forms of settlement is in the orientation of their development: sub-divided enclosures develop internally, and linear complexes develop externally. Linear complexes are the most common type of complex farmstead, with a total of 134 being recorded on the database, outnumbering the 61 sub-divided enclosures by more than 2:1. A further 50 complex farmsteads could not be placed into either category as their plans were

Romanch2.indd 32

2ndC A.D.

4thC A.D.

insufficiently complete. All variants of complex farmsteads peak in numbers during the second and third centuries a.d. (fig. 2.16). Geographically, both sub-divided enclosures and linear complexes may be found in relatively close proximity, for example around the Fen edge and in the Ouse Valley, though there are concentrations of linear complexes in the Thames Valley and around the estuaries of the Severn and the Humber (fig. 2.17). The form of these types of settlement suggests that they may have been primarily

14/09/2016 12:46:21

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

80

linear complexes

sub-divided enclosures

33

complex farmsteads (uncertain)

70 no. of complex farms

60 50 40 30 20 10 0 virgin establishment

fig.

enclosed seƩlement

unclassiĮed farms/ Įeld system

2.18.  The origins of complex farmsteads

involved in livestock husbandry, although the presence of corndryers in many of them also suggests an arable component to the economy. In many of the linear complexes the enclosures are contiguous, and on some sites funnelled entrances can be observed. The systems appear highly suited to managing herds of cattle or flocks of sheep where the animals could be herded, counted and separated. Trackways running alongside the settlement would have aided the control of their movement beyond its confines. This evidence may, in part, explain the regular siting of many linear complexes in river valley and estuarine locations, places where natural pastures and water sources could have been exploited. The increasing number of linear complexes through the later first and second centuries a.d. may be interpreted as evidence for an expansion of pastoral farming, or an intensification of animal management in areas where there was increasing pressure on land availability owing to heightened arable production. A relatively large number of complex farmsteads can be shown to have developed from pre-existing domestic settlements or field systems (fig. 2.18). Eleven sub-divided enclosures developed from simple enclosed settlements which, to some extent, indicate that these were simply more developed forms of enclosed farmsteads. However, it is also notable that more than half of the known complex farmsteads developed in areas with no evidence of immediately preceding activity. Although much of this land may have previously been used as grazing, the establishment of such farmsteads suggest that these landscapes were becoming increasingly organised and managed by the second century a.d. It has become increasingly clear that complex farmsteads, either in the form of large and more developed enclosed farmsteads, or the more extensive linear complexes, were a particular characteristic of the Roman period, with a distinct distribution largely restricted to the Central Belt and parts of the North-East. Their appearance perhaps suggests changes to the way that local

Romanch2.indd 33

open seƩlement

landscapes were being managed and how livestock was being husbanded. It is likely that they were related to an increasing focus upon the production of surplus for market. VILLAS Villas have been understood first as high-status ‘Romanised’ dwellings, then as economic entities, settlements that were better able to manage and exploit their surrounding landscape to generate surplus wealth (see papers in Branigan and Miles 1989). More recently, greater emphasis has been placed on the role of the villa as a medium for transmitting a set of cultural ideals and as a social arena for receiving and entertaining guests (Scott 2004; Taylor 2011; Millett 2014). Much work still needs to be undertaken on their roles as ‘producers’ and ‘consumers’, and their place in the settlement hierarchy. In total, 326 settlements classed as villas are recorded in the database. Other sites of possible ‘villa status’ exist, but these lack detailed excavation on the building(s) (see Chs 3–7 for specific cases). Millett (2014) has suggested that villas in Roman Britain may number around 2000, although these still only represent perhaps as little as one per cent of all settlement. Villas clearly constitute more than one per cent of the settlements in our dataset, and are almost certainly over-represented compared with other farmsteads, because of their archaeological visibility and their long history of excavation. The overall distribution of villa settlements is relatively well known (e.g. Mattingly 2006, 379– 99), being particularly well represented through the Central Belt region, from the Cotswolds, across the Chilterns and up into the river valleys that drain into the Fens, a distribution similar to that of the majority of complex farmsteads. The distribution of excavated villas correlates well with the distribution of all known or suspected villas recorded on the NMR, though some disparity is evident, particularly south of the Humber where excavated examples are

14/09/2016 12:46:22

34

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

2.19.   Distribution of excavated villas in relation to NMR ‘villas’

comparatively rare (fig. 2.19). Villas also cluster along the chalk downland in the South region, particularly along the North and South Downs and into Hampshire, although these are areas where complex farmsteads are comparatively rare. Beyond these core areas, villa settlement is more sparsely spread across East Anglia, Lincolnshire and Yorkshire, and across the West Midlands into Wales. Other areas are almost completely devoid of villas, including northern England and along the west coast in Cheshire, Devon and Cornwall. Absences in the south and east of the country, such as in the Weald and along the lower and middle Thames valley, are more localised. The isolation of villas as a separate ‘settlement category’ is probably more apparent than real, since many developed from other types of farmstead. The number of villas constructed during the Roman period is shown in fig. 2.20, and illustrates the considerable increase in new builds during the second century a.d. Around half of all villa establishments in each century have been recorded as developments from earlier settlements. This is almost certainly an under-representation since many excavations, particularly those undertaken pre-1990, placed much emphasis upon discovering

Romanch2.indd 34

the plan and date of the villa structures, rather than exploring the possibility of pre-villa activity. While some villas may be de novo establishments, it is very difficult to establish these conclusively on the basis of negative evidence. Even the so-called ‘protopalace’ phase at Fishbourne, West Sussex, is now known to have been preceded by an earlier Roman and late Iron Age phase of occupation (Manley and Rudkin 2005). Many other excavation reports mention the recovery of late Iron Age or early Roman pottery that pre-dates the first phase of villa construction, but information regarding the context of such material has often been lacking. In fact, of the villa sites with proven pre-villa settlement activity, the great majority (80) have their origins in the late Iron Age or earlier, demonstrating that most sites which became villas had long periods of prior activity and development. Villas, then, represent significant investments of wealth designed to monumentalise the pre-existing settlement (Millett 2014). The changing density and distribution of villas across the country between the later first and fourth century a.d. is shown in fig. 2.21. It is widely known that much of the earliest villa construction in Britain occurred on the south

14/09/2016 12:46:34

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

new establishment/previous domes c ac vity not iden

35

ed

villa developed from earlier domes c focus 80 70

no. of villas

60 50 40 30 20 10 0 1stC AD

Romanch2.indd 35

2ndC AD

3rdC AD

4thC AD

fig.

2.20.  The origins of villas (from date of earliest villa establishment)

fig.

2.21.   Kernel density distribution of excavated villas from the late first to fourth century a.d.

14/09/2016 12:47:21

36

THE RURAL SETTLEMENT OF ROMAN BRITAIN enclosed villa

20

complex villa

18 16 no. of villas

14 12 10 8 6 4 2 0 Central Belt

fig.

South

North-East

East

Central West Upland Wales and the Marches

2.22.   Number of villas able to be classified by settlement morphology type by region

a

b N

0

100 m 1:2000

fig. 2.23.  Plans of late Roman ‘complex’ villa at Barton Court Farm, Oxon (Miles 1986) (a) and mid-Roman ‘enclosed’ villa at Chilgrove 2, West Sussex (Down 1979) (b)

coast of Sussex and in north Kent (cf. Rudling 1998; Millett 2007a). The early Sussex villas included some particularly grand examples, such as Fishbourne, Southwick, and Pulborough, each placing an emphasis upon the design of central courtyards and reception areas, architectural aspects more readily paralleled in Gaul and the Mediterranean (Taylor 2011, 181). These sites were certainly the exception, interpreted as representing investments of wealth by local elite groups with each, potentially, the centre of a large estate (Cunliffe 1973, 79). The early Kent villas, however, are fairly evenly spaced along the major river valleys of the Medway, Cray and Darenth, which flow north into the Thames estuary (see also Taylor 2011, fig. 1). Some have interpreted

Romanch2.indd 36

these as the residences of those exploiting and exporting local resources such as Kentish ragstone, quarried along the border of the North Downs and the Weald (Boyce 2007; Elliot 2014). By the mid-second century a.d., the number of villas found in the Central Belt region is approximately the same as that in the South, while some investment was also starting to be made further north and west. The number of villas found in the Central Belt region continues to increase into the fourth century a.d., while those in the South and in other regions reach a peak in the third century. As discussed above, relatively few villas have information regarding their immediate settlement and landscape context. However, of those that do, it is notable that those from the Central Belt tend

14/09/2016 12:47:23

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

to be of complex form (complex villas) and those from the South tend to be of enclosed form (enclosed villas) (fig. 2.22), in general following the dominant patterns already demonstrated by the farmsteads in those regions (for complex and enclosed nomenclature see under respective farmstead types above). It is important to note that the categorisation of a villa as either complex or enclosed does not reflect its relative size or status. For example, the small corridor villa-house at Barton Court Farm, Oxfordshire, was constructed in the late third century a.d. within a reorganised sub-divided enclosure, with several areas divided up into spaces for different farmstead activities (Miles 1986; fig. 2.23a). This site shows no more evidence for architectural refinement than the enclosed villas at Chilgrove 2 (Down 1979; fig. 2.23b), Batten Hanger (Magilton 1991; Gardner 2009), and Beddingham (Rudling 1998), all located on the chalk downland in Sussex, each set within a simple rectilinear enclosure, following the layout of many other farmsteads in that region. If excavation were to place a greater focus on the landscape context of villas, it would greatly enhance our understanding of them. It is commonly assumed that villas were highly

37

productive economic entities, but, for the most part, we do not know how the inhabitants of villa settlements generated their wealth (Millett 2007a; Taylor 2011). Perhaps it was through arable production and processing, as is indicated at sites like Yewden, Buckinghamshire (Eyers 2011), or Northfleet, Kent (Andrews et al. 2011), or the production of building materials as at Ashtead, Surrey (Bird 2014). It is also uncertain whether the wealth used to construct villas in the first instance was produced through the workings of the settlement or through money gained from elsewhere; only with a better understanding of their settlement context can we make more informed interpretations about how villas may have functioned. NUCLEATED SETTLEMENTS: ROADSIDE SETTLEMENTS, MILITARY VICI AND VILLAGES Nucleated settlements are clearly of importance for any understanding of the wider RomanoBritish rural economy, yet remain relatively unknown through a lack of recent academic

fig. 2.24.   Distribution of excavated nucleated settlements (vici along Hadrian’s Wall and the Stanegate are omitted)

Romanch2.indd 37

14/09/2016 12:47:35

38

THE RURAL SETTLEMENT OF ROMAN BRITAIN

attention and synthesis. The parameters regarding what is included in this project as a ‘nucleated settlement’ have been set out in Chapter 1, with the term encompassing all larger settlements (generally 3 ha and above), except for the major urban centres (including civitas capitals and coloniae) and defended ‘small towns’. The classification includes a wide range of sites, generally referred to in excavation reports either as ‘small towns’ or ‘villages’, often with little in the way of differentiation between the two (Hanley 1987; R.F. Smith 1987; Burnham and Wacher 1990). This project has sub-classified smaller nucleated settlement primarily into villages and roadside settlements, based exclusively on whether or not the site had direct association with the Roman road network. Civilian settlements that originated next to forts – the so-called military vici – are also included in this group of settlements (except for those in the Hadrian’s Wall/Stanegate area; see Ch. 1). These were similar to roadside settlements in that they developed along major roads. However, because of their association with military sites, these settlements may have functioned quite differently to other roadside settlements and deserve to be examined as a distinct group. In this chapter, they are included and summarised with roadside settlements, but they are given more detailed, individual attention in the regional chapters, particularly those for the North-East (Ch. 7), the North (Ch. 9), and Upland Wales and the Marches (Ch. 11). For further discussion of

the layout and typology of vici, the reader is referred to Sommer (2006) and, for a gazetteer and summary discussions of settlements in Northern England, to Bidwell and Hodgson (2009). As a group, nucleated settlements are broadly distributed across England and Wales, though the density of sites is far from even, with particular concentrations of roadside settlements in parts of the East Midlands and East Anglia, and those sites defined as villages being more prevalent in parts of Gloucestershire, Wiltshire and Somerset. Military vici represent the primary form of nucleated settlement in the north and west (fig. 2.24). ROADSIDE SETTLEMENTS AND MILITARY VICI The vast majority of the nucleated sites recorded in the project database are those that developed alongside the major metalled roads, where their form is generally described as being ‘ribbon-like’, signifying the influence of the road. This can, however, be an oversimplification of the matter. A combination of excavation and geophysics at Westhawk Farm, Kent, for example, has shown that the roadside settlement there included small bounded plots of land, perhaps equating with individual properties, on the west side of the road, but an uneven and irregular distribution of features to the east, where the road from Lympne joins the junction at the settlement (Booth et al. 2008; fig. 2.25). There are many examples of roadside settlements and some military vici, which, rather

N

Geophysics anomalies - archaeological Geophysics anomalies - ferrous Excavated area fig.

Romanch2.indd 38

0

200 m 1:5000

2.25.   Geophysical survey plan of the roadside settlement at Westhawk Farm, Kent (Booth et al. 2008)

14/09/2016 12:47:38

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

than developing along one stretch of road, spread along two or more roads that intersected in the vicinity. For example, geophysical survey at the Welsh forts at Cefn Caer and Caer Gai, both in Gwynedd, has shown evidence for settlement ‘ribbons’ along several roads radiating from the forts (Hopewell and Burman 2007; fig. 2.26), while at Stamford Bridge in East Riding,Yorkshire, at least three roads converged, with excavations there demonstrating how densely the settlement features can build up in the area of the crossroads (Roe 2009; fig. 2.27).

39

However, without large-scale excavation, geophysical survey or cropmark evidence, the plans of these settlements are very difficult to determine. Sites located around London, such as at Staines and Brentford, are particularly disadvantaged in this regard, and our understanding of them is only developed through a synthesis of usually small-scale, targeted excavations (Canham 1978; Jones 2010). Unfortunately, many of these excavations were undertaken at intervals many years, sometimes decades, apart employing different excavation and publication strategies,

Caer Gai farm and outbuildings

Agger

Vicus N

Fort Annexe

Barracks Bathhouse Annexe Bypass road

se

dd

riv

e

Road

Road

Di

su

Mansio Side-road and industrial area 0

200 m 1:5000

fig.

2.26.   Geophysical survey plan of the fort and vicus at Caer Gai, Gwynedd (Hopewell and Burman 2007) Road

Wicker-lined pits Enclosures N

0

100 m 1:2000

fig.

Romanch2.indd 39

2.27.   Excavation plan of the roadside settlement at Moor Lane, Stamford Bridge, Yorks (Roe 2009)

14/09/2016 12:47:39

40

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Bre

thought to have been the site of a mansio (Winbolt 1924), was built on the south side of the River Arun. As noted above, the further extent of this settlement was revealed with evidence from fieldwalking, geophysics and small-scale excavation, indicating occupation expanding south from the enclosure along the road for some distance (Luke and Wells 2000; Thompson 2006a). One inference that might be drawn from this example is that the road system and the waterways were used in conjunction for moving people and goods, and perhaps even livestock, although direct evidence for the use of rivers for transport is still fairly minimal (see Chs 5, 6 and 7 in particular). Another possibility is that tolls were paid to cross the river, which may have provided at least some level of economic stimulus to the settlement. As just discussed, some roadside settlements may well have grown to a considerable size (e.g. Old Sarum (Sorviodunum), which possibly covered 36–45 ha (Moffat 2010)), and were likely to have been just as important as those nucleated settlements which at some point received masonry or earthwork enclosures. However, there are also a number of such settlements that were located within relatively close proximity to major walled/ defended towns, and must have had a strong relationship with them. Across the Roman province, there were 37 roadside settlements located within 10 km of a defended town, 14 of these lying within just 3 km, possibly in some cases part of a continuous ribbon development. Lying 2 km south of the major walled town at Catterick (Cataractonium), and dating between the

ham

Ice no

tten

Ve nt a

x x

w x

Occupation? Romano-British leat Road w Wharf? Causeway Masonry building Temple x Burial(s) Metalworking Tanning Crop-processing Mill/’maltings’ complex?

ru m

making synthesis difficult if not impossible. Elsewhere, extensive geophysical survey in the Vale of Pickering has revealed linear Roman settlements extending over several kilometres along a trackway (Powlesland et al. 2006). The scale of these settlements, and the presence of a track rather than a metalled road, suggests that they are perhaps best considered as villages (see below, p. 41 and Ch. 7). The size of nucleated settlements is also difficult to define, though surface survey techniques have been successfully undertaken at some sites to demonstrate the spread of Roman material culture along roadsides, reflecting the approximate extents of the settlements (e.g. Alfoldean, West Sussex: Luke and Wells 2000; Shiptonthorpe, Yorkshire: Millett 2006). At Great Walsingham in Norfolk, various metal-detecting and fieldwalking surveys (and very limited excavation) have suggested a settlement perhaps spread over 50 ha, immediately south of where the Roman road crosses the River Stiffkey (Norfolk HER 42850). However, estimations of the extent of settlements based purely upon such surveys could be potentially misleading, with some of the material culture perhaps representing manure scatters or outlying farmsteads (e.g. Garton 2008). A considerable number of roadside settlements and military vici are located at the crossing of rivers, as for example at Scole on the Norfolk/ Suffolk border, which lies on either side of the River Waveney (Ashwin and Tester 2014; fig. 2.28). At Alfoldean, West Sussex, a large ditched enclosure with internal masonry structures,

(Extent

un

cle

x

x

ar)

x (E

xt

en

t

Needh

am/Dit

unclear)

chingh

am

N

Temple site

River Waveney

x Camulodunum

?

0

500 m 1:10000

fig. 2.28.   Interpretative plan of the roadside settlement at Scole, either side of the River Waveney on the Norfolk/ Suffolk border (Ashwin and Tester 2014)

Romanch2.indd 40

14/09/2016 12:47:40

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

later first century a.d. and the middle of the fourth century, the roadside settlement at Bainesse, North Yorkshire, has provided evidence for domestic activity running alongside Dere Street. Previous excavations have revealed property plots, including timber and masonry structures, along with the remains of a pottery kiln (Busby et al. 1996; Wilson 2002), while current excavations in advance of new motorway construction is showing almost continuous activity along Dere Street which stretches for a number of miles (Wilson pers. comm.). Field systems were also found to back on to the enclosure plots, with geophysical survey showing that they ran for at least 500 m and perhaps further towards Cataractonium (Speed 2006). At Dringhouses, c. 3 km south-west of the colonia at York, evidence for successive timber structures, hearths, and cremation burials, has been excavated, perhaps also from within enclosed plots similar to those seen at Bainesse (Ottaway 2011). Another probable roadside settlement lies c. 2 km south of Silchester Roman town at Latchmere Green, where the Roman road splits south-east and south-west to Chichester and Winchester respectively. Settlement remains have been identified here from an array of surface finds, including quantities of pottery, ceramic building

41

materials, quernstones, slag, glass, animal and human bone, covering an area of around 6 ha (Corney 1984; Fulford and Creighton 1998). Small-scale excavation indicated that the settlement, which had late Iron Age origins, was principally inhabited between the late first/second century a.d. and the late fourth century, while a subsequent pipeline excavation revealed the flint foundations from three walls, along with the presence of pottery and tile wasters, and smithing slag (Brading 2011). As with many of the other examples, it remains uncertain if we are looking at ‘separate’ settlements or else elements of extensive, though not necessarily intensive, ribbon developments stretching from the larger urban core. It is only with greater exploration of the hinterlands of the main Roman towns that we will be able to address such questions. VILLAGES The sites defined in this project as villages include all nucleated settlements that do not appear to have developed along, or in relation to, the major Roman road system. As such it is a very broad category, and it must be emphasised that the meaning of the term ‘village’ in this context is quite different from the village in later medieval or

Building 3

Building 1 Early track

Building 2

N

Buildings Excavation trenches 0

200 m 1:5000

fig. 2.29.   Earthwork survey plans with locations of excavation trenches within village settlements at Chisenbury Warren, Wiltshire (McOmish et al. 2002; Fulford et al. 2006) and Chalton, Hampshire (Cunliffe 1977)

Romanch2.indd 41

14/09/2016 12:48:08

42

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Kiln Roman structures Roman Major Later Iron Age enclosures Roman cemeteries

N

0 fig.

200 m 1:5000

2.30.   Plan of excavated features in the Roman village settlement at Mucking, Essex (Lucy and Evans 2012)

modern times, which developed under entirely separate historical and social circumstances (Millett 2014). Villages are in one sense an agglomeration of farmsteads, and it is the evidence for multiple centres of domestic activity that separates villages from complex farms, which had a single domestic focal point. Villages could spread over considerable areas, with some of those visible as earthworks on Salisbury Plain extending well over 20 ha (McOmish et al. 2002; Fulford et al. 2006). As discussed above, the distribution of villages appears to be quite restricted (see fig. 2.24), but this is likely to reflect the fact that these types of settlement are very difficult to identify without large-scale excavation or survey. Their apparent concentration in parts of central southern England is in part the result of the levels of earthwork preservation on the Chalk downland, with good examples revealed at places like Chisenbury Warren on Salisbury Plain and Chalton on the Hampshire Downs (McOmish et al. 2002; Cunliffe 1977; fig. 2.29). Most of these downland earthwork sites have received very little or no

Romanch2.indd 42

intrusive archaeological investigation, though some villages elsewhere have been more extensively excavated. Archaeological investigations over 18 ha at Mucking on the South Essex Thames Estuary revealed a major Iron Age settlement that was extensively redeveloped after the Roman conquest into a series of enclosures and droveways with two main areas of domestic settlement (Lucy and Evans 2012; fig. 2.30). Excavations on this scale remain very much a rarity, though recent geophysical survey, fieldwalking, evaluation and excavations over 60 ha at Love’s Farm St Neots, Cambridgeshire, revealed a landscape of trackways, fields and 2–3 zones of domestic occupation, with nucleation appearing to develop during the early Roman period (Oxford Archaeology East 2012). As can be seen from these examples, villages can cover very large areas and, where the evidence is available, appear to have developed quite organically, often from Iron Age origins. However, just how these settlements functioned in relation to the surrounding landscape and other settlements has yet to be determined, the majority having received little investigation.

14/09/2016 12:48:10

MORPHOLOGICAL CLASSIFICATION AND OVERVIEW

SUMMARY This chapter has presented an overview of the major types of Roman rural settlement found across England and Wales. It is recognised that such settlement is inherently varied, often poorly understood and represents a continuum of forms and scales that on many occasions defy any further categorisation. Nevertheless, the mass of mainly developer-funded excavations of the past 25 years has produced a significantly increased dataset of Roman rural sites to work with, and within this certain broad classes of settlement have been identified, defined here on the basis of morphology, size, and context, as well as architecture and finds assemblages, which will be drawn upon in the following chapters. Those settlements classified as farmsteads (i.e. without villa architecture) have

Romanch2.indd 43

43

seen the most notable increase in the volume of excavation and publication during this time. These have been further categorised, primarily on the basis of form, in order to provide a framework for additional analyses to be presented in subsequent chapters, based upon shared characteristics of chronology, economy, and social status. The distribution of all excavated rural settlement across England and Wales has been shown to be far from even, but the number and spread of sites is sufficient to reveal very distinct regional patterns, which will be explored in Chapters 4 to 11. Prior to this, the following chapter will explore further the settlements themselves, to take account of the great variability in building forms and functions that exist between the different types of site and across the different landscapes of the Roman province.

14/09/2016 12:48:11

Chapter 3

BUILDINGS IN THE COUNTRYSIDE By Alexander Smith

INTRODUCTION

At either end of this continuum, from singleroomed timber or mass-walled structures to large multi-courtyard masonry complexes, there were obviously huge differences, both architecturally and socially, but between these extremes was a myriad of often more subtle variation, perhaps reflecting different individual, cultural and economic choices. With all due acknowledgement of this complex variation, there have been a number of attempts at broad classification, at least of those buildings or complexes explicitly defined as villas. Richmond’s (1969) account of villa plans has formed the basis for comparative analysis in many subsequent studies (e.g. Hingley 1989, 30; Perring 2002, 72; Martins 2005, 3), comprising ‘cottage’, ‘corridor’, ‘winged-corridor’ and ‘courtyard’ types. Smith’s (1998) subsequent critique of this system proposed a more complex typology based upon perceived social expression of the architecture, and included hall houses, row houses, courtyard houses and multiple variants thereof. Within all such classification, however, there are inherent issues with blurred boundaries between types, and with the recognition that most buildings will have complex architectural histories. In many cases such histories have not been recognised during excavation, either because of restricted areas of investigation and/or significant truncation, or else the buildings were investigated prior to the implementation of modern archaeological techniques, and at best we are left with plans depicting the final stages of active occupation. The current study of rural settlement in Roman Britain has taken into account those excavated buildings/complexes interpreted as villas in all their different forms, generally following the broad classificatory system of Richmond (1969) just outlined (see Villa architecture p. 71). However, it is immediately apparent that these only represent a significant minority of architectural forms in the Romano-British countryside, with the vast bulk of the buildings in most areas being of much less complex form, albeit still with considerable variation. With the mass of mainly developerfunded excavations, we are now in a position to create an account of Romano-British rural architecture that is not totally dominated by villa buildings, and which demonstrates pronounced levels of regional and local diversity.

The study of rural architecture has often been at the forefront of previous accounts of the RomanoBritish countryside (e.g. Rivet 1969; Branigan 1977b; Morris 1979; Hingley 1989; Scott 1993; Smith 1998; Perring 2002; Martins 2005; Taylor 2011). The archaeological remains of buildings in rural settlements have been classified and used to provide frameworks through which significant insight has been gained into aspects of social and economic expression, as well as giving a deeper understanding of wider regional characteristics. With some notable exceptions (e.g. S. Clarke 1998; Taylor 2001; 2013), these studies have concentrated on one broad category of elite rural architecture – the Roman villa. Definitions of the term ‘villa’ can vary significantly from one author to another (see Chs 1 and 2), but – in archaeological terms – it is generally taken as meaning rural buildings of the Roman period with architectural characteristics associated with high-status prominent display, such as tiled roofs, painted plaster walls, mosaic floors, under-floor heating and often with associated bathhouses (Hingley 1989, 21; King 2004a, 349; Roymans and Derks 2011b, 2). Smith, however, advocated a far wider approach (1998, 11). For him the ‘villa’ embraced almost any domestic rural architecture within a Roman province that was unlike ‘native’ forms, and thus accentuated the binary ‘Roman’ and ‘non-Roman’ division inherent in the concept of ‘Romanisation’, which has been critiqued at length within wider theoretical debates about the reproduction of social identities in the Roman period (e.g. Woolf 1998; Mattingly 2004; Van Oyen 2015). Eleanor Scott (1993), in her comprehensive gazetteer of Roman villas in Britain, was equally wide ranging in her interpretative parameters, listing over 1600 entries, based not only on excavated buildings but also on any presence of material such as roof tile, wall plaster and window glass in a rural context. However, this work failed to take into account the wider settlement context and the various possible interpretations of the surface material. Both Smith (1998, 11) and Scott (1993, 6) did recognise that there was an architectural continuum across time and space rather than a strictly defined and easily recognised set of classifiable buildings. 44

Romanch3a.indd 44

15/09/2016 15:55:03

BUILDINGS IN THE COUNTRYSIDE

45

fig. 3.1.   Distribution of all excavated late Iron Age and Roman rural buildings showing percentages of settlements in each region with evidence for defined buildings

THE ARCHITECTURAL DATASET

2% 2% 1% 2% 2% 1% 3% 3% 4% 4%

The current study has collated information on farmstead (all) farmstead (all) 6175 different buildings from 1563 sites across roadside settlement 8% roadside settlement England and Wales, ranging in date from the late 8% villa 41% villa Iron Age to late Roman period. As shown in fig. village 41% village 3.1 this architectural evidence is well spread vicus vicus across the regions, being recorded in c. 60 per cent hillfort 19% hillfort of South, Central Belt, East and North-East religious site 19% religious site industrial site region settlements and up to 80 per cent of industrial site isolated building settlements in the North, the latter due mainly to isolated building greater preservation of earthworks in upland areas. 20% 20% Most of the recorded buildings come from sites interpreted as farmsteads, with approximately fig. 3.2.  The context of buildings: percentage of total equal numbers from complex (628) and enclosed number of buildings by site type (n=6175) (674) farmsteads, followed by roadside settlements, villa complexes, villages, military vici, hillforts and In terms of building density, this remains fairly a number of ‘non-settlement’ sites (fig. 3.2). The modest at 2.4 buildings per settlement across the emphasis on non-villa farmstead buildings is whole dataset, although this does vary by region, important for achieving a more balanced with both the North and Upland Wales and understanding of Romano-British rural Marches exceeding four buildings per settlement architecture across the whole country, which, as for reasons outlined above (i.e. better preservation Taylor has demonstrated for parts of the East of earthworks) (fig. 3.3a). There are also obvious Midlands and Wroxeter hinterlands, can in turn and expected differences in the quantity of lead to a more nuanced appreciation of local and regional community identity (Taylor 2013). buildings by settlement type (fig. 3.3b), with a

Romanch3a.indd 45

15/09/2016 15:55:12

THE RURAL SETTLEMENT OF ROMAN BRITAIN

(b) (b)

5 4 3 2 1 0

8 7 6 5 4 3 2 1 0

fig. 3.3.  

6

9 average no. of buildings per settlement

(a) (a)

average no. of buildings per settlement

46

farmstead farmstead farmstead (all) (enclosed) (complex) (n=1866) (n=501) (n=245)

village (n=59)

Average number of buildings per site by (a) region and (b) settlement type (n=total number of settlements)

much higher average number in nucleated settlements, followed by villa complexes then farmsteads. The nature of complex farmsteads, with extensive evidence for multiple differentiated zoning, is undoubtedly behind an increased architectural presence over enclosed farmsteads, although the differences are not huge. The overall number of buildings recorded, although considerable, would of course only represent a small fraction of the total count originally found across the late Iron Age and Roman countryside. Buildings, which are defined here as constructions with walls and roofs at a suitable scale for human entry, have only been included in the dataset if they are of a recognisable form, defined most often by specific arrangements of postholes, beam slots or masonry foundations. There are at least a further 450 settlements where potential elements of building foundations were noted but with no discernible patterns, or else there are what appear to be building platforms, but no obvious traces of sub-structure. It has been suggested in some regions (e.g. Thames Valley: Allen et al. 1984; Booth et al. 2007, 36) that masswalled (e.g. turf) buildings may have become more common during the late Iron Age and Roman periods, with a resultant increased difficulty in archaeological recognition (though possibly indicated by penannular gullies; see below, p. 51). Perring (2002, 98–106) describes an array of such earth, cob and clay construction techniques for Romano-British buildings, though definite examples are encountered only sporadically, and then mostly in better preserved urban environments.

Romanch3a.indd 46

villa vicus (n=53) roadside (n=326) settlement (n=182)

table

3.1: recorded

characteristics of buildings

Building characteristic

Variables recorded

Form

rectangular, circular

Structural material

masonry, timber/mass-walled

Differentiation of space multi-room building Chronology

overall use of building, form and material

Architectural elaboration

tiled roof, painted plaster, architectural stonework, window glass, tessellated floor, hypocaust

Specific building type

aisled building, cellared building, villa types

ARCHITECTURAL RECORDING In a large-scale project such as this it has not been possible to record detailed information on each and every building, and instead base-line characteristics have been recorded as set out in table 3.1. It is appreciated that such relatively coarse level data will omit many more subtle developments, such as the differentiation between part- and full-masonry (although drystone and mortared masonry buildings are differentiated where possible), and circular and oval buildings, though such variables are not always possible to determine, and even then may not always be defined as such within reports. Nevertheless, the overall quantity of data is sufficient to enable a fairly robust characterisation of rural architectural type and form, on a chronological, geographical and site/landscape context basis.

15/09/2016 15:55:12

BUILDINGS IN THE COUNTRYSIDE

BUILDING FORM AND MATERIAL Rural architecture in Roman Britain is characterised by great complexity in physical form, from simple, single-roomed timber or masswalled roundhouses to palatial, multi-courtyard, rectilinear, masonry villas. At the most basic level, this can be broken down into the fundamentals of morphology and building material, as outlined above: circular vs rectilinear and timber vs masonry. The functional and ideological impacts of such dichotomies have been discussed at length in various archaeological, historical and architectural studies (e.g. Steadman 2006; Pope 2008; Bradley 2012), and the broad evidence from rural settlements in late Iron Age and Roman Britain will now be outlined. CIRCULAR AND RECTANGULAR ARCHITECTURE The main domestic architectural form in Britain prior to the Roman invasion was the roundhouse, which had origins in the early Bronze Age, and reached maximum numbers across the country in the middle Iron Age (Pope 2008, 14). Thereafter, the numbers of circular buildings in many parts of the south and east appear to have declined significantly, or at least were built in a style that is not easily recognisable in the archaeological record. However, it is recognised that roundhouses did continue in use into at least the early Roman period, and remained very numerous in the north and west throughout the third and fourth centuries a.d. (Hingley 1989, 31; Perring 2002, 51; Taylor 2007, 31; see Chs 9–11). A recent study of roundhouses in Wales revealed an exceptionally long-lived history, from as early as the later third millennium b.c. to the later first millennium a.d., with one-third of the sites in the study having their origins in the Iron Age and most of these appearing to continue in occupation through to the Roman period (Ghey et al. 2007).

A total of 2659 circular, or at least curvilinear, buildings from 813 sites were recorded in the current dataset, representing 43 per cent of the total number of structures. Although not all of these would have been domestic roundhouses, it is likely that the greater percentage were, with a relatively small number of others having specific functions noted such as agricultural buildings, workshops and shrines (see Building function below). On a national level there is a steady reduction in the number of sites with circular architecture over time, with the most noticeable period of decline being observed across the late Iron Age to early Roman transition, when there is fall from 40 to 24 per cent of settlements containing such buildings (fig. 3.4). As will be shown in Chapters 4 to 11, however, there are significant regional variations to this national pattern, with, for example, the major period of decline in the relative frequency of circular buildings being the second century a.d. for the East region and the third/fourth centuries for the North. In a general discussion on architectural geometry, Steadman (2006, 128) argued that circularity in plan was often a characteristic of freestanding, widely spaced, single-room houses in pre-industrial societies. With a greater desire for physical differentiation of space came a change from single-room to multi-room houses and a resulting change from circular to rectangular plan shapes, since the latter made it easier to pack together rooms in plan and offered greater flexibility in the event of a requirement to expand the overall dimensions of the building (e.g. multistorey) (ibid.). The gradual change from predominantly circular to predominantly rectangular architecture in most parts of Roman Britain is likely to be linked at least in part with such a need for greater architectural differentiation and flexibility, as buildings (circular and rectangular) became increasingly used as tools for

Circular

800

47

Rectangular

700

no. of sites

600 500 400 300 200 100 0 Late Iron Age fig.

Romanch3a.indd 47

late 1stC AD

2ndC AD

3rdC AD

4thC AD

3.4.  Use of circular and rectangular buildings over time

15/09/2016 15:55:13

48

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 3.5.  Distribution of all excavated circular and rectangular buildings, c. late Iron Age–late Roman

Romanch3a.indd 48

15/09/2016 15:55:25

BUILDINGS IN THE COUNTRYSIDE

Iron Age

1st century A.D.

2nd century A.D.

3rd century A.D.

49

4th century A.D. Circular building Rectangular building Circular and rectangular buildings Other excavated rural site Roman road

fig.

Romanch3a.indd 49

3.6.  Distribution of sites with different architectural forms from the late Iron Age to the 4th century a.d.

15/09/2016 15:56:34

50

THE RURAL SETTLEMENT OF ROMAN BRITAIN

% of buildings

more complex expressions of cultural identity and hierarchy (see Multi-space architecture below). The 3517 rectangular buildings from 1079 sites represent 57 per cent of the known number of structures in the current dataset. Unlike circular buildings there is far more direct evidence for a multitude of different functions other than purely domestic, partly because of the increased scope of such buildings for division into multiple rooms that could be utilised for different purposes (see Building function below). Although comparatively rare, there are many instances of rectilinear structures dating to the late Iron Age, occurring sporadically across most parts of England and Wales, and forming just under 5 per cent of all buildings in use at this time (fig. 3.4). Traditionally many Iron Age rectangular buildings are thought to have had non-domestic functions, often interpreted as shrines (e.g. those in Danebury hillfort; Cunliffe 1995), though usually only on the basis of their incongruity among other forms of contemporary architecture. In a review of the evidence, however, Moore (2003) pointed out that these buildings may have served a multitude of functions, including domestic occupation and as animal byres. The well-known post-built rectangular buildings at Goldcliff on the Gwent Levels are mostly of middle Iron Age date (Bell et al. 2000), and further buildings of a similar nature and date have recently been found on the Sussex Downs at Peacehaven (Hart 2015, 175). In both instances the buildings have been suggested as seasonal huts, used for temporary human and animal occupation in a society based upon transhumance (Moore 2003, 53; Hart 2015, 180). Most late Iron Age rectangular buildings recorded in the current dataset are quite insubstantial structures, and of uncertain use, such as the small posthole building from Bluntisham in the Ouse Valley (Burrow and Mudd 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

rectangular

2010). However, larger, more defined structures of late Iron Age date do exist, as seen at Newhouse Park, Chepstow, on the margins of the Severn Levels (not too far from the buildings at Goldcliff), where major excavations revealed a substantial, multi-phase rectangular building with a metalled floor, associated with much domestic debris and dated to the early first century a.d. (Robic and Ponsford 2008; see fig. 3.16 below). The increased adoption of rectilinear forms over time on a national basis is clear, these eventually becoming the dominant form across most of the Roman province by the second to third centuries a.d. (fig. 3.4). As with circular buildings, however, the situation is highly variable across the different regions, with the South for example having a dominance of rectangular buildings by the later first century a.d. and architecture in the South-West region always remaining predominantly curvilinear (including oval and ‘boat-shaped’ buildings; see Ch. 11). The distribution of circular and rectangular architecture can be seen in fig. 3.5. Both forms are found across the country, though clearly at differing densities with rectangular buildings dominating in much of the south and east, and circular buildings to the north and west, though there is also a highly significant cluster in and around the Fenland region. As already stated, there was a gradual, though piecemeal, shift in most areas from predominantly circular to predominantly rectangular architecture, and this is seen geographically in fig. 3.6. Even during the late Iron Age, parts of the south-east display a much greater diversity of architectural form than those areas further west and north, and it is the south-east that sees the greatest proliferation of rectangular architecture during the later first century a.d. The major expansion of rectangular buildings across the province can be seen in the circular

fig. 3.7.  Relative frequency of circular and rectangular buildings according to major settlement type (n=total number of buildings identified)

Romanch3a.indd 50

15/09/2016 15:56:35

BUILDINGS IN THE COUNTRYSIDE

second century a.d., and by this time it is only parts of north-east England, north-west Wales and Cornwall where circular buildings still dominate, with only the latter area remaining in this situation by the fourth century. However, even during the later Roman period, there are areas such as the Fenlands and its western periphery that persist with a relatively strong, circular, architectural signature (see Ch. 5). The distribution of building forms across England and Wales shown in figs 3.5 and 3.6 encompasses all forms of rural settlement, and thus the apparent proliferation of rectangular buildings from central Wales and parts of the north-west, for instance, is largely due to the presence of military vici rather than a reflection of ‘native’ rural architecture. The proportion of circular versus rectangular architecture according to the major settlement types is shown in fig. 3.7, and in part correlates with general chronological patterns, with settlement types more resonant of the Iron Age having far greater proportions of circular buildings. However, the national disparity between farmstead types seen here is not purely chronological, as the number of rectangular buildings does not reach parity with circular buildings on enclosed farmsteads until the fourth century, whereas on complex farmsteads they surpass them by the second century and become ever more dominant thereafter. This suggests that changing architectural styles were generally concurrent with changes in settlement form across the countryside, although the picture, as ever, is very mixed and there can be substantial variation in form even within a relatively limited area, with different settlements adopting different architectural styles over time. Furthermore, within the broad classification of circular and rectangular buildings there was also little homogeneity, with a great deal of differentiation in complexity and use of space, along with building materials and architectural elaboration. BUILDING MATERIALS AND APPEARANCE The extensive variety of building materials and techniques used in rural Romano-British architecture has really only started to be fully appreciated since the advent of developer-funded archaeology in 1990, and the resultant huge increase in non-villa buildings being investigated. Even then, much of the discussion has still revolved around villas and townhouses (e.g. Johnson and Haynes 1996), although Perring’s work on the Roman house in Britain did provide an account of variable building materials and methods of construction (Perring 2002, 80–110). Many recent site excavation reports also enter into

Romanch3a.indd 51

51

considerable detail on this matter, such as the 36 mostly timber-framed buildings from the major roadside settlement at Wilderspool, Warrington (Rogers and Garner 2007) and the 62 timber buildings from the inland port at Camp Ground, Colne Fen, Cambridgeshire, including platform, beam-slot buildings and post-built structures (Evans 2013b, 236–77; see Building function below). The national scale and scope of the current project has meant that specific construction details have not been systematically recorded for all buildings in the dataset, and instead a basic division between timber/mass-walled and masonry structures has been catalogued, as displayed in fig. 3.8. Prior to the Claudian conquest, timber or masswalled architecture was almost exclusive across much of Britain, mostly in the form of roundhouses and, indeed, many Roman period roundhouses retained similar construction techniques (though see discussion of masonry circular buildings below). Most of these buildings were only represented by penannular gullies, usually interpreted as drainage gullies around the exterior of the superstructure. In some cases the building’s walls were defined by ring grooves for solid split timbers, or else stakeholes used in wattle and daub construction (Pope 2008, 17). However, in other cases, there was no indication as to the building’s superstructure, either through truncation of timber slots/posts or else use of a mass-walling technique such as turf. Despite this uncertainty over the construction material, the term ‘timber building’ will be used throughout the remainder of this volume in order to describe such buildings, and to differentiate them from those structures built using some form of masonry. The 3370 timber buildings in the current dataset represent 55 per cent of the total number of structures, which is almost certainly a significant under-representation, as they are often quite insubstantial (especially box-framed and beamslot structures; see discussion of buildings at Staines, Ch. 4) and were rarely identified prior to the middle of the twentieth century. The predominance of such buildings in the Iron Age has already been noted, and they were still very common during the Roman period, when a variety of new carpentry techniques were introduced, including timber framing, that ensured ever more substantial and sophisticated wooden buildings were capable of being built (Goodburn 1992, 197; Perring 2002, 83). The widespread distribution of timber buildings across England and Wales is shown on fig. 3.8, dominating the excavated building record in most areas. Exceptions include places where there is a ready supply of good building stone, such as in the Cotswolds, although

15/09/2016 15:56:35

52

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 3.8   Distribution of all excavated timber/masswalled and masonry buildings

Romanch3a.indd 52

15/09/2016 15:56:47

BUILDINGS IN THE COUNTRYSIDE

fig.

3.9.  Distribution of circular masonry buildings

the apparent lack of timber buildings here may be just an artefact of the historical concentration of excavations on villas in this area. Overall, timber was undoubtedly the main construction material used within most rural buildings in Roman Britain, and the variety of techniques employed within these structures is quite evident; however, systematic analysis of these construction methods lies beyond the scope of this study. Masonry buildings are defined as those having at least stone footings, even if the remainder of the building may well have been cob or timber framed. A total of 2713 such ‘stone’ buildings were recorded, only apparent in most areas during the Roman period, although drystone masonry structures (predominantly roundhouses) did occur in parts of Dorset, Cornwall, north-west Wales and north-east England from the late Iron Age onwards (fig. 3.8). The first significant development of masonry buildings (including the first mortared structures) occurred in the mid- to late first century a.d. in parts of the south-east, and to a lesser extent in the Central Belt. As with rectangular buildings, it is not until the second century a.d. that masonry buildings become more widespread across the country, including the

Romanch3a.indd 53

53

north and west, although here they generally still remained as drystone structures outside of military contexts. Most masonry buildings in the dataset were rectangular though 570 circular masonry structures were also recorded, the distribution of which demonstrates distinct clusters (fig. 3.9). Those to the north and west represent for the most part a continuation of late Iron Age masonry roundhouse tradition, with diameters typically somewhat smaller than those of average timber roundhouses, such as at Middle Gunnar Peak, Barrasford, Northumberland, where three excavated buildings had thick walls (1–1.5 m) of irregularly shaped blocks with a rubble and earth core and internal diameters ranging from 4 m to 7.5 m (Jobey 1981). In contrast, many of the circular masonry structures in the south and east, including the major concentration in and around the Nene and Upper Ouse Valleys, were often substantial and occasionally mortared buildings, as for example at the roadside settlement of Higham Ferrers in the Nene Valley, where four structures of second century a.d. date were revealed with tightly packed ironstone foundations and internal diameters up to 10.7 m (giving an interior area of 90 m²)

15/09/2016 15:56:54

54

THE RURAL SETTLEMENT OF ROMAN BRITAIN

(Lawrence and Smith 2009, 52; cf. FriendshipTaylor and Friendship-Taylor 1997). For the most part (though not in all areas; see Ch. 6), such buildings are likely to have developed from preRoman architectural traditions, perhaps being used as active agents in the creation of distinct community identities (Taylor 2013). In terms of associated settlement context, timber structures account for almost 75 per cent of buildings within farmsteads, representing everything from major agricultural barns to postbuilt roundhouses. There is no overall evidence for complex farmsteads having a higher percentage of masonry buildings, being for the most part associated with rectangular post-built and beamslot structures, although occasionally including larger aisled buildings, themselves with a substantial timber component. Villa complexes are, unsurprisingly, associated with very high levels of masonry architecture (76 per cent of buildings in these sites), though this is in part due to a lack of excavation beyond the main villa building in many cases. Where investigation has been more extensive, for example around the periphery of the Ingleby Barwick villa in County Durham, a range of timber and masonry buildings is often encountered (Willis and Carne 2013). Within nucleated settlements, timber architecture is dominant (60 per cent) at villages, while a much more mixed range of masonry and timber buildings is encountered in roadside settlements and military vici. Architectural elaboration The nature of any superstructure and elaboration within rural buildings is quite poorly understood in most cases, especially outside of villa complexes, where, unsurprisingly, most of the academic attention has been focused (Johnson and Haynes 1996; Smith 1998; Perring 2002). As discussed above, such villa buildings are defined by the presence of features such as tiled roofs, painted plaster walls, tessellated floors and hypocaust systems, though there are still many settlements incorporating buildings (including occasional circular structures) with some of these attributes without necessarily being defined as villas. The most frequently encountered evidence is for tiled-roof buildings, interpreted by the quantity of roofing material (ceramic or stone), which is attested at c. 20 per cent of sites where no villa has been defined, including 109 farmsteads, 67 roadside settlements, 15 military vici and 33 religious sites. Evidence for painted plaster walls and glazed windows is encountered far less often (c. 12 per cent of non-villa sites), while tessellated floors and hypocausts are very rare (c. 4 per cent), with most of the pavements occurring in religious sites (all Romano-Celtic temples), then roadside

Romanch3a.indd 54

settlements, and only four being attributed to settlements defined as farmsteads (e.g. Towcester, Northants; Turland 1977). These settlements, along with many of the 15 ‘farmsteads’ with evidence for hypocausts, are usually regarded as ‘possible villas’, but generally lack good evidence. Some of the farmsteads included here, however, have hypocausted buildings within their complexes but no domestic villa range, such as the small tworoom bathhouse/caldarium at Faverdale, Darlington (Proctor 2012), and the building interpreted as a masonry workshop with a heated room for drying at Fosters End Drove, East Winch, Norfolk (Lally and Nicholson 2008). At Chilton Fields in Oxfordshire, a very small (1.91 × 1.62 m internally) heated building has been suggested as a hot bath (Pine and Preston 2015, 23–9), though it has more in common with the caldarium from Faverdale, and another similar structure is known at Ingleby Barwick in County Durham (Willis and Carne 2013). The Chilton Fields example lay over 25 m from what is interpreted as a very modest three-roomed ‘villa’ building (discussed in Ch. 5). Most of the building refinements discussed above are from sites that concentrate in the same geographic regions as villa buildings, all being parts of the same suite of architectural elements that developed within parts of the province from the earliest post-conquest period onwards. Outside this southern and eastern zone such embellishments are very rare, generally only occurring in military vici, roadside settlements and the odd remote villa building. BUILDING FUNCTION Architectural space can be utilised in many different ways, even within a single building (e.g. see Perring 2002, chapters 8–11). Attempts to determine function, both of the building as a whole and individual components, can be extremely problematic, especially when it is noted that many buildings and rooms may have been multi-functional or at least had different uses within their lifespan. Furthermore, any thorough analysis of the use of space in buildings requires the position of doorways to be known, and this is rarely the case in Roman Britain. Functional interpretation in general relies upon the presence of certain features such as recognisable furniture/ embellishments (e.g. hearths, ovens, flooring, bathing pools, etc.), environmental remains and artefactual remains, though other factors such as structural parallels and context also often come into play. Even with these strands of evidence, our understanding of the more detailed use of space within buildings is rarely above the level of vague supposition. Analysis of a Roman aisled building

15/09/2016 15:56:55

BUILDINGS IN THE COUNTRYSIDE

55

a

Barn

Living

0

20 m

N

1:500 b

Port Basin R

R

R

W

R R

R M

R Paddock Yards

R

Shr?

R G? (? ck Tra

East Road

st We

S

Shr

G

)

R?

G W S M R Shr

Granary Warehouse Shop / Storage shed Mill Residence Shrine ‘Official / Public space’

S Field 0

100 m 1:2000

fig. 3.10.  Internal functional areas within Structure 6 at Langdale Hale (a) and functional designation of buildings within the Romano-British port settlement at Camp Ground, Colne Fen (b) (Evans 2013b)

at North Wanborough in Hampshire sought to produce a social reconstruction, dividing the building into male and female areas on the basis of different categories of finds (Hingley 1989, 43–5). However, the choice of artefacts to reflect different

Romanch3a.indd 55

genders was entirely subjective (e.g. male = keys, knives, spears and female = combs, shuttles and spindlewhorls), and little account was taken of the taphonomic processes that led to the distribution of this ‘rubbish’ in the main part of the building.

15/09/2016 15:56:55

56

THE RURAL SETTLEMENT OF ROMAN BRITAIN

In some cases, the levels of preservation, extent of excavation and meticulous post-excavation analysis can provide a reasonable level of confidence in attempting to determine general building function within settlements, as seen in the recently published account of a farmstead and nucleated inland port on the edge of the Cambridgeshire Fens at Colne Fen (Evans 2013b; fig. 3.10). Within the farmstead at Langdale Hale, two timber post-built and beam-slot buildings are interpreted as being residential on the basis of associated finds, one (Structure 17) even tentatively suggested as being connected with inhabitants undertaking an administrative role (on the basis of writing/tally equipment), and the other (Structure 6) incorporating an agricultural barn (ibid., 169–70). The much larger inland port at Camp Ground had 62 identified timber buildings with a variety of building styles (noted above) and with various interpretations as residential houses, granaries, warehouses, mills, shops and a shrine (ibid., 432– 4). Many of these interpretations are, as with Langdale Hale, based upon associated finds’ signatures, while one, a 12 × 28 m beam-slot structure, was suggested as a warehouse purely on the basis of its waterside location and scale (ibid., 440). Even considering the scale and comparative thoroughness of this archaeological investigation, the excavators still admit that some of the functional classification of buildings was quite arbitrary (ibid.). Attempts to determine function within many of the rural buildings in this dataset are hampered by such coarse subjectivity and, indeed, many do not have any particular use assigned to them at all. Nevertheless, there are broad categories of function allocated to some buildings, and these will be examined briefly now. DOMESTIC Assessing the number of buildings used for domestic occupation is problematic for reasons discussed above, though it is likely that a high proportion of the 5772 buildings on 1480 settlements with evidence for domestic occupation would at least partly fall into this category. Dominic Perring’s ‘The Roman House in Britain’ (2002) is probably the most comprehensive account of different forms of domestic architecture, yet aside from fairly limited discussion of roundhouses, is very much orientated towards the houses of the upper echelons of Romano-British society (urban town houses and rural villas), mainly in the south and east of the country. Identifying domestic dwellings further down the social scale is in some ways more problematic, in part due to their more insubstantial nature. Yet these often single-roomed structures, whether circular or rectangular, have been recorded with

Romanch3a.indd 56

increasing frequency within recent developerfunded archaeology, and are discussed below and in the regional analyses of Chapters 4 to 11. Determining a domestic function for such buildings is usually by default, with an assumption that ‘household’ debris in the vicinity is associated, though of course it is highly likely that they were multi-functional spaces for the most part, incorporating a range of craftworking and agricultural aspects (see Taylor 2001, 51–2). Specific evidence for a domestic function within buildings may come through occasional preservation of features like hearths or even painted plaster walls. There are over one hundred explicitly recorded examples of settlements with single-room buildings containing in situ hearths, such as a group of late Roman rectangular stonefounded buildings in the roadside settlement at Navenby, Lincolnshire, which each had hearths, floor surfaces and drains (Palmer-Brown and Rylatt 2011). In general, a higher proportion of circular buildings have been noted to contain internal hearths (e.g. at Higham Ferrers, Northants; Lawrence and Smith 2009, 52–7), but this does not appear to have been due to better overall levels of preservation and instead may result from different arrangements for cooking and/or heating. In rural settlements throughout the late Iron Age and Roman periods, it is more common to find evidence for hearths that were not contained within recognised buildings, potentially because of the truncation of mass-walling or more ephemeral structural features that might have stood around them, but perhaps also because some cooking took place outside the main domestic buildings. At the upper end of the scale, a series of hearths were noted to the west of the north range at Fishbourne Palace, believed to have been part of a kitchen garden and possibly for outdoor dining (Cunliffe 1998; M. Jones 2007). Within more modest contexts, there were at least two late Roman stone-built domestic ovens set into the top of earlier midden deposits in a farmstead at Castle Farm, South Cadbury, Somerset, associated with a cobbled surface, but away from a number of post-built buildings (Leach and Tabor 1996, 11). In addition to the simple, single-room buildings just discussed, there are of course many more complex structures that seem certain to have included domestic components. These buildings, which include those defined as villas, are discussed in more detail below, but here it is worth just reiterating the multi-functional nature of such architecture. A recently excavated and wellpreserved late Roman masonry building at Nesley Farm, Tetbury, Gloucestershire, originally comprised a single-roomed rectangular structure (15 × 7 m), with the later addition of two smaller

15/09/2016 15:56:56

BUILDINGS IN THE COUNTRYSIDE

rooms to the south (A.J. Roberts 2013; see fig. 3.21). The two smaller rooms contained painted plaster walls, and one a potential recessed fireplace, and were clearly residential, while the main room appears to have been divided into agricultural/industrial activity at one end (identified by a corndryer/smoker), and what is suggested as a kitchen at the other, on the basis of a well-defined hearth and a ‘drainage pot’ (ibid., 35). Such a close association of domestic and agricultural/industrial functions within a single architectural framework, whether multi-roomed or not, appears to be relatively commonplace within rural Romano-British settlements. AGRICULTURE Agricultural production lies at the heart of Romano-British rural settlement, both economically and undoubtedly socially. In terms of its architectural signature, it has just been remarked upon that many buildings would have had both domestic and agricultural functions, although there is also a range of structures that appear to be specifically designed for agrarian and/or pastoral needs, including animal shelters, arable processing facilities and storage facilities. The only detailed national study of such buildings remains Morris’s Agricultural Buildings in Roman Britain (1979), which assessed the available evidence, including a gazetteer of known structures. Since then many more rural agricultural buildings have been excavated, and a summary of these now follows. Livestock shelters There are only 27 settlements (mostly farmsteads, but including five villas) where structures have been positively interpreted in excavation reports as livestock shelters, sometimes specifically stated as cattle byres or horse stables. In most ways such interpretation is entirely subjective (e.g. ‘stables’ at Babraham, Cambs.; Armour 2007), since many are designated as potential stock shelters within the parameters of what appear to be agricultural building/enclosure complexes, though some do have more positive evidence to support identifications. At Cefn Graeanog, a small mid- to late Roman enclosed farmstead in Gwynedd, phosphate analysis has suggested that some buildings may have been used as livestock shelters (Fasham et al. 1998), while a fourth century a.d. masonry building at Bradley Hill, Somerton, Somerset, was interpreted as a byre on the basis of potential stall divisions along with a drain running along the west side of the building (Leech 1981, 189). Notwithstanding the many rural structures that could possibly have functioned as livestock shelters (including the general agricultural

Romanch3a.indd 57

57

buildings discussed below), the striking paucity of positively identified stock buildings may suggest that the normal arrangement was to over-winter stock in open hay yards rather than dedicated structures. General agricultural buildings and crop processing Large numbers of buildings have been assigned a general ‘agricultural’ function, mainly because they have minimal numbers of associated domestic objects, although this argument is less convincing in regions further to the north and west where the general levels of material culture decrease significantly (see Chs 9–11). Some of these buildings do have greater evidence for agrarian facilities, usually in the form of internal corndryers or occasionally other features such as threshing floors (e.g. Dalton Parlours and Wattle Syke in West Yorkshire, where buildings contained both; Wrathmell and Nicholson 1990; Martin et al. 2013), although these are rarely observed (see vol. 2 for more details). Out of 357 rural settlements across the country with evidence for corndryers, just 66 (18.5 per cent) had at least one such structure located within a building (e.g. Yewden villa, Bucks: Eyers 2011; Stanwick, Northants: Crosby and Muldowney 2011) (fig. 3.11). Although at some other settlements, such as the farmstead at Pineham North, Northamptonshire (Carlyle 2007), and the villa at Minster in Thanet, Kent (Moody 2010), there are suggestions that ovens had possible shelters built over them, it seems that in most areas, and in most periods, the norm was for corndryers to be freestanding, albeit sometimes explicitly associated with adjacent storage or threshing structures (e.g. Frost Hill (Site 44), Bullock Down, East Sussex: Rudling 1982). Of those settlements where there is evidence for corndryers within buildings, 60 per cent were located within villa complexes (often in aisled buildings; see below, p. 66), and most of these date to the later Roman period (e.g. Yewden), with a concentration in the Central Belt region, particularly to the south-west (fig. 3.11). Not all of these could be classified as ‘dedicated agricultural buildings’, as in half the cases the corndryers were inserted directly into the main villa building or associated bathhouse, often cutting through mosaic floors, such as at Butleigh and Ilchester Mead, in Somerset (Martin and Driscoll 2010; Hayward 1982). This well-observed phenomenon is often taken – quite reasonably – as suggesting a fundamental change in the social and economic trajectories of such villas, though quite why the inhabitants felt the need to insert a structure more often sited externally within the main domestic building remains more of a puzzle.

15/09/2016 15:56:56

58

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Corndryers

Villa house N

Corndryer

0

100 m 1:2000

fig. 3.11.  Distribution of excavated sites with corndryers (those contained within buildings highlighted in red) and plan of Yewden villa, Bucks (Eyers 2011)

Storage Perhaps the most commonly stated use for agricultural buildings within excavation reports is storage, both of arable produce and other elements such as tools and carts. Granaries of some sort must have existed at many rural farmsteads though they are very hard to define, being specifically noted at only 41 settlements, including

Romanch3a.indd 58

13 farmsteads, 17 villas, 7 roadside settlements, 3 military vici and 1 village. In many cases such interpretation is again somewhat arbitrary, though others are more securely based, with evidence such as quantities of carbonised grain found in a fourth-century masonry barn in a villa complex at Great Weldon, Northamptonshire, which was destroyed by fire (Smith et al. 1990). A recent

15/09/2016 15:57:03

BUILDINGS IN THE COUNTRYSIDE a

59

b

Camp Ground, Cambs Lullingstone, Kent c Walls Cut feature 0

20 m 1:500

Great Weldon, Northants

fig. 3.12.  Distribution of excavated sites with four-post structures and buildings interpreted as granaries and plans of selected ‘granaries’ (Evans 2013b; Philp and Mills 1991; Smith et al. 1990)

excavation at Bredon’s Norton in theWorcestershire Avon Valley revealed part of what appeared to be a bathhouse, which was later used as granary. This was suggested by a high concentration of spelt wheat grain preserved on the floor of one room, under a dense layer of wood charcoal, seemingly when the building burnt down in the later fourth/ early fifth century a.d. (grain radiocarbon dated

Romanch3a.indd 59

345–430 cal AD, 95% confidence) (T. Allen et al. 2015). The evidence of carbonised grain preserved in conflagrations in both of these examples highlights the difficulties in interpreting agricultural storage, which may have occurred, even as a secondary function, in any number of buildings, but in most cases would leave no archaeological trace.

15/09/2016 15:57:10

60

THE RURAL SETTLEMENT OF ROMAN BRITAIN

In structural terms, the buildings that have been specifically suggested as granaries are quite varied, including both timber beam-slot and postholedefined structures and substantial masonry barns, with most of the latter being associated with villas (fig. 3.12). A more defined form, which ultimately stems from a military context (Gentry 1976), comprises parallel rows of beam slots or masonry sleeper walls (or piers in the case of some such as Lullingstone in Kent; Meates 1979) which raised the floor up, as seen at settlements like Camp Ground, Colne Fen in Cambridgeshire, Westminster Sports Field, Horton Kirby in Kent and Cleevelands, Bishop’s Cleeve in Gloucestershire (Evans 2013b; Philp and Mills 1991; Joyce 2010a). The distribution of buildings interpreted as granaries, as shown in fig. 3.12, is largely concentrated in the Central Belt region, with another group around London and a broad spread to the north-west, with most of the latter being associated with military vici and roadside settlements. Assuming that such buildings were genuinely utilised for large scale storage of grain, this perhaps points to a general increase and centralisation of arable production in this central region, at least from the second century a.d. onwards, from when most of the buildings are dated, a proposition that will be discussed at length in volume 2. A far more widespread form of structure that is traditionally interpreted as being used for grain storage is the ubiquitous four-post structure, a feature more commonly discussed in Iron Age studies (Reynolds 1974; Cunliffe 2005, 394; Van der Veen 2007, 116). There is generally much less discussion of four-post structures within the Roman period, largely owing to the perception that they rarely occur on Roman sites, though they were noted by Morris, who saw them as representing a delay in the adoption of Roman farming methods (1979, 31). Furthermore, when they are encountered in Romano-British contexts an alternative functional explanation is sometimes sought, such as at Langdale Hale in Cambridgeshire where they are suggested as mills (Evans 2013b, 69). Although strictly outside the architectural remit of this chapter (i.e. they are not included in the general count as ‘buildings’), it is worth noting that they have been recorded on 114 settlements in the current database, of which 80 (70 per cent) potentially were in use into the Roman period, and 36 (32 per cent) definitely so, with some certainly utilised during the third and fourth centuries a.d., such as at Sedgeford in Norfolk and Amesbury in Wiltshire (SHARP 2014; Seager Smith and Fitzpatrick 2000). The continued use of such structures is probably indicative of grain storage (if indeed this was their main function) at a household

Romanch3a.indd 60

level, as opposed to the larger scale, perhaps more ‘commercial’ granaries discussed above. Other agricultural buildings There are of course other types of rural agricultural building, including those defined as mills, smokehouses, threshing sheds etc. In general the archaeological evidence for such buildings is quite meagre, though our evidence for mills is starting to increase to the point where it is now believed that they were a common feature in the RomanoBritish landscape, at least during the later Roman period (Shaffrey 2015). Most of this evidence comes from the millstones themselves (ibid.); there are still relatively few positively identified mill buildings. These are recorded in just seventeen rural sites in the current database, nearly all of which are villas and roadside settlements, the majority suggested as watermills, such as at the villa complex at Woolaston in the Forest of Dean, which had a building with millstones built into the floor and evidence for a double aqueduct or leat traced as an earthwork running up the valley slope (Fulford and Allen 1992, 201). Perhaps the most spectacular example, comprising 3–4 watermills, is from what is thought to be a roadside settlement at Ickham in Kent, probably dating to the later Roman period (Bennett et al. 2010). Typically, the evidence is more subtle, as at the roadside settlement at Scole on the Norfolk/Suffolk border, where a watermill was postulated on the basis of a leat and faint structural remnants (Ashwin and Tester 2014, 195–6). Overall it is believed that watermills were comparatively rare (Peacock 2013, 113; Shaffrey 2015), and on the basis of the incidence of millstones it is likely that many more non-water-powered mill buildings are yet to be identified; indeed, many of the ‘general agricultural buildings’ referred to above could have fulfilled such a role. One probable example of an animalpowered mill was revealed in the roadside settlement at Stanwick, Northamptonshire, where a large (11.5 m diameter) circular masonry building contained the remains of two tread-mills with wear marks caused by hooves (Crosby and Muldowney 2011, 126). INDUSTRIAL WORKSHOPS In addition to agricultural buildings, there is a whole range of structures that have been interpreted as representing workshops associated with craftworking or industrial activities. These have been recorded on 134 settlements across most of the regions, with the majority, unsurprisingly, being found in roadside settlements and specialised industrial sites, where there is also the greatest diversity of associated industry (table 3.2; fig. 3.13).

15/09/2016 15:57:11

BUILDINGS IN THE COUNTRYSIDE

table

3.2: buildings

61

interpreted as workshops: associated site types and industries

Site type Metalworking Pottery production

Tile production

Other

Not specific

Total

Farmstead

21 2

0 0 8 31

Villa

11

0

0 1 15

27

Roadside settlement

19

3

0

30

Vicus

2 0

Industrial

10 14

Other (religious, village, hillfort)

5

Total

0

5

7

4 3 2 33 0

68 19

3

1 1 3 0

1

6

5 8 34 134

N

Kiln Well

0

20 m 1:500

fig. 3.13.  Distribution of excavated sites with buildings interpreted as workshops and plan of late Roman pottery workshop at Stibbington, Cambs (Upex et al. 2008)

Romanch3a.indd 61

15/09/2016 15:57:18

62

THE RURAL SETTLEMENT OF ROMAN BRITAIN

By far the most common associated industry is metalworking (probably because other crafts such as wood and leather working do not generally leave an archaeological trace), with 38 per cent of these buildings specified as being connected with iron production and/or smithing. Those structures interpreted as smithies (or at least having some evidence for smithing) are quite varied, ranging from a circular building defined by gullies with an internal hearth at the nucleated ‘village’ settlement of Gill Mill, Oxfordshire (Booth and Simmonds 2012) to a more substantial rectangular masonry

building with furnaces, slag, hammerscale, a quenching tank and smithing tools found at the roadside settlement at Ashton, Northamptonshire (Hadman and Upex 1975). About all these structures have in common is the presence of hearths/ovens/furnaces and distinct concentrations of smithing slag, occasionally with hammerscale also recorded. Pottery workshops, where noted, generally seem to have been fairly insubstantial timber structures (cf. Swan 1984, 46–8), such as the sub-rectangular hollow and posthole-defined building lying 5 m

N

0

20 m 1:500

fig. 3.14.  Distribution of excavated sites with structures interpreted as religious in nature (shrines and RomanoCeltic temples) and plan of shrine at Rutland Water, Rutland (Carlyle 2011a)

Romanch3a.indd 62

15/09/2016 15:57:26

BUILDINGS IN THE COUNTRYSIDE

north of the kilns at Meole Brace in Shropshire (Evans et al. 1999). However, more substantial buildings did exist, as for example at Stibbington in Cambridgeshire, where a rectangular masonry building of fourth-century date was associated with two external kilns and contained possible emplacements for potters’ wheels (Upex et al. 2008; fig. 3.13), and at Fosters End Drove in Norfolk where there was a masonry building interpreted as a workshop with a heated room for drying (Lally and Nicholson 2008). Another more substantial building was revealed at Alice Holt in Hampshire, where a large timber-framed structure covered a number of early Roman kilns, with additional drying sheds and other related buildings in the vicinity (Lyne 2012). This covering structure was used to suggest that pottery production had been a full-time occupation at that time, as indoor kilns and heated drying sheds prevented problems of ‘winter potting’ (ibid., 136). On the basis of associated finds, context and features, other possible workshops have been associated with tile manufacture, saltworking, boneworking, shale production, brewing, tanning and fulling, and all, as may be expected, vary significantly in form. A large number of buildings have also been provided with a general ‘workshop’ tag without any specific reasoning, this being particularly common on villa complexes. Although there almost certainly were buildings functioning in an industrial or workshop capacity on villa estates, as seen by the number of structures or rooms associated with metalworking, without specific evidence the ‘workshop’ label attached to many of those listed here should perhaps be viewed with some caution.

63

cent located away from any obvious associated settlement (fig. 3.14). Detailed analysis of such constructed sacred space will be presented in volume 3 of this series, but such structures can be broadly divided into the architecturally defined ‘Romano-Celtic’ temple, characterised by an inner central space (cella) surrounded by an ambulatory, and a far more heterogeneous group of buildings/ structures with broad interpretations of a religious nature, defined here as ‘shrines’. These range from small ‘household’ shrines within villa complexes to substantial masonry buildings, though most are quite modest structures, characterised by circular or polygonal architecture in over 50 per cent of cases, such as the circular building from Rutland Water Habitat Creation, Lagoon B, which was associated with articulated animal burials and many finds including a lead curse tablet and a fragment of a figurine of Minerva (Carlyle 2011a; fig. 3.14). There are also many examples of ‘shrines’ that contain minimal or no architectural elements, but are interpreted as religious on the basis of ‘unusual’ characteristics in the finds assemblages, such as at Rothwell Haigh, Leeds, where a well within an enclosure contained complete pottery vessels, a yew bucket, ash drinking vessels, querns, shoes, articulated animal remains, and a human adult skull (Cool and Richardson 2013). Sacred sites defined by enclosures but with no internal shrine/temple building are relatively common, forming over 20 per cent of all 132 religious sites identified from

N

RELIGIOUS ARCHITECTURE Constructed sacred space appears to have been very rare in the pre-Roman Iron Age, being interpreted on only about ten sites across the country (Smith 2001, 67). Furthermore, with the exception of outstanding examples such as the circular timber shrine at Hayling Island, Hampshire (King and Soffe 1998), many of these interpretations are highly subjective and the structures are not all that well dated. Overall, this suggests that religious observance at this time was not spatially segregated to any great degree, or at least was not defined in architectural terms, as seen perhaps reflected in the proliferation of evidence for structured or placed deposits within and around settlement contexts (Chadwick 2012; Smith 2014). This situation changes quite markedly in many areas of the south and east during the early to mid-Roman period when there is a relative proliferation of religious structures, identified at 155 rural sites within the current dataset, 44 per

Romanch3a.indd 63

Inner precinct Truncated area within shrine

Outer precinct Road surface Possible entrance?

0

20 m 1:500

fig. 3.15.  Plan of religious enclosure with monumental entrance at Higham Ferrers, Northants (Lawrence and Smith 2009)

15/09/2016 15:57:26

64

THE RURAL SETTLEMENT OF ROMAN BRITAIN

southern England. However, in some cases there may well still be an architectural element to the site, as at Higham Ferrers in Northamptonshire, where a roadside shrine was defined by drystone wall, but with a monumental mortared stone foundation for an entrance facing across the Nene Valley (Lawrence and Smith 2009; fig. 3.15). Most previous studies of Romano-British religious sites have been almost wholly concerned with temples of Romano-Celtic type (e.g. Lewis 1966; Woodward 1992; Smith 2001), much in the same way that previous analyses of rural settlement have largely concentrated on villa buildings, and indeed to some extent the two share quite similar distributions across the country. However, as with domestic buildings, the mass of mainly developerfunded excavation since 1990 has revealed that Romano-British religious architecture, notwithstanding the huge problems with interpretation, is far more complex and varied than previously supposed. AN ARCHITECTURAL CONTINUUM: BUILDING TYPES Most previous studies of Romano-British rural architecture have focused upon building typology, usually concentrating on villa types, with ‘nonvilla’ buildings often restricted to roundhouses and aisled buildings (e.g. Rivet 1969; Smith 1998;

Cotswold Community, Glos

Broughton, Bucks

N

Eton, Berks

Perring 2002). Hingley (1989, 30–6) took a slightly more wide-ranging approach, including single room and 2–3 room structures while other, recent, studies have emphasised the often complex and blurred boundaries between building types (e.g. King 2004a, 349–50). The current overview of rural architecture has so far concentrated on basic form, material and function, and will now conclude with a broad assessment of this continuum of building types, from the simplest of structures to major courtyard complexes. SINGLE-SPACE ARCHITECTURE The majority of sites in the database (992; 63 per cent) with evidence for architecture contain what appear to be single-roomed buildings only, which account for 3195 structures (fig. 3.16). These are particularly prevalent on farmsteads (85 per cent of farmsteads with architecture), with only slightly higher levels noted on enclosed rather than complex farmstead types. The prevalence of simple undifferentiated structures on farmsteads is partly explained by the fact that over 67 per cent of all buildings on such sites were circular in form, which as stated above, were far less likely to contain identifiable room division than rectangular buildings (although there are notable exceptions). However, even of the 278 farmsteads with only evidence for rectangular buildings, 77 per cent contained solely single-room structures. This may

Langdale Hale, Cambs

Rhiwgoch, Harlech, Gwynedd

Newhouse Park, Chepstow 0

Ashton Keynes, Wilts

20 m 1:500

fig. 3.16.  Plans of ‘single-roomed’ buildings (Allen 2013; Atkins et al. 2014; Evans 2012; Kenney 2012; Powell et al. 2008a; Powell et al. 2010; Robic and Ponsford 2008)

Romanch3a.indd 64

15/09/2016 15:57:27

BUILDINGS IN THE COUNTRYSIDE

in part be attributable to a lack of evidence for building partition, which need leave very little or no trace in the archaeological record, but the high representation probably also genuinely reflects typical vernacular rural architectural traditions across all regions of late Iron Age and Roman Britain. The apparent lack of more complex structures on c. 40 per cent of roadside settlements with architectural evidence is almost certainly in part due to the limited scale of excavation of these settlements: Rust (2006) has demonstrated the complexity and chronological variability of building forms within ‘small towns’, which he attributes to economic and social factors. Nevertheless, it remains likely that simple, single-space buildings (masonry and timber) were very common on such sites, though often interspersed with slightly more elaborate structures, such as at Scole in Suffolk (Ashwin and Tester 2014, 219). The settlement at Shepton Mallet, lying astride the Fosse Way, contained a reasonably wide variety of building types, including an apsidal-ended, rectangular, masonry-footed building (X) with no apparent sub-division, but interpreted as both a barn and

fig.

Romanch3a.indd 65

65

domestic dwelling on the basis of associated finds and a large deposit of charred spelt wheat at one end (Leach with Evans 2001, 309). Cellared buildings Most (76 per cent) single-room buildings were of timber, usually either post-built or of beam-slot construction, excepting the unknown number of possible mass-walled structures that may have been defined by gullies (see above, p. 51). On 40 sites, however, there was evidence for a particular type of (mostly) single-spaced structure, with floor levels below the ground surface. A total of 105 such ‘cellared buildings’ have been recorded, though they are far from a homogeneous group, and are generally distinct from the Anglo-Saxon Grubenhauser, where evidence points to a suspended floor above the sunken cavity. The Roman cellared buildings are found sporadically over parts of south, central and north-east England, with particular concentrations of building numbers in north-west Kent, west Yorkshire and around Dorchester in Dorset (fig. 3.17; see Chs 4 and 7). Where such structures have been attributed a function, it is usually agricultural or occasionally

3.17.  Distribution of excavated sites with Roman cellared buildings

15/09/2016 15:57:34

66

THE RURAL SETTLEMENT OF ROMAN BRITAIN

industrial in nature, as for example in the ‘village’ settlement at Wattle Syke in West Yorkshire, where the fifteen late Roman cellared buildings were suggested as being used predominantly for crop processing, along with animal skinning, drying grain, cooking, antler working, and smithing (Martin et al. 2013). However, a number of the Kent examples did appear to be domestic structures (e.g. at Monkton; Bennett et al. 2008).

AISLED BUILDINGS An architectural tradition that spread across large parts of Roman Britain, and which has been the focus for much academic attention (e.g. Taylor 2001; 2013; Cunliffe 2013b), is the aisled building or hall. These are essentially timber-framed buildings (often with masonry footings) with a series of regularly spaced posts (usually earthfast

N

Shiptonthorpe, Yorks

Dunkirt Barn, Hants 0

20 m 1:500

fig. 3.18.  Distribution of excavated sites with aisled buildings and selected aisled building plans (Cunliffe and Poole 2008f; Millett 2006)

Romanch3a.indd 66

15/09/2016 15:57:42

BUILDINGS IN THE COUNTRYSIDE

or on post-pads) along the main axis to support the roof, though there is still some debate about elements of their reconstruction (e.g. King 1987; Mackreth 1996). They are generally considered to have been multi-functional buildings, incorporating agricultural, industrial and domestic aspects, with most starting off as a single, open space (e.g. at Shiptonthorpe, Yorks; Millett 2006), and later being partially sub-divided into what are usually termed ‘developed’ or ‘modified’ aisled halls (Cunliffe 2013b, 98; e.g. Dunkirt Barn, Abbotts Ann, Hants; Cunliffe and Poole 2008f) (fig. 3.18). Taylor’s study of the internal space of aisled buildings in the East Midlands indicated evidence for cooking and eating located centrally within one end of the mid-Roman open hall, while agricultural/ craft activities were practised in another part (Taylor 2001, 51–2). The addition of wellfurnished domestic rooms and general architectural elaboration in the later Roman period (the ‘developed’ hall) produced what is argued as a hybrid form of architecture expressing a plurality of identity (Taylor 2013, 179). The current project dataset has recorded evidence for 219 aisled buildings on 168 sites, spread across much of the country, though they are fairly sparse in the north and west (fig. 3.18). The huge range in scale of such buildings is immediately apparent, from as small as 50–60 m² at places like Yaxley, Peterborough (Phillips 2014) to huge masonry-footed buildings such as at Rivenhall in Essex, which covered over 800 m² (Rodwell and Rodwell 1993). While some of these variations in scale may be due to functional differences (see analysis in Ch. 4), Cunliffe has argued that size is more likely to be a reflection of social status (2013b, 98), which is also suggested by the notable differences between the size of such buildings on villa and non-villa sites (see below, p. 69). The pre-Roman origin of aisled buildings in Britain has been commented upon previously (Hingley 1989, 39; Perring 2002, 53), yet despite

67

the mass of developer-funded work over the past 25 years, the building at Gorhambury, Herts, remains the only well-dated example (notwithstanding those dating to the early Iron Age: Moore 2003, 49–53), although two timber aisled buildings at Furfield Quarry in Kent could potentially pre-date the conquest (Howell 2014). An increasing number of aisled buildings can now be dated to the later first century a.d., especially around the Thames estuary and in parts of the East Midlands, though the greatest expansion in their numbers took place in the second century (fig. 3.19), albeit with a geographic bias towards the east of the country. The growth continues into the third century when they reach their numerical and geographic maximum, and then there is an overall sharp decline in the fourth century, though with a greater concentration of surviving examples in certain central and western areas. Context of aisled buildings The association of aisled buildings with villa complexes has been well noted, with suggestions that they represented the homes of estate workers (Richmond 1969, 65) or extended families of higher status (Smith 1963, 12; cf. Hingley 1989, 41–5). Just over half of the excavated examples from the current dataset are from villas (87 sites), many suggested to have been used for domestic purposes, often alongside other activities. The relationship of aisled buildings to villas is quite varied, as shown in fig. 3.20. Fourteen sites show clear evidence for one or more aisled buildings being in use prior to the construction of the main villa building, which was typically of corridor or winged corridor form (see below, p. 71. The aisled hall usually formed the main residential house up to this time, and often continued in some form after the main villa building had been constructed (e.g. Brading on the Isle of Wight: Cunliffe 2013a). That there was not always a transition from aisled

180

no. of buildings in use

160 140 120 100 80 60 40 20 0 Late Iron Age fig.

Romanch3a.indd 67

Late 1stC AD

2ndC AD

3rdC AD

4thC AD

3.19.  Number of aisled buildings in use over time

15/09/2016 15:57:42

68

no. of villa sites with aisled buildings

THE RURAL SETTLEMENT OF ROMAN BRITAIN

50 45 40 35 30 25 20 15 10 5 0 Precedes Develops into main development of villa villa building

fig.

Incorporated into single architectural layout

3.20.  Distribution and relationship of excavated aisled buildings associated with villas

hall to villa, however, is shown by the site at Shakenoak Farm in Oxfordshire, where the aisled building actually supplanted an existing winged corridor house that had fallen into decline (Brodribb et al. 2005). More commonly, the aisled hall itself is seen to develop into a building with ‘villa’ attributes, usually by insertion of quite lavish ‘private’ suites of rooms at one end, with mosaic floors, hypocausts and associated bathhouses, which lay in contrast to the ‘public’ open hall of the remainder of the

Romanch3a.indd 68

Added to villa complex

building. Sometimes external facades were even added to make them appear more like corridor houses (Taylor 2013, 179). It is usually in the late Roman period that such architectural transformations of aisled halls occurred, with a particular concentration in the downlands of central southern England, seen at sites such as Meonstoke and Stroud in Hampshire (King 1987; Williams 1909; fig. 3.20). The distinction between such elaborate ‘developed’ aisled halls and those buildings classified as ‘cottage’ or ‘corridor’ villas

15/09/2016 15:57:50

BUILDINGS IN THE COUNTRYSIDE

with similar architectural components, is probably not significant in an economic sense, but instead reflects different individual and wider cultural choices in certain parts of the country (see villa architecture below). In most cases, aisled buildings were seen to be a component of the villa complex rather than the main residential unit (often positioned at rightangles to one side of the main villa house; see below, p.73), and here they are more often argued to have had an agricultural or industrial function, though a mixed use is still quite likely. In some of the major courtyard and ‘palatial’ villas (e.g. Woodchester, Glos) they are actually incorporated into a single architectural whole, though usually sited well away from the primary high-status domestic core. Away from villas, a range of aisled buildings has been revealed at farmsteads (51 sites), roadside settlements (17 sites), villages (5 sites) and military vici (3 sites). Most of these other settlements lay outside of the South region, which had a much higher proportion of its aisled buildings associated with villas (77 per cent, compared with 45 per cent in the Central Belt and 36 per cent elsewhere; see Chs 4 and 5). In general the size of aisled buildings on these ‘non-villa’ sites tends to be much more modest (average of 230 m² rather than 380 m² at villas), corroborating the suggestion by Cunliffe (2013a) noted above that social status was the most important contributory factor to the scale of such structures. Furthermore, at most (82 per cent) of the non-villa sites the aisled buildings were ‘open’ rather than ‘developed’ halls, and relatively few of these had any function specified. However, there are eleven sites with open halls where a residential or mixed use function has been proposed, including Claydon Pike in Gloucestershire, where one of the two aisled buildings contained embellishments in the form of a tiled roof and painted plaster walls (Miles et al. 2007). MULTI-SPACE ARCHITECTURE A greater level of architectural division within buildings is, as discussed above, intrinsically linked with the adoption of rectilinear structures, which allows more flexibility in building size and form (Steadman 2006, 128). Such flexibility is important as architecture became fully incorporated into packages of cultural and individual expression and hierarchy within many parts of Roman Britain, particularly within the upper echelons of society. Such association of multi-space architecture with what may be regarded as higher status rural sites (notably those buildings termed villas) is shown by their comparative rarity on farmsteads, which account for just 96 of the 571 sites where such buildings are recorded. Notwithstanding the

Romanch3a.indd 69

69

inherent problems in identifying room divisions noted above, the apparent lack of such differentiation within most buildings on farmsteads, circular or rectangular, is striking, and is perhaps indicative of a continuation of preexisting ways of dwelling, as Taylor has suggested when comparing use of space within Iron Age roundhouses and open aisled buildings (Taylor 2001, 51–2). Such intra-building spatial analysis lies beyond the scope of this study, and in many cases the levels of truncation and/or recording preclude any detailed comments on building use. The developmental history of most multiroomed buildings in the current dataset has not been recorded in detail, though it is known that many developed from single-space buildings, as for example with the developed aisled halls discussed above. Some may have remained as quite simple structures, with only 2–3 rooms, while others grew more complex, on occasion developing into elaborate villa houses. Such is the constant state of architectural flux in many regions that classification of building ‘types’ within this range is quite problematic, with often at best only the final ‘type’ being recognised. Nevertheless, it is possible within existing frameworks of building classification (e.g. Richmond 1969; Hingley 1989; Smith 1998) to provide a brief outline of such architectural variation on a provincial basis, while more detailed regional accounts are provided in Chapters 4 to 11. Multi-room farmstead buildings There are 96 farmsteads that contain one or more multi-room buildings, in both timber and masonry, but which do not appear to have developed sufficient levels of architectural embellishment (mosaic floors, hypocausts etc.) to ever be classified as ‘villas’. As can be seen from fig. 3.21, the majority (62 per cent) lie within a single region – the Central Belt – and they will be discussed more fully in Chapter 4. Their comparative rarity outside this region is noteworthy, especially their paucity throughout most of the South region (especially the southeast), and it is clear that they do not merely follow the same distribution patterns as the more elaborate multi-room buildings defined as villas. The proliferation of such buildings in the Central Belt appears to be consistent with a trend towards greater architectural variety and dynamism in rural settlement form here, perhaps reflecting in some ways the existence of more complex hierarchies of social status in at least parts of this region. The five farmsteads with multi-roomed buildings from the North region are quite different from most of those further south, and include two where circular stone buildings of second–fourth century a.d. date were internally divided by

15/09/2016 15:57:50

70

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

b

c

N

Nesley Farm, Tetbury

Claydon Pike

Vineyards Farm, Charlton Kings 0

20 m 1:500

fig. 3.21.  Distribution of excavated farmsteads with multi-room buildings and plans of buildings in Gloucestershire at Charlton Kings (Rawes 1991), Claydon Pike (Miles et al. 2007) and Nesley Farm, Tetbury (Roberts 2013)

Romanch3a.indd 70

15/09/2016 15:57:57

BUILDINGS IN THE COUNTRYSIDE

stakeholes (e.g. Huckhoe, Northumberland: Jobey 1959). There were also three farmsteads in Cumbria with rectangular buildings of late Roman date, the rooms being divided by a mixture of fairly flimsy stone and timber foundations (e.g. Old Brampton and Wolsty Hall; Blake 1960). Villa architecture As stated in the introduction to this chapter, with some notable exceptions (e.g. S. Clarke 1998; Taylor 2013), most of the previous work on Romano-British rural architecture has concentrated almost entirely on the analysis of villa buildings. This overview has attempted to shift the focus back towards the architecture of the masses, however insubstantial that may sometimes have been, and within this to emphasise the variety and complexity of building forms across the country. However, although they may only have made up a very small proportion of the total of rural buildings, villa buildings do remain of fundamental importance to our understanding of social and economic life within Roman Britain, themes that will be explored more fully in later chapters and in subsequent volumes. In a purely architectural sense, the variety of buildings that fall under the term ‘villa’ is considerable, both in terms of the building plan itself and in the levels of embellishment and furnishing. Traditionally this heterogeneity has often been explained as representing varied levels of ‘Romanization’ (Richmond 1969), though more recently it is cultural, social and economic factors that are more often advanced (e.g. Smith 1998; Perring 2002, 50–1). In a study of the variability of villa architecture in the east of England, Martins (2005) argued that any explanation of diversity should go beyond broad economic and social reasons, and a ‘consumer’ model is proposed, highlighting individualistic choices made by people within elite society, albeit recognising that

Total (all phases)

71

there would have been tensions between the values of individuals and the collective traditions of prevailing groups. Taking into consideration the individual variation just outlined, most (68 per cent) of the 326 settlements recorded with villa architecture in the current dataset can be placed within one of the broad categories first outlined by Richmond (1969) (figs 3.22 and 3.23). At one end of this spectrum are those developed aisled hall houses discussed above that have evidence for mosaic flooring, heated rooms and bathhouses, along with what have been termed ‘cottage houses’ (Richmond 1969, 52–3; Hingley 1989, 37), where the emphasis is on a range or block of rooms, sometimes with a transverse internal corridor. As with the aisled halls, what elevates such buildings to most archaeological definitions of ‘villa’ status are the associated embellishments (heated rooms, mosaics, bath suites etc), and therefore the distinction here between some masonry ‘multiroom farmstead buildings’ (or even certain ‘undeveloped’ aisled halls) and ‘cottage villas’ is somewhat arbitrary, dependent as it is upon these additional refinements, many of which may just have been a matter of personal preference. Nonetheless, although many of those buildings classified as hall or cottage villas may be relatively modest in size when considered on an empirewide scale (over 60 per cent having less than ten rooms), they are still comparatively exceptional within the countryside of Roman Britain in terms of the investment of wealth in the design and layout, acquisition of building materials, and skilled construction techniques required in their erection. As stated above, there is a blurred architectural continuum, and it must be remembered that even multi-room masonry buildings without any additional refinements would have been distinct from the vast majority of buildings within Romano-British rural settlements.

Total (final form)

80 70

no. of sites

60 50 40 30 20 10 0 cottage/hall house fig.

Romanch3a.indd 71

corridor

winged corridor

courtyard

palatial

3.22.  Types of villa within the dataset by total number and by final form

15/09/2016 15:57:58

72

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Dunkirt Barn, Hants Thurnham, Kent

0

20 m 1:500

N

Brough Field, Derbys

Woodchester, Glos

Llantwit Major, Glam 0

100 m 1:2000

fig. 3.23.  Plans of selected villa types (Cunliffe and Poole 2008f; Ling et al. 1990; Lawrence 2006; Nash-Williams 1953; O’Neil 1955)

Romanch3a.indd 72

15/09/2016 15:57:59

BUILDINGS IN THE COUNTRYSIDE

73

60

no. of villas in use

50 40

cottage/hall house corridor

30

winged corridor courtyard

20

palatial

10 0 1st C AD fig.

2nd C AD

4th C AD

3.24.  Chronology of villa types

As discussed in Chapter 2, at least half of all villas developed from pre-existing settlements, with about 40 per cent of later first-century villas having late Iron Age origins. It has sometimes been stated that such early villas were built on a large scale and generally became smaller through time (e.g. Millett 2014, 4), though, with the exception of the early villas in Sussex (notably the huge complex at Fishbourne) and a few in north Kent (e.g. Eccles), the vast majority of these first-century buildings were of the relatively modest ‘cottage’ type (fig. 3.24). Moreover, a substantial proportion of these went on to develop into larger, more elaborate buildings over time, mostly corridor and winged corridor houses, which form the most distinctive, most numerous and most widespread types of villa across Britain (fig. 3.25). Such buildings are defined by a range of rooms opening on to a corridor, typically with one or more of them containing mosaic floors and underfloor heating, with either an attached or a separate bathhouse, and some could be remarkably similar in plan. The addition of wings at either end added symmetry to the frontal aspect of the villa, and such winged corridor buildings became the most common form of elite rural housing from the late second century onwards, contemporary with similar developments in northern Gaul (Taylor 2011, 181). Most of these types of villa have between 10 and 20 rooms, though as ever there is huge variety, with, for example, one small villa at Yarford, Kingston St Mary in Somerset initially comprising just four rooms linked by a corridor (King 2004b), while another villa at The Mount, Maidstone, Kent contained over 28 rooms, most set between two opposing corridors (Houliston 1999). At the upper end of the scale of villa architecture were those defined by ranges of rooms set around a defined courtyard, often with a gated entrance. Such courtyard villas generally developed from earlier corridor villas, with new ranges and outbuildings added piecemeal over time, including the quite persistent inclusion of aisled buildings arranged at right-angles to one side of the main villa

Romanch3a.indd 73

3rd C AD

house (e.g. Bignor, Sussex; see Ch. 4, fig. 4.44a). In some cases there were more distinct episodes of major reconstruction, such as at Bucknowle in Dorset, where significant modifications were made to create a more unified building complex at the start of the fourth century a.d. (Light and Ellis 2009). Courtyard villas reached full development during the later third and fourth centuries a.d. and have a very distinctive distribution, for the most part being concentrated in parts of Gloucestershire, Wiltshire, Somerset and Dorset (fig. 3.25). The scale of architectural investment in such settlements was considerable, and although, as Millett states, ‘the buildings themselves provide little or no information about how the wealth deployed to construct them was created’ (2014, 4), it is likely, based upon the relatively high levels of ancillary buildings, corndryers and agrarian tools (see regional chapters and vol. 2), that at least part of the wealth used to maintain these complexes was generated by agricultural production. The uppermost end of the villa architectural scale is represented by huge, usually multicourtyard building complexes with at least 40 rooms (over 70 in the case of Darenth in Kent) that must surely have served as properties of the highest levels of elite society, in Britain or other parts of the empire. With the exception of Fishbourne, which is a complete architectural anomaly in its British chronological context, all of these ‘palatial’ type villas date to the later Roman period, most belonging to the fourth century a.d., and all developed from earlier more modest complexes. They tend to cluster in the same areas as courtyard villas, particularly in the Cotswolds area (e.g. Woodchester and Turkdean), and in the Lower Nene Valley around the walled town of Durobrivae (Cotterstock and Castor; see Ch. 5). Sarah Scott’s analysis of such ‘super-elite’ villas indicated that the British examples were part of a wider phenomenon across many parts of the Roman Empire, most dating to the period c. a.d. 320–380, a time of containment and stability along the empire’s frontiers (Scott 2004, 40).

15/09/2016 15:58:00

74

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

3.25.  Distribution of different types of excavated villas

Specific historical evidence indirectly suggests ownership of British villas by the super-rich from other parts of the empire, yet the growth of such elaborate complexes does not appear to have been initiated by developments on the continent, as many Roman villas in Britain began their expansion somewhat earlier (late third/early fourth century; ibid., 42, 47). The ultimate fate of most of these and other villa buildings in the fifth century is still rather ill understood, but by the later fourth century many had clearly started to shift in function away from an emphasis purely on elite social display towards more utilitarian aspects, as demonstrated by the twenty examples of corndryers inserted into the main villa buildings, often through mosaic flooring (see Agricultural buildings above). Of the 214 villas thought to have been in active use of some kind in the late fourth century, only 46 (21 per cent) are explicitly stated as having some form of continued use into the fifth century (assessed in the relevant regional chapters), although this is partly due to inevitable problems of dating in this period. Most buildings probably gradually decayed, sometimes over considerable periods of time, as the effects of rural economic deterioration, or as Gerrard argues (2013, 13), the agricultural ‘re-alignment’, took full effect.

Romanch3a.indd 74

FROM NATIONAL OVERVIEWS TO REGIONAL SYNTHESES Chapters 2 and 3 have provided broad overviews using data derived from England and Wales for two of the major elements of the Romano-British countryside to be discussed in this volume – the settlements themselves and the buildings contained therein. Throughout both the aim has been to attempt to redress the balance of study away from the elite villas and back towards the experiences of the greater mass of population, a redress which has only been made possible through utilisation of the wealth of modern archaeological excavation reports, both published and ‘grey’ literature. It has been appreciated that there is considerable diversity in settlement type and form as well as in architectural expression within different parts of the Roman province, and this is analogous with variations in the wider landscape infrastructure along with farming practices and in the use of material culture. The following Chapters 4 to 11 will explore this regional diversity, looking at chronological developments within all of these elements of rural life, and examining more subtle patterns within different intra-regional landscape zones.

15/09/2016 15:58:07

Chapter 4

THE SOUTH By Martyn Allen

After the Central Belt, the South is the second largest region, covering just over 24,254 km². It encompasses the modern administrative counties of Kent, Greater London, Surrey, East Sussex, West Sussex, Hampshire, the Isle of Wight, and Dorset, while it also encroaches on parts of Essex, Hertfordshire, Buckinghamshire, Berkshire, Wiltshire, Somerset and Devon (fig. 4.1). The nature of the landscape The region is characterised by four main geographical units: the chalk downs, the Weald, the Hampshire Basin, and the London Basin, which can be broken down further into smaller landscape zones, based upon Natural England

fig.

character areas (fig. 4.2). The bulk of the chalk downs rises across much of southern Wiltshire and northern Hampshire, stretching south-west to the Dorset coast, and to the east to the North and South Downs, which together encircle the Weald. The chalk downs comprise rolling hills, capped with shallow but mildly alkaline soils which can be exploited for both arable and pasture. The downs are also drained by numerous river valleys containing highly fertile, alluvial soils and are particularly suited to arable agriculture. Many of the chalkland river valleys run through the Hampshire Basin, to drain into the Solent to the south, and through the London Basin to the north, to drain into the Thames and the Thames Estuary. These estuarine regions constitute a

4.1.  The South region in relation to modern county boundaries

75

Romanch4.indd 75

19/09/2016 12:11:34

76

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

4.2.   Constituent landscape zones of the South region

mixture of geologies, essentially a variety of clay and sand ‘brickearth’ deposits that tend to be more acidic than the chalkland soils, but can be quite fertile in places. Although numerous rivers drain into the south coast, the London Basin is dominated by the Thames and its tributaries. The Thames enters the South region as it runs south through the Chilterns and into a wide floodplain extending across much of modern Berkshire and Greater London. Though this landscape is more acidic than the chalkland and is more susceptible to seasonal influences, such as winter flooding, it still provides suitable land for both arable and pastoral farming. The Weald, however, is much less accommodating of intensive arable agriculture, consisting of an inner High Weald, predominantly of sandstone, and an outer Low Weald, formed of heavy clay, both of which are separated from the chalkland by a ridge of Greensand. The Weald is largely wooded today, with its name thought to derive from the Old English for ‘forest’ (Onions 1966). Just how far it was covered by wood in the Roman period is uncertain, but a recent survey of the pollen evidence suggests that it consisted of a mixture of woodland and grassland pastures (Rippon et al. 2015). At the

Romanch4.indd 76

south-eastern periphery of the Weald, the land runs down to the wetlands of Romney Marsh. Much of the latter has been gradually reclaimed during and since the medieval period, and it is uncertain how much was exploited and settled during the Roman period (Rippon 2002). The South dataset The South region is populated by 897 individual site records, comprising almost one quarter of the total dataset. The sites are spread widely, if unevenly, across the region with notable concentrations occurring north and south of the head of the Thames estuary, along the Sussex Coastal Plain, and in the hinterlands of some of the Roman towns in the region, for example, around Ilchester, Dorchester and Silchester (fig. 4.3, top). It is important to note here that, while the majority of sites have been excavated since 1990 under development-led conditions, the dataset also includes the results of twentiethcentury antiquarian fieldwork (e.g. Wolseley et al. 1927; Curwen 1933; Parsons and Curwen 1933; Holleyman 1936), university-based research (e.g.

19/09/2016 12:11:42

THE SOUTH

77

fig. 4.3.  Kernel density of South region records (n=897) and all excavation records (1910–2010) from National Monuments Records (NMR) index (n=17,582)

Romanch4.indd 77

19/09/2016 12:11:54

78

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Cunliffe 2000; Cunliffe and Poole 2008a), and ‘rescue archaeology’ undertaken in the later 1970s and 1980s (e.g. Fasham 1980; 1983; 1987; Philp 1973; 1984; Smith et al. 1998). Another aspect to consider is the expansion in the scale of some of the work that has been undertaken in the South, particularly since 1990. While most of the sites in the dataset stem from relatively small-scale, development-led work, an important number derive from major infrastructural projects. Significantly, these have moved the emphasis away from individual, sitebased investigations towards multi-period, landscape-scale excavations. While, individually, such projects provide a better contextualisation of Roman rural settlements, collectively, their scale is now beginning to influence the overall distribution pattern of settlements within the region. Projects such as High Speed One (HS1; Booth et al. 2011), Heathrow Terminal 5 (Lewis et al. 2010), the Dorchester by-pass (Smith et al. 1998) and the East Kent Access Road (Andrews et al. 2015) have expanded the evidence base in areas where very little was known before, such as on the eastern stretch of the Wealden Greensand ridge along which HS1 now runs. Despite the increased number of sites generated by developer-funded archaeology, gaps in our knowledge of the Roman rural settlement pattern remain. Comparison of our site distribution with the National Monuments Records (NMR) database shows that the two datasets broadly correlate (fig. 4.3). Nonetheless, the area of greatest variation that stands out is within Greater London. It is clear from the NMR data that Greater London has received a substantial number of excavations during the twentieth and early twenty-first centuries, primarily due to the rapid rate of urban expansion. Yet, in the hinterland of the provincial capital of Londinium, Roman rural settlement sites are notably scarce, particularly in the areas away from the major Roman roads. A number of sites have been identified more recently around the Greater London/Essex border, largely through extensive gravel quarrying (e.g. Howell et al. 2011; Lyons 2012), but otherwise the roadside settlements, such as those at Brentford and Old Ford, provide much of our evidence for rural settlement around Londinium. The relative absence of rural settlement evidence in these low-lying areas appears to be real, further exemplified by the recent publication of the Olympic Park excavations around the Lea Valley where, despite large areas being opened up for investigation, almost no evidence for Roman farming settlement was encountered (Powell 2012). The large expanse of the Weald is another area with few settlement sites. Consisting of a

Romanch4.indd 78

comparatively infertile terrain, the Weald is known mostly for its early Roman iron production sites, as well as for some sites involved in quarrying and tile manufacture, all demonstrating the importance of the natural resources available in the region and the suitability of its environment to support such industries (Cleere and Crossley 1985; Hodgkinson 2008; see vol. 2). There is emerging evidence that arable expansion onto the Low Weald took place in the later Roman period; a watching brief at Eastlands Farm near Burgess Hill revealed the remains of a T-shaped corndryer dating to the later fourth century a.d. (Sawyer 1999), but such sites are currently rare. It is possible that much of the Wealden population in the Roman period was more commonly engaged in industry than farming, though this would not explain how food production and procurement was managed in the Weald. It may be that settlements whose economic basis perhaps involved a mixture of small-scale pastoralism, woodland management and hunting are simply not as archaeologically visible as the known industrial sites in the area. It remains to be seen whether future survey and excavation will uncover further evidence of farming settlements. Roman Rural Settlement patterns It is important to recognise that the rural settlement of Roman Britain did not exist in isolation from the provincial infrastructure of towns, military sites and roads, much of which became established in the second half of the first century a.d. The South region contained a number of major urban centres, most notably the provincial capital at Londinium (London), five civitas capitals, Calleva Atrebatum (Silchester), Durovernum Cantiacor um (Canterbury), Dur novaria (Dorchester), Venta Belgarum (Winchester), Noviomagus Reginorum (Chichester), and the colonia at Camulodunum (Colchester). Other major settlements included the ports at Clausentum (Bitterne), Rutupiae (Richborough) and Portus Dubris (Dover), the latter two of which became forts of the Saxon Shore, and the walled towns at Durobrivae (Rochester) and Lindinis (Ilchester). Alongside Rutupiae and Portus Dubris, Saxon Shore Forts were also erected in the third century a.d. at Regulbium (Reculver), Portus Lemanis (Lympne), Anderitum (Pevensey Castle), and Portus Adurni (Portchester Castle). The precise role which the major towns and ports played in administering their rural hinterlands and the scale of their economic impact upon them is much debated. However, while it might be assumed that the presence of urban and military settlements would have significantly impacted upon rural populations, the degree to which this occurred

19/09/2016 12:11:54

THE SOUTH

79

fig. 4.4.   Excavated late Iron Age/Roman rural settlement in the South region in relation to Roman roads and urban centres 100% 90% 80%

% of sites

70%

hillfort

60%

village

50%

roadside selement

40% 30%

villa

20%

farmstead (all)

10% 0%

fig.

4.5.   Relative frequency of main settlement types by landscape zones (with >20 sites) in the South region

probably varied across different regions of the country (cf. Taylor 2013). The 897 rural site records from the South region relate to 627 individual settlements (fig. 4.4; table 4.1). Farmsteads and villas are counted as single site records, while nucleated settlements may include a number of site records reflecting the greater extent of some of these settlements. The

Romanch4.indd 79

remainder include industrial sites, field systems, isolated buildings and sites of a religious nature, where no evidence for domestic settlement was encountered during their excavation. Farmsteads dominate the dataset (70.6 per cent), followed by villas, which are represented with a greater frequency in the South than in other regions (18.0 per cent). Nucleated settlement is comparatively

19/09/2016 12:12:02

Romanch4.indd 80

  5   1

  9   6   2   3  36 113  43   9   1  14   2  30  45  23  28  51 443

Isle of Wight

Isles of Portland and Purbeck

New Forest

South Coast Plain and Hampshire Lowlands

London Basin

North Kent Plain

Greater Thames Estuary

High Weald

Low Weald

Romney Marsh

Wealden Greensand

Hampshire Downs

North Downs

South Downs

South Wessex Downs

Total

  8   2

 4  1

  26

 5

(111)

  13

 2 (43)

  7

 1

  6

  7

 1

  4

  30

17

 2

  7

 2

 2

(19)

 1

 1

10

 4

 1

 2

113

 13

  9

 10

  7

 17

  3

  3

 10

  8

 10

  1

 13

  3

Villa

Dorset Heaths

 6

Farmstead (open)

 24

Farmstead (enclosed)   1

Wessex Vales

Blackdowns

Farmstead (complex)

34

 3

 3

 6

 6

13

 1

Roadside settlement  2

21

10

 1

 1

 2

 2

 1

 2

 2

Village

2

1

1

Vicus

of major settlement types by landscape zone in the south region

Farmstead (all)   4

4.1: number

Landscape zone

table

14

 5

 1

 1

 1

 1

 2

 2

 1

Hillfort

627

 82

 38

 34

 56

 48

  2

 24

  2

 13

 61

138

 46

  3

  4

 11

 12

 42

 11

Total

13.1

6.1

5.4

8.9

7.7

0.3

3.8

0.3

2.1

9.7

22.0

7.3

0.5

0.6

1.8

1.9

6.7

1.8

   %

80 THE RURAL SETTLEMENT OF ROMAN BRITAIN

19/09/2016 12:12:03

THE SOUTH

well represented in the region, predominantly in the form of roadside settlements which are relatively common around London and along Watling Street, stretching eastward across the North Kent Plain to Canterbury and Richborough at the coast. Roadside settlements are also recorded in the Low Weald, particularly on its southern and western margins, areas where farmsteads are not well attested. Village settlement is most notable in the western part of the South region, occurring in particular on the South Wessex Downs (Salisbury Plain), mostly owing to the preferential survival of identifiable earthworks (McOmish et al. 2002; Fulford et al. 2006). Variation in the frequency of different settlement types across different landscape zones is illustrated in fig. 4.5.This shows that considerable differences occur in the distribution of those sites that developed villa architecture, for example. The North Downs, the Wealden Greensand and the Wessex Vales each contain relatively high numbers of villa sites, though they are very poorly represented in the London Basin. Considering the extent of gravel extraction that has occurred on the lower Thames terraces, the general absence of villa settlement appears to reflect a real difference between this area and the Upper Thames Valley where this settlement type is more common (cf. Booth et al. 2007, 42–79). Of course, one aspect of the settlement archaeology which is masked by taking these data at face value is the change that occurred through time across the region, and to understand the intra-regional variation more clearly we must first assess the chronological data. Regional chronologY Along with coins, the well-studied, regional pottery industries in the South (e.g. Alice Holt wares, New Forest wares, Dorset Black-Burnished wares (BB1), etc.), and the comparatively widespread consumption of early Roman continental imports,

81

such as samian, has enabled a relatively high resolution in dating a large proportion of rural settlements. Combining these data from all sites in the South facilitates the construction of a broad chronology which illustrates the relative proportions of settlements that were inhabited from the late Iron Age through to the end of the fourth century a.d. (fig. 4.6). The data show an increase by about 25 per cent in the number of sites occupied from the late Iron Age into the later first century a.d. The period from the later first century a.d. to the mid-second century then sees a peak in settlements, after which numbers gradually reduce to the later fourth century when the figure reverts to a similar number to that recorded for the late Iron Age. The late first/second century a.d. peak in rural settlements in the South is similar to that in the East region, but is quite different to that seen across the Central Belt where the high point is not reached until the late second/third century a.d. (see Chs 5 and 6). However, observing these chronological trends across such large and diverse landscapes undoubtedly masks a considerable degree of inter-regional variation, as evidenced by the synthesis of dating patterns across different landscape zones in the South. These demonstrate a distinct geographical patterning, broadly on an east–west alignment (fig. 4.7). Settlements in the eastern part of the region – the North Kent Plain, the North Downs, the Wealden Greensand, and the Low Weald – all show increasing settlement numbers from the late Iron Age through to the first half of the second century a.d., after which numbers decline steadily through to the second half of the fourth century a.d. Remarkably, the number of settlements of later fourth-century a.d. date recorded in the Low Weald and on the Wealden Greensand is less than half that known to have been inhabited during the late Iron Age. An increasing frequency of settlement also

500

no. of selements

400 300 200 100 0

fig. 4.6.   Number of excavated settlements in use over time in the South region (NB: the M–LIA data only include sites that continued into the late Iron Age)

Romanch4.indd 81

19/09/2016 12:12:04

82

THE RURAL SETTLEMENT OF ROMAN BRITAIN

80 60 40

South Wessex Downs (n=67)

20

Wessex Vales (n=37)

% se lements in use

100 80 60 40

North Kent Plain (n=57)

20

North Downs (n=29)

0

0

100

100

80 60

South Downs (n=33)

40 20 0

South Coast Plain and Hampshire Lowlands (n=43)

% se lements in use

% se lements in use

% se lements in use

100

80 60 40

Wealden Greensand (n=37)

20

Low Weald (n=23)

0

% se lements in use

100 80 60 40 20

London Basin (n=135) Hampshire Downs (n=50)

0

fig.

4.7.  Variations in settlement chronology by selected landscape zone within the South region

characterises the London Basin, the South Downs, the South Coast Plain and the Hampshire Downs from the late Iron Age into the late first century a.d. However, the rate of decrease thereafter is generally less marked than in the Weald and in North Kent. It is possible that the reduction in settlement numbers around the periphery of the Weald was related to the decline in the Wealden iron industry, thought to have occurred from towards the middle of the third century a.d. (Hodgkinson 2008). Two of the most important iron-producing sites identified at Bardown and Beauport Park are thought to have ceased production between a.d. 220 and a.d. 240 (Cleere and Crossley 1985, 84–5), events that occurred soon after abandonment of the Classis Britannica fort at Dover (Philp 1981, 94–7). Whether these changes were directly related is uncertain, and it is difficult to gauge how they may have impacted upon settlement numbers. It is possible that the declining number of farmsteads can be explained by an increase in settlement nucleation into the later Roman period. However, this is also difficult to substantiate from the available evidence. Two of the more recent and more extensively excavated roadside settlements in Kent at Westhawk Farm

Romanch4.indd 82

near Ashford and Springhead near Southfleet have demonstrated evidence for considerable settlement decline from the early or middle part of the third century a.d. A substantial contraction in both settlements occurred at this time, as evidenced by a reduction in pottery usage and coin loss, as well as building demolition, with almost complete abandonment by the mid-fourth century a.d. (Booth et al. 2008, 394; Andrews et al. 2011, 208–9). Of course, the situations in London and Canterbury at this time would have had a considerable effect on the amount of traffic on the road between the two towns, and the settlements that were located on it. That Watling Street was not the bustling thoroughfare it was in the early Roman period is also indicated at Ospringe where a large mixed-rite cemetery appears to have been used almost exclusively between the second and third century a.d. (Whiting et al. 1931). The dating of the cemetery suggests that its associated settlement may have been in decline or even abandoned by the beginning of the fourth century a.d. Overall, the data show that a pattern of settlement decline into the fourth century a.d. occurred across much of the South region, though further west the rate of settlement abandonment appears to have been far less marked.

19/09/2016 12:12:06

THE SOUTH

There, in the South Wessex Downs and the Wessex Vales, the peak in settlement numbers does not occur until the late third/early fourth century a.d., at least 150 years later than in all the landscape zones to the east. In fact, the settlement chronologies of the Wessex sub-regions are more closely aligned to those of areas immediately to their north in the Central Belt, such as the Cotswolds (see Ch. 5). This pattern exists despite the fact that the South Wessex Downs form part of the same chalkland zone as the Hampshire, North and South Downs. The variation in dating across the chalkland suggests that the dynamics of settlement establishment and abandonment were perhaps more influenced by social or political factors than by geological and environmental constraints. Settlement continuity and abandonment Taking account of settlements that were established earlier in the Iron Age, fig. 4.8 shows that around 65 per cent of sites occupied in the late Iron Age (c. first century b.c.–early first century a.d.) were founded during that period. This figure is substantially higher than the frequency of settlement establishments seen in the centuries that followed, the rate reducing consistently into the fourth century a.d., when only c. 3 per cent of the sites occupied in that phase were also founded then. The proportion of sites (32.5 per cent) that were newly set up in the late first century a.d. appears relatively high, but, rather than being a direct consequence of the Roman conquest, this might perhaps be seen as the continuation of a longer-term pattern of settlement expansion that was already underway during the late Iron Age. This possibility has been previously suggested for parts of the Sussex Coastal Plain and Kent (Davenport 2003; Hill 2007, 24). Not only did the period from the first century b.c. see a high frequency of new settlements

83

being established, but there was also a high degree of settlement continuity through to the second century a.d. with a correspondingly low level of settlement abandonment. As already demonstrated by the chronologies of all settlements in the South, there exists considerable intra-regional variation, particularly in the establishment of new farmsteads in the late Iron Age, which ranges between c. 85 per cent in the Weald to only c. 35 per cent on the Hampshire Downs (fig. 4.9). The ratio of settlement establishments and abandonments in each phase on the Hampshire Downs and the South Wessex Downs appears to have been more balanced compared with landscapes zones to their east, which display comparatively high numbers of foundations in the late Iron Age and the late first century a.d., but more frequent abandonments in the second and third century a.d. The very high incidence of settlement abandonment at the end of the fourth century a.d. appears to signify a major change in the settlement pattern, but it is no doubt partly a consequence of the invisibility of conclusive fifth-century a.d. dating evidence on many late Roman sites. Certainly, the loss of the major pottery industries in southern Britain with their datable wares, in parallel with the dramatic reduction in coin loss by the beginning of the fifth century a.d. has influenced these data to an unknown degree. Through the use of radiocarbon-dating and the identification of post-Roman material culture, 29 farmstead/villa sites show continued occupation beyond the fourth century a.d. and, of these, 15 provided evidence for activity that might be identified as, culturally, early ‘Anglo-Saxon’. At Laleham, Spelthorne, an upsurge in activity was noted by increasing quantities of material into the late Roman period, which the excavator related to renewed economic prosperity at Staines (Taylor-

% of sites 'established new'

% of sites 'abandoned'

100

% of sites

80 60 40 20 0 Late Iron Age (n=307) fig.

Romanch4.indd 83

Late 1stC AD (n=413)

2ndC AD (n=444)

3rdC AD (n=393)

4thC AD (n=321)

4.8.   ‘New’ and ‘abandoned’ settlements over time in the South region

19/09/2016 12:12:07

84

THE RURAL SETTLEMENT OF ROMAN BRITAIN

100

Low Weald and Greensand Ridge

% of sites

80 60 40 20 0 Late Iron Age Late 1stC AD (n=35) (n=47) 100

2ndC AD (n=50)

3rdC AD (n=38)

North Downs and Coastal Plain

100

60 40 20 2ndC AD (n=77)

3rdC AD (n=61)

Late Iron Age Late 1stC AD (n=32) (n=38)

London Basin

100

2ndC AD (n=38)

3rdC AD (n=33)

South Wessex Downs

80 % of sites

% of sites

80 60 40 20

60 40 20

0

0 Late Iron Age Late 1stC AD (n=81) (n=105)

2ndC AD (n=107)

3rdC AD (n=93)

Late Iron Age Late 1stC AD (n=35) (n=43)

2ndC AD (n=50)

3rdC AD (n=56)

4.9.   ‘New’ and ‘abandoned’ settlements over time in selected landscape zones of the South region

Wilson 2002). Early Saxon pottery was also discovered, but without any obvious break in occupation or change in site organisation. Indeed, two features contained both late Roman and early Saxon pottery, indicating that both types were being used at the same time. In most other cases, the evidence for postRoman continuity on farmsteads and villa sites is far more scanty. Better evidence for continuity is occasionally observed on religious or funerary sites, such as at Tolpuddle Ball, Dorset, where radiocarbon dating of several inhumations showed that the cemetery was in use from the fourth century a.d. to the seventh century (Hearne and Birbeck 1999). The relationship between this cemetery and the nearby Roman farmstead is difficult to explain, considering the latter appeared to be abandoned at the end of the fourth century a.d. Farmsteads: MORPHOLOGY, CHRONOLOGY AND DISTRIBUTION The settlement patterns noted across the region are complicated by the diverse range of farmsteads that developed in different parts of the region. Just under 40 per cent of farmsteads in the South region have been classified according to the morphological criteria of open, enclosed and complex, as set out in Chapter 2. It must be noted here that some of the classified farmsteads counted in table 4.1 may relate to the same sites. For example, if a farmstead began as an open settlement and then later developed into an

Romanch4.indd 84

40 0

Late Iron Age Late 1stC AD (n=46) (n=70)

fig.

60 20

0

100

Hampshire Downs

80 % of sites

% of sites

80

enclosed settlement, both phases of the site are counted. Enclosed farmsteads are the most frequently identified type in the South, dominating the sample with 64 per cent of sites, with lesser numbers of open farmsteads, at 11 per cent, and complex farmsteads, at 25 per cent (fig. 4.10). Analysis of the relative frequencies of the three main forms of farmstead over time shows that the relative proportions of enclosed and complex types changed markedly across the Roman period, while open farmsteads were poorly represented throughout. The general trend sees a reduction in the relative frequency of enclosed farmsteads from the late Iron Age to the fourth century a.d., coupled with an almost inverse increase in the frequency of the complex farmsteads over the same period (fig. 4.11). Enclosed farmsteads are overwhelmingly the dominant farmstead type in the late Iron Age, but by the fourth century a.d. enclosed farmsteads and complex farmsteads are more equal in number. Open farmsteads are primarily recognised in the late Iron Age, constituting between 12.5 per cent and 15 per cent of the identified settlement prior to the Roman conquest, but by the late first century a.d. their relative frequency falls below 5 per cent and they remain very poorly represented compared with other types of farmstead thereafter. Together, these chronological developments are illustrated in the distribution maps in fig. 4.12, which show the gradually evolving settlement landscape of the South region.

19/09/2016 12:12:09

THE SOUTH

Enclosed

no. of farmstead sites idenfied

30

Complex

85

Open

25 20 15 10 5 0

4.10.   Number of open, enclosed and complex farmsteads in selected landscape zones of the South region

% of farmsteads identified by morphology

fig.

fig.

80

open farmstead (n=19)

enclosed farmstead (n=112)

60

40

20

0

4.11.   Relative frequency of farmstead types in use over time in the South region

While enclosed farmsteads reduce in number over time in most areas, they appear to have been consistently present across the chalk downland, particularly on the Hampshire and South Wessex Downs, but also towards the head of the Thames estuary and on some of the gravel terraces of the Middle Thames (fig. 4.13). In the eastern part of the region, the predominance of enclosed farmsteads on the North and South Downs also extends onto the Wealden Greensand ridge and to a lesser extent on to the Low Weald. A few settlements classified as complex farmsteads are located in the region, such as the site at Mackie Avenue, near Hassocks, but generally these are fairly undeveloped compared with those seen on the Thames gravel terraces (Mullin et al. 2010). Complex farmsteads and open farmsteads, in contrast, are rarely found on the chalk downs, but instead are far better represented on the Thames

Romanch4.indd 85

complex farmstead (n=43)

gravel terraces, being prevalent around the major river systems of the Thames, the Colne, and the Kennet. By the third century a.d., complex farmsteads became the dominant form of farmstead in the London Basin, with a proliferation of large settlement complexes, such as those found at Hengrove Farm near Staines (Poulton 2007) and the Imperial College Sports Ground site near Harlington (Crockett and Nowell 1998). These types of site are frequently located on lowlying ground with enclosure systems integrated by trackways with external field systems (e.g. Johnstone and Bowden 1985; Wallis and Waughman 1998; Biddulph et al. 2007; fig. 4.14). By contrast, the few complex farmsteads that are located on the chalk downs tend to be smaller settlements, perhaps with a particular focus upon livestock management, such as at Winnall Down/Easton Lane near Winchester

19/09/2016 12:12:11

86

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 4.12.  Distribution of farmstead types in use over time in the South region (top left: late Iron Age; top right: late 1st and 2nd century a.d.; bottom left: 3rd century a.d.; bottom right: 4th century a.d.)

(Fasham 1985; Fasham et al. 1989) or Rowbury Farm, Wherwell (Cunliffe and Poole 2008e) (fig. 4.15). However, a few larger complex farmsteads do occur on the downs, such as at Owslebury (Collis 1970) and at Dunkirt Barn (Cunliffe and Poole 2008f) (fig. 4.16). A different pattern is beginning to emerge to the west of the chalk downland. Only six farmsteads identified to morphological class are located on the mixed sandstone and limestone formations of the Wessex Vales, but these are exclusively sites of complex farmsteads. Enclosed farmsteads are apparently completely absent from this landscape zone, while there is evidence that the sites at Lyde Road, Yeovil (Clelland 2011), and Royal Naval Air

Romanch4.indd 86

Station, Yeovilton (Lovell 2005) developed from earlier open farmsteads from the mid–late Iron Age and continued through the Roman period as larger complex settlements more akin to those found on the Thames gravels. The virtual absence of open farmsteads in the Roman period is difficult to interpret. It is possible that this pattern is truly representative and that, as a form of settlement, they become largely extinct after the Iron Age. But the issue may also be one of visibility. Open settlements of the mid–late Iron Age in the South generally present a commonly recognised form, usually consisting of a number of associated roundhouses and adjacent pit groups with no obvious sign of surrounding boundary

19/09/2016 12:12:40

THE SOUTH

fig.

4.13.  Distribution of farmstead types in relation to geology in the South region

table

4.2: proportion



of farmsteads classified over time in the south region

Late Iron Age

Late first century a.d.

Second century a.d

Third. Fourth century a.d century a.d.

No. dated farmsteads classified

120

117

105

75

46

Total no. dated farmsteads

271

337

341

262

186

% classified

44.3

marking, good examples of which can be found at Ashford Prison, Spelthorne (Carew et al. 2006; see South Case Studies), and Westhampnett, Area 5 (Fitzpatrick et al. 2008). However, excavation of the few open farmsteads that date to the postconquest period in the South suggests that their settlement characteristics may have been quite different to those known from the later Iron Age. At Tolpuddle Ball in Dorset, Roman-period settlement activity was indicated by the construction of a small masonry, rectangular building with evidence for shale-working, corndrying, and animal-processing, which are all comparatively visible activities in the archaeological record (Hearne and Birbeck 1999). Similarly, at Foxholes Farm, near Hertford, evidence for

Romanch4.indd 87

87

34.7

30.8

28.6

24.7

metalworking and agricultural-processing was found alongside slight traces of timber buildings (Partridge 1989; fig. 4.17). At both of these sites, late Iron Age enclosure boundaries were not maintained post-conquest, but had these settlements not been fully excavated, their classification as ‘open’ would have been problematic (see Ch. 2 for issues with ‘fragmentary sites’). It is also notable that ten sites in the South are recorded simply as ‘isolated buildings’, with little further evidence to characterise them. Without substantial associated evidence for activities such as industrial or agricultural processing, which are highly visible in the archaeological record, many open settlements are likely to remain unclassified.

19/09/2016 12:12:47

88

THE RURAL SETTLEMENT OF ROMAN BRITAIN a

b

Wells Waterholes/wells

Possible waterhole

N

0

200 m 1:5000

c

fig. 4.14  Plans

of complex farmsteads at Pingewood, Burghfield, Berkshire (Johnstone and Bowden 1985) (a), Beam Washlands, Dagenham, Greater London (Biddulph et al. 2007) (b), and Chigborough Farm, Essex (Wallis and Waughman 1998) (c) a

b

N

0

200 m 1:5000

fig. 4.15  Site plans of complex farmsteads on the Hampshire Downs at Winnall Down/Easton Lane (Fasham 1985; Fasham et al. 1989) (a) and at Rowbury Farm, Wherwell (Cunliffe and Poole 2008e) (b)

Between the late Iron Age and the later Roman period, the proportion of farmsteads in the South region that are classified, whether as open, enclosed or complex, steadily reduces from just under 45 per cent to below 25 per cent (table 4.2). When we consider that the total number of

Romanch4.indd 88

farmsteads also falls across the period we are left with a much-reduced sample of classified farmsteads for the fourth century a.d. Only more extensive excavations of late Roman settlements will help us to understand better the overall settlement landscape in the South region.

19/09/2016 12:12:49

THE SOUTH

89

a

N

b

0

200 m 1:5000

fig. 4.16  Site plans of larger complex farmsteads at Owslebury (Collis 1970) (a) and at Dunkirt Barn (Cunliffe and Poole 2008f) (b) in Hampshire

A number of settlements in the South region show evidence for significant changes to their form and layout through time. However, the developmental sequences of many sites are very difficult to disentangle, a problem made all the more difficult where there has been disturbance and truncation. Evidence of abandonment and dislocation of settlement is, however, more evident from the sample. Several sites show good evidence for late Iron Age open settlements being abandoned and overlain by field systems or open pasture in the early Roman period, particularly in low-lying areas such as in the Middle Thames valley (i.e. Ashford Prison, Middlesex (Carew et al. 2006); cf.

Romanch4.indd 89

Preston 2012) or on the Sussex Coastal Plain (i.e. Westhampnett Bypass (Fitzpatrick et al. 2008); Titnore Lane, Goring-by-sea (Clarke 2012)). It is possible that the changes evident at these sites reflect corresponding, significant alterations in wider land use soon after the conquest. Just how widespread such changes were is difficult to gauge, but repeated examples in these low-lying areas suggest that they may have been extensive. The majority of settlements where development is recorded relate to farmsteads that acquired villa architecture, with 46 villa sites in the South displaying clear evidence for a settlement phase prior to villa construction.

19/09/2016 12:12:49

90

THE RURAL SETTLEMENT OF ROMAN BRITAIN Before A.D. 50

c. A.D. 50-250

Burials Waterhole c. A.D. 250-400

Field boundary Farm buildings?

After c. A.D. 400

Sunken huts Posthole buildings 0

200 m 1:5000

fig.

4.17.  Phased site plan for the open Roman-period farmstead at Foxholes Farm, Hertford (Partridge 1989)

The emergence and development of villas in the South In the South region villas have a long history of study, exemplified by the number of late antiquarian/early twentieth-century excavations recorded on the project database. These represent a minimum number, of course, since more villa sites are known from cropmark or field survey evidence (cf. Scott 1993). Others that have been insufficiently excavated have not been included in this study. The dating of villa sites, in terms of their initial construction and development over time, is central to our understanding of their social and economic role in the Romano-British countryside of the South, but this remains problematic in numerous cases. Several excavations have revealed elements of villa complexes that may only be later additions to the original building, leaving the dates of their initial establishment unknown (e.g. Fordcroft, Orpington (Philp and Keller 1995), and Teynham (Wilkinson 2011), both in Kent). Of the 113 villa sites in the dataset in the South region one-third can, at best, only be approximately dated to a broad period of occupation, but with no certainty of the dates of their construction and abandonment, and some cannot be dated at all. With the remaining two-thirds of sites, dating evidence indicates that initial villa constructions occurred between the later first century a.d. and the mid-fourth century a.d. In fig. 4.18 the chronologies of all the dated villas are shown divided between sites that are located in the eastern (Kent, East Sussex, West Sussex, Greater

Romanch4.indd 90

London and Surrey) or western (Hampshire and IoW, Berkshire, Dorset, Devon, Wiltshire and Somerset) half of the region. This division demonstrates substantial variability between different sites, including the period of occupation prior to the development of villa architecture, the date of their earliest villa building, and the length of subsequent use. Later first century a.d. villa establishments are predominantly restricted to the North Kent Plain along Watling Street and in the adjoining river valleys of the North Downs, and similarly along the Sussex Coastal Plain (fig. 4.19). The early coastal plain villas are a generally well-known phenomenon, argued by some as lavish investments by local elite groups in order to associate themselves with a new political authority (Cunliffe 1973, 79; Rudling 1998). It may be no coincidence that first-century a.d. villa establishments were located on the Sussex and Kent Coastal Plains, areas that would have sat at the interface between Britain and the continent, forming the main access routes for trade and other traffic across the channel. Investments of wealth in these areas would have been important for demonstrating local positions of power, though the development of these villas must also be considered against the context of late Iron Age elite activity and the central role of longdistance trade networks (Creighton 2000; Hill 2007). High-status, pre-conquest activity at these early villa sites has so far only been discovered at Fishbourne where the fill of a ditch produced feasting waste, luxury artefacts and imported dining wares (Manley and Rudkin 2005).

19/09/2016 12:12:50

THE SOUTH

91

Date BC/AD 500

400

300

200

100

01

500

400

300

200

Langstone Bramdean Brading Dinnington Halstock Sparsholt Tarrant Hinton

date of villa construction

Rockbourne date of villa construction

Charminster Ilchester Mead Myncen Farm Cox Green Thornford Meonstoke Pamphill Honeyditches Shide, Newport Combley Fullerton Dewlish Holcombe Liss Dunkirt Barn Downton Bucknowle Houghton Down Grateley South Wellhead

Figure 4.18: Chronology of occupation on villa sites in the South region

Romanch4.indd 91

100

fig.

10

-100 Fishbourne Ashtead Angmering Walberton Thurnham Eccles Northfleet Faversham Park Street Chilgrove 1 Minster-in-Thanet Sandwich villa Southwick Chichester Harbour Chalk Snodland Farningham III Teynham Walton-on-the-Hill Cobham Park Horton Kirby Lullingstone Goring Folkestone Darenth Littlehampton Allen's Farm, Plaxtol Crofton, Orpington Bignor Roman villa Little Oakley Sidlesham Watergate Barcombe Hull Place, Sholden The Mount, Maidstone Beddington Keston Rapsley, Ewhurst Chiddingfold Batten Hanger Beddingham Worplesdon Hale Road, Farnham Little Chart Farningham II Upmarden Chilgrove 2

-100

Date BC/AD

4.18.  Chronology of occupation on villa sites in the South region

19/09/2016 12:12:51

92

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

4.19.  Distribution of villas over time in the South region (top left: late 1st century a.d.; top right: 2nd century bottom left: 3rd century a.d.; bottom right: 4th century a.d.)

a.d.;

Although northern Kent and southern Sussex led the way with most of the earliest villas, the eastern half of the South region then saw a substantial reduction in their number after a peak in the later third century a.d. (fig. 4.20). Villa settlement in the western half, in contrast, did not peak until the early fourth century a.d. and, by the late fourth century a.d., more villas were occupied overall in the west than in the east. Rudling (1998) has observed that several of the first-century a.d. Sussex villas, particularly those on the coastal plain, had declined by the end of the third century a.d., with a number abandoned by the beginning of the fourth century a.d., while those on the chalk downland of Hampshire and the Greensand of West Sussex, notably at Bignor, tended to

Romanch4.indd 92

witness increased investment and expansion over the same period. Explanations for the decline have focused upon coastal raiding, which may have affected the fortunes of some settlements, but not necessarily all. In the second century a.d., villa construction had spread across the chalk downland into Hampshire and Dorset, including a group located just south of Lindinis (Ilchester), a town well known for having a cluster of villas in its hinterland (King forthcoming). However, a number of the second century a.d. sites were relatively modest establishments, certainly compared with some of the earlier Sussex villas. It was not until the third or fourth century a.d. that settlements such as Ilchester Mead (Hayward 1982), Dinnington

19/09/2016 12:13:17

THE SOUTH

93

Kent, Essex, Greater London, East Sussex, West Sussex and Surrey Berkshire, Devon, Dorset, Hampshire, Somerset, and Wiltshire 20 18 16 no. of sites

14 12 10 8 6 4 2 0 AD43-100 fig. 4.20.  Chronological

AD100-150

AD150-200

AD200-250

AD250-300

>AD300

variation in date of villa construction between eastern and western areas of the South region

Late first-second century A.D. field system

Late Iron Age enclosed farmstead

N

Third century A.D. villa

Fourth century A.D. villa

0

80 m 1:1250

fig.

4.21.  Phased plans of the site at Beddington, Surrey (Howell 2005)

(King forthcoming), and Rockbourne (Hewitt 1968), developed into grander establishments. Most of the sites that received villa architecture in the later Roman period (third and fourth century a.d.) had long histories of prior occupation.

Romanch4.indd 93

In some cases, for example at Bucknowle (Light and Ellis 2009), Grateley South (Cunliffe and Poole 2008c), and Beddington (Howell 2005; fig. 4.21), settlements established in the Iron Age continued in use throughout the Roman period.

19/09/2016 12:13:19

94

THE RURAL SETTLEMENT OF ROMAN BRITAIN

40

% of villas

30 20 10 0

fig.

4.22.  Villas as a percentage of the total number of settlements in different landscape zones of the South region

The expansion, development and decline of villa settlement shows a considerable degree of variability through time across the South region, with some evidence that villas were more commonly found in particular landscapes (fig. 4.22). This variation may reflect differences in site visibility between geologies, or the level of fieldwork undertaken on other types of settlement, particularly where developer-funded excavations are more or less intensive. While these are important factors, the availability and suitability of exploitable land must also have been a consideration for villa owners. Villas are recorded in their greatest relative frequency on the Wealden Greensand, mostly towards the northern edge of the Weald, with the lowest in the London Basin. Today, the suitability of the soils on the Greensand for arable farming varies between areas of lighter, better-drained land located in or near the river valleys, and areas of sandy or heavy clay soils that tend to be covered in heath and rough grazing vegetation. Variation in the local environment may have been an attractive prospect for villa owners. Applebaum remarked long ago that villa owners may have had a preference for land at the margins of different soils, where arable, woodland, and other environments might be exploited together (Applebaum 1958, 69–70), a point also raised by Frere in relation to Bignor villa (Frere 1987, 265). The assumption that arable farming produced the bulk of the wealth invested into Romano-British villas is perhaps an over-generalisation (see Millett 2007a; 2014), and the differences observed in villa location perhaps suggest that the sources of their wealth, if derived locally, varied between sites and between landscapes. It has been stated previously that villas should not necessarily be seen in isolation from other forms of farming settlement (e.g. complex, enclosed or open farmsteads), because their

Romanch4.indd 94

classification is determined by their architecture rather than their overall settlement form (see Ch. 2). It is unfortunate that many villa excavations simply do not provide information on wider features, due to their focus on the buildings. The recent work at Brading on the Isle of Wight is a rare exception, where geophysical survey exposed a complex of late Iron Age and Roman field boundaries, surrounding the late Roman courtyard villa (Cunliffe 2013a; fig. 4.23). No villas in the South can be strictly identified as being complex in form, though some variation in the layout of villa settlements can be highlighted. A number of villas include masonry walls that link separate buildings and thus provide a centrally enclosed space, thereby crystallising the rectilinear form of the complex. The villas at Sparsholt and Stroud, both in Hampshire, exhibit this type of arrangement with masonry buildings attached externally to the central walled area (Johnston and Dicks 2014; fig. 4.24). Bignor villa in West Sussex is one of the best examples of a villa in the South that developed from a loose array of masonry structures in the early Roman period, to being formally enclosed by a wall, and eventually to become a complex, double-courtyard villa in the fourth century a.d. (Aldsworth and Rudling 1995). The villas at Halstock in Dorset (Lucas 1993) and Rockbourne in Hampshire (Hewitt 1968) show similarities with Bignor in their final form, encircling a courtyard with ranges around three sides (see more plans in the villa building section below). While some villa settlements present evidence of enclosure, others provide evidence for internal division. At Keston in Kent the villa complex in its third-century a.d. phase includes at least two ditches that together partly enclose the main domestic area, while at the same time dividing it from a temple-mausoleum area immediately to the north (Philp et al. 1991; fig. 4.24). Within the

19/09/2016 12:13:20

THE SOUTH

95

N

0

Limit of geophysical survey

200 m 1:5000

fig. 4.23.  Geophysical survey results at Brading villa, Isle of Wight, showing late Iron Age and Roman enclosures (Cunliffe 2013a)

a

b Aisled building

Baths

Shafts Well

North cemetery

N

Timber building House

Villa House Barn?

Fencelines Lim

it o

Hall

Timber building

0

f ex

cav

atio

n

100 m 1:2000

fig. 4.24.  Plans of villas at Sparsholt, Hampshire (Johnston and Dicks 2014) (a) and Keston (third century a.d. phase), Kent (Philp et al. 1991) (b)

main complex, fences delineate a loosely arranged set of masonry structures, setting out different areas of space and, perhaps, activity. Through a re-examination of the plan of the villa at Rapsley in Surrey, Smith (1980) argued that a central wall, which divided the complex, signified evidence for dual-ownership of the villa. Of course, the question of villa ownership is highly debatable and is largely

Romanch4.indd 95

tangential to the archaeological evidence, but it is important to observe that many of the villas in the South region were laid out or developed in different formats that varied over time. It is difficult to quantify the range of villa forms in the South due to the varied and often restricted results produced by many villa excavations. The main problem with identifying the form of villa

19/09/2016 12:13:21

96

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

N

0

0

50 m

200 m 1:5000

1:1000 b

0

100 m 1:2000

0

200 m 1:5000

fig. 4.25.  Excavation and interpretative plans of late Roman villas at Houghton Down (Cunliffe and Poole 2008b) (a) and Dunkirt Barn (Cunliffe and Poole 2008f) (b) in Hampshire

Romanch4.indd 96

19/09/2016 12:13:22

THE SOUTH

complexes is that, although many appear to have been ‘open’ in plan, the full extent of the settlement is rarely observed. Here, cases in point are illustrated by the Hampshire villas at Houghton Down (Cunliffe and Poole 2008b), Grateley South (Cunliffe and Poole 2008c), and Dunkirt Barn (Cunliffe and Poole 2008f), where evidence for much larger bounded enclosures have been identified from geophysical survey and cropmarks, each encompassing something in the region of 15 ha around the main villa complexes (fig. 4.25). Undoubtedly, these were settlements of considerable size that would probably have extended out to wider field-systems. However, if the villa buildings had been investigated in isolation, as so many sites of this type were, the wider form of the settlement would have gone unrecognised. Nucleated settlements Roadside settlements and villages are the main forms of rural, nucleated settlement found in the South, although a few Iron Age oppida and hillforts are also included in the database because they have late Iron Age phases of activity (though not those that later develop into major Roman towns, such as Silchester, which are beyond the scope of this project). Roadside settlements are relatively evenly distributed across the region, being particularly prevalent on the major roads leading to and from London. Other village settlements are mainly, though not exclusively, recorded in the western half of the region and, as mentioned, this is primarily due to the exceptional survival and survey of earthworks and cropmarks on Salisbury Plain and the South Wessex Downs (McOmish et al. 2002; Fulford et al. 2006), which may have no bearing on their real distribution across the South. One notable absence from the dataset is the excavated evidence for vici associated with the Saxon Shore forts in the South. Very little work has been focused on their identification, though recent aerial and geophysics survey around Richborough demonstrates that these settlements could have been extensive (Small 2002). In total, 68 nucleated settlements have been recorded from the South, with some known from a number of different excavations (see Ch. 2). In terms of broad dating, the late Iron Age includes a mixture of villages, hillforts and oppida, and though these are classified separately in this way, their respective character and function within the landscape may not have been very different from one another. Roadside settlements, of course, did not develop until the layout of the major road system when they became the predominant form of nucleated settlement from the later first century a.d. through to the later fourth century a.d.

Romanch4.indd 97

97

Roadside settlements In the South, nearly half of the nucleated settlements developed alongside roads. These vary between foci located at crossroads or road/ river intersections and ribbon developments, sometimes located close to other major urban settlements (see Ch. 2). The full extent of roadside settlements is often poorly recognised, though some estimates gained through survey suggest that a wide range may have existed: 9 ha at Alfoldean (Thompson 2006a), 10 ha at Westhawk Farm (Booth et al. 2008; Chapter 2, fig. 2.25), 14 ha at Neatham (Millett and Graham 1986), 18 ha at Yeovil (Reed 2000), 25 ha at Shapwick (Papworth 1994; fig. 4.26) and up to 36–45 ha at Old Sarum (Sorviodunum) (James 2002). Unfortunately, only eight roadside settlements have seen single excavations in excess of 0.5 ha. These include Elms Farm at Heybridge (Atkinson and Preston 2000), Syon Park, Brentford (Cowie et al. 2013), Springhead (Andrews et al. 2011) and Westhawk Farm (Booth et al. 2008), which together stand out as the most extensively excavated examples in the South. Even though the excavated areas of these sites still do not cover the full extent of the settlements, they provide useful impressions of their form (additionally facilitated at Westhawk Farm by extensive geophysical survey). Although their layouts differ, each of these settlements shows evidence for side lanes, ‘property’ plots with buildings, hearths and wells, shrines and areas set aside for industrial and/or agricultural processing. Eight roadside settlements show some evidence for activity prior to the Roman conquest and, presumably, before the construction of the road alongside which they continued to develop. At Springhead, late Iron Age enclosures and a ‘processional way’ leading from the river up the slopes overlooking the springs are all thought to have been elements of a ritual centre (Andrews et al. 2011, 13–31), whereas at Westhawk Farm a small cemetery, including a high-status burial, is the only landscape feature conclusively dated preconquest, but which may have served as an important focus for the later development of the settlement (Booth et al. 2008, 27–34). While ritual activity dating to the late Iron Age has been identified at both of these sites, categorical evidence for pre-conquest domestic settlement is more elusive. Only at Old Sarum, Sorviodunum, is late Iron Age settlement attested, albeit within the defended hilltop enclosure that later became incorporated into the roadside settlement during the Roman period (Rahtz and Musty 1960; James 2002; Moffat 2010). This bears some similarity with Poundbury, due east of Dorchester, though the evidence for settlement here does not appear

19/09/2016 12:13:23

98

THE RURAL SETTLEMENT OF ROMAN BRITAIN

N

Cropmark Magnetic anomaly 0

200m 1:5000

fig.

4.26.  Plan of the roadside settlement at Shapwick, Dorset (Papworth 1994)

to have been focused on the nearby road. A possible late Iron Age defended site at Judd’s Hill, Kent, is also perhaps a precursor to the roadside settlement at Ospringe, Durolevum, best known for its large, mixed-rite, Roman-period cemetery (Harding 2003; Wilkinson 2008). While the vast majority of roadside settlements in the South demonstrate occupation commencing in the later first century a.d. and continuing through to the second half of the fourth century a.d., this broad chronology masks much of the fluctuations in settlement activity that occurred over time. The decline in settlement activity at Westhawk Farm and Springhead through the late third and early fourth century a.d. has already been discussed in this chapter, yet our understanding of these settlements is well served by large-scale excavations. At sites with smaller areas of excavation, establishing their chronology is more problematic. The dating evidence from a range of sites in Staines, Surrey, shows that different areas of the Roman settlement were in use at different times in its history. Located along the London to Silchester road on a number of gravel islands, where the road crosses the River

Romanch4.indd 98

Thames close to its confluence with the Colne, Staines (Pontibus) was an important roadside settlement at a vital river crossing, which was established very soon after the conquest and continued to be occupied into the later fourth century a.d. figure 4.27 presents the date ranges of fourteen occupation sites in Staines, displayed from left to right as they were located from the west, at the crossing of the Thames, running eastward alongside or a little set back from the road. The central and westernmost sites constitute the earliest evidence for activity, mostly dating at least from the pre-Flavian period. Sites in the eastern part of the settlement come into use during the second century a.d. and appear to be used for burials rather than domestic occupation (Hayman and Ayres 2001). However, from the early third century a.d. the central and westernmost sites all appear to have been affected by periodic flooding so that, by the middle of the third century a.d., settlement activity closest to the river was abandoned (Crouch and Shanks 1984; McKinley 2004; Jones 2010). By the later third century a.d., evidence for occupation across the settlement as a

19/09/2016 12:13:23

THE SOUTH

99

riverside

450 400

year / AD

350 300 250 200 150 100 50 10

change in nature of occupaƟŽŶ fig. 4.27.  Variation in dating evidence from different excavations in Staines (sites leading eastward from the river [left to right])

whole is characterised by less intensive activity, though the easternmost sites begin to show evidence of occupation into the fourth century a.d. (Hayman 2000). The shifting spatial patterns at Staines suggest that considerable fluctuations in the form, character and extent of roadside settlements could have occurred over time, and this point must be considered when these larger and more complex settlements are represented by a few or even only a single excavation. Villages Villages are represented by twenty individual settlements, predominantly in the western half of the South region with ten being located on the South Wessex Downs. The identification of these village settlements is mostly due to their visibility as earthworks on the chalk downland, which led to a number being investigated prior to the 1930s (Pitt-Rivers 1887; 1892; Cunningham 1913; Hawley and Goddard 1924; Kivell 1926). Some of the settlements are extensive and are larger than many of the roadside settlements in the region. Those within the military training area of Salisbury Plain have recently been subject to detailed survey and grouped as either ‘compact’ villages with sizes ranging from 26 ha (Charlton Down) to 6 ha (Knook Down West) and 3.5 ha (Knook Down East), or ‘linear’ villages such as Chapperton Down, Orcheston Down, Cheverall Down, Chisenbury Warren (5 ha) and Coombe Down (2 ha) (McOmish et al. 2002, 87–108; Fulford et al. 2006). In terms of their form, the Wessex Downland settlements appear to have been agglomerations of compounds with associated, sub-rectangular ‘hut’ platforms, which developed quite organically, with

Romanch4.indd 99

no obvious centre or focal point, while field systems/lynchets can usually be identified at their peripheries (fig. 4.28). The ‘linear’ villages display similarities with roadside settlements in the way that they develop along a trackway, causing the settlements to develop in more of a linear form as a result, with partly excavated examples at Castle Farm (Tabor 2004), Chapperton Down (Malim and Martin 2007), Cheverall Down and Chisenbury Warren (McOmish et al. 2002, 99, 102). Aerial and ground survey of the Salisbury Plain Training Area has revealed extensive evidence of field systems, with the molluscan evidence from excavations of lynchets pointing to a marked intensification of cultivation in the RomanoBritish period (Bradley et al. 1994, 108, 111; McOmish et al. 2002, 100–4, fig. 4.1; Fulford et al. 2006; fig. 4.29). The importance of water storage and conservation in this relatively dry environment is demonstrated by the evidence for the provision of dams, cisterns and ponds (McOmish et al. 2002, 103–4, figs 4.6, 4.18). On the Hampshire Downs, the site at Chalton, surveyed and excavated in the 1970s by Cunliffe (1977), presents a comparable type of settlement. Though not as large as some of the South Wessex agglomerations, perhaps covering around 4–5 ha in extent, Chalton also exhibits ‘hut’ platforms, integrated trackways and surrounding field systems. In many ways these types of settlements bear some similarity with the complex farmsteads that dominate the river valleys to the north of the chalk landscape, particularly along the Thames, and it may be that they simply represent a slightly different, perhaps larger, form of agglomerated settlement in these chalkland landscapes.

19/09/2016 12:13:24

100

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

N

b

c

0

200 m 1:6000

fig. 4.28.  Site plans of Knook Down East (a), Charlton Down (b) and Cheverall Down (c) on the Salisbury Plain (McOmish et al. 2002, 92, fig. 4.7; 96, fig. 4.11, 102, fig. 4.16)

Romanch4.indd 100

19/09/2016 12:14:48

THE SOUTH

101

15 0

150

Charlton Down

Upavon Down

Church ditches Compton Down

150 5

0

Old

Nu

rse

100

ry D

itch

1km 1:25,000

fig. 4.29.  Aerial survey plan of the village settlements at Charlton Down, Upavon Down, and Compton Down, Wiltshire, showing multiple field systems and trackways (McOmish et al. 2002, 107, fig. 4.21)

N

Cleavel Point

0

200 m 1:5000

fig.

4.30.  Plan of first-century a.d. settlement enclosures at Ower, Poole Harbour, Dorset (Sunter and Woodward 1987)

Romanch4.indd 101

19/09/2016 12:14:53

102

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Other ‘village’ sites include three that probably also functioned as ports: Hamworthy and Ower in Poole Harbour (Sunter and Woodward 1987; fig. 4.30), and the Royal Manor Arts College site on Portland on the Dorset coast (Palmer 2009). These sites are difficult to classify, but they appear to have also been engaged in substantial, nonagricultural activities with evidence for pottery production (Sunter and Woodward 1987), shale manufacturing (ibid.), salt production (Jarvis 1993; Coles and Pine 2009), and the processing of fish and wild fowl (Palmer 2009; see also Maltby and Hamilton-Dyer 2012). The continental pottery imports identified at these sites suggest that maritime trade played a part in their economies. Further variation in village settlement characteristics can be seen at Monkton on the Isle of Thanet. Here a road-scheme excavation revealed part of an extensive settlement with at least 23 ‘cellared buildings’ and ancillary structures, located alongside a trackway, and with a small rectangular structure on the settlement’s western fringes, which was interpreted as a shrine (Bennett et al. 2008). The agglomeration of so many cellared buildings dating to the Roman period is

fig.

Romanch4.indd 102

unusual for Britain. The settlement originated in the later first century a.d. and was primarily active during the second and third centuries a.d., producing considerable evidence for a diversified economy including the processing of arable surplus, metalworking and bone-working. On the basis of more limited excavation at Tothill Street, Mount Pleasant, it may even have extended c. 1 km further to the east (Nielsen 2010). Evidence from other local sites, in particular the recent and extensive excavations along the East Kent Access road-scheme (Andrews et al. 2015), suggests that the southern part of the Isle of Thanet may have been relatively densely occupied during the Roman period, perhaps giving rise to a number of nucleated settlements. Buildings As is becoming clear from the excavated evidence, there is a great variety of settlement types across the South region, and this diversity can be examined further through the remains of late Iron Age and Romano-British rural buildings. In total, 1042 structures from 340 sites, spread relatively

4.31.  Distribution and frequency of rural buildings on excavated sites in the South region

19/09/2016 12:15:01

THE SOUTH

103

100

% of selements

80 60 40 20 0

fig.

4.32.  Frequency of settlements with evidence for buildings by different landscape zones in the South region

evenly across the region, are recorded on the project database (fig. 4.31), providing a sample that ranges from small timber buildings to the late first century a.d. palatial villa at Fishbourne, which remains one of the most grandiose buildings ever to have graced Roman Britain (Cunliffe 1971). However, there is considerable variation in the percentage of settlements that produce evidence for structures in the different sub-regions, from as low as 29 per cent of sites on the Hampshire Downs to as high as 81 per cent in the Wessex Vales (fig. 4.32). Equally, the number of buildings identified on different sites can vary significantly, depending on conditions of preservation and the size of the settlement. As we shall see, the types of site where buildings have been identified, the form of the buildings, their

table

4.3: quantification

construction techniques and the materials used, were all contributing factors to their recognition and rate of survival. Context, form and construction Of the 1042 buildings recorded from sites in the South, over 30 per cent derive from farmsteads, over 20 per cent from villas, and over 15 per cent from farmsteads with good evidence for later villa phases (table 4.3). Nucleated settlements also contribute a significant proportion of the sample, with roadside settlements and villages accounting for 13.5 per cent and 11.1 per cent respectively. The remaining structures are spread among sites that could only be categorised as ‘isolated buildings’, industrial sites, and religious sites.

of buildings at different site types in the south region

Note: It was not always possible to identify timber and masonry buildings with certainty so the numbers of each may not equal the total Site type (n=no. of sites)

Romanch4.indd 103

Circular Rectangular Timber Masonry Total buildings buildings buildings buildings buildings no. % no. % no. % no. % no. % average

Farmstead (n=140) Villa (n=90) Farmstead/villa (n=25) Roadside settlement (n=20) Village (n=16) Hillfort (n=4) Isolated building (n=6) Industrial (n=16) Shrine (n=5) R-C temple (n=18)

158 56.2 165 21.7 251 45.7 71 14.5 323 31.0 2.3 4 1.4 209 27.5 6 1.1 207 42.4 213 20.4 2.4 35 12.5 122 16.0 58 10.6 98 20.1 157 15.1 6.3 27 9.6 114 15.0 95 17.3 44 9.0 141 13.5 7.1 36 12.8 80 10.5 92 16.8 24 4.9 116 11.1 7.3 11 3.9 4 0.5 14 2.6 1 0.2 15 1.4 3.8 0 0.0 7 0.9 2 0.4 4 0.8 7 0.7 1.2 3 1.1 32 4.2 26 4.7 9 1.8 35 3.4 2.2 2 0.7 6 0.8 3 0.5 5 1.0 8 0.8 1.6 5 1.8 22 2.9 2 0.4 25 5.1 27 2.6 1.5

Total

281 761 549 488 1042

19/09/2016 12:15:02

104

THE RURAL SETTLEMENT OF ROMAN BRITAIN

no. of sites with circular buildings

no. of sites with rectangular buildings

200

no. of sites

150

100

50

0 Late Iron Age fig.

Late 1stC AD

2ndC AD

4thC AD

4.33.  Number of sites with circular and/or rectangular buildings over time in the South region

Roadside settlements and villages stand out as having the highest average number of buildings per site, reflecting their status as nucleated settlements with higher populations of people and, presumably, animals that required stabling, and, perhaps, a greater diversity of commercial and industrial activities compared with smaller farming establishments. Further differences emerge between different forms of settlement in terms of the ratio of circular to rectangular buildings. While circular and rectangular forms are found in broadly equal quantities at farmsteads, at all other site types rectangular forms dominate. Circular buildings are rare on villa sites, but they are slightly better represented in villages than in roadside settlements. These variations partly result from chronological factors, since the number of sites with circular architecture gradually decreases in the South after the late Iron Age, whereas sites with rectangular architecture become more common by the later first century a.d. (fig. 4.33). The establishment of early villas and roadside settlements in the later first century a.d. certainly impacts on the proportion of sites with rectangular architecture. However, the uptake of rectangular architecture is also readily noticeable on farmsteads and these show a very similar trend to the overall pattern of change. The use of masonry and timber architecture also varies significantly between different types of site. It is of course very difficult to identify which buildings were ‘fully’ masonry-constructed and those that had masonry foundations with a timber superstructure. Timber-based (and mass-walled) buildings are also more difficult to identify than masonry structures, tending to rely on the presence of well-preserved postholes, beam slots and/or drainage gullies, and are probably relatively poorly represented in the archaeological record (see Ch. 3). Only at villas is masonry architecture, unsurprisingly, used

Romanch4.indd 104

3rdC AD

extensively in preference to timber building, whereas on all other types of site, timber tends to dominate. However, the uptake of masonry as a basic construction material can be traced through the dating of sites where it is first recorded. The use of stone is restricted to three sites in the late Iron Age, all of which are located in south Dorset: Maiden Castle (Wheeler 1943), Compact Farm, Worth Maltravers (Graham et al. 2002) and Rope Hole Lake, Corfe Castle (Sunter and Woodward 1987). Each of these examples comprises similarly sized, drystone, circular buildings whose close geographic distribution suggests that they formed part of a local architectural tradition. At Rope Hole Lake, Corfe Castle, the presence of circular masonry structures was first evidenced in the early or middle Iron Age and appeared to continue into the late Iron Age. Based upon the finds recovered, the site appears to have been primarily engaged in shale manufacture, and similar evidence was derived from two similar structures at Compact Farm. The Maiden Castle example was later interpreted, based upon its finds assemblage, as a late Iron Age shrine (Drury 1980; Wait 1985). In the late first century a.d. masonry buildings are almost exclusively restricted to the east of the region, predominantly on villa sites, but also on roadside settlement and, to a lesser extent, on farmstead, temple and industrial sites (fig. 4.34). In the second century a.d., their distribution spreads right across the region, particularly on sites on the chalk downs and in the Wessex Vales. By the third century a.d., masonry architecture was well established in the eastern half of the region with, by then, very few sites adopting masonry architecture for the first time, unlike further west where its uptake was now considerable. It is notable that very few roadside settlements developed masonry architecture for the first time later than the later first century a.d., with examples

19/09/2016 12:15:03

THE SOUTH

fig.

105

4.34.  Distribution of excavated masonry buildings over time in the South region

100%

%masonry

%mber

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

South Wessex Wessex Vales North Downs Hampshire Downs (n=19) (n=28) and Coastal Downs (n=16) Plain (n=43)

The Weald (n=33)

South Downs London Basin and Coastal (n=108) Plain (n=23)

fig. 4.35.  Relative frequency of masonry and timber buildings found at farmsteads within selected landscape zones in the South region (n=no. buildings)

on farmsteads, particularly those that developed into villa establishments and, to a lesser extent, villages and industrial sites occurring in the second and third century a.d. The intra-regional variation in the use of timber or masonry buildings at farmsteads is also evident from fig. 4.35, which shows a high proportion of masonry structures on farmsteads on the South Wessex Downs and the Wessex Vales, while timber

Romanch4.indd 105

buildings tend to dominate in most other zones to the east. This pattern partly results from chronological factors, owing to the comparative lack of early Roman farmsteads from the western areas of the South region where timber building might be expected. Timber buildings overwhelmingly dominate on farmsteads in the London Basin and the South Downs and Coastal Plain, even into the later Roman period.

19/09/2016 12:15:11

106

THE RURAL SETTLEMENT OF ROMAN BRITAIN a

b

N

0

20 m 1:500

fig. 4.36.  Plans of internally partitioned posthole structures at Winnall Down, Hampshire (Fasham 1985), (a) and late first-century a.d. timber-framed structure at Chichester Harbour, Sussex (Rudkin 1986) (b)

Types of timber buildings Owing to the poor survival of evidence of timber buildings in the archaeological record, it is difficult to appreciate their full range and variety. This issue has recently been brought into sharper focus by the discovery of later first-century a.d. clayfloored areas set alongside the Roman road at the Prudential 1989 site in Staines, which provided evidence of box-framed timber buildings with no trace of postholes or beam slots (Jones 2010). Since the basic foundations of timber buildings are often difficult to identify, the recognition of internal division is even more challenging. In the majority of cases, we might assume that timber buildings were single-roomed, with flimsier partitions leaving little trace. This seems to have been the case with the five rectangular buildings excavated at Park Brow on the South Downs in the 1920s. Though thought to have been of simple, wattle and daub construction, some painted plaster and window glass were also recovered (Wolseley et al. 1927). Rare examples of timber structures with internal partitions include at least two first–second century a.d. posthole buildings at Winnall Down, Hampshire (Fasham 1985), and several firstcentury a.d. rectangular, beam-slot structures in the hillfort at Cadbury Castle (Barrett et al. 2000). Excavations at the roadside settlements at Bourne Place, Neatham (Graham 1992), and at Elms Farm, Heybridge (Atkinson and Preston 2000), have revealed floor plans of structures that may have consisted of more than one room, while the

Romanch4.indd 106

enigmatic, cellared buildings excavated at Monkton, Thanet, also provide indications of internal partitioning (Bennett et al. 2008). At Chichester Harbour a timber building, dating to the later first century a.d., with corridors surrounding a range of at least two rooms, was replaced in the second century a.d. by a large, masonry-footed, aisled building (Rudkin 1986; fig. 4.36). Sites with evidence for cellared buildings are generally rare, though they are located widely across the South region, from Essex to Dorset (see Ch. 3, fig. 3.17 for national distribution). Their distribution, however, shows a marked concentration in eastern Kent, notably on the Isle of Thanet, where the recent discovery of at least 23 buildings at Monkton, predominantly dating to the second and third century a.d., has been followed by further examples at Tothill Street, Mount Pleasant (Nielsen 2010) and along the East Kent Access Road scheme (Andrews et al. 2015). Together, they suggest a localised building tradition on the Isle of Thanet. Farmsteads with masonry buildings Sites with insubstantial timber buildings may well be heavily under-represented in the archaeological record compared with masonry structures, which are reported from 22 per cent of farmsteads. While there has been an assumption on the part of some excavators that the presence of a masonry building signals a villa, this is often on the basis of limited investigation. In the first place we should consider buildings of unmortared, drystone footings (to

19/09/2016 12:15:11

THE SOUTH

107

N

0

Hearth

20 m 1:500

fig.

4.37.  Site plan of the late Roman building complex at Woodhouse Hill, Dorset (Field 1965)

Thruxton

North Warnborough

Clanville

N

Houghton Down

Stroud

0

20 m 1:500

fig.

Romanch4.indd 107

4.38.  Plans of aisled buildings with later modifications (Cunliffe 2013b, 99, fig. 6.2)

19/09/2016 12:15:13

108

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

4.39.  Distribution of aisled buildings across the South region

a

b

N

0

20 m 1:500

fig. 4.40.  Plans of timber aisled buildings at Furfield Quarry, Boughton Monchelsea, Kent (Mackinder 2006) (a), and Bower Road near Smeeth, Kent (Diez 2006) (b)

support a timber-framed or cob superstructure) such as the extensively excavated complex of firstto fourth-century structures found at the site of Woodhouse Hill, Studland, Dorset, which possibly formed part of a larger, nucleated settlement (fig. 4.37). The simple, rectangular cottages lacked any associated refinement such as tessellated flooring, window glass, or painted wall plaster (Field 1965).

Romanch4.indd 108

One of the more recognisable structural forms common to the Roman period is the aisled building, the definition of which is detailed in Chapter 3 (fig. 4.38). While tending to be a feature of villa complexes, aisled buildings are also represented on nine farmsteads, two roadside settlements and a village site in the South (fig. 4.39). Of the buildings at the nine farmsteads, six

19/09/2016 12:15:20

THE SOUTH

a

109

b

N

0

20 m 1:500

fig. 4.41.  Plans of cottage/strip-house villas at Houghton Down, Hants (Cunliffe and Poole 2008b) (a) and Shillingstone, Dorset (Corney and Robinson 2007) (b)

a

0

20 m 1:500

b

c

fig. 4.42.  Plans of winged-corridor villas at Barcombe, Sussex (Rudling et al. 2011) (a), Lullingstone, Kent (Meates 1979) (b), and Sedgebrook Field, Plaxtol, Kent (M. Davies 2009) (c)

Romanch4.indd 109

19/09/2016 12:15:20

110

N

THE RURAL SETTLEMENT OF ROMAN BRITAIN

b

a

N

N

c

0

100 m 1:2000

fig. 4.43.  Plans of courtyard/‘palatial’ villas at Eccles, Kent (Detsicas 1989) (a), Fishbourne, Sussex (Cunliffe 1971) (b), and Darenth, Kent (Philp 1984) (c)

were masonry-footed, while those at Bower Road near Smeeth and Boughton Monchelsea, Kent, were timber-built (Diez 2006; Mackinder 2006; fig. 4.40). Booth et al. suggest that the Bower Road example displays only ‘similarities with aisled buildings’, and argue that the postholes actually represent a wall line, with the surrounding trench being a ditch rather than a wall footing (Booth et al. 2011, 275). The example at Boughton Monchelsea may be of a similar type, perhaps reflecting a construction tradition that was local to Kent (Booth pers. comm.). Only two other aisled buildings of timber are recorded from the South, one at the roadside settlement at Neatham, Hampshire (Millett and Graham 1986), the other at the villa at Little Oakley, Essex (Barford 2002). Those with masonry footings may have either had masonry walling to roof level or timber superstructures. The aisled building at Meonstoke, Hampshire, with its fallen but well-preserved gabled end, provides evidence of such a masonry-constructed aisled building in the South (King 1987). The care

Romanch4.indd 110

taken in the coursing of the flint and tile in the elevations of this building suggests that the exterior of the walls were not plastered, but meant to be seen as decoration. At least ten aisled halls, mostly on villa sites in Hampshire, have been interpreted as the main residential building of the settlement, an indication of their social significance in the region. Indeed, the impressive aisled building at West Blatchington, West Sussex was the only masonry building at the site. As Cunliffe (2013b, 98) has suggested, the size and function of these structures may provide further indications of the social status of their owners and of their role within settlements. Indeed, there does appear to be a correlation between the larger examples and those apparently used for residential purposes; aisled buildings measuring in excess of 600 m² tend to be interpreted as domestic buildings, while those smaller than 600 m² are more commonly thought to have had an agricultural function (the latter is often supported by the presence of agricultural features, such as corndryers).

19/09/2016 12:15:21

THE SOUTH

a

111

1st - 2nd C A.D.

2nd - 3rd C A.D. ?Aisled building (Barn 1)

Early ditch

?Shed 1

Unfinished baths Abandoned furnace

Period IIE VILLA

Shed 1

Barn 1

Baths

4th C A.D. Shed 2

Barn 2 0

100 m 1:2500

b 1

2

5

3

6

4

7

8

fig. 4.44.  Plans showing the development of the villas at Bignor, West Sussex (Aldsworth and Rudling 1995) (a), and Rockbourne, Hampshire (Hewitt 1968) (b)

Romanch4.indd 111

19/09/2016 12:15:22

112

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Villa buildings In Chapter 3, we outlined the main villa building types: cottage-style/strip-house (fig. 4.41: (a) Cunliffe and Poole 2008b; (b) Corney and Robinson 2007), corridor/winged-corridor (fig. 4.42: (a) Rudling et al. 2011; (b) Meates 1979; (c) M. Davies 2009), courtyard and palatial (fig. 4.43: (a) Detsicas 1989; (b) Cunliffe 1971; (c) Philp 1984). Analysis of their incidence in the South shows that all types are represented between the late first and the fourth century a.d. Of the

few that can be assigned to the later first century a.d., the diminutive cottage-style or strip-house type villa building is marginally the most common, while the second century a.d. sees a significant increase in the number of corridor and wingedcorridor villas across the region. The overall number of cottage/strip-house villas and the larger courtyard types also increases in number but to a lesser extent, a trend that continues into the third century a.d., after which the total number of villas decreases slightly. By the fourth

0

20 m 1:500

fig.

Romanch4.indd 112

4.45.  Plan of the ‘villa’ complex at Chiddingfold, Surrey (Cooper et al. 1984)

19/09/2016 12:15:22

THE SOUTH

century a.d., only courtyard villas and those classified as ‘palatial’ increase in number, representing late Roman developments of earlier establishments, such as at Bignor (Aldsworth and Rudling 1995) and Rockbourne (Hewitt 1968), where significant expansion and investment in these villa complexes occurred (fig. 4.44). A detailed analysis to demonstrate the diversity of villas in the South region is beyond the scope of this chapter, but the largely unparalleled buildingcomplex at Chiddingfold, Surrey, is an example of a site that may have performed a religious rather than secular function (Cooper et al. 1984; Bird 2002; fig. 4.45). The distribution of villas according to their final form varies across the South (fig. 4.46). In broad terms there appears to be comparatively less diversity across the North Downs and its hinterland, where corridor and winged-corridor villa houses dominate the sample. Cottage-style villas are generally rare, at least at sites where they occur without further development, while a few of the larger establishments appear to be located on prominent, riverside locations, such as

fig.

Romanch4.indd 113

113

at Northfleet (Andrews et al. 2011), Darenth (Philp 1973; 1984), and Eccles (Detsicas 1989). In the Wessex sub-region and on the Hampshire Downs and South Downs, villas are generally more varied, with more examples of the smaller, cottage-style and the more elaborate courtyard villas. Scott (2004) has suggested that the investment of wealth may be linked to competitive display on behalf of local elite families/ communities. This is difficult to qualify based purely upon the form of villa houses alone, but the level of aggrandisement found at sites in different landscapes of the South may, to some extent, reflect the degree to which economic success was being invested into the villa building. Other additions to settlements, such as bathhouses, or some of the more elaborate circular, polygonal or octagonal masonry buildings found in the South, also played a prominent role in expressing levels of wealth and the social identity of the occupants, such as at Bax Farm, Teynham (Wilkinson 2011) and the circular temple-mausoleum at Keston (Philp et al. 1991; see also Meates 1979; fig. 4.24b).

4.46.  Distribution of excavated villas according to their final ‘type’ in the South region

19/09/2016 12:15:29

114

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Landscape context and infrastructure topography The location of settlements in the South varied topographically according to type. Farmsteads and roadside settlements are more commonly located on lower-lying ground, particularly the latter, with less than 10 per cent being situated on land higher than 100 m OD. Village settlements in the South tend to show a very different topographic pattern to roadside settlements, with nearly half located at heights above 100 m OD. This variation is largely explained, on the one hand, by the relationship of roadside settlements with river crossings (see above, p. 97 and Ch. 2) and, on the other, by the relatively high elevation of the larger, chalk-downland villages found on Salisbury Plain and in the Wessex hinterland. Villas in the South are located in a range of different landscape zones. Over 11 per cent are found explicitly in coastal locations, such as at Fishbourne, West Sussex, where the high-status complex was situated around the head of Chichester Harbour (Cunliffe 1971). Other villas with prominent coastal locations include Brading (Cunliffe 2013a) and Gurnard Bay (Motkin 1990), both on the Isle of Wight. The majority of villa sites were located in river valleys, with some placed on low-lying land close to the banks of rivers or streams, such as Fullerton, Hampshire (Cunliffe and Poole 2008d), and Snodland, Kent (Ocock and Syddell 1967; Birbeck 1995), while others were situated some way up valley slopes with potential panoramic views, of which Bignor

0 2 4 6 8 10 12 14 16 18 no. of sites

fig.

Romanch4.indd 114

enclosed farmsteads (n=112) 200+ 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

complex farmsteads (n=43)

height above ordnance datum (m)

height above ordnance datum (m)

height above ordnance datum (m)

villas (n=113) 200+ 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

(Aldsworth and Rudling 1995) and Plumpton (Allen 1984) are good examples. Of course, it is very difficult to know for certain that the landscape was a deciding factor in the placement and orientation of villa buildings, but the precise situation of many villa sites is unlikely to have been mere coincidence. At East Wear Bay, Folkestone, Kent, the winged-corridor villa was constructed at the top of a cliff and orientated to overlook the English Channel (Winbolt 1925; Coulson 2013), while the villa at Minster in Thanet would have had a view over the Wantsum Channel and out to sea past Richborough (Perkins and Parfitt 2004). Considerable differences can be observed between the spot height ranges of excavated examples of villas, open farmsteads, enclosed farmsteads and complex farmsteads (fig. 4.47). Few villas are found above 100 m OD, though many are located on elevated ground, particularly those on the chalk downlands. Open farmsteads almost exclusively occupy lower ground, which is, as already observed, reflected by their distributions in the Middle Thames Valley and on the south coast plain. Complex farmsteads also tend towards low-lying ground, with most being recorded 10 km) perhaps suggest that a level of isolation was a desired factor in their location, for example the villas at Fifehead Neville (Oliver 1928), Hinton St Mary (Painter 1964; 1965), and Shillingstone (Corney and Robinson 2007), which are all located on the Stour floodplain in Dorset. The villa at Chiddingfold in Surrey has already been mentioned with regard to its possible religious function (Bird 2002), but its unusual position on a hilltop, along with its location over 13 km from the nearest known road, may further suggest that a level of isolation was a deliberate complex farmsteads

open farmsteads

30

no. of sites

25 20 15 10 5 0

distance from major road/km fig.

4.48.  Distances of villas and farmstead types from the major road network in the South region

70 60 complex farmstead (n=43)

% of sites

50 40

enclosed farmstead (n=112)

30

villa (n=113)

20 10 0 within 2km

within 5km

within 10km

within 15km

within 20km

fig. 4.49.  Distances of villas, enclosed farmsteads and complex farmsteads from a major walled town in the South region

Romanch4.indd 115

19/09/2016 12:15:31

116

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

4.50.  Riverside distribution of villas in north Kent

decision. However, more importantly, perhaps, these villas show that distance from a major road was not an inhibiting factor in their development. The same may also be true of the distance between villas and major walled towns. Less than a quarter of villas are located within 10 km of a major town, and there is no greater association between villas and towns than for other types of farmsteads (fig. 4.49), unlike the situation nationally (see Ch. 12). A slightly higher proportion of complex farmsteads are closer to towns compared with villas and enclosed farmsteads, which, if significant, may be associated with agricultural provisioning. These data are somewhat surprising, however, given the known distributions of villas around some towns, such as Ilchester. The relationship between towns and villas may have been stronger around some urban settlements than others. Although some villas were located at relatively long distances from the major roads, they may well have been connected to the transport infrastructure by trackways or by rivers. Of the villa sites recorded across the South, 22 (19.5 per cent) are found within 200 m of a river or stream and a further 17 (15 per cent) within 400 m; it is noteworthy that many of these sites are located in

Romanch4.indd 116

north Kent, where the rivers drain northwards into the Thames estuary (fig. 4.50). The seemingly close relationship of villas and rivers must at least in part reflect where the best agricultural land was, along with provision of drinking water for cattle, although use of the river as an economic resource is also likely in many cases. Although we do not know the extent of the navigability of the rivers or changes in their courses and how much they might actually have been used for transport, some villas do show direct evidence for the use of the riverside. Northfleet villa in Kent currently lies just under 200 m from the banks of the River Ebbsfleet, due south of its confluence with the Thames, and recent excavations have shown evidence for harbour facilities lying just 10 m from the complex (Andrews et al. 2011). Substantial evidence for malting as well as the recovery of millstones suggests that the Northfleet villa not only used the river to export beer and perhaps bread and/or flour, but also to operate a watermill nearby. More conclusive evidence for the use of streams for milling at villa establishments is found at Fullerton in Hampshire, where the River Anton was canalised to run past the villa house southwards to a mill (Cunliffe and Poole 2008d). This would

19/09/2016 12:15:38

THE SOUTH

117 Enclosure

Limit of excavation

Trackway

Trackway

Field system/ enclosures

Field system/ enclosures 0

200 m 1:5000

fig. 4.51.  Plan of North Bersted, West Sussex (Taylor et al. 2013), showing linear trackway leading to enclosure with field system

have been a considerable investment, implying the extensive processing and export of an arable surplus. Non-villa farmsteads of differing type show little variation in their distance from rivers. Open farmsteads, in particular, are almost all found within 2 km of a river and it is rare to find complex and enclosed farmsteads more than 3 km from a river, though small numbers are located as far as 6 km away. In between settlements, roads and rivers, the landscape would also have been veined by numerous tracks and droveways. Some 298 sites include evidence for trackways ranging across all the main site types, though they have a particular affinity with complex farmsteads, since trackways form part of their classification criteria (see Ch. 2). Generally, the visibility of trackways depends upon the maintenance of drainage ditches on either side of the track, though some routes are defined by hollow-ways, which are formed by gradual erosion of the track surface over time. Trackways can function in the same way as, or in conjunction with, field-systems in applying order on the landscape, by defining boundaries and enforcing the direction of travel. Indeed, Booth (2011a, 6–8) suggests that the evidence for clearly defined trackways could reflect changes in property ownership, agrarian development, increased mobility, and/or new modes of transport. A long linear trackway at North Bersted on the Sussex Coastal Plain can be seen extending from a substantial enclosure, while also forming the primary boundary for a large-scale co-axial field

Romanch4.indd 117

system (Taylor et al. 2013; fig. 4.51). The formalisation of a trackway at the site of Lea Farm near Hurst in Berkshire saw the abandonment of a semi-enclosed settlement and the development of a new complex farmstead (Manning and Moore 2011; fig. 4.52). It would seem that the laying out of new trackways had the potential to influence fundamentally changes and developments in local settlement patterns, as well as reflecting social integration as networking avenues between settlements (e.g. Booth 2011a, 3). Field systems The analysis of field systems in the South increases our understanding of how the population managed and organised land beyond the settlement itself. Unfortunately, fields are poorly understood compared with settlements, because they are generally perceived as being relatively unimportant and tend not to draw the same attention through commercial archaeology or research-led excavations (though see Fowler 2002, 127–60; Chadwick 2007; 2008; Allen 2008). While there has been concentrated, large-scale mapping of field systems on chalk landscapes in the South region, such as the Danebury Environs, Salisbury Plain and the South Downs (Palmer 1984; McOmish et al., 2002, fig. i.1; Carpenter 2008), the excavated evidence for late Iron Age or Roman field systems has been recorded from 201 sites widely dispersed across the South region (fig. 4.53). The distribution of these sites does not completely mirror the distribution of all rural sites

19/09/2016 12:15:39

118

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

0

100 m 1:2000

b

fig. 4.52.  Plan of Lea Farm, Hurst, Berkshire (Manning and Moore 2011), showing late Iron Age enclosed farmstead (a) and an early Roman complex farmstead with trackway (b)

Romanch4.indd 118

19/09/2016 12:15:40

THE SOUTH

fig.

4.53.  Distribution of sites with excavated late Iron Age/Roman field systems in the South region

recorded on the database, and there exist some areas where field systems are poorly represented. Analysis of the spot heights of excavated field systems demonstrates that the vast majority have been identified on low-lying ground, providing a useful comparative dataset to chalk downland surveys. Clusters of sites with field systems appear along the gravel terraces of the River Kennet and the middle Thames, around the Thames estuary, on the South coast plain, and in eastern Kent, and this may be a reflection of the intensity of developer activity in these areas. Some sites are also known from urban hinterlands such as around Chichester, Dorchester and Ilchester. While the low-lying plains and river valleys were host to considerable areas of land division, upland areas, notably on the chalk, were also extensively cultivated and, where the survey evidence has been tested by excavation, this consistently shows use of the field systems during the Roman period (Allen 2008 for Danebury Environs; Carpenter 2008, 26, for the South Downs; Fulford et al. 2006 for Salisbury Plain). Mapping of the field systems west of the villa at Fullerton, Hampshire shows their integration with the two farmsteads at Flint Farm and Rowbury Farm. Both originated in the Iron

Romanch4.indd 119

119

Age, but only Rowbury shows evidence of late Iron Age and early Roman occupation (Crutchley 2008, fig. 2.13). Direct dating of field systems is problematic, relying heavily upon the recovery of pottery, generally heavily weathered from re-working, from the lynchets and field ditches. Assemblages tend to be small when fields are located at a distance from the settlements, where domestic refuse is less concentrated. Pottery from the topsoil can also be useful for dating periods of land use, but less valuable for identifying episodes of change and reorganisation of field systems, though here the molluscan evidence can be very informative, as on Salisbury Plain (Fulford et al. 2006, 12–21 (surface collection), 143–52 (molluscan evidence)). Of the field systems that have been dated to periods of use, distinct chronological patterns can be observed. Across the South region as a whole, the number of sites at which field systems are identified broadly doubles between the late Iron Age and the later first century a.d. This number then steadily decreases through to the fourth century a.d. until it reaches a level last observed in the late Iron Age. Within these data, however, there is considerable intra-regional variation, as is demonstrated by the

19/09/2016 12:15:47

120

THE RURAL SETTLEMENT OF ROMAN BRITAIN

60

London Basin

South Wessex Downs and Wessex Vales

50

no. of sites

40 30 20 10 0 Late Iron Age Late 1stC AD fig.

3rdC AD

4thC AD

4.54.  Number of excavated field systems in the London Basin and the Wessex landscape zones over time

evidence from the London Basin compared with the Wessex Downs and neighbouring Vales (fig. 4.54). A substantial expansion in the number of fields is evident by the later first century a.d. in much of the eastern part of the South region, while the number of sites with fields in the Wessex regions, by contrast, increases only marginally through to the third century a.d., with a slight reduction into the fourth century a.d. The distinctiveness of these patterns is similar to the regional settlement chronologies already discussed and may suggest that the intensity of land division partly correlates with the density of settlement. Variations in field shape and size provide subtle indications of the differing ways in which the land was organised. Numerous small fields perhaps suggest that the land was more intensively managed, while larger fields may indicate a more extensive system of land management (Van der Veen and O’Connor 1998). However, our ability to identify large fields is very dependent on large, open-area excavation, or extensive aerial or geophysical survey. An example of large fields dating to the Roman period has been discovered at the site of Thanet Earth, Kent, where an extensive area of land is divided by a few long field boundaries (Rady 2010). Although the site may have suffered to a degree from truncation by later activity, including cultivation, there is almost no evidence for domestic settlement from the largest part of the excavation, which covered c. 45 ha. Field systems that enclose small pockets of land are more commonly identified than large fields, but these systems can also spread over relatively wide areas. The fields at Long Lane near Ickenham, Greater London, are interpreted as stock pens, though the presence of charred plant remains and quernstones suggests that arable farming was also practised in the vicinity (Lakin 1994). Perhaps an intensive system of land management was in place, involving a rotational system of crop husbandry

Romanch4.indd 120

2ndC AD

alongside seasonal pasturing of livestock to feed off the cereal stubble after harvest. Such an arrangement would also allow for re-fertilisation of the fields through manuring (Van der Veen and O’Connor 1998, 132–4). It is possible that smaller fields and more intensive land management regimes reflect denser populations of people with limited availability of land. Another system of small, co-axial fields has been identified at the roadside settlement site at Syon Park, Brentford (Cowie et al. 2013), where small fields and a droveway connected the road and the riverside, allowing people and livestock access to the water but with restricted movement on the adjacent land (fig. 4.55). While there appears to have been a great deal of variation in the presence, form, and size of fields across the South, the overall reduction in the number of sites with field systems between the first and fourth century a.d. may indicate that larger, but less archaeologically visible, fields were becoming more common over time, not that less land was under cultivation. However, this appears not to have been the case in the Wessex region, which shows evidence for gradually increasing numbers of sites with fields up to the third century. It is rare to observe changes in field arrangements at individual sites, where large, open areas of excavation are required. However, Bestwall Quarry, Dorset, provides a relevant example (Ladle 2012). Probably better known as a production site of SE Dorset Black-Burnished ware (BB1), excavation plans of the surrounding area reveal a constantly changing landscape, where a closely divided, late Iron Age/early Roman field system gradually became more open in the third and fourth centuries. Just how far this type of change occurred elsewhere is unknown, but the reduction in the number and spacing of field boundaries may partly account for the reducing number of sites with evidence for fields across the South region.

19/09/2016 12:15:48

THE SOUTH

Roman Road

121

Oven

Roman buildings

Graves Fields

Trackway

Palaeochannel

0

100 m 1:2000

fig. 4.55.  Plan of Syon Park, Brentford (Cowie et al. 2013), showing co-axial field system and trackway between the road and palaeochannel

Settlement hierarchies: the social and economic basis of settlements The settlement and landscape evidence has, so far, shown an extraordinary level of intra-regional and chronological variation across the South region, with clear differences between the eastern and western halves. To determine how much of this variation was rooted in the social and economic activities of the inhabitants of these settlements, we must take account of the considerable wealth of material culture and environmental remains that has been generated from excavations. These data are analysed here at both the regional and sub-regional level to broadly characterise the different types of settlements that have been considered in this chapter. Material culture Pottery Pottery is by far the most ubiquitous type of artefact recovered from sites in the South region. Unfortunately, the number of sites where the pottery assemblage has been quantified varies between different types of settlement. Basic sherd counts of pottery fragments are reported from 69

Romanch4.indd 121

per cent of farmsteads, 63 per cent of villages, 57 per cent of roadside settlements, and only 37 per cent of villas, and even fewer site reports include the weight of the pottery. The low proportion of villas is undoubtedly due to the number of pre1990 excavations, when it was perhaps considered less important to directly compare material culture assemblages from different sites. There is also a lack of quantification of particular types of pottery. Less than 50 per cent of villas that have produced samian also provide a corresponding sherd count, though this rises to 68 per cent of farmsteads. Despite the obvious improvements that could be made in the reporting of pottery assemblages, a considerable dataset does exist, allowing for intersite type comparisons of pottery consumption to be made. Overall, roadside settlements and villages produce the greatest quantity of samian ware, averaging 451 and 325 sherds per site respectively. These compare to site averages of 138 sherds from villas, 76 sherds from complex farmsteads, and 46 sherds from enclosed farmsteads. The discrepancy between different types of site is also mirrored by the average number of mortaria sherds. Again, roadside settlements and villages are the most productive, averaging 218 and 106 sherds respectively, while villas produced an average of 38 sherds, complex farmsteads 19 sherds, and

19/09/2016 12:15:49

122

THE RURAL SETTLEMENT OF ROMAN BRITAIN

enclosed farmsteads 21 sherds. The much higher average sherd counts from roadside settlements and villages no doubt reflect their status as nucleated settlements, where greater numbers of people would have been using and depositing pottery. If some nucleated settlements also functioned as market centres, then this too would potentially account for higher sherd counts, since greater numbers of pottery vessels would more often be circulated around the settlement. Some of the differences between settlement types may be partly influenced by chronological factors. A large number of enclosed farmsteads are late Iron Age in date, and many were abandoned before samian was widely circulated in the south region. The similarity of the average sherd counts of mortaria between complex and enclosed farmsteads, however, suggests that the adoption of some Roman wares was more equal. Variation in the consumption of samian ware may partly reflect social and economic status, particularly at villas compared to farmsteads, if samian ware was more greatly valued (see Willis 2011). This is likely if we consider the quantity of samian as a percentage of the overall pottery assemblage from different sites, which is rarely above 2–3 per cent at most sites. Small finds Overall, sites across the South region tend to be relatively productive in terms of artefacts other than pottery. Yet within individual assemblages, there exists considerable variation in the frequency of occurrence of different major classes of finds (fig. 4.56). Coins stand out as the most commonly occurring class, being represented on almost 50 per cent of sites, and they are followed by foodprocessing items (querns and millstones), brooches, knives and other tools, and items

explicitly associated with textile-processing (e.g. loomweights, spindlewhorls, sewing needles, wool combs and weaving boards). Thereafter, an ‘intermediate group’ of finds is found on around 20 per cent of sites. This includes items of personal adornment, such as finger rings, hairpins, and bracelets, as well as toilet instruments such as nail cleaners and tweezers. The least frequently found artefacts are religious objects, writing equipment and lighting equipment. The ordering of commonly found artefacts (e.g. coins, querns, brooches, etc.) against infrequently recovered artefacts (e.g. lighting equipment, religious objects, etc.) is generally the same in each landscape zone in the South, though with some variation in their relative frequencies. The South Wessex Downs stands apart from the London Basin and the North Kent Plain, for example, with most artefact classes being more commonly recovered (fig. 4.57). Exceptions to this are coins and hairpins, which occur in relatively equal numbers on both North Kent Plain and South Wessex Downs sites. Variation in frequency of occurrence can also be observed by examining individual categories of artefact across all landscape zones. For example, agricultural tools and foodprocessing objects are more frequently reported from sites on the Hampshire Downs and on the South Wessex Downs, though the differences between these and other landscape zones is less marked for food-processing items (fig. 4.58). Finger rings, on the other hand, are better represented on sites on the South Wessex Downs and in the Wessex Vales, rather than on the Hampshire Downs, yet weighing objects such as steelyards – mostly used for trade and exchange – are slightly more commonly found on sites in the Wealden Greensand, where few finds are generally

% of sites with object present (n=897)

50 40 30 20 10 0

fig.

Romanch4.indd 122

4.56.  Frequency of major artefact categories on all sites in the South region

19/09/2016 12:15:51

THE SOUTH

recovered. These intra-regional variations between different categories of artefact are unlikely to be explained by factors relating to preservation and visibility alone, because they are made of similar materials. Rather, they may reflect real differences in the frequency or intensity of certain activities or occupations, as well as in fashion, in different areas. Clear variation between broad classes of rural settlements can also be observed through the percentage of each producing different classes of finds (fig. 4.59). In general, roadside and village settlements produce most types of artefacts, while

% of sites with object present

70

London Basin (n=221)

123

farmsteads tend to be the least productive class of site. In the case of coins, villas behave similarly to roadside settlements and villages, but they are more like farmsteads where items such as brooches or artefacts associated with textile-processing are concerned. Household items, knives/tools and toiletry objects are well represented on roadside settlements, while items of personal adornment such as hairpins and bracelets tend to be well represented on villas. Artefacts of a more utilitarian nature, however, such as food-processing (quernstones) and textile-processing (needles,

North Kent Plain (n=91)

South Wessex Downs (n=90)

60 50 40 30 20 10 0

fig.

4.57.  Frequency of major artefact categories recovered across selected landscape zones in the South region

20 15 10 5 0

% of sites with object present

25

food-processing (querns/millstones) 60

% of sites with object present

% of sites with object present

agricultural tools 30

16 14 12 10 8 6 4 2 0

50 40 30 20 10 0

weighing objects

% of sites with object present

ger rings

fig.

Romanch4.indd 123

35 30 25 20 15 10 5 0

4.58.  Frequency of selected categories of artefact on sites in selected landscape zones in the South region

19/09/2016 12:15:55

124

THE RURAL SETTLEMENT OF ROMAN BRITAIN

roadside settlement (n=22)

% of sites with object present

100

village (n=11)

villa (n=75)

farmstead (n=264)

80 60 40 20 0

fig.

4.59.  Frequency of major artefact categories on different types of rural settlement in the South region

villa (n=75)

complex farmstead (n=36)

enclosed farmstead (n=72)

open farmstead (n=10)

% of sites with object present

100 80 60 40 20 0

fig.

4.60.  Frequency of major artefact categories on villas and farmstead types in the South region

loomweights, spindlewhorls, etc.) items stand out on farmsteads. One explanation for the difference in the representation of utilitarian objects between villas and farmsteads is that excavations of the former have tended to concentrate on the residential buildings rather than on peripheral structures and areas where utilitarian activities may have been carried out (fig. 4.60). There is comparatively little variation between farmsteads of different types, though a greater number of complex farmsteads tend to produce items of personal adornment, such as brooches, bracelets and rings, compared with their enclosed and open counterparts. Other classes of finds, however, are very poorly represented on all farmsteads in comparison to villas, such as security items (locks and keys), writing equipment (styli), recreational

Romanch4.indd 124

items (dice, game counters, etc.) and religious objects (clay pipe figurines, lead tablets). When these classes of artefacts are recovered on farmsteads they often come from burial contexts, such as a late first century a.d. inhumation from Northumberland Bottom near Gravesend, which contained 23 gaming pieces, an associated board and two bone dice, as well as other high-status items (Allen et al. 2012). Such evidence emphasises the social importance and value placed upon items such as these. Writing equipment is especially rare from farmsteads, the recovery of three styli from the enclosed settlement at Alington Avenue in Dorset being a notable exception (Davies et al. 2002), suggesting that much of the rural population remained illiterate and had little need for writing. This site is very close to Dorchester, so it is perhaps not typical of rural settlements in general.

19/09/2016 12:15:57

THE SOUTH

Environmental Remains Animal bones The South region can also be characterised through its environmental record, particularly of its zooarchaeological and archaeobotanical assemblages, both of which have seen a significant number of assemblages being recorded over the past 25 years. The animal bone dataset includes 266 assemblages with at least 100 identified specimens (NISP) from 183 sites. Where possible these assemblages have been separated into chronological phases of: late Iron Age (30 assemblages), late Iron Age/early Roman (41), early Roman (48), middle Roman (25), and late Roman (68). When assemblages cannot be assigned to date in this way, they are simply phased as ‘Roman’ (54). Smaller assemblages are recorded on the database and are available through the online resource, but are omitted from the analysis in this chapter. The assemblages included represent a considerable dataset for rural settlement in the South region (see Maltby 2010 for an overview of urban data). However, it must be noted that the distribution of faunal assemblages is very uneven and is dominated by the number from sites on the South Wessex Downs (26 per cent), the London Basin (24 per cent), the Hampshire Downs (17 per cent), and the North Kent Plain (14 per cent). This distribution is primarily due to favourable survival on the chalk downland and in the alluvial river valleys. The Weald, by contrast, is almost completely devoid of faunal assemblages, firstly because settlement evidence is sparse and, secondly, the underlying clay soils are more acidic, providing poor preservation conditions for most organic remains. The concentration on the southern part of the Hampshire Downs is primarily due to the exceptionally large faunal assemblage from the settlement at Owslebury, which spans the middle Iron Age to the end of the Roman period (see Collis 1968; 1970, for overviews of the site and Maltby 1987 for the animal bone report). Another assemblage from a rural settlement at Oakridge II, Hampshire, close to Owslebury, is also exceptionally large, consisting of nearly 25,000 fragments from an estimated 840 animals of 38 different species (Maltby 1994). However, this group of animal bones derives exclusively from whole or partial skeletons deposited into the shaft of a single well. A simple comparison of these two faunal assemblages exemplifies the greatest problem affecting any synthesis of animal bone data. The composition of each assemblage is context-specific: it can reflect one or a range of cultural practices; it will be influenced by postdepositional processes; and each must be

Romanch4.indd 125

125

interpreted accordingly. Just how far the placement of animal body parts into the well at Oakridge II reflects ‘normal’ consumption patterns is, of course, highly questionable. In order to counter some of these issues, assemblages that appear to be anomalous are omitted from broad surveys of the type undertaken here, but are highlighted to demonstrate their individual importance. In the areas where faunal assemblages are more common, there exists considerable variation in cattle, sheep/goat and pig relative frequencies, but the general trends are brought into sharper focus when the data from farmsteads located in the London Basin, the Hampshire Downs and the South Wessex Downs are viewed together (fig. 4.61). Of the 29 farmsteads located in the London Basin, 24 include assemblages that produce 50 per cent or more cattle remains. On farmsteads on the Hampshire Downs this drops to only 8 assemblages out of 32, and on the South Wessex Downs to only 4 out of 25 assemblages. Of the eight farmsteads in the Hampshire Downs with higher cattle proportions, six are located close to one another in the Basingstoke area due south of Silchester. Allowing for such exceptions there is an otherwise clear difference between the higher proportions of cattle in the London Basin compared to a dominance of sheep/goat on the chalk downlands of Hampshire and South Wessex. While pig remains are generally uncommon on most farmsteads compared to cattle and sheep/ goat, this is particularly true of sites on the South Wessex Downs south to Dorset where they generally account for less than 10 per cent of the assemblage. In this context, the discovery of high proportions of pig remains, particularly of cranial elements and teeth, at the nucleated settlement of Ower on the Dorset coast is intriguing. Linking to the evidence for salt production at the site, Maltby (2006, 119–20) has drawn attention to the possibility that the remains represent the processing of pigs, with cured bacon and other pork products then being transported out of the settlement. While we can see differences in representation of the main domestic taxa between different landscape zones in the South, irrespective of the type of settlement, when we come to consider assemblages from the different types of settlement, disregarding the sub-regional context, we find that the relative frequencies of major livestock taxa are remarkably similar (fig. 4.62). Only at villa and roadside settlements are cattle noticeably more frequent overall compared to sheep/goat, with pig bones being more frequent at villa sites, a feature commonly cited as reflecting a preference for pork among higher-status or more ‘Romanised’ sites (e.g. King 1991). Assemblages with the highest frequencies of pig bones come from the early Roman palace at

19/09/2016 12:15:57

Romanch4.indd 126

Poundbury Farm

Fordington Bo…om (LIA/ER)

Fordington Bo…om (ER)

Tolpuddle Ball (LR)

Maiden Castle Road

Tolpuddle Ball (LIA/ER)

Suddern Farm (ER)

Whitcombe (LR)

Grateley South (LIA)

Flagstones

Grateley South (ER)

Durrington Walls

Figheldean (LIA/ER)

Beddington

Nazeingbury

Eton Area 16 (ER)

Hengrove Farm

Eton Area 16 (LIA)

Innova Park, Enfield

Marnel Park (LR)

North Shoebury

Thorpe Lea Nurseries (LIA)

Coldharbour Quarry

Laleham, Spelthorne

West Thurrock

Kingsmead Quarry (LR)

Wood Lane, Slough

Lake End Road West Marnel Park (ER)

Maldon Road, Colchester

Reading Business Park

Stra‘ord Market Depot

Thorpe Lea Nurseries (LR)

Lincoln Road, Enfield

Thorpe Lea Nurseries (ER)

Runfold Farm

Kingsmead Quarry (LIA/ER)

Manor Farm, Horton

Foxholes Farm

Pingewood

Moor Hall Farm Ship Lane, Aveley

mean %NISP

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

Whitcombe (LIA/ER)

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

Stonehenge Area C1

Upavon Airfield

Teffont Evias

Rotherley

Fordington Bo…om (LR)

High Post, Salisbury

Suddern Farm (LIA)

Suddern Farm (LR)

Alington Avenue

Old Kempsho Lane Viables Farm Ructstalls Hill (LR) Park Prewe Hospital Ashley Cowdery's Down (ER) Brighton Hill South Overton Winnal Down Thruxton (LR) Ructstalls Hill (ER) Woolbury Thruxton (ER) Houghton Down (LIA) Balksbury Camp (LR) Oakridge II/IV Dunkirt Barn Micheldever Wood Houghton Down (ER) Thruxton (LIA) Rowbury Farm (ER) Ne lebank Copse Bloswood Lane Cowdery's Down (LIA/ER) Fir Hill, Bossington Rowbury Farm (LIA/ER) Old Down Farm Ructstalls Hill (LIA) Balksbury Camp (LIA) Houghton Down (LIA/ER) Kennel Farm Balksbury Camp (ER) Twyford Down

mean %NISP

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

Beach's Barn

Figheldean (LR)

mean %NISP

126 THE RURAL SETTLEMENT OF ROMAN BRITAIN

London Basin

pig

sheep/goat

ca™le

Hampshire Downs

pig

sheep/goat

ca le

South Wessex Downs

pig

sheep/goat

ca…le

fig. 4.61.  Relative frequency of major livestock taxa from farmsteads in the London Basin, Hampshire Downs and South Wessex Downs landscape zones

19/09/2016 12:15:58

THE SOUTH

Fishbourne, and the later villas at Lullingstone and Liss. At the latter site, it was suggested the high proportion of pig could imply a local focus on pig husbandry, with feasting being indicated in some features (Hamilton-Dyer 2008). Pig bones are also recovered in greater quantities at sites where a religious focus is clear, such as at the temples at Hayling Island and Chanctonbury Ring, both of which provided evidence for the very specific selection and deposition of pig carcass parts, suggesting that pork meat played a crucial role in ritual feasting practices at these sites (King 2005, 337–44). After the three major livestock taxa, equids are the next most common mammals represented in most faunal assemblages. While mules and donkeys have been identified in Romano-British assemblages (see vol. 2), the vast majority of bones are likely to derive from horses (Johnstone 2008). Horse bones are generally better represented on farmstead and village sites than on roadside settlements and villas. High proportions of horse have been recovered from a few farmsteads such as Copse Farm, West Sussex (Bedwin and Holgate 1985), while evidence for horse-breeding is also becoming increasingly recognised from the bones 60

127

of immature animals found on farmsteads. Dog bones are regularly found on all types of site, but generally in small quantities, while cat bones are sparsely represented, though sometimes identified as whole or partial burials in prominent features, such as the six animals placed in the well at Oakridge II (see above, p. 127). The bones of wild mammals are equally rare, but there appears to be a significant difference between the frequencies of deer and, to a lesser extent, hare bones recovered from villas compared with other types of site. Antler is frequently found on most types of site where it may have been collected after being shed from stags and bucks, though the identification of butchered deer bones on numerous villa sites suggests that hunting may have been a regular pastime of local elites (Allen 2015). Plant remains In total, 324 phased assemblages of plant remains have been recorded from 284 sites in the South. The remains of spelt wheat dominate most cereal assemblages and are the most frequently identified of all cultivated plant taxa across the South (fig. 4.63). Barley and oats are also relatively common,

cale

sheep/goat

pig

enclosed farmstead n=54

complex farmstead n=26

farmstead (all) n=132

mean %NISP

50 40 30 20 10 0 village n=16

fig.

open farmstead n=7

villa n=37

roadside selement n=40

4.62.  Inter-site type comparison of mean relative frequencies of major livestock taxa in the South region

% of assemblages present (n=324)

100 80 60 40 20 0

fig. 4.63.  Percentage presence of cultivated plant taxa on sites in the South region where archaeobotanical remains have been identified (includes both charred and waterlogged assemblages)

Romanch4.indd 127

19/09/2016 12:16:01

128

THE RURAL SETTLEMENT OF ROMAN BRITAIN

% of assemblages present

100

North Kent Plain (n=40)

South Coast Plain (n=31)

80 60 40 20 0

100 % of assemblages present

London Basin (n=86)

Hampshire Downs (n=35)

South Wessex Downs (n=32)

80 60 40 20 0

fig. 4.64.  Percentage

presence of cultivated plant taxa on sites in selected landscape zones in the South region (includes charred and waterlogged remains)

though barley tends not to dominate assemblages as much as spelt wheat, while oats are rarely found in any quantity and are commonly referred to in specialist reports as a ‘weed’ growing alongside the main wheat or barley crops (see vol. 2 for more detailed analysis). Oats may well have been selected for livestock foddering, and therefore do not feature as much in assemblages that reflect crops intended for, or disposed after, human consumption. Livestock foddering was the interpretation of a proportionally high representation of oats at Innova Park, near Enfield, where a small settlement with a droveway and enclosures were identified on the banks of the River Lea (Ritchie 2008). Less frequently recovered cereal taxa include emmer wheat, freethreshing wheat (such as club wheat and bread wheat), and rye. The cultivation of emmer wheat is generally considered to have been common in prehistory, but replaced by spelt wheat in many areas during the Iron Age (M. Jones 1981). Some late Iron Age/early Roman assemblages in the South suggest that a few settlements continued to place an emphasis upon emmer. Although it seems clear that spelt wheat was the dominant cultivar of the Roman period in the South, identifying the relative abundance of different plant taxa is fraught with difficulties. Glume wheats such as spelt and emmer tend to be well represented because the chaff is largely composed of glume bases, which are removed at a relatively late stage of processing, while the chaff

Romanch4.indd 128

of free-threshing cereals, such as barley and bread/ club wheat, is largely represented by rachis, which are removed earlier in the processing sequence (Hillman 1981; Van der Veen and Jones 2006, 219). Increased use of radiocarbon dating is now also demonstrating contamination from later farming practices (e.g. medieval) (Pelling et al. 2015). With these issues in mind it is worth observing the intra-regional differences seen across the South in terms of the relative presence of cultivated plants in botanical assemblages (fig. 4.64). Examination of the data across the five landscape zones with the most sites (London Basin, South Coast Plain, North Kent Plain, the Hampshire Downs and the South Wessex Downs) shows that, at about 80 per cent, the percentage of assemblages in which barley is identified varies very little. Beyond the ubiquitous cereals of spelt and emmer, however, some considerable variation can be identified in lesser-grown crops, which appear to have been favoured, or less favoured, in different zones. In the London Basin, rye is identified in nearly 20 per cent of assemblages but in no more than 6 per cent of assemblages in any other of the main landscape zones. On the South Coast Plain, free-threshing wheat (c. 45 per cent), pulses (c. 52 per cent) and flax (c. 20 per cent) are comparatively well represented, while on the North Kent Plain, emmer wheat (c. 58 per cent) and fruits (c. 33 per cent) stand out as being better represented compared to other sub-regions.

19/09/2016 12:16:03

THE SOUTH

% of assemblages present

% of assemblages present

villa (n=29)

farmstead (all) (n=194)

129

roadside selement (n=35)

100 80 60 40 20 0

100

complex farmstead (n=35)

enclosed farmstead (n=58)

open farmstead (n=9)

80 60 40 20 0

fig. 4.65.  Percentage presence of cultivated plant taxa on different site types in the South region (includes charred and waterlogged remains)

Most horticultural crops such as vegetables and fruits are likely to be under-represented because they require waterlogged or other favourable conditions, such as mineralisation, for preservation (Van der Veen et al. 2007). Fruit seeds are best represented at villa sites, though a slightly higher proportion of assemblages from roadside settlements have fruit remains (as well as herbs and nuts) compared with all other farmsteads (fig. 4.65). Four pit or well features dating from the late third century to the mid-fourth century a.d. at the roadside settlement of Neatham, Hampshire, produced the remains of cherries, plums, walnut, and coriander (Murphy 1986, 149–50). Apple and bullace may have been cultivated, but edible wild fruits, including sloe, blackberry, elder, rose, hawthorn and hazel were also abundant in the samples. At the Tobacco Dock site, Shadwell, on the eastern outskirts of London, remains of fig, elderberry and carrot were recovered from late third- and fourth-century a.d. waterlogged deposits (Branch 2011, 141–3). As well as distinctions that may be related to cultural choice or differing degrees of affluence, such as the wider range of fruits detected at villas, there also appear to be chronological factors at play. The higher proportion of open farmsteads with emmer wheat tend to be settlements at which occupation had ended by the beginning of the Roman period, while the later Roman occurrence of a greater representation of fruits and vegetables at roadside settlements perhaps suggests that such

Romanch4.indd 129

foods had become more common in diets by the third and fourth centuries a.d. The combination of the analyses of aspects of the material culture, the zooarchaeological and the botanical data from settlements in the south region complements and enriches the differentiation of types of settlement on the basis of their ground plan, at the same time as revealing change over time. Case Studies: Middle Thames Valley and the Hampshire Downs During the late Iron Age, the Middle Thames Valley contained a number of small, open settlements located on the flood plain. Many of these appear to have been swiftly abandoned and replaced by new, more complex, settlement forms and co-axial field systems during the later first century a.d. These changes partly coincide with the construction of the main London to Silchester road and the development of the important roadside settlement at Staines (Pontibus) by a crossing point of the Thames. Staines is located c. 43 km east of Silchester, and c. 30 km west of London. Using data from all the excavated settlements in this sub-region, these changes in the rural landscape are argued here to reflect intensification in the production and marketing of domestic livestock. A very different pattern of settlement development is observed on the chalk downs of Hampshire. Here, small, enclosed

19/09/2016 12:16:05

130

THE RURAL SETTLEMENT OF ROMAN BRITAIN

farmsteads were prominent during the late Iron Age and early Roman periods, perhaps involved in relatively small-scale arable and pastoral farming. By the later Roman period, villas were more common, alongside new types of field system and a proliferation of corn-drying structures, found at a range of different settlements. Together, the evidence suggests that an intensification of arable farming occurred in this sub-region. An important distinction between the two case study areas is that the Middle Thames Valley has been mainly investigated through developerfunded work, while the Hampshire study area is heavily dominated by the research of Barry Cunliffe on the Danebury environs (Cunliffe and Poole 2008a). However, it is the purpose of these case studies to show that these two sub-regions clearly contrast in terms of the nature and timing of the changes that took place, with each appearing to show very different developments in farming techniques and strategies. The Middle Thames Valley The Middle Thames Valley case study area covers approximately 300 km², containing 42 site records

fig.

Romanch4.indd 130

on the project database, though fourteen of these relate to individual excavation records from the roadside settlement at Staines (fig. 4.66). In total 25 farmsteads are represented, twelve of which have evidence for associated field systems, while three records relate specifically to field systems. Excavated villas are completely absent in the subregion. One funerary site is also present, a small cremation cemetery at Prospect Park, Harmondsworth, which lies close to the junction of two Roman roads and may relate to nearby roadside settlement (Farwell et al. 1999). Geographically, the area focuses on the gravel terraces of the Rivers Thames and Colne, with most of the settlements being located within close proximity of one of these rivers or their tributaries. The number of farmsteads in use in the subregion remained fairly static between the late Iron Age and the third century a.d., with only a slight drop in the fourth century. However, a number of open farmsteads were either abandoned or had changed in form by the later first century a.d., at the point when an increasing number of complex farmsteads were developing. The open settlements were quite similar in plan, generally consisting of

4.66.  Distribution map of sites in the Middle Thames Valley case study area

19/09/2016 12:16:11

THE SOUTH

of settlement expansion and increased complexity can be observed at the sites of Kingsmead Quarry, Horton (Chaffey 2009), Cranford Lane, Harlington (Elsden 1996), and at Hengrove Farm, due south of Staines (Poulton 2007). The latter expanded in the later first century a.d. into a large complex complete with multiple enclosures and trackways surrounded by field systems. Together, these developments demonstrate a repeated pattern of considerable settlement reorganisation during the early Roman period in this sub-region (fig. 4.68).

small clusters of timber-built roundhouses, sometimes with four-post structures, while others included small enclosed areas within the main settlement (e.g. Hengrove Farm (Poulton 2007); Ashford Prison (Carew et al. 2006); Imperial College Sports Ground (Crockett and Nowell 1998); fig. 4.67). A number of these gravel terrace sites show good evidence for significant reorganisation of the settlement, with some transforming into complex farmsteads, or else for replacement by rectilinear field systems. Examples a

131

b

c

Neolithic Monument

0

200 m 1:5000

fig. 4.67.  Plans of three late Iron Age open farmsteads in the Middle Thames Valley at Hengrove Farm, Staines, Surrey (Poulton 2007) (a), Ashford Prison, Spelthorne, Surrey (Carew et al. 2006) (b), and Imperial College Sports Ground, Harlington, Greater London (Crockett and Nowell 1998) (c)

a

b

c

0

200 m 1:5000

Palaeochannel

fig.

4.68.  Plans of three Roman-period complex farmsteads in the Middle Thames valley at Hengrove Farm, Staines, Surrey (Poulton 2007) (a), and Cranford Lane, Harlington, Greater London (Elsden 1996) (b), and Kingsmead Quarry, Horton, Berkshire (Chaffey 2009) (c)

Romanch4.indd 131

19/09/2016 12:16:13

132

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Late Iron Age/early Romano-British Iron Age settlement Bronze Age field system Settlement

Eastern field systems

0

200 m 1:5000

fig. 4.69.  Plan of Heathrow Terminal 5 showing middle Iron Age open settlement and LIA/ER complex settlement and field system (Lewis et al. 2010)

Alongside settlement expansion, the introduction of co-axial field systems implies considerable changes to the way in which land and resources were managed. Although these changes accompany the emergence of the Thames gravel complex settlements, evidence for fields predating the conquest have been identified at Heathrow Terminal 5 (Lewis et al. 2010) and at Cippenham, Slough (Preston 2012). However, during the first century a.d. there was a substantial increase in the frequency of sites with evidence for field systems. In some cases, the evidence clearly demonstrates a post-conquest origin for the field boundaries, with, on occasion, some field systems expanding over later Iron Age farmsteads, indicating significant changes in local settlement and land-use patterns. Assessment of the extent of these new field systems is impossible without large-scale excavation or partially excavated cropmark evidence (for example, at Mayfield Farm, East Bedfont: Jefferson 2003). Perhaps the best example was revealed during the excavations at Heathrow Terminal 5, which covered over 75 ha (Lewis et al. 2010; fig. 4.69). Here, a middle Iron Age open settlement, consisting of numerous roundhouses and associated enclosures, was either developed or abandoned and reoccupied by a complex farmstead during the late Iron Age/early Roman period. At the same time, a rectilinear field system covering several hundred square meters to the east of the settlement was laid out on a different alignment over an earlier Bronze Age field system. At Ashford Prison, Spelthorne, a similar field system dating to

Romanch4.indd 132

Roman Iron Age Bronze Age

0

200 m 1:5000

fig. 4.70.  Plan of Ashford Prison, Spelthorne (Carew et al. 2006)

the later first century can also be seen cutting through a late Iron Age open settlement (Carew et al. 2006; fig. 4.70), and it is quite possible that the change in land use found at this site was related to the development of the large complex farmstead at Hengrove Farm located only 400 m to the north. In addition, at Wood Lane, Slough, an early Roman field system extends from a later Iron Age enclosed settlement which remained in use into the first half of the third century a.d. (Ford 2003). Alongside Heathrow Terminal 5, this site is one of the few examples of settlements in the sub-region that continued across the late Iron Age/Romano-

19/09/2016 12:16:16

THE SOUTH

British transition, but which incorporated new forms of land division instead of being replaced by them. Evidence for later Roman developments is more limited, though the construction of enclosure complexes around large droveways is identified from third- and fourth-century a.d. phases at Imperial College Sports Ground and at Heathrow Terminal 5 (fig. 4.71). These types of site are reminiscent of ‘ladder complexes’, but they are not evident in this sub-region prior to the third century a.d. The droveway at Imperial College Sports Ground measured over 40 m across, while the Heathrow example was around 100 m wide, positioned between a series of linear enclosures and a possible field system. Both of these routeways must have been able to support substantial traffic if required. Domestic activity was identified at Imperial College Sports Ground from late Roman timber-lined wells and deposits of pottery and animal bone (Crockett and Nowell 1998), but it appears to have been more elusive at Heathrow (op. cit.), where settlement evidence was found to the east of the droveway complex, perhaps suggesting that it was primarily used for the trafficking and corralling of large numbers of livestock. It is notable that evidence for arable processing varies across the case study area. Only relatively recently have remains of small corndryers been identified at Coldharbour Quarry, Thorpe (Margetts and Robertson 2013), and Eton Area 16 (Allen 2013), both of which contained quantities of charred grain. Evidence for cereal processing was also identified at Kingsmead Quarry, Horton (Chaffey 2009, 59–60). However, evidence for an increase in arable farming at Heathrow Terminal 5 in the late Iron Age/early Roman period, associated with the imposition of the rectilinear field system, is partly attributed to the production of fodder crops for livestock consumption. Carruthers highlights the introduction of cultivated oats in the late Iron Age as a ‘significant advance’, being well suited to acidic, damp soils, and the increasing proportion of barley in the Roman period as being linked to the increased use of horses (Carruthers 2010, 35). In addition, an increase in grass pollen in the mid– late Iron Age perhaps reflected an intensification of hay production, which lasted through to the end of the Roman period, while evidence for the storage of hay was identified in early–mid-Roman deposits and was argued to have supported livestock through the winter (ibid. 34). Archaeobotanical samples were analysed from a further eleven sites in the case study area, all of which were lacking in cereal remains, with no evidence for arable processing being identified.

Romanch4.indd 133

133

a

N

Movement

0

200 m 1:5000

b

Movement

Late Romano-British enclosures Early - mid Romano-British Late Iron Age/early Romano-British fig. 4.71.  Plans of later Roman droveway complexes in the Middle Thames Valley at Heathrow Terminal 5 (Lewis et al. 2010) (a) and Imperial College Sports Ground, Harlington (Crockett and Nowell 1998) (b)

On the whole, the considerable evidence for trackways, enclosures and waterholes on many of the sites in the case study area suggests a greater emphasis upon pastoral farming, with a focus on the exploitation of the extensive pastures available on the Thames gravel terraces. The extent to which arable and pastoral farming featured in the local agricultural economy may also be partially measured by the evidence for processing and consumption of agricultural remains at Staines, assuming that the roadside settlement became a local market centre. Unfortunately, only one archaeobotanical assemblage of substance has, so far, been analysed and reported on in any detail from Staines, that from the Former Central Trading Estate site (McKinley 2004). The largest quantity of charred

19/09/2016 12:16:18

134

THE RURAL SETTLEMENT OF ROMAN BRITAIN

cereal remains from the site was recovered from the floor of a second-century a.d. masonry structure suggesting that it was, at least for some part of its history, used as a granary (ibid. 43). The site also produced remains of bread wheat, spelt, emmer, barley and rye, indicating the importation of a range of crops to the settlement, since the location of Staines itself and the nature of the soils would have been unsuitable for the cultivation of these crops. However, some evidence for local cultivation of plants was provided by the remains of peas, beans, strawberries, elderberries and Prunus sp. (sloes, cherries, plums, etc.) recovered from waterlogged deposits, and it is possible that a local ‘market garden’ economy prospered in Staines during the early phases of the settlement. The identification of fig seeds is also evidence for longer distance trade networks (ibid. 48). Although the identification of a granary represents the storage of surplus cereal production, there is no evidence for large-scale cereal processing at Staines, as suggested for other roadside settlements in Britain (e.g. Springhead, Kent: Andrews et al. 2011). At present, there is a total lack of evidence for corndryers from the settlement, despite a relatively large number of sites being excavated. This correlates with the paucity of grain-drying structures noted from the rest of the

table

4.4: proportions

case study area. The local landscape and the availability of meadowland appears more suited to the grazing of livestock than large-scale cereal production. Indeed, Bird has suggested that Staines may have functioned as a centre for cattle trading, ultimately serving London (Bird 1996, 224; see also Lewis et al. 2010, 298). The faunal assemblages recovered from sites in Staines and from farmsteads within the case study region are generally dominated by cattle bones, though the samples vary between sites (table 4.4). For example, higher frequencies of sheep/goat remains have been recovered from the rural settlements at Hengrove Farm and Eton Area 16, and they were notably abundant from the Elmsleigh Centre site in Staines, particularly from early Roman deposits. Pig bones are generally better represented at sites in Staines compared to farmsteads, often at the expense of sheep/goat remains, though again the data demonstrate the variability of the samples even within the roadside settlement. Significantly, cattle bones from all of the assemblages recovered from Staines show evidence for distinctive cleaver butchery marks (cf. Chapman 1984, 115–17; McKinley 2004, 29). Cattle scapulae frequently showed evidence for multiple chops around the fossa and trimming marks along the spine, indicative of meat-filleting,

of the three main domesticates from sites in the middle thames valley case

study area, divided between farmsteads and staines (pontibus)

Farmsteads Thorpe Lea Nurseries, Egham Eton Area 16 Kingsmead Quarry, Horton Manor Farm, Horton Thorpe Lea Nurseries, Egham Eton Area 16 Lake End Road West Kingsmead Quarry, Horton Thorpe Lea Nurseries, Egham Coldharbour Quarry, Thorpe Wood Lane, Slough Laleham, Spelthorne Hengrove Farm, Staines

Phase Late Iron Age Late Iron Age LIA/ER Early Roman Early Roman Early Roman Middle Roman Late Roman Late Roman Late Roman ‘Roman’ ‘Roman’ ‘Roman’

Total NISP  183  756  210  242  155 1317  116  242  317  107  251  136  494

%cattle 52.5 47.8 58.6 77.3 72.9 43.4 62.1 75.6 70.7 57.0 59.0 57.4 46.8

%sheep/goat 33.3 43.5 33.3 14.9 21.9 49.2 34.5 19.8 26.2  6.5 34.7 27.2 48.6

%pig 14.2  8.7  8.1  7.9  5.2  7.4  3.4  4.5  3.2 36.4  6.4 15.4  4.7

Staines sites (roadside settlement)

Phase

Total NISP

%cattle

%sheep/goat

%pig

Early Roman Early Roman Early Roman Middle Roman Middle Roman Late Roman Late Roman Late Roman Late Roman Late Roman

1184  448  428 1451  582  515   89  186  433  640

64.3 52.2 25.7 66.5 62.9 68.9 58.4 55.9 55.4 50.8

26.2 38.6 62.9 23.6 23.4 19.4 27.0 41.4 18.5 40.6

 9.5  9.2 11.4  9.9 13.7 11.7 14.6  2.7 26.1  8.6

Friends Burial Ground site Former Central Trading Estate Elmsleigh Centre 1975–78 Friends Burial Ground site Former Central Trading Estate Former Central Trading Estate 42–54 London Road Old Police Station, 10–16 London Rd Friends Burial Ground site Elmsleigh Centre 1975–78

Romanch4.indd 134

19/09/2016 12:16:19

THE SOUTH

while blade punctures suggested that shoulders were being hung, perhaps for smoking or salting. Cattle vertebrae tended to have been axially chopped where carcasses had been split, while numerous hyoid bones and marks on the posterior of a number of mandibles suggest that the removal of tongues was common. Similar cleaver butchery marks have also been found on sheep/goat and pig bones, and horse butchery has been noted from several assemblages, highlighting the organised marketing of livestock. These types of mark are particular to towns and forts in Roman Britain and are indicative of large-scale and rapid processing of livestock (Dobney 2001; Maltby 2007). The redistribution of body parts within the settlement is illustrated by the presence or absence of cattle horncores. Although good evidence for cattle-processing remains can be found at the Friends Burial Ground and Elmsleigh Centre sites, horncores were largely absent, whereas at the Former Central Trading Centre, relatively large quantities were recovered, indicating the movement of horn around the settlement for tanners and/or horn-workers after primary butchery. It is important to note that there is very little evidence that these distinctive butchery patterns were employed on rural farmsteads in the subregion, where butchery evidence is recorded by specialists as being scarce or restricted to knifefilleting or skinning marks. Articulated limbs also appear to have been more commonly recorded, possibly indicating that carcasses were less intensively processed (Chaffey 2009, 54). At Thorpe Lea Nurseries, Egham, for example, skinning marks were common on bones from a number of species, while horncores were generally not removed from cattle skulls (Iles and Clark 2012, 175). Butchery practices on these sites differed from the carcassprocessing techniques found at Staines, leaving less frequent markings on the bones as a result. In summary, the excavated evidence from the Middle Thames Valley case study area indicates that considerable expansion and reorganisation of settlement occurred during the late Iron Age/ Romano-British transition, in particular the abandonment of open settlement and an increase in complex farmsteads. At the same time, there is more evidence for field boundaries, with extensive rectilinear field systems revealed by large-scale excavations, dividing farm land into more regular and organised plots. Features that relate to the husbandry and trafficking of livestock are also encountered more regularly, such as waterholes and tracks or droveways, these becoming more prevalent in the later Roman period. This evidence can be viewed in parallel with the development of the roadside settlement at Staines, where an emphasis upon large-scale livestock processing, as

Romanch4.indd 135

135

shown by butchery evidence, suggests a role as a market centre. In contrast, evidence for processing of arable surplus is comparatively rare, although evidence for grain storage is present. Overall, the Middle Thames Valley appears to have developed a more intensive livestock husbandry economy through the Roman period. However, there is very little evidence for wealth being reinvested in architecture, as reflected by the absence of excavated villas from the case study area. This can be contrasted with the chalk downs to the south, where there is much greater evidence for investment in arable agriculture and the more common development of villas alongside larger roadside settlements. The Hampshire Downs The Hampshire Downs case study area covers approximately 700 km², around twice the extent of the Middle Thames Valley study area, and contains 38 site records in total (fig. 4.72). This sub-region lies predominantly within modern Hampshire but also crosses into the eastern side of Wiltshire, on an area of chalk downland within which flow the courses and tributaries of the Rivers Avon and Test. Settlements include 29 farmsteads, 7 of which develop into villas, 2 roadside settlements and 1 village settlement. Evidence for field systems is recorded at eight sites. Many of the settlements are located within a triangle of three Roman roads. Located at the three crossing points of these roads are the two roadside settlements of East Anton and Old Sarum (Sorviodunum), and the civitas capital at Winchester (Venta Belgarum), which lies just beyond the case study area. Chronological evidence from the Hampshire Downs indicates a small increase in the number of farmsteads in use from the late Iron Age to the fourth century a.d. It is notable that open farmsteads are absent from the dataset, unlike in the Middle Thames Valley, and the late Iron Age settlement pattern is dominated by enclosed farmsteads, many of which are of the easily recognisable, ‘banjo’ enclosure form. This type of site has been described in Chapter 2, but essentially consists of a sub-circular ditched boundary that funnels out from the entrance. Not all enclosed settlements of late Iron Age date conform to this type, though others are generally of a similar size and tend to have either funnelled entrances or trackways leading to them, implying some level of controlled access (Cunliffe and Poole 2000b; 2008d; 2008e; fig. 4.73). Complex farmsteads do make an appearance during the Roman period, though these are few in number, while villas develop initially in the second century a.d., becoming most prominent in the third and fourth centuries a.d.

19/09/2016 12:16:20

136

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

4.72.  Distribution of sites in the Hampshire Downs case study area b

a

N

0

c

100 m 1:2000

fig.

4.73.  Plans of late Iron Age enclosed farmsteads in the Hampshire Downs region at Nettlebank Copse, Wherwell (Cunliffe and Poole 2000b) (a), Rowbury Farm, Wherwell (Cunliffe and Poole 2008e) (b), and Fullerton (Cunliffe and Poole 2008d) (c)

Romanch4.indd 136

19/09/2016 12:16:29

THE SOUTH

137

Late Iron Age

Early Roman: Phase A

Early Roman: Phase B

Late Roman: Phase C

0

500 m 1:10000

fig.

4.74.  Plans of Dunkirt Barn, Abbots Ann (Cunliffe and Poole 2008f), showing the settlement’s development from the late Iron Age banjo enclosure to the late Roman villa

It has been noted previously in this chapter that settlements across the Hampshire and Wiltshire downland show strong evidence for continuity and development into the third and fourth centuries. Many of the sites with late Iron Age banjo enclosures exhibit continued occupation through the Roman period, though not necessarily within the banjo itself, and some of these settlements underwent considerable reorganisation. At Grateley South, geophysical survey, supported by small-scale excavation, revealed settlement features covering around 15 ha, at the centre of which a banjo enclosure was encircled by a complex of field boundaries and other enclosures (Cunliffe and Poole 2008c). The late Iron Age phase is thought to have been succeeded by a large rectilinear enclosure defined by a more substantial ditch, with field systems to the south that appear to have been associated with both early and late Roman features. In the fourth century a.d., a villa complex of four or five large masonry structures was constructed that respected the central area where the banjo enclosure stood. Phased plans of the site at Dunkirt Barn, Abbots Ann, demonstrate a very similar pattern of development to that seen at Grateley South (Cunliffe and Poole 2008f). The core of this settlement is shown by survey to have been c. 14 ha, and, despite distinct phases of development being identified, excavation showed that the settlement was continuously occupied from the first century b.c. to the end of the fourth century a.d. Once more the position of the banjo

Romanch4.indd 137

enclosure was respected by later Roman features, suggesting its continued importance at these sites (fig. 4.74). Other examples of this type of settlement development can be observed within or beyond the case study area, such as at Bramdean located to the east (Perry 1972; 1982; 1986). Of the nineteen rural settlements in the case study area that were occupied in the late Iron Age, thirteen continued into the fourth century, most undergoing substantial reorganisations in form. In addition to these developments, over half of the 34 sites in the area (including nucleated settlements) also adopted corndryers, quite different to the Middle Thames Valley where such structures were rare. Almost all the corndryers in this case study area were late Roman in date, though possible second-century examples were excavated at Houghton Down and Grateley South (Cunliffe and Poole 2008b; 2008c). The two structures at Houghton Down were built in the widely recognised T-shaped form, while at Grateley South a more developed, V-shaped corndryer was constructed. These features may suggest an increased focus on the processing of surplus grain, implying an expansion in arable agriculture. Of the eighteen sites with corndryers, twelve produced archaeobotanical evidence which showed that spelt wheat consistently dominated most samples, though some variation between different sites is evident. Three plant assemblages included samples with high proportions of germinating grain, providing good evidence that

19/09/2016 12:16:29

138

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Corndryer

Corndryer

Corndryer

0

100 m 1:2000

fig. 4.75.  Plan of East Anton, Hampshire, showing trackway, enclosures and several corndryers (Firth 2011)

the corndryers were being used as malting ovens (Campbell 2008a, 169–72; Campbell 2008b, 162; Campbell 2008c, 203–4). However, Marijke van der Veen has argued that corndryers were multifunctional structures (Van der Veen 1989), a point that is supported by evidence from Grateley South where the twin flues of the V-shaped corndryer each contained significantly different proportions of sprouting grain, suggesting that each assemblage was being processed for different products (Campbell 2008a, 174). Although spelt wheat was the most common cereal identified in most assemblages, barley also appears to have been common in many samples. It is possible that spelt and barley were both grown for malting, such as at Dunkirt Barn where a mixed-grain malt may have been preferred (Campbell 2008c, 204). At the roadside settlement of East Anton, excavation of a peripheral area revealed activity beginning in the first century a.d. but with evidence for a major intensification of agricultural/ industrial activity from the mid-third century (Firth 2011). At least twelve malting kilns/ corndryers, dating c. a.d. 240–400+, were excavated across the site, ten of which were sampled for archaeobotanical material (fig. 4.75). The assessment results showed very high numbers of cereal grains, principally of barley and wheat, and wood charcoal (Allen 2011). Unfortunately, very little else is known about the roadside

Romanch4.indd 138

settlement at East Anton since excavation has been limited. However, the number of corndryers found in such a restricted area, assuming that their use was broadly contemporary, indicates that substantial quantities of grain were being processed, either being malted for the production of beer or charred for consumption or storage (cf. Reynolds and Langley 1979; Van der Veen 1989). Evidence for storage structures is more difficult to identify than corndrying structures (see Ch. 3), though several sites are recorded with potential evidence of this type of agricultural feature. Fourpost structures have been identified in Romanperiod phases at Woolbury (Cunliffe and Poole 2000a), Balksbury Camp (Wainwright and Davies 1995), and Ashley (Neal 1980), which may be indicative of comparatively small-scale storage facilities more commonly associated with later Iron Age practices, while aisled halls at the late Roman villas at West Dean (Master 1885) and Houghton Down (Cunliffe and Poole 2008b) may have provided much larger storage facilities. While there is good evidence for an increased emphasis upon arable processing on the Hampshire Downs, it is difficult to determine how much of the landscape in the study area was turned over to arable production. Excavated evidence for Romanperiod field systems is lacking and we are currently more reliant upon cropmark evidence. In the 1980s, aerial photography recorded a regular system of rectilinear fields on the chalk at Cockey Down near Salisbury. A pipeline excavation in an area of the site revealed evidence for domestic activity spanning the Iron Age and RomanoBritish periods (Lovell 1999). Owing to the restricted nature of the excavation it was unclear if settlement was continuous, though late Roman activity comprised a corndryer, boundary ditches and a possible ditched trackway. Unfortunately, without further excavation the dating and phasing of the field system cannot be determined. A possible exception to this comes from the cropmark evidence adjacent to the villa at Fullerton where a configuration of elongated strip fields, each averaging 40 × 150 m in area, can be seen running parallel to the River Anton behind the villa complex, which was established in the mid-third century a.d. (fig. 4.76). The lynchets of the field system were found to be slight and denuded by later disturbance, making them difficult to date. However, the orientation of the field system and its proximity to the villa led Cunliffe to suggest that they may have been ‘the home fields of the estate’ (Cunliffe and Poole 2008d, 143). As mentioned earlier in this chapter, the river at Fullerton had been canalised to run in front of the villa to operate a mill located to the south, while a hall house was converted during the middle of the

19/09/2016 12:16:30

THE SOUTH

N

Strip fields

Mills

r Anto n R i ve

Roman cana l

Villa

139

comparative lack of corndryers and storage facilities at many of these settlements suggests that these changes were concerned with an intensification of livestock husbandry rather than arable expansion, and it was shown through zooarchaeological evidence that the roadside settlement at Staines likely provided an important redistribution centre for the trade and exchange of animals and their associated products. The chalk downs of Hampshire and Wiltshire, by contrast, appear to have witnessed an expansion in arable agriculture over the same period of time, with a number of settlements engaging in large-scale crop-processing. Once more, towns and roadside settlements provided marketing opportunities, with wealth clearly being reinvested back into some rural settlements, which later began to adopt villa-style architecture. REGION SUMMARY

0

200 m 1:5000

fig. 4.76.  Plan of the villa and field systems at Fullerton, Hampshire (Cunliffe and Poole 2008d)

fourth century, with a corndryer and ovens being inserted. This occurred at the same time as the construction of a winged-corridor villa house, with its baths and extensive mosaic flooring, implying that a substantial investment of wealth occurred at the site, probably financed by the agricultural activities of the settlement (ibid. 166). Case study Summary The case studies presented here provide evidence for two distinct sub-regions, both of which demonstrate evidence for considerable changes in the form and use of domestic space and in wider patterns of land use from the later Iron Age through to the end of the Roman period. These changes have been shown to be fundamentally linked to the agrarian economy, and were probably associated with the development of markets through the emergence of roadside settlements and larger towns, which were established at the junctions of major communication routes. However, the developments observed in each case study area can be seen to have differed markedly. Open settlements were replaced by large enclosure and trackway complexes on the gravel terraces of the Middle Thames Valley, and some sites witnessed the establishment of extensive droveway systems, enabling the trafficking of large numbers of livestock across the landscape. A

Romanch4.indd 139

Owing to its proximity to the continent, its numerous urban centres, villas, and late Roman forts, the South region has long been a focus for studies of Roman Britain. But what is clear from the results presented in this chapter is that the rural settlement pattern across the region was highly varied. There were considerable differences in the types of settlements located in different landscape zones, as well as in the rate of their establishment and abandonment. This is exemplified by the rapid expansion and decline in settlement activity in the easternmost part of the region, compared to the slower, but more sustained settlement expansion in the western part. It is possible that this variation occurred due to a range of factors, including differences in social organisation, political stability, fluctuations in the local and regional economy, and differences between landscapes. Although patchy, the available historical evidence suggests that the political situation across the region was complicated and changeable, made all the more complex by the involvement of the Roman state, both before and after the conquest of a.d. 43. Considerable changes in the settlement pattern, alongside the rapid adoption of continental imports, on the South Coast Plain in the first century a.d. has been argued to reflect the growing power of elite groups (Cunliffe 1973; Davenport 2003). Certainly, the establishment of a number of early, elaborate villas with clear evidence for Mediterranean influence indicates major investments of wealth. Other direct results of Roman influence are indicated in the eastern part of the region, such as the increase in iron production in the Weald, which has been suggested to have been orchestrated by the Classis Britannica

19/09/2016 12:16:31

140

THE RURAL SETTLEMENT OF ROMAN BRITAIN

(Cleere and Crossley 1985). The construction of Watling Street and the establishment of London and Canterbury as major centres for commerce must have also facilitated trade and exchange, which was perhaps responsible for the establishment of early villas along the Medway and Darenth rivers. Early Roman economic development in the south-east led Mattingly (2006, 386) to state that much of this land became ‘prime real estate for investment by incomers to Britain’. Although it is difficult to gauge the actual movement of people, it does appear likely that the economic developments were at least partly responsible for the rapid expansion of rural settlement in that part of the South region. The rise in settlement numbers in the east is matched by an almost equally dramatic decline in the later third and early fourth centuries. At the same time, a number of villas were also abandoned on the Sussex Coastal Plain (Rudling 1998). Political instability and threats from piracy have been identified as possible factors for these changes and, certainly, the establishment of the Saxon Shore Forts along the south coast appears to have been a consequence of the insecurity of the third century (Mattingly 2006, 241–3). In addition, the town walls of London also received bastions in the fourth century and a riverside wall to complement the land-side defences, though whether these reflected a response to real threats or were intended to reflect the status of the settlement is uncertain (ibid. 330–3). The importance of Watling Street appears to have waned during this period, and the decline of the Wealden iron industry from the midthird century may also have impacted upon settlements, as has been suggested at Westhawk Farm (Booth et al. 2008). However, while the number of settlements was reducing, the great majority of villas in Kent continued to be occupied into the fourth century, and some villas, such as Bignor, received significant levels of investment during this period. Overall, the changes witnessed

Romanch4.indd 140

in the eastern part of the South region were not uniform, and it is possible that they reflected developments in land use and economic reorganisation, perhaps signifying a greater intensification of agricultural production in the fourth century a.d. (cf. Gerrard 2014). Compared to the situation in the east, the settlement pattern in the western part of the region is quite different. Here, settlement numbers also increased from the late Iron Age to the early Roman period, though this was far more gradual and, rather than declining after the second century a.d., was maintained into the late third and early fourth centuries. Other developments also took longer to become widespread in the western part of the region such as the adoption of masonry buildings, which were not common until the third century. The large, agglomerated villages found on the Salisbury plain seem to typify much of the settlement evidence in this area, and the inhabitants there were clearly engaged in extensive arable cultivation of the chalk-downs. These settlements appear to represent a long-lived pattern of settlement, originating in the Iron Age and maintained right through the Roman period. The regional chronology of the area is similar to adjacent areas of the Central Belt, particularly between the Fosse way and the Severn Estuary, and between Ilchester and the Cotswolds (see Ch. 5; and Mattingly 2006, 393–9, fig. 14). In one sense, the chronological and interregional settlement patterns in the South are quite clear, and there are obvious differences between the eastern and the western parts of the region. These may reflect considerable differences in lifestyle, farming, trade, and many other aspects of daily life. However, even at this scale, some of the nuances of this variation remain masked, and the two case studies, of the Middle Thames Valley and Hampshire Downs, have demonstrated regional variability within specific landscapes in far greater detail.

19/09/2016 12:16:31

Chapter 5

the central belt By Alexander Smith

The Central Belt is by far the largest of the regions defined within this project; at 32,459 km², it incorporates over 25 per cent more land than the next largest zone, the South. The region cuts a significant swathe though central Britain, including parts of 23 English counties along with most of Gwent and Glamorgan in South Wales (fig. 5.1). THE NATURE OF THE LANDSCAPE As would be expected given the scale of the region, the Central Belt contains an extensive and diverse group of geographic areas, ranging from the Vale of Taunton and the Quantock Hills in south-west England and the Vale of Glamorgan in

fig.

South Wales, extending up to and including the Fens in the east of England (fig. 5.2). Large areas of the region are low lying, particularly the Gwent Levels, Somerset Levels and the East Anglian Fens, and these very distinctive landscapes have had a profound effect upon elements of the late Iron Age and Romano-British settlement pattern. Elsewhere there are several major river-valley landscapes such as the Nene and Ouse Valleys, which drain the gently rolling claylands of the West Anglian Plain, the Severn and (Warwickshire) Avon Vales, the Thames and (Bristol) Avon Vales and the Trent Valley and Rises. It is within these wide valley zones that a large proportion of our data on Roman rural settlement lies, though this has been at least partly brought about by patterns

5.1.  The Central Belt region in relation to modern county boundaries

141

Romanch5.indd 141

15/09/2016 17:50:03

142

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

5.2.   Constituent landscape zones of the Central Belt region

of modern development and mineral extraction, particularly quarrying of the river terrace gravels (see The Central Belt dataset below). The valleys are interspersed with hillier landscapes including the carboniferous limestone Mendips, the oolitic limestone Cotswolds, and the chalk hills of the Berkshire and Marlborough Downs and the Chilterns. While the Somerset Levels and the Fens are highly acidic burial environments, most of the region is characterised by less acidic soils, meaning that faunal remains, human skeletal material and botanical evidence are often very well preserved. The Central Belt as defined in this project is largely coterminous with the Central Zone of the Fields of Britannia project, which utilised similar cultural and physical characteristics in its creation (Rippon et al. 2015, 182; see Ch. 1). These generally fertile lands have also been previously defined in terms of the medieval landscape, being characterised by open fields and villages that developed between the eighth and twelfth centuries (ibid.; Roberts and Wrathmell 2000). It is clear then that despite the significant variation in individual landscape zones, there are broadly shared characteristics across the Central Belt

Romanch5.indd 142

region that provide at least some level of cultural and economic coherence, one that appears to have persisted for a long period of time. THE CENTRAL BELT DATASET Around 40 per cent of all records from the current project are found within the Central Belt, amounting to over 1500 separate database site entries. As these records derive exclusively from intrusive archaeological fieldwork, it can be assumed that the concentration of data in this region is in large part due to significant levels of modern development (including mineral extraction), and also to the fact that many areas have had a long history of archaeological research, as suggested by the distribution map of archaeological reports by date in Chapter 1 (see fig. 1.1). In terms of other data sources, Taylor’s national distribution maps of Roman rural settlement showed distinct concentrations of cropmark and soilmark sites mostly following the same river valley and Fen edge locations as the excavated dataset (Taylor 2007, 14–5, fig. 3.3; see Ch. 12, fig. 12.3). The only significant exception

15/09/2016 17:50:11

THE CENTRAL BELT

143

fig. 5.3.  Kernel density of Central Belt region records (n=1509) and all excavation records (1910– 2010) from National Monument Records (NMR) Index (n=14,622)

Romanch5.indd 143

15/09/2016 17:50:24

144

table

THE RURAL SETTLEMENT OF ROMAN BRITAIN 5.1: number

of excavated sites and density (per km²) by landscape zone in central belt region

Landscape zone Area (km²) No. of records 273

14

0.0513

1109

40

0.0361

Bristol, Avon Valleys and Ridges

836

52

0.0622

Charnwood

174

2 0.0115

1640

67 0.0409

Bedfordshire Greensand Ridge Berkshire and Marlborough Downs

Chilterns Coal Measures Cotswolds

200

4

2881

0.0200

120 0.0417

Dean Plateau and Wye Valley

825

32

0.0388

Gwent Levels

150

22

0.1467

Lincolnshire and Rutland Limestone

1260

75

0.0595

Mendip Hills

303

12

0.0396

Mid-Somerset Hills

420

19

0.0452

Midland Clay Pastures

1717

41

0.0239

Midvale Ridge

445

40

0.0899

Rockingham Forest

510

22

0.0431

2102

146

0.0695

650

26

0.0400

South Wales Valleys

1100

8

0.0073

Thames and Avon Vales

2532

165

0.0652

The Fens

3807

87

0.0229

Trent Valley and Rises

4571

159

0.0348

350

22

0.0629

Severn and Avon Vales Somerset Levels and Moors

Vale of Glamorgan Vale of Taunton/Quantock Fringes West Anglian Plain Yardley-Whittlewood Ridge

478

10

0.0209

3509

302

0.0861

337

17

0.0504

was in the hills and moors of south-west Somerset, where quantities of cropmark/soilmark data were much higher than those of the excavation record. The spread of records across the region is, as would be expected, far from even, as shown in fig. 5.3 and table 5.1. In terms of density of excavation records by landscape zone, the Gwent Levels (which include the immediate vicinity of the wetland area) stands out above all other areas, with over 20 sites recorded, despite this being the smallest zone in the region. Although almost 30 per cent of these records are of non-settlement features (mostly field/drainage systems), it is clear that this wetland zone was well occupied, particularly during the mid- to late Roman period. Elsewhere, the West Anglian Plain has the greatest density (and overall number) of records, while at the other end of the scale, the South Wales Valleys and The Fens appear relatively sparsely occupied, though the picture for the latter is somewhat

Romanch5.indd 144

Density of records per km²

misleading as site records are concentrated on the raised islands and on the immediate periphery. When viewed spatially (fig. 5.3, top), major concentrations of records are found to the north of Cambridge around the southern Fen edge (see Cambridgeshire Fen edge case study below), the western Fen edge near Peterborough, southern Lincolnshire around Sleaford, the southern Ouse Valley and Nene Valley north of Northampton, parts of the Upper Thames and Severn Valleys, the central Cotswolds, and in the Avon Valley around Bath and Bristol. Most of these concentrations are in relatively low-lying zones, in river valleys and on the edge of the Fens. Their relative density can at least partly be attributed to patterns of modern development, as seen when comparisons are made with the density of all recorded excavations in the National Monuments Record (NMR) Index between 1910 and 2010, shown on fig. 5.3. In areas such as the Cambridge environs,

15/09/2016 17:50:24

THE CENTRAL BELT

Peterborough, parts of the Oxfordshire Thames Valley, the Severn Valley around Gloucester and the Bristol Avon Valley, the patterns of excavation suggest a correlation with high levels of development, whether in the form of housing/ commercial, infrastructure or mineral extraction. On the other hand there are other zones with high concentrations of Roman rural settlement records, such as the Nene and Ouse Valleys and south Lincolnshire, that do not appear to have had spectacularly high levels of excavation (though there are increased concentrations around Milton Keynes and Bedford), and in these areas at least we might consider that there was a genuinely very high density of Roman rural settlement. There are of course many variables that influence the patterns of the excavated data, but it seems inescapable to conclude that Roman rural settlement was not evenly distributed across the region, and the major fertile valleys and Fen edge zones might possibly have had higher levels of activity, associated with their greater capacity for agricultural productivity (see Ch. 12 for wider discussion on settlement density).

145

ROMAN RURAL SETTLEMENT PATTERNS Romano-British settlement within the Central Belt region is known to have been immensely varied, characterised by a mixture of farmsteads and villas of different forms, as well as numerous roadside settlements and villages. Besides many walled ‘small towns’, major centres within the region include the legionary fortress and associated canabae at Caerleon (Gwent), coloniae at Glevum (Gloucester) and Lindum (Lincoln), and other towns that served as administrative centres at Venta Silurum (Caerwent), Corinium Dobunnorum (Cirencester), Ratae Corieltavorum (Leicester), and Verulamium (St Albans) (fig. 5.4). The 1509 records from the current dataset for the Central Belt region equate to 1090 individual rural settlements, others comprising multiple records for nucleated sites or wider landscape features such as field systems or isolated agricultural buildings. As stated above, these excavated settlements were not equally distributed across the region, and are broken down by site

fig. 5.4.   Excavated late Iron Age/Roman rural settlement in the Central Belt region in relation to Roman roads and urban centres

Romanch5.indd 145

15/09/2016 17:50:32

Romanch5.indd 146

of major settlement types by landscape zone in the central belt region

Farmstead Farmstead Farmstead Farmstead Villa Roadside Village Vicus Hillfort Total no. % of total (all) (complex) (enclosed) (open) settlement settlement settlement

5.2: number

21

0

3

2

5

2

1

0

1

28

2.57

4

8

0

0

0

43

3.94

17 2 3 0 4 2 1 0 0 24 2.20

Midvale Ridge

2

1

3

0

0

18

1.65

91 10 16 4 11 6 2 0 2 110 10.09 13 0 8 1 5 2 0 0 0 17 1.56

Vale of Glamorgan

14 0 3 0 3 0 0 0 0 17 1.56 874 135 138 32 156 82 24 1 17 1090

Yardley-Whittlewood Ridge

228 63 49 11 21 14 4 0 1 252 23.12

Total

West Anglian Plain

Vale of Taunton/Quantock Fringes 9 2 2 1 1 0 0 0 1 10 0.92

47 6 4 1 1 4 2 0 0 54 4.95

The Fens

101 22 14 5 11 5 5 0 0 117 10.73

0 0 0 0 0 3 0 1 1 5 0.46

0

Trent Valley and Rises

Thames and Avon Vales

South Wales Valleys

0

14

Somerset Levels and Moors 4

88 15 9 5 10 5 3 0 2 103 9.45

Severn and Avon Vales

7 0 2 0 6 1 0 0 0 12 1.10

26 4 3 0 2 1 0 0 2 30 2.75

Midland Clay Pastures

Rockingham Forest

15 0 1 0 5 1 0 0 0 17 1.56

2 0 0 0 2 4 0 0 0 7 0.64

0

Mid-Somerset Hills

Mendip Hills

2

34

1

15 0 3 1 0 0 0 0 1 16 1.47

Lincolnshire and Rutland Limestone

8 0 2 0 4 3 0 0 1 15 1.38

49 2 8 1 40 11 2 0 2 96 8.81

4 0 0 0 0 0 0 0 0 4 0.37

34 1 6 0 13 3 0 0 0 43 3.94

2 0 0 0 0 0 0 0 0 2 0.18

Gwent Levels

Dean Plateau and Wye Valley

Cotswolds

Coal Measures

Chilterns

Charnwood

Bristol, Avon Valleys and Ridges 26 3 0 0 6 5 1 0 3 40 3.67

Berkshire and Marlborough Downs

0 0 1 0 0 0 8 0.73 Bedfordshire Greensand Ridge 7 0 0

Landscape zone

table

146 THE RURAL SETTLEMENT OF ROMAN BRITAIN

15/09/2016 17:50:32

% of sites

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

147

hillfort village roadside settlement villa farmstead

fig. 5.5.   Relative frequency of main settlement types by landscape zones (with >20 sites) in the Central Belt region (n=no. of sites)

type and landscape zone in table 5.2. As in other parts of Britain, there is an almost total dominance of sites categorised as farmsteads, forming over 75 per cent of all settlements in the region as a whole, followed by villas (14 per cent) then roadside settlements (7 per cent) and villages (2 per cent). A small number of hillforts remained in use into the Roman period, while only a single excavated vicus is recorded from the Central Belt region, at Abergavenny in South Wales, notwithstanding the earlier phases of settlements outside of short-lived early Roman military bases, including the legionary fortress at Usk. The proportion of different settlement types does vary by landscape zone, however, with, for example, the Fens comprising almost 90 per cent farmsteads, and the Cotswolds having a much higher percentage (39 per cent) of settlements with villa architecture (fig. 5.5). The proportion of nucleated settlement always remains relatively low, though roadside settlements are quite numerous in the Lincolnshire and Rutland Limestone zone (e.g. Sleaford and Navenby), forming over 17 per cent of all settlement, mostly positioned on the major north–south road from Durobrivae to Lincoln. This is a similar proportion to that found in the East region, and appears to be part of a particularly intricate network of nucleated centres focused on the road network in the wider eastern area (see Ch. 6). REGIONAL CHRONOLOGY As with most other regions, the primary method of dating for Roman sites in the Central Belt is through the pottery assemblages, with the range of finewares commonly encountered on many settlements having quite a nuanced chronological

Romanch5.indd 147

resolution. The Central Belt lies on the fringes of the distribution for some of these finewares, such as Hadham red-slipped wares in the east and New Forest slipped wares to the south and west, but there are two major production centres within the region, in the Nene Valley and Oxfordshire, whose products are widely distributed throughout (cf. Tyers 1996, 168–78). Although detailed analysis of pottery types has not been attempted for this project (though see below, p. 185, and vol. 2), the spread of fineware pottery (including imported wares) on Roman period sites in the Central Belt region is quite evident, being present in the assemblages of at least 88 per cent of the 1044 sites where pottery was quantified. In sites where only coarse wares are noted (possibly due to the small size of the assemblage), chronological resolution can be that much poorer, with, for example the date of Severn Valley ware ranging from possible pre-conquest to the fourth century a.d.; this can disproportionately affect our understanding of the chronology of ‘lower order’ farmsteads in particular. Radiocarbon dates are increasingly utilised to refine ceramic chronologies, and in some cases to push the limits beyond those that are understood on the basis of pottery alone (e.g. at Horcott Quarry, Glos; Hayden et al. 2012; see Ch. 1, fig. 1.4). Other datable finds categories such as brooches and especially coins are also particularly well represented in Central Belt sites, so that, on the whole, our understanding of regional chronological trends for most areas is fairly robust. In terms of a broad regional chronology for the Central Belt, there appears to have been a steady rise in the number of settlements in use from the

15/09/2016 17:50:33

148

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Elsewhere, the Severn and Avon Vales to the west appear to have had a more rapid rate of settlement growth up to the later second century a.d., followed by a significant decline in the fourth century. The Upper Thames and Bristol Avon had a similar pattern in some senses, though there was a greater rise in settlement numbers during the late Iron Age, a peak in the early second century a.d. and a more gradual decline during the later Roman period. The differences between these western valley zones show how complex and varied the situation was, even within adjacent areas. Nevertheless, they all showed evidence for significant settlement expansion (and transformation) during the later first to early second centuries a.d., potentially associated with newly emerging urban centres at places like Cirencester and Gloucester (e.g. Booth et al. 2007, 374).

late Iron Age until the later second century a.d., after which there is a slight decline, this becoming more pronounced only during the later fourth century a.d. (fig. 5.6). There are of course significant intra-regional variations to this pattern, as shown by four sample landscape zones in fig. 5.7. In the hillier regions to the south and west, for instance, there is a steady increase in the number of settlements in use from the late Iron Age, reaching a peak in the later third/early fourth century, about a hundred years later than in most other areas. This is very similar to the situation seen in the adjoining South Wessex Downs and the Wessex Vales within the South region (see Ch. 4), thereby suggesting distinct wider territorial patterns; in all these areas the rise in settlement numbers is at least partly due to the increased construction of villas at this time. 800

no. of settlements

700 600 500 400 300 200 100 0 Mid-Late Late Iron AD50 Iron Age Age 100

AD100 - AD150 - AD200 - AD250 - AD300 - AD350 150 200 250 300 350 400

no. of settlements

90

Cotswolds (n=104)

Severn and Avon Vales (n=108)

30 20 10

West Anglian Plain (n=268)

200

75 60 45 30 15

160 120 80 40 0

l

Romanch5.indd 148

40

0

0

fig.

The Fens (n=54)

50 no. of settlements

80 70 60 50 40 30 20 10 0

no. of settlements

no. of settlements

fig. 5.6.   Number of settlements in use over time in the Central Belt region (NB: the M–LIA data only include sites that continued into the late Iron Age)

h

l

b

l

d

5.7.  Variations in settlement chronology by selected landscape zone within the Central Belt region

15/09/2016 17:50:34

THE CENTRAL BELT

All of the eastern zones of the Central Belt were at their height in terms of settlement numbers during the later second century, with numbers in the Fens in particular dropping off quite sharply after this point. The mid–later second century a.d. settlement peak in the Fens may have been broadly associated with construction of major drainage systems (possibly including Car Dyke, though see Simmons and Cope-Faulkner 2004) designed to reclaim much of the landscape. This was also the period when a nucleated settlement was established at Stonea (c. a.d. 130) in the middle of the Fens, which has often been regarded, albeit with fairly minimal evidence, as the administrative seat of an imperial estate (Jackson and Potter 1996; Malim 2005, 127; Upex 2008, 180; cf. Fincham 2002). The subsequent third

149

century a.d. decline in settlement numbers would seem likely to be associated with increased flooding in the eastern Fens (Upex 2008, 194), though sites on the Fen edge would appear to have remained relatively constant (see Cambridgeshire Fen edge case study below) . Settlement continuity and abandonment The graphs in figs 5.6 and 5.7 show how overall settlement numbers increased and decreased over time, but what they do not show are the relative levels of establishment and abandonment over time, which can provide some measure of the degrees of continuity in site use. figure 5.8, for instance, illustrates that during the late Iron Age, c. 60 per cent of the 457 sites in use across the

% of sites 'established new' during phase

% of sites 'abandoned' during phase

100 90

% of sites in use

80 70 60 50 40 30 20 10 0 Late Iron Age (n=457) fig.

Late 1stC AD (n=574)

2ndC AD (n=753)

3rdC AD (n=726)

5.8.   New’ and ‘abandoned’ settlements over time in the Central Belt region

Chilterns (n=43)

100

100

90

90

80

80 % of all sites in use

% of all sites in use

Trent Valley and Rises (n=110)

70 60 50 40 30

70 60 50 40 30 20

20

10

10

0

0 LIA

LIA

AD50 - AD100 - AD150 - AD200 - AD250 - AD300 - AD350 100 150 200 250 300 350 400

100

90

90

80

80

70

70

60 50 40 30

50 40 30 20

10

10 LIA

Romanch5.indd 149

60

20 0 AD50 - AD100 - AD150 - AD200 - AD250 - AD300 - AD350 100 150 200 250 300 350 400

AD50 - AD100 - AD150 - AD200 - AD250 - AD300 - AD350 100 150 200 250 300 350 400

The Fens (n=54)

100

% of sites in use

% of sites in use

Cotswolds (n=96)

fig.

4thC AD (n=656)

0 LIA

AD50 - AD100 - AD150 - AD200 - AD250 - AD300 - AD350 100 150 200 250 300 350 400

5.9.   ‘New’ and ‘abandoned’ settlements over time in selected landscape zones of the Central Belt region

15/09/2016 17:50:36

150

THE RURAL SETTLEMENT OF ROMAN BRITAIN

region were apparently ‘new’ (i.e. no indication of continuity of occupation from an earlier phase in the excavated area) settlement foundations, with the remaining 40 per cent representing continuations from the middle Iron Age. Most of the late Iron Age settlement continued in use past the conquest although 20 per cent were thought to have been ‘abandoned’ or had at least shifted in location before this date. The number of ‘new’ settlements established across the region steadily decreases over the course of the Roman period, while the levels of ‘abandonment’ gradually start to increase from the later third century, before an apparent wave of abandonment in the later fourth century, where it is thought that 84 per cent of sites in use did not continue into the fifth century a.d. (though see below, p. 151). The pattern for the Central Belt as a whole disguises many distinct chronological variations that can be seen within the different landscape zones of the region, as shown in fig. 5.9. In the Trent Valley and Rises, for example, a large lowland plain in the north-east of the region, there is a greater level of continuity of settlement from the middle to late Iron Age, though many of these

fig.

Romanch5.indd 150

settlements were then abandoned before the Roman conquest; there is subsequently a high proportion (50 per cent) of ‘new’ settlement attributed to the early Roman period (later first century a.d.). In the Chilterns, by contrast, a very high proportion (83 per cent) of settlement appears to have been established de novo during the late Iron Age, with only 17 per cent of sites occupied during this period having had any middle Iron Age precursor (cf. Thompson 2015). This pattern of increasing settlement numbers during the late Iron Age is also observed in wider parts of southern and eastern England where it has been associated with piecemeal population movement (Hill 2007, 23) and a reorganisation of the rural landscape into larger farming units (Bryant 2000, 16). Most of the Chiltern settlements continued uninterrupted into the Roman period and it is not until the fourth century that there is widespread evidence for settlement abandonment. In the Cotswolds there is a relatively high proportion (56 per cent) of late first-century a.d. settlements ‘newly’ established compared to most areas, with a corresponding high level of site abandonment (24 per cent of sites in use) during

5.10.   All excavated farmsteads (open, enclosed, complex and unclassified) in the central belt region

15/09/2016 17:50:44

THE CENTRAL BELT

the preceding late Iron Age, similar to the pattern of the Trent Valley and Rises. There then appears to be a period of settlement stability and growth into the later Roman period, with more settlements being established than being abandoned, at least until the fourth century a.d. (a similar pattern to that in the adjacent Wessex Downs in the South region; see Ch. 4). The Fens, on the other hand, appear to have evidence of more pronounced change, with a peak of new settlement in the second century and increased abandonment thereafter, for reasons outlined above. All the different landscape zones appear to have had a major phase of ‘abandonment’ during the later fourth century a.d. This is more likely to be a product of the vagaries of the dating evidence for this late Roman/early post-Roman period than a genuine pattern of complete settlement collapse, as discussed in Chapter 12 for the province as a whole. Explicit evidence for such continuity is decidedly lacking in most parts of the Central Belt, however, being noted on just 57 sites and heavily biased towards the west of the region (see Ch. 12, fig. 12.26). Such evidence is usually based upon radiocarbon dating, which is still rarely used for most Roman period sites, and so any resulting patterns must be viewed with great caution. FARMSTEADS: MORPHOLOGY, CHRONOLOGY AND DISTRIBUTION The distribution of all 874 farmsteads within the Central Belt is shown in fig. 5.10. Chronologically, these settlements conform to the regional and intra-regional patterns outlined above, principally because they form the dominant settlement type

151

within nearly all landscape zones. Unfortunately the majority (70 per cent) of farmsteads in the Central Belt region could not be morphologically classified in any phase for reasons outlined in Chapter 2 (e.g. minimal scale of excavation and/or truncation of archaeological features). The landscape zones incorporating major river valleys generally have a higher percentage of classified farmsteads (up to c. 50 per cent in the West Anglian Plain; see fig. 5.11), partly owing to the larger scale of excavations in these areas, often associated with extensive mineral extraction (see Methodology, Ch. 1). Of those excavated farmsteads where more diagnostic plans were available, there are approximately equal numbers having complex (135) and enclosed (138) phases, and very few (32) having any recognisable ‘open’ stage of development (notwithstanding the problems in interpreting such sites; see Ch. 2) (see examples in fig. 5.12). The distribution of different farmstead forms across the landscape zones is quite varied, with most (twelve of the nineteen zones where there was any farmstead classification) having greater numbers of enclosed farmsteads, some, such as the Vale of Glamorgan, Cotswolds and Chilterns, overwhelmingly so (see table 5.2). There is a distinct clustering of complex farmsteads around the southern and western Fen edge and along the Nene and Ouse Valleys, while further west they also dominate in the Upper Thames and Severn Valleys. Farmsteads with unenclosed phases are undoubtedly under-represented and show no particular geographic affinities, though they did represent two of the five classified farmsteads on the Berkshire and Marlborough downs.

% of farmsteads classified

60 50 40 30 20

Open Enclosed Complex

10 0

fig. 5.11.  Percentage of farmsteads classified by selected landscape zone in the Central Belt and breakdown of classification (n=total number of farmsteads in zone)

Romanch5.indd 151

15/09/2016 17:50:44

152

THE RURAL SETTLEMENT OF ROMAN BRITAIN

domestic focus

‘Open’ farmstead: Marsh Leys, Kempston, Beds

N

Enclosed farmstead: Kempsford Stubbs Farm, Glos

N

Complex farmstead: Cleveland Farm, Ashton Keynes, Wilts

0

200 m 1:5000

fig. 5.12.  Site plans of selected open (Marsh Leys, Beds; Luke and Preece 2011), enclosed (Stubbs Farm, Glos; Miles et al. 2007) and complex (Cleveland Farm, Wilts; Powell et al. 2008a) farmsteads in the Central Belt region

Romanch5.indd 152

The reasons behind the differential spread of farmstead types across the region probably relate to a combination of cultural and economic factors (discussed in Ch. 2 and below), while there are also very distinct chronological patterns relating to the frequency of such settlements within the landscape. figure 5.13 displays the relative percentage of the three major farmstead types in use over time within the Central Belt. It demonstrates a marked decline in the proportions of enclosed farmsteads in use, from a peak of 62 per cent of all classified farmsteads in the late Iron Age to a low of 22 per cent during the late Roman period. A corresponding rise is noted in the proportion of complex farmsteads, which account for just 17 per cent of farmsteads in the mid- to late Iron Age and reach a peak of over 74 per cent in the third century a.d. Open farmsteads, as just discussed, were rarely encountered, though they did form over 27 per cent of classified farmsteads in the mid–late Iron Age period, which appears to represent the tail end of a general long-term shift from open to enclosed settlements in many central areas of Britain during the Iron Age (Knight 2007). The most dramatic period of change, certainly in terms of the expansion of complex farmsteads, was the later first to mid-second century a.d. These farmsteads then reached their peak in terms of overall numbers (102) during the later second century, subsequently undergoing a slight decline, which became more pronounced during the fourth century, mostly in line with the general patterns of farmstead chronology noted above. Only a few landscape zones within the Central Belt have sufficient classified farmsteads to enable a more detailed intra-regional analysis, as shown in fig. 5.14. With the exception of the West Anglian Plain the numbers are still relatively low, yet they do demonstrate quite varied patterns. Most similar are the western valleys (Upper Thames, and Severn and Avon), which have a substantial rise in the number of complex farmsteads at the start of the second century a.d., with a significant drop in the fourth century, especially in the Severn and Avon vales. In the West Anglian Plain there is a more pronounced drop in the number of enclosed farmsteads from a high point in the late Iron Age, and a steadier rise in complex farmsteads, which peak during the later second century a.d., much the same as in the Upper Thames Valley. The situation in the Trent Valley and Rises is markedly different, with a peak in numbers of enclosed farmsteads during the later first century a.d., and a significant fall only during the third century a.d., when for the first time they are outnumbered by complex farmsteads. The small number of classified farmsteads from this landscape zone necessitates caution in placing too much

15/09/2016 17:50:46

THE CENTRAL BELT

open farmstead (n=32)

enclosed farmstead (n=138)

153

complex farmstead (n=135)

% of classified farmsteads in use

100 90 80 70 60 50 40 30 20 10 0

fig.

5.13.  Relative frequency of farmstead types in use over time in the Central Belt region

Severn & Avon Vales 16

16

14

14

12

no. of farmsteads

no. of farmsteads

Thames & Avon vales 18

12 10 8 6 4

10 8 6 4 2

2 0

0 M-LIA

LIA

M-LIA

AD50 - AD100 AD150 AD200 AD250 AD300 AD350 100 -150 -200 -250 -300 -350 -400

LIA

M-LIA

fig.

LIA

AD50 - AD100 AD150 AD200 AD250 AD300 AD350 100 -150 -200 -250 -300 -350 -400

50 45 40 35 30 25 20 15 10 5 0 M-LIA

LIA

AD50 - AD100 AD150 AD200 AD250 AD300 AD350 100 -150 -200 -250 -300 -350 -400

5.14.  Chronological distribution of farmstead types in use in selected landscape zones in the Central Belt

significance on such patterning, but it does again demonstrate the intra-regional diversity apparent in the rural settlement pattern. The wider regional spread of farmstead types over time is shown in fig. 5.15, where the late Iron Age concentration of complex farmsteads around the Fens and in the Nene and Ouse Valleys is most evident. During the period of greatest settlement expansion in the region, in the later second century a.d., the complex farmsteads spread further west, though, apart from areas of the Upper Thames and Severn Valley just noted, they tend to be fairly widely spaced. It is notable that to date no certain complex farmsteads have been found to the west of Gloucester, though possible examples exist on the edge of the Gwent levels at

Romanch5.indd 153

West Anglian Plain

no. of farmsteads

no. of farmsteads

Trent Valley and Rises 10 9 8 7 6 5 4 3 2 1 0

AD50 - AD100 AD150 AD200 AD250 AD300 AD350 100 -150 -200 -250 -300 -350 -400

places like Nash (Meddens and Beasley 2001) and Rumney Great Wharf (Fulford et al. 1994), where regular systems of drainage ditches have been found, similar to those in parts of the Somerset Levels across the Bristol Channel. The overall distribution of farmstead types changed little during the later third century a.d., though enclosed farmsteads became even scarcer in most areas except South Wales. By the later fourth century the major focus of complex farmsteads was again limited to parts of the east, while further west, the concentrations in the Thames and Severn Valleys dwindled significantly, though it is of note that of those that remained, most lay relatively close to the major urban centres of Gloucester and Cirencester (see Ch. 12). More

15/09/2016 17:50:47

154

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 5.15.  Distribution of farmstead types in use over time in the Central Belt (top left: Late Iron Age; top right: later 2nd century a.d.; bottom left: later 3rd century a.d.; bottom right: later 4th century a.d.)

nuanced accounts of this late Roman period are difficult to establish, as not only is there less evidence for active settlement, but there is also a significant reduction in the proportion of farmsteads that can be classified (down to 23 per cent from a peak of 37 per cent in the late Iron Age), a phenomenon found across much of Britain, and possibly related to the greater visibility of enclosed farmsteads which, as just demonstrated, tend to dominate during the late Iron Age/early Roman period. Variations in morphological types The discussion of farmstead morphology across England and Wales in Chapter 2 noted the many different variants within the major classifications

Romanch5.indd 154

of farmstead types. Enclosed farmsteads, for instance, ranged greatly in size, shape and the construction methods used for the outer enclosing bank. In the Central Belt, enclosed farmsteads in all landscape zones were typically rectangular in form (56 per cent), while others were irregular (18 per cent), curvilinear (7 per cent) or D-shaped (5 per cent). Where known, all were of earthwork construction (ditch and presumably bank, but this was not always demonstrated), though occasionally supplemented by masonry walling such as in the late Roman farmstead at Claydon Pike, Gloucestershire, which contained a drystone wall on a shelf along the inside of the main enclosure ditch (Miles et al. 2007, 181; fig. 5.16). As with the South region, most (87 per cent) enclosed

15/09/2016 17:51:20

THE CENTRAL BELT a

155 b

N

Claydon Pike, Glos

Old Shifford Farm, Oxon 0

100 m 1:2000

5.16.  Enclosed farmsteads at (a) Claydon Pike, Glos (Miles et al. 2007) and (b) Old Shifford Farm, Oxon (Hey 1995) fig.

farmsteads appear to have been established as ‘new’ settlements, sometimes then developing a complex form, though at least eight developed from ‘open’ farmsteads and two (including Claydon Pike noted above) actually developed into enclosed farmsteads from complex farmsteads during the late Roman period. The vast majority of enclosed farmsteads were under c. 1 ha in size, with the few larger farmsteads often utilising earlier hillfort enclosures, such as the early Roman buildings lying against the innermost rampart of an Iron Age coastal promontory fort potentially covering over 7 ha at Porthkerry Bulwarks in the Vale of Glamorgan (Davies 1973). Commensurate with their relatively modest size, most contained only one ditched circuit, though on occasion even small enclosed farmsteads were contained within two or three concentric boundaries (e.g. at Banbury, Oxon; see fig. 5.18 below). At other sites, such as at Old Shifford Farm in Oxfordshire, elements of concentricity are limited to certain parts of the settlement, in this case enclosing what is thought to have been the main domestic focus (Hey 1995, 111; fig. 5.16). Complex farmsteads also encompassed many variants, being broadly divided into sub-divided enclosures (e.g. Rudgeway Lane, Glos.; Hart and McSloy 2008) and linear complexes (e.g. NIAB Huntington Road, Cambridge: Luke 2014), the distinction being marked by internal development for the former (an enclosure divided into many spaces), and external development for the latter (an agglomeration of different enclosures; see Ch. 2 for details). Both types are well represented in the Central Belt, albeit with linear complexes (80) far outnumbering sub-divided enclosures (29) in most landscape zones, except the Severn and Avon Vales, where the numbers are more or less even

Romanch5.indd 155

(fig. 5.17). As outlined in Chapter 2, the predominance of linear complexes is linked to an increase in landscape management, and particularly with livestock farming. Their prominence within many of these Central Belt zones appears to have resulted from attempts at widespread landscape control, all designed to increase levels of agricultural productivity (see vol. 2 for a more detailed analysis). The development of complex farmsteads has already been discussed in Chapter 2, and the patterns in the Central Belt as a whole broadly reflect the wider picture from England and Wales, with c. 56 per cent (74 sites) appearing to originate as ‘new’ establishments (mostly in the late Iron Age and early Roman period) and the remainder developing from existing farmsteads or occasionally from field systems. A recently excavated example from Banbury quite clearly shows such development, in this case in a very abrupt manner, with the small (c. 0.1 ha) late Iron Age/early Roman concentric enclosed farmstead (if it is to be interpreted as such and not as a stock enclosure) giving way to an extensive series of conjoined rectilinear enclosures by the start of the second century a.d.; in the enclosures were a number of probable circular structures and two corndryers (Simmonds 2014; fig. 5.18). There is a degree of intra-regional variation in settlement development, with the mass of complex farmsteads in the valleys of the West Anglian Plain having evidence for longer histories of occupation than are seen in many other landscape zones, c. 30 per cent being specifically recorded as developing from enclosed farmsteads. In the Upper Thames Valley by contrast there appears to have been more ‘implantations’ of new complex farmsteads, mainly during the second century a.d. on land that seems to have become more actively managed, a

15/09/2016 17:51:20

156

THE RURAL SETTLEMENT OF ROMAN BRITAIN

N

0

100 m 1:2000

Sub-divided enclosure at Rudgeway Lane, Walton Cardiff, Tewkesbury

Linear complex at NIAB, Huntingdon Road, Cambs

fig. 5.17.  Distribution of complex farmstead types in the Central Belt and site plans of farmsteads at Rudgeway Lane, Glos (Hart and McSloy 2008) and NIAB Huntington Road, Cambridge (Luke 2014)

Romanch5.indd 156

15/09/2016 17:51:28

THE CENTRAL BELT

0

0 1:2000 1:2000

157

100 m 100 m

fig. 5.18.  Site plan of Banbury Flood Alleviation Scheme, Banbury, Oxfordshire, showing change from enclosed to complex farmstead (Simmonds 2014)

development also manifested in the widespread cutting of ditched trackways through the landscape at this time (Booth 2011a). VILLAS IN THE CENTRAL BELT Just under 50 per cent of all villas within the current dataset from England and Wales lie within the Central Belt region, although, as seen in fig. 5.19 and in table 5.2 above, the distribution of

Romanch5.indd 157

these 156 settlements is firmly weighted towards specific landscape zones, notably the Cotswolds and to a lesser extent the Chilterns and parts of the West Anglian Plain. Such concentrations of villas, particularly in the Cotswolds, are a wellknown phenomenon, though many of these sites have been subject to minimal, often antiquarian, investigation, and so do not always have useful levels of detail. As such, and in parallel with the South region, there remain a number of villas

15/09/2016 17:51:29

158

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

5.19.  Distribution of complex, enclosed and unclassified villas in the Central Belt 140 120 continued villa occupation

no. villas

100 80

development from farmstead

60

"new" villa establishment

40 20 0 AD50 100 fig.

AD100 - AD150 - AD200 - AD250 - AD300 - AD350 150 200 250 300 350 400

5.20.  Number of villas in use over time in the Central Belt with indications of development

where evidence is too poor to ascertain the date of establishment or abandonment, although over 90 per cent are provided with at least some level of chronological detail, other than just ‘Roman’. The development of villas over time on a national basis has been discussed in Chapter 2 (see fig. 2.21), and is shown for the Central Belt in fig. 5.20. Just ten villas can be assigned to the later first century a.d., some being part of the early group around Verulamium (e.g. Gadebridge Park: Neal 1974), others located around the

Romanch5.indd 158

fringes of the Nene and Ouse Valleys (e.g. Piddington: Friendship-Taylor and FriendshipTaylor 2013) and a small number found further west. These early western villas, which include Shakenoak Farm in Oxfordshire (Brodribb et al. 2005) and The Ditches, North Cerney in Gloucestershire (Trow et al. 2009), were situated within existing late Iron Age power centres, namely the ill-understood North Oxfordshire Grim’s Ditch and the oppidum at Bagendon. Altogether at least five out of the ten early villas had definite

15/09/2016 17:51:38

THE CENTRAL BELT

159

a Arable Paddocks

N

?Strip fields

Ancillary building Bathhouse

Arable Trackway

?ARABLE Domestic Farmyard enclosure

Bakery OPEN Well

Mound

Paved Yard

CORRALLING AREA Domestic and Farmyard enclosures Gravel pit

Detached bathhouse

Timber barn/cowshed

D

ro

ve

PASTURE

w

ay

ov

ew

ay

?Strip fields

Dr

0

200 m 1:5000

b

N

0

100 m 1:2000

fig. 5.21.  Plans of (a) ‘Complex villa’ at Roughground Farm, Glos. (Allen et al. 1993) and (b) ‘Enclosed villa’ at Whitton Lodge, South Glamorgan (Jarrett and Wathmell 1981)

evidence for an Iron Age precursor, and indeed many of those settlements that developed villa architecture at a later date could trace their origins back to the pre-conquest period. As will be discussed below, the early villa buildings of the Central Belt were all fairly modest compared to the major early villas of the Sussex coastal plain, yet would still have been strikingly different to

Romanch5.indd 159

contemporary vernacular architecture, and probably represent the fledgling adoption of what were perceived to be higher status Roman building forms by certain members of the indigenous elite. The first half of the second century heralded the start of a more rapid growth of villas in the Central Belt, though at first this was largely concentrated in the same areas as the earliest villas, with a few more

15/09/2016 17:51:39

160

THE RURAL SETTLEMENT OF ROMAN BRITAIN

isolated establishments such as at Llandough in South Wales (Owen-John 1988) and Shipham in Somerset (Barton 1963/4). The major growth in the number of villas more widely across the region appears to have occurred from the later second century a.d. onwards, with particular concentrations in the Cotswolds, where numbers continued to increase right up until the middle of the fourth century. The later fourth century heralds no particular decline in the number of villas in use within most landscape zones, though there are signs that the nature of occupation was changing in at least some settlements, with, for example, at least thirteen villas (mostly in the west of the region) having corndryers inserted into the main buildings (see Ch. 3). The only zones to see major reductions in the numbers of villas in use at this time were the Chilterns to the south-east and the Vale of Glamorgan in South Wales, the pattern in the former being consistent with the evidence for a general decline in numbers of villas in parts of south-east England during the later Roman period (see Ch. 4). As discussed at length in Chapters 2 and 3, the umbrella term ‘villa’ encompasses a wide range of different settlements, all ostensibly of relatively higher status than farmsteads, but nevertheless of extremely varied form, presumed social status and economic base. The architectural variability within the Central Belt will be discussed below, but in terms of wider settlement morphology it would seem that – where known – most villas follow the same patterns as Central Belt farmsteads in the mid–late Roman period, in being characterised by an association with networks of enclosures and trackways. The distribution of these 23 ‘complex villas’ is shown in fig. 5.19, with distinct concentrations in the Upper Thames Valley and central Cotswolds around Cirencester and Gloucester, and a few further east in the Nene and Ouse Valleys. In most cases the complex morphology of the outer settlement is only known wholly or partly through geophysics or cropmarks, such as at Harnhill 5.7 km east of Cirencester where part of a Roman villa complex was revealed with outer enclosures and trackways covering at least 5 ha (Wright 2008) or at Roughground Farm in the Upper Thames where the villa complex stretches over 15 ha, and could conceivably be regarded as a villa and ‘estate village’ (Allen et al. 1993; fig. 5.21; see below, p. 161). The nine ‘enclosed villas’ are more widely distributed, though they do appear to have been more common in areas with a dominance of enclosed farmsteads, such as the Chilterns and Vale of Glamorgan. The modest villa at Whitton Lodge on a Boulder Clay plateau in South Glamorgan originated as a late Iron Age enclosed

Romanch5.indd 160

farmstead (0.42 ha), with the villa buildings continuing to occupy the same enclosed space between the second and early fourth centuries a.d. (Jarrett and Wathmell 1981; fig. 5.21). An aerial photographic survey in 1996 revealed two concentric outer ditches to the main enclosure, extending c. 25–30 m from it, with a paved roadway leading from the main entrance (NPRN 227673). Two other potentially related but undated enclosures were found over 150 m to the east but there is nothing to suggest an extensive conjoined ‘complex villa’ settlement. The overall relative proliferation of villas within at least parts of the Central Belt landscape is a defining characteristic of the region, and although most cannot be said with any certainty to have been centres of extensive agricultural estates, many are likely to have been integral to wider developments of increasing agrarian productivity. Even within the Cotswolds villas are still fewer than other farmstead types (which almost certainly remain under-represented), and the social and economic relationships between these settlements will be explored more fully below. Pivotal to the functioning of all of these ostensibly smaller farming settlements, however, is their relationship with larger nucleated settlements. NUCLEATED SETTLEMENTS: ROADSIDE SETTLEMENTS, VILLAGES AND OPPIDA Of the three major types of nucleated settlement detailed in Chapter 2, roadside settlements and villages are relatively well accounted for in the Central Belt, while settlements remaining as vici attached to Roman forts only appear in the far western extremities, in the valleys of South Wales (fig. 5.22; see also Ch. 11). During the late Iron Age there are also a number of often extremely extensive, if mostly somewhat dispersed, settlements loosely grouped as ‘oppida’, which have generally received relatively little archaeological attention. On a national basis many of these places were seen to develop into major Roman towns (e.g. Verulamium) and are therefore beyond the scope of this project, though ‘enclosed oppida’ have been included at Abingdon and Salmonsbury, both sites developing into, or adjacent to, Roman nucleated settlements. The only extensive ‘territorial’ type oppidum within the Central Belt to be included is at Bagendon in the Cotswolds, which covered over 200 ha, defined by a number of linear earthworks. Recent excavations and survey have indicated a complex sequence of development with middle Iron Age origins, reaching its height in the mid-first century a.d. with an area of well-organised intensive occupation covering 15 ha and a number of outlying elements (e.g. The Ditches) in what was

15/09/2016 17:51:39

THE CENTRAL BELT

161

fig. 5.22.  Distribution of all excavated Roman nucleated settlement in the Central Belt region, in relation to walled towns

a large polyfocal complex (Moore 2014). Sites such as Bagendon were undoubtedly centres of production, exchange and power in the late Iron Age, and many of these functions were subsequently invested in nucleated settlements during the Roman period, from major urban centres through to extensive roadside ‘towns’ and even to some extent the ‘village’ type settlements. Villages The 24 aggregated rural farming settlements loosely defined as villages occur in ten of the landscape zones within the region, mostly in very low numbers and scattered along the major valleys of the Rivers Ouse, Upper Thames and Severn. Their appearance in similar locations (and landscape contexts; see below, p. 175) as complex type farmsteads is hardly surprising given that villages can be essentially regarded as expanded versions of such settlements, with multiple divisions of space into different activity areas (including zones of occupation), interspersed with trackways and surrounded by field systems. As outlined in Chapter 2, most village-type settlements are generally difficult to identify, especially without

Romanch5.indd 161

geophysics, earthwork or cropmark data, although with increased large-scale investigation in areas like north-west Cambridgeshire, more are now starting to be recognised (e.g. two major ‘village’ settlements north of Cambridge at Longstanton; Evans et al. 2006; see case study below). In terms of their chronological development in the Central Belt, there is a distinct rise in numbers during the early second century a.d. (to nineteen settlements from twelve in the late Iron Age), but thereafter remain fairly static, even up to the later fourth century a.d. There is a high degree of continuity from the late Iron Age onwards, with ‘new’ villages not found after the early second century a.d., and very few instances of such settlements being abandoned prior to the later fourth century a.d., the middle–late Iron Age Somerset lake villages and the substantial settlement of similar date at Humberstone, Leicester, being among the few notable examples. The nature of those settlements defined as villages within the Central Belt was quite diverse, and – as stated in Chapter 2 – they should not be confused with villages as defined during the medieval or modern period. As indicated above,

15/09/2016 17:51:47

162

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Excavated area

N

Stream

Villa ‘Village’

0

200 m 1:5000

fig. 5.23.  Cropmarks of the ‘village’ settlement at Lockington, Leicestershire, in relation to the adjacent villa and excavated site to the north-west (showing second–third century a.d. phase) (Thomas et al. 2013)

Romanch5.indd 162

15/09/2016 17:51:47

Romanch5.indd 163

1:7500

500 m

fig. 5.24.  Site plan of extensive excavations of the nucleated Roman ‘village’ settlement at Gill Mill, Ducklington, Oxfordshire, with later Roman phases highlighted (Booth and Simmonds 2012)

0

Mid–late and Late Roman Phases 4a and 4b Other features Road surface Archaeological layer

N

THE CENTRAL BELT 163

15/09/2016 17:51:50

164

THE RURAL SETTLEMENT OF ROMAN BRITAIN

some such as Abingdon developed out of what are generally perceived to be late Iron Age centres of power (oppida), while at least five are thought to have been directly associated with villas, including that at Lockington in Leicestershire (Clay 1985; Thomas et al. 2013; fig. 5.23). Much of this site is known through cropmark evidence alone, though recent large-scale excavations on the periphery revealed a related, complex development of enclosures, trackways and buildings, with continued activity from as early as the fifth or fourth century b.c. to the fourth century a.d., though with a radical reorganisation in the late Iron Age (ibid.). The spatial association of the Lockington settlement with the villa (dated second to fourth century a.d.) is striking, and along with other settlements such as Kingscote (Glos), Kings of Wessex School, Cheddar (Somerset) and possibly Roughground Farm, Glos (see above, p. 159), it could be viewed as an ‘estate village’ associated with the villa complex, where the agricultural workforce was concentrated as part of attempts to increase productivity and control (Hingley 1989, 102; Mattingly 2006, 382–3; see also discussion of the roadside settlement at Stanwick, Northants, below). Unfortunately, most of these other sites are very poorly understood, with information largely derived from cropmarks and fieldwalking alongside relatively limited excavation. As such it is not only difficult to ascertain just how widespread such nucleated rural settlements were, but also to understand their socio-economic functions and their relationships with other settlements. One settlement defined here as a village that has received considerable archaeological attention as the result of extensive quarrying over many years is Gill Mill, which lies on the gravels of the lower Windrush Valley in Oxfordshire (Booth and Simmonds 2012; fig. 5.24). Fairly limited later Iron Age activity preceded an extensive settlement established in the valley bottom in the early second century a.d., covering at least 10 ha and characterised by regular, ditched, rectangular enclosures around a large open area near the centre, suggested as perhaps a market place (ibid.). The economy of the settlement is thought likely to have been heavily geared towards cattle rearing and marketing and elements of its distinctive morphology find parallels in some other nucleated settlements including Fleet Marston, Aylesbury, Buckinghamshire, where a number of regular, ditched, rectangular enclosure systems lie on the periphery of what is thought to be a major roadside settlement along Akeman Street (Lythe 2009; Oxford Archaeology 2002; HS2 2013; fig. 5.25). The morphological similarities between Gill Mill and Fleet Marston highlight the somewhat arbitrary

Romanch5.indd 164

N

0

200 m 1:5000

fig.

5.25.  Magnetometer survey of regular ditched enclosures probably associated with the Roman roadside settlement at Fleet Marston, Aylesbury, Buckinghamshire (HS2 Ltd/GSB Prospection Ltd; HS2 2013)

division between our different classes of nucleated settlement, with at least some villages probably performing a similar range of functions to roadside settlements. Furthermore, although these villages may not have lain upon the main Roman road system, most are still arranged around major trackways, often at junctions, suggesting that transportation routes were still significant in their development. In terms of area covered, most of the Central Belt villages where we have any indication were less than 10 ha, making them somewhat smaller than the average roadside settlement, though there was still considerable variation and overlap between the two. Roadside settlements The 82 roadside settlements of the Central Belt are far more widely distributed across the region, as seen plotted in fig. 5.22 in relation to the defended ‘small towns’, which also form part of the network of nucleated roadside centres but which fall outside of the scope of this volume (see Ch. 1). Some regions did have higher percentages of roadside settlements relative to the number of villas and farmsteads, notably Lincolnshire and Rutland Limestone and the Cotswolds, but it remains uncertain if this greater apparent density was a genuine historical pattern and, if so, what the social or economic implications of this were. In the case of the former zone, many of the nucleated settlements lay upon a major north–south route running along the eastern side of the province, which may have facilitated their growth and development. As discussed in Chapter 2, there are also considerable differences in our knowledge base of many of these nucleated sites, with relatively few

15/09/2016 17:51:51

THE CENTRAL BELT

165

N

Road/stone layer

0

200 m 1:5000

fig. 5.26.  Excavation site plan of roadside settlement at Higham Ferrers, Northants (Lawrence and Smith 2009) and geophysics plot of the roadside settlement at Somerdale, Keynsham, Bath and North-East Somerset (Archaeological Surveys Ltd 2012)

being comprehensively investigated. About 80 per cent of those in the Central Belt had less than 0.5 ha excavated, with many either being sampled by evaluation trenching only (e.g. Stilton, Cambs: Thompson 2006b), or investigated through a series of small-scale interventions over many years (e.g. Dymock, Glos: Catchpole et al. 2007). The overall morphology of many of these settlements is therefore poorly understood, though it does appear that there was a considerable variation in scale,

Romanch5.indd 165

with some sites being traced alongside the road system for over 1 km (e.g. Shepton Mallett: Leach with Evans 2001; Whitewalls: Wilmott and Shipp 2006), and others appearing to be more concentrated over a few hundred metres (e.g. Baydon, Wilts: Fowler and Walters 1981). Where sufficient investigation has occurred, or where it has been possible to conduct wide-scale geophysical survey, many roadside settlements in the Central Belt are seen to cover 10–20 ha, and seem to have

15/09/2016 17:51:54

166

THE RURAL SETTLEMENT OF ROMAN BRITAIN

been defined by a series of ditched plots, fronting onto the main road, often with side lanes and groups of buildings located both at the roadside and sometimes towards the rear of the plots, as illustrated by the examples from Higham Ferrers, Northants (Lawrence and Smith 2009) and Keynsham in Bath and North-East Somerset (Archaeological Surveys Ltd 2012) (fig. 5.26). The variety in form and functions of these buildings is explored more fully below, but they reflect the wide range of activities occurring within roadside settlements, which often includes evidence for fairly intensive industrial and agricultural practices, alongside both formal and informal religious practice and undoubtedly a commercial function as markets and centres for the collection of taxes (see vol. 2). Some roadside settlements appear to form specialised industrial centres, such as Ariconium on the Dean plateau (ironworking; Jackson 2012), Caergwanaf in the Vale of Glamorgan (ironworking; Burnham 2004, 263–4) and Charterhouse in the Mendips (leadworking; Fradley 2009), while others owe their primary development to their role as ports, either inland (e.g. the Fenland port at Camp Ground, Cambs, and the river port at Crandon Bridge in Somerset) or on the coast such as at Sea Mills (Abonae). The development of roadside centres is obviously dependent on the creation of the road system, which in this central region incorporates parts of most of the major roads in Roman Britain, including Akeman Street, Watling Street and Ermine Street. The Fosse Way road cuts through the region from Somerset in the south-west to its termination at Lincoln to the north-east, and along it were strung ten roadside settlements recorded in the current dataset, major urban centres at Cirencester and Leicester, and at least six walled ‘small towns’, although some of the latter were significantly smaller than some of the unwalled nucleated sites (e.g. Dorn, whose walls encompassed just c. 4 ha). As most of the major roads in the region would seem to have been established during the mid- to late first century a.d., it is not surprising that this period sees the first expansion of roadside settlement, particularly in the Cotswolds along Akeman Street and the Fosse Way. However, these were not all ‘new’ sites as at least 22 (42 per cent of those sites ‘established’ during the later first century a.d.) demonstrated some kind of previous activity during the late Iron Age. In a few cases, such as at Ashton in the Nene Valley in Northamptonshire (Dix and Hadman 1984), these may well have already been substantial settlements, or else were functioning as preRoman industrial sites such as at Charterhouseon-Mendip, where lead mining was undoubtedly the reason why the road was built in the first place.

Romanch5.indd 166

Most roadside settlements with evidence for pre-conquest activity, however, seem to have developed from simple farmsteads, and it is doubtful in many of these cases whether the preexistence of such minor settlements had any real effect upon the routes of the major roads. However, there is now limited evidence that at least some of these roads may have followed existing, wellconstructed, Iron Age routeways, as seen at Sharpstones Hill, Shropshire (Malim and Hayes 2010; cf. Copeland 2011, 47). The subsequent development of roadside settlements within the Central Belt follows a similar pattern to that of villages, with a distinct increase in numbers of sites during the early second century a.d. (most of the few Severn and Upper Thames Valley examples date from this period), but thereafter remaining fairly constant, though there were still plenty of internal fluctuations in the fortunes of settlements during this time. Extensive excavations at Stanwick in the Nene Valley revealed a roadside settlement that developed in the late first/second century a.d. from a series of late Iron Age enclosures and trackways (Crosby and Muldowney 2011). It was one of a string of such sites lying upon a road that appears to have run from Irchester to Titchmarsh; these included the settlement at Higham Ferrers, just 3 km south of Stanwick. However, unlike Higham Ferrers, the settlement at Stanwick was set back from the major road, with a number of smaller metalled lanes being identified (ibid., 62) (fig. 5.27). Major changes occurred at Stanwick during the later Roman period, culminating in the development of a winged corridor villa in the later fourth century, fronted by a large enclosure, which cut across earlier boundaries and buildings, as other nearby buildings declined in importance or went out of use. This clearly signified a major shift in the social and economic focus of the settlement, perhaps then developing as a late villa estate ‘village’ in the same way as sites like Lockington, discussed above. Elsewhere, the fourth century saw decline and abandonment of many other roadside settlements, principally in parts of the east at smaller sites like Tort Hill in Cambridgeshire and Kempston in Bedfordshire, but also over in South Wales at Cowbridge and Great Bulmore, where reduction in the level of occupation was representative of a more widespread pattern of decreasing intensity of settlement activity in the fourth century. Only eleven settlements had specific evidence for continuation into the fifth century, with four of these being in Somerset, where such post-Roman activity has generally been well observed (e.g. Gerrard 2009). Radiocarbon dates from burials (Leach with Evans 2001, 28, 45) and

15/09/2016 17:51:55

THE CENTRAL BELT

167

N

Main road through Nene Valley

Layers Stone layers/roads Projected roads

0

200 m 1:5000

fig. 5.27.  Plan of mid-second to early third century a.d. phase of the roadside settlement at Stanwick, Northants (Crosby and Muldowney 2011)

archaeomagnetic dates from hearths (Holbrook 2011, 42) within the roadside settlement at Shepton Mallet indicated some kind of continued occupation right up until the sixth or seventh century a.d., albeit as the excavators suggest, ‘in much reduced circumstances and within the context of a new social and political structure’ (Leach with Evans 2001, 323). BUILDINGS As befits the scale of the Central Belt region, and its proportion of all records in the current dataset, over 43 per cent of all recorded architecture from the project is located here, accounting for 2681 buildings from 676 sites (fig. 5.28). This substantial dataset shows that there is no single characteristic architectural style within the region, but rather a complex amalgam of different traditions, some quite local and firmly rooted in pre-Roman styles and others more widespread, reflecting patterns found across the Roman Empire.

Romanch5.indd 167

CONTEXT, FORM AND CONSTRUCTION Most buildings from across the region derive from farmsteads, except in a few areas such as the Cotswolds and Chilterns where buildings associated with villa complexes dominate. In terms of absolute numbers, farmsteads account for 1045 buildings, while 583 come from villa complexes and 742 from roadside settlements. As discussed in Chapter 3, at the simplest level buildings can be divided into approximately circular and rectangular forms, with the Central Belt as a whole being dominated by the latter, which account for c. 60 per cent of all structures. When viewed by settlement type, however, farmsteads and villages have much higher proportions of circular buildings (which occur in 59 per cent and 77 per cent of these site types respectively) than villas and roadside settlements (17 per cent and 23 per cent), and enclosed farmsteads have the highest percentage of all (in 82 per cent of sites as opposed to 56 per cent of complex farmsteads). This is in part dictated by chronological factors, as illustrated in fig. 5.29, which shows the number of sites with circular and

15/09/2016 17:51:56

168

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

5.28.  Distribution and frequency of rural buildings on excavated sites in the Central Belt region

rectangular buildings over time. For the region as a whole there is a gradual decline in numbers of sites with circular buildings and a more pronounced rise in the presence of rectangular buildings, which become the dominant form by the early second century a.d. Although all landscape zones follow a similar basic trend, there are notable differences, with for example many more sites from the West Anglian Plain (including the Nene and Ouse Valleys) maintaining circular buildings into the fourth century, albeit usually alongside rectangular buildings and in much smaller numbers by this point. In contrast, the Cotswolds very quickly changed from an almost complete dominance of circular buildings in the late Iron Age (on an admittedly very small number of sites with evidence for buildings) to one very much dominated by rectangular buildings by the end of the first century a.d., in a period even before the subsequent proliferation of villa buildings. The basic architectural traditions (morphology and building material) of all landscape zones in the Central Belt are presented in table 5.3. The particular dominance of circular buildings in eastern zones such as the Fens, Trent Valley and

Romanch5.indd 168

Rises and West Anglian Plain is obvious, although even within these areas rectangular architecture becomes the norm by the later Roman period. In terms of building materials, timber (or masswalled; see Ch. 3) construction is more common within the region as whole, the dominance being total within the late Iron Age when there is no tradition of masonry architecture. The latter was introduced in the second half of the first century a.d. in early villas and some roadside settlements, though it remained extremely rare in most areas until the second century, when masonry buildings were found in over 50 per cent of all settlements with architectural evidence. The geographic patterns with regard to primary construction materials in some ways follow similar lines to those relating to building morphology, with most of those areas having a higher proportion of rectangular buildings also having high proportions of buildings at least partially built in stone. However, certain landscape zones also have relatively high numbers of circular masonry buildings of Roman date, particularly in and around the Nene and Upper Ouse Valleys in the West Anglian Plain (see Ch. 3, fig. 3.9), where

15/09/2016 17:52:05

THE CENTRAL BELT

169

Central Belt (all landscape zones)

no. of sites

400 300 200

Circular Rectangular

100 0 Late Iron Late 1stC 2ndC AD Age AD

3rdC AD

West Anglian Plain

80

80

60

60

no. of sites

no. of sites

Cotswolds

4thC AD

40 20 0

40 20 0

Late Iron Late 1stC 2ndC AD 3rdC AD 4thC AD Age AD

Late Iron Late 1stC 2ndC AD 3rdC AD 4thC AD Age AD

fig. 5.29.  Number of sites with circular and/or rectangular buildings over time in the Central Belt region as a whole, and in the Cotswolds and West Anglian Plain landscape zones

table

5.3: architectural

Landscape zone

Total no. No. rectangular No. circular No. timber No. masonry buildings buildings buildings buildings buildings

Bedfordshire Greensand Ridge

15

7

8

13

2

Berkshire and Marlborough Downs

47

43

4

14

32

137

131

6

19

108

Bristol, Avon Valleys and Ridges Charnwood Chilterns Coal Measures Cotswolds

1 0 1 1 98 84 14 35 6

3

3

6

331 277 54 67

0 54 0 263

Dean Plateau and Wye Valley

47

35

12

20

27

Gwent Levels

13

8

5

8

5

Lincolnshire and Rutland Limestone

79

45

34

50

28

Mendip Hills

40

39

1

5

34

Mid-Somerset Hills

62

58

4

4

58

Midland Clay Pastures

42

9

33

34

8

Midvale Ridge

56

43

14

13

42

Rockingham Forest

70

43

27

23

47

157

86

71

97

49

Severn and Avon Vales Somerset Levels and Moors

96

27

69

69

26

South Wales Valleys

14

14

0

4

10

186

119

67

106

79

98

29

69

86

11

Thames and Avon Vales The Fens Trent Valley and Rises

165

65

100

119

46

Vale of Glamorgan

67

42

25

32

36

Vale of Taunton/Quantock Fringes

17

8

9

12

5

793

377

416

549

240

44

14

30

29

15

West Anglian Plain Yardley-Whittlewood Ridge Total

Romanch5.indd 169

traditions within the central belt region

2681 1606 1076 1415

1225

15/09/2016 17:52:05

170

THE RURAL SETTLEMENT OF ROMAN BRITAIN % Masonry buildings

% Timber buildings

% of building types

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

fig. 5.30.  Relative frequency of masonry and timber buildings found at Roman period farmsteads within selected landscape zones in the Central Belt region (as percentages of all buildings, minimum 14 buildings)

over 50 per cent of 123 such buildings were located, including 30 at a single site at Stanwick (Crosby and Muldowney 2011). This appears to represent a relatively localised architectural tradition and is not just restricted to ‘lower status’ or specialist buildings (e.g. shrines or workshops), as suggested by the roughly circular building with a tessellated floor at Ringstead in the Nene Valley (Jackson and Parry 1980). The more typical correlation between areas with higher proportions of rectangular buildings and masonry construction is in part due to the presence of villas and, to a lesser extent, roadside settlements, where rectangular stone buildings appear more commonplace. If just farmsteads of the Roman period are considered the situation is more mixed, as shown in fig. 5.30. In general the regions further to the east still have much greater percentages of buildings in timber, with the Fens in particular having no masonry construction at all. Although stone buildings are known from this region, they are largely restricted to villas on the periphery (e.g. Little Oulsham Drove, Feltwell, Norfolk: Gurney 1986) and the monumental buildings of the nucleated settlement at Stonea (Jackson and Potter 1996), with the notable exception of a stone-footed building in the ‘village’ at Langwood Farm on the Fen Island of Chatteris (Evans 2003b). This must have been due mainly to the relative lack of suitable sources of building stone, since in the areas of higher limestone plateaus a little to the west (Yardley, Whittlewood Ridge and Rockingham Forest zones in Northants), the percentage of masonry buildings within farmsteads was at its highest for the whole Central Belt.

Romanch5.indd 170

Further to the west, the hillier zones (e.g. Cotswolds) in particular have very high proportions of masonry farmstead buildings, again probably due in no small measure to better availability of building stone. The architecture (rectangular and circular) of the major western valleys is far more timber dominated, particularly in the Severn Valley where almost 90 per cent of farmstead buildings did not seem to have any masonry component, despite close proximity to the Cotswolds. This suggests that, as with areas further east, the majority of buildings on farmsteads would have utilised only very local materials for construction purposes, and longer range movement of building stone was largely restricted to higher status settlements, notably villas. Furthermore, although a slightly higher percentage of complex type farmsteads than enclosed farmsteads in the region do have masonry buildings, the difference is not great (26 per cent as opposed to 19 per cent of site type), indicating that any variations in status between farmstead types (see discussion below) did not always find expression in masonry architecture. TYPES OF BUILDINGS IN THE CENTRAL BELT Over 60 per cent of settlements with records of architecture in the Central Belt contained evidence for single-roomed buildings only, this rising to 80 per cent of farmsteads in particular. The difficulties of identifying room division, especially in more insubstantial timber/masswalled buildings, have been discussed in Chapter 3, but it does seem that the vast majority of structures associated with farmsteads were fairly

15/09/2016 17:52:06

THE CENTRAL BELT a

171 b

N

Phase 1 Phase 2

0

20 m 1:500

fig. 5.31.  Plans of Roman multi-room buildings at (a) Newhouse Park, Chepstow, South Wales (Robic and Ponsford 2008) and (b) Horcott Quarry, Gloucestershire (Hayden et al. 2012)

simple in nature. Most (57 per cent) of the 271 sites with evidence for multi-roomed buildings comprised villa complexes, with another 14 per cent being found in 38 roadside settlements. The 60 farmsteads explicitly noted with multiroomed buildings were spread throughout the region, with two concentrations in the Cotswolds and in the West Anglian Plain, particularly parts of the Nene and Upper Ouse Valleys in the same zones as the circular masonry buildings discussed above, suggesting a particularly rich and diverse architectural tradition in this area. The great majority of these multi-roomed farmstead buildings were of masonry construction, such as the second–third century a.d. three-roomed ‘winged’ building from Horcott Quarry on the Upper Thames gravels (later replaced by a simple two-roomed rectangular structure (Hayden et al. 2012)) and the similarly dated three-roomed building at Newhouse Park, Chepstow (Robic and Ponsford 2008) (fig. 5.31). Evidence for timber structures with internal partitioning is far more scarce though examples do occur when preservation allows, for example in a large rectangular timber building of posthole and beamslot construction at Stoke St Mary, Taunton in Somerset (Dawson et al. 2003). Aisled buildings are relatively common in parts of the Central Belt, with over 100 examples recorded from 80 sites in the region (see Ch. 3, fig. 3.18 for distribution). Aside from the single late Iron Age example from Gorhambury, all others date from the Roman period, reaching their widest distribution in the second–third centuries a.d. Unlike in the South region, aisled buildings in the Central Belt are almost as commonly found on farmsteads as they are on villa complexes, the great majority of the farmsteads being of complex form where classified. They also appear in nine roadside settlements, three villages, an industrial site, a temple complex and as an apparently isolated building, thereby demonstrating the

Romanch5.indd 171

relative contextual diversity of this architectural form in this region. Such diversity is also seen in the construction material, with both timber (36) and stone-footed (54) structures well represented. The average size of Central Belt aisled buildings is much less than those of the South (272 m² as opposed 438 m²), probably reflecting the greater variety in terms of context and construction, though the same pattern of larger buildings being equated with residential and/or mixed use, and smaller buildings with agricultural use is evident. Villa buildings Chapter 3 highlighted the blurred architectural continuum from simple single-room structures to palatial multi-courtyard villa complexes, and it is within the Central Belt that this diversity is at its greatest. Villas have already been discussed above as a settlement type, but in terms of their architecture, this region not only contains a significant variety in form and scale, but also has the least distinct boundaries between those buildings that archaeologists have classified as villas and those that are seen by most to fall somewhat short (notwithstanding the variable definitions of the term villa; see Ch. 3). This is usually seen in terms of a lack of embellishments (mosaics, heated rooms, etc.), although, as just discussed, the Central Belt did have a number of farmsteads with multi-roomed masonry buildings, and some of these did have evidence for elaboration, such as painted plaster walls or tiled roofs. This blurred villa/non-villa boundary is such that there are over 50 farmsteads in this region that have been labelled as ‘possible villas’, sometimes just on the basis of a lack of excavation on the building (e.g. Hill Farm, Haversham, Bucks: Mudd 2006), but in other cases because the building lacks certain refinements. One example at Alfred’s Castle, Ashbury, Oxfordshire, on the Berkshire Downs may have started life as an aisled hall in the

15/09/2016 17:52:07

172

THE RURAL SETTLEMENT OF ROMAN BRITAIN a

b

c 0

20 m 1:500

1993 projected walls 2002 excavated walls

fig. 5.32.  Plans of Roman buildings at (a) Alfreds Castle, Oxon (Gosden and Lock 2013), (b) Chilton Fields, Oxon (Pine and Preston 2015), and (c) Croughton, Northants (Payne 2012)

late first/early second century a.d., but by the start of the third century it had developed into a sixroomed building with tiled roof and some evidence for painted plaster walls, though with no other obvious signs of architectural embellishment (Gosden and Lock 2013; fig. 5.32a). Another recently excavated late Roman building at Chilton Fields in the Oxfordshire Thames Valley originally comprised a simple rectangular masonry structure, possibly of two storeys, with an additional room/ stairwell on both sides, perhaps added at a later date, along with a small timber extension and a possible corridor (Pine and Preston 2015) (fig. 5.32b). The masonry foundations were said to have been very crude, with chalk and mortar floors, but the building had at least some painted plaster walls and a tiled roof (ibid., 15-19). Furthermore, lying over 25 m to the east was a very small hypocausted structure suggested to have contained a heated pool, and it was partly on

Romanch5.indd 172

this basis that the excavators interpreted the site as a modest Roman villa (ibid., 76). Ultimately, whether the addition of a heated room/structure or mosaic floor to such buildings would have elevated the perceived social status of the inhabitants within the local community is uncertain, though presumably quite likely. There is little doubt that such features would have entailed significant expenditure and were perhaps deemed unaffordable or even unnecessary in some cases, when such multi-roomed masonry buildings would already have stood out quite significantly from the majority of building forms in the Roman countryside. Within the upper echelons of the architectural continuum, there is nevertheless a whole range of buildings to which the term villa has been applied. The major types of villa have been discussed in Chapter 3 and are detailed by Central Belt landscape zones in table 5.4. At a relatively

15/09/2016 17:52:07

THE CENTRAL BELT

table

5.4: villa

173

architecture within the central belt region (final form of villa)

Landscape zone

Cottage/ Corridor Winged Courtyard Palatial Hall corridor

Berkshire and Marlborough Downs

2

Bristol, Avon Valleys and Ridges Chilterns

2

Uncertain

Total

2

1

5

3

1

6

1 6 4 2 13

Cotswolds

1 4 11 9 6 9 40

Dean Plateau and Wye Valley

1

2

1

4

Lincolnshire and Rutland Limestone

1

2

1

4

Mendip Hills

1 1 2

Mid-Somerset Hills 3 2 5 Midland Clay Pastures Midvale Ridge

1

1

2

1 1 2 4

Rockingham Forest

1 1 1 2 1 6

Severn and Avon Vales

1 2 5 2 10

Somerset Levels and Moors Thames and Avon Vales

1

1

2

1 2 2 2 4 11

The Fens 1 1 Trent Valley and Rises

1

1

1

8

11

Vale of Glamorgan 1 1 1 2 5 Vale of Taunton/Quantock Fringes

1

6 4 3 1 7 21

Yardley-Whittlewood Ridge

1 1 1 3

Total

17 25 36 26 8 44 156

modest scale, and where there is the highest degree of ambiguity with buildings such as that at Alfred’s Castle, are the developed aisled halls and cottagestyle villa buildings, such as at Croughton, Northamptonshire (Payne 2012; fig. 5.32c), which usually do have elements like mosaic pavements and/or heated rooms. At least twelve more elaborate villas in the Central Belt are thought to have developed from these forms of buildings, although there were a total of seventeen cottage/hall villas that remained as such in their final form. This is especially notable in the West Anglian Plain, which, despite having very high concentrations of settlement records, has somewhat less evidence for development even of winged corridor villas, let alone the more extensive courtyard types, although these were more prevalent on the adjacent low limestone plateau of the Rockingham Forest, towards the town of Durobrivae. The most widespread forms of villa across the Central Belt, as across the remainder of the province, are the corridor and winged corridor types, with many courtyard villas developing from such buildings (fig. 5.33). Not all of these villas developed in such a way, however, and some could remain as fairly modest structures, such as the

Romanch5.indd 173

1

West Anglian Plain

7-roomed late Roman corridor building at Yarford, Kingston St Mary in the southern Quantock Hills (King forthcoming), which marked the southwestern limit of sustained villa building nationally. Parts of south-western Britain have already been shown in Chapter 3 to have the most examples of excavated courtyard type villas, such as at Keynsham in Somerset (Cox 1998) and Atworth in Wiltshire (Erskine and Ellis 2008). These would seem to represent a distinctly higher socio-economic order, at least in terms of the investment involved in their development, and, in the Central Belt, are found in considerable numbers in the Cotswolds in particular, along with other hillier landscape zones. At the uppermost end of the scale of villa architecture lie those termed ‘palatial’, often arranged around two or more courtyards and usually comprising 40 to 50 rooms or more where known. Eight out of the twelve records of ‘palatial’ villas are located within the Central Belt, with six of these lying in the Cotswolds. All of these Cotswold ‘super villas’, such as Woodchester and Chedworth, date in their most opulent form to the fourth century a.d., though they seem to have developed from more modest origins.

15/09/2016 17:52:07

174

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

5.33.  Distribution of excavated villa types within the Central Belt

N

0

100 m 1:2000

fig. 5.34.  Plan of parchmarks and geophysical survey results showing outline of multi-courtyard villa complex at Turkdean, Glos. (Holbrook 2004)

Romanch5.indd 174

15/09/2016 17:52:16

THE CENTRAL BELT

The developmental history of the impressive triple-courtyard villa at Turkdean, c. 19 km northeast of Cirencester, is less well understood, being known mainly from parchmarks and geophysical survey, though small-scale trenching indicated occupation at the site from the second to late fourth century a.d., and possibly into the early fifth (Holbrook 2004, 66; fig. 5.34). Such villas represent a huge investment of wealth in this region from the very top levels of society within the Roman Empire. Away from the Cotswolds, the only two villas to fall into this category are Castor and Cottersock in the Nene Valley to the north and south of Durobrivae (Upex 2001; 2011). The nature and development of both are ill understood, though the Roman buildings at Castor are thought to have covered an area of 3.77 ha and the site is argued to have been a large administrative complex, taking over from Stonea in the Fens during the later third century a.d. Overall, the complex hierarchy of architecture within the Central Belt, from the mass of simple single-roomed structures to the Cotswold ‘supervillas’, highlights the degree of social and economic diversity within the region, which is also indicated by the associated material culture within settlements, discussed below. LANDSCAPE CONTEXT AND INFRASTRUCTURE TOPOGRAPHY Topographic factors appear to have been major influences on the development of different types of rural settlement, as shown through the spot height analysis graphs in fig. 5.35. Farmsteads in general

0 2 4 6 8 10 12 14 16 18 20 no. of sites

fig.

Romanch5.indd 175

200+ 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

Villas

height above ordnance datum (m)

200+ 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

appear to have favoured lower lying ground – the valley slopes and gravel terraces – though when different farmstead types are looked at, quite different patterns emerge. Enclosed farmsteads are scarcer on the lowest ground and relatively more prevalent in ‘upland’ areas (i.e. over 100 m OD) compared with complex farmsteads. The preference for complex farmsteads to be located in the major river valleys has already been noted, and in areas such as the West Anglian Plain this is even more pronounced, with nearly all examples being found near the Fen edge and on the gravel terraces of the Nene and Ouse Valleys, rather than on the surrounding higher claylands (some settlements seemingly ‘zoned’ and arranged to maximise agricultural output and access to transportation; e.g. at Broom, Beds: Cooper 2005, fig. 25). In other areas with fewer sites the differences are less marked, although the general preference for enclosed farmsteads to be located in higher landscape zones such as the Cotswolds has already been demonstrated. Such differences in landscape context may well be related to differences in economic function and social structure noted below. Those settlements defined as villas have a much more varied landscape context within the Central Belt as a whole, being distributed across nearly all bands of elevation. Their greater frequency in ‘upland’ zones compared to other settlements (e.g. six above 200 m OD) is in part due their prevalence in the Cotswolds, Chilterns and Berkshire and Marlborough Downs, though even among the smaller numbers in the Upper Thames and Avon Vales, villas occur more frequently on relatively higher ground, with 45 per cent found above 100 m compared with 11 per cent of farmsteads. In the West Anglian Plain, however, Enclosed farmsteads

height above ordnance datum (m)

height above ordnance datum (m)

Complex farmsteads

175

0 2 4 6 8 10 12 14 16 18 20 no. of sites

200+ 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9 0 2 4 6 8 10 12 14 16 18 20 no. of sites

5.35.  Spot height analysis for villas, complex farmsteads and enclosed farmsteads in the Central Belt region

15/09/2016 17:52:17

176

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Charterhouse in the Mendip Hills and Ariconium (Weston under Penyard) on the Dean plateau. Unlike in the South region (see Ch. 4), villages in the Central Belt are just as common at lower levels as roadside settlements, being especially prevalent in some of the major river valleys and at the Fen edge, much the same as complex farmsteads. The few that do occur on higher ground, at Knighton Bushes Lambourn and Kingscote, Gloucestershire, are likely to have been intimately associated with villa estates.

there are less obvious differences, with many villas sited on lower-lying ground such as at Redlands Farm, which lay on a sand island in the floodplain of the River Nene (Biddulph et al. 2002), and the newly discovered villa at Itter Crescent, east of Durobrivae, close to the western Fen edge (Henley et al. 2012). The two ‘palatial’ villa complexes in this region at Cottersock and Castor both lay at low levels, within a few kilometres of the western bank of the River Nene, though Castor was elevated in relative terms compared to the surrounding landscape, sited upon a terraced hill slope. Most of the grandest villas in the Cotswolds appear to have been carefully sited on the slopes of valleys with extensive views across the landscape, such as at Woodchester and Great Witcombe, which both lay towards the edge of the Cotswold escarpment. Such an active integration of villa and landscape has been highlighted by Scott (2004, 53), who argued that their setting was just as important a factor in the inherent desire to impress as the architectural design of the villa building itself. Roadside settlements were another major settlement type with specific topographic associations, being much more prevalent at lowerlying levels. This is primarily because, as discussed in Chapter 2, there was a strong association of such sites with rivers, and particularly river crossings. Over 50 per cent of roadside settlements in the Central Belt were located in lower-lying riverine contexts, with most of those on higher ground generally being of smaller size and/or associated with mineral extraction such as villas (n=156)

35

TRANSPORTATION: RIVERS, ROADS AND TRACKWAYS The association of roadside settlements with river crossings highlights the potential importance of rivers as transportation routes during the Roman period (cf. J.E. Jones 2012, 86–96), yet direct evidence for such use is still decidedly sparse. The extensive settlement at Camp Ground, Colne Fen is thought to have been an inland port, distributing cargo across the Fens and further afield (Evans 2013b, 451), while excavations of the Car Dyke in the near vicinity revealed large amounts of pottery in the base of the canal and a possible substantial timber warehouse adjacent to it (Macaulay 2012). The canal linked major rivers in the region, and seems very likely to have been associated both with the local Horningsea pottery industry in the second–third centuries a.d., and also with the wider distribution of agricultural produce from the region, although the interpretation and history of this monument remains complex (cf. Simmons and Cope-Faulkner 2004; see case study below). All Roman farmsteads (n=739)

% of site type

30 25 20 15 10 5 0 up to 0.25 35

0.25 - 0.5

0.5 - 1

Roman complex farmsteads (n=132)

1-2

2-5

5+

Roman enclosed farmsteads (n=119)

% of site type

30 25 20 15 10 5 0 up to 0.25 fig.

Romanch5.indd 176

0.25 - 0.5

0.5 - 1

1-2

2-5

5+

5.36.  Distances of Roman farmsteads and villas from major roads in the Central Belt region

15/09/2016 17:52:17

THE CENTRAL BELT

177

region. The problems of using the current Roman road system (as defined within the National Monuments Record) for any form of spatial analysis have been highlighted in Chapter 4, and yet there are a number of clear patterns in the data for the Central Belt region. figure 5.36 shows the distances of Roman-period farmsteads and villas from the major Roman roads in the region and the most immediately striking observation is the apparent close relationship between them. The great majority of both farmsteads (77 per cent) and villas (89 per cent) in this region lie within 5 km of the road system, although this may in part be affected by biases in development-led archaeology (i.e. many Roman roads lie under under/adjacent to modern roads, which attract more development). There are, however,

Elsewhere in the Central Belt, there is evidence of a timber wharf sealed by layers of alluvium at a roadside settlement on the Fen edge at Stilton (Wessex Archaeology 2006), and other possible wharfs at Stanton Low, Bucks (Woodfield and Johnson 1989), Chepstow, Monmouthshire (J.E. Jones 2012, 79) Grendon Lakes and Stanwick Silt Pond, Northants (Humphrey 1998; Jackson 1984) and Barland’s Farm on the Gwent Levels (Nayling and McGrail 2004). While it remains uncertain how extensively the major rivers were used as navigable highways, it was the Roman road system that formed the main communication network within the new province, and this seems to have been of fundamental importance in terms of the development and expansion of settlement within the Central Belt

N

0

200 m 1:5000

fig.

Romanch5.indd 177

5.37.  Plan of Roman settlement and trackway at NIAB, Huntingdon Road, Cambs (Luke 2014)

15/09/2016 17:52:18

178

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Cirencester

N

A417

Am

pn ey B

roo k

River C hu rn

Er

South Cerney

in

m et

re

St

Cotswold Community

Th am

Neigh Bridge

es

)

Rive r

(T

Cleveland Farm

19

Latton Lands

Somerford Keynes

A4

Spratsgate Lane

Court Farm

Ashton Keynes Cricklade

Swil lB

rook

Crop marks Roman road 0

2 km 1:50000

fig. 5.38.  Cropmarks showing Romano-British settlements and linking trackways, south of Cirencester in the Upper Thames Valley (after Powell et al. 2010, fig. 1.6)

Romanch5.indd 178

15/09/2016 17:52:18

THE CENTRAL BELT

differences noted between classified farmstead types, with complex farmsteads behaving far more like villas and a much higher percentage (29 per cent) of enclosed Roman-period farmsteads being located over 5 km from the road network. As with villas, this suggests greater integration of the Central Belt complex farmsteads into the communication systems of the province, probably in turn linked to greater social and economic integration. The major Roman roads are of course only one component of the transportation network within the province, the bulk of which comprises extensive networks of trackways, usually visible only when defined through parallel drainage ditches (shown in fig. 5.37). Such features are frequently traced as cropmarks in aerial photographs within the region, but have only been recorded here when they form components of excavated sites. Of the 506 sites from the Central Belt with explicit evidence of trackways, 441 were directly associated with settlements, with the remainder either being part of religious or industrial sites, or else forming elements of wider field systems. They are most prevalent in the Fens and Upper Thames and Avon Vales, where they were recorded on over 50 per cent of settlements, though were also noted fairly frequently (30–45 per cent) in most other areas. Their relative scarcity (22 per cent) in the Trent Valley and Rises is not easily explained, as the landscape zone has a reasonable number of settlement sites, with an average percentage classified. It could at first glance indicate a relative lack of settlement integration in at least parts of this zone, although extensive concentrations of enclosures, trackways and wider droveways observed as cropmarks in parts of the Trent Valley suggest that this was not the case (Bishop 2006, 4). Alternatively, it has been suggested that earlier enclosure and trackway boundaries in this region were ignored or altered to less archaeologically discernible forms (e.g. hedges) in the later Roman period (Taylor 2006, 11), which may account for their apparent lack of visibility in at least some of the excavated data. Most of the trackways in the Trent Valley and Rises zone were associated with farmsteads and roadside settlements, with a particular affiliation with complex farmsteads, consistent with the wider regional picture. Associations of such farmsteads with trackways is hardly surprising given the morphological nature of these settlements, which often integrate track- or droveways within or alongside the networks of enclosures, as demonstrated by the Roman settlement at NIAB, Huntingdon Road, Cambridgeshire (fig. 5.37). Yet trackways are also a feature associated with 33 enclosed farmsteads

Romanch5.indd 179

179

in the region (24 per cent of the total as opposed to 66 per cent of complex farmsteads), sometimes being attached to the main enclosure (e.g. Ash Plantation Site 2, Cambs: Abrams and Ingham 2008) and at other times lying some distance away and of uncertain relationship (e.g. A419 Blunsdon By-pass, Wilts: Brett and McSloy 2011). The dating of trackways is problematic, especially away from the domestic core, owing to the dearth of associated diagnostic artefacts, and so they have not been explicitly recorded as a separate dated feature in this project. However, by general association with settlements, it is seen that trackways were certainly a feature during the late Iron Age, particularly in parts of the West Anglian Plain and Fens, although it was not until the later first and particularly the early second century that such features became more widespread across the region, as explicitly noted by Booth for the Upper Thames Valley (Booth 2011). When excavated sites are viewed together with cropmark data, wider landscapes of interconnected farmsteads can sometimes be discerned, as in the Upper Thames south of Cirencester in fig. 5.38. These are occasionally shown to join up with the major Roman roads as seen at Court Farm, Latton, where a Roman trackway linking a farmstead with Ermin Street had evidence for cart-wheel ruts (Mudd et al. 1999, 126). The increase in numbers of trackways during the later first and second century a.d. points to greater levels of interconnectivity from this time, and also implies a greater emphasis on defining and dividing up the landscape, at least in some areas, which is concurrent with other changes in the landscape, for example in the development of field systems. FIELD SYSTEMS The landscape of most of rural central England today is overwhelmingly characterised by integrated systems of fields interspersed with roads, farmsteads, towns and villages. It has been recently demonstrated in the Fields of Britannia project that large parts of these fieldscapes are the products of sequential developments over thousands of years, within which there do not appear to have been any significant periods of abandonment (Rippon et al. 2015). As with so many other elements of the rural landscape these developments were far from uniform, and even from the genesis of widespread field systems in the mid–late Bronze Age, the variety in their form and scale seems to have been considerable (Yates 2007). In Taylor’s study of Roman rural settlement he examined their wider landscape contexts and explicitly noted such variation in the field systems across different landscapes (Taylor 2007, 55–72).

15/09/2016 17:52:19

180

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 5.39.  Distribution of sites with excavated and dated field systems in the Central Belt (top left: late Iron Age; top right: 2nd century a.d.; bottom left: 3rd century a.d.; bottom right: 4th century a.d.)

Within our Central Belt region these included the co-axial, strip fields of the Trent Valley, which appear to have developed piecemeal from the late Iron Age, and localised axial field systems aligned to the long sinuous trackways found in the Fenland region (ibid., 62–5). Further west, the Upper Thames Valley was characterised by networks of enclosed fields around settlements and trackways with ‘open’ areas beyond, while the Berkshire Downs was dominated by narrow co-axial strip fields (ibid., 66–7; cf. Levick 2015). The study of field systems relies heavily upon the existence of cropmarks found from aerial photography, as shown in the Upper Thames landscape in fig. 5.38. However, there are increasingly large-scale excavations that incorporate great expanses of field system, thus enabling a

Romanch5.indd 180

more detailed picture of their development to emerge. This is well demonstrated by excavations in the Upper Thames at Roundhouse Farm, Marston Meysey, Wiltshire, where extensive areas of middle Iron Age occupation over 25 ha were replaced by systems of fields and trackways during the late Iron Age/early Roman period, which partly re-used the earlier Iron Age landscape boundaries (Cass et al. 2015). The tensions apparent between such macro- (i.e. large-scale mapping projects) and micro-scale (excavations) analysis of field systems have been recently explored by Chadwick, who argued that the integration of these two approaches is essential for creating more nuanced accounts of these important landscape features, which ‘held great social, historical and symbolic significance’ (Chadwick 2013, 26).

15/09/2016 17:52:53

THE CENTRAL BELT

181

N

Farmstead

0

200 m 1:5000

fig.

5.40.  Plan of second century (Simmonds et al. 2011)

a.d.

farmstead and associated field system at Tubney Wood Quarry, Oxon

N Lazybeds

0

200 m 1:5000

fig.

5.41  Plan of extensive second–third century a.d. field system at Eye Quarry, Cambs (Patten 2004)

Within the current project, there are some 300 sites from the Central Belt region with records of excavated field systems, distributed as shown in fig. 5.39. Some of these are explored in relation to the surrounding NMP cropmark data in the case study below, but, for the most part, analysis is restricted to an assessment of chronology and associated context, albeit framed within the variable landscape zones that make up the region. Dating the field systems is problematic for the same reasons as for trackways noted above – namely, the lack of associated diagnostic dating material. Nevertheless, over 80 per cent of the excavated field systems in the region had some approximate dating, with a steady rise in numbers until a peak in the second century a.d., unsurprisingly quite similar to the chronological

Romanch5.indd 181

trajectory of the overall settlement pattern. The individual landscape zone patterns, however, do not always follow those of settlements, as demonstrated by the Upper Thames and Avon Vales, where excavated field systems apparently reached their highest number during the third century a.d. (whereupon many start to silt up), whereas settlement numbers peaked during the early second century (see above, p. 148). The chronological maps of field systems shown in fig. 5.39 reveal distinct concentrations in the east of the Central Belt during the late Iron Age, and a relative scarcity to the west at this time. There is subsequently a big expansion of field systems across the region in the later first and second centuries a.d., with many of the eastern zones (e.g. Trent Valley and Rises, Fens and West

15/09/2016 17:52:54

182

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Anglian Plain) showing field systems at their most extensive during this period, partly because of a significant increase in agricultural exploitation of the Fens and the higher claylands landscapes. The decline in the amount of evidence for field systems in these eastern areas during the later Roman period may be due to greater use of less archaeologically visible forms of boundary (e.g. hedges), as noted above with regard to trackways in the Trent Valley (Taylor 2006, 11), but increased flooding on lower lying ground in the Fens might also have been a factor.

Half of the excavated Central Belt field systems recorded in the current dataset were explicitly associated with farmsteads, with approximately half of these seen to be integrated with the settlement (e.g. Tubney Wood Quarry, Oxon: Simmonds et al. 2011; fig. 5.40) and others noted some distance away (e.g. Parnwell, Peterborough: Webley 2007). Without sufficient area of excavation or associated cropmark or geophysical evidence, the extent of the field systems around individual settlements remains largely unknown, but it was certainly not the case that all landscapes were ‘filled in’ with integrated

12 10

no. of sites

8 6 4 2 0

fig.

Romanch5.indd 182

5.42.  Distribution and chronology of sites with evidence for lazybeds in the Central Belt region

15/09/2016 17:53:02

THE CENTRAL BELT

networks of fields as found throughout most of the region today. Multi-hectare excavated sites like Tubney Wood, for instance, illustrate that field boundary ditches are often just traced in the immediate area around the settlement, and then much of the landscape appears ‘open’, although of course hedges may have provided further division. Occasionally more extensive landscapes of what appear to be very organised and integrated field systems are subject to excavation, such as that on a gravel terrace near the Fen edge at Eye Quarry, Cambridgeshire (Patten 2004; fig. 5.41), and yet even in such landscape zones, there are still large areas of land that do not appear to have had any division (see case study below). The functions of most of these field systems remain unknown, but are assumed to be for a mix of arable and pastoral use, sometimes with other features such as watering holes and potential fodder stands suggesting an emphasis on stock. One feature that seems to be almost exclusive to parts of the Central Belt (a few others are known mainly in the East region) is the system of narrow, parallel trenches known as ‘lazybeds’, as found within one of the fields at Eye Quarry and shown on fig. 5.41. Although not numerous, with just seventeen examples excavated to date within the region, these lazybeds are concentrated in eastern parts of the Central Belt, around the Fen edge and in the main river valleys. One of two areas of Roman lazybeds found during recent large-scale excavations in north-west Cambridge had evidence of seasonal irrigation capacity, with water introduced into downslope feeder-channels from wells on top of the gravel ridge (Evans pers. comm.). Unfortunately, there is little environmental evidence associated with such features to help determine function, although it is likely that they are associated with growing horticultural crops such as cabbages, carrots, etc. Other systems of parallel bedding-trenches with associated postholes

183

have been excavated across 7.5 ha at Wollaston in the Nene Valley, and had pollen evidence to support the existence of viticulture (Brown et al. 2001), although this remains distinct from the lazybed system (see vol. 2). Chronologically, lazybeds seem most prevalent in the early to mid-Roman period, and particularly the second century a.d., at the height of settlement and field system expansion in that eastern part of the Central Belt (fig. 5.42). Their use within this zone is just another example of the many broad differences apparent in rural settlement and land use between the eastern and western parts of the region. SETTLEMENT HIERARCHIES: THE SOCIAL AND ECONOMIC BASIS OF SETTLEMENTS The emphasis so far in this chapter has been on the morphology, chronology and landscape context of rural settlements within the Central Belt, and very clear patterns have emerged indicating both elements of similarity and heterogeneity. There appears to have been a series of quite distinct broad settlement types, albeit with blurred typological boundaries, but in order to determine if there were any correlating social and economic differences, the evidence from associated material culture and environmental remains must be taken into account. MATERIAL CULTURE The Central Belt is especially rich in late Iron Age and Romano-British material culture, as seen by the distribution of Portable Antiquities Scheme (PAS) finds in the region (Brindle 2014, 3, fig. 3). Over 81 per cent of Central Belt sites produced records of ‘small finds’ of some sort, while 70 per cent have quantified records of ceramic assemblages.

% of sites (n=1090)

70 60 50 40 30 20 10 0

fig.

Romanch5.indd 183

5.43.  Frequency of major artefact categories on all sites in the Central Belt region

15/09/2016 17:53:03

184

THE RURAL SETTLEMENT OF ROMAN BRITAIN

% of sites with object present

90

Cotswolds (n=93)

Thames and Avon Vales (n=118)

The Fens (n=53)

80 70 60 50 40 30 20 10 0

fig. 5.44.  Frequency of major artefact categories recovered across selected landscape zones in the Central Belt region

This does not of course mean that all remaining sites were bare of any material culture, but that the specialist information was not available (see methodology problems, Ch. 1). Pottery in particular suffers from variable levels of recording, as such material was noted on nearly all (c. 95 per cent) sites in the Central Belt, but not always quantified. The presence (if not quantification) of certain fine and specialist wares was usually noted with greater consistency, indicating that Gaulish samian was recovered from at least 66 per cent of sites, mortaria from 51 per cent and amphorae from 30 per cent. These wares were found throughout the Central Belt landscape zones, though there were notable differences in their occurrence at different site types (see below, p. 185). Aside from pottery, the frequency of occurrence of other finds categories across the region is shown in fig. 5.43. Methodological issues associated with such quantification will be explored in some depth in subsequent volumes, but notwithstanding these problems, it is clear that coins dominate other small finds, being recorded at almost 60 per cent of sites. Brooches form by far the largest category of personal adornment, while other relatively commonly recorded finds in the region comprise more utilitarian object such as knives and tools, as well as objects associated with food processing (quernstones) and textile processing (spindlewhorls, loomweights etc.). Other, more specialist objects, such as those concerned with religious practice, recreation, writing and lighting equipment all remain comparatively rare, although there are differences in the proportions of most of these finds categories within different landscape zones and between different classes of settlement. The geographic variation in the presence/ absence of finds within the Central Belt is highlighted by the three different zones selected

Romanch5.indd 184

for comparison in fig. 5.44. In the Cotswolds, for instance, over 80 per cent of settlements had records of at least one coin, compared with just 47 per cent in the Fens, although this must be at least partly because of the different chronological patterns in these zones (i.e. settlement numbers in the Cotswolds reach their height in the later Roman period when there is the greatest evidence for coin use). Most other landscape zones, such as the Thames and Avon Vales, were around average for the region, though there does appear to have been a distinct drop off in the central northern zones of the region (Midland Clay Pastures and Trent Valley and Rises), where coins were found on just 30–40 per cent of settlements. The Fenland was one of just two landscape zones in the region where coins were not the dominant find type, the other being another low-lying wetland area, the Gwent Levels. In both of these zones, brooches were encountered on more sites than coins, possibly hinting at differences in the economy of these wetland areas, although, as stated above, chronological factors were also important, since there was fairly rapid decline in settlement numbers in both areas during the fourth century a.d., when coin finds are otherwise at their most common and brooch wearing had declined. These zones are also comparatively well represented by objects associated with food and textile processing, while most other finds categories are fairly sparsely represented. The Cotswolds and other mainly ‘upland’ zones such as the Berkshire and Marlborough Downs and the Chilterns, on the other hand, have consistently higher than average percentages of most find types, particularly items associated with recreation, writing and religion. This is mainly due to the relatively high representation of villas in these zones, yet such objects do also seem to percolate down the social scale in these areas.

15/09/2016 17:53:04

THE CENTRAL BELT

185

100 90

% of sites present

80

roadside settlement (n=82)

70 60

villa & farmstead/villa (n=156)

50

farmstead (n=808)

40

farmstead final form complex (n=121)

30

farmstead final form enclosed (n=96)

20 10 0 samian fig.

amphora

5.45.  Frequency of samian, amphora and mortaria recovery within settlement types in the Central Belt region

The differences noted in the occurrence of object types across the landscape zones of the Central Belt in part reflect differences in the types of settlement to be found there. The relationship between object categories and settlement types in the Central Belt is shown in figs 5.45 to 5.47. Pottery of course remains the most commonly found artefact across all sites, though there are major differences between site assemblages that may relate to aspects of settlement economy and social status (cf. Booth 2004). As noted above, only certain ceramic wares have been systematically recorded and analysed in this volume, and these are shown for the Central Belt in fig. 5.45. The distribution and use of samian pottery in Roman Britain and other western provinces has been recently comprehensively reviewed by Willis (2011), primarily as an index of social and economic processes and practice. He demonstrated contrasting patterns of samian consumption at different categories of site, concluding this to be ‘a sensitive indicator of their status and identity’ (ibid., 227). Within his study, most ‘rural sites’ were grouped together, yet, even in terms of basic presence/absence of samian, it is clear that there is significant divergence between different forms of rural settlement (see also other regional chapters). Within the broad settlement categories, roadside settlements in the Central Belt, unsurprisingly, have the highest percentages of sites with samian present (92 per cent), followed by villas (a site category which here include farmsteads that develop into villas) and then farmsteads. However, within the latter there are major differences in classified types, with the proportion of complex farmsteads receiving samian almost matching that of roadside settlements, and far outstripping enclosed farmsteads. Although the quantity of samian at complex farmsteads may be much less than at nucleated sites (see vol. 2 for analysis),

Romanch5.indd 185

mortaria

they clearly had much greater access to this pottery than enclosed farmsteads, or even apparently than villas, although this probably results in part from the poor recording of ceramic assemblages in older excavation reports on villas. Similar basic patterns occur with the presence of another imported ware, amphorae, along with specialist ‘Roman’ mortaria. In all three ceramic categories, complex farmsteads out-perform enclosed farmsteads, suggesting certain socioeconomic differences and greater levels of connectivity, also hinted at by other finds categories. The frequency of other object categories within different settlement types is shown in figs 5.46 and 5.47. As with pottery there is a clear hierarchy of frequency of occurrence for most object categories, with roadside settlements at the top, then villas and farmsteads, although there are certain finds groups where distinctions are less pronounced. Roadside settlements and villas, for instance, are quite proximate in terms of the percentages of these sites having evidence for coins, tools/knives, household, security and equine/ transport objects, though the numerical density of these finds may still be quite different (see below, p. 187). Roadside settlements particularly stand out with items of food processing, which occur on c. 60 per cent of sites as opposed to c. 35 per cent of both farmsteads and villas. The association of millstones with roadside settlements is particularly pronounced in this region; they occur on 23 per cent of such sites, as opposed to 17 per cent of villas and just 5 per cent of farmsteads. This does suggest that centralised cereal processing was more likely to have been carried out at nucleated settlements and villa estates, though there are also notable differences within farmstead types, with 23 per cent of complex farmsteads containing millstones as opposed to 3 per cent of enclosed

15/09/2016 17:53:04

186

% of sites with object present

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig.

villa & farmstead/villa (n=156)

farmstead only (n=808)

5.46.  Frequency of major artefact categories on different types of rural settlement in the Central Belt region

% of sites with object present

fig.

roadside settlement (n=82)

100 90 80 70 60 50 40 30 20 10 0

100 90 80 70 60 50 40 30 20 10 0

villa & farmstead/villa (n=156)

farmstead final form complex (n=121)

farmstead final form enclosed (n=96)

5.47.  Frequency of major artefact categories on villas and farmstead types in the Central Belt region

High status objects at sites with >100 coins 100%

'Rich' Farmsteads (n=17)

'Rich' Villas (n=25)

Other Farmsteads (n=531)

Ligh ng Equipment

Military Fings Security object and Weaponry

% of sites with object present

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

Wri ng Equipment

fig. 5.48.  Occurrence of selected high status objects on ‘rich’ farmsteads and villas compared to other farmsteads in the Central Belt (as % of site types)

Romanch5.indd 186

15/09/2016 17:53:08

THE CENTRAL BELT

table

5.5: quantity

187

of object types by settlement category expressed as mean number of objects per

hectare of excavation (excludes evaluations and watching briefs)

Object type

Roadside Villas All farmsteads settlement (72 sites) (474 sites) (59 sites)

Coin

427.09 358.22 10.37

Household object

50.03

Brooch Food processing Knife/tool Hairpin

4.52 0.55

28.10 27.77 1.93

2.49

1.71

22.47

2.08

1.60

20.15 31.57 1.50

1.03

2.71

14.37 38.03 0.58

12.54

0.54

12.20

Enclosed farmsteads (65 sites)

0.57

2.09

0.49

0.48

Recreation object

9.35

5.16

0.09

0.07

0.16

Toilet/cosmetic implement

9.26

11.05

0.47

0.53

0.27

Weighing object

8.76

3.01

0.21

0.28

0.18

Bracelet

8.58 19.60 0.65

0.82

0.48

Textile-processing

7.69 12.89 0.81

0.59

2.03

Agricultural tool

6.52

5.26

0.21

0.20

0.25

Finger ring

6.08

9.03

0.29

0.34

0.21

Security object

4.72

7.98

0.20

0.21

0.14

Military fitting and weaponry

4.19

3.14

0.50

0.30

3.26

Equine/transport equipment

3.89

4.15

0.14

0.12

0.16

Writing equipment

3.15

7.13

0.12

0.13

0.09

Religious object

2.80

1.68

0.12

0.17

0.04

Lighting equipment

0.74

0.70

0.04

0.07

0.00

farmsteads. In this way, as in many others, complex farmsteads appear to perform on a similar economic if not social level to villas. figure 5.47 compares the percentage of villas and major classified farmstead types in the Central Belt on which different object type categories are recovered. As just stated, most object categories occur in similar percentages of complex farmsteads and villas, and can even be better represented at complex farms in some cases. Major differences, however, lie with objects associated with areas like recreation, religion and the household (e.g. furniture fittings), pointing to quite different social values between occupants of the different types of settlement, which also correlates with the architectural differences. Simple enclosed farmsteads generally have a much more restricted range of object types than other sites, with material from just a few categories like textile processing and food processing occurring at levels similar to those in other smaller rural settlements. The foodprocessing equipment in the simple enclosed farmsteads consists almost entirely of quernstones, which are widespread. The somewhat anomalously high percentage of ‘military’ objects from enclosed farmsteads in the region is largely because

Romanch5.indd 187

19.48

Complex farmsteads (98 sites)

slingstones are recorded under this category (see Appendix 3), as otherwise objects of Roman military equipment remain very rare. The graphs shown in the figures above denote only the presence or absence of particular object types within different forms of settlement, and indicate nothing about their relative frequency. Such figures are heavily affected by factors such as area of excavation, so table 5.5 presents the mean number of objects from the different site types by hectare of excavation. Coins were the most numerous object type, particularly in roadside settlements and to a lesser extent in villas, where they were overwhelming more common than in farmsteads by a factor of up to 40. Complex farmsteads again out-performed enclosed farmsteads, although the paucity of coins on the latter is at least partly to do with the comparative rarity of these sites in the later Roman period when coin use was at its height in the countryside. Nevertheless, even in complex farmsteads the numbers of coins are generally very low, suggesting that activities involved with monetary transactions (e.g. marketing or tax collection) were very rare on such sites. There are of course exceptions to this, with a small number of farms producing much

15/09/2016 17:53:08

188

THE RURAL SETTLEMENT OF ROMAN BRITAIN

higher numbers of coins, usually also alongside much larger numbers of other object types of perceived ‘high status’, as shown in fig. 5.48. An example is a substantial complex farmstead (or perhaps a village?) at Cleveland Farm, Ashton Keynes in the Upper Thames Valley, which produced 1339 coins along with many items of personal adornment and objects associated with religion and writing, although the quantity of objects recovered here was greatly aided by widespread use of metal detectors (Powell et al. 2008a). Clearly then, there were some non-villa farming settlements in the Central Belt that in terms of material culture at least, seem to have performed at much higher social and economic levels than others. ENVIRONMENTAL REMAINS: THE ANIMAL AND PLANT ASSEMBLAGES

mean %NISP

An important measure of the economic and social structure of rural settlements can also be provided by the study of associated environmental material (animal and plant remains), which can provide crucial information on the variability of farming practices. This regional variability in farming has been recently demonstrated across parts of central England for the early medieval period, where it was suggested to have been influenced by a combination of geological and cultural factors (Rippon et al. 2014). As with the material culture, the environment data for this project will be fully analysed in subsequent volumes, and so just a brief outline is presented here, in relation to particular settlement types and landscape zones of the Central Belt.

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

pig

Faunal remains The animal bone assemblages for the Central Belt region are the richest of any in Britain, partly due to soil conditions which are generally favourable for the preservation of such material. A recent comprehensive account of Iron Age and Roman faunal assemblages in central England has highlighted distinct chronological and regional patterns in the data (Albarella et al. in prep.), much of which is also reflected in the current dataset, which consists of 461 assemblages of over 100 NISP. Among the major domesticated species (cattle, sheep/goat and pig), cattle are dominant across the region (53 per cent of main domesticates), followed by sheep/goat (38 per cent) then pig (9 per cent). The proportion of sheep/goat is somewhat larger than Albarella’s patterns for central England, largely because his dataset included major urban centres, where sheep/goat are generally far less numerous (Albarella et al. in prep., fig. 6.2). There are substantial variations in this regional pattern, seen in fig. 5.49, with some (mostly valley) landscapes having greater proportions of cattle (up to 65 per cent in the Bristol, Avon Valley and Ridges) and much higher percentages of sheep/ goat in zones like the Fens and the Cotswolds (43 per cent and 47 per cent respectively). The Cotswolds and the Berkshire and Marlborough Downs remain the only landscape zones with higher proportions of sheep/goat than cattle, due to the particular suitability of these ‘upland’ areas for sheep farming. Pigs remain a fairly minor component of the assemblage in all regions, though they range up to 15 per cent in some hillier

sheep/goat

cattle

fig. 5.49.  Relative frequency of major livestock taxa across all Central Belt landscape zones where there are five or more assemblages of >100 NISP (mean percentages from sites in each zone)

Romanch5.indd 188

15/09/2016 17:53:09

THE CENTRAL BELT

189

explained by the late Iron Age emphasis in the date of enclosed farmsteads, though the patterns barely seem to change for those sites occupied throughout the Roman period, and so seem to relate more to different farming regimes, undoubtedly associated with their location in significant numbers in higher landscape zones. The much greater dominance of cattle on complex farmsteads reflects the development of these sites in the major river valleys, with large areas of lush meadow land pasture. The very form of complex farmsteads appears to reflect an emphasis on livestock management, and cattle herds seem to have been an important aspect of this, used for a variety of purposes including traction for ploughing arable fields (discussed in vol. 2). Villas in the region also seem to have been largely cattle dominated, though they have by far the highest percentage of pig, a pattern seen too in urban assemblages (Albarella et al. in prep.), undoubtedly reflecting a more Mediterranean-influenced diet. These general patterns, by region and settlement type, of course disguise a variable situation, with a number of sites exhibiting ‘extremes’ that may be dictated by specific economic, cultural, ritual or even personal factors. An excavation at Nash on the Gwent Levels, for example, revealed a midRoman agricultural landscape incorporating areas of possible seasonal occupation, with a substantial animal bone assemblage completely dominated (93.2 per cent NISP) by cattle (though this did include a small number of animal burials), which was interpreted as evidence for a specialist cattle management and production centre, potentially for local military supply (Meddens and Beasley

zones (e.g. Chilterns), partly because of their association with villas during the Roman period (and possibly the greater presence of woodland), but mostly owing to their greater numbers in these zones during the late Iron Age. The proportions of cattle, sheep/goat and pig do not stay constant over time, as has been demonstrated previously on numerous occasions (e.g. King 1999; Albarella 2007). In particular it has often been noted that the proportion of cattle increases during the Roman period, and this pattern certainly exists in the Central Belt data, though it is only really seen from the second century onwards (fig. 5.50). This increase is observed across most landscape zones, particularly in the major river valley areas, although in the Cotswolds there actually appears to be an increase in sheep/goat at this time, perhaps linked to greater pastoral specialisation within differing landscape zones. The Fenland also appears to have had steadily increasing sheep/goat proportions from the early Roman period, becoming the dominant taxon in the third–fourth centuries a.d. Some of the differences noted in the faunal assemblages across the landscape zones are clearly associated with the patterns of settlement found in those areas, as noted in relation to the correlation between villas and relatively high representation of pigs. figure 5.51 shows these different associations across the main settlement types, with cattle clearly dominating over sheep/goat in complex farmsteads and villas (and indeed in most unclassified farmsteads), whereas the proportions are far more equal in nucleated sites and enclosed farmsteads. The latter association may be partly

Central Belt (n=461) 60

mean % NISP

50 40 cattle

30

sheep/goat

20

pig

10 0 LIA (1st C BCmid 1st C AD)

60

LIA/ER (1st C BC/AD)

ER (1st-2ndC MR (2nd-3rdC LR (3rd-4thC AD) AD) AD)

Cotswolds (n=37)

60 50 mean % NISP

mean % NISP

50

Thames & Avon Vales (n=72)

40 30 20 10

40 30 20 10

0

0 LIA (1st C LIA/ER (1st ER (1stBC-mid 1st C BC/AD) 2ndC AD) C AD)

MR (2nd3rdC AD)

LR (3rd4thC AD)

LIA (1st C LIA/ER (1st ER (1stBC-mid 1st C BC/AD) 2ndC AD) C AD)

MR (2nd3rdC AD)

LR (3rd4thC AD)

fig. 5.50.  Relative frequency of major livestock taxa over time in the Central Belt, Cotswolds and Thames & Avons Vales (NISP mean percentages)

Romanch5.indd 189

15/09/2016 17:53:10

190

THE RURAL SETTLEMENT OF ROMAN BRITAIN cattle

70

sheep/goat

pig

60

mean %NISP

50 40 30 20 10 0 farmstead (complex) (n=100)

farmstead (all) (n=315)

villa (n=45)

village (n=12)

roadside settlement (n=56)

farmstead (enclosed) (n=58)

fig. 5.51.  Relative frequency of major livestock taxa across different settlement types in the Central Belt (mean percentages from site types)

2001, 167). Most of the assemblages with very high proportions of sheep/goat (70 per cent NISP+) were from religious sites such as Uley or Haddenham, though the roadside settlement at Higham Ferrers in the Nene Valley is exceptional, with sheep amounting to 79 per cent of the main domesticates even in the non-shrine assemblage (Strid 2009). The sheep slaughter patterns and concentration of ewes from this assemblage suggested a focus on sheep dairy production in this area (ibid. 289). As stated above, pig rarely exceed 15 per cent of NISP, though a substantial late Roman assemblage from the main villa building at Castle Copse in Wiltshire contained 66 per cent pig bones (Hostetter and Howe 1997), and a much larger assemblage (6824 NISP) from a conquest-period shrine at Hallaton in Leicester was almost entirely of pig, undoubtedly reflecting its use in religious rituals at the site (Score 2011). Substantial variation may also be seen with other species, both domesticated and wild, though for the most part these rarely form a significant component of any assemblage. Horses are most dominant on farmsteads, particularly complex farmsteads, where there is also greater evidence for horse breeding, discussed in volume 2. Dogs form a constant but low percentage of faunal assemblages at most sites, although they are more readily found as associated bone groups (ABGs), whether these are perceived to be of a ritual nature or not (cf. Morris 2011). Domestic cat bones have been recorded on 83 sites across the region, always in very low numbers, but seemingly far more common during the later Roman period, particularly in villas and roadside settlements. Wild mammals are exceptionally rare at most types of settlement, though deer (especially red deer) in particular stand out at villas, where they are argued to represent hunting as a sign of high social status (Grant 1989; Albarella et al. in prep.).

Romanch5.indd 190

Plant remains A total of 536 rural settlements from the Central Belt region contained some form of archaeobotanical report, usually on charred plant remains, though some had waterlogged samples. At a basic level of presence/absence, the range of main arable crops and selected other (mostly) cultivated plants is presented in fig. 5.52, with some regional variations expressed in fig. 5.53. As expected, spelt wheat (Triticum spelta) is the crop found most often on rural sites in all Central Belt landscape zones except in parts of the north-east (Lincs and Notts), where barley (Hordeum vulgare L.) appears on slightly more sites. However, in terms of relative abundance within charred plant assemblages, spelt is overwhelmingly dominant in all areas, with higher levels of barley only found on fourteen sites, all farmsteads. Emmer (Triticum dicoccum) and free-threshing wheats (Triticum aestivum) usually form quite minor components on c. 30 per cent of sites, though on occasion they are present in greater quantities, which in the case of emmer wheat most often occurred within late Iron Age–early Roman assemblages, such as at the farmstead at Wardy Hill on a Fen island in Cambridgeshire (Evans 2003a). The generally greater frequency of emmer wheat in this eastern part of the Central Belt is notable, occurring on almost 50 per cent of sites in the Fens and West Anglian Plain (see case study below), with a similar incidence of free-threshing wheat. As with emmer the latter rarely forms the most dominant arable crop in any zone, being recorded as such on just eleven sites, which often date to the later Roman period such as the farmstead at West Street, Bedminster, Bristol (Young 2006). The only other frequently encountered crop is oats (Avena sativa), though this usually occurs in fairly small quantities and with uncertainty as to whether it is cultivated or wild. A farmstead at Old Parks

15/09/2016 17:53:11

THE CENTRAL BELT

191

% of sites present (n=536)

80 70 60 50 40 30 20 10 0

% of sites present

90 80 70 60 50 40 30 20 10 0

% of sites present

fig. 5.52.  Percentage presence of cultivated plant taxa on sites in the Central Belt region where archaeobotanical remains have been identified (includes both charred and waterlogged assemblages)

90 80 70 60 50 40 30 20 10 0

Lincolnshire and Rutland Limestone (n=25) Trent Valley and Rises (n=54)

The Fens (n=38) West Anglian Plain (n=142)

fig. 5.53.  Percentage

presence of cultivated plant taxa on sites in selected landscape zones in the Central Belt region (includes charred and waterlogged remains)

House, Ashby de la Zouch, Leicestershire, did produce a large quantity of oat grains, particularly during the Roman phase, and it seems likely to have been cultivated in this instance, perhaps as a fodder crop (Jones and Dingwall 2002). A range of minor crops is also occasionally encountered across the Central Belt, such as flax (Linum usitatissimum), used for oil and for making textiles, which was found on 41 sites, including some from possible flax retting pits (used to separate plant fibre from stem) in a late Roman farmstead at Old Shifford Farm, Standlake, Oxon (Hey 1995). The cultivation of flax appears almost exclusively limited to the Fens and major river valleys, occurring on over 15 per cent of sites with archaeobotanical assemblages in the Upper Thames. Rye (Secale cereale), probably used as a

Romanch5.indd 191

food grain, cover crop and forage crop, was identified on 54 sites, more widely distributed across the Central Belt, though most commonly encountered in the Fens. Samples containing rye appear to be mostly of the mid- to late Roman period, including charred remains from a fourthcentury granary at the villa at Great Weldon, Northants (Smith et al. 1990). Other food plants include pulses (mostly Celtic bean, broad bean, pea and lentil), fruits (e.g. sloe, crab apple, plum, cherry, blackberry, etc.) and a range of other horticultural crops (e.g. carrots, celery, beets, cabbages, etc), though the survival of the last two in particular is generally dependent upon waterlogged conditions and so they are probably quite under-represented in most areas (Van der Veen et al. 2007, 193). Although still rare,

15/09/2016 17:53:12

192

THE RURAL SETTLEMENT OF ROMAN BRITAIN

% of sites

(a)

% of sites

(b)

90 80 70 60 50 40 30 20 10 0

100 90 80 70 60 50 40 30 20 10 0

roadside settlement (n=44) farmstead/villa & villa (n=48) farmstead (n=425)

farmstead complex (n=91) farmstead enclosed (n=53)

fig. 5.54.  Percentage presence of cultivated plant taxa in (a) major settlement types and (b) farmstead types within the Central Belt Region (includes charred and waterlogged remains)

horticultural crops are most commonly found on sites of the West Anglian Plain, which correlates well with the evidence for bedding trenches in this area, noted above. The sites with evidence for the widest range of fruits, pulses and horticultural crops generally date to the mid–late Roman period; these assemblages represent culinary innovations, incorporating many of the 50 new food plants identified for the Roman period by van der Veen, who argued that they were used in the creation and maintenance of group and individual identities (Van der Veen 2008, 106). Most of these newly imported ‘elite’ food varieties were seen to be more common on military sites and major urban centres, where they occurred soon after the conquest, filtering out to higher status rural sites (villas) during the later Roman period (ibid.). Further differences in the archaeobotanical assemblages of rural settlement types can be demonstrated in the current dataset for the Central Belt (fig. 5.54). For the most part spelt wheat is dominant in all settlement types, though differences can be seen between complex and enclosed farmsteads, spelt occurring in 93 per cent of the former’s archaeobotanical assemblages as opposed to less than 70 per cent of the latter’s. Emmer wheat on the other hand forms a consistent 45 per cent of both. Free-threshing (or bread/club) wheat is consistently more often identified at roadside settlements (43 per cent) and villas (42 per cent), while among farmstead types, it occurs in complex farmsteads at a similar level (40 per cent), a much

Romanch5.indd 192

higher incidence than on enclosed farmsteads (24 per cent). A similar situation can be seen with most minor food plants, particularly rye and pulses, and also horticultural crops in general, although here there is a more distinct representation on roadside settlements. Some of the differences noted between the food-plant assemblages of different site types can be linked with chronological factors, with enclosed farmsteads in particular being more common in the late Iron Age/early Roman period. Nevertheless, it does appear that roadside settlements, villas, and to some extent complex farmsteads cultivated, or at least consumed, a wider range of food crops, and many of the inhabitants of these sites probably had a much more varied diet than those at other rural settlements. These distinctions can probably be associated with differences noted in the faunal assemblages and material culture, suggesting a complex social and economic hierarchy of settlements that varied across different landscape zones and evolved over time. CASE STUDY: THE CAMBRIDGESHIRE FEN EDGE Within the entire Central Belt region there are a few landscapes that stand out not only in terms of the density of excavation records, but also in how recently these records have been published/ disseminated. The Cambridgeshire Fen edge is one of these areas, and has been chosen for a more detailed case study because there is now a growing

15/09/2016 17:53:12

THE CENTRAL BELT

fig.

5.55.  The Cambridgeshire Fen edge case study area

wealth of modern, well-excavated and recorded data, which should enable us to examine at a finer resolution some of the intricate and variable patterns of rural settlement and land use. A c. 40 × 30 km (c. 1200 km²) zone has been selected, straddling two landscape zones, the Fens and the West Anglian Plain, and incorporating two Roman walled small towns, at Godmanchester (Durovigutum) and Cambridge (Duroliponte). The zone contains 108 records from the current dataset (within the Central Belt), comprising 72 farmsteads, four villas, four roadside settlements, five villages, five pottery production sites, two religious sites and twelve field systems (fig. 5.55). Prior to the introduction of PPG16 in 1990, just one of these records had been disseminated, and almost 80 per cent are dated between 2006 to 2014, albeit a few of them publishing accounts of much earlier excavations (e.g. New Fen Drove, Earith; Evans 2013b, 459–64). This mass of newly reported excavation data can be used to complement earlier wide-scale fieldwalking and cropmark surveys, together suggesting that Iron Age/Roman settlements occurred at very regular intervals across much of the landscape (suggested as every c. 300–500 m in Evans 2008, 181–6).

Romanch5.indd 193

193

The Fenlands in particular have had a very long and fruitful tradition of fieldwalking and cropmark surveys, revealing extensive areas of Roman settlement and field systems linked by trackways and waterways (Hall and Coles 1994). This landscape has been the subject of much academic debate, particularly with regard to the postulated existence of a Roman imperial Fenland estate, first fully articulated by Salway (1970) and expanded upon by others such as Jackson and Potter (1996) and Malim (2005). The basis upon which such interpretation is founded – the construction of apparently centralised and widespread drainage systems, mass expansion of settlement during the second century a.d., a lack of villas, and desire to control the salt industry – has been heavily criticised by Millett (1990, 120), Taylor (2000), and Fincham (2002). In particular it has been argued that, contrary to earlier thinking, much of the southern and western Fen margins were already well populated in the late Iron Age, and the expansion of settlement into more marginal areas during the early Roman period was mainly due to the natural lessening of salt-water flooding (Taylor 2000, 651). Nevertheless, even if the concept of a directly run imperial estate no longer

15/09/2016 17:53:19

194

THE RURAL SETTLEMENT OF ROMAN BRITAIN

seems tenable, it has still been argued that Roman authorities took a very active interest in the exploitation of the Fen landscapes (Fincham 2002, 75), and there are some signs of this from recent excavations on the Fen edge. In particular, there is evidence for large-scale drainage of the Rhee Lake marsh-inlet at Colne Fen in the early Roman period (Evans 2013a, 50), which must have been a considerable undertaking, and there are ‘military-type’ raised sill granaries/warehouses at Camp Ground, Bullock’s Haste and Waterbeach (cf. Evans 2013b, 274). It may have been the case,

as suggested by Mattingly (2006, 385), that some part of the region was retained as ager publicus or ‘public land’, with native Britons allowed occupancy in return for rents and taxes levied in terms of produce such as salt and animal products. The case study area incorporates the southern edge of this well-studied Fenland landscape, along with the somewhat less well synthesised landscapes further south, including the gravel terraces of the Rivers Ouse and Cam, and the higher intervening claylands, much of which has been subject to exceptionally large-scale recent excavation. Recent

N

19th Century quarries

0

200 m 1:5000

fig. 5.56.  Huge area excavation of Romano-British rural settlement in north-west Cambridge (Evans and Newman 2010)

Romanch5.indd 194

15/09/2016 17:53:25

THE CENTRAL BELT

evaluations and excavations by Cambridge Archaeological Unit in north-west Cambridge, for example, covered 150 ha and revealed at least five Roman settlements to the south of the main Cambridge to Godmanchester road, most densely packed along a band of gravel (to maximise ‘dryliving’; Evans pers. comm.) between zones of heavier clay (http://www-cau.arch.cam.ac.uk/ NWC.htm) (fig. 5.56).

5.58). This is in contrast to the area around Godmanchester, which had no known Iron Age settlement background (Jones 2003, 186), and around which late Iron Age settlement is comparatively rare. Further south along the Ouse Valley there is increasing evidence for later Iron Age occupation, both along the river terraces and on the surrounding higher claylands, which have been shown by limited excavation and wider scale field surveys to have been quite densely populated at this time (Timby et al. 2007b, 55). A recent road scheme evaluation along a 28 km corridor north-west of Cambridge indicated that the claylands in this area were first colonised at any scale during the middle Iron Age, with communities starting to be able to farm the heavy soils of these highly fertile landscapes (Evans and Standring 2012, 100–2). Further recent road scheme and housing developments on the boulder claylands west of Cambridge have also revealed extensive, managed landscapes of later Iron Age date, with over half of the fourteen identified sites in this area having some kind of demonstrable pre-conquest origins (Abrams and Ingham 2008; Wright et al. 2009). The majority of classified farmsteads in this clayland zone were of simple enclosed type (e.g. Poplar Plantation; fig. 5.59), while complex farmsteads were more widely distributed across the claylands, river terraces and Fens, usually comprising irregular systems of conjoined enclosures such as at Little Common Farm (fig. 5.59). Around 50 per cent of all enclosed farmsteads within the case study area did not

SETTLEMENT PATTERNS AND CHRONOLOGY The broad chronological patterns of the 85 settlements (farmsteads, villas and nucleated sites) in the study area are shown in fig. 5.57, with wider landscape developments illustrated in figs 5.58, 5.60 and 5.62. Although there is widespread evidence for middle Iron Age occupation in the area (e.g. Timby et al. 2007b, 54–6; Evans 2013a), just 22 settlements in our dataset had evidence of activity continuing from this period into the late Iron Age or beyond, these being a mix of open, enclosed and complex farmsteads. There then appears to have been a significant increase in the number of ‘new’ settlements during the late Iron Age, with some of these, such as Tort Hill (Ellis and Hughes 1998) being established in the context of existing field systems. The distribution of late Iron Age settlement shows a relatively high level of occupation within the Fenland zone (71 per cent of all Fen sites in the study area in use at this time), as well as further south in the vicinity of Cambridge, which is known to have had Iron Age origins (Alexander and Pullinger 1999, 17) (fig. (a)

% of sites 'abandoned' during phase

(b)

% of sites 'established new' during phase

100 90 80 70 60 50 40 30 20 10 0

60 50

% of sites in use (n=85)

no. of settlements in use

195

40 30 20 10 0 M-LIA

LIA

AD50 - AD100 AD150 AD200 AD250 AD300 AD350 100 -150 -200 -250 -300 -350 -400

(c)

Open

no. of farmsteads in use

16

Enclosed

LIA

AD50 - AD100 AD150 AD200 AD250 AD300 AD350 100 -150 -200 -250 -300 -350 -400

Complex

14 12 10 8 6 4 2 0 M-LIA

LIA

AD50 -100 AD100 150

AD150 200

AD200 250

AD250 300

AD300 350

AD350 400

fig. 5.57.  Chronological patterns of case study settlements: (a) number of settlements in use; (b) settlement ‘establishment’ and ‘abandonment’; (c) farmstead types in use

Romanch5.indd 195

15/09/2016 17:53:28

196

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

5.58.  Settlement development in the case study area in the late Iron Age with selected sites named in the text a

b N

0

100 m 1:2000

fig. 5.59.  Late Iron Age farmsteads on the Cambridgeshire claylands (Wright et al. 2009): (a) Complex farmstead at Little Common Farm; (b) Enclosed farmstead at Poplar Plantation

appear to continue much beyond the mid-first century a.d., and of those that did, nearly all were confined to the central boulder claylands. The post-conquest period saw no great rise in overall settlement numbers, though this disguises a period of apparent major settlement upheaval, with c. 20 per cent of the 48 sites in use being

Romanch5.indd 196

newly established during this time and a further 20 per cent being abandoned by the end of the first century a.d. This remains the period with the highest levels of abandonment or at least dislocation until the fourth century a.d., and while it does not appear to have been catastrophic, it is not unreasonable to suppose some kind of

15/09/2016 17:53:35

THE CENTRAL BELT

197

fig. 5.60.  Settlement development in the case study area in the early Roman period with selected sites named in the text

association of this pattern with either the Claudian conquest or the Boudican revolt and the subsequent increased military presence that has been argued for the area (Malim 2005, 42; though see Ch. 6). All of those abandoned settlements that could be classified were enclosed farmsteads, while most new farmsteads were of complex type, albeit of far more regular form than was typical in the late Iron Age (e.g. see fig. 5.63). There is at least one incidence (Roxton Road West, Beds: Timby et al. 2007b) when an enclosed farmstead was transformed into something more complex, and indeed this was a period when sixteen settlements (c. 19 per cent of the total) are recorded as undergoing some form of transformation, such as at Bob’s Wood, 3 km north-west of Godmanchester, where an existing farmstead was modified into a typically more formalised and integrated enclosure system (Hinman 2012). At Camp Ground on the Fen edge there appears to have been direct continuity from the middle Iron Age through to the late Roman period, though the Iron Age enclosure compounds gave way to a more indefinite group of fields/enclosures at some point in the mid- to later first century a.d., before a complete reorganisation

Romanch5.indd 197

into a large nucleated port settlement in the early second century a.d. (Evans 2013b, 203). The second century was the period when settlement numbers reached their height in the study area, as with the wider regions around it (fig. 5.60). By this time Godmanchester and Cambridge appear to have become thriving, albeit fairly modest, urban centres, both probably developing from Roman military sites, with a range of industries and agricultural facilities. All the nucleated settlements in the area seem to have been in existence by this time, most developing in some form from earlier smaller farming settlements, although the two substantial ‘village’ settlements (which contained major, probably stone-footed building complexes) revealed by evaluation trenching, geophysics and cropmarks at Longstanton, appear to date from the second century onwards (Evans et al. 2006), hinting at the greater propensity for settlement nucleation during the mid–later Roman period. Lying c. 6 km further south, just c. 2 km to the north-west of Cambridge, was an extensive settlement fronting onto the Godmanchester road covering over 5 ha (NW Cambridge Site IV; see fig. 5.56), which developed from a small complex

15/09/2016 17:53:41

198

THE RURAL SETTLEMENT OF ROMAN BRITAIN

no. of objects per ha excavation

60 50 coins

40

bracelets 30

hairpins military fittings/weaponry

20

writing equipment recreation objects

10

weighing objects

0 Langdale Hale, Earith, Colne Fen

Vicar's Farm, West Cambridge

Other complex farmsteads (mean value)

fig. 5.61.  Comparison of selected finds assemblages from ‘rich’ complex farmsteads at Langdale Hale and Vicar’s Farm with mean value from 14 other complex farmsteads in the case study area

farmstead of mid–late Iron Age date. It is hard not to see its major development in the later first and second centuries a.d. as being related to the development of the town at Cambridge, though as the post-excavation work for this site is on-going, the specific social and economic relationships between the settlements remain uncertain. However, it was one of at least twelve settlements that have been revealed within 5 km of Cambridge (further sites are recorded to the south of the town beyond the Central Belt region, e.g. Addenbrooke’s: Evans 2008), that appear to have developed quite rapidly in the early Roman period. These include a number of complex-type farmsteads characterised by systems of enclosures and linked by trackways/ droveways, such as at Vicar’s Farm, c. 1 km west of Cambridge, which appears to have been a place for the storage and processing of locally grown cereals (Lucas 2001; see fig. 5.63(a)). In all, it indicates a highly organised and managed landscape, seemingly geared up for intensive agricultural and at some levels (e.g. pottery production) industrial production. The hinterland of Godmanchester, 23 km north-west of Cambridge, only really started to be developed from the beginning of the second century a.d., with farmsteads such as Rectory Farm lying just 800 m north-east of the town and seemingly linked by a road and field boundaries (Lyons et al. forthcoming; Jones 2003, 188). Both this and the longer-lived settlement at Bob’s Wood, 3 km north-east of the town overlooking Ermine Street, appear to have been quite wealthy farmsteads by this time, with aisled buildings, networks of enclosures and substantial pottery and small finds assemblages; both later developed into relatively modest villas that otherwise appear quite rare in this area. Further north along the Fen edge there also appear to have been major developments in the early second century a.d., with establishment of the inland Fen port at Camp Ground, Colne Fen

Romanch5.indd 198

(see Ch. 3, fig. 3.10), and the transformation of a nearby farmstead at Langdale Hale into a series of formally arranged enclosures along a major trackway, associated with ironworking, hide processing and large-scale crop processing (Evans 2013b, 47; fig. 5.63(c)). The specialist nature of the Langdale Hale site is reminiscent of the Vicar’s Farm settlement, with both sites also having relatively large assemblages of pottery, coins and certain other finds compared with many other farmsteads (fig. 5.61). This suggests a special socio-economic role for such settlements, perhaps incorporating markets and/or tax collection points, linked to their close proximity to important nucleated settlements (see discussion of ‘rich’ farmsteads above). However, while Langdale Hale has been suggested as having specific associations with state supply networks (ibid.), Vicar’s Farm seemingly has no indication of ‘official’ status. The development of Fen edge sites like Camp Ground and Langdale Hale, together with others, including a large but ill-understood nucleated settlement at New Fen Drove, appears to coincide with a major expansion of settlement deeper into the Fens. It is also the period when the Car Dyke canal was thought to have been constructed (see above, p. 176), which, for the Cambridgeshire part at least, is likely to have been used for river transport, part of a wider system of Fenland rivers/ canals. The increased exploitation of these Fenland waterways as a transport network has been suggested as being connected with military supply (Evans 2013b, 451), with agricultural produce from the region being collected along with other goods and then shipped north via ports like Camp Ground. This would at least partly explain the apparently strong ‘official state interest’ in the Fenland area from this time, as noted above. However, explicit epigraphic evidence of state involvement in the Fens remains to be discovered. Unlike in the central Fens, the settlement pattern on the Fen edge and throughout most of

15/09/2016 17:53:42

THE CENTRAL BELT

199

fig. 5.62.  Settlement development in the case study area in the late Roman period with selected sites named in the text

the study area does not appear to contract too drastically during the later Roman period (fig. 5.62). Indeed the third century a.d. was the high point of settlements such as Camp Ground, while modest villa buildings were constructed at Bob’s Wood, Rectory Farm and Whitehills (an associated late Roman cemetery was excavated at Watersmeet: Nicholson 2006) around Godmanchester. Activity within the town itself appears to have been most intensive during the second and third centuries a.d., though possible episodes of flooding occurred during the mid-third century, and there is evidence for contraction in some areas later in that century, contemporary with the construction of the first masonry defences (Jones 2003, 187–8). Nevertheless there was clearly continuing activity in the town at least until the end of the fourth century, and further corridor villas of probable fourth-century date were built along the Ouse Valley, 10 km to the east at Fen Drayton (Mortimer 1995; Evans 2013b, 474) and within an Ouse tributary 14 km to the south-west at Great Staughton (Greenfield et al. 1994); there are also a number of other known or suspected villas within the valley for which there is minimal information.

Romanch5.indd 199

The situation around Cambridge appears more varied in the later Roman period, with some sites having been abandoned by the early/mid-second century (e.g. High Cross and Addenbrooke’s) and others like Vicar’s Farm to the west being most intensively occupied in the later third century (Lucas 2001). A third-century villa was hinted at by the concentrations of high-status building material found during excavations at the northwest Cambridge development (Site VII), but there was little evidence of much activity into the fourth century (Evans and Newman 2010). Other possible mid–late Roman villas in the vicinity, at places like Milton (Reynolds 1994) and King’s Hedges School, Arbury (Alexander and Trump 1969; HER 05424 and 05411), remain quite poorly understood, though they do hint at increased elite architectural investment at this time. Two other farmsteads (Site II/RB1 and Traveller’s Rest Sub Site) excavated as part of the north-west Cambridge development were abandoned or had changed use by the end of the second century (Evans 2015), while the larger roadside settlement (Site IV) underwent a gradual contraction during the third century a.d. with all activity ceasing by the mid-fourth century. Overall,

15/09/2016 17:53:49

200

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a N

0

100 m 1:2000

b

c

N N fig. 5.63.  Roman

complex farmsteads at (a) Vicar’s farm, Cambridge (Lucas 2001), (b) Knobbs farm, Earith (Evans 2013b), and (c) Langdale Hale, Earith (Evans 2013b), showing outer enclosures/paddocks and trackways

Romanch5.indd 200

15/09/2016 17:53:49

THE CENTRAL BELT

it seems that after a highpoint in the mid-Roman period (corresponding with the maximum growth of Cambridge and the main period of production of the local Horningsea pottery industry), there was a more pronounced decline in fortunes, with over half of all settlements around the town appearing to be in decline or being abandoned before the end of the third century. Whether this represents a genuine decline in the intensity of exploitation of the landscape, or else an ‘economic re-alignment’ with many smaller farming units giving way to fewer larger (villa?) estates, as suggested by Lucas (2001, 153) remains uncertain, though the latter seems more likely. There is certainly no evidence for an expansion of the town at Cambridge at this time (i.e. no indication that surrounding populations were moving into the town), though the earlier earthwork defences were replaced by a stone wall and ditch by the midfourth century, and occupation is thought to have continued in some form at least into the fifth century (Burnham and Wacher 1990, 249). Whatever the situation, it suggests a somewhat different trajectory of rural development to that of the Ouse Valley to the west where there is a stronger fourth-century tradition overall, perhaps

201

linked to a more active maintenance of the trade networks up through to the port at Camp Ground, which still appears to have been well used at this time. Nevertheless, all parts of the study area remain fairly modest, in terms of the numbers and types of settlement in use, compared with the western Fen edge around Durobrivae; here many villas developed in the later Roman period, suggesting that the Lower Nene Valley was the more vibrant social and economic zone during the late Roman period. FARMING THE LAND The river and Fen gravel terraces of this part of Cambridgeshire are very well suited to both pastoral and arable farming, while the heavier soils of the claylands have now been proven to be equally well exploited, although with more of an emphasis on animal pasture (Abrams and Ingham 2008, 119). There are various different strands of evidence that may be used in attempts to reconstruct aspects of this agricultural economy, including settlement morphology, wider landscape infrastructure (fields, paddocks, droveways etc.), agricultural tools and structures, and a whole suite

fig. 5.64.  Mid–late Roman farmstead at Fenstanton in relation to surrounding cropmarks in the lower Ouse Valley (cropmark data © Historic England)

Romanch5.indd 201

15/09/2016 17:53:56

202

THE RURAL SETTLEMENT OF ROMAN BRITAIN

of environmental remains. Many of these have already been touched upon in the broader context of the Central Belt region overall, and will be fully explored on a wider basis in a subsequent volume, but here these elements will be used to create a brief synopsis of the variability of farming practices on a much smaller scale. The farming landscape: farmsteads, fields and trackways It has been noted above that from the early Roman period onwards most farmsteads in the case study area were characterised by systems of regular conjoined enclosures, often structured around trackways/droveways that are presumed to have linked different settlements with fields and wider open pastures (e.g. complex farmsteads shown in fig. 5.63). The nature of these sites suggests that they were geared up for livestock management, with many of the enclosures containing waterholes and presumably being used as paddocks, as demonstrated by the recently excavated site at NIAB, Cambridge (see fig. 5.37 above). National Mapping Programme (NMP) cropmark data for parts of this Fen edge area have revealed many

other examples of what appear to be Romanperiod complex farmsteads, adding valuable contextual information about the landscapes around some of the excavated sites. An example in the Lower Ouse Valley revealed a major droveway with adjacent paddocks/enclosures running north from a partially excavated mid–late Roman farmstead at Fenstanton, which comprised ditches, pits, waterholes and a probable timber building (Nicholson 2004) (fig. 5.64). The farmstead lay just 400 m north of a major Roman road and it seems likely that the road, farmstead, droveway and enclosures were all connected into a wider system associated with movement of animals from the open floodplain pasture, possibly up into the farmstead and through to the major road to be herded towards Cambridge or Godmanchester. figure 5.65 shows the extent of the NMP cropmark data in the wider region north-west of Cambridge in relation to the excavated sites, with those cropmarks interpreted as Iron Age/Roman settlements highlighted in red and those suggested as enclosures or field systems in blue. The field systems in particular seem more densely clustered on and around the gravel terraces and lower valley

fig. 5.65.  National Mapping Programme (NMP) ‘Iron Age and Roman’ cropmark data (red=‘settlement’; blue=‘fields/enclosures’) in relation to excavated sites to the north and west of Cambridge (cropmark data © Historic England)

Romanch5.indd 202

15/09/2016 17:54:02

THE CENTRAL BELT

slopes (e.g. fig. 5.66c), while settlements seem equally well represented on the higher claylands (though there are indications of a preference for occupation sites on gravels beside clays in this area: Evans and Standring 2012, 100–3). In many ways this reflects the excavated evidence, with nine out of the twelve field systems being located at these lower levels, suggesting that much of this

203

area was highly organised into an integrated agricultural landscape. However, it is clear that even in these lower-lying zones field systems did not spread everywhere, as an excavated area of over 50 ha in extent lying immediately to the west of the nucleated settlements at Longstanton, on the interface between the third terrace gravels and Amptill Claylands, was almost completely devoid

a

b

0

100 m 1:2000

c

N

0

200 m 1:5000

fig.

5.66.  Plans of excavated Roman agricultural landscapes at (a) Low Fen, Fen Drayton (Mortimer 1995) and (b) The Fields (Wright et al. 2009), along with (c) NMP cropmark of Iron Age/Roman field system in the Lower Ouse Valley (NMR 15285; © Historic England)

Romanch5.indd 203

15/09/2016 17:54:03

204

THE RURAL SETTLEMENT OF ROMAN BRITAIN

of Roman features (Hunt and Paul 2015). In addition, it is clear that some field systems did occur on the higher claylands, such as the intermittent pockets of fields observed to the west of Cambridge that were seemingly aligned upon the main Roman road system, probably dating from the second century a.d. (Wright et al. 2009, 87). Two superimposed rectangular field systems were excavated at The Fields in this boulder clay landscape, the latter system comprising an unusual grid arrangement, with the ‘compartments’ (9 × 13–17 m) representing either small fields or animal pens (ibid., 57) (fig. 5.66b). It is assumed that many of the field systems in the study area, as elsewhere, relate to the practice of arable agriculture, though, unfortunately, rarely is there any direct environmental evidence for this. There are, however, five examples of the ‘lazybed’ system of parallel bedding-trenches discussed above, which appear to be related to horticulture (fig. 5.66a). This concentration of c. 20 per cent of all recorded Roman lazybed systems from England and Wales within such a relatively small area suggests growing horticultural crops was a specialisation in this region, which is at least partially borne out by the evidence of the plant remains (see below, p. 205). Agricultural structures and objects Settlement morphology, trackways and fields provide a crucial context for the variable farming practices in the region, but further details are dependent upon associated structures, finds and in particular environmental remains. In terms of structures there are buildings interpreted as granaries at three settlements, two along the Fen edge at Camp Ground and Knobbs Farm and another at the farmstead/villa at Rectory Farm just outside Godmanchester. In addition, twelve aisled buildings on seven sites possibly served an agricultural function, while a large timber building at Car Dyke, Waterbeach may have been used for grain storage as environmental remains from the adjacent canal sediments included quantities of insect pests of stored grain (Macaulay 2012). All of these structures hint at the large-scale storage and movement of grain in the area, while corndryers at eight sites, mostly complex farmsteads, suggest extensive processing of arable crops, though some may have also been used in the malting process, as indicated at Newmarket Road, Cambridge, where large quantities of malting waste came from the oven (Wallis 2012). Corndryers, along with drying racks, threshing floors and granaries were also found in Godmanchester, indicating that at least some of the crop processing (and probably malting) was carried out in ‘urban’ environments (Jones 2003, 188).

Romanch5.indd 204

There is considerable evidence for milling in the study area, with fragments of over 50 millstones recovered from eleven sites, most a mixture of complex farmsteads and nucleated settlements, including sixteen examples from Langdale Hale, Earith, which, as mentioned above, is interpreted as a specialised (and ‘official’?) cereal-processing farmstead, associated with the nearby port at Camp Ground (Evans 2013b, 171). A further twelve millstones also came from the fourthcentury villa at Great Staughton (Greenfield et al. 1994), while others are known from Vicar’s Farm, both also possibly specialised processing centres. Agricultural tools are reasonably rare finds in any Romano-British context, but were recorded on ten sites in the area, again a mixture of complex farmsteads and nucleated settlements, and included a sickle, shears, and ploughshare from Lower Cambourne (Wright et al. 2009) and a hoe, rake prong, and socketed pruning hook from Love’s Farm St Neots (Oxford Archaeology East 2012). In all, the evidence suggests that many of the complex farmsteads and nucleated settlements, along with the few villas in the area, were organised for extensive arable cultivation of the gravel terraces and valley sides, most probably in a mixed arable farming economy integrated with livestock management. Faunal remains The evidence for animal remains is typically quite rich in this region, with data recovered from around two-thirds of all settlements in the case study area. There are 45 phased animal bone assemblages with over 100 NISP, 37 from farmsteads (twenty complex, six enclosed), and the remainder from a few nucleated settlements, a single modest villa (Bob’s Wood), a shrine (Haddenham), a field system and a pottery production site. For the case study area as a whole cattle clearly dominate (52 per cent of CSP), then sheep/goat (41 per cent) with a very small proportion of pig (6.9 per cent). There are some assemblages, particularly in parts of the Ouse Valley area and on the Fen island of Ely, where sheep/goat do form the majority, though usually only by a very small margin. Exceptions include the late Iron Age settlement and Roman period shrines at Haddenham in the Fens (up to 80 per cent sheep/goat) and a mid-Roman assemblage from a probable complex farmstead just to the east of Godmanchester (61 per cent sheep/goat: Wait 1990). There are clear differences between farmstead types in terms of the proportion of the main domesticated species, as demonstrated above for the Central Belt as a whole, with cattle dominating at complex farmsteads and sheep/goat at enclosed

15/09/2016 17:54:03

THE CENTRAL BELT

farmsteads. For the most part this pattern is the result of chronological factors, with nearly all enclosed farmsteads that have faunal assemblages being dated to the late Iron Age, when there is a clear bias towards sheep husbandry. In contrast, however, in four out of the five late Iron Age assemblages from complex farmsteads, cattle form the dominant taxon. The pronounced increase in the proportion of cattle from the midRoman period onwards, as observed for the wider region, is certainly seen in this study area, concurrent with the major expansion of complex farmsteads, nucleated settlement, and as far as the chronological evidence will allow, field systems and corndryers. This suggests that the increase in the importance of cattle was part of a wider-scale shift to more expansive arable agriculture, with the animals being used for traction and manuring. Plant remains It has been suggested above that an increase in arable agriculture during the mid–late Roman period may have been one of the primary drivers behind large-scale changes in the rural landscape in this zone and the wider Central Belt region. Charred plant remains were recovered from 66 sites in the Cambridgeshire Fen edge case study area, and there is no doubt that spelt wheat formed the primary arable crop, though emmer wheat was present in much higher proportions than elsewhere in the Central Belt, being identified in over 50 per cent of assemblages. During the Roman period emmer was rarely more than a very minor component, and may well have been a contaminant rather than an actively cultivated crop, though there are several instances of it being the dominant crop in late Iron Age assemblages, such as at Hurst Lane Reservoir, Ely and Wardy Hill, which both contained large amounts of emmer chaff, almost certainly representing cropprocessing waste. The more frequent occurrence of emmer during the late Iron Age helps to explain the greater association of this crop with enclosed farmsteads. Later Roman assemblages from this farmstead type show the typical spelt dominance, thereby suggesting that there were eventually no major differences in the cultivation of primary cereal types between different types of farmstead. Free-threshing wheat also occurred in a higher percentage of assemblages here than in many other areas in the Central Belt, being found on almost 60 per cent of complex farmsteads and nucleated settlements, occasionally in some quantity, such as in the late Roman farmstead at Harradine’s Farm, Woodhurst (Williams 2011). Flax and broad bean were also found at this site and it was suggested (ibid.) that there was a system

Romanch5.indd 205

205

of crop rotation, which may well have been the case at 21 other settlements where, in addition to the major wheats and barley, flax, rye and/or pulses were also found. Rye appears only to have been added to this suite of minor crops in the third and fourth centuries a.d. Evidence for horticultural crops is fairly limited, though given the number of ‘lazybed’ cultivation trenches noted above they are likely to have been quite widespread. Waterlogged remains from a pond at the farmstead/villa at Rectory Farm just outside Godmanchester produced macrofossils of beet, marigold, grape, fig, fennel and opium poppy, along with other plant and tree species reminiscent of a Mediterranean style garden, indicating the high status of the occupants of the site (Lyons et al. forthcoming). The fig is certainly likely to be an imported plant, though grapes are known to have been cultivated in the wider region (e.g. at Wollaston; see above, p. 183) and another possible grape pip has been recovered nearby at a potentially extensive farming settlement at Ely Road/Hill Close, Milton (Rees 2009). Finally, in terms of environmental remains, it is worth noting that at least five excavated sites in the case study area are argued to have evidence for hay meadows (e.g. Eaton Socon: Gibson 2005; Little Paxton Quarry: A. Jones 2011), based on plant and insect remains. Hay would have been an increasingly important commodity for ensuring winter animal fodder, especially with the growth of the two main towns and other larger nucleated settlements, and so it is quite likely that extensive areas of managed hay meadow existed within the major river valleys of the Cam and the Ouse and around the Fen edge. SUMMARY: ROMAN RURAL SETTLEMENT IN CAMBRIDGESHIRE FEN EDGE This brief account of late Iron Age and Roman rural settlement on the Cambridgeshire Fen edge has demonstrated the variability of landscape use and development even within a relatively limited area, though all within a broader framework of more widespread change that occurred as the region became fully incorporated within the Roman province. A reasonably high density of mixed late Iron Age settlement was found throughout the Fens, valleys and higher claylands, and there is evidence for considerable disruption to this pattern within the 50 years following the Claudian conquest, possibly associated with the aftermath of this conquest or with the Boudican revolt almost 20 years later. Post-conquest developments were, nevertheless, highly localised, with many new or transformed settlements around Cambridge dating from the mid-first century a.d., and other transformations further west, in the

15/09/2016 17:54:04

206

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Ouse Valley and Fen edge, belonging to the later first and early second centuries. The wider landscape soon developed into a mosaic of trackways, fields and open grassland areas with complex farmsteads particularly dominant in the wide river valleys and broad Fen edge gravels. Some of these developed into specialised agricultural processing and, perhaps, market centres, linked to local nucleated centres and associated with the intensification of arable production, contributing to the wider ‘bread basket’ of central Roman Britain. Longer distance trade of agricultural and other produce is also hinted at by the development of roads and waterways, along with the establishment of the Fenland port at Camp Ground. Such developments continued into the later Roman period, when there is some evidence for increased elite investment in villa architecture, though seemingly not to the same extent or at the same scale of opulence as in the Lower Nene Valley or further west in the Cotswolds. The decline in farming settlement numbers around Cambridge at this time may reflect a centralisation of the landscape into larger villa estates, although the excavated evidence for such settlements remains fairly minimal. Over 40 per cent of the sites in the study area indicate some activity continuing to at least the end of the fourth century a.d., while reports on the farmstead/villa at Rectory Farm (Lyons et al. forthcoming) and the farmstead at Childerley Gate (Abrams and Ingham 2008) suggest occupation, if only limited, into the fifth century. Given the problems with dating this period noted above, it is probable that many other late Roman sites also continued into the fifth century if not beyond, although it is likely that the late Roman agricultural patterns underwent significant re-alignment, as suggested by the subRoman (fifth century) faunal assemblage from Childerley Gate, which reverted from cattle to sheep/goat dominance (ibid., 120). Most of the data used in this case study have been derived from the recent proliferation of very extensive excavations in and around Cambridgeshire Fen edge, and, used together with other data sources such as the NMP cropmark survey, are helping to transform our understanding of the variability of the Romano-British countryside, even on a relatively localised scale. Many of the large-scale archaeological investigations, such as at North-West Cambridge and NIAB Huntington Road, are still in the early phases of post-excavation analysis, and, together with the many future planned projects, will no doubt further revolutionise our understanding not only of this landscape, but also of the diversity of the wider Central Belt region.

Romanch5.indd 206

REGION SUMMARY The large Central Belt region encompasses a diverse range of landscapes, including major river valleys, clay vales and plateaus, chalk and limestone hills and three substantial areas of wetland. As would be expected, there is a corresponding diversity in the density, chronology and character of Roman rural settlement, though there is also a degree of cultural and economic homogeneity as expressed through the range of material culture, architecture, landscape infrastructure and settlement types. It is essentially what may be regarded as ‘typical’ Romano-British countryside, in having large numbers of farmsteads surrounded by trackways and field systems, as well as the greatest concentration of settlements with villa architecture and a range of larger nucleated settlements, mostly sited upon the major Roman road system. The region also contains a high proportion of the Roman province’s major urban centres, including at least four civitas capitals, which must have had a significant influence on underlying rural settlement development. The Central Belt was clearly well populated in the late Iron Age, with a high percentage of settlements to the east of the region in particular being established de novo at this time. The impact of the Roman conquest (or Boudican revolt) is hard to detect in most areas, though landscapes like the Trent Valley and Rises and parts of the West Anglian Plain do appear to have had a slightly greater dislocation of settlement from the late Iron Age (see discussion, Ch. 12). In most areas, however, it is not until the start of the second century a.d. that major developments occurred, reflected in changes in settlement morphology, architecture, landscape infrastructure and agricultural practice. In particular the rise in number of complex farmsteads, and corresponding increases in trackways and field systems, suggest a greater emphasis on landscapes that were more actively managed for agricultural productivity. It appears that during the second century a.d., complex farmsteads spread from the valleys of the West Anglian Plain (Nene and Ouse), where many had evidence for longer histories of occupation, to the main western valleys of the Upper Thames, Avon and Severn, where many seem to have been new ‘implantations’, indicating wide-scale changes in these landscapes. Settlement numbers as a whole peaked during the later second century a.d., a period which also corresponded with a major growth in the number of villas across much of the region.The architecture of ‘non-villa’ settlements also demonstrated an increase in diversity by this time, with rectangular buildings now being the norm and a big increase

15/09/2016 17:54:04

THE CENTRAL BELT

in the evidence for masonry structures, sometimes with multiple rooms. However, despite this move towards what may be argued as more ‘Roman’ types of buildings, there remained considerable diversity within the region, with circular architecture, for example, remaining quite prominent in parts of the West Anglian Plain. The later Roman period sees perhaps the greatest differentiation between landscape zones of the Central Belt. To the east the Fens (though not the Fen edge) exhibit a significant decline in settlement numbers from the early/mid-third century a.d., while numbers are at their height in the Cotswolds during the early fourth century. A large proportion of these were villas, including a number of ‘super-villas’, clearly belonging to the uppermost members of society. Such apparent concentration of wealth in a relatively small area

Romanch5.indd 207

207

cannot be directly related to any changes in agricultural practice, though the general decline in rural settlement numbers in the fourth century (particularly notable in the Severn and Avon Valleys adjacent to the Cotswolds) may have been part of large-scale landscape reorganisation at this time, perhaps including a move to more centralised estates. Overall, the sheer number of excavated Roman rural settlements in the Central Belt has enabled a much greater appreciation of the depth of diversity within a ‘core’ zone of Roman Britain. The differences in settlement types are just one aspect, and can be related to other patterns in material culture and environmental material, that at least start to reveal something of the mosaic of communities that lived in the heart of the Roman province.

15/09/2016 17:54:04

Chapter 6

THE EAST By Alexander Smith

The region defined here as the East is among the smallest of those within the current project, at 11,393 km². It conforms in large parts to the area of England identified as East Anglia, comprising most of Norfolk and Suffolk, along with the southeastern portion of Cambridgeshire, northern halves of Essex and Hertfordshire, and a very small part of Bedfordshire (fig. 6.1). The northern half of the region is traditionally associated with the late Iron Age territory (and subsequent Roman civitas) of the Iceni, while to the south lay areas usually attributed to the Trinovantes and the Catuvellauni (see Ch. 12 for discussion of such ‘tribal’ areas). These broad ‘tribal’ groups figured largely in classical literary accounts of Britain during the later Iron Age and early Roman period (e.g. Tacitus Annals XIV, 31–7), particularly the

fig.

Claudian conquest itself and the Boudican revolt of a.d. 60/1. Such accounts have subsequently become firmly embedded in many academic syntheses of the region (e.g. J. Davies 2009), so that much of the early development in particular has been framed through a pseudo-historical narrative, perhaps at times a little too much to the detriment of the primary archaeological evidence. THE NATURE OF THE LANDSCAPE The specific make-up of the region has been determined by six distinctive landscape zones (fig. 6.2), including the coastal saltmarshes and lowland heaths of North Norfolk and the Suffolk Coast, along with the sandy heaths, valleys and chalk rivers of Breckland, immediately to the east

6.1.  The East region in relation to modern county boundaries

208

Romanch6.indd 208

06/09/2016 17:04:52

THE EAST

fig.

209

6.2.   Constituent landscape zones of the East region table

6.1: number

of excavated sites and density (per km²) by landscape zone in the east region

Landscape zone Breckland

Area (km²)

No. of records

Density of records per km²

1019.26 32 0.0314

East Anglian Chalk 838.70 55 0.0656 East Anglian Plain

6343.87

171

0.0270

North Norfolk

1793.23

27

0.0151

Suffolk Coast and Heaths

845.00

9

0.0107

The Broads

553.17 8 0.0145

of the Fens. The largest area of wetlands in the region comprises the Norfolk Broads, while by far the biggest single landscape zone (6343 km²) is that defined as the East Anglian Plain, a plateau dissected by a patchwork of streams and widely spaced river valleys and interspersed with extensive areas of boulder clay, which become far more undulating to the south and south-west. The final zone is the East Anglian Chalk, comprising lowlying chalk hills dissected by river valleys. Although the low-lying coastal areas of the region are very acidic, soils in other areas are generally less destructive, meaning that some excavated sites have produced good environmental evidence.

Romanch6.indd 209

THE EAST DATASET There are a total of 302 records for the East region, accounting for 182 different settlements, along with ‘isolated’ field systems, agricultural buildings, religious sites, and funerary sites. The density of these records across the different landscape zones is shown in table 6.1, revealing generally higher concentrations of excavated sites to the west, within Breckland and the East Anglian Chalk. When this density is viewed geographically, however, there is also a major concentration towards the south of the East Anglian Plain and north Essex in particular (fig. 6.3. top), where

06/09/2016 17:05:00

210

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 6.3.  Kernel density of East region records (n=302) and all excavation records (1910–2010) from National Monument Records (NMR) Index (n=4707)

Romanch6.indd 210

06/09/2016 17:05:12

THE EAST

211

fig. 6.4.  ‘Roman’ NMP cropmark data in the Norfolk Broads/coast in relation to later Roman coastal forts at Caistoron-Sea and Burgh Castle (cropmark data © Historic England)

Romanch6.indd 211

06/09/2016 17:05:27

212

THE RURAL SETTLEMENT OF ROMAN BRITAIN

road scheme excavations have revealed many sites along the A120 (Roman Stane Street; Timby et al. 2007a) and development work at Stansted airport has uncovered extensive landscapes of Roman farmsteads and field systems (Havis and Brooks 2004; Cooke et al. 2008). There is also a group of nine nucleated roadside settlements in this part of north Essex and into Hertfordshire that often have multiple records (e.g. seven records for Great Dunmow along Stane Street), thereby contributing towards the higher overall density of records in this area. Other concentrations of records occur around Baldock in Hertfordshire, to the south of Cambridge in the Cam Valley, and around Thetford in the heart of Breckland, while much of the coastal areas of North Norfolk, The Broads and the Suffolk Coast have fairly minimal excavated evidence. When compared with the National Monument Records (NMR) index of excavations (1910– 2010; fig. 6.3), there are distinct differences, with the two most densely excavated zones around Norwich and Ipswich corresponding with areas of only slightly higher than average Roman rural settlement records. The relative lack of records around Norwich in particular is surprising, as although the city did not develop from a Roman town, the civitas capital of the Iceni at Caistor lies just c. 5 km to the south. Nevertheless, despite differences such as these, the overall distributions of all excavations and Roman rural settlement records are quite similar, both exhibiting distinct paucities to the north and east of the region, especially on the coastlines (in part dictated by the National Park status of the Norfolk Broads, with little resulting development). It is known, however, that these were far from deserted landscapes, as extensive cropmark surveys in parts of Norfolk in particular have revealed dense patterns of interlinked settlements, fields and trackways, ostensibly of Roman date, as revealed for parts of The Broads in fig. 6.4. The wider distribution of cropmarks/ soilmarks in the East region purporting to be of Roman date is shown by Taylor’s figure 3.3 (2007, 15, 49; see Ch. 12, fig. 12.3), which demonstrates table

6.2: number

Landscape zone East Anglian Chalk

26

East Anglian Plain

80

North Norfolk

11 5

Total

Romanch6.indd 212

The East region contained a single major Roman urban centre, the civitas capital of Venta Icenorum at Caistor St Edmund, Norfolk, which was fairly modest in size compared with most other civitas capitals. However, the important late Iron Age oppidum, short-lived legionary fortress and subsequent colonia at Colchester lay just to the south of the region, and the whole area is particularly noteworthy for its proliferation of Roman ‘small towns’ (Brown 1995; fig. 6.5). Many of these ‘small towns’ developed from preconquest settlements, with Baldock and Braughing being classified as oppida owing to the scale of their late Iron Age activity (Pitts 2010). Only one of the Roman ‘small towns’ was later surrounded

Farmstead Farmstead Farmstead Farmstead Roadside Vicus Villa (all) (complex) (enclosed) (open) settlement 20

The Broads

ROMAN RURAL SETTLEMENT PATTERNS

of major settlement types by landscape zone in the east region

Breckland

Suffolk Coast and Heaths

abundant evidence from much of Norfolk, though with an almost complete break further south in Suffolk and north Essex, both as a result of differing soil types and variable local authority records of cropmark data.Taylor also demonstrated the widespread occurrence of finds scatters as evidence for Roman settlement in the region (ibid., 16, fig. 3.4), and this has been increased massively by the Portable Antiquities Scheme (PAS) records of Roman finds, which show particularly dense concentrations of finds across much of the East region (Brindle 2014; Walton 2012). The distribution pattern of excavated Roman rural settlements recorded in the East is therefore one that is very much biased by patterns of modern development, with over 85 per cent of the records being created since 1990, largely as a result of developer-funded work. Nevertheless, despite such geographic bias, there is still enough detailed information to allow characterisation of the form, character and development of these farmsteads, villas, villages and roadside settlements across the region from the late Iron Age to the end of the Roman period.

1 143

3

3

1

6

2

1

1

12

17

3

19

1

3

1

3

0

2

0

1

0 22

0 0 27

6

Village Total

5 1 26

1

3 10

28

1

102

1

16

1

7

0 1 2 29

2 14

3 182

06/09/2016 17:05:27

THE EAST

213

fig. 6.5.   Excavated late Iron Age/Roman rural settlement in the East region in relation to Roman roads and urban centres

by walls (Great Chesterford) and two by earthwork enclosures (Brampton and Chelmsford), the rest comprising undefended (or at least partially ditched) roadside settlements of varying scales, forming a comprehensive network of nucleated sites similar to that in parts of the Midlands. Some of these roadside settlements developed from early Roman forts, while later Roman military activity was restricted to four coastal forts at Walton Castle, Burgh Castle, Caister-on-Sea and Brancaster, all built during the third century a.d. The 182 Roman rural settlements recorded as part of the current project for the East are dominated by farmsteads, forming over 78 per cent of all sites, rising to 86 per cent in the East Anglian Chalk, though falling to just 64 per cent in North Norfolk, where the admittedly smaller corpus appears somewhat more diverse (table 6.2; fig. 6.5). Those settlements with villa architecture are relatively rare, especially compared to further south and west, with just fourteen excavated examples being recorded, though many more possible villas are recorded on the National Monuments Record (NMR), often defined by cropmarks and field survey. As just discussed, roadside settlements are particularly numerous in

Romanch6.indd 213

this region, though nucleated settlements located away from the major Roman road network (defined here as villages) are far scarcer. Vicus settlements were limited to those relating to the late Roman forts at Caister-on-Sea and Burgh Castle, as well as a nucleated settlement at Brancaster in North Norfolk, which was associated with a fort in the third century but had earlier origins, perhaps associated with an earlier military phase (Hinchliffe and Sparey-Green 1985; see below, p. 225). REGIONAL CHRONOLOGY Chronological data are available for all except four sites in the East dataset, enabling a fairly robust assessment of development over time. The great majority of these sites have been dated through their ceramic assemblages (especially finewares, which generally have greater chronological resolution), with use of radiocarbon dating being noted in less than 5 per cent of records (see Ch. 1, fig. 1.4). Nevertheless, finewares are well represented on many settlements in the East, both imported (e.g. samian recorded from c. 75 per cent of sites) and British wares, the latter including products from kilns at Hadham, Nene Valley, Colchester and Oxfordshire.

06/09/2016 17:05:34

214

THE RURAL SETTLEMENT OF ROMAN BRITAIN

140

no. of selements in use

120 100 80 60 40 20 0

fig. 6.6.   Number of settlements in use over time in the East region (NB: the M–LIA data only include sites that continued into the late Iron Age)

no. of settlements in use

16 14 12 10 8 6 4 2 0

East Anglian Chalk (n=28) 18 16 14 12 10 8 6 4 2 0

no. of settlements in use

no. of settlements in use no. of settlements in use

East Anglian Plain (n=102) 90 80 70 60 50 40 30 20 10 0

20 18 16 14 12 10 8 6 4 2 0

North Norfolk (n=17)

fig.

6.7.  Variations in settlement chronology by selected landscape zone within the East region

The overall chronological trajectory of settlements in use is shown in fig. 6.6, demonstrating a peak during the later first century a.d. and then decline in numbers from the third century onwards. The variations to this pattern in terms of different landscape zones are shown in fig. 6.7, with major differences demonstrated by a much more pronounced second-century peak of activity in North Norfolk, very similar to that in the Fenlands to the west (see Ch. 5), and a much less noticeable decline during the later Roman period in those landscapes to the west of the region, with settlement numbers in Breckland actually peaking in the third century. The early Roman peak and late Roman decline further south and east is similar to that noted in other

Romanch6.indd 214

Breckland (n=26)

parts of Essex and the south-east (see Ch. 4), with the decline around Colchester suggested as potentially associated with a movement of the rural population within the walls (Holbrook 2010). In terms of settlement continuity in the region, it is clear that the late Iron Age was a period of major expansion, with only 15 per cent of sites occupied at this time having origins in the middle Iron Age, the remainder being apparently established as ‘new’ settlements (fig. 6.8). Most (90 per cent) of these continued in occupation into the early Roman period, when there was also a considerable number of additional settlements established all across the region. This was, however, a period of significant settlement disruption, with almost 20 per cent of sites being ‘abandoned’;

06/09/2016 17:05:36

THE EAST

100

% of sites 'established' during phase

215 % of sites 'abandoned' during phase

90

% of sites in use

80 70 60 50 40 30 20 10 0 Late Iron Age (n=84) fig.

Late 1stC AD (n=125)

3rdC AD (n=113)

4thC AD (n=97)

6.8.   ‘New’ and ‘abandoned’ settlements over time in the East region

some, such as Addenbrooke’s south of Cambridge, being replaced by field systems (Evans 2008). Another site nearby at Babraham near the River Granta (tributary of the River Cam) was also abandoned during the later first century a.d., with the focus of activity shifting to the west to a series of ditched enclosures and agricultural buildings (Armour 2007). This suggests a major reorganisation of the landscape rather than widespread abandonment, which is corroborated by a further nineteen sites (c. 15 per cent of total) recorded as undergoing transformations at this time, such as at Kilverstone, Thetford, where a major enclosure system was laid out (Garrow et al. 2006). It would be tempting to attribute such a state of flux to the upheavals and aftermath of the Boudican revolt in a.d. 60/1, though the problem with this lies in that, notwithstanding a small group of sites around Thetford, the majority of those settlements showing indications of ‘abandonment’ were located in the southern half of the region (e.g. Colemans Farm, Rivenhall End: Roy 2003), well away from the ‘Icenian heartlands’ where one might expect devastation to be more apparent. Archaeological evidence for such ‘territory blighted by the Boudican revolt’ (Mattingly 2006, 384) is in fact quite limited in the countryside of East Anglia, restricted for the most part to ‘exceptional’ sites such as the enclosure complex at Fison Way in Thetford, which was levelled around the time of the revolt (Gregory 1991). While this lack of evidence may be partly due to problems of broad ceramic dating of sites, it is likely, as Aldhouse-Green (2006, 227) has pointed out, that Roman retaliation in the aftermath of the Boudican revolt was quite patchy and targeted to specific sites of political importance, with the vast majority of the countryside left untouched. In fact, in North Norfolk in particular most rural settlements show evidence for distinct continuity

Romanch6.indd 215

2ndC AD (n=127)

until the later second century at least, although the relatively high proportion of farmsteads established ‘new’ in the later first century a.d. (or after a short hiatus such as at Sedgeford: SHARP 2014) could conceivably be linked with wider scale landscape reorganisation following the revolt (Frere 2000). The mid- to late Roman period saw an ever decreasing number of ‘new’ settlements being established (with none post-dating a.d. 200 in North Norfolk), and an increasing number of settlements being abandoned, particularly from the later third century a.d. Unlike a number of the early Roman ‘abandoned’ settlements, these sites do not appear to have been succeeded by field systems at this time, although at least one site, at Clay Farm, Trumpington, Cambs, had a possible double-ditched late Roman funerary enclosure constructed over the earlier farmstead (Phillips and Mortimer 2011). A total of 78 sites (42 per cent of all sites) were occupied until at least the end of the fourth century a.d., though only eleven of these were said to have continued into the fifth century or later, including the nucleated settlement at Baldock (Fitzpatrick-Matthews 2014) and the well-known villa at Rivenhall in Essex, which was thought to have been in use until the sixth century a.d., although the sequence here is quite poorly dated (Rodwell and Rodwell 1993; cf. Rippon et al. 2015, 9). FARMSTEADS: MORPHOLOGY, CHRONOLOGY AND DISTRIBUTION The distribution of all 133 excavated farmsteads in the East region is shown in fig. 6.9, with broad chronological patterns highlighted in fig. 6.10. There is a clear later first century a.d. emphasis in the number of farmsteads in use, with a steady decline after this point, which, if genuine, must reflect a fundamental shift in the way that the rural landscape was organised and farmed, particularly

06/09/2016 17:05:37

216

6.9.   All excavated farmsteads (open, enclosed, complex and unclassified) in the East region

no. of farmsteads

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig.

100 90 80 70 60 50 40 30 20 10 0

6.10.   Number of excavated farmsteads in use over time in the East region

in the later Roman period when the drop in farmstead numbers is more pronounced. As with surrounding regions, only c. 35 per cent of farmsteads (46 sites) could be classified in terms of their plan, although a further 20 per cent (27 sites) had more speculative evidence enabling classification with a much lesser degree of certainty (fig. 6.9). Most of the classified farmsteads (including those with less certainty) were enclosed, followed by complex farmsteads and only six having any recognised ‘open’ phases of development

Romanch6.indd 216

(table 6.2). The majority of all these farmsteads lay to the south and west of the region, with those outside this area almost all being of the enclosed form (fig. 6.9), although cropmark evidence from the parts of Norfolk referred to above does suggest that more complex farmsteads may have existed to the north and west (see fig. 6.4). The general distribution of excavated complex farmsteads appears to represent the easternmost fringe of their main Central Belt concentration (see Chs 2 and 5), since they are particularly

06/09/2016 17:05:44

THE EAST

% of classified farmsteads

80

complex farmsteads

enclosed farmsteads

217 open farmsteads

70 60 50 40 30 20 10 0

fig.

6.11.   Relative frequency of farmstead types in use over time in the East region

N

a

b

Structure 0

100 m 1:2000

fig. 6.12.   Development of the farmstead at Kilverstone, Thetford, Norfolk from (a) late Iron Age (?)open to (b) early/mid-Roman complex settlement (Garrow et al. 2006)

Romanch6.indd 217

06/09/2016 17:05:45

218

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

p

N

Haverhill Business Park, Sturmer

b

s s

s

s s s p

Pond

s

Structure

s

MTCP site, Stansted airport

0

100 m 1:2000

fig. 6.13.   Plans of complex farmsteads in Essex at (a) Haverhill Business Park, Sturmer (Gardner 2004) and (b) MTCP site, Stansted airport (Cooke et al. 2008)

Romanch6.indd 218

06/09/2016 17:05:46

THE EAST

219

a

b

N

c

0

100 m 1:2000

fig. 6.14.   Plans of enclosed farmsteads from the East region at (a) Great Notley, Essex (late Iron Age/early Roman) (Brooks and Holloway 2006), (b) Lobs Hole, Stevenage, Herts (late Iron Age) (Hunn 2005) and (c) Wighton, Norfolk (late Iron Age) (Gregory and Gurney 1986)

prevalent in the river valleys of the East Anglian Chalk, including those in the Cam Valley south of Cambridge, such as at Addenbrooke’s (Evans 2008) and Clay Farm, Trumpington (Phillips and Mortimer 2011). As stated above, these Cambridgeshire sites were unusual in appearing to go out of use by the end of the early Roman period, along with a number of other settlements in the local environs (see Cambridgeshire Fen edge case study, Ch. 5); the abandonment of these sites has been argued by Evans (2008, 135) to have been associated with the establishment of villas in the near vicinity at places like Shelford (HER CAM 57). Most complex farmsteads in the region follow the wider pattern of increased visibility during the middle to later Roman period (fig. 6.11). Half of all these farmsteads were seen to have developed from more simple forms of settlement, as at Kilverstone, Thetford, Norfolk, where an apparent open settlement of late Iron Age date was totally transformed in the early Roman period into a complex farmstead of conjoined enclosures and trackways (fig. 6.12; Garrow et al. 2006). The developmental trajectory from ‘simple’ to complex was not always followed, however, as seen at Strood Hall in the south of the East Anglian Plain, which developed from a very regular system of early Roman enclosures and trackways to midand late Roman phases characterised by timber/

Romanch6.indd 219

mass-walled structures, pits and middens, seemingly unenclosed by any ditched boundaries (Timby et al. 2007a, 84; see Ch. 2, fig. 2.5). The later Roman farmstead at Kilverstone was slightly unusual in the region in being characterised by a system of conjoined enclosures (linear complex), rather than by a single sub-divided enclosure (fig. 6.13), a type that formed the majority of complex farmsteads in the East region (see Ch. 2, fig. 2.17 for overall distribution). Such sub-divided enclosures are more typically found in areas where simple enclosed farmsteads are dominant and can be viewed as developments of such farmsteads, perhaps associated with more elevated social status and/or with differing economic trajectories. The 27 enclosed farmsteads are far more widely spread across the East region, and vary considerably, though most tend to be rectilinear in form and fairly modest in size (average 0.5 ha), as at Lobs Hole, Herts (Hunn 2005), and Wighton, Norfolk (Gregory and Gurney 1986) (fig. 6.14b and c). A small (0.35 ha) late Iron Age enclosed farmstead at Great Notley, Essex, was later replaced on a larger scale (min 0.63 ha) in the later first century a.d., but there was no evidence for any more complexity in its layout, and the excavators suggested that occupation may only have been ‘sporadic’ (Brooks and Holloway 2006). Almost all enclosed farmsteads were encompassed

06/09/2016 17:05:47

220

THE RURAL SETTLEMENT OF ROMAN BRITAIN

by a single ditch, though the ACS site at Stansted did have a concentric outer ditch along at least two of the sides (Havis and Brooks 2004). The number of enclosed farmsteads in use demonstrates a distinct peak in the later first century a.d., thereafter undergoing quite a steep decline, especially in the late Roman period when very few remain. The only enclosed farmstead known to have been constructed during the late Roman period was the LTCP site at Stansted airport, which lay just to the south of an earlier complex farmstead that had been abandoned (Cooke et al. 2008, 155). This site had two distinct parts to the enclosure ‘divided’ by an area of metalling, and was clearly home to a variety of activities including burial, animal management, crop processing and domestic occupation (ibid., 156–7). Unfortunately, as with other regions, the proportions of farmsteads that can be morphologically classified in the later Roman period drops to an even lower level, falling to just c. 25 per cent. It has previously been argued that such a situation may well reflect a move to more

fig.

Romanch6.indd 220

open-style farmsteads that are generally far more difficult to detect, and in the East region there are some slight indications of this with sites like Strood Hall discussed above, and at Hartismere School, Eye, in Suffolk (Craven 2012). VILLAS IN THE EAST East Anglia has long been noted for its relative lack of villas compared with areas further south and west (Mattingly 2006, 384; J. Davies 2009, 187), and very few of these have been comprehensively excavated and published. The situation is somewhat better in northern Essex with sites like Great Holts Farm and Chignall, but overall our knowledge of villas and their surroundings in the East region remains fairly scant. In total there are just fourteen excavated villas in the current dataset, though a further nine have been recorded as ‘possible villas’, mostly on the basis of associated building material and excavator’s interpretation. One such ‘possible villa’ was identified through evaluation trenches and geophysical survey at Harlowbury, c. 1.5 km from

6.15.   Distribution of excavated villas in the East region, compared with ‘villas’ recorded in the NMR index

06/09/2016 17:05:54

THE EAST

the major Romano-Celtic temple at Harlow, defined as a complex of buildings within a system of enclosures, and probably related to the illunderstood nucleated settlement near to the temple (Dicks and Chadwick 2010). The distribution of all the certain and possible excavated villas in the East is shown in fig. 6.15 (including those farmsteads that developed into villas), in relation to an additional 88 ‘villas’ from the region as defined by the National Monuments Record (NMR). Even among the far more numerous NMR ‘villas’ there is a distinct geographic bias towards the south and west, and this is also where all of the more ‘definite’ examples are located, indicated by excavation evidence or clear cropmarks. Nearly all of the NMR ‘villas’ to the north and east are defined through finds scatters or are buildings of uncertain nature, suggesting that the actual pattern of villa scarcity in this area is likely to be genuine. One of the only certain excavated villas in this north-east area was discovered fairly recently c. 1.5 km south of the civitas capital at Caistor, comprising a modest late Roman corridor villa building, along with substantial aisled structures and a highly unusual masonry Y-shaped building, 150 m from the villa and potentially preceding it (Bowden 2011; see fig. 6.27). Villas are not only reasonably rare within much of the East region, they are also founded comparatively late (compared to further south), with the earliest starting in the second century

221

a.d.

and the peak of occupation being reached during the later third century (fig. 6.16). Although continuity of occupation cannot always be demonstrated, almost all of the villas where we have evidence developed from existing farmsteads, at least four of which originated in the preconquest period. The late Roman emphasis of villas is quite different to the pattern for farmsteads, suggesting that, in parts of the south of the region at least, villa estates may have been established at the expense of existing farmsteads, perhaps as part of wider-scale landscape reorganisation. Limited evidence exists for the immediate context around villa buildings in the region, though elements of what has been interpreted as a substantial planned estate were found in excavations to the south of a large courtyard villa at Chignall in Essex near the floodplain of the River Can (C.P. Clarke 1998). These included stock enclosures, outer fields, a possible detached bathhouse, cemetery, and timber structures of agricultural and domestic function, presumably used by the workers of the estate (ibid.) (fig. 6.17). Elsewhere, what would appear to be similar ranges of enclosures and buildings were observed at villas such as Great Holts Farm, Gestingthorpe, Rivenhall, and Little Wymondley, though rarely do any coherent plans of the villa complex emerge, and specialist information on finds and environmental material that might provide more information on the functioning of the estate is often lacking.

no. of villas in use

12 10 8 6 4 2 0 AD50 100

AD100 - AD150 - AD200 - AD250 - AD300 - AD350 150 200 250 300 350 400

600 500

Date BC/AD

400 300 200 100 10 -100

date of villa establishment

fig.

Romanch6.indd 221

6.16.   Chronological patterns of excavated villas in the East region

06/09/2016 17:05:55

222

THE RURAL SETTLEMENT OF ROMAN BRITAIN (cropmark of) Roman villa

N

0

100 m 1:2000

fig. 6.17.   Plan of excavations south of the villa at Chignall, Essex (C.P. Clarke 1998), revealing part of the outer complex

NUCLEATED SETTLEMENTS: ROADSIDE SETTLEMENTS, VICI AND VILLAGES Roadside settlements The East region dataset contains records relating to 28 nucleated settlements lying upon the main Roman road system, representing a far higher proportion of the overall settlement number than in any other region (17 per cent as opposed to 7.5 per cent in the Central Belt). As discussed in Chapter 2, such settlements probably formed a key economic element of much of central and southern Roman Britain and their regular spacing along the roadways of the east meant that few farmsteads would have lain more than a day’s journey away (c. 20 km) (fig. 6.18). The economic and social functions of roadside settlements in the East will

Romanch6.indd 222

be explored more fully below, but here it is important to just reiterate the huge variety in form, scale and development of the places that fall within this broad category of settlement (cf. Ch. 2). Such settlements in the East were usually located at the junctions of major roadways, with ditched enclosure plots, timber and sometimes masonry buildings, pits, quarries and other features spread out along the road frontages, with occasional evidence for side lanes. Elements of ditched boundaries encompassing at least parts of the settlement were revealed at Kelvedon (Rodwell 1988) and Wixoe (Atkins 2013), though seemingly not to the same scale as the earthwork defences at Brampton or Chelmsford. Rarely does there appear to be any central focus, though occasional ‘public’ buildings are hinted at, such as the

06/09/2016 17:05:56

THE EAST

fig.

6.18.   Distribution of nucleated settlement in the East region

substantial masonry buildings at Baldock (Stead and Rigby 1986) and Braughing (Partridge 1977). It is generally very difficult to determine the overall extent of these settlements, especially as very few have had any large-scale excavation (just seven sites over 1 ha) and/or geophysical survey. Where extents are known, or at least estimated, they range from c. 6–7 ha at Harlow to an exceptional c. 80 ha at Baldock (albeit quite dispersed), with most in the range of 15–30 ha. The settlement at Baldock, located in a bowl in the north-eastern tip of the Chilterns, was remarkable in many ways, and has been quite comprehensively explored over a number of years, with 7 ha of excavation and 25 ha of geophysical survey (Stead and Rigby 1986; Burleigh and Fitzpatrick-Matthews 2010; fig. 6.19). At other sites, however, such as Coddenham in Suffolk and Great Walsingham in Norfolk, there has been very little excavation, though on the basis of extensive fieldwalking and metal-detector surveys the areas of these settlements have been estimated to be in the range of 50–60 ha (though this possibly includes finds from manure scatters in surrounding fields).

Romanch6.indd 223

223

The considerable size of many of the roadside settlements, together with the often limited areas of investigation, ensures that their development over time is rarely well understood. At least twelve showed some evidence of pre-conquest activity, mostly quite insubstantial in nature, though sites like Baldock and Braughing in Hertfordshire were important late Iron Age settlements (often classified as oppida), with evidence for high-status burials suggesting that they were elite centres of power. A possible roadside settlement at Thetford, interpreted on the basis of enclosure ditches, timber structures, pits and corndryers lying near to a main Roman road (Andrews and Penn 1999) may also have succeeded an important late Iron Age centre of power, although it seems to have remained a fairly modest and essentially ‘rural’ settlement, possibly developing from the Iron Age site after a lengthy hiatus. The early Roman development of some roadside settlements appears to have been stimulated by the Roman military, with first-century a.d. forts postulated at Kelvedon, Coddenham, Pakenham, Saham Toney and Billingford, forming a line up to Colchester and from there into the heart of East

06/09/2016 17:06:04

224

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Building/possible Temple Enclosure Cemetery Road Quarry 0

500 m 1:10000

fig. 6.19.   Plan of Roman roadside settlement at Baldock, Herts. (after Burleigh and Fitzpatrick-Matthews 2010, fig. 18)

Anglia, perhaps providing some level of monitoring during the post-Boudican period (J. Davies 2009, 150). Whether associated with the military or not, most settlements grew fairly rapidly following the development of the road system in the later first century a.d., though it was not really until the end of the first or early second century that many reached a more substantial size, often acquiring a degree of planning, however rudimentary this may have been. At Scole in the Waveney Valley for example, mid–later first century a.d. evidence was restricted to cess and rubbish pits, roundhouses and posthole structures, while during the early– mid-second century some measure of centralised planning is thought to have appeared (Ashwin and Tester 2014, 217). Scole, like most roadside settlements in the East, appeared to flourish in the second and third centuries a.d., with many small-scale developments, but no major alterations in form, scale or complexity. At Needham in Norfolk and Radwinter in Essex there was little late Roman evidence, but this is more likely to reflect the very small scale of archaeological investigation than a genuine cessation of activity. Most roadside settlements

Romanch6.indd 224

continued to be occupied until the end of the fourth century, although at least eight have evidence to suggest they were in decline long before this. Most investigations within the settlement at Long Melford in Suffolk concluded that all significant activity ceased after the mid-third century, though occupation of sorts continued until the end of the Roman period (e.g. Craven 2008). More typically the period of decline extends from the mid-fourth century, as for example at Braintree, Kelvedon and Hacheston. The dwindling fortunes of at least some roadside settlements at this time seem to parallel a wider rural settlement decline, and certainly does not suggest that the inhabitants of abandoned farmsteads were all flocking to these larger centres. The situation of such sites in the post-Roman period has recently been discussed by Fitzpatrick-Matthews (2014), who highlighted how the post-Roman evidence from Baldock may not be quite as exceptional as it currently seems, with fifth-century data awaiting to be discovered from other sites. As it stands, only three other roadside settlements in the east apart from Baldock are argued to have had fifth-century occupation, Wixoe, Billingford and Saham Toney.

06/09/2016 17:06:05

THE EAST

225

N

Late Roman fort

Eastern settlement

0

200 m 1:5000

fig. 6.20   Plan of nucleated settlement to the east of the coastal fort at Brancaster, Norfolk (Hinchliffe and SpareyGreen 1985)

Other nucleated settlement In addition to nucleated settlements of varying sizes sited upon the main Roman road system, there was a variety of other often substantial settlements in the East region. A series of smallscale excavations at Felixstowe in Suffolk revealed parts of a large Roman port, possibly extending over c. 30–40 ha including a road/street system and masonry-footed buildings (Fairclough 2011). There would certainly have been other ports around the East Anglian coastline, probably including Hunstanton in North Norfolk, which lay near the end of the Peddars Way Roman road, and from where goods may have been traded across the Wash (J. Davies 2009, 190). Lying 9 km to the east, and overlooking the coastal saltmarshes, was a nucleated settlement at Brancaster, revealed by cropmarks and geophysical survey as a system of rectilinear ditched boundary plots and streets covering over 23 ha (fig. 6.20; Hinchliffe and Sparey-Green 1985; Wessex Archaeology 2014). These lay to the east of a late Roman coastal fort (with another area of settlement to the west), but seem to have originated earlier, in the later second century a.d., probably before the foundation of the fort in the early third century. Although an earlier military site has been postulated, this remains unproven and so it remains uncertain if the settlement at Brancaster was a military vicus from the start, or else a coastal roadside settlement (port?) with a later associated fort. This may well have been the case at Felixstowe, the long-lived Roman port settlement noted above, which was only later possibly associated with a fort established

Romanch6.indd 225

at Walton Castle (Fairclough 2011). Two other coastal forts at Burgh Castle and Caister-on-Sea both have excavated elements of what are presumed to be the associated vici (though activity at Caister also started prior to the establishment of the fort), and cropmarks indicate a particularly dense zone of settlement, trackways and fields surrounding them both (see fig. 6.4). Finally there are three fairly ill-understood settlements that appear to be of substantial size, but which do not appear to be orientated on the main Roman road network and so are here interpreted as villages. These sites, at Handford Road, Ipswich, RAF Lakenheath and Wenhaston, all span the whole Roman period with possible late Iron Age origins, and had excavated features comprising timber buildings, ditches, pits and wells. Plough marks immediately north of the RAF Lakenheath settlement indicate associated cultivation (Craven 2010), characteristic of what would seem to be essentially agglomerated agricultural settlements. BUILDINGS The East region records produced a total of 385 buildings from 107 excavated settlements. With just over two buildings per settlement, the region falls just below the average for England and Wales (2.5 buildings) but above that of the South (1.9 buildings per settlement). The evidence is not evenly distributed across the region, with a higher proportion of settlements to the south and west

06/09/2016 17:06:06

226

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig.

6.21.   Distribution and frequency of rural buildings on excavated sites in the East region

fig.

6.22.   Distribution and frequency of circular and rectangular architecture in the East region

containing evidence for buildings (fig. 6.21), though even here the number of examples generally remains quite low, the maximum being 22 in the nucleated settlement at Braughing, Herts (Partridge 1977; 1981).

Romanch6.indd 226

Rectilinear buildings are far more numerous than circular structures across the region, with the latter being concentrated in parts of the south and west (fig. 6.22). The overall dominance of rectilinear structures starts from the later first

06/09/2016 17:06:15

THE EAST Circular

70

227

Rectangular

60 no. of sites

50 40 30 20 10 0 Late Iron Age Late 1stC AD fig.

2ndC AD

3rdC AD

4thC AD

6.23.   Number of sites with circular and/or rectangular buildings over time in the East region a

River Roding b N

0

20 m 1:500

0

200 m 1:5000

fig. 6.24   Plans of late Roman (a) workshop and (b) granary/storehouse at Frogs Hall Borrow Pit, Takeley, Essex (Ennis 2006)

century a.d., and becomes much more apparent from the second century onwards, although circular buildings still appear right through into the fourth century (fig. 6.23). This does not always seem to have been due to inherent conservatism, as in at least six of the eleven sites with evidence for late Roman circular structures, there was no earlier evidence for such buildings. Indeed, at both Frogs Hall Borrow Pit, Takeley in Essex (Ennis 2006) and Kilverstone in Norfolk (Garrow et al. 2006), the late circular buildings appear to replace earlier rectangular structures. Furthermore, the function of circular buildings in the later Roman period, as far as it is possible to tell, was more likely to be non-domestic, as for example the shrines at Kelvedon (Rodwell 1988) and Hockwold (Gurney 1986), and the ‘workshops’ at Takeley (fig. 6.24(a)) and Kilverstone, interpreted as such on the basis of metalworking debris. Nevertheless, there were a few examples of what do appear to have been domestic roundhouses

Romanch6.indd 227

occurring in late Roman contexts, with five being recorded alongside three rectangular buildings within the complex farmstead at MTCP, Stansted (Cooke et al. 2008; see fig. 6.13b). FARMSTEADS Most of our evidence for buildings in the East region comes from farmsteads, both in terms of the number of sites and total number of structures (table 6.3). The chronological patterning of circular and rectilinear buildings at farmsteads generally follows that of the region as a whole (i.e. a rise in number of sites with rectangular buildings and decline in sites with circular buildings; fig. 6.23), except it is not until the second century that the latter came to dominate, and then not quite to such a marked degree. Complex farmsteads are far more likely to be associated with rectilinear buildings than enclosed farmsteads, although this is mostly dictated by chronological factors as the situation is more mixed in the mid- to late Roman

06/09/2016 17:06:16

228

THE RURAL SETTLEMENT OF ROMAN BRITAIN

table

6.3: site

Site type

context of excavated buildings in the east region

Circular Rectangular No. circular No. rectangular No. timber No. masonry buildings buildings buildings buildings buildings buildings (no. sites) (no. sites)

Farmstead

31 35 100

73 165

8

Farmstead/villa

2 6 3 22 7 18

Villa

0 9 0 23 1 22

Roadside/vicus

8 21

17 107

90

26

Village

1 2 2 5 7 0

Religious site

3 7 3 15 5 13

Agricultural building

1

5

2

10

12

0

Isolated building 0 3 0 3 0 3 Total

46 88 127 258 287

a

c

b a

90

b

c

N

Strood Hall, Essex

Melford Meadows, Norfolk 0

Brandon Road, Norfolk

20 m 1:500

fig. 6.25  Plans of farmstead buildings at (a) Strood Hall, Essex (Timby et al. 2007a), (b) Melford Meadows, Norfolk (Mudd 2002) and (c) Brandon Road, Norfolk (Bales 2004)

period, exemplified by the differing but contemporary structural forms noted above at the Stansted MTCP site. As discussed in Chapter 3, determination of specific building function is often not possible, especially given the usual scant physical remains, though most are assumed to have incorporated domestic occupation on the basis of hearths and/ or associated finds. Others are likely to have been workshops, as noted above for Kilverstone and Takeley, and agricultural buildings, such as another building recorded at Takeley, which was interpreted as a post-built granary/storehouse on the basis of charred grain/seeds (including pea and field bean) from many of the postholes (fig. 6.24b; Ennis 2006, 37–40). Additional agricultural buildings from farmsteads in the East include a substantial, post-built granary from Elveden By-pass, Suffolk (Booth 2014, 363–5), and a late first–second century timber building interpreted as a cattle byre from Rayne in Essex (Smoothy 1989). Over 95 per cent of farmstead buildings in the East appear to be of timber or mass-walled

Romanch6.indd 228

construction (see Ch. 3), with circular structures most often defined by drainage gullies and rectangular buildings by a mixture of postholes and beam slots. There are many problems in identifying some of these buildings, exemplified by what would appear to be a substantial structure of third-century date from Strood Hall in Essex, defined only as a rectilinear hollow (14 × 8 m) and infilled with midden material (Timby et al. 2007a, 89; fig. 6.25a). Of the seven farmsteads with building(s) having at least some form of masonry foundation, at four (Feltwell, Coggeshall, Bottisham and Great Wilbraham) these were thought to have been agricultural buildings or shrines associated with wider villa estates, though mostly with fairly minimal supporting evidence. Elsewhere, structures were suggested as possible workshops at Fosters End Drove and Sheringham Shoals (the former incorporating a heated room for drying; Lally and Nicholson 2008), and a building of uncertain function with cobble foundations was revealed at Maltings Lane, Witham. The latter site

06/09/2016 17:06:17

THE EAST

may also have been part of a villa estate as significant quantities of window glass, tesserae, tile, opus signinum and painted plaster were found in the vicinity (Robertson and Davis 2004). Such elaboration is virtually non-existent at other farmsteads in the East, though six others did have evidence for window glass. The vast majority of all farmstead buildings, timber or masonry, would appear to have been quite simple, single-room structures, at least from the surviving sub-surface evidence, though the occasional existence of more sophisticated buildings is hinted at by a substantial (c. 10 × 6 m) three-room timber building of beamslot construction, dating to the early Roman period, at Melford Meadows, Brettenham, Thetford (Mudd 2002; fig. 6.25b). The most substantial structures to be found on eastern farmsteads are aisled buildings, of which eight have been recorded from six sites, all defined as complex farmsteads where classification was possible. Three of these sites comprise farmsteads around Thetford, with two buildings from Brandon Road interpreted as agricultural barns dating to the second/third century a.d. (Atkins and Connor 2010; fig. 6.25(c)). All of the other aisled buildings were of a similar date, though three continued into the fourth century. They were of ‘undeveloped’ type, and quite modest in size (most under 200 m²) compared with some of those structures found on villa complexes. As with Brandon Road, it is likely that all were utilised as agricultural barns, one at Beck Row, Mildenhall containing a corndryer/malting oven and suggested as a possible malting house (Bales 2004). Nevertheless, however modest these may seem compared to certain other aisled buildings across the country, they still would have been strikingly different to other buildings on farmsteads in the East, suggesting perhaps a special social status. VILLA COMPLEXES The fourteen excavated villa sites from the East region contained 48 buildings, including 33 ancillary buildings in addition to the main villa houses. Compared to areas further west and south, the scale of most villa houses was quite modest, with five coming under the ‘cottage/hall’ type category, such as at Exning in Cambridgeshire where the villa and associated bath suite developed from an earlier aisled building during the later third century a.d. (Webster 1987; fig. 6.26a). The main late Roman villa building at Great Holts Farm was particularly modest, with the timberframed structure only really being assigned to ‘villa status’ on the basis of overall form and the presence of an attached bathhouse (Germany 2003). Nevertheless, there were some grander and more extensive villa building complexes in the

Romanch6.indd 229

229

East, such as an unexcavated courtyard villa near the floodplain of the River Can at Chignall in Essex, which was revealed by cropmarks (C.P. Clarke 1998; see fig. 6.17). Another example is suspected at Stanton Chair, Ixworth in Suffolk, which developed from an aisled or corridor building in the later Roman period (Maynard and Brown 1936). Large winged-corridor buildings have also been revealed at Rivenhall in Essex (Rodwell and Rodwell 1993) and Gayton Thorpe in Norfolk (de Bootman 1998), both associated with other buildings, the latter including an adjacent winged building to the south on a slightly different alignment (fig. 6.26b). One of the more recent villa excavations in the East lies 1.5 km south of Caistor Roman town in Norfolk, comprising a fairly modest corridor building (6+ rooms) dating to the late Roman period (Bowden 2011). Lying 150 m south of the villa was a highly unusual Y-shaped structure, with two splayed wings 13.5 m long, converging on a rectangular room, with an apsed room behind (fig. 6.27). The function of this building is uncertain, though it has been suggested as a summer dining room (triclinium) or shrine, and it may well be earlier than the main villa building (ibid.). Most ancillary buildings in villa complexes are considerably less exotic, comprising for the most part structures presumed to have been used for agricultural purposes, such as the six aisled buildings revealed on four villa sites. One of these at Rivenhall was particularly massive; at 880 m² it is among the largest such buildings known from the country and surely a huge statement of the wealth and social standing of the occupants of the site. Another agricultural structure, lying 250 m from the main villa at Stebbing Green in Essex, was of ill-defined, probable sill-beam construction, and appears to have associated with the production of malt for brewing, a potentially important part of the villa’s economy (Bedwin and Bedwin 1999). NUCLEATED SETTLEMENTS Nucleated settlements (roadside settlements, military vici and villages) in the East region contained evidence for many structures, with a diverse array of different building types (table 6.3). The great majority (123) of buildings from nucleated sites were recorded from roadside settlements, with over 86 per cent of these being rectilinear in form, including an aisled building recorded at Braintree in Essex (Garwood and Lavender 2000). The remaining seventeen circular structures comprise three shrines and scatters of domestic roundhouses, all of the latter dating to the late Iron Age or early Roman period, prior to or during the early stages of the development of roadside settlements.

06/09/2016 17:06:17

230

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

b

N

N

Exning, Cambs

0

20 m 1:500

Gayton Thorpe, Norfolk fig. 6.26.   Plans of villa buildings from (a) Exning, Cambs (Webster 1987) and (b) Gayton Thorpe, Norfolk (de Bootman 1998)

Most buildings from roadside settlements were fairly modest timber structures, as typified by the group excavated from Scole on the Norfolk/ Suffolk border, where seventeen timber buildings were constructed using a mix of earthfast posts, sill beams and post-pads (Ashwin and Tester 2014, 219). Although none of these was on a grand scale, the recovery of well-preserved roof timbers re-used in a late second-century malting complex did testify to a high level of carpentry skills. One ‘strip building’ was revealed, with what is suggested as commercial space occupying a

Romanch6.indd 230

portico to the front facing the road, and workshops and domestic space behind (ibid.). Such buildings are frequently observed at some of the larger urban centres in Roman Britain, such as at London and Colchester, though they were more common in the early Roman period (Perring 2002, 41). Seven posthole and/or beam-slot buildings were also excavated in the settlement at Wixoe on the eastern side of the River Stour, although more sophisticated masonry buildings (including a possible bathhouse or administrative building) were also indicated by geophysical

06/09/2016 17:06:18

THE EAST

231

N “Corridor villa”

Approximate location of tiled building

Aisled building

Aisled building

Winged building

0

100 m 1:2000

Winged building

Aisled building

0

20 m 1:500

fig.

Romanch6.indd 231

6.27.   Plan of villa complex south of Caistor, Norfolk (Bowden 2011)

06/09/2016 17:06:18

232

THE RURAL SETTLEMENT OF ROMAN BRITAIN

survey and quantities of building material, which included roof tile, wall plaster and fragments of opus signinum (Atkins 2013). About half of all East region roadside settlements had evidence for at least one masonry building, though many of these were only partially revealed and seemingly of fairly simple construction, probably with timber superstructures. Stonefounded shrines/temples were attested at Scole, Crownthorpe, Hockwold and Great Dunmow, while definite bathhouses were located at Braughing, Long Melford and Great Walsingham. The provision of bathing establishments implies a certain level of ‘Roman’ social infrastructure that was usually reserved for larger towns and cities (along with villas), although there is no way of knowing if they were for general public usage, or more restricted, perhaps for state officials. Other masonry buildings of some status include a possible mansio within the settlement at Baldock, the multi-roomed building at Wixoe mentioned above, and two stone-built hypocausted buildings found near the road frontage at Pakenham (Tester 2002). With the possible exception of the latter, these have all been interpreted as ‘official’ buildings, and there is scant evidence for highstatus domestic dwellings within the core areas of such settlements. LANDSCAPE CONTEXT AND INFRASTRUCTURE TOPOGRAPHY AND GEOLOGY The East region landscape is generally fairly lowlying, with some higher ground to the south-west, though rarely rising above 130 m OD. As with other regions, topographic factors appear to have had at least some influence on the location of settlements within this landscape, and this is of course inextricably linked to underlying geology and soils. The boulder clays of the East Anglian Plain clearly form the most extensive zone and have the largest number of settlements, although these are generally concentrated in or near the river valleys that dissect this landscape. The settlements lying further west favour the lower chalk zone, particularly where this is cut through by the wide river valleys such as those of the Cam or Granta. Notwithstanding the major issues of bias associated with the wider settlement patterns noted earlier in this chapter, the general predilection of settlements in the region for relatively lower-lying ground is clear, with almost 70 per cent of all sites lying below 50 m OD. Within different types of settlement, however, divergent patterns emerge. In particular, complex farmsteads and roadside settlements are more

Romanch6.indd 232

likely to be found at very low levels (especially 10–19 m OD), often on sand and gravel riverterraces, much as in other regions of central and southern Britain. The only exceptions to this, in terms of complex farmsteads, are a relatively tightly defined group in the vicinity of Roman Stane Street (modern A120), including MTCP site at Stansted airport, which lie on the higher Essex claylands at 83–105 m OD. The concentration of eastern roadside settlements at relatively low levels is partly because c. 75 per cent of them were located at a river crossing or alongside river banks, and even those few found at higher levels (e.g. Great Dunmow and Braughing; 75–77 m OD) were still found close to rivers (see Ch. 2). Villas tend to cluster at a slightly higher level, though again often still sited in river valleys such as at Castle Hill, Ipswich, which lay just below the crest of a low hill in the Gipping Valley (Wessex Archaeology 2003). Enclosed farmsteads likewise seem to have been more common at higher levels, certainly more than the majority of complex farmsteads, and were the most dominant settlement type above 50 m OD, being particularly prevalent on the boulder claylands. TRANSPORTATION: RIVERS, ROADS AND TRACKWAYS Despite the emphasis on habitation in the lowerlying river valleys, and the fact that over 60 per cent of all settlements lay within 2 km of a significant river, evidence for use of the inland waterways remains almost non-existent in the region. Excavations at Cranfield Mill in Ipswich revealed some evidence for what was argued as ‘Roman river management’ (Medlycott 2011a, 41), while timber wharfs/jetties were noted along the river fronts in the roadside settlement at Scole (Ashwin and Tester 2014, 222) and in the defended Roman town at Brampton (Knowles 1977). The fact that so many roadside settlements were situated at river crossings might suggest that rivers were utilised far more than this evidence indicates (though the instigation of tolls was also probably a factor; see Ch. 2), but in terms of transport links they were always likely to have been secondary to the major road network, which appears to have been quite comprehensive in this region. Almost all the excavated settlements in the East region dataset lay within 10 km of the Roman road network as defined by the National Monuments Record, while a more intimate association is suggested for villas and complex farmsteads, with over 70 per cent of both types lying within 2 km, as opposed to just 50 per cent of enclosed farmsteads. A similar situation was observed for the Central Belt region, the pattern perhaps explained by the greater economic and social draw

06/09/2016 17:06:19

THE EAST

of the road network within these parts of the Roman province, and in particular a close connection of villas and complex farmsteads with roadside settlements. More localised movement through the landscape is also revealed by trackways on 110 sites, fairly well distributed throughout the East region. Trackways were specifically noted on 71 settlements (c. 65 per cent of total), most dating to the Roman period, though possible trackways of pre-conquest date were recorded at five sites, including Shropham Quarry in Norfolk where a track/droveway was revealed leading towards an enclosed late Iron Age/early Roman settlement on a low hill (Carlyle 2011b). The extent of trackways around settlements and across wider landscapes is rarely revealed through excavation, though the cropmark surveys in parts of Norfolk have revealed areas of integrated trackways, fields and settlements spreading over hundreds of hectares (see fig. 6.4 above). FIELD SYSTEMS The high agricultural potential of much of the East region landscape is well noted, with almost all of it covered by open fields during the medieval period, enabling it to become one of the most prosperous regions in Britain (Williamson 2005, 13). Today it remains the principal grain-producing

233

area of the country. The scale and intensity of agricultural production in the Romano-British period is less well known, although there is considerable evidence for field systems in many parts of the region. A total of 60 sites with excavated field systems is included in the current dataset for the region, many of which lay in or near to the river valleys dissecting the East Anglian Plain. A large proportion (42 per cent) of these field systems has direct association with farmsteads. They often just comprise a series of ditches radiating out from the main settlement, though rarely traced as far as those at the MTCP site, Stansted, which ran in very straight lines from the farmstead for over 300 m (Cooke et al. 2008; fig. 6.28). Elements of field systems were also associated with four villas and nine roadside settlements, the latter often recorded immediately outside the settlement core, suggesting that the occupants of such nucleated sites were generating agricultural produce themselves rather than relying wholly on outlying farmsteads. The remaining twenty excavated field systems had no obvious associated settlement and rarely form any coherent pattern, though this is largely to do with the limited overall area of excavation. As with trackways, it is only really when cropmark evidence is sought that the wider picture of Romano-British fields can be discerned, as seen with the extensive

N

0

200 m 1:5000

fig. 6.28.   Plan of field system ditches radiating out from late Roman settlement at MTCP site, Stansted (Cooke et al. 2008)

Romanch6.indd 233

06/09/2016 17:06:19

234

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig.

6.29.   ‘Roman’ field systems in the Norfolk Broads as revealed through NMP cropmarks (for location see fig. 6.4) (cropmark data © Historic England)

areas of small co-axial fields on the boulder clay adjacent to the Norfolk Broads, which are believed to be of later Iron Age and and/or Roman origin (fig. 6.29). These would seem quite different from larger fields revealed on the boulder clay around Stansted, perhaps hinting at differences in the agricultural regimes. As discussed in previous chapters, accurate dating of field systems remains very problematic, though there is good evidence from the East for many having origins in the Iron Age or earlier. Independently dated excavated examples (i.e. not dated purely by association with settlement) appear to be most numerous in the late first and second centuries a.d. (up to sixteen in use), with a steady decline thereafter (to ten in use by fourth century). This apparent ‘abandonment’ (or at least silting up of ditches) of many field systems prior to the end of the Roman period is found across much of the central and southern regions, though as elsewhere there is some evidence that the alignment of these fields (especially in the Boulder Clay of the East Anglian Plain) continued to shape the landscape for many centuries to come (Rippon et al. 2015, 169–81).

Romanch6.indd 234

SETTLEMENT HIERARCHIES: THE SOCIAL AND ECONOMIC BASIS OF SETTLEMENTS While there may have been a relative paucity of high-status villas and a low level of architectural elaboration within much of the East, there is a particularly large and rich finds assemblage from the region, along with considerable levels of information derived from environmental remains, although the latter is somewhat concentrated towards the south and west. When assessed together with the settlement and landscape evidence, these data provide great scope for exploring variations in economy and social structure, which in turn leads to a greater understanding of the inter-relationships between rural settlements. MATERIAL CULTURE The general range of objects from the East region is similar to that from the South and Central Belt, and has much the same order of prevalence, i.e. with pottery and coins being the most common objects and categories such as lighting, recreation

06/09/2016 17:06:28

% of all sites (n=182)

THE EAST 80 70 60 50 40 30 20 10 0

6.30.   Frequency of major artefact categories on all sites in the East region table

6.4: distribution

of selected object categories across landscape zones of the east region

(%

of total no. object type)

Equine/transport equipment

fig.

235

Writing equipment

Weighing object

Textile processing

Security object

Religious object

Recreation object

Lighting equipment

Knife/tool

Hairpin

Food processing

Coin

Brooch

Bracelet

Landscape zone

Agricultural tool



Breckland

12.3 10.42 6.64 11.13 2.31 16.6 6.17 8.23 0

East Anglian Chalk

18.03 14.63 12.9

1.64 29.6 1.09 26.78 6.29 9.2

East Anglian Plain

65.57 71.4 64.94 54.71 71.54 62.83 77.59 68.27 27.45 72.13 45.6 75.54 53.32 81.71 62.07

3.95 17.69 17.44 12.03 18.67 72.55 24.59 8.8 15.22 8.29 5.71 17.24

North Norfolk 3.28 3.33 13.9 27.56 8.46 2.74 4.06 4.02 0

1.64 15.2 7.07 10.66 5.71 11.49

Suffolk Coast and Heaths

0

0.82 0.22 1.52 2.62 0

0.38 0.15 0.8 0

0.8 1.09 0.95 0.57 0

The Broads 0 0 0.1 0.03 0 0 0 0 0 0 0 0 0 0 0 Total no. of objects

122 451 1914 28609 130 1313 665 498 51 61 125 184 422 175 87

and writing being the scarcest. Pottery has been recorded from all settlements (though just 70 per cent was quantified), with samian and amphorae being noted on a higher percentage of settlements here than in any other region (75 per cent and 40 per cent respectively); mortaria were approximately equal with the Central Belt, being found on c. 50 per cent of sites in both regions. The frequency of occurrence of most nonpottery object types is also much higher in this region, with, for example, 74 per cent of sites containing coins as opposed to 58 per cent in the Central Belt and less than 50 per cent in the south (fig. 6.30). The high prevalence of what may be termed ‘elite’ material culture (e.g. certain personal objects such as hairpins and finger rings) in particular has been noted for the region, with Mattingly arguing that such portable wealth (as

Romanch6.indd 235

opposed to architectural wealth) was part of ‘specific local patterns of adaptation to Roman rule’ (2006, 385) in terms of the creation and maintenance of social relations and identity. The quantities of objects within selected categories found in the East region are shown in table 6.4, shown as percentages recovered from the different landscape zones. As would be expected given that this is the largest zone with most settlement data, the greatest volume of almost all object types come from the East Anglian Plain, though elsewhere there are certain patterns that may hint at intra-regional differences in social and economic structures. North Norfolk in particular appears to have very few objects associated with agriculture (these deriving mostly from Brancaster), perhaps correlating with an almost total lack of millstones to suggest less

06/09/2016 17:06:29

236

fig.

% of sites with object present

100

% of sites with object present

THE RURAL SETTLEMENT OF ROMAN BRITAIN

100

80 60 40 20 0

80 60 40 20 0

villa (n=14) farmstead (n=143)

villa (n=14) complex farmstead (n=22) enclosed farmstead (n=28)

6.31.   Frequency of major artefact categories on different types of rural settlement in the East region

intensive arable cultivation in this area. Elsewhere, there is some evidence that more intensive agricultural regimes operated in parts of Breckland and the East Anglian Chalk, with these areas not only having relatively high proportions of agricultural tools and food-processing objects (including millstones), but also being provided with quite a number of aisled buildings (for storage) and corndryers. Many of the differences between landscape zones noted in table 6.4 result from variations in their settlement profiles because, as with other areas of southern and central Britain, there is a distinctly uneven distribution of different object classes and quantities between settlement types. As noted above, the most ubiquitous class of artefact found across all site types remains pottery, yet assemblages quantified to any degree were only noted for 80 per cent of farmsteads, 55 per cent of nucleated settlements and just 42 per cent of villas. Records of certain other types of object appear more ‘complete’, and here we see, for example, 100 per cent of both villas and roadside settlements having quantified evidence for Roman coinage (fig. 6.31). In terms of absolute numbers of coins, roadside settlements completely dominate, accounting for 85 per cent of the c. 28,600 coins recorded for the East region, including six sites with quantities over 1000. Although this was largely due to the use of metal detectors on excavations, particularly at sites in Norfolk and Suffolk, the sheer volume of coins from these settlements must surely be a reflection of their important role as market centres.

Romanch6.indd 236

roadside /vicus (n=29)

Most other object categories are also more frequently present at roadside settlements, reflecting both their larger populations and their wide range of different social and economic functions. Brooches are particularly prevalent, occurring on almost 80 per cent of such sites and accounting for 73 per cent of the total number for the East. Along with the prevalence of other personal items, this points to a particular emphasis on personal display, as might be expected of settlements where movement and circulation of people would be commonplace. Food-processing objects were recorded at 75 per cent of roadside settlements yet, unlike in the Central Belt (and to some extent the South), very few had specific evidence for millstones. That at least some of them functioned as cereal-processing centres, however, is indicated by the thirteen fragments recovered at Scole, where a possible water mill was noted, along with a malting complex (Buckley 2014). Commercial activity is also attested by relatively high incidences of objects associated with weighing and writing (in addition to the coinage), while a high level of textile processing is hinted at by the many spindlewhorls, needles, loomweights (though these are generally thought to date to the later Iron Age/earliest Roman period) and shears found at seventeen roadside settlements. There are also occasional examples of wool combs and weaving tablets on such sites, suggesting that all stages of textile processing were being carried out, with sheep perhaps being driven in from the surrounding countryside (see Faunal remains below and see vol. 2). The whole eastern

06/09/2016 17:06:31

THE EAST

region and Norfolk in particular has been argued as being especially important for sheep-rearing and textile manufacture, with suggestions that an imperial woollen mill recorded at ‘Venta’ in the Notitia Dignitatum was actually at Caistor-byNorwich (Venta Icenorum) rather than Winchester as usually proposed (Manning 1966; Crummy 2010). The recovery of religious objects such as figurines and model axes on over 60 per cent of roadside settlements provides some clues about the spiritual focus that such sites could provide, as is also indicated by the presence of structures interpreted as shrines/temples on a number of such sites (e.g. Great Dunmow, Baldock, Crownthorpe). The many religious objects from the settlement at Harlow Holbrooks, including gilt bronze letters, bronze leaves and miniature axes, were probably associated in some way with the Romano-Celtic temple lying less than 0.5 km distant, with some indications (from 200 fragments of copper-alloy sheet) that it may have actually been a manufacturing centre for such items (Conlon 1973). As elsewhere, villas in the East were generally less well represented than roadside settlements in terms of the total range of different objects recovered, although they were more commonly represented in some categories. In particular, a greater percentage of villas contained certain items associated with personal adornment/ grooming (bracelets, finger rings, toilet objects), along with recreation (dice, gaming counters) and security (keys, lock parts), all reflecting a preoccupation with personal appearance, lifestyle and protection. Villas were also best represented with knives/tools, which is perhaps surprising given the considerable evidence for industrial practices at most roadside settlements (pottery production and metalworking in particular). However, the outer ‘working’ complexes of most villa estates (and where many of the East villa excavations were concentrated) may have incorporated a range of different craftworking activities, as seen with the evidence for boneworking, ironworking and bronzeworking at the villa estate at Gestingthorpe, Essex (Draper 1985). As expected, the East region farmsteads generally had a much more restricted range of object types, with 40 sites (28 per cent of total) having evidence for just pottery alone. Nevertheless, the proportion of farmsteads with evidence for coinage is still quite high at c. 60 per cent, which is considerably greater than in farmsteads in the South or Central Belt (40–48 per cent), perhaps influenced by the high density of roadside settlements in this region. Although, as with other

Romanch6.indd 237

237

regions, most farmsteads still had low numbers of coins recovered (overall less than 10 per cent of the East total) this increased presence may hint at greater integration into a monetary economy. However, the overall role of coinage in RomanoBritish society is still not well understood, and it has been recently reiterated that their primary function (in the later Roman period at least) was far more intrinsically linked with urban rather than rural life, in particular tax collection and/or commercial transactions (Walton and Moorhead 2015, 6; see vol. 2). Among the different farmstead categories, complex farmsteads do appear somewhat more likely to have coins than enclosed types (fig. 6.31), and have a higher density of coins per hectare of excavation, though differences are not particularly marked and may well be linked more to chronology (i.e. the fact that more complex farmsteads were of late Roman date) than variations in coin usage. Nevertheless there is a greater representation of all finds types at complex farmsteads, and in terms of items associated with food processing (notably querns), they even surpass villas, both as a proportion of sites with such objects and overwhelmingly with actual numbers of objects. Five complex farmsteads also had millstone fragments, probably reflecting their roles as agricultural processing centres, as for example at Melford Meadows in Thetford, where the farmstead covered c. 2 ha and included at least one aisled building, along with agricultural tools and millstone fragments (Mudd 2002; Pine 2010). ENVIRONMENTAL EVIDENCE The environmental evidence from excavated sites in the East region is somewhat patchy, with a far greater number of plant and especially animal bone assemblages coming from the south and west (in parts of Essex, Herts and Cambs) than further north and east (Norfolk and Suffolk). In terms of faunal remains, such geographic bias has been noted previously (Albarella et al. in prep.), and seems to be at least partly due to the greater extent of acidic soils in the North Norfolk, Suffolk and Breckland heaths and the Norfolk Broads, though it is also due to the relative lack of excavated settlements in these zones. Faunal remains In total, 99 phased animal bone assemblages of over 100 NISP have been recorded from the East region and, despite the geographic bias just mentioned, there are some slight trends noted in the frequency of main domesticates according to different landscape zones (fig. 6.32). As a whole, the assemblages from the East region are very

06/09/2016 17:06:31

238

THE RURAL SETTLEMENT OF ROMAN BRITAIN cattle

70

sheep/goat

pig

60

mean %NISP

50 40 30 20 10 0 East Anglian Plain East England (n=96) East Anglian Chalk (n=60) (n=21)*

Breckland (n=10)

fig. 6.32.  Relative frequency of major livestock taxa across in the East region and in selected landscape zones (mean percentages from sites in each zone) (*excludes large assemblages from Great Chesterford temple)

cattle

70

sheep/goat

pig

Mean %NISP

60 50 40 30 20 10 0 LIA/ER (1stC BC/AD) n=22 fig.

ER (1st-2ndC AD) MR (2nd-3rdC AD) LR (3rd-4thC AD) n=17 n=10 n=22

6.33.   Relative frequency of major livestock taxa over time in the East region

much dominated by cattle, more so than any other region except the North. This dominance is even more marked in the East Anglian Plain, or at least in the southern part of this zone where most of the data are located, though there are indications that sheep/goat (and to a much lesser extent pigs) are more common on the chalk landscapes to the west. This is particularly the case in Breckland, where it has been noted that the light soils are very much suited to sheep-rearing, and were successfully used for this purpose during the medieval period (Crummy 2010, 76). A greater emphasis on cattle herds to the south, if genuine, ties in well with that observed in the adjacent London Basin zone in the South region, where the vast majority of settlement assemblages were firmly cattle based, and may be associated with supply networks to London and Colchester (see Ch. 4). An increase in proportions of cattle over time during the Roman period is well documented, both for central/eastern England and across the province (Albarella et al. in prep.; King 1999). This was certainly the case for the current East region dataset, though the high point appears to have been reached by the mid-Roman period, with the situation changing little thereafter (fig. 6.33).

Romanch6.indd 238

Sheep/goat proportions are rather more stable throughout, though with slight evidence for a late Roman decline, while pigs, always a distant third, become significantly less important after the late Iron Age/early Roman period (first century b.c./ a.d.). This is in part influenced by the very high pig percentages from sites within the Braughing oppidum (up to 49 per cent at Skeleton Green), though late Iron Age phases of farmsteads such as the MTCP site at Stansted also have proportions of pig well in excess of 20 per cent. In addition to evidence for slight intra-regional and chronological variation in the faunal remains, there are also some notable differences in the assemblages from different settlement types (fig. 6.34). Unfortunately only four phased assemblages derived from villa sites, two from the same settlement at Chignall, where excavations were located to the south of the main courtyard villa. The assemblages from here (especially that of the late Roman phase) together with those from Great Holts Farm and Stebbing Green, all in Essex, were completely cattle dominated, while the very low percentages of pig compared to those from villas further south and west may be explained because most assemblages were from locations well

06/09/2016 17:06:32

THE EAST cattle

sheep/goat

239 pig

80 70 mean %NISP

60 50 40 30 20 10 0 villa (n=4)

enclosed farmstead (all) complex farmstead (n=8) (n=49) farmstead (n=21)

roadside settlement (n=28)

fig. 6.34.   Relative frequency of major livestock taxa across different settlement types in the East region (mean percentages from site types)

removed from the main villa buildings. There was very little difference noted between the two major types of farmstead, both being almost equally cattle dominated, though roadside settlements did have a markedly higher proportion of sheep/goat, similar to the situation in the Central Belt (see Ch. 5). This may well be linked with the relatively strong association of such sites with objects of textile processing noted above, all perhaps being part of a system of larger scale cloth production. Within these broad regional and intra-regional trends there are of course many individual variants, reflecting local circumstances, choices of the farming communities and post-depositional factors. A late Iron Age/early Roman farmstead on a chalk hill at Edix Hill, Barrington, for instance, had relatively high proportions of sheep/goat (60 per cent of main domesticates) and was interpreted as being geared up toward livestock farming, with ditched and fenced boundaries defining droveways and stock enclosures (Malim 1997). Elsewhere, an especially cattle-dominated assemblage was noted from the roadside settlement at Braintree (83 per cent; Havis 1993), while the major Roman temple sites at Harlow and Great Chesterford have substantial faunal assemblages, renowned for their dominance of sheep/goat (85 per cent and 99 per cent respectively; France and Gobel 1985; Medlycott 2011b). Plant remains Almost 40 per cent of all records from the East region included some form of report on the plant remains (predominantly charred but occasionally waterlogged material), relating to 128 different settlements. Although there is still a bias towards the south and west, the records are more widely dispersed than those for the animal remains, with a small number of samples from North Norfolk and the Suffolk Coast and Heaths. The general range and proportion of crops present on sites in the East are quite consistent

Romanch6.indd 239

with those for the central and southern regions, being dominated by spelt wheat in all landscape zones. Spelt appears to have been grown and processed with some intensity over the course of the Roman period, especially to the south and west of the region where often abundant chaff and occasional whole spikelets (e.g. Tunbridge Lane, Bottisham; Kenny 2012) were recorded in most of the corndryers where environmental sampling had taken place. Both emmer wheat and free-threshing wheat were noted in c. 25 per cent of plant assemblages, almost always in lesser quantities than spelt, and free-threshing wheat in particular was very rare in Breckland and North Norfolk. After spelt, barley is the next most commonly encountered cereal crop, and is equally well represented in North Norfolk, occasionally being the dominant crop on site, as at Massingham Road, Rougham, where a sample from an early/ mid-Roman ditch contained barley grains and rachis nodes along with one possible wheat grain (Wilson et al. 2012). Barley was also the predominant cereal type identified at Norwich Road, Caister-on-Sea, on the north-east Norfolk coastline, c. 200 m south-east of the coastal fort, its prevalence there argued as being due to its greater tolerance of salt (Albone 2006). Oats are also just as commonly found in much of Norfolk, though less so in other areas, and in most cases there is the usual uncertainty as to whether it is the wild or cultivated form. Charred plant remains from a second–third century a.d. corndryer at Foxley Road, Norfolk, contained mostly oat grains, though these may have represented fuel for the oven (Wilson et al. 2012). A range of other types of crops was encountered sporadically, each recorded in less than 15 per cent of East region assemblages, though with some intra-regional differences. Rye, for example, was found in approximately a third of assemblages from Breckland, possibly part of a wider concentration as it was also well noted in the

06/09/2016 17:06:33

240

THE RURAL SETTLEMENT OF ROMAN BRITAIN

80

farmstead (n=107)

roadside settlement (n=16)

70

% of sites

60 50 40 30 20 10 0

fig. 6.35.  Percentage presence of cultivated plant taxa in farmsteads and roadside settlements within the East region (includes charred and waterlogged remains)

adjacent Fenland zone (see Ch. 5). Fruits and other horticultural crops are rarely encountered in assemblages, largely owing to the relative lack of waterlogged and/or mineralised remains. However, a range, including plum, apple/pear, blackberry, grape, elder berry, sloe, bullace and wild celery, was found from deposits in the roadside settlements at Baldock and Scole, while waterlogged remains from a well at a farmstead at Hinxton in Cambridgeshire included plum, sloe and cherry stones. Evidence for grape was found at six sites, while vine pollen from Scole indicates the presence of vineyards somewhere near to the settlement (Wiltshire 2014, 416). The rich array of fruit remains from the roadside settlement at Scole correlates with evidence from other similar sites, such as Baldock, suggesting a slightly more varied diet than that typically found at farmsteads, although the range and prevalence of most arable crops remains quite similar across both settlement types within their landscape zones (fig. 6.35). One of the few possibly notable differences concerns the greater percentage of oats present in assemblages from roadside settlements, this perhaps serving as animal fodder for horses in particular. Some indication of the mechanisms of food production and distribution is provided by environmental samples taken from excavations at the roadside settlement of Wixoe in Suffolk, where it was suggested that part-processed crops were imported from the hinterland, with the final stage of processing occurring in the town (Atkins 2013). As yet there is little evidence for granaries within such nucleated settlements, with the possible exception of an aisled building from Braintree, though as mentioned above, millstone fragments did occur in some roadside settlements, and quernstones were generally quite plentiful, including over 60 fragments from Icklingham, West Stow (West 1989). This all suggests that food

Romanch6.indd 240

processing, whether mechanised or not, was an important and widely practised activity at these larger population centres. There is little in the plant assemblages to suggest any major differences between the two most common farmstead types, except that material from complex farmsteads appears to have been somewhat more dominated by spelt at the expense of other wheat varieties. These sites are also relatively better represented with rye, flax and fruits, hinting at greater diversification, perhaps as part of crop rotation. Unfortunately there are just five assemblages from villas in the east, and most of these had a very low number of samples. An exception was at Great Holts Farm in Essex, where 213 bulk samples were collected, most with low densities of plant remains, but some within the aisled granary suggesting a highly organised agricultural regime, with wheat, barley and pulses all shown to be grown and stored as separate crops (Germany 2003). REGION SUMMARY The East may be one of the smallest regions of the current project, and be lacking in excavated data for large parts, but there is enough to recognise distinct variations in the character of Roman rural settlement within different landscape zones. In all areas there is evidence for major continuity of settlement from the late Iron Age into the early Roman period, though with evidence for disruption and transformation in some parts during the later first century a.d. The excavated sites in the south of the region reached their peak in numbers at this time, though further north this peak is not reached until the mid-Roman period, with a fairly pronounced decline thereafter, especially in North Norfolk.

06/09/2016 17:06:33

THE EAST

As elsewhere it is a region dominated by small farming settlements, which appear to undergo a particularly steep decline in numbers during the later Roman period, probably signifying a fundamental shift in the way that the rural landscape was organised and farmed. This was a period in which certain farmsteads were increasing in size and complexity (developing into complex farmsteads), especially in the south and west, perhaps becoming centres of increased agricultural production, associated with extensive arable fields of primarily spelt wheat. The growth of villa architecture at this time was undoubtedly associated with such changes, though most buildings remained relatively modest compared to those further south and west, and there is little evidence that they were ever a regular part of the landscape in the north-east of the region. The relative lack of complex farmsteads and villas (notwithstanding the undated evidence from cropmarks) in this area, together with differences noted in field systems, arable crops and certain objects associated with food processing, hints at quite measurable differences in the farming landscapes between the north and further south and west, which in turn may reflect distinctive cultural landscapes. The differences may also imply varying political realities, with the northern parts of the region possibly having greater levels of state control associated with supply of the military to the north, where East Anglian pottery demonstrates some trade connection (see vol. 2). Such a scenario would fit in well with the context of military sites being established at Brancaster and Caister-on-Sea from the early third century a.d., as well as the concentration of roadside settlements, which may have acted as collection and distribution points.

Romanch6.indd 241

241

The key to any understanding of Roman settlement patterns in the East region is the network of these nucleated roadside settlements, which is among the densest anywhere in Britain, though similar to adjacent parts of the Central Belt. Many of these originated with some pre-conquest activity and continued until at least the end of Roman period, although there was a noticeable decline in fortunes from at least the mid-fourth century a.d. onwards. It is clear that these nucleated sites fulfilled a range of roles, including industrial production, commercial opportunities and provision of religious foci, though just how much of a local administrative function they had is somewhat more uncertain. There is little evidence for highstatus dwellings, though relatively high-status finds (hairpins, finger rings, metal and glass vessels, etc.) are far more abundant, hinting at differences in the way that wealth and social status were expressed compared to regions further south and west. This well-developed network of roadside settlements may be a significant factor in the relatively high prevalence of material culture in farmsteads in the East compared to any other region. Does the high proportion (in terms of occurrence if not quantity) of farmsteads with coins, brooches and household objects, for instance, indicate an increased integration with the growing market economy centred on the nucleated roadside sites? Or are they part of a regional manifestation of how social relations and identity were formed and maintained by local farmers, as Taylor (2013, 186) has demonstrated for other parts of the Roman province? These themes will be explored in subsequent volumes, and the propositions are, of course, not mutually exclusive; both indicate the importance of material culture for our understanding of rural communities in this region.

06/09/2016 17:06:34

Chapter 7

THE NORTH-EAST By Martyn Allen

The North-East region covers c. 14,282 km², stretching from Tyne and Wear in the north to Lincolnshire and Nottinghamshire in the south. In addition to these counties, the region includes all or parts of Derbyshire, East Riding, South Yorkshire, West Yorkshire, North Yorkshire, and County Durham (fig. 7.1). The nature of the landscape The North-East region is made up of fourteen landscape zones based upon Natural England character areas (fig. 7.2). The geography of the region is heavily influenced by the Humber estuary, which flows into the North Sea. Immediately to the north of the estuary lies Holderness, a coastal marshland that is mirrored

fig.

to the south by the Lincolnshire Marshes. The estuary is fed by five major rivers – the Hull, Trent, Don, Aire, and the Ouse – which drain the Vales of York and Mowbray, and the Humberhead Levels. Upland regions include the North York Moors and the chalk hills of the Yorkshire Wolds, which are separated by the Vale of Pickering. The North-East region is divided from the Pennines by the Magnesian Limestone Belt, which is generally covered by fertile and free-draining calcareous soils, while a second limestone plateau is located at the northern point of the region and is separated from the North York Moors by the Tees Lowlands. To the west of the Magnesian Limestone, the Coal Measures of South Yorkshire lie in the Central West project region (see Ch. 8). Although outside the North-East, this landscape zone

7.1.  The North-East region in relation to modern county boundaries

242

Romanch7.indd 242

15/09/2016 17:28:20

THE NORTH-EAST

fig.

7.2.   Constituent landscape zones of the North-East region

contains a distinctive settlement pattern of enclosed farmsteads and field systems which is similar to that found on the limestone, as well as the forts and vicus at Castleford, on the same road as Doncaster and Tadcaster (Bidwell and Hodgson 2009, 133–6). This point serves as a reminder that the boundaries of the project regions should not be seen as clear divisions, but simply as a means for organising the data. The North-East dataset The dataset for the North-East contains 363 site excavation records, which account for 258 settlements as well as ‘isolated’ field systems, religious sites, burial sites and industrial sites. The site records are distributed unevenly across the region, with the greater concentrations tending to be located in the low-lying vales and on the Magnesian Limestone. The density of site records for each landscape zone is presented in table 7.1, which confirms the high frequency of sites located on the Magnesian Limestone Belt, while other concentrations occur on Holderness, in the Vale of Pickering, and on the North Lincolnshire Coversands and Clays (cf. fig. 7.3, top).

Romanch7.indd 243

243

The high proportion of excavated sites identified on the Magnesian Limestone is mirrored by a high volume of excavations recorded in that area on the National Monuments Records (NMR) database; the effect of developer-funded excavations on the North-East region is readily apparent (fig. 7.3b). Limestone quarrying on the Magnesian Belt and gravel quarrying on the Humberhead Levels has been responsible for the identification of a number of sites in the east and south of the region. Excavations resulting from pipeline projects have been even more productive, contributing 62 sites (17 per cent) to the regional dataset, many of which are located on the northern side of the Humber Estuary. The vast majority of sites recorded from Holderness, East Riding, for example, were identified during the excavation of the Easington to Ganstead natural gas pipeline (Flintoft and Glover 2009). In addition to excavated evidence, aerial photography and geophysical survey has been fundamental in revealing a range of settlements and field systems, such as the extensive, linear trackway and enclosure system in the Vale of Pickering, where small-scale excavations have demonstrated a late Iron Age/Romano-British

15/09/2016 17:28:30

244

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 7.3.  Kernel density of North-East region records (n=363) and all excavation records (1910– 2010) from National Monument Records (NMR) Index (n=3434) (excluding data from York)

Romanch7.indd 244

15/09/2016 17:28:45

THE NORTH-EAST

table

7.1: number

of sites and density (per km²) by landscape zone in the north-east region

Landscape zone

Area (km²)

No. of records 7

Density of records per km²

Durham Magnesian Limestone Plateau

447.00

Holderness

869.32 32 0.0368

Humber Estuary

285.68

9

0.0315

1723.53

48

0.0278

859.60

18

0.0209

Humberhead Levels Lincolnshire Coast and Marshes Lincolnshire Wolds

0.0157

844.86

17

0.0201

North Lincolnshire Coversands and Clay Vales

1318.61

40

0.0303

North York Moors and Hills

1864.34

18

0.0097

Sherwood

534.57 7 0.0131

Southern Magnesian Limestone

1367.62

59

0.0431

Tees Lowlands

1004.99

24

0.0239

425.50

13

0.0306

Vale of York and Mowbray

1627.16

29

0.0178

Yorkshire Wolds

1108.85

20

0.0180

Vale of Pickering

date (Powlesland et al. 2006). A number of studies of the cropmark evidence from the region have also been undertaken, covering the Sandstone of South Yorkshire and North Nottinghamshire (Riley 1980), the Magnesian Limestone Belt (Roberts 2010) and the Yorkshire Wolds (Stoertz 1997). Each of these surveys has confirmed considerable evidence for late Iron Age and Roman-period enclosure, trackways, and field systems, much of which is not represented in the project database, but must be considered alongside it. The distribution of known cropmark sites of possible Roman date has also been mapped by Taylor, who highlighted additional concentrations of enclosed settlement evidence in the Vale of Pickering and on Holderness (Taylor 2007, 24–5, 44). In contrast to areas where evidence for Romano-British rural settlement is more abundant, the North York Moors have received comparatively little attention in terms of survey or excavation. Here, excavated sites are restricted to the periphery of this landscape and only become more numerous again on the Tees Lowland to the north. Roman rural settlement patterns The North-East region contains a number of major urban centres, including the civitas capital of Isurium Brigantum at Aldborough, the possible civitas capital of Petuaria at Brough (cf. Wacher 1969; 1995, 394–8; Hunter-Mann et al. 2000), and the coloniae of Eburacum at York and Lindum at Lincoln. The origins of Roman York and Lincoln lay with the establishment of the legionary

Romanch7.indd 245

245

fortresses in the later first century a.d., and a number of other Roman forts and accompanying vici are also known at varying points along the road system, with notable examples located at Malton and Doncaster. These two sites may also be considered alongside the other small towns found in the region, though it is uncertain whether they ever became fully established as walled towns (Buckland and Magilton 1986; Wenham and Heywood 1997). Other walled ‘small towns’ are located at Horncastle, Caistor, Tadcaster (Calcaria), and Catterick (Cataractonium), and it is important to also consider the role of the military in the development of a number of these sites, including Catterick and, potentially, Horncastle (Bidwell and Hodgson 2009, 141–5, 162–4). Of the 258 settlements recorded for the NorthEast region, 213 are farmsteads. However, the percentage of farmsteads compared to other settlement types varies between different landscape zones (table 7.2; fig. 7.4). For example, in Holderness and on the Lincolnshire Coast and Marshes all the sites recorded are farmsteads, while in the Vale of York and Mowbray and on the Yorkshire Wolds, farmsteads represent only 64 per cent and 47 per cent of sites respectively. Settlements with villa architecture are rare, with only twenty sites recorded for the region. However, many more potential villa sites are recorded on the National Monuments Record (NMR), with a distinct concentration in the Lincolnshire Coversands and Clays between Lincoln and the Humber Estuary (see Villas below). Roadside settlements are represented by thirteen sites and military vici are represented by five sites. The latter only include sites with direct evidence for the

15/09/2016 17:28:53

246

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 7.4.   Excavated late Iron Age/Roman rural settlement in the North-East region in relation to Roman roads and urban centres

table

7.2: number

Landscape zone

of all settlement types by landscape zone in the north-east region

Farmstead Farmstead Farmstead Farmstead Villa Roadside Village Vicus Oppidum Total % (all) (complex) (enclosed) (open) settlement

Durham Magnesian Limestone Plateau Holderness Humber Estuary

6

3

1

6

2.3

30 1 3 3 30 11.6 5

2 4

1

2

2.3

Humberhead Levels

26

6

Lincolnshire Coast and Marshes

11

1

11

4.3

Lincolnshire Wolds

10

1

13

5.0

2

2

6

3

30 11.6

16 3 3 1 1 21 8.1 North Lincolnshire Coversands and Clay Vales North York Moors and Hills Sherwood

15

Vale of Pickering

2

1

16

6.2

5

17

1

3

1

43 16.7

17 4 1 2 3 2 1 23 8.9 8

2

1

1

1

1

11

4.3

Vale of York and Mowbray 16

4

2

1

6

1

1

25

9.7

8

4

2

6

1

2

17

6.6

Yorkshire Wolds Total

Romanch7.indd 246

3

6 4 6 2.3

Magnesian Limestone Belt 39 Tees Lowlands

3

1

213 33 45 10 20 14 6 4 1 258 100.0

15/09/2016 17:29:01

THE NORTH-EAST

no. of selements in use

associated development of a civilian settlement alongside a fort, although this is not always straightforward to determine. For example, the role of the military in the early development of Piercebridge is poorly understood (Cool and Mason 2008). Here, a large, stone-built fort was certainly constructed by the middle of the third century a.d., but appears to have been off-set from the main road and civil settlement. Millett stated that the secondary position of the fort indicated that Piercebridge was a small town rather than a vicus, implying that the settlement developed independently (Millett 1990, 147). However, several authors have since argued, based upon comparisons with other sites in the north, that a first-century a.d. fort must have existed lying somewhere on or close to the River Tees (Bidwell and Hodgson 2009, 147; Ottaway 2013, 106).

fig.

Regional chronology Dating evidence was available from 252 sites, mostly through pottery and coinage, and an increasingly important corpus of radiocarbon dates (see Ch. 1). The chronological data show an increase in settlement numbers from the late Iron Age to a peak in the second half of the second century a.d. (fig. 7.5). Thereafter, a decline can be observed through to the later fourth century a.d., when settlements numbered around onethird of that seen in the late Iron Age. However, there are clear intra-regional differences within this pattern (fig. 7.6). The settlement chronology in Holderness, for example, shows a distinct decline in settlement numbers from a peak in the late Iron Age. In contrast, settlement numbers peak in the second century a.d. on the Southern Magnesian Limestone and on the North Lincolnshire Coversands and Clays, before

200 180 160 140 120 100 80 60 40 20 0

7.5.   Number of settlements in use over time in the North-East region

North Lincolnshire Coversands and Clays (n=20)

40 35 30 25 20 15 10 5 0

no. of sites

no. of sites

30 25 20 15 10 5 0

no. of sites

Holderness (n=30) 16 14 12 10 8 6 4 2 0

Southern Magnesian Limestone (n=42)

fig.

Romanch7.indd 247

247

Vale of York and Mowbray (n=27) no. of sites

25 20 15 10 5 0

7.6.  Variations in settlement chronology by selected landscape zone within the North-East region

15/09/2016 17:29:05

248

THE RURAL SETTLEMENT OF ROMAN BRITAIN

100 90 80

sites established

% of sites

70 60 50

sites abandoned

40 30 20 10 0 Late Iron Age Late 1stC AD (n=142) (n=156)

fig.

2ndC AD (n=187)

3rdC AD (n=167)

4thC AD (n=127)

7.7.   ‘New’ and ‘abandoned’ settlements over time in the North-East region

b

a

Phase 4

Phase 3

Roman road N

0

200 m 1:5000

fig. 7.8.   Site plans of Roman Ridge, West Yorkshire (Roberts et al. 2001), showing (a) the late Iron Age/early first century field system, trackway and enclosures, and (b) the line of the Castleford to Tadcaster Roman road

declining into the fourth century a.d., while in the Vale of York and Mowbray, settlement numbers also increase into the second century a.d., but show little evidence for decline thereafter, possibly because of the influence of major urban centres at York and Aldborough (see Ch. 12). The intra-regional differences in chronology may have been partly influenced by land in the south coming under Roman control in the later part of the first century a.d. With Brigantia continuing to hold political power, much of the northern part of the North-East can be equated as being culturally ‘Iron Age’ well into this period. Around one-third of the settlements occupied in the late Iron Age continued in use from the middle Iron Age, suggesting that this was a period of both continuity and expansion (fig. 7.7). In the late first century a.d., only 18 per cent of the 156 sites in use were established ‘new’ at that time, considerably less than the 60% in the East region, for example (see Ch. 6), where there was much greater evidence for settlement expansion in this period. Nevertheless, the construction of new roads and military establishments, particularly along the

Romanch7.indd 248

western and eastern fringes of the Humberhead Levels and the Vale of York and Mowbray, appear to have caused some disruption to settlement (Roberts 2010, 71–2). At Roman Ridge, West Yorkshire, in the first century a.d., a field-system with a trackway and two enclosures was transformed by the construction of a Roman road running north from Castleford to Tadcaster (Roberts et al. 2001; fig. 7.8). The deposition of second to fourth century pottery at the site suggests that continued or renewed domestic activity occurred nearby (ibid.). At 10–14A Hall Gate, Doncaster, evidence for a wattle fence, a gully and a ditch, were later sealed by a road that was constructed around a.d. 70 when the Roman fort of Danum was built (Archaeological Services WYAS 2008). Similar evidence has also been identified at Glen Garth, Hayton, where the construction of a road at the end of the first century or the beginning of the second century led to the reorganisation of a small settlement, which became realigned towards the highway (Halkon et al. 2015). However, while the new road network affected some settlements, cropmark evidence suggests that its impact did

15/09/2016 17:29:07

THE NORTH-EAST

249

N

0

100 m 1:2000

fig. 7.9.   Plan of the twin enclosure complex at Allerton Park Quarry, North Yorkshire (Ross 2009), established in the mid-third century a.d.

not result in any substantial reorientation of field systems (Roberts pers. comm.). Some of the changes that took place in the later first century a.d. may have been partly responsible for the increase in newly established settlements in the second century a.d., and it is possible that these changes brought about new opportunities for trade and exchange. At Gibraltar Farm, KingstonUpon-Hull, a settlement established on the northern bank of the Humber in the mid-second century a.d. soon began to receive continental goods, including samian and Moselkeramik, perhaps suggesting that the inhabitants were exploiting river-borne trade links (Tibbles and Steedman 1997; Van de Noort and Ellis 2000). Fewer new settlements originated after the second century a.d., though the establishment of enclosure complexes at Allerton Park Quarry (Ross 2009; fig. 7.9) and Swaythorpe Farm, Kilham (Mackey 1998), appear to have been exceptions. In the third and fourth centuries an increasing number of settlements were being abandoned, while at Hensall Quarry (Weston 2013) and Crossgates, Seamer (MAP Archaeological Consultants 2001), both the settlements and their associated field systems went out of use, suggesting that wholesale changes in land use were occurring in some areas. It is possible that soil exhaustion was a factor in the decline of some settlements, with Hensall Quarry and Crossgates both being located on sandy, slightly acidic, soils, which may not have been suitable for the cultivation of some crops. In total, 106 sites were occupied in the second half of the fourth century a.d., though little can be said about settlement activity in the early fifth century a.d. Some excavators have suggested that

Romanch7.indd 249

the occurrence of very late fourth-century pottery vessels, such as Crambeck wares and Huntclifftype jars, may indicate continuity into the fifth century. However, this is very difficult to prove owing to the kiln sites being abandoned before the end of the fourth century (Corder 1989a; 1989b). Ottaway points out that the Crambeck industry probably ceased once the army was no longer being paid, while the reappearance of regional hand-made wares at this time represents a return to a ceramic style not seen since the Iron Age (Ottaway 2013, 319–20). Unfortunately, these vessel types are difficult to date with any degree of accuracy. Late Roman coinage dating to the end of the fourth century a.d. (Reece period 21: a.d. 388–402) has been found on twelve sites, mostly villas and roadside settlements, with large numbers being recovered from Beadlam villa (Neal 1996), Langton villa (Corder and Kirk 1932), Shiptonthorpe (Millett 2006), and Hayton (Halkon et al. 2015) – the latter two benefitting from extensive metal-detecting surveys. Of course, the recovery of Reece period 21 coinage does not demonstrate evidence for fifth-century activity, and none of these sites otherwise provided evidence that they continued beyond the fourth century. Even at sites where clear evidence for fifthcentury activity has been demonstrated, there is uncertainty whether it represents continuous occupation from the late Roman period. At Malton, the fort was abandoned in the fourth century, though some of the masonry buildings to the south appear to have been re-used in the fifth and sixth centuries a.d. (Wenham and Heywood 1997). At Heslerton, the late Roman ladder settlement appears to have been abandoned prior to the establishment of an Anglian settlement and

15/09/2016 17:29:08

250

THE RURAL SETTLEMENT OF ROMAN BRITAIN

cemetery (Powlesland et al. 2006). However, a nearby late Roman shrine continued to be used through to the seventh century, providing a focus for the early development of the Anglian settlement (Powlesland 1998). Some sites were clearly abandoned by the end of the fourth century, but show some evidence that activity recurred in the early medieval period, such as at Wattle Syke (Martin et al. 2013), Sewerby Cottage Farm, Bridlington (Fenton-Thomas 2009), and Melton A63 (Fenton-Thomas 2011). While conclusive evidence for domestic occupation on late Roman sites in the fifth century is generally lacking, lone burials dating to the early medieval period do appear to have been a feature at some sites, such as at Dalton Parlours (cf. I. Roberts 2013, 300). Farmsteads: morphology, chronology and distribution The distribution of all 213 farmsteads in the North-East region is presented in fig. 7.10, and their combined chronological data are shown in fig. 7.11. There is very little difference in the total numbers of farmsteads in use between the late Iron Age and the second half of the second

fig.

Romanch7.indd 250

century a.d., after which they decline through to the end of the fourth century. In total, 88 farmsteads have been classified in terms of their morphology. Enclosed farmsteads are most common, found relatively frequently on the Magnesian Limestone Belt compared to the low-lying Vale of York and Mowbray and the Humberhead Levels, where only a few excavated examples exist. Naburn in the Vale of York stood as one such rare example (Jones 1988; Roberts pers. comm.), but, more recently, the Asselby to Pannal pipeline has revealed evidence for settlement enclosure and field systems that bear some resemblance to those found on the Magnesian Limestone, the Coal Measures, and the Sherwood Sandstones to the south (Gregory et al. 2013). A number of sites excavated in these landscape zones can be identified as complex farmsteads, such as at Topham Farm, Sykehouse (Roberts 2003). Certainly, complex farmsteads are fairly widespread across the North-East in general, and it appears that the lack of enclosed and complex settlements in the Vale of York and Mowbray may have been due to poor visibility, an issue that is now being partly rectified through the identification

7.10.   Distribution of all farmsteads in the North-East

15/09/2016 17:29:16

THE NORTH-EAST

of cropmark sites (Ottaway 2013, 62–3). The few open farmsteads in the region are mostly located in the northernmost part, with other sites located on Holderness, in the Vale of Pickering, and on the Magnesian Limestone Belt. As with most other regions, the chronological data from classified farmsteads show an increase in the frequency of complex farmsteads over

251

time, and a reduction in enclosed and open farmsteads (fig. 7.12). Open farmsteads are predominantly late Iron Age or early Roman in date, and very few are present in the second century a.d. or later. Although some appear to have been abandoned in the early Roman period, such as Thorpe Thewles in the Tees Lowlands (Heslop 1987), others developed into enclosed or

no. of farmsteads

160 140 120 100 80 60 40 20 0

fig.

7.11.   Numbers of farmsteads in use in the North-East over time

80

open farmstead (n=16)

enclosed farmstead (n=50)

complex farmstead (n=33)

% of classified farmsteads in use

70 60 50 40 30 20 10 0

fig.

7.12.   Relative frequency of farmstead types over time in the North-East First century A.D.

Second century A.D.

Third century A.D.

(a)

Third century A.D.

(b)

N

0

200 m 1:5000

fig.

Romanch7.indd 251

7.13.   Phase plans of High Wold, Bempton Lane, Bridlington (Roberts 2009)

15/09/2016 17:29:18

252

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

b

N

0

100 m 1:3000

fig. 7.14.   Plans of two types of complex farmstead in the North-East at (a) Faverdale, Darlington (Proctor 2012) and (b) Newbridge Quarry, Pickering (Richardson 2012)

complex farmsteads, such as at High Wold, Bridlington (Roberts 2009; fig. 7.13), and Faverdale, Darlington (Proctor 2012). At Faverdale, an extensive open settlement consisting of several roundhouses with hearths and stock enclosures, developed into a complex farmstead in the second century a.d. (fig. 7.14a). Defined by a network of conjoined rectilinear enclosures and trackways, with a principal enclosure located on a high spur of land, this type of settlement is a form more typically found in the Central Belt during this period (see Ch. 5). In addition, the construction of a small, two-roomed stone building with a hypocaust and painted-plaster walls, probably a small bathhouse, also suggests some degree of affluence at the site.

Romanch7.indd 252

N

Trackway

Roundhouse

0

100 m 1:3000

fig. 7.15.  Plan of Holmfield Interchange, Site Q, showing late Iron Age/early Roman D-shaped enclosure and the late Roman complex farmstead (Brown et al. 2007)

15/09/2016 17:29:20

THE NORTH-EAST

a

253

Enclosed farmstead

0

100 m

N

1:2000 b

Enclosed farmstead

fig. 7.16.  Plans of two late Iron Age/early Roman enclosed farmsteads integrated with a rectilinear field system at Heslington East, (a) Area A1 and (b) Area A2 (Antoni et al. 2009)

Romanch7.indd 253

15/09/2016 17:29:20

254

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

b Potential earthworks Former ditch line

0

200 m 1:5000

fig. 7.17.  Site plans of Wattle Syke, West Yorkshire, showing (a) the late Iron Age–middle Roman enclosure complex (village), and (b) the late Roman (fourth century a.d.) open settlement (Martin et al. 2013)

Second century A.D.

Fourth century A.D.

N

Third century A.D.

Fifth century A.D.

0

200 m 1:5000

fig.

7.18.   Phase plans of the settlement at Parlington Hollings, West Yorkshire (Roberts et al. 2001)

Alongside those complex farmsteads consisting of a major, sub-divided enclosure, settlements that were characterised by systems of enclosures lining a trackway also became more common in the Roman period. These types of site are commonly referred to as ‘ladder settlements’ owing to their

Romanch7.indd 254

linear morphology (e.g. Stoertz 1997). However, they varied considerably in size and in terms of the density of settlement features and material culture they included. The larger examples recorded on the database have been classified as villages, but smaller settlements are better understood as

15/09/2016 17:29:21

THE NORTH-EAST

complex farmsteads, as at Newbridge Quarry, Pickering (Richardson 2012; fig. 7.14b). The 45 enclosed farmsteads in the North-East vary considerably in size and shape, though the majority (56 per cent) are rectilinear in form. They ranged from small, single-ditched enclosures measuring around 0.1 ha, up to the doubleditched, 3.5 ha curvilinear enclosure at Tattershall Thorpe (Seager Smith 1998). While most enclosed farmsteads were fairly simple settlements, some are found to have developed over time. For example, the small D-shaped enclosure at Holmfield Interchange, Site Q, was incorporated into a larger complex farmstead in the late second century a.d. (Brown et al. 2007; fig. 7.15). Cropmarks showed that the farmstead was associated with other settlements in the area, via an adjoining trackway and fields. The integration of enclosed farmsteads within field systems appears to have been relatively common, as at Heslington East where two small, enclosed farmsteads of late Iron Age/early Roman date were found to utilise field ditches as part of the settlement’s boundaries (Antoni et al. 2009; fig. 7.16). Further cropmark evidence suggests that some of these systems extended over several

fig.

Romanch7.indd 255

255

kilometres, for example at Tickhill and Went Hill, South Yorkshire (Roberts 2010, 32). In the late Roman period, some farmsteads show evidence for substantial reductions in size and complexity. At Wattle Syke, a modestly sized, open farmstead with cellared buildings was established in the late third century a.d. where a much larger, complex farmstead with multiple enclosures had been present in the later first and second centuries, possibly after a short hiatus in activity (Martin et al. 2013; fig. 7.17). At the nearby site at Parlington Hollings, a complex farmstead and field system was reduced in size in the fourth century, with only a single enclosure and a boundary ditch being maintained (Roberts et al. 2001; fig. 7.18). While many settlements show evidence for continued activity into the late Roman period, changes, such as those witnessed at Wattle Syke and Parlington Hollings show that this was not uniform. Villas in the North-East Twenty villa sites are recorded in the North-East dataset and these are widely distributed across the region (fig. 7.19). This pattern is somewhat

7.19.  Distribution of excavated villas in the North-East compared to ‘villas’ recorded in the NMR index

15/09/2016 17:29:30

256

THE RURAL SETTLEMENT OF ROMAN BRITAIN

500

year BC/AD

400 300 200 100

a

0 -100

b ological ed villas in the iest date of nd (B) number me

no. of villas in use

date of villa construc on

16 14 12 10 8 6 4 2 0

fig. 7.20.   Chronological patterns of excavated villas in the North-East: (a) earliest date of villa construction and (b) number of villas in use over time

different from the distribution of villas recorded on the NMR, which shows a dense concentration of sites on the Lincolnshire Coversands and Clays, representing the northernmost tip of the so-called ‘villa zone’ of central England (Mattingly 2006, 480, fig. 17). The disparity between the two datasets highlights the lack of villas that have been comprehensively excavated and published. Of those that are recorded on the database, the quality of information available is highly varied. Some sites can only be regarded as ‘possible villas’, such as Wharram Le Street and Wharram Grange, which were primarily investigated using magnetometry and fieldwalking (Rahtz et al. 1986). Other sites have suffered from a protracted post-excavation process, such as Beadlam villa, where paper archives and material assemblages became widely dispersed, with some since being lost (Neal 1996). However, the situation is beginning to improve with more recent excavations, as with the publication of the villa at Ingleby Barwick (Willis and Carne 2013). Chronological data from villa sites shows that around half were constructed in the second century a.d., with the other half in the third century, and most developed from an earlier farmstead (fig. 7.20). Although it is not always easy to demonstrate continuity of occupation, seven villas show evidence for pre-villa occupation stretching back to the late Iron Age. This chronological pattern is somewhat different from that of farmsteads, which reduce in number into the fourth century, in part

Romanch7.indd 256

reflecting the development of some into villas. However, the small number of excavated villas means that their social and economic importance is difficult to assess. Villas in the North-East appear to have been quite varied in their size and complexity, ranging from the large courtyard building at Scampton, containing at least 40 rooms (Illingworth 1808), to the small, ‘cottage-style’ villa at Rudston (Stead 1980). Unfortunately, there is very limited evidence for the immediate context around villa buildings. An exception can be found at Welton villa, where excavations in advance of gravel quarrying revealed an extensive landscape of enclosures, trackways, aisled barns, corndryers, and a possible shrine/mausoleum (Mackey 1999; fig. 7.21). Similar evidence for an extensive settlement complex has been revealed around the Roman villa at Ingleby Barwick, where a network of rectilinear enclosures and trackways was laid out on a sand and gravel terrace just south of the River Tees (Willis and Carne 2013). Despite these useful examples, most villas provide very little evidence for how the settlement functioned, which requires greater emphasis on the recovery of finds and environmental assemblages. Nucleated settlements Roadside settlements The thirteen roadside settlements recorded in the database are unevenly distributed along the major

15/09/2016 17:29:32

THE NORTH-EAST

257

a Corndryers

Site of shrine/mausoleum?

Corndryer

Timber buildings

Corndryer Corndryers Corridor house N

Aisled barn Corndryer

0

200 m 1:5000

b

N

Winged-corridor villa

fig.

7.21.   Plans of the villas at (a) Welton Wold (Mackey 1999) and (b) Ingleby Barwick (Willis and Carne 2013)

road network, with a particular concentration found between Brough and York, though it is likely that other sites are yet to be identified (fig. 7.22). Despite their comparative rarity in the NorthEast, roadside settlements probably performed a range of economic roles, though our understanding of their function is very limited. The majority are recognised by contiguous plots of land along the roadside, which are divided in places by trackways running off the main road (Halkon et al. 2015; fig. 7.23). Most provide evidence for buildings, pits and wells, and some, such as Bainesse, North Yorkshire, have field systems backing onto ‘property plots’ (Wilson 2002). In most cases, roadside settlements in the North-East do not appear to have been enclosed, though it is rare for their full size to be determined, particularly without extensive geophysics or cropmark

Romanch7.indd 257

evidence, such as at Shiptonthorpe where the settlement covered around 10 ha (Millett 2006). It is also uncertain whether these settlements had a central focus, which, presumably, would have been important if markets or local administration were a feature. Geophysical survey and small-scale excavation at East Park, Sedgefield, identified a single rectangular timber building sited in an otherwise open space in the centre of the settlement, which was suggested as having a ‘public’ function, perhaps as a shrine or a marketplace (Hale 2010). Chronological data show that some roadside settlements developed from late Iron Age foci (fig. 7.24), such as the large, double-ditched enclosure identified at Hayton (Halkon et al. 2015), or the ritual complex at Nettleton and Rothwell (Willis 2013). At both these sites, the late Iron Age

15/09/2016 17:29:33

258

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

7.22.   Distribution of nucleated settlement and military sites in the North-East

N

Ro

m

an

ro

ad

Trackways Possible stone structures 0

200 m 1:5000

fig.

Romanch7.indd 258

7.23.   Interpretative plan of the geophysical survey results at Hayton (Halkon et al. 2015)

15/09/2016 17:29:44

THE NORTH-EAST

259

25

no. of sites in use

20 15 10 5 0

fig.

7.24   Chronological development of nucleated settlement in the North-East

elements were incorporated into the Roman roadside settlements, demonstrating their continued significance. A possible late Iron Age shrine could have provided a ritual focus at Rudstone Dale, Newbald, where a timber structure was found to contain ten neonate burials and two animal burials, a calf and a lamb, placed around a central hearth (Wood 2011). Other sites appear to have been founded upon ‘virgin’ ground, as at Winteringham on the south side of the Humber, where early activity is suggested to have been stimulated by the military (Stead 1976). The position of forts nearby at Hayton and Kirmington may also have helped the civilian settlements to flourish in the early Roman period (Halkon et al. 2015; Jones and Whitwell 1991). Of course, sites such as these blur the boundary between roadside settlements and military vici, where civilian settlements are located very close to the fort itself (see below). The decline and abandonment of roadside settlements in the North-East is also poorly understood. Most continued to be occupied into the fourth century a.d., and the lack of fourth-century material at Dringhouses, near York (Ottaway 2011), East Park, Sedgefield (Hale 2010) and Bishop Grosseteste College, Newport (Wragg 1995), is probably due more to the restricted scale of excavations than a genuine lack of activity. At Shiptonthorpe, gravel spreads dating to the late fourth century possibly represent the latest buildings in use at the settlement. Here, Millett suggests that a reordering of space along the roadside points to a reduction in the local population, while the road may have continued to be used into the early Saxon period, influencing the development of the early medieval village at Shiptonthorpe (Millett 2006, 307–8). Vicus settlements As with roadside settlements, excavated evidence from military vici is equally patchy. Only five are included in the North-East dataset, although other known examples such as Castleford lie just

Romanch7.indd 259

outside the region; the location of this site on the road between Doncaster and Newton Kyme shows that it would have been an important settlement in the south-west of the region. Excavations at Welbeck Street have revealed part of the vicus at Castleford, including some good examples of early Roman timber strip buildings that front the road leading south from the fort (Abramson et al. 1999). The vicus settlement at Newton Kyme is not included in the database because the very limited excavations carried out at the site have been restricted to the fort, and are unpublished (Bidwell and Hodgson 2009, 136–8). However, aerial photography has shown an extensive roadside development of enclosures, buildings and trackways, running for 600 m south of the fort (Boutwood 1996; fig. 7.25). Unfortunately, the lack of excavation means that the relationship between the fort and the civil settlement is not well understood. The plan of the settlement at Newton Kyme represents one of the few more extensive overviews of the form and size of a vicus in the North-East. Catterick (Cataractonium) is a comparatively well-understood site owing to extensive excavations over an area where there has been very little modern development (Wilson 2002). Catterick itself is not included in the database because it had clearly developed into a walled town at least by the fourth century a.d., the defences of which were integrated with those of the fort that sat adjacent to it (Bidwell and Hodgson 2009, 144, fig. 59). Two sites that are included in the database as military vici are Doncaster (Danum) and Malton (Derventio). These two settlements are far less well understood compared to Catterick, but each has produced indications that they too developed into ‘military towns’ in the later Roman period. Boundary ditches excavated at Doncaster were suggested to have marked the western extent of the settlement, and it was argued that the recovery of limestone and finely tooled sandstone from the base of two

15/09/2016 17:29:45

260

THE RURAL SETTLEMENT OF ROMAN BRITAIN

N

Roman fort

Possible Neolithic henge monument and/or Roman amphitheatre Vicus settlement Road

0

200 m 1:5000

fig.

7.25.   Plan of cropmarks showing the fort and vicus at Newton Kyme, North Yorkshire (Boutwood 1996)

of the ditches could have derived from a ‘defensive’ wall (Buckland and Magilton 1986, 42, 210–13; Bidwell and Hodgson 2009, 131–2, fig. 53). Further, slim, evidence for vicus enclosure has been noted at Malton, between the fort and the River Derwent. Here, several phases of building construction took place during the second and fourth centuries, which appear to coincide with renovations of the fort, signifying substantial investment in the settlement (Mitchelson 1964; Wenham and Heywood 1997).

Romanch7.indd 260

In contrast to most military vici in the North region, which were largely abandoned by the end of the third century a.d., particularly along Hadrian’s Wall, the vici in the North-East region mostly continued into the fourth century (Bidwell and Hodgson 2009, 33–4). This seems to reflect the transition of some into regional market centres, which certainly occurred at Catterick and possibly at Doncaster and Malton as well. Such continuity into the fourth century is consistent with the dating of the roadside settlements in this region.

15/09/2016 17:29:46

THE NORTH-EAST

The end of the fourth century sees the abandonment of most of the North-East military vici, though there is some evidence for continuity, or re-emergence of activity into the fifth century at Piercebridge (in the far north of the North-East region), where activity contracts to areas within and immediately around the fort (Cool and Mason 2008), and at Malton, where Roman buildings appear to have been re-used in the fifth and sixth centuries (Wenham and Heywood 1997). Other nucleated settlements A number of nucleated settlements that are not associated with the main Roman road system are also present in the North-East. The region contains an important late Iron Age oppidum at Stanwick, North Yorkshire, consisting of an earthwork enclosure covering nearly 350 ha. Dating the site to the middle of the first century a.d., Mortimer

261

Wheeler argued that it was constructed by the Brigantes tribe as a defensive response to the Roman Conquest (Wheeler 1954). However, excavations in the 1980s led to the suggestion that the elaborate entranceways and walled ramparts were more consistent with displays of status and prestige by a pro-Roman client kingdom (Haselgrove et al. 1990). Stanwick was abandoned soon after the conquest, as was the extensive late Iron Age settlement at Dragonby, Lincoln, where the recovery of Roman ballista bolts and spearheads were interpreted as evidence for an attack (May 1996). However, while Stanwick was never reoccupied, activity at Dragonby had been re-established by the end of the first century a.d. Excavations there revealed a flourishing settlement that continued into the fourth century, with evidence for stone-footed, aisled buildings, stockkeeping, craftworking and trade.

45

N

40

Area of excavation

40 45 50

55

40

0

500 m 1:10000

fig.7.26  Plan

Romanch7.indd 261

of cropmarks showing the late Iron Age/Roman ladder complex at Burton Fleming (Tabor 2009)

15/09/2016 17:29:47

262

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Ladder settlements have been discussed previously in relation to complex farmsteads, though some expanded to a size where they are better understood as villages. They have been widely recognised through aerial photography of cropmarks in the North-East, particularly on the Yorkshire Wolds (Stoertz 1997). Several examples can be seen to follow the contours of valley slopes, such as at Burton Fleming (Tabor 2009; fig. 7.26) and Heslerton (Powlesland et al. 2006), both of which originated in the later Iron Age, implying that the local topography was fundamental to their development. Both these sites consist of a large number of contiguous enclosures running along both sides of a trackway, most probably linked to the corralling and long-distance movement of domestic livestock (cf. Giles 2007). It is important to note here that the long trackway identified by Powlesland at Heslerton, which was in use during the late Iron Age, appears to be the same as the hypothesised Roman road (Margary route 816) running between the fort and vicus at Malton/ Norton, along the northern foothills of the Yorkshire Wolds, to the signal station on the coast at Filey. Other than the discovery of small sections

fig.

Romanch7.indd 262

of metalled road surfaces at Malton and Filey, no traces of it have otherwise been discovered in between (Margary 1955, 424–5). It is very likely that the route between the two Roman settlements was the late Iron Age trackway, which continued to be used throughout the Roman period. Buildings The North-East dataset includes records for 644 buildings from 156 settlements, an average of just over four buildings per settlement, close to twice the mean number recorded in the South and East regions. Excavated buildings are widely distributed across the region, being well represented in the Tees Lowlands, but are less concentrated to the south of the Humber and on the Magnesian Limestone (fig. 7.27). Circular buildings are nearly twice as numerous as rectilinear buildings, though the distribution of the two forms is generally similar (fig. 7.28). Holderness and the North York Moors lack rectilinear buildings, which may be due to a relative lack of post-conquest settlement in these areas. Sites with circular buildings were more

7.27   Distribution of all buildings in the North-East

15/09/2016 17:29:58

THE NORTH-EAST

263

fig. 7.28  Distribution of circular and rectangular buildings in the North-East

Romanch7.indd 263

15/09/2016 17:30:15

264

THE RURAL SETTLEMENT OF ROMAN BRITAIN

circular

100

rectangular

90 80 no. of sites

70 60 50 40 30 20 10 0 Late Iron Age

3rdC AD

4thC AD

7.29.   Use of circular and rectangular buildings in the North-East over time

common than those with rectangular buildings until the third century a.d., and they continued to be constructed into the fourth century, perhaps reflecting elements of conservatism (fig. 7.29). At The Bridles, Barnetby le Wold, the construction of a timber roundhouse within an enclosure in the fourth century a.d. appears to have followed a long tradition of circular building use that stretched back into the late Iron Age (Allen and Rylatt 2002). Only two sites with evidence for rectangular buildings date to the late Iron Age, at Faverdale, Darlington (Proctor 2012), and Pig Hill (Northern Archaeological Associates 2004), both in County Durham. The function of these structures is uncertain; the example at Faverdale was located to the south of the main domestic area and it may have been an ancillary building. The construction of rectilinear buildings only becomes more common from the late first century a.d., though the earliest examples that appear to have been used for domestic habitation seem to occur at roadside settlements and military vici. Farmsteads Most buildings in the North-East dataset are found at farmsteads, where circular, timber/masswalled buildings were particularly dominant (table 7.3). Although the chronological pattern of building use on farmsteads is similar to that for all settlements, the data show that the uptake of rectangular building was far less pronounced, though circular and rectilinear buildings are recorded in roughly equal numbers of sites in the third and fourth centuries. Rectilinear buildings are better represented on complex farmsteads than enclosed farmsteads, and are almost entirely absent from open farmsteads (fig. 7.30). This pattern is partly chronological, owing to the increasing number of complex farmsteads present in the middle and late Roman period when rectilinear buildings were also more common. Of the eight complex farmsteads with buildings

Romanch7.indd 264

2ndC AD

circular

25

rectangular

20 no. of sites

fig.

Late 1stC AD

15 10 5 0 complex farmstead

enclosed farmstead

open farmstead

fig. 7.30.   Use of circular and rectilinear buildings on farmsteads of different types

dating to the fourth century, none are circular in plan. The roundhouse excavated within the late Roman complex farmstead at Holmfield Interchange, Site Q, is dated by a few sherds of calcite-tempered pottery and it was very uncertain whether this related to the occupation of the house (Brown et al. 2007, 69). The contemporary use of circular and rectilinear buildings on some farmsteads may suggest differences in function and possibly prestige afforded to different areas of the settlement – for example, at Crossgates, Seamer, four late Iron Age, timber roundhouses lay within a sub-divided enclosure and were added to in the first century a.d. by a second enclosure that contained a rectangular, limestone building. Timber buildings vastly outnumber masonry structures on farmsteads, with many circular buildings defined by drainage gullies, and some providing evidence of postholes for internal support. On many sites, these gullies can be seen to cut one another, indicating the gradual replacement of these buildings over time. Postholes that mark out the external walls of roundhouses tend to be short-lived, though a good, late Iron Age example can be seen at Welton Wold (Mackey 1999). The majority of

15/09/2016 17:30:18

THE NORTH-EAST

table

Site type

7.3: buildings

265

in the north-east region

Circular Rectangular No. circular No. rectangular No. timber No. masonry buildings buildings buildings buildings buildings buildings (no. sites) (no. sites)

Farmstead Farmstead/villa Villa

105 36 350 62 384 26 10 10 34 54 33 55 1 8 1 25 0 26

Roadside settlement 3 9 10 44 36 17 Vicus

0 3 0 26 4 22

Village

3 1 21 10 23 8

Shrine

0 1 0 2 0 2

Industrial Total

0 2 0 5 3 2 122 70 416 228 483 158

a

b N

N

Wall footings on residual bank

0

20 m 1:500

fig. 7.31.   Plans of drystone masonry structures on the North York Moors at (a) Percy Rigg, Kildale (Close 1972), and (b) Great Ayton Moor (Tinkler and Spratt 1978)

rectilinear timber buildings are of posthole construction, though a few have been identified from post-in-ditch foundations, such as those recorded at Thorpe Hall, Eastrington (Wood 2011). Many of these buildings were small and fairly simple structures, and the identification of their remains is likely to be adversely affected by truncation. Stone buildings are very poorly represented on farmsteads, found on only 16 per cent of sites with evidence for structures. In the North York Moors, localised examples of drystone, circular buildings have been identified at Percy Rigg, Kildale (Close 1972), Crag Bank, Kildale (Close et al. 1975), Great Ayton Moor (Tinkler and Spratt 1978), and Roxby site 2 (Inman et al. 1985), though subtle differences in construction technique occur between sites, such as the use of internal paving (fig. 7.31). These structures all date to the late Iron Age, with those at Crag Bank and Roxby site 2 continuing to be used in the Roman period.

Romanch7.indd 265

Most masonry structures on farmsteads are rectilinear in form, and are commonly unmortared, single-room structures, such as those identified in Edlington Wood, South Yorkshire (Corder 1951). Signs of significant investment or elaboration in masonry are very rare. Only two buildings with hypocausts are known from farmsteads in the North-East, one at Heslington East (Roskams and Neal 2012), and another at Faverdale, Darlington (Proctor 2012). The lack of a main villa house at these two sites means that they are classified here as ‘farmsteads’, but they clearly display indications of status and perhaps should not be considered much differently from modest villas elsewhere in social and economic terms. The building at Heslington East was located near a masonry ‘tower’, which was interpreted as a monumentalisation of the western entrance into the settlement. A stone-walled octagonal building at land off Horkstow Road, South Ferriby, may

15/09/2016 17:30:19

266

THE RURAL SETTLEMENT OF ROMAN BRITAIN

a

b

c

Possible corndryer

N

0

20 m 1:500

fig. 7.32.  Post-built ‘structures at (a) Melton A63 (Fenton-Thomas 2011), (b) Cedar Ridge, Garforth (Owen 1998), and (c) Stile Hill, Colton (Archaeological Services WYAS 1995)

Romanch7.indd 266

8 7 6 no. of sites

have been a shrine or a memorial (Clay 2006). Such investments in masonry building were exceptional on farmsteads in the North-East. The specific function of most buildings is usually indeterminable without further evidence, though many are likely to have been primarily for domestic use. Possible agricultural buildings have been excavated at Melton A63 (Fenton-Thomas 2011), Cedar Ridge, Garforth (Owen 1998), and Stile Hill, Colton (Archaeological Services WYAS 1995). Dating to the second and third centuries a.d. these buildings were timber, post-built constructions, with the example at Melton A63 formed of closely aligned postholes, indicating that it held a raised floor (fig. 7.32). All these buildings were located close to corndryers, which perhaps suggests their use as granaries. Other buildings may have been ancillary workshops, such as the late Roman post-built structure at Thorpe Hall (Wood 2011), or the late Iron Age beam-slot structure previously mentioned at Pig Hill (Northern Archaeological Associates 2004), both of which produced considerable evidence for ironworking. Cellared buildings have been identified at six sites, and those at Wattle Syke (Martin et al. 2013) and Dalton Parlours (Wrathmell and Nicholson 1990) have produced evidence that they were used for a range of functions, including crop processing, grain drying, animal skinning, antlerworking, cooking, and smithing. However, at Welton Wold (Mackey 1999) and Sewerby Cottage Farm, Bridlington (Fenton-Thomas 2009), the insertion of corndryers into cellared buildings indicates that some were used more specifically for agricultural purposes.

5 4 3 2 1 0 coage/hall

fig.

corridor

winged corridor

courtyard

uncertain

7.33.  Villa building forms in the North-East

Villa complexes The few villas excavated in the North-East contain a total of 81 buildings, the majority of these being ancillary to the main domestic residence. As with other areas of the country, villa houses ranged considerably in size and complexity (fig. 7.33). Small corridor buildings with minimal evidence for interior refinement have been excavated at Scruff Hall Farm, Drax (Wilson 1965), and Welton Wold (Mackey 1999), while similarly modest striphouses, such as at Rudston (Stead 1980) and Holme House (Harding 2008), incorporated bathhouses and tessellated flooring (fig. 7.34). The villa at Scampton is a somewhat grander and more complex affair, though, unfortunately, it was largely excavated in the late eighteenth century, and the dating evidence is poor (Illingworth 1808). Aerial photography of the site suggests that the settlement complex comprised an array of

15/09/2016 17:30:21

THE NORTH-EAST

a

267

b

N

c

d

0

20 m 1:500

e

f

0

100 m 1:2000

fig. 7.34.  Comparative plans of six villa houses: (a) Scruff Hall Farm, Drax (corridor) (Wilson 1965), (b) Welton Wold (corridor) (Mackey 1999), (c) Rudston (strip-house) (Stead 1980), (d) Holme House (strip-house) (Harding 2008), (e) Winterton (courtyard) (Stead 1976), and (f) Beadlam (courtyard) (Neal 1996)

Romanch7.indd 267

15/09/2016 17:30:22

268

THE RURAL SETTLEMENT OF ROMAN BRITAIN

buildings, including a bathhouse, arranged round two courtyards. The substantial courtyard villa at Winterton, North Lincolnshire, was also subject to antiquarian excavations, though more recent investigations in the 1950s showed that the settlement developed gradually from the late firstcentury a.d. farmstead into a far more sophisticated complex (Stead 1976). By the end of the second century a.d., the settlement boasted several buildings with hypocausts, including a bathhouse, two aisled buildings, a workshop and other structures, together arranged around a central courtyard. A similar complex arrangement of buildings is also found at Beadlam villa, where two winged-corridor houses were set side on, once more with other structures to form a central courtyard (Neal 1996). Aisled buildings have been identified on six villas in the North-East. Most of these appear to have been ancillary to the main villa house, apart from at Chapel House Farm, Dalton-on-Tees, where a winged-corridor villa developed from an earlier aisled structure (Brown 1999). Otherwise, these buildings range from the comparatively large, masonry-footed structures at Dalton Parlours and Ingleby Barwick, to the more modest timber buildings at Welton Wold. Circular or oval masonry buildings have been found at five villas. The example at Langton was tentatively interpreted as a mill, and was associated with the insertion of an apparent threshing-floor nearby (Corder and Kirk 1932). An agricultural or an industrial function may be supported by evidence from Ingleby Barwick, where a large timber, circular structure (c. 8 m diam.), rebuilt in stone during the fourth century a.d., included a compacted earthen floor built over an oven (Willis and Carne 2013). Evidence for iron-smithing in the circular structures at Winterton and Beadlam suggest that some may have been workshops, and were potentially important elements in the economy of these villas (Stead 1976; Neal 1996). Nucleated settlements Evidence for buildings at nucleated settlements have been recorded from roadside settlements, military vici and villages, and include a range of different structural forms (table 7.3). Rectangular buildings are far more common than circular buildings at roadside settlements and vici, both in terms of the number of sites and the overall number of buildings, though the use of timber and masonry construction is more equally represented. In contrast, circular buildings are more common at villages, though this only relates to three sites, with Dragonby being the only one with rectilinear structures standing alongside circular ones. Eight of the Dragonby examples were masonry

Romanch7.indd 268

constructed, with some showing evidence of central postholes (May 1996). In contrast, late Iron Age and early Roman buildings at Wattle Syke (Martin et al. 2013) and Low Caythorpe (Fraser and George 2013) were exclusively roundhouses. The dominance of rectilinear buildings at roadside settlements signifies the different architectural traditions of these settlements compared to villages. Four of the post-built roundhouses at Rudstone Dale, Newbald, dated to the late Iron Age, prior to the site’s development as a roadside settlement. It is perhaps significant that the only circular structure found during the Roman phase contained the burial of a sheep or goat and a human cremation within its central area, and a neonate within the north-west arc of its putative wall, suggesting that it may have been a shrine (Wood 2011). Architecture at Bainesse (Site 46) consisted of mostly rectilinear, timber buildings, located alongside a smaller number of masonry structures (fig. 7.35). Their alignment along the road, along with the recovery of ironand boneworking waste, and a wide array of artefacts, suggests that they may have performed a mixture of domestic, commercial and industrial roles (Wilson 2002). The timber buildings found here used a combination of beam-slot and posthole construction, a technique that also appears to have been typical in other roadside settlements, such as at Shiptonthorpe (Millett 2006). Masonry buildings tend to be more common than timber structures at military vici, though this partly reflects an apparent desire for stone architecture in the immediate vicinity of the forts. Outside the third-century a.d. fort at Piercebridge, a number of large masonry buildings were constructed, including a temple dedicated to Jupiter Dolichenus (Cool and Mason 2008). At Malton, masonry buildings were built just south of the fort, enclosed in an annexe between the defences and the River Derwent. In the fourth century a.d., a major rebuilding programme in the area saw the addition of eight new structures, including a town house with a fine mosaic floor and painted plaster, a bathhouse, and a kiln building. Finds recovered from the site included a large number of high-status items, such as hair pins, finger rings, bracelets, toilet equipment, and a considerable quantity of vessel glass (Mitchelson 1964; Wenham and Heywood 1997). These ‘annexes’ are largely restricted to the military vici, having more in common with larger towns than rural nucleated settlement. This finding seems to be consistent with Bidwell and Hodgson’s suggestion that some vici developed into regional market centres, perhaps existing as ‘military towns’ in the late third and fourth centuries a.d., such as at Catterick (Bidwell and Hodgson 2009, 33–4).

15/09/2016 17:30:22

THE NORTH-EAST

a

269

N

Corndryer

Dere Street

Projected timber buildings Gravelled surfaces

0

20 m

Masonry building

1:500

N

b

Aisled building

Pits

Enclosure ditch

fig. 7.35.  Plans of buildings at roadside settlements: (a) timber and masonry structures at Bainesse (Site 46) (Wilson 2002) and (b) a beam-slot and posthole structure at Shiptonthorpe (Millett 2006)

Romanch7.indd 269

15/09/2016 17:30:24

270

THE RURAL SETTLEMENT OF ROMAN BRITAIN enclosed farmsteads height above ordnance datum (m)

height above ordnance datum (m)

complex farmsteads 100+ 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9 0

1

2

3 4 5 6 no. of sites

7

100+ 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

8

0

0

fig.

1

2

3 4 5 6 no. of sites

7

8

3 4 5 6 no. of sites

7

3 4 5 6 no. of sites

7

8

100+ 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9 0

1

2

8

7.36.   Spot height analysis for major settlement types in the North-East

Landscape context and infrastructure Topography As outlined at the beginning of this chapter, the North-East region consists of a mixture of lowlying Vales, the upland landscapes of the North York Moors and the chalk Wolds of Yorkshire and Lincolnshire. Much of the region, other than the upland areas, is covered with alluvial deposits of clay and till. As has been shown previously, rural settlement tends to favour the fertile vales and the Magnesian Limestone. Excavated sites tend to be more visible on the chalk Wolds compared to other upland areas, such as the peatland of the North York Moors, a pattern that is supported by cropmark evidence (e.g. Stoertz 1997). The settlement pattern is, of course, heavily biased by modern factors (see Chs 1 and 12) as well as cultural influences, such as the development of towns, military sites, and the major roads. However, there does appear to have been a strong relationship between rural settlements and the underlying geology of the North-East region. The preference of farmsteads in the region for low-lying ground is demonstrated by spot height data, which shows that 74 per cent were sited below 60 m OD, and 40 per cent located on land below 20 m OD, though variations emerge for particular types of site (fig. 7.36). Complex farmsteads are frequently encountered on lowland areas, particularly on superficial till and clay geologies. In comparison, enclosed farmsteads

Romanch7.indd 270

2

roadside settlements/vici

100+ 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

height above ordnance datum (m)

height above ordnance datum (m)

villas

1

were more commonly located on higher ground, with a number of sites found on land over 100 m OD. The highest are farmsteads on the North York Moors, such as Percy Rigg, Great Ayton Moor, Crag Bank, the Levisham Moor enclosures, and Roxby sites 1 and 2, sited between 190–280 m OD. These sites tend to include evidence for stock pens and were probably engaged in upland pastoral farming. As seen in other regions, roadside settlements and military vici are consistently located at low levels, especially below 40 m OD, and generally close to rivers, where their forts were strategically sited. Villas, on the other hand, were more evenly distributed at different heights. Beadlam villa, for example, was located on slightly raised ground next the River Riccal, which may have been an important influence on the siting of the villa. In contrast, Langton, Wharram Grange and Wharram Street all occupied higher ground at the edge of the Yorkshire Wolds. Transportation: rivers, roads and trackways The Humber estuary formed an important conduit for the movement of goods and people into and out of the North-East, with the town of Petuaria at Brough and the roadside settlement at Winteringham on the opposite bank long thought to have been of naval importance (Wacher 1969; 1995; Stead 1976). Certainly, epigraphic evidence suggests that the Ouse was navigable and that York was a port of trade (Hawkes et al. 1946, 67; RIB

15/09/2016 17:30:25

THE NORTH-EAST

271

a Trackway

Enclosed farmstead

Trackway Complex farmstead

Trackway

0

400m

N

1:8000

b

Trackway

Trackway

fig. 7.37.  Plans of long-distance trackways with associated field systems and settlements at (a) Holmfield Interchange (Brown et al. 2007), and (b) Swillington Common, West Yorkshire (Roberts et al. 2001)

Romanch7.indd 271

15/09/2016 17:30:26

272

THE RURAL SETTLEMENT OF ROMAN BRITAIN

trackway have recently been identified near Bramham (Weston and Roberts 2015) and near Collingham (Gregory et al. 2013, 138–41). Field systems Aerial photography has produced considerable evidence for late Iron Age and Romano-British field systems in many areas of the North-East region. The visibility of cropmarks is greatest on the Sherwood Sandstones, the Yorkshire Wolds, and the Magnesian Limestone Belt, where previous studies have demonstrated a variety of field forms (Riley 1980; Stoertz 1997; Roberts 2010). The evidence is far sparser in the Vale of York and on the Humberhead Levels, where the alluvium and glacial clay deposits have created a masking effect, reducing visibility (Taylor 2007, 44), as well as a warping effect that alters the appearance of some field systems (Roberts pers. comm.). A total of 120 excavated sites in the North-East include evidence for field systems. These are widely distributed across the region, with a notable concentration occurring on the Magnesian Limestone Belt. Just over half of all these sites (63) are associated with farmsteads, and two with villas. Often field systems have been identified only from fairly insubstantial traces of ditches, though, where cropmark evidence survives, some settlements can be observed as parts of a much wider agricultural landscape. The complex of ‘brickwork’ fields at Dunstan’s Clump on the Sherwood Sandstones, now seen as a classic example of this type of land management, included numerous settlements interspersed among the extensive pattern of parallel strip-fields (fig. 7.38).

N

I.653; RIB III.3195), though archaeological evidence for the putative Roman harbour is limited (Ottaway 1993, 69, 85). In general, however, the extent to which rivers were used as transport routes in the North-East is difficult to assess, owing to a lack of evidence for wharfs/ jetties. At Perrins Cottages, Fiskerton, a possible docking area was tentatively suggested from the identification of a limestone, rubble surface, mixed with quantities of late Roman material, which could have functioned as an access point to the River Witham (Palmer-Brown 1994). Beyond the major rivers, it appears that riverine transport was not important, particularly for the rural population, who were perhaps more dependent on the road system and local trackways. A relatively high proportion of villas were located within 2 km of a major road (60 per cent), implying that easy access to these highways was desired. In contrast, only 22 per cent of complex farmsteads were located that close to the road network, which is perhaps surprising given the clear association of this type of settlement with major roads in other regions in the south and east of the country (see Chs 5 and 6). However, trackways were recorded on 75 per cent of complex farmsteads, highlighting the importance of the transport network beyond the major roads in the North-East. In some areas, cropmark evidence shows trackways running over considerable distances, surrounded by field systems and linking individual settlements (Brown et al. 2007; Roberts et al. 2001; fig. 7.37). Dating evidence from some settlements suggests that many of these landscapes had continued from the late Iron Age, though two examples of metalled sections of late Roman rural

0

2 km 1:50000

fig.

Romanch7.indd 272

7.38.   Plan of Romano-British ‘brickwork’ fields at Dunston’s Clump, Babworth (Garton 1987)

15/09/2016 17:30:27

THE NORTH-EAST

Fieldwalking around one of the enclosure settlements produced evidence for manuring in the fields closest to the settlement, indicating their use for arable cultivation (Garton 1987; 2008). Determining whether fields were under pasture or being cultivated is notoriously difficult, particularly since fields may have supported both livestock and crops at different times. Open areas of land have been identified on the Magnesian Limestone Belt by Roberts, who suggested that they may have been for communal use, perhaps for livestock grazing, as at Wattle Skye where a large, open space was encircled by several enclosure complexes (Roberts 2010, 26). The division of open land has also been identified on the Yorkshire Wolds at Cat Babbleton Farm, where Romanperiod pit-alignments have been traced over long distances, perhaps also signifying rights over grazing (Cardwell 1989). Of course, in order to distinguish between pastoral and arable usage of fields, much depends upon the recovery of suitable environmental evidence. On the Humberhead Levels near Doncaster, a rare preserved assemblage of waterlogged wood, pollen, and insect remains, recovered from late Iron Age/early Roman field ditches at Balby Carr, clearly showed that they were used for grazing livestock (L. Jones 2007). Dung beetles were relatively common, while the pollen evidence indicated the presence of grassland with hedges and patches of woodland. Coppiced roundwood, found with tool marks, may also have been used for fencing, perhaps for stock control. The site was waterlogged and appears to have been unsuitable for arable agriculture, which likely occurred on higher, drier ground. Although numerous botanical assemblages from the North-East demonstrate that cereal farming was widespread (see Plant remains below), evidence for the cultivation of specific field systems is more sporadic. At North Thoresby, Lincolnshire, phosphate levels in the fills of a number of regularly spaced ditches were consistent with manuring. The excavators speculated that the site was used for vine cultivation (Webster et al. 1967), though no pollen or macrobotanical evidence was available to support their interpretation, and the layout of the field system may simply have been consistent with drainage. Further possible evidence for horticultural trenches was also identified at Burnby Lane, Hayton, close to the valley floor, though as at North Thoresby, the lack of corroborating environmental evidence means that we do not know how this land was actually used (Halkon et al. 2015). Greater insights into the arable use of field systems could be achieved through the analysis of weed assemblages from farmsteads and field systems, and further research should be directed towards this end. Field systems

Romanch7.indd 273

273

and paddocks have also been recorded at roadside settlements, where ditched boundaries extended behind property plots lining the road (e.g. Wilson 2002). The presence of fields at these sites would support Millett’s suggestion that roadside settlements were essentially agricultural and pastoral in character, but also formed central places where produce could be gathered and processed (Millett 2006, 309). The dating of field systems in the North-East, as with other regions, remains problematic, though a considerable number of sites have been shown to originate in the late Iron Age and continue into the Roman period. Field systems were most numerous in the second century a.d., after which they reduced in number into the fourth century. The apparent decline in the number of field systems in use into the late Roman period is a pattern in common with the South and East regions. However, while many Romano-British field systems may have gone out of use, some appear to have influenced the form of the subsequent medieval landscape. Recent analysis of the medieval furlongs at Dunstan’s Clump shows that they were aligned with the primary boundaries of the Roman brickwork fields, which ran parallel to each other in an east–west direction. This suggests that, although the brickwork field system at Dunston’s Clump probably went out of use by the end of the Roman period, the boundaries could have survived as earthworks, providing a template on which the medieval open field system was lain (Rippon et al. 2015, 214). However, while this may have been the case at Dunston’s Clump, the extent to which this was repeated elsewhere is uncertain. Settlement hierarchies: the social and economic basis of settlements MATERIAL CULTURE It has been stated that trade and exchange in the North-East predominantly occurred on a local basis throughout the Roman period (Ottaway 2013, 146–9, 204–8). Although regional and long distance supply networks were facilitated by the army, there appears to have been minimal economic integration between the population of the countryside and those in urban and military centres. While broad differences in the type and quantity of ‘Roman’ commodities has been pointed out, the growing body of evidence from rural farmsteads is less well understood, and is here outlined further in order to better understand the different forms of settlement in the North-East.

15/09/2016 17:30:27

274

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Pottery As in the south and east of England, the most ubiquitous type of artefact recovered from Roman sites in the North-East is pottery. In the early Roman period, hand-made vessels common in the Iron Age continued to be used on rural sites, and there appears to have been little appetite for new ‘Roman’ forms of food preparation or dining customs (Ottaway 2013, 148). Later Roman ceramics, notably the fourth-century wares from Crambeck, are far more widely distributed among rural settlements, though the success of the Crambeck industry is largely attributed to its supply links with the military (Wilson 1989; Tyers 1996, 16, 74; Ottaway 2013, 290–1). The low level of trade in ‘Roman’ commodities between rural farmsteads and the towns and military bases in the North-East can be examined through the recovery of pottery on different types of site. Samian ware has been recovered in significantly higher proportions from roadside settlements and military vici, particularly at the latter where the use and trade of samian can be associated with the military (fig. 7.39). If anything, the vici data are under-represented since these counts do not include the exceptional ceramic assemblages from Piercebridge and Castleford (the latter is technically outside the region), which skew the data so heavily in favour of vici that the comparatively tiny quantities coming from farmsteads and villas become almost undetectable. However, compared to samian ware, mortarium sherds are recovered in far more equal quantities

average no. of sherds per site

samian 400 350 300 250 200 150 100 50 0 roadside vicus (n=7) villa (n=8) complex enclosed farm (all) settlement farmstead farmstead (n=90) (n=13) (n=23) (n=23)

average no. of sherds per site

mortaria 40 35 30 25 20 15 10 5 0 roadside vicus (n=7) villa (n=8) complex enclosed farm (all) settlement farmstead farmstead (n=90) (n=13) (n=23) (n=23)

fig. 7.39.  Recovery of samian and mortaria from different types of site in the North-East, based upon average sherd counts per site (not including Piercebridge)

Romanch7.indd 274

between different types of site, which might suggest that the adoption and use of mortaria by rural communities was more consistent with those living along the road network. The distinction between the consumption of both samian and mortaria on villas and complex farmsteads, compared to enclosed farmsteads also draws attention. It is perhaps surprising that there is very little difference between the average sherd counts deriving from villas and complex farmsteads, yet both of these settlement types produce much greater quantities than that recovered from enclosed farmsteads. On closer inspection, the data from complex farmsteads is inflated by pottery from three sites: Faverdale, Heslington East, and Burnby Lane, Hayton. These are exceptional sites in terms of their material culture and architecture, and clearly stand out from other farmsteads. These sites remind us that there was probably little difference in terms of social standing and wealth between the higher-status, complex farmsteads and the more modest villas. Although in general a low level of economic integration between settlements on the road network and farmsteads may have been apparent, the data suggest that there existed some variation, with more evidence for trade and exchange being detectable on some farmsteads than others. Small finds The range of objects from the North-East closely mirrors those found in other regions, with quern stones, coins, knives and other tools featuring prominently. At roadside settlements, military vici and villas, most artefact types occur at similar frequencies, and are consistently more common than at farmsteads (fig. 7.40). Coins, for example, are only recorded at 28 per cent of farmsteads, compared to 87 per cent of roadside settlements and 86 per cent of villas. By far the greatest numbers of coins have come from the roadside settlements at Shiptonthorpe and Hayton (Millett 2006; Halkon et al. 2015). However, these datasets are skewed by large numbers of coins from extensive metal-detecting programmes, alongside coinage recovered from excavations and fieldwalking. The largest brooch assemblage comes from the village at Dragonby, Lincolnshire, which included a considerable number of late Iron Age La Tène types (May 1996). Although coin and brooch assemblages are frequently small or absent from farmsteads, two notable exceptions are Burnby Lane, Hayton (Halkon et al. 2015), and Heslington East (Roskams and Neal 2012). These settlements are, perhaps significantly, located close to major nucleated centres and, as discussed above with regards to pottery, were probably higherstatus farmsteads more akin to villas.

15/09/2016 17:30:28

fig.

% of sites with object present

100 90 80 70 60 50 40 30 20 10 0

% of sites with object present

THE NORTH-EAST

100 90 80 70 60 50 40 30 20 10 0

roadside selement/vicus (n=15)

villa (n=14)

complex farmstead (n=40)

farmstead (n=194)

enclosed farmstead (n=31)

7.40.   Frequency of major artefact categories on different types of rural settlement in the North-East region

Items concerned with Roman styles of dress and personal display, such as finger rings, toiletry objects and hair pins are almost completely absent from farmsteads, but are comparatively well represented at roadside settlements and villas. Combs for personal use have been recovered from fifteen sites, almost exclusively nucleated settlements and villas, including wooden examples from Shiptonthorpe (Millett 2006) and Rossington Bridge (Buckland et al. 2001). The only examples of bone combs recovered from farmsteads have occurred at Melton wastewater works, near Brough (Bishop and Westwood 2004), and Chase Farm, North Killingholme (Humberside Archaeology 1991). In contrast to personal-use items, utilitarian objects, such as food-processing items and knives/ tools, are more equally recovered from each of the three main settlement types, indicating the need for these object classes on farmsteads over other forms of material culture. Quernstones are notably more widespread across the Tees Lowlands and the Magnesian Limestone Belt in comparison to coins and brooches. Millstones are also widely distributed across the region, but have been found on only eighteen sites, seven of which are nucleated settlements that may have operated as centres where agricultural produce could be gathered and processed. Other settlements with high numbers of millstones include the village settlement at Wattle Syke on the Magnesian Limestone Belt (seventeen), and the villa at Ingleby Barwick in the Tees Lowlands (six), and these sites may have

Romanch7.indd 275

villa (n=14)

275

similarly acted as centres for the centralised processing of grain in these sub-regions. Equipment for textile processing is very well represented at nucleated settlements, including the exceptional quantity of spindlewhorls, loomweights and sewing needles recovered from Dragonby (May 1996), and the two weaving combs and four weaving tablets at Malton (Wenham and Heywood 1997). The concentration of textile-processing artefacts at roadside settlements, military vici and villages may be a reflection of commercial activity, as opposed to household craft, which would involve the centralisation of sheep management, or at least their wool. Other potential indicators of commercial activity are comparatively sparse. Writing equipment has been identified at fourteen sites, mostly villas, though greater numbers of styli and seal boxes have been recovered from nucleated settlements, especially Dragonby (May 1996), Bainesse (Wilson 2002) and Shiptonthorpe, the latter of which contained two silver fir writing tablets (Millett 2006). Weighing objects, such as steelyards, are equally rare, coming from eighteen sites, though farmsteads more commonly produce this type of artefact. Religious objects appear most frequently on nucleated settlements, occurring in their greatest number at Dragonby, which contained two figurines, one of Mars Gradivus and the other of Mars Ultor, as well as ‘votive’ plaques and two Iron Age miniature shields. The votive finds were interpreted as being associated with a Mars cult

15/09/2016 17:30:31

276

THE RURAL SETTLEMENT OF ROMAN BRITAIN

centre, or at least were indicative of a shrine (Alcock 1996; May 1996, 603). As well as nucleated settlements, eight farmsteads have also produced religious items, including a sceptre head in the form of Mars from Blackhills Farm and The Hollys, Wickenby (Hall 2008), and a pipe-clay figurine from Millfield Farm,Wheldrake (Robinson 2009), perhaps reflecting religious observance being carried out at individual households. In addition to items of a more ‘classical’ type, a number of sites include apparent religious objects that could have been worn, such as a phallus amulet from Burnby Lane (Halkon et al. 2015), a perforated boar’s tusk from Melton A63 (FentonThomas 2011), and a drilled dog tooth pendant from Newbridge Quarry, Pickering (Richardson 2012). As in other regions, villas clearly stand out in terms of their profile of material culture compared to farmsteads in the North-East. Artefacts relating to security (locks, keys, etc.) and recreation (gaming counters, dice, etc.) are fairly common finds in villa assemblages, but are exceptionally rare on farmsteads, reflecting differences in lifestyle and attitudes towards protecting property. There is little difference between the profiles of material culture of enclosed farmsteads and complex farmsteads, although the range of artefacts from complex farmsteads is slightly greater, with some producing lighting and writing equipment for example. Perhaps more surprising is the poor representation of agricultural tools (ards, spades, and scythes, etc.) and equine/ transport-related equipment (cart-fittings, hipposandals, etc.) at all types of farmsteads, considering that arable farming and livestock husbandry would have been the preoccupation of most of the population. Items associated with ploughing are particularly rare, being restricted to an ard tip recovered from Malmo Road, Hull (Tibbles 1992), and a ploughshare from Gunhills, Armthorpe (Richardson 2008). However, harvesting and horticultural tools, such as scythes, pitchforks, pruning hooks and spade shoes, are slightly more common. Environmental evidence The environmental evidence from excavated sites is unevenly distributed across the North-East, with a greater number of animal bone and plant assemblages located in the central and southern parts of the region. These assemblages are predominantly restricted to low-lying land in the river valleys, but also on the Magnesian Limestone and the chalk where the alkalinity of the soils are good for bone preservation, but less so for botanical remains unless they are charred.

Romanch7.indd 276

Animal bones A total of 90 well-dated, animal bone assemblages of over 100 identified specimens (NISP) have been recovered from sites located widely across the North-East. The largest assemblages derive from Piercebridge, Dragonby, Winterton villa, Wattle Syke, and Dalton Parlours, though there is a considerable lack of faunal assemblages from the Sherwood Sandstones, the North York Moors, and the Lincolnshire Coast and Marshes, which prevents a thorough examination of intra-regional variability. Despite these restrictions, a broad comparison of animal bone assemblages between different landscape zones shows some distinct variations in the relative frequencies of the cattle, sheep/goat and pig bones (fig. 7.41). The Vale of York and Mowbray is heavily dominated by cattle bone, being almost twice as frequent as those of sheep/goat, whereas assemblages from Holderness and theYorkshire Wolds produce a higher frequency of sheep/goat remains. This is perhaps a reflection of the greater suitability of the chalk Wolds and the coastal lowlands for sheep-grazing, while settlements in the Vale of York may have been concerned with the supply of cattle to York itself, where their bones have been recovered in considerable quantities (O’Connor 1988, 75–81). An increasing frequency of cattle remains over time also occurred in the North-East, as with other regions in the south and east of England, and is a pattern which has been previously documented from national trends (King 1984; 1999). Sheep/goat remains clearly dominate over cattle in the late Iron Age and in the early Roman period, but are overtaken by cattle in the middle Roman period, and continue to be the most common domesticate into the fourth century a.d. (fig. 7.41). Within this shift of emphasis, some assemblages continue to exhibit high sheep/goat percentages into the later Roman period, particularly on the Magnesian Limestone at Wattle Syke (Martin et al. 2013) and Parlington Hollings (Roberts et al. 2001), where a preference for mutton appears to have been maintained, perhaps alongside a greater emphasis on wool exploitation. A slight increase in the frequency of pig bones is also evident over time; they tend to be more common on sites in the Vale of York and Mowbray, though exceptionally high proportions of pig bone have been recorded at Cedar Ridge, Garforth (44 per cent) (Owen 1998). Alongside the evidence for intra-regional and chronological variation in the animal bone assemblages, there are also some distinct differences between different site types (fig. 7.41). While assemblages from villas are disappointingly rare, they do show a clear preference for cattle, with some including a relative abundances of pig

15/09/2016 17:30:31

THE NORTH-EAST cale

60

sheep/goat

277

pig

mean %NISP

50 40 30 20 10 0 Vale of York and Mowbray (n=8)

Southern Magnesian Limestone (n=15)

60

North Lincolnshire Coversands and Clay Vales (n=10)

Holderness (n=8)

Yorkshire Wolds (n=8)

mean %NISP

50 40 30 20 10 0 LIA (1st C BC-mid 1st C AD) n=7

LIA/ER (1st C BC/AD) n=13

ER (1st-2ndC AD) n=16

MR (2nd-3rdC AD) n=12

LR (3rd-4thC AD) n=19

Mean %NISP

60 50 40 30 20 10 0 villa (n=6)

farmstead (complex) (n=18)

farmstead (enclosed) (n=10)

farmstead (all) (n=60)

roadside selement /vicus (n=12)

village (n=6)

fig. 7.41.   Relative frequency of cattle, sheep/goat and pig: (a) by landscape zone; (b) by period; and, (c) by major settlement type (mean percentages from sites with >100 NISP)

bones, most notably at Dalton Parlours (15 per cent) and Holme House (31 per cent in the early Roman phase). The preference for pork on villa sites is also found in other regions and is a welldocumented feature nationally (King 1984; 1991; 1999). Dalton Parlours and Holme House also produced relatively high proportions of sheep/goat bones, which is at odds with assemblages from other villas. It is uncertain whether this reflects a real difference in consumption patterns, or whether the villa assemblages were affected by residuality from earlier phases of occupation. Unlike pottery consumption (see above, p. 274), there is very little difference in the proportion of major domesticates between different types of farmsteads, though there is significant variation between individual assemblages. Faunal assemblages from the villages at Dragonby, Low Caythorpe and Wattle Syke all produced a consistent pattern of high sheep/goat frequencies, though roadside settlements and military vici, while being heavily cattle-dominated overall, also greatly vary between different sites. It may be significant that the highest proportions of cattle bone were identified in assemblages from the vici at Doncaster (Buckland and Magilton 1986; Chadwick and Burgess 2008), Piercebridge (Cool and Mason 2008) and Rossington Bridge (Buckland et al. 2001), which may reflect the importance of cattle to the military (cf. King 1999). In contrast, the faunal material recovered from the roadside settlement at Mount Pleasant

Romanch7.indd 277

House, Nettleton and Rothwell, was overwhelmingly dominated by remains of sheep/goat, which was possibly associated with religious activity occurring in the vicinity of the shrine identified at the site (Willis 2013). Few faunal assemblages in the North-East have been recovered from overtly religious contexts, though special mention must be given to the Iron Age chariot burial and nearby shrine at Ferry Fryston (Brown et al. 2007; fig. 7.42). The chariot burial itself dated to the second century b.c., though radiocarbon dating of cattle remains recovered from the upper fills of the enclosing ditch demonstrated that the monument was a site of ritual significance well into the Roman period, with most of the remains having been deposited in the third and early fourth centuries a.d. (Bates et al. 2007). The bones derived from a minimum of 162 cattle, but were almost exclusively skulls and right forelimbs, articulating from the shoulder to the foot, with little evidence for butchery marks, suggesting careful methods of carcass dismemberment (ibid.). Tooth wear patterns showed that most of the animals were slaughtered at either one-and-a-half or two-and-a-half years old, indicating periods of intensive culling during the late summer/autumn months. It is therefore possible that the slaughter may have been a seasonal event, with deposits accumulating over a period of time. The lack of carnivore gnawing on the bones demonstrated that the body parts were deposited quickly, or at least were protected from

15/09/2016 17:30:34

278

THE RURAL SETTLEMENT OF ROMAN BRITAIN

N

Early-prehistoric feature

Early-prehistoric features

Cattle bones placed in ditch Iron Age chariot burial Neolithic henge Possible RomanoBritish shrine

P

0

die

me

tos

l va

-ha

ha

50 m 1:1000

fig. 7.42.   Religious monuments at Ferry Fryston showing location of cattle deposits around the Iron Age chariot burial (Brown et al. 2007)

scavengers, leading Orton (2006) to suggest that the remains may have been curated for some time prior to being deposited in a single episode. In terms of the sources of origin of the cattle, strontium isotope analysis showed that many were not raised on the Magnesian Limestone, where the site is located, but were imported from further afield (ibid.). The cattle chosen for deposition may have been traded from a variety of sources, or possibly the rituals undertaken at the site were fulfilled by gatherings of people arriving with cattle over long distances (either as live animals for ritual slaughter or as carcass parts). Another striking aspect of the site is the absence of Roman material culture, a feature also raised by Hodgson who suggested ‘that the participants in ceremonies here were detached from Roman provincial culture’ (Hodgson 2012, 52). Although the true nature of the activities being undertaken are uncertain, the Ferry Fryston material is a prime example of the important spiritual role held by livestock throughout the Roman period. Alongside the three main domestic livestock species, a range of other animals make up the remaining fauna in animal bone assemblages from the North-East, the most common of which are equids. Horse bones almost always form less than 20 per cent of faunal assemblages, though this varies from site to site. Of the sites that contain over 10 per cent horse bones, eight are complex farmsteads and only two are enclosed farmsteads. The use of horses for riding perhaps outweighs their use as pack animals, though increasing

Romanch7.indd 278

evidence for other equid species is now emerging. Reanalysis of late Iron Age equid specimens from Thorpe Thewles by Johnstone identified evidence for donkey and mule in the assemblage, which she suggested might have been imports from the continent (Johnstone 2004, 246). Other examples of possible mules have been identified at Hayton fort (ibid.) and a late first/second-century a.d. burial in the vicus at Healam Bridge (Ambrey et al. 2011), suggesting that these animals were being utilised by the Roman military. Dog bones are found on 75 per cent of sites, though rarely in any great quantity, unless recovered as associated bone groups, as discovered at Dalton Parlours where two partially articulated dog skeletons were found in a pit deliberately lined with bones from pigs and sheep/goats, possibly forming an elaborate burial rite. Cat bones are far rarer, and have only been identified on 15 per cent of sites. Bone from wild species is equally sparse, though a considerable quantity of red deer bones was recovered from the complex farmstead at Heslington East, suggesting that hunting was undertaken to some degree (Roskams and Neal 2012). Butchery marks found on a red deer scapula at the site indicated that it may have been from a shoulder of venison hung for smoking or salting. It is clear that antler, particularly from red deer, was widely utilised for tool manufacture. In particular, late third–early fifth century a.d. deposits of worked antler at Wattle Syke suggest that this was an important part of the economy of the site during this period (Martin et al. 2013).

15/09/2016 17:30:35

THE NORTH-EAST

% of assemblages

100

roadside selement/vicus (n=18)

279

farmstead (n=125)

90 80 70 60 50 40 30 20 10 0

fig. 7.43.  Presence of arable crops and other plant taxa from roadside settlements/vici and farmsteads in the North-East

Plant remains Of the 153 sites with recorded plant remains most are, unsurprisingly, biased towards the lowland areas of the region, being most common on the Lincolnshire Coversands and Clays (12 per cent), the Southern Magnesian Limestone (12 per cent), the Humberhead Levels (12 per cent), and in the Tees Lowlands (11 per cent). There are a disappointingly small number of samples located on the chalk of Yorkshire and Lincolnshire Wolds, and the alkaline soils here may have hindered preservation. The range and proportion of plant taxa identified from floral assemblages is highly consistent with other regions, with spelt wheat dominating, although barley is the best represented cereal taxa in terms of its overall presence (fig. 7.43). Barley, however, is rarely found in abundance within samples, exceptions including Site 20–4 (West) on the Asselby to Pannal pipeline, where a late Iron Age pit produced a large quantity of processed barley grain, mixed with a few oats. Only at Dragonby has barley been found in greater quantities than spelt wheat in samples across different phases of the site (May 1996). Emmer wheat is mostly identified on sites south of the Humber, though is also present at a few sites in the Tees Valley. When emmer wheat is identified in relative abundance it is generally from late Iron Age or early Roman samples, such as at Raymoth Lane, Worksop (Palmer-Brown and Munford 2004) and Stenigot Reservoir, Donington-onBain (Lindsey Archaeological Services 1997). In comparison, free-threshing wheat is widely distributed across the North-East, though it consistently features in low proportions. It only appears to have been the dominant cereal taxa in samples recovered from 8–10 High Street, Doncaster, where it was identified alongside rye and oat grains (Chadwick and Burgess 2008).

Romanch7.indd 279

As in other regions, oat remains are fairly ubiquitous finds in archaeobotanical samples (identified at 58 per cent of sites), but nearly always represent a very minor proportion of the assemblages. In most reports, the archaeobotanical specialist dismisses the finding of oat as a weed growing alongside other cereals. However, the recent excavation of the complex farmstead at Rossington Colliery, South Yorkshire, found oat remains alongside spelt and barley in abundant quantities, suggesting that it may have been deliberately cultivated and processed in the late Roman period, perhaps as a fodder crop (Roberts and Weston 2016). In contrast, rye is very poorly represented, found on only 8 per cent of sites, and appears to have been a genuinely marginal crop in the North-East, while pulses are clearly more common, present on nearly 20 per cent of sites. Broad beans are the common type of pulse identified, though lentils have been recovered from the Waterdale site at Doncaster (Davies 2013), and cultivated peas were recovered in some quantity from a late Roman ditch at Site 2 on the Asselby to Pannal pipeline (Gregory et al. 2013). Fruit remains have only been identified from 10 per cent of sites, reflecting the fact that they tend to require waterlogged conditions for preservation. Large quantities of hawthorn and elderberry at Dragonby perhaps reflect seasonal gathering, though the Waterdale site at Doncaster produced a range of fruit and nuts, including imported and comparatively rare foods, such as fig, grape, date, walnut and stone pine (Davies 2013). At Frenchgate (Site DG) in Doncaster, a secondcentury a.d. pit included 1400 apple seeds, estimated to have been from 150–300 apples, alongside 23 plum stones, and a few seeds of sloe, elderberry and grape. The high number of apple seeds was interpreted by the archaeobotanist as possible waste from cider production, though the

15/09/2016 17:30:37

280

THE RURAL SETTLEMENT OF ROMAN BRITAIN

feature was also considered to have been a cess pit (Williams 1986, 198). Although this is an outstanding example, the more frequent recovery of horticultural foods, such as fruits and pulses, appears to distinguish roadside settlements from farmsteads. Otherwise, the presence of most plant taxa is similar between the two types of site, with differences potentially lying in the scale of processing and consumption rather than the range of foods eaten. In general, there is little difference between the main settlement types in terms of the presence of different plant taxa. One of the few differences between enclosed and complex farmsteads is that a much greater proportion of the former have produced emmer wheat (42 per cent), and, as mentioned previously, this appears to have been more common on late Iron Age sites. Only six villa sites have records of plant remains, and their poor representation is further hampered by small samples. The exception is found at Ingleby Barwick where a wide range of plant taxa included freethreshing wheat, oats, flax and Celtic bean, while the weed seed assemblage suggested that the heavier clay soils were being tilled (Huntley 2013b). REGION Summary As we have seen throughout this chapter, the North-East was a diverse region with a range of landscapes that fostered quite different patterns of settlement and land use. These reflect localised, cultural traditions which persevered and developed from the Iron Age, and it may be tempting to associate this intra-regional diversity with the late Iron Age tribal groups that are thought to have held political power across the region – the Brigantes, the Corieltauvi, and the Parisi. Caution must be exercised here, however, since our knowledge of these groups is based upon sketchy historical records written from a Roman perspective (see Ch. 12 for discussion of ‘tribes’). We have very little idea as to how far the ‘everyday’ activities of the rural population were a reflection of their political identity, if at all. Instead, it may be more productive to look beyond the tribal history of the region and assess the diversity of the archaeological record in its own right with regard to farming activity, trade and exchange, and the adoption of ‘new’ building forms and material culture. The Roman conquest of this region was a gradual process, taking place over a number of decades in the second half of the first century a.d. (Bidwell and Hodgson 2009). Excavated evidence has shown that existing settlements were quickly swept aside in advance of road or fort construction. Yet, the impact of the Roman military on the wider landscape is less obvious. Many late Iron

Romanch7.indd 280

Age field systems appear to have been maintained into the second century, retaining their form and orientation, and it is not until the late Roman period that more significant changes occurred. The presence of the military appears to have affected patterns of farming, industry, and trade, over a long period of time. The establishment of fortresses and coloniae at Lincoln and York may have been responsible for stimulating the expansion of rural settlement in the Vale of York and Mowbray and the Lincolnshire Coversands and Clays during the early Roman period. However, this pattern is quite different in landscapes where Roman influence was minimal. In Holderness, settlement numbers declined after the late Iron Age, while in the Yorkshire Wolds the distinctive ladder settlements seen across the chalk appear to have continued from the late Iron Age into the Roman period, with little evidence for disruption or change (Stoertz 1997, 52–4). The increasing frequency of cattle bones over time may have reflected their economic importance to the towns, roadside and military settlements. Certainly, sites with higher frequencies of cattle tend to be located in landscape zones through which the major roads ran. In contrast, a greater emphasis on sheep and goats in the Yorkshire Wolds and on Holderness may be tied to more traditional forms of pastoral farming. A telling feature of the ladder settlements in the Wolds is their association with the local topography, as seen at Heslerton and Burton Fleming, with their long droveways and enclosure complexes, indicating that long-distance movement of livestock may have been important. The cattle bone evidence from Ferry Fryston is also highly suggestive of seasonal transhumance of people and animals, while the lack of Roman material culture implies the presence of communities who were not fully engaged with Roman systems of trade and exchange. The lack of economic integration between settlements on the road network and outlying farmsteads is partly borne out in the patterns of material culture consumption. That roadside settlements were centres where agricultural surplus could be gathered, processed, and redistributed, is indicated by the presence of corndryers and millstones, while textile equipment is noticeably more common than on other types of site, perhaps pointing to the commercial importance of wool. The presence of mules on sites on the road network is also significant for the transportation of goods and produce, while imported and locally cultivated, horticultural plant foods indicate that the diets of some inhabitants of roadside settlements were relatively diverse. Some farmsteads and villas also clearly benefited from

15/09/2016 17:30:37

THE NORTH-EAST

the Roman economic system, such as at Faverdale and Heslington East, where higher numbers of coins and other finds have been noted. In addition, some imported pottery types were more widely adopted than others on farmsteads, which perhaps suggests that new forms of food preparation and dining filtered into the countryside, but on a limited and varied scale. Major financial investment can be seen in the architecture of some of the larger villas, such as at Scampton and Beadlam, while more recent excavations have improved our understanding of the landscape context of some villas. Investigations at Ingleby Barwick and Welton Wold have revealed evidence for wider ‘estates’ in which these villas may have operated.

Romanch7.indd 281

281

Overall, the North-East is a region that lies on the edge of the ‘core zone’ of Roman Britain, connecting the wealthy, arable-based communities of the Central Belt to the military-controlled, uplands of the north. The establishment of roads, towns and forts brought with it new opportunities for the local population. Yet the region has a distinctive character, with forms of settlement not seen elsewhere in Britain, and there is a strong level of continuity from the late Iron Age. We get a sense of conservatism prevailing in the countryside, though the adoption of new Roman fashions varied considerably between different settlements; this is an aspect that will be examined further in volumes 2 and 3 of this series.

15/09/2016 17:30:37

Chapter 8

THE CENTRAL WEST By Tom Brindle

The area defined by the project as the Central West extends over 18,364 km² and incorporates elements of several modern administrative counties. The region includes the greater parts of Herefordshire, Shropshire, Cheshire, Merseyside, Derbyshire, West Yorkshire and approximately half of South Yorkshire, although the only counties to be included in their entirety are the metropolitan counties of Greater Manchester and the West Midlands. In the west the region extends into Wales to include small parts of Flintshire, Wrexham, Powys and Monmouthshire, and in the north very small sections of Lancashire and North Yorkshire. The northern extents of Worcestershire and Warwickshire fall within the south of the region, and in the east the region includes a very small section of Nottinghamshire (fig. 8.1).

fig.

THE NATURE OF THE LANDSCAPE The region contains considerable variation in terms of topography, geology and land use. Topographically, there are low-lying areas, principally in the north-west, in parts of Shropshire, Cheshire, Greater Manchester and Merseyside; plateaus, particularly in the south-west, and upland terrain, primarily in the north-east. However, the superficial division of the region into lowland, plateau and upland does injustice to the immense variation within each of these landscapes, even between areas that appear similar topographically, as shown by the large number of distinctive landscape zones within the region (fig. 8.2). While much of the region’s land use is rural and agricultural, there are some extensively

8.1.  The Central West region in relation to modern county boundaries

282

Romanch8.indd 282

07/09/2016 16:24:03

THE CENTRAL WEST

fig.

8.2.   Constituent landscape zones of the Central West region

developed areas, and the region includes some of the largest cities and urban conurbations in England and Wales outside London, including the West Midlands conurbation, Greater Manchester, Liverpool, Leeds, Sheffield, Bradford, Coventry and Stoke-on-Trent. Soils within the region are primarily acidic, resulting in poor survival of environmental remains, particularly bone. THE CENTRAL WEST DATASET The project database contains 200 records of excavations relating to 183 individual sites in the Central West region (the disparity reflecting a small number of sites that are represented by multiple records). These sites are mapped in fig. 8.3 (top), yet it is crucially important to be aware of the limitations of the excavated data, particularly with regard to their geographical distribution. There appears to be a strong tendency for sites to focus on the low-lying terrain of the region, while the upland areas of Herefordshire and the Pennines are starkly under-represented. This undoubtedly in large part reflects a lowland focus for excavations rather than the genuine ancient settlement pattern,

Romanch8.indd 283

283

as indicated by the distribution of excavations of all periods recorded by the National Monuments Record (fig. 8.3). Indeed, Taylor (2007, 14) has mapped a large number of (mainly unexcavated and undated) earthworks from the southern Pennines, and a large number of earthwork sites from the Peak District have been subject to survey and surface artefact collection, with many being assigned Iron Age to Roman dates (cf. Makepeace 1998). The apparent focus on the lowland terrain may therefore to a certain extent be illusory. While settlements are widespread across the lowlands, there are some notable clusters within these areas: for instance, a number of sites are situated in the vicinity of the civitas capital at Wroxeter, Shropshire (Viroconium), and also in close proximity to the walled ‘small town’ at Wall, Staffordshire. While clusters in these areas may reflect increased settlement density in the hinterlands of these centres, they are perhaps more likely to represent areas in which development has focused. A number of the sites near Wroxeter reflect development-led work associated with the A5/A49 Shrewsbury bypass (Ellis et al. 1994) as well as the research-focused work of the Wroxeter

07/09/2016 16:24:10

284

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 8.3.  Distribution of excavated Roman rural sites (n=183) and all excavation records (1910–2010) from National Monument Records (NMR) Index (n=4507) in the Central West region

Romanch8.indd 284

07/09/2016 16:24:24

THE CENTRAL WEST

Hinterland Project (Gaffney et al. 2007). Several of the sites near Wall were excavated as a result of road development, notably during the construction of the M6 Toll (Powell et al. 2008b), and in Merseyside there is a cluster of sites, several of which were also excavated ahead of road improvement (Cowell and Philpott 2000). What is notable, however, is that although the region encompasses some of the largest cities in the country outside London, the vast amounts of development associated with these urban centres have contributed very few new Romano-British rural sites (fig. 8.4). This clearly does not reflect a lack of excavation in these areas, as the distribution of excavations recorded by the NMR illustrates (fig. 8.3). It is somewhat unclear whether the almost total absence of rural sites from the builtup areas of the region reflects genuinely low-level settlement density, or whether such sites, which are rarely represented by large numbers of finds and which may already have been destroyed or damaged during earlier episodes of development, often go unrecognised during small-scale excavations in built-up areas. Nevertheless, it is perhaps instructive that large infrastructure

fig.

Romanch8.indd 285

285

projects, such as the construction of the M6 Toll, have also produced relatively few Romano-British sites compared with what might be expected in some other regions (Powell et al. 2008b; British Archaeology 2002, 5). In particular, a watching brief associated with a continuous 68 km long gas pipeline in the region, running north-west to south-east through parts of Cheshire, Shropshire and Staffordshire, resulted in the addition of just two sites to the project database (one a single cremation, the other a small group of pits; Taylor 1999). With topsoil stripping of an easement approximately 30 m wide, this represents an area of over 200 ha, and the dearth of late Iron Age and Roman-period sites recognised during this work is noteworthy, suggesting that at least parts of the landscape saw comparatively low levels of occupation during the Roman period. In order to explore intra-regional diversity within the settlement pattern, throughout the course of this chapter the region’s sites have been divided into six spatial groups. Compared with some regions, the number of excavated sites in the Central West is relatively low, and it has not been possible to group them meaningfully by individual

8.4.   Distribution of excavated Roman rural sites in relation to built-up areas

07/09/2016 16:24:32

286

fig.

THE RURAL SETTLEMENT OF ROMAN BRITAIN

8.5.   Distribution of sites in the Central West by six sub-regions

table

8.1: number

and density of sites in the central west arranged by sub-regions

Landscape zone

Area (km²)

No. of sites

Coal Measures

2283

32

0.014

Midlands Plateau/South Derbyshire Claylands

3852

44

0.011

Central Herefordshire

1512

17

0.011

Mosses and Meres/Shropshire Hills/Wrexham

4452

44

0.010

Mersey Basin/Flintshire

2327

20

0.009

Southern Pennines/Pennine Fringe

3927

26

0.007

Total

18351 183

Natural England landscape zones. Such an approach is made more problematic by the extension of the region into parts of Wales, where there is no equivalent to the Natural England areas (see Ch. 1). For these reasons six new zones have been established, based in the main upon the existing Natural England zones, but adding adjacent parts of Wales where applicable, and merging Natural England areas where there would be too few sites to consider individually (fig. 8.5; table 8.1). The dividing lines between these six zones do not always represent firm, clearly distinguishable physical boundaries, and the

Romanch8.indd 286

Density of sites per km²

0.010

borders are to a certain extent arbitrary, yet they provide a pragmatic means of assessing the variability in the settlement pattern within the Central West region. ROMAN RURAL SETTLEMENT PATTERNS There are two major Roman-period centres in the Central West region, Wroxeter (Viroconium Cornoviorum) and Chester (Deva). Both sites had Roman military origins, established as legionary

07/09/2016 16:24:40

THE CENTRAL WEST

table

8.2: major

287

site types in the central west region displayed by sub-regions

Landscape zone Farmstead Villa Roadside Vicus Hillfort settlement Midlands Plateau/South Derbyshire Claylands

24

Coal Measures

21 0 0 2 0

Mosses and Meres/Shropshire Hills/ Wrexham

20

7

4

2

1

15

Southern Pennines/Pennine Fringe

12

1

1

1

1

10

Central Herefordshire

10 3 1 0 1

Mersey Basin/Flintshire Total

4

0

0

9 2 2 1 1 96 14 12 6

fortresses, the former during the a.d. 50s, and the latter during the 70s. However, whereas Chester was to remain a permanent legionary base with an accompanying and extensive canabae, occupied until at least the late fourth century a.d., the legionary fortress at Wroxeter was abandoned in around a.d. 90, after which it was redeveloped and became the civitas capital of the Cornovii. Away from these major centres, there were, in addition, a number of nucleated sites of the type usually defined as ‘small towns’, situated along major Roman roads. These include the ‘walled towns’ of Kenchester (Magnis), Alcester (Alavna), Wall (Letocetum), Water Eaton (Pennocrucium), Whitchurch (Mediolanum), Little Chester (Derventio) and Rocester. These ‘walled towns’ fall outside the remit of this study, although it is worth noting that they have a southern focus and do not seem to have been a feature of the northern half of the Central West. Besides the walled towns, however, there were also a number of large undefended roadside settlements, which have a wider distribution, and several of these are represented by records on the project database. As well as the long-lived fortress at Chester, several auxiliary forts and marching camps are known, and the Roman army maintained a strong presence, predominantly in the north of the region, particularly during the later first and second centuries a.d. Most forts had associated civilian settlements, vici, and in the north these seem to have taken the place of the walled ‘small towns’ of the south. Excavated examples on the database include those associated with the auxiliary forts at Middlewich, Northwich, Manchester, Castleshaw, Castleford and perhaps also Adel. The example at Castleford is best seen as a vicus associated with a north–south network of forts extending outside the border of the region, into the North-East, including Newton Kyme and Doncaster, perhaps all established at around the same time towards the end of the first century a.d.

Romanch8.indd 287

1

Other site

4

10 9

2 5 51

(see Ch. 7). In addition to the excavated examples on the database, limited interventions have also been undertaken in vici associated with forts at Ilkley, Slack, Templeborough, Chesterfield and Brough-on-Noe, although these are not well understood (Roberts pers. comm.). Unlike Chester, most forts were abandoned during the second century a.d., although a small number maintained garrisons into the late Roman period, and Manchester, for instance, witnessed activity into the fourth century (Bidwell and Hodgson 2009, 97; Jones 1974). Of the 183 sites broadly classified as being of ‘rural’ character and recorded on the project database, 132 have been identified as domestic settlements of some description, the remainder comprising industrial sites, field systems or areas of funerary activity. table 8.2 presents broad classifications of the settlements by the six subregions, and it is clear that farmsteads are overwhelmingly the most common type of domestic site across the region as a whole, accounting for 73 per cent of domestic sites. Nucleated sites on the database are confined to undefended roadside settlements (9 per cent of settlements) and military vici (5 per cent), with villages absent, although as noted above ‘walled towns’ are a common feature of the archaeology of the south of the region. Villas occur, although they are few in number (fourteen excavated examples – 11 per cent of settlements), and these are unevenly distributed, being clearly focused on the Mosses and Meres/Shropshire Hills/Wrexham sub-region, with an emphasis on the south which is, broadly speaking, similar to the distribution of walled ‘small towns’. Of the few sites broadly defined as villas in the north of the region, those at Loushers Lane, Warrington, Cheshire (Rogers and Garner 2007; Hinchliffe et al. 1992), and Pentre Farm, Flint, Flintshire, are very atypical, probably functioning as official buildings associated with local administration rather than rural villas. Indeed, Pentre Farm has

07/09/2016 16:24:40

288

THE RURAL SETTLEMENT OF ROMAN BRITAIN

long been recognised as being in some way involved with the control of lead production (O’Leary 1989), and recent excavations (completed after data collection was finished) have indicated the presence of an extensive roadside industrial settlement immediately to the south-east of the ‘villa’ (Chris Martin pers. comm.), suggesting that the building was part of or closely associated with this nucleated site. The remaining element of the settlement pattern consists of a small number of hillforts with Iron Age origins that continued to be occupied in some way during the late Iron Age/ Roman period within the region. REGIONAL CHRONOLOGY As farmsteads account for the large majority of settlements recorded from the Central West it is inevitable that any attempt to display rural settlement chronology in graphical form will be biased towards these sites, and for this reason the chronological pattern shown by each of the main settlement types are discussed here independently. However, as for the most part villas are likely to have been involved in some way with agricultural production (though perhaps not the site at Pentre Farm, Flint), and as in some cases villa buildings represented late developments at sites that would otherwise be considered farmsteads, the region’s fourteen villas have been included as ‘farming settlements’ here. As in some other regions, the dating of Iron Age and Roman period rural sites in the Central West is often hindered by the low dating resolution afforded by pottery (cf. Ch. 1). Much of the region appears to have been largely aceramic during the late Iron Age (Philpott and Adams 2010, 179–80; Gaffney et al. 2007, 223, 280), and in areas where pottery was in use ceramic traditions often continued unchanged into the early Roman

period, making it difficult to distinguish between sites occupied during the late Iron Age and those established during the Roman period (Roberts pers. comm.). The chronological patterns presented below must therefore be regarded with a certain degree of caution. figure 8.6 presents the chronology of farming settlements, showing the number of settlements in use over time, suggesting a rise in the number of sites occupied during the late Iron Age, with this increase continuing until the second half of the second century a.d. This is followed by a slight decline in the number of settlements occupied at around the turn of the third century, although the numbers then remain relatively stable until the end of the third/early fourth century, when there is a decline that continues until the end of the Roman period. This broad regional pattern masks important intra-regional trends, however, and there are distinct differences between some of the subregions. To illustrate this, fig. 8.7 presents the chronological data from two sub-regions with very different chronological profiles, the Mosses and Meres/Shropshire Hills/Wrexham area, and, on the opposite side of the Pennines, the Coal Measures. Both regions are represented by a very small number of farmsteads of mid- to late Iron Age date, yet whereas the number of farmsteads in use during the late Iron Age and first century a.d. increased in both areas, the Coal Measures appears to have witnessed much more considerable settlement expansion at this time. In both areas the number of settlements continued to rise until the mid- to late second century, when broadly similar numbers of settlements were occupied in both sub-regions. The early third century saw a decline in the number of settlements in both areas, yet this reduction appears to have been more acute

no. of selements (n=110)

90 80 70 60 50 40 30 20 10 0

fig.

Romanch8.indd 288

8.6.   Number of farming settlements in use over time in the Central West region

07/09/2016 16:24:42

THE CENTRAL WEST

in the Coal Measures, heralding a period of marked settlement abandonment that continued unchecked until the end of the Roman period (a phenomenon previously noted by Hodgson 2012, 53). In contrast, the number of settlements in use in the Mosses and Meres/Shropshire Hills/ Wrexham area fluctuated slightly during the course of the third century, but there does not appear to have been a major episode of settlement decline until the mid-fourth century. Although the second century a.d. may be regarded as the period of peak activity across the region, the differences between the two subregions discussed above are further emphasised if we consider the evidence for the dates at which settlements apparently emerged, rather than just noting the broad number of settlements in use over time (fig. 8.8). Considerably more settlements in the Coal Measures produced evidence for an Iron Age (or earlier) origin, and there appears to have been a particular episode of settlement

20

Coal Measures (n=21)

289

expansion during the late Iron Age in this zone. Many of these sites continued to be occupied throughout the early Roman period, and as the number of settlements continued gradually to increase, this resulted in a landscape that was at its most densely populated (in terms of the number of settlements) during the second century. Although some farmsteads in the Mosses and Meres/Shropshire Hills/Wrexham area were established during the Iron Age, the period in which most farmsteads emerged seems to have been considerably later than in the Coal Measures, and the second century appears to have been the period of major settlement expansion in this area. The reasons for the striking chronological differences between sites in these two parts of the Central West are uncertain, and it is at present unclear whether the variance reflects genuine differences in the settlement patterns of the two areas or an artificial phenomenon caused by the aforementioned sub-regional variation in the use

Mosses and Meres/Shropshire Hills/Wrexham (n=27)

no. of selements in use

18 16 14 12 10 8 6 4 2 0

fig.

8.7.   Number of farming settlements in use over time in selected sub-regions of the Central West

Coal measures (n=21)

Mosses and Meres/Shropshire Hills/Wrexham (n=27)

16 14

no. of sites

12 10 8 6 4 2 0 Late Iron Age

Late 1stC AD

2ndC AD

3rdC AD

4thC AD

fig. 8.8.   Establishment dates for farming settlements in the Coal Measures and the Mosses and Meres/Shropshire Hills/Wrexham area

Romanch8.indd 289

07/09/2016 16:24:44

290

THE RURAL SETTLEMENT OF ROMAN BRITAIN

of ceramics, and the subsequent effect this has had on the dating of sites. Taking the evidence at face value, however, it would appear that there were distinct geographical differences in settlement chronology, even within these two sub-regions. There is considerable chronological variation between the other major classes of settlement within the region – the military vici, roadside settlements and hillforts. As vici are civilian settlements associated with military sites, they typically emerged shortly after the foundation of the forts to which they were attached, usually from around the final quarter of the first century a.d. Some vici apparently went out of use at broadly the same time as activity at the forts ceased, as at Northwich (Jones 1972) and Castleshaw (Redhead 1999), although in some cases the civilian settlement appears to have extended beyond the period of military occupation, as, for example, at Middlewich (Garner and Reid 2012; Williams and Reid 2008). At Castleford the fort and vicus were redeveloped during the second half of the third century a.d., becoming a defended civilian settlement that continued until the end of the fourth century a.d. (Rush et al. 2000; Abramson et al. 1999; Cool and Philo 1998). The sites defined as roadside settlements form an amorphous group, comprising sites of varying character. Some, like military vici, developed adjacent to or in close proximity to forts, but these are differentiated from vici as the civilian settlements emerged after the forts were abandoned, as at Greensforge (A. Jones 1999; Webster 1981). In some cases, however, the relationship between the civilian settlements and the military activity is uncertain, for example at Children’s Hospital, North Street, Derby (Higgins 1999). In other instances civilian roadside settlements developed out of other military sites, such as those interpreted as possible supply depots at Holditch (Rogers and Garner 2007; Charlton 1962), Wilderspool (Rogers and Garner 2007; Hinchliffe et al. 1992) and Redhill, Telford (Mann 2011; Browne and Boon 2004). Occasionally early military phases or military associations are indicated but ill understood, as at Wigan, where the recent discovery of a bathhouse (Miller and Aldridge 2011) has confirmed the presence of early Roman military activity, as also suggested by earlier interventions (Jones and Price 1985). Other nucleated settlements take the form of ‘ribbon’ developments along roads, sometimes not far from other large settlements, for example at Heath Road (Hannaford and Mason 1991), 2 km south of the walled town at Whitchurch. Despite their varying character and mixed origins, the available dating evidence suggests that the region’s roadside settlements emerged during broadly the

Romanch8.indd 290

same period. None has produced evidence for Iron Age origins, and all appear to have been established during the later first or early second centuries a.d. In most cases the settlements appear to have been relatively long lived, although several appear not to have continued beyond the end of the third century a.d. Others, including Meole Brace (Ellis et al. 1994; Hannaford and Philpotts 1994), Redhill, Telford (Mann 2011; Browne and Boon 2004), Heronbridge (Mason 2004) and Greensforge (A. Jones 1999; Webster 1981), have produced evidence for fourth-century occupation, although the precise nature of the activity is not always clear. None of the region’s roadside settlements have produced evidence for continuity beyond the Roman period. The site at Meols (Griffiths et al. 2007), on the north coast of the Wirral peninsula, represents something of an anomaly as it is so very poorly understood. The site is known principally through chance finds, but it was evidently an important centre throughout the late Iron Age and Roman periods. The rich range of artefacts recovered, which include objects such as third-century b.c. Carthaginian coins and other exotic artefacts, suggest that the site was an important and long-lived harbour serving an Atlantic trade route that linked western Britain to the Mediterranean via Brittany (Philpott and Adams 2010, 177). The final class of rural settlement within the region are hillforts, of which there are four examples on the database. In all cases the late Iron Age/ Romano-British activity recorded represents continuity of (or renewed) occupation at sites that originated as hillforts during the mid-Iron Age, although the character of occupation is usually uncertain, and the length of duration of the continuity into the late Iron Age and Roman period varies. At Midsummer Hill, Herefordshire (Stanford 1981), and the Berth, Shropshire (Morris and Gelling 1991), for example, activity appears to have ceased during the first century a.d. At Castle Steads, Bury (Fletcher 2005), some sort of activity apparently continued into the second century a.d., whereas at Old Vicarage, Mellor, Stockport, occupation, possibly of high status, may have continued until the fourth century (Noble et al. 2008). Further examples saw excavation in the early or mid-twentieth century, and as a consequence the available information is sometimes limited. The multivallate site at Nescliffe, Shropshire (Hume and Jones 1960) represents a further Iron Age hillfort with poorly understood evidence for Roman-period occupation, and a large hoard of early fourth-century coins was also recently recovered from the vicinity of the site (Wigley pers. comm.). At Bury Walls, Shropshire, excavations in the 1930s resulted in the discovery

07/09/2016 16:24:44

THE CENTRAL WEST

of a masonry building of suspected Roman date within the interior of the hillfort (Morris 1932), and this has been suggested as a possible late Roman temple (Murdie et al. 2003). Almondbury Hillfort, near Huddersfield, has been regarded by some as a likely stronghold of Cartimandua and as the Camulodunum recorded in the Ravenna Cosmography (Faull 1981, 116; Varley 1976; Challis and Harding 1975, 120), although evidence of occupation in the later Iron Age has not been confirmed by excavation (Varley 1976).

291

RURAL SETTLEMENT FORM The form of rural settlements in the Central West has previously been characterised by Taylor (in his sections on the West Midlands and the NorthWest), with enclosed farmsteads being recognised as the dominant settlement type, principally identified through aerial photography (Taylor 2007, 43, 47–8). Of the 110 excavated farmsteads and villas recorded on the project database for the region, 40 sites (36 per cent) had plans that were

Whitwood Common, West Yorks

Methley Site 1, West Yorks

Legh Oaks Farm, High Legh, Cheshire

Bulcliffe Wood, Netherton, West Yorks

0

100 m 1:2000

fig. 8.9.  Plans of enclosed farmsteads of rectilinear (Methley Site 1; Roberts and Richardson 2002), irregular (Whitwood Common; Burgess and Roberts 2004), D-shaped (Bulcliffe Wood, Netherton; Grassam and Roberts 2008) and curvilinear (Legh Oaks Farm, High Legh; Nevell 2002) form

Romanch8.indd 291

07/09/2016 16:24:44

292

THE RURAL SETTLEMENT OF ROMAN BRITAIN

sufficiently detailed and extensive to allow confident classification by form, and, of these, the large majority (85 per cent) were small, dispersed, settlements of enclosed form. There were, in addition, a further twelve potential enclosed farmsteads which had insufficiently clear or complete plans for such a classification to be made with confidence. Although enclosed settlements dominate the settlement pattern of the region, there is considerable variation among them, with a range of forms represented. Enclosures of rectilinear form are most common, accounting for 56 per cent of those assigned a form, while enclosures of irregular (28 per cent), D-shaped (11 per cent) and curvilinear (6 per cent) shape also occur (fig. 8.9). Most enclosures appear to have been of single-ditched construction, although substantial double-ditched enclosures, as for example, at Sharpstones Hill Site E, Shropshire (fig. 8.10; Bain and Evans 2011; Barker et al. 1991), occur in small numbers (eight in total), with an apparent emphasis on the west of the region. The relatively small number of well-understood sites prevents a confident understanding of the intra-regional and chronological patterns for the form of enclosures, and the pattern within the region was complex. Philpott (2000a, 183–4), focusing on the Cheshire/Merseyside area, suggested that double-ditched curvilinear enclosures may be Iron Age and those of singleditched, rectilinear form of Roman date, highlighting a contrast with the pattern in Shropshire, where rectilinear enclosures sometimes originated in the late Iron Age. The limited sample of excavated sites makes the identification of firm patterns difficult, and there are insufficient numbers for statistics to be of value, although, at least in the west of the region, the excavated evidence broadly supports Philpott’s suggestion. The rectilinear enclosure at Ochre Brook, in Merseyside (Cowell 2009; Philpott 2000b), for example, appears to have been a second-century establishment, whereas in Shropshire the rectilinear settlements at Hay Farm, Eardington (Hunn 2000) and Sharpstones Hill Site E (Bain and Evans 2011; Barker et al. 1991) had origins in the late Iron Age, continuing on into the Roman period. East of the Pennines, in the Coal Measures, many of the rectilinear enclosures seem to have had Iron Age origins, and some did not continue long into the Roman period, yet others, such as Bullerthorpe Lane (Roberts et al. 2001) may only have been established in the second century a.d. Many irregularly shaped enclosures (seemingly more common in the east) also appear to have been of a range of dates; some seem to have had Iron Age origins, as at Whitwood Common

Romanch8.indd 292

0

100 m 1:2000

fig. 8.10.   Plan of farmstead at Sharpstones Hill Site E, Shropshire (Bain and Evans 2011)

(Burgess and Roberts 2004), while others were of Roman date, as at Killingbeck Hospital, Leeds (Timms 2012). Although Philpott (2000a, 183) has observed a late prehistoric date for double-ditched curvilinear enclosures in Cheshire and Merseyside, most of the region’s sites with double-ditched enclosures were of rectilinear form and appear predominantly to have been of Roman date. The number of ditches alone is therefore a poor guide to settlement chronology. Indeed, in the Coal Measures it has been suggested that double-ditched enclosures may typically have been products of two phases: an initial large-ditched enclosure (often of late Iron Age date), subsequently redefined when it was integrated into a later field system complex (Roberts pers. comm.). In this part of the region few double-ditched enclosures are thought to have been created with both circuits as primary features, although a second/third century example from Low Common, Whitwood, is one such example (Burgess and Roberts 2004). The variable and complex range of enclosures across the region makes assigning a date to sites based upon their morphology alone extremely problematic, and it would certainly be dangerous to extrapolate from one part of the region to another. Rural settlements of complex form make up a very minor component of the settlement pattern within the region, with only four (10 per cent) of the classified sites assigned to this morphological group with any degree of confidence, although an additional nine unclassified sites have been regarded as potential examples. There are too few

07/09/2016 16:24:45

THE CENTRAL WEST

293

0

100 m 1:2000

Mid-Iron Age Romano-British

fig. 8.11.   Site plan of a mid-second to mid-third century a.d. complex farmstead, developing from a mid-Iron Age open and subsequently enclosed settlement, north of Langley Mill, West Midlands (Powell and Ritchie 2008)

certain examples to identify any meaningful patterns within their geographical distribution. While complex farmsteads never approached enclosed sites in terms of their frequency within the region, their number appears to have increased from the second century a.d. Again, these sites do not represent a homogeneous group; there are considerable differences between them in terms of their precise form and chronology, and they all underwent considerable change and development over time. Some had Iron Age origins; the site north of Langley Mill (fig. 8.11), appears to have experienced a complex sequence of development, emerging first as an open, then subsequently enclosed settlement in the mid-Iron Age, followed by contraction during the late Iron Age and early Roman periods, and then by extensive reorganisation and reoccupation between the midsecond and mid-third centuries, when a series of enclosures, some with structures, were linked to a wider system of fields and other boundaries (Powell and Ritchie 2008).

Romanch8.indd 293

At Loushers Lane (Rogers and Garner 2007; Hinchliffe et al. 1992), a late Iron Age enclosed farmstead was remodelled in the late first century a.d. when a series of enclosures were established containing a range of building types, all associated with a lane leading to the newly constructed villa, and the site continued to see further reorganisation over time. Other complex sites produced little evidence for Iron Age activity, and that at East of Birmingham Road Nurseries, Shenstone (Simmonds 2008) appears to have been established in the mid-second century a.d., although it also produced evidence for a phase of major reorganisation, this time in the mid-third century, with ditches recut and the establishment of new enclosures. Only two open settlements have been recorded from the region, although this low number must certainly result from the lack of major features such as enclosure ditches, which means that such sites are very difficult to identify. The two known examples are quite different. Court Farm, Halewood, Knowsley (Adams 1997), in

07/09/2016 16:24:46

294

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Early second century A.D.

Late second century A.D.

0

Early/mid third century A.D.

100 m 1:2000

fig. 8.12.  Sequence of development at Bullerthorpe Lane, Swillington, where a third-century building and stock enclosure replaced an earlier enclosed settlement (Roberts et al. 2001)

Merseyside, was represented by a large group of sub-rectangular buildings of mid- to late Roman date, and there is no evidence that it was ever enclosed. In contrast, Bullerthorpe Lane, Swillington, West Yorkshire (Roberts et al. 2001), had an unenclosed third-century phase, when it was represented by a single building and a stock enclosure, which may have replaced an earlier farmstead of enclosed form (fig. 8.12). BUILDINGS The evidence for buildings in the Central West region amounts to 302 individual structures recorded from 72 sites (table 8.3). Taylor (2007, 31) has previously noted the presence of rectilinear buildings at rural sites in the region, and these stand out as being by far the most common type of building within the region with 202 examples excavated (c. 80 per cent of those recorded). However, rectangular buildings are very unevenly distributed across the different classes of site, the majority (61 per cent) being recovered from roadside settlements and military vici. This largely reflects the large size of these nucleated sites, and a correspondingly high number of buildings from each individual settlement. Excavations at the roadside settlement at Wilderspool, Warrington, for example, recorded a series of 36 rectangular buildings, although not all of these were contemporary (fig. 8.13; Rogers and Garner 2007). Circular buildings are very uncommon at

Romanch8.indd 294

nucleated settlements, comprising just 3 per cent of all buildings associated with these types of site; the only roadside settlement with buildings of circular form is Longdales Road, King’s Norton, Birmingham, and the only vicus is Middlewich, where possible pre-Roman roundhouses and an oval building of Antonine date were identified. Outside of the nucleated settlements circular buildings are far more common, accounting for 41 per cent of the buildings at farming settlements and hillforts (although of the region’s hillforts only Old Vicarage, Mellor, produced good evidence for structures). However, this pattern is skewed by the inclusion of villas in the farming settlements, and these sites are unsurprisingly far better represented by rectangular architecture than by circular buildings. At farmsteads without villa architecture circular buildings do outnumber those of rectangular form, yet, while they form the greatest proportion of buildings from farmsteads, they do not completely dominate. In terms of chronology the regional pattern follows the provincial pattern closely, with a reduction in the frequency of sites with circular buildings over time, as rectangular forms become increasingly common. The most striking period of decline in the use of circular architecture appears to be during the late Iron Age to Roman transition, where there is a drop from 43 per cent of sites with circular buildings to 23 per cent (fig. 8.14). By the first century a.d. sites producing rectangular buildings slightly outnumber those with circular buildings, although the balance is relatively equal.

07/09/2016 16:24:46

295

ur se

of Ri ve r

THE CENTRAL WEST

d te ca n u Tr

by

co er rm fo

ted

nca Tru

by

ip r Sh

al

Can

te

hes

nc Ma

Roman Road Projected building

0

100 m 1:2000

fig. 8.13.   Plan of buildings at Wilderspool, Warrington, during the period c. a.d. 250–270 (Rogers and Garner 2007)

table

8.3: number

Site type

of buildings from different site types in the central west region

Total no. buildings

No. circular buildings

No. rectangular buildings

Farmstead 84 46 Farmstead/villa 48 Roadside settlement

99

10 3

38 38 96

Hillfort 5 5 0 Vicus Other Total

48 1 47 18 1 17 302 66 236

However, by the second century many more sites with evidence for buildings produce rectangular buildings, with circular buildings remaining relatively uncommon from then onwards. This broad chronological pattern of course masks differences between the different classes of settlement, and to a certain extent reflects an increase over time of those sites that produce rectangular buildings, namely roadside settlements, military vici and villas. However, if we look at

Romanch8.indd 295

farmsteads exclusively, the same temporal trend is clearly evident, the principal difference being that circular architecture remains important longer (until the second century a.d.) than the overall regional pattern suggests (fig. 8.15). The materials used in the construction of buildings vary across the different classes of site. At farmsteads, buildings were overwhelmingly of timber (or mass-walled; see Ch. 3) construction, only 14 per cent of structures incorporating stone.

07/09/2016 16:24:47

296

THE RURAL SETTLEMENT OF ROMAN BRITAIN circular

% of settlements with buildings

50

rectangular

45 40 35 30 25 20 15 10 5 0 Late Iron Age (n=37)

fig.

Late 1stC AD (n=56)

2ndC AD (n=83) 3rdC AD (n=86) 4thC AD (n=68)

8.14.   Use of circular and rectangular buildings in the Central West over time

60

circular

rectangular

% of farmsteads

50 40 30 20 10 0 Late Iron Age Late 1stC AD (n=37) (n=56) fig.

3rdC AD (n=86)

4thC AD (n=68)

8.15.   Use of circular and rectangular buildings on farmsteads in the Central West over time

The few farmsteads with stone buildings (ten examples) are widely dispersed across the region, with examples in Clwyd Powys, Shropshire, Herefordshire and Warwickshire, although there is a small cluster in the Derbyshire/South Yorkshire area. Where masonry structures do occur at farmsteads they are usually rectangular and of mortared construction, sometimes with evidence for tiled roofs and occasionally for wall plaster, as at Ripley, Derbyshire (Palfreyman and Ebbins 2012). At Wattscliffe, Harthill Moor, Derbyshire, the only farmstead in the region to have produced stone-built circular houses, somewhat different construction techniques were used, and the walls of one house were constructed using large gritstone boulders (Makepeace 1998). Timber was also the most commonly used material for buildings at nucleated settlements, although most of these settlements also included some examples of masonry buildings, these being revealed in large numbers at certain sites (e.g. Holditch and Heronbridge). In contrast to the buildings from farmsteads, which are usually interpreted as

Romanch8.indd 296

2ndC AD (n=83)

domestic houses, buildings at nucleated sites have a wide range of functions and are frequently suggested as workshops, shops and taverns. Villas are the only class of site within the region where masonry-built structures outnumber those of timber, although this is scarcely surprising given that a well-appointed masonry-built structure is a typical characteristic of this settlement type. Yet sites identified as villas do also display a range of buildings, both in terms of form and in the material used. All of the region’s villas display complex sequences of development, and the provision of a full villa building often only occurs after several generations of activity during which settlement focuses around other types of building. At Loushers Lane, Warrington, the construction of the villa building in the late first century a.d. was associated with the remodelling of a previously occupied farmstead (Rogers and Garner 2007; Hinchliffe et al. 1992). The villa at Acton Trussell, Staffordshire (Penk Valley Archaeological Group 1997; 2004), appears to have gone through various transformations, initially developing from late Iron

07/09/2016 16:24:50

THE CENTRAL WEST

297

0

20 m 1:500

fig. 8.16.  Plan of building at Pentre Farm, Flint, during the late second/early third century associated with a recently discovered roadside settlement (O’Leary 1989)

0

a.d.,

probably

20 m 1:500

fig.

8.17.   Plan of a corridor villa with a detached bathhouse at Hales, Staffordshire (Goodyear 1974)

Age roundhouses, incorporating timber-built rectangular buildings in the late first century a.d., with stone buildings added in the second century. Further extensions and the addition of a bathhouse occurred during the second to third century, and the villa experienced a complete rebuild during the mid-fourth century. In terms of our understanding of the development of their form, several of the villas from the region are too poorly understood to provide much information about their plan. Based on the eight examples that could be classified by villa form, almost all appear to have been of relatively modest scale. Three were cottage buildings (e.g. Pentre Farm, Flint: O’Leary 1989; fig. 8.16), two were corridor (e.g. Hales, Staffordshire: Goodyear 1974; fig. 8.17) and three were winged-corridor types (e.g. Eaton-by-Tarporley, Cheshire: Mason 1983; fig. 8.18). Although the evidence is very

Romanch8.indd 297

0

20 m 1:500

fig. 8.18. Plan of a winged corridor villa at Eaton-byTarporley, Cheshire, during its final phase (Mason 1983)

07/09/2016 16:24:51

298

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fragmentary, the villa at Loushers Lane, Wilderspool, is regarded as being of courtyard form in its final phase (Rogers and Garner 2007, 42), which would make it the most elaborate of the villas from the region. The small numbers prevent any meaningful assessment of the distribution of these different classes of villa within the region. Compared with some of the most luxurious courtyard and palatial villas known from elsewhere within the province (particularly in the South and the Central Belt), the Central West villas seem relatively unimpressive, although in the context of the broader settlement pattern of the region these sites must have stood out as rich and impressive examples of architecture. LANDSCAPE CONTEXT AND INFRASTRUCTURE TOPOGRAPHY The bias in the excavated data towards the lowerlying areas of the region has already been noted above, and it would be wrong to assume that areas without excavated sites were entirely uninhabited. Taylor (2007, 14, fig. 3.2), for instance, has mapped large numbers of settlements recorded as earthworks, cropmarks and soil marks from upland parts of the region, particularly in the Marches, although few of these sites have been dated through excavation. It is therefore likely that the lowland focus in the excavated data reflects, at least to a certain extent, greater numbers of excavations associated with increased development

in these areas. Nevertheless, the modern settlement pattern also exhibits a strong preference towards the plateaus and river valleys, and it seems likely that the emphasis on these areas was a genuine characteristic of the Romano-British settlement pattern. Collectively, the sites from the region appear to be evenly distributed across a range of heights, ranging from 9 m above sea level up to a height of 313 m. However, there are some striking differences between some of the classes of site in terms of their situation within the landscape. It is notable, for example, that of the region’s six military vici, four occupy positions below 50 m, and two are located above 100 m. These sites do not occupy ‘mid-level’ terrain. Of course this reflects the fact that these nucleated settlements are directly associated with forts, which always occupy strategic locations, and all of the low-lying forts appear to have controlled the area around major river confluences, whereas those occupying prominent positions commanded routes through the terrain. A further striking aspect of the topographic distribution between different classes of sites is a clear preference for villas to occupy elevated positions, usually on slopes overlooking valleys (fig. 8.19). Of the fourteen villas, ten (71 per cent) occupy positions at 75 m above sea level or higher. Eight of these (57 per cent) are located at elevations between 75–100 m, compared, for example, with just 14 per cent of farmsteads, the majority of which lie below 75 m. It is striking that the two villas which do occupy lower terrain

0

fig.

Romanch8.indd 298

villas

height above ordnance datum (m)

height above ordnance datum (m)

farmsteads 300+ 290-299 280-289 270-279 260-269 250-259 240-249 230-239 220-229 210-219 200-209 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

2

4

6 8 10 no. of sites

12

14

300+ 290-299 280-289 270-279 260-269 250-259 240-249 230-239 220-229 210-219 200-209 190-199 180-189 170-179 160-169 150-159 140-149 130-139 120-129 110-119 100-109 90-99 80-89 70-79 60-69 50-59 40-49 30-39 20-29 10-19 0-9

0

1

2

3 4 no. of sites

5

6

7

8.19.   Spot height analysis of farmsteads and villas in the Central West region

07/09/2016 16:24:52

THE CENTRAL WEST

appear to be of a somewhat different character to many of the other villas. The villa at Loushers Lane is early (established c. a.d. 75), situated adjacent to the roadside settlement at Warrington, and it is thought to be closely associated with the nucleated site (Rogers and Garner 2007, 43–4). Pentre Farm, Flint, is believed to have been directly linked to the control of lead mining (O’Leary 1989, 50–1), and its likely association with a recently discovered roadside (probably industrial) settlement has been noted above. Both of these buildings were therefore potentially official residences occupied by people responsible for the administration of places or resources, and their position is governed by the location of the activity they are associated with. The reasons more typical villas tend to be situated where they are within the landscape are unclear. It may at least partly be aesthetic, with locations chosen because of the views they afforded, although perhaps the selection of elevated locations was another component in the armature of villa construction, possibly for reasons of architectural tradition, and perhaps as a further indicator of social status, associated with the desire for the building to be seen from particular positions within the landscape. If, perhaps, villa owners were supported by rent from other farmsteads, then there would also be less need for them to be situated near the fertile valley bottoms (Higham 1993, 52). It is difficult to assess how important Mediterranean traditions would have been for the wealthy occupants of Britain, but the focus for high-status dwellings on hill slopes may have been a fashionable symbolic expression of power. TRANSPORTATION: RIVERS, ROADS AND TRACKWAYS The Roman infrastructure evidently had a significant impact on the development of the settlement pattern within the region. As well as the establishment of legionary bases and auxiliary forts, the military consolidation of the area involved the construction of roads, resulting in the transformation of the landscape, the new roads becoming the focus for the region’s nucleated sites. All such settlements within the region occupy road junctions or sit alongside Roman roads. Most other sites within the region are also within relatively close proximity to a road, almost all lying within 10 km and the large majority within 5 km. In terms of the farming settlements, villas appear to have a closer association with roads than farmsteads; half of all villas are located within 1 km of a road, compared to 26 per cent of farmsteads. Beyond the major Roman roads, a fairly large proportion (nearly 30 per cent) of farmsteads and

Romanch8.indd 299

299

0

100 m 1:2000

fig. 8.20   Plan of a trackway leading to a mid- to late Roman farmstead at Billingley Drive, Thurnscoe, South Yorkshire (Neal and Fraser 2004)

villas produced evidence for transport links in the form of trackways, these being features of both the late Iron Age and Roman period landscapes. A number of these features are represented by earthworks or cropmarks recognised during aerial photography, although at several sites traces of trackways/roads were excavated, typically leading to the entrances of settlements. At Billingley Drive, Thurnscoe, South Yorkshire (Neal and Fraser 2004), a ditched trackway led to the entrances of two mid- to late Roman enclosures (fig. 8.20); at St Donat’s, Burghill/Upper House Farm, Moreton on Lugg, Herefordshire, a metalled trackway or minor road, flanked by ditches, linked the site to Watling Street, a major Roman road (Wainwright and Rogers 2007); at Preston Farm, Shropshire a metalled trackway led to the gated entrance of the settlement enclosure (Ellis et al. 1994). As well as roads and trackways, rivers formed another potential means of transport and communication, although there is little direct evidence for their use from the region and navigable waterways appear not to have been especially important in terms of the location of most farmsteads, villas or roadside settlements. The exception is for forts and associated vici, which, as discussed above, are more likely to occupy lower-lying river valleys, allowing the control of waterborne traffic and bridging points, as well as facilitating supply by boat. FIELD SYSTEMS Field systems were a further element of the landscape surrounding many settlements, and features identified as being parts of field boundaries, usually represented by linear ditches, were present at 31 sites (22 per cent), almost all of them farmsteads. In addition, a further nineteen sites had excavated features that were interpreted as elements of field systems but where no associated domestic focus has been found. The excavated

07/09/2016 16:24:52

300

THE RURAL SETTLEMENT OF ROMAN BRITAIN

field systems are fairly widely distributed across the region, although there are few examples from the northern part of the Mosses and Meres/ Shropshire Hills/Wrexham area, and only one from the Mersey Basin zone. It is possible, however, that this geographical imbalance may reflect the reduced availability of cropmark evidence from this region rather than a genuine difference in their distribution (Philpott 2000a, 192). SETTLEMENT HIERARCHIES: THE SOCIAL AND ECONOMIC BASIS OF SETTLEMENTS. A CASE STUDY FROM THE CHESTER AND WROXETER HINTERLANDS The above sections have presented a summary of the broad settlement pattern for the Central West, yet it is important to recognise that within the region there are some demonstrable differences that are bound up with localised historical developments; the Roman conquest and the

fig.

Romanch8.indd 300

establishment of military bases, industrial sites and urban centres appear to have affected the countryside in different ways. There is insufficient space here to consider this in detail for the entire region, but a case study focusing on Chester and Wroxeter, with a particular focus on the social and economic basis of settlements within these areas, exemplifies some important nuances within the region. This comparative case study focuses on excavated sites within a 30 km radius of each of these major centres respectively (fig. 8.21). The areas around Chester and Wroxeter are two of the best represented areas in the region for excavated sites, but the evidence is still sparse compared to many other regions, and the fairly limited evidence for rural settlement has already been given considerable attention, notably by Gaffney et al. (2007) for the Wroxeter area and by Carrington (2012) and Philpott (2000a; Philpott and Adams 2010) for the area around Chester. Readers requiring more detail than the brief overview presented here are therefore directed towards these detailed surveys.

8.21.  The Chester and Wroxeter case study area

07/09/2016 16:25:07

THE CENTRAL WEST

SETTLEMENT PATTERNS Wroxeter, Viroconium Cornoviorum, was the civitas capital of the Cornovii during the Roman period, probably replacing an earlier tribal centre, possibly the hillfort known as the Wrekin (White and Barker 1998, 39; Philpott and Adams 2010, 170). Chester, Deva, may have have lain within the same civitas, being referred to as a polis of the Cornovii by Ptolemy (Petch 1987, 115), although the true extent of the Cornovian territory is not known (see Ch. 12 for discussion of ‘tribes and civitates’). While at a broad level the rural Romano-British settlement pattern around Chester and Wroxeter can be described as very similar, characterised by dispersed, small, enclosed farmsteads with broadly similar architectural traditions, there are a number of subtle differences within the hinterlands of the two centres that potentially reflect important differences in the way the countryside was organised and exploited. Of critical importance to this are the fundamental differences between Chester and Wroxeter, the former a long-lived legionary fortress, the latter a civitas capital. Chester was occupied by a permanent military force (whose numbers fluctuated according to military campaigns and troop movements), at least until the mid-third century, after which military occupation of the fortress appears to have continued at a reduced level (Bidwell and Hodgson 2009, 72). The fortress had a large and mixed extra-mural settlement, the canabae, populated by soldiers’ dependents and those involved with providing services and goods to the army. Legionary fortresses are assumed to have been surrounded by a large area named the prata legionis, territory held by the legion used for grazing cavalry mounts, herds and flocks, and the provision of natural resources and raw materials

14

301

(Philpott and Adams 2010, 207; Carrington 2012, 344–5; Mason 1988, 164–5). While the precise extent of the prata surrounding Chester is uncertain, where the area of other prata have been calculated they could amount to more than 500 km² (Carrington 2012, 344), and it has been suggested that Chester’s prata may have been situated to the north and east of the fortress, perhaps including all of the Wirral (Mason 1988, 180). Chester’s hinterland might therefore have included a substantial area that was under the direct control of the military, as opposed to the territory around Wroxeter, which, once the site had transformed from a legionary base into a civitas capital, was probably all under the control of the civitas Cornoviorum, governed by local elites. To what extent can the differences between these major sites be recognised in the countryside? One of the most notable differences relates to the chronology of settlement, and the apparently low population density of Cheshire during the late Iron Age, compared with Shropshire, has previously been noted (Philpott 2000a, 192; Philpott and Adams 2010, 175) (fig. 8.22). While this difference may partly reflect variance in the intensity of aerial reconnaissance and crop regimes in the two areas, as well as greater susceptibility for soils in the Severn Valley to produce cropmarks, based on the excavated evidence alone there are notably more sites from the Wroxeter area than from the Chester area with evidence for occupation in the late Iron Age, even though the two areas are represented by broadly similar numbers of excavated sites. While the Chester sites have an overwhelming tendency to be established between the late first to early second century a.d., there seems to be greater evidence for continuity between the late Iron Age and Roman periods around Wroxeter. The seemingly rapid increase in the number of Chester

Wroxeter

12

no. of sites

10 8 6 4 2 0

fig.

Romanch8.indd 301

8.22.   Chronology of farming settlements in the Chester and Wroxeter hinterlands

07/09/2016 16:25:08

302

THE RURAL SETTLEMENT OF ROMAN BRITAIN

farmsteads near Chester, from around the time the legionary fortress was established, is probably linked to the need for increased local production to meet the needs of the military occupation, and a form of colonisation, perhaps officially encouraged, has been suggested (Carrington 2012, 388; Philpott and Adams 2010, 207). One site near Chester appears to have strong associations with the fortress. The roadside settlement at Heronbridge, just 2 km south of Chester, established during the late first century, produced material culture of similar character to that of the canabae, and the discovery of tombstones and evidence for a substantial and impressive funerary monument, similar to examples recorded from Chester, indicate that at least some of the inhabitants were prosperous (Philpott 2006, 71). Given the proximity of the roadside settlement to Chester it is possible that it formed part of the prata legionis, although there is no direct evidence to support the suggestion. A further site with clear military links is Holt, situated adjacent to the River Dee around 12 km south of Chester. The site produced pottery and tiles, and is thought to have been established during the late first century as a manufacturing and supply depot for the legionary fortress. A site at Ochre Brook, approximately 7 km north of the River Mersey, and 23 km northeast of Chester, also revealed evidence for a relatively short-lived phase of tile production during the 160s and stamps indicated that the tiles were also being manufactured for the XX Legion at Chester. In contrast to the site at Holt, however, there is little evidence that the site was military in origin, and morphologically the site has all of the characteristics of a ‘normal’ enclosed farmstead. The identity of the owner of the site is of course uncertain, but Philpott has speculated that it may have been a retired veteran who leased land for the purpose (Philpott 2000a, 197). A further feature of Chester’s hinterland is the presence of a number of nucleated sites, many of which appear to have had an industrial focus. A range of industries are represented at these settlements, including pottery, iron and other metal production, although the extraction of salt was of major regional importance and is likely to have been the principal focus at Northwich (Jones 1972), Middlewich (Williams and Reid 2008) and Nantwich (Arrowsmith and Power 2012). The latter two sites in particular have produced a wealth of evidence for salt extraction, including brine tanks, lead salt pans and large quantities of briquetage. Several of these sites have strong military associations, most notably the vici at Middlewich and Northwich, whereas Warrington and Nantwich may also have been under official control (Carrington 2012, 387–8). The widespread

Romanch8.indd 302

distribution of Iron Age Cheshire briquetage (Morris 1985) indicates that at least one of these sites was involved in salt production in the preRoman period, and it is likely that the Roman military took control of and intensified production in some cases. The establishment of new nucleated industrial sites in the late first century, at the same time as the area sees a rapid increase in the number of farmsteads, is viewed as a direct response to the need to supply the Roman military machine, not just at Chester, but also the forts to the north (Carrington 2012, 387). The two roadside settlements in the vicinity of Wroxeter (Meole Brace, approximately 6 km to the west and Redhill, Telford, 16 km to the east) are of a somewhat different character. Meole Brace produced limited evidence for pottery production, but it has more characteristics of a roadside market than the industrial centres to the north (Ellis et al. 1994). Redhill, Telford, on the other hand, emerged out of a military supply depot and had evidence for metalworking, although there are no indications that it was a major production site (Mann 2011). MATERIAL CULTURE There is a clear hierarchy in both the Wroxeter and Chester hinterlands in terms of access to material culture. In both areas nucleated settlements produced evidence for access to a range of artefacts, including large quantities of imported pottery and a wide array of objects including coins, brooches and various other dress accessories (fig. 8.23). Although not as well represented by artefacts as roadside settlements, the finds assemblages from villas in the Chester and Wroxeter areas have produced a range of objects, with artefacts such as coins, brooches, other dress accessories and household equipment occurring frequently. While poorly represented by objects compared to the nucleated settlements, the villas in the two areas stand out as being considerably better represented by most artefact types than farmsteads; however, all classes of object are relatively rare at these types of site. However, whereas the nucleated settlements and villas in the Chester and Wroxeter hinterlands appear to behave in a broadly similar way in terms of the range of artefacts they produce, farmsteads from the two areas differ. One such difference can be recognised in pottery assemblages within the two areas, with perhaps the most obvious difference being the distinctive types of coarse wares present at farmsteads in the two areas; sites around Chester tend to be dominated by Cheshire Plain wares, with those near Wroxeter chiefly represented by Severn Valley wares. This difference evidently reflects the marketing distribution of pots in these different fabrics; Cheshire Plain wares were

07/09/2016 16:25:08

THE CENTRAL WEST

farmsteads (n=26)

% of sites with object present

100

villas (n=8)

303

nucleated settlements (n=7)

90 80 70 60 50 40 30 20 10 0

fig. 8.23.  Frequency of major artefact categories on different types of settlement in the Chester and Wroxeter hinterlands

50

Chester (n=14)

Wroxeter (n=12)

% of sites present

40 30 20 10 0

fig.

8.24.   Frequency of major artefact categories on farmsteads in the Chester and Wroxeter hinterlands

produced at Wilderspool, among other places, whereas Severn Valley ware production is known from Meole Brace, west of Wroxeter. There are further differences, however. Fieldwalking and excavation by the Wroxeter Hinterland Project, as well as elsewhere, revealed generally low numbers of ceramics in the area around the town, and the authors noted that the quantity of this material declines rapidly with distance from the city (White and Gaffney 2003, 223). This appears to represent continuity of an Iron Age cultural trend for favouring organic over ceramic vessel types (Gaffney et al. 2007, 280). Although pottery was used in the Severn Valley area during the Iron Age, its scarcity suggests that it may traditionally have been used only for certain activities or in certain socially prescribed contexts that continued into the Roman period (Wigley pers. comm.). Inconsistencies in the recording of pottery from excavated sites make detailed comparison difficult,

Romanch8.indd 303

but impressionistically the situation is somewhat different for the area around Chester. More farmsteads in the Chester hinterland consistently produce greater quantities of Roman pottery, pottery is more widely distributed, and sites appear to be better represented by imported pottery than farmsteads near Wroxeter, suggesting that pottery vessels performed a wider role in social practices in this area during the Roman period. This is despite a similar background of only very limited pottery-use during the Iron Age (Philpott and Adams 2010, 179–80). This pattern is also visible through other artefacts recovered from farmsteads. While it is true that in both areas finds are scarce compared to roadside settlements and villas, objects are undoubtedly more frequently recovered from the Chester hinterland (fig. 8.24); more sites in this area produce objects such as coins and brooches, and they tend to produce more of them.

07/09/2016 16:25:10

THE RURAL SETTLEMENT OF ROMAN BRITAIN

Notably, writing equipment is absent from all excavated farmsteads surrounding Wroxeter, yet three rural sites in the Chester area have produced styli (Irby, Wirral: Philpott and Adams 2010; Ochre Brook: Philpott 2000b; Plas Coch, Wrexham: N.W. Jones 2011), implying literacy in at least some parts of the countryside in this area. Philpott suggests the stylus from Ochre Brook may relate to accounts kept by civilian contractors supplying the military with tiles (Philpott 2000a, 197). Although the excavated sites in the Chester hinterland apparently produce more objects than those around Wroxeter, the relative scarcity of artefacts at excavated sites in both areas makes this difficult to demonstrate emphatically. However, the pattern is supported and emphasised by finds recorded by the Portable Antiquities Scheme (PAS). Firstly, there are substantially more artefacts recorded on the PAS database from the Chester area than Wroxeter (excluding objects and coins from hoards, 1423 artefacts from Chester, against 869 objects from Wroxeter – an almost 40 per cent difference). While it is difficult to establish the extent to which this reflects variance in levels of metal detecting and reporting of metal-detector finds, there are also clear differences in the proportions of the types of objects recovered by metal detectorists from the two areas; coins make up 59 per cent of finds on the PAS database from the Chester hinterland, and just 38 per cent of finds from the area around Wroxeter (fig. 8.25). Brooches, however, form roughly similar proportions of artefacts from both areas. There are several potential reasons for the apparent differences in the circulation of at least some aspects of material culture in the two landscapes. One is that the legionary fortress was a source of more wealth than Wroxeter, and that farmsteads in its hinterland benefited from the flow of consumable goods entering Chester through military trade networks. There are, however, also implications for our understanding of potentially different cultural attitudes in the two areas. This subject has been given recent attention by Jeremy Taylor (2013), who has suggested that for members of a traditional agricultural society in the Wroxeter hinterland, ‘Roman’ material culture may have been an irrelevance, and rural identity and status may have been tied into traditional communal activities and the ownership and display of cattle (Taylor 2013, 185; Gaffney et al. 2007, 280). If, as suggested by Carrington (2012, 388) and Philpott and Adams (2010, 207), the landscape around Chester was subject to colonisation by migrants (whether veterans, migrants from elsewhere in Britain or elsewhere in the empire),

Romanch8.indd 304

60%

Wroxeter (869 objects)

Chester (1423 objects)

50% 40% % of finds

304

30% 20% 10% 0% Proporon of coins

Proporon of brooches

fig. 8.25.   Proportion of coins and brooches among all finds recorded by PAS (excluding objects from hoards) in the Chester and Wroxeter hinterlands

then the greater numbers of coins and other artefacts circulating within this landscape may represent a more mixed population than that in the hinterland of Wroxeter. Such a scenario might also explain the limited distribution of ceramics in the Wroxeter area, where such mixed populations were perhaps confined to the city, smaller towns and roadside settlements. The scarcity or absence of artefacts from some sites in the Chester hinterland perhaps suggests that the demographic here was complex, and a local society seems likely to have continued to exist alongside a migrant population, as suggested by the presence of roundhouses at Birch Heath, Tarporley (Fairburn 2003). Given the pattern in the material culture, it seems paradoxical that villas within the region cluster around Wroxeter rather than Chester. As we have seen, two of the three excavated villa buildings near Chester (Pentre Farm, Flint and Lousher’s Lane, Warrington) are atypical in that they appear likely to have had official functions associated with the administration of resources and/or places. The only other example from near Chester, Eaton-by-Tarporley (Mason 1983), may be regarded as a more typical rural villa. All three of the Chester villas appear to have had relatively early origins, initially constructed between the late first and late second centuries a.d. Wroxeter, in contrast, is surrounded by five excavated villas, and a further seven have been recognised through other techniques (White and Gaffney 2003, 223). The dating evidence for these villas is of variable quality, but on current evidence the initial villa buildings appear to have been constructed similarly early, between the late first and the early third century a.d. The reason for the difference in the distribution of villas is unclear. It is possible that

07/09/2016 16:25:11

THE CENTRAL WEST

there are more in the Chester area awaiting discovery, and the poor susceptibility of the soils for the formation of cropmarks compared with the area around Wroxeter may help account for the pattern. However, it is also possible that the situation reflects social and economic differences in the landscapes surrounding the legionary fortress and the civitas capital. If a substantial part of the Chester hinterland was requisitioned for military supply much of the area may have been populated by migrant tenants or a servile workforce, and this may have restricted the opportunities for the display of private wealth through the construction of villa buildings. The greater number of villas near Wroxeter perhaps represents ownership of land by the Cornovian elite, who may have capitalised on the economic opportunities created by the requirements of the army at Chester and elsewhere in the north. There may also have been a degree of differentiation in the way in which the two areas were perceived in terms of social status, with it perhaps being more prestigious to construct villas in the environs of the civitas capital than the legionary fortress (Wigley pers. comm.). It is noteworthy in this regard that the villa at Whitley Grange, to the west of the Wroxeter, may have functioned as a hunting lodge, rather than a permanently occupied site (Gaffney et al. 2007, 141). While some differences between Chester and Wroxeter may be recognised in the archaeology of their respective hinterlands, other differences are manifested less clearly in the archaeology of the surrounding countryside. There is, for instance, little evidence for any distinction between the two areas in the late Roman period, when the two centres appear again to have experienced different circumstances. Wroxeter apparently flourished as an urban centre until at least the end of the fourth century, with occupation of some sort continuing into the fifth century and beyond (White and Barker 1998, though see now Lane 2014 for an alternative interpretation). Chester, however, seems to have witnessed more mixed fortunes, and although activity continued in some form into the fourth century, the level of occupation was considerably reduced (Bidwell and Hodgson 2009, 72). There is at present, however, little evidence to suggest that sites in the hinterlands of the two centres were affected differently by these late Roman developments. Rural sites in both areas diminish in numbers between the end of the second and the late fourth century, and while marginally more of the Wroxeter hinterland sites continue for longer than those in the Chester area the numbers are so small and the dating evidence so variable that this may perhaps be of little significance.

Romanch8.indd 305

305

RURAL SETTLEMENT ECONOMIES The evidence for the economic basis of settlements around Wroxeter and Chester is in both areas hampered by soil conditions, which are very unfavourable for the preservation of bone, and only six sites in the Chester hinterland and two in the Wroxeter area produced faunal remains. Even at these sites, animal bones were typically recovered in very small numbers, and only Pentre Farm and the roadside settlement at Nantwich (both in the Chester area) produced assemblages with more than 100 specimens. At Pentre Farm a midRoman assemblage suggested an emphasis on cattle, followed by sheep/goat and then pig, with horse and hare also identified. At Nantwich there was a similar emphasis on cattle, again followed by sheep/goat and pig, while other species included horse, dog, red and roe deer, and domestic fowl. There is insufficient faunal evidence from farmsteads in the hinterland of either centre to allow comment on the types of animals favoured for consumption at low-status rural sites, although across the region as a whole the limited data (just three farmsteads with more than 100 specimens) suggest an emphasis on cattle, followed by sheep/ goat and then pig. While there is little direct evidence from faunal remains, work in the Wroxeter area has indicated a general focus on pastoralism, suggested by the geographical relationship between settlement enclosures and soils more suited to grazing than arable production, with perhaps the exception of those fields immediately adjacent to the city walls that seem to have been used for market gardening (White and Gaffney 2003, 223; Gaffney et al. 2007, 254–7). At Duncote Farm, for instance, approximately 2 km north-east of Wroxeter, there was evidence for the reorganisation of the existing landscape during the second century a.d., when coaxial fields were established (Ellis et al. 1994), and these have been suggested as possible horticultural plots (White et al. 2013, 177). However, very few (just four) sites in the Wroxeter area have yielded botanical remains, and in most cases these were recovered in very small quantities. At Ellesmere Road, Shrewsbury (Booth 1996), significant quantities of spelt chaff and rye were recovered, as well as some barley and oats, although even here it was suggested that the initial stages of crop processing may have been carried out elsewhere. There is very little evidence of intensive arable production in the area as a rule. The impression of a pastoral focus for the Wroxeter hinterland sites is complemented by the presence at Wroxeter of an area of presumed industrial pitting recognised through geophysics, thought likely to represent processing of secondary animal products through tanning, fulling or dyeing,

07/09/2016 16:25:11

306

THE RURAL SETTLEMENT OF ROMAN BRITAIN

although hard evidence is currently lacking. These features, combined with the presence of a large enclosure interpreted as a large cattle market in the town, suggest that the city and its hinterland may have been focused on the intensive rearing of livestock, perhaps also supplying the northern military sites with cattle and other animal products (Gaffney et al. 2007, 254–6, 283; Carrington 2012, 380). A slightly greater number of sites from the area around Chester have produced botanical evidence, although the numbers are still small, and in most cases the evidence is relatively minor, with most evidence coming from nucleated sites such as Middlewich (Garner and Reid 2012; Williams and Reid 2008) and Nantwich (Arrowsmith and Power 2012). Spelt wheat appears to have been the dominant crop, but the presence of a wide range of cereals, including emmer, bread wheat, barley and rye, has been suggested as evidence for a lack of specialisation in the region (Stallibrass 2011, 117). While the presence of corndryers at Heronbridge (Mason 2004), Plas Coch (N.W. Jones 2011) and possibly the Eaton-by-Tarporley villa (Mason 1983) indicate large-scale processing of grain in the area, Carrington has suggested that the principal products of the Chester hinterland, perhaps provided as a tax in kind, were animals and animal products (Carrington 2012, 395). Of note in this regard is evidence from Kingsley Fields, Nantwich, where hook-damaged cattle scapulae, often recovered from military and urban sites in the north of England (Dobney 2001, 40–1), suggest hung beef shoulders, which were perhaps brined and dried in Nantwich (Arrowsmith and Power 2012). Skulls of cattle, sheep and red deer were also present and may represent skinned carcasses. The importance of the nucleated sites with evidence for salt production, Northwich, Nantwich and Middlewich, may therefore be about more than the production and distribution of salt as a commodity in its own right, being also associated with the use of salt for processing carcasses and preserving joints of meat in order to supply the military. While a range of cereals was recovered from the farmstead at Birch Heath, Tarporley (Fairburn 2003), a broader economic basis is suggested by the presence of a number of sherds from ceramic cheese presses, suggesting cheese production dependent on cattle or sheep rearing (Carrington 2012, 385). CASE STUDY SUMMARY The above discussion has sought to present some of the intra-regional diversity within the rural settlement pattern of the Central West. There are notable differences between the landscapes surrounding Wroxeter and Chester, which reflect

Romanch8.indd 306

the particular historical circumstances that led these two sites, originally both legionary fortresses, to develop very differently. While the earlier sections of the chapter provided a descriptive overview of the character of the Romano-British settlement pattern in the region, this case study has highlighted some important nuances, indicating that the hinterlands of Wroxeter and Chester, despite their geographical proximity, must have felt like very different places. REGION SUMMARY The Central West region includes a wide variety of different landscape zones, with some evidence for corresponding variability in the density of late Iron Age and Roman settlement. In particular, low numbers of excavated sites within the extensively developed urban areas, or from major infrastructure projects such as the M6 Toll and major pipeline routes, suggest that the landscape in parts of the region may have been relatively empty of permanent occupation, perhaps only utilised occasionally, such as for seasonal grazing. The late Iron Age and Roman rural settlement pattern of the Central West can be broadly characterised as being chiefly represented by dispersed, enclosed, farmsteads. Villas were certainly found within the region, but they were uncommon and typically of comparatively modest form. Aside from the major Roman urban and military centres of the region, Wroxeter and Chester, nucleated settlements were distributed along the region’s major Roman roads; in the south, many were walled ‘small towns’ or undefended civilian settlements, yet in the north most were vici associated with Roman forts, or seem otherwise in some way to have had military connections. Several of these northern nucleated sites appear to have performed important industrial functions (including salt production), likely to be associated with supplying the legionary fortress at Chester, as well, perhaps, as forts on the northern frontier. This highlights one of the principal characteristics of this region, as lying on the junction between the agricultural heartland of Roman Britain (Central Belt; Ch. 5), with its mass of villas, roadside settlements and complex farms, and more upland zones (the North and Upland Wales and Marches; Chs 9 and 11) with – in general – smaller, dispersed, enclosed farmsteads, and a much greater and longer lasting Roman military presence. Our understanding of the chronological development of rural settlements in the Central West is hindered by the often limited dating evidence, a consequence of the regional scarcity of material culture, especially during the late Iron

07/09/2016 16:25:11

THE CENTRAL WEST

Age, which makes it difficult to identify and confidently date sites. In general terms, however, there appears to have been an increase in the number of sites occupied between the late Iron Age and the end of the second century a.d., when the numbers of settlements in use peaked, before a decline during the late third and fourth centuries. There are, however, some distinctive subregional differences, regarding both chronology and other aspects of the evidence, reflecting the range of physical landscapes within the region as well as local historical developments. For example, sites to the east and west of the Pennines appear often, on the basis of the existing dating evidence, to have developed quite differently. Late Iron Age/ first century a.d. sites have been identified from across the region, yet, in the Coal Measures to the east, there seems to have been considerably greater settlement expansion during this period than is apparent further west. Many late Iron Age sites from all landscapes continued into the Roman period, yet from the third century onwards there appears to have been a period of more marked settlement abandonment in the Coal Measures, as opposed to the gradual pattern of settlement disuse observed in the west. This is part of a pattern previously noted in South Yorkshire, which suggests the collapse during the late Roman period of long-lived communities with traditions dating back to the Iron Age (Hodgson 2012, 53). Although the evidence suggests that the west of the region may have been less densely occupied than the east during the late Iron Age, there is, in

Romanch8.indd 307

307

some areas, strong evidence that pre-existing Iron Age cultural traditions persisted into the Roman period. Roman rural sites in the Severn Valley, near the civitas capital at Wroxeter, typically produce very small quantities of Roman ceramics, apparently reflecting continuity of a cultural preference towards the use of organic vessels, with rural sites in this area betraying little evidence for engagement with the Roman city (Gaffney et al. 2007, 280). This is in contrast to the situation in the landscape surrounding the legionary fortress at Chester, where rural sites often produce a greater range of material culture, suggesting that the population of this area included at least some groups of people with different cultural values to many of those inhabiting the Wroxeter hinterland. The chronology of sites near Chester, many of which appear to have been established shortly after the foundation of the legionary fortress, suggests that this part of the region saw early Roman settlement expansion, possibly involving migrant populations, associated with the requirements of supplying the legionary base, and possibly Roman forts further afield. While at first glance the character of the Roman settlement pattern of the region outside of the towns and military sites might seem uniform and relatively unvaried, dominated as it is by enclosed farmsteads, with few villas, even fewer complex farmsteads and an absence of villages, this masks a complex situation with locally distinctive chronologies and social and economic differences that were influenced by a range of circumstances.

07/09/2016 16:25:11

Chapter 9

THE NORTH By Tom Brindle

The North region is large, extending over an area of 19,818 km². Its northern boundary is formed by the English/Scottish border and it contains the entirety of the modern administrative counties of Cumbria and Northumberland, the unitary authorities of Tyne and Wear, most of County Durham and Lancashire and the westernmost third of North Yorkshire (fig. 9.1). Hadrian’s Wall, which extends east from the Solway Plain in Cumbria to Wallsend, North Tyneside, is of course one of the most notable aspects of the Roman archaeology of the North. Its presence, alongside a large number of Roman forts, has resulted in a traditional emphasis on the archaeology of the Roman military in the region; the difficulty of recognising and dating Romano-British rural sites in the area has meant that in the past these sites

fig.

attracted less attention (Philpott 2006, 62; Hingley 2004, 343). However, our understanding of the region’s character during the Roman period has improved considerably as a result of more recent efforts to integrate aspects of the military and rural archaeology (e.g. Hodgson et al. 2013; Proctor 2009). THE NATURE OF THE LANDSCAPE The North incorporates a diverse range of landscapes, with some extreme topographical differences. These include low-lying areas such as the Lancashire Plain and Valleys, the West Cumbria Coastal Plain, the Solway Basin and the Eden Valley in the west, and the North Northumberland

9.1.  The North region in relation to modern county boundaries

308

Romanch9.indd 308

06/09/2016 17:27:13

THE NORTH

fig.

9.2.   Constituent landscape zones of the North region

Coastal Plain and Northumbria Coal Measures in the east (fig. 9.2). Much of the region is also of upland character, including the Border Uplands of Northumberland as well as the rugged mountains of the Cumbria Fells and Dales and the North Pennines and the Yorkshire Dales. The soils of the region are principally acidic, particularly in lowlying areas where the excavated evidence is concentrated, and this restricts the amount of faunal and archaeobotanical remains available from the region, although pollen evidence provides some information about the ancient environment. THE NORTH DATASET The North dataset includes 138 records for 123 distinct sites (15 records relate to sites with multiple records). These include 99 settlements and 24 non-domestic sites, most comprising those associated with burial and industry, along with field systems. The geographical distribution of sites is far from even, with much of the region being very poorly represented by excavated data (fig. 9.3; table 9.1). There are very few sites from some parts of the region characterised by upland

Romanch9.indd 309

309

landscapes such as the North Pennines, the Yorkshire Dales, the Forest of Bowland or the Cumbria Fells and Dales. The areas best represented by data are the low-lying Solway Basin, the Eden Valley and the Northumbria Coal Measures. The distribution of excavated rural sites mapped in fig. 9.3 (top) is, however, a poor guide to the true distribution of Romano-British settlement in the region, primarily reflecting the areas in which excavation has been focused. figure 9.3 reveals how excavations of all periods have focused principally on the region’s lowlands, as well as the area around Hadrian’s Wall. The Northumbria Coal Measures, for example, have seen a number of excavations, particularly to the north of Newcastle, often as a result of residential development and mineral extraction. In the Solway Basin in Cumbria a number of sites have been excavated in advance of residential development around Carlisle, but the susceptibility of the sand and gravel geology of the area to form cropmarks has also resulted in a number of excavations targeting sites initially identified through aerial photography. The relatively high frequency of excavated Iron Age and Romano-British sites in the Border Uplands largely reflects the particular

06/09/2016 17:27:21

310

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 9.3.  Distribution of excavated Roman rural sites (n=123) and all excavation records (1910–2010) from National Monument Records (NMR) Index (n=2908) in the North region

Romanch9.indd 310

06/09/2016 17:27:39

THE NORTH

table

9.1: number

of sites and density (per km²) by landscape zone in the north region

Landscape zone

Area (km²)

No. of sites

978.13

21

0.0215

809.56

11

0.0136

1565.92

20

0.0128

Solway Basin Eden Valley Northumbria Coal Measures North Northumberland Coastal Plain

Density of sites per km²

377.58

3

0.0079

Border Uplands

3950.80

28

0.0071

Lancashire Plain and Valleys

1642.24

11

0.0067

West Cumbria Coastal Plain

499.74

2

0.0040

Pennine Dales Fringe

873.03

3

0.0034

Cumbria Fells and Dales

3461.12

10

0.0029

North Pennines

2145.63

6

0.0028

Forest of Bowland

1114.85

3

0.0027

Yorkshire Dales

2399.84

5

0.0021

research interests of George Jobey who, during the second half of the twentieth century, excavated a number of Iron Age and Romano-British rural sites in this part of the region. Whereas the general trend towards a greater density of sites in the valley systems and low-lying areas may reflect genuine concentrations of settlement in these fertile areas (cf. Mattingly 2006, 418) the uplands were certainly by no means as sparsely populated as the excavated evidence suggests. Taylor (2007, 12) has shown how the North-West and the North-East are, by far, the two regions of England best represented by upstanding earthworks, with good preservation of these in the region’s uplands, which have seen less destruction through arable farming or housing development. Although usually undated, sites represented by earthworks and cropmarks together made up 85 per cent of the potential evidence for RomanoBritish settlement within Taylor’s dataset for the North-East and North-West (Taylor 2007, 12), and the excavated evidence formed only a minor component. While the excavated data provide little information about the density of settlement in the North, however, they do provide contextual and chronological information that is rarely available from sites recognised primarily through aerial photography or field survey. This is particularly the case in an area in which finds scatters form a minor component of the evidence, because of a general scarcity of artefacts at rural settlements in the Iron Age and Roman periods (ibid., 41–3). ROMAN RURAL SETTLEMENT PATTERNS The Roman archaeology of the North is dominated by military sites, notably the forts associated with Hadrian’s Wall and the major communication

Romanch9.indd 311

311

routes of the region (fig. 9.4). Villas are almost completely absent from the region, although sites characterised as farmsteads at Old Brampton (Blake 1960) and Old Durham (Richmond et al. 1944; Wright and Gillam 1951; 1953) are contenders. Indeed, the latter may relate to a known distribution of villas in south-east County Durham in the North-East region (Ch. 7), and may therefore be regarded as a northern outlier of this group, rather than an entirely isolated occurrence. The civilian towns and villages that are a feature of the settlement pattern further south in the province are absent, although large nucleated civilian settlements, vici, were associated with almost all of the permanent forts (see Sommer 2006 for an up-to-date assessment of this settlement type). At least two of these, Carlisle in Cumbria, and Corbridge in Northumberland, may eventually have had civic functions as tribal capitals (Mattingly 2006, 261). In terms of the rural settlement pattern, farmsteads, as elsewhere, represent by far the most common type of settlement recorded from the region; the 81 excavated farmsteads account for 82 per cent of settlements (table 9.2; fig. 9.4). It is important to note here that military vici are falsely under-represented in the dataset, as for reasons of pragmatism (i.e. numbers of archaeological interventions are too large) vici associated with forts along Hadrian’s Wall and the Stanegate (e.g. Vindolanda) were excluded from data collection. The Border Uplands and the Solway Basin, through which the route of Hadrian’s Wall runs, therefore include greater numbers of vici than table 9.2 and fig. 9.4 suggest. Aside from farmsteads and vici, other types of settlement are very scarce indeed. A single site has been classified as a roadside nucleated settlement, at Walton-le-Dale, Lancashire (Gibbons and

06/09/2016 17:27:39

312

THE RURAL SETTLEMENT OF ROMAN BRITAIN

fig. 9.4.  

Distribution of excavated late Iron Age/Roman rural settlements in the North region in relation to Roman roads and urban military sites (excluding excavated vici associated with forts on Hadrian’s Wall and the Stanegate)

table

9.2: number

of sites by type in each landscape zone in the north region

Landscape zone Farmstead Roadside Vicus Other Total % of total settlement site settlement Border Uplands

26

0

0

2

28

23%

Northumbria Coal Measures

16

0

4

0

20

16%

Solway Basin

14

0

2

5

21

17%

North Pennines

5

0

0

1

6

5%

Lancashire Plain and Valleys

5

1

3

2

11

9%

Eden Valley

4

0

4

3

11

9%

Cumbria Fells and Dales

4

0

2

4

10

8%

North Northumberland Coastal Plain

3

0

0

0

3

2%

Yorkshire Dales

3

0

0

2

5

4%

West Cumbria Coastal Plain

1

0

0

1

2

2%

Pennine Dales Fringe

0

0

2

1

3

2%

0

0

0

3

3

2%

81

1

17

24

123

100%

Forest of Bowland Total

Howard-Davis 2001; Pickering 1957), although the importance of this site may have been as a harbour with likely military connections (Gibbons and Howard-Davis 2001). The remaining 24 non-

Romanch9.indd 312

domestic sites comprised six industrial sites, mostly with military associations, eight funerary sites, primarily associated with vici, five field systems, three caves and two shrines.

06/09/2016 17:27:49

THE NORTH

313

70 60

no. of selements

50 40 30 20 10 0 Late Iron Age fig.

AD50-100

AD200-300

AD300-400

9.5.   Number of settlements in use over time in the North region

REGIONAL CHRONOLOGY It is necessary to preface any discussion of the chronological development of Romano-British settlement in the North with a caveat regarding the quality of dating evidence for rural settlements in the region. Whereas there has been a marked increase in the adoption of absolute dating techniques in recent years, particularly radiocarbon dating, many of the sites excavated prior to the end of the twentieth century are dated based exclusively on ceramic evidence. Unfortunately, ceramics are a poor dating tool for late Iron Age and Romano-British rural sites in the north. Much of the region appears to have been aceramic during the Iron Age, and most rural sites produce little pottery, even when occupied during the Roman period. Where Iron Age pottery does occur it is often imprecisely dated, and very long-lived pottery traditions mean that it is usually impossible on the basis of pottery form or fabric to identify the point within the Iron Age when vessels were in use (Sherlock 2012; Cunliffe 2005, 212; Hodgson pers. comm.). These issues clearly limit the confidence with which many farmsteads in the region can be dated, and it is likely that many sites occupied during the late Iron Age and Roman periods have gone unrecognised due to a lack of dating evidence. For other settlement types, even those without radiocarbon dates, the basis for our understanding of site chronologies is on a sounder footing, as most have a Roman military connection and therefore had access to a range of more closely datable material culture. This lack of precision for the region’s sites means that the chronological patterns shown in figs 9.5–9.8 are presented here by century, rather than by half-century as in some chapters in this volume. The variable quality of dating evidence also means that the discussion of settlement chronology that follows must be regarded as tentative.

Romanch9.indd 313

AD100-200

The above caveats noted, the broad chronological pattern for settlement within the region, incorporating all types of domestic site, is for a steady rise in the number of settlements occupied between the late Iron Age and the end of the second century a.d. It is somewhat unclear whether the apparent peak in sites occupied during the second century a.d. reflects a genuine increase in settlement numbers, or whether the pattern represents an increased number of sites receiving Roman pottery at this time, perhaps misleadingly suggesting a second-century origin for some. Whatever the reality, the increase in settlement seems to have been relatively short lived, and there appears to have been a dramatic reduction in the number of sites occupied during the third century a.d., followed by continued settlement abandonment during the fourth century (fig. 9.5). This broad trend of course masks important differences in the chronologies of the two main classes of site within the region, with the chronological development of military vici and farmsteads being quite different for the most part (figs 9.6 and 9.7). Vici grew up adjacent to forts and their fortunes were therefore inextricably linked to the Roman military situation in the north; they therefore tend to emerge during the late first to early second century a.d., during the various phases of military consolidation of this period. The decline of military vici in this region during the second half of the third century is a well-recognised phenomenon (Bidwell and Hodgson 2009, 33–4). The region’s farmsteads, on the other hand, have more varied chronologies. Some produce evidence suggesting continuity between the Iron Age and Roman periods (e.g. Baldhowend, Matterdale, Cumbria: Loney and Hoaen 2005; Kennel Hall Knowe, North Tynedale, Northumberland: Jobey 1978), whereas others, as far as we can tell from the abovementioned limited

06/09/2016 17:27:51

314

THE RURAL SETTLEMENT OF ROMAN BRITAIN 18 16

no. of selements

14 12 10 8 6 4 2 0 Late Iron Age fig.

AD50-100

AD100-200

AD200-300

AD300-400

9.6.   Number of vici in use over time in the North region 60

no. of selements

50 40 30 20 10 0 Late Iron Age fig.

AD50-100

AD200-300

AD300-400

9.7.   Number of farmsteads in use over time in t