Preface and Acknowledgements
The English Landscapes and Identities project (known hereafter by its acronym EngLaId) was a five-year project which ran between 1st August 2011 and 31st July 2016. It was funded by an ERC Advanced Grant (269797) awarded to Chris Gosden and we are very happy to acknowledge the support of the European Research Council. The researchers on the project were Chris Gosden (principal applicant), Anwen Cooper (prehistory), Miranda Creswell (artist), Tyler Franconi (Roman period), Chris Green (GIS), Letty ten Harkel (early medieval), Zena Kamash (Roman period), and Laura Morley (research coordination). The project looked at the long-term history of the English landscape from 1500 bc to ad 1086, combining evidence on landscape features, such as track-ways, fields, and settlements, with the distribution of metalwork. The project examined a crucial period of English landscape history from the start of the settled agricultural landscape to the medieval world, which was directly ancestral to that of modernity. Working from the Bronze Age to the early medieval period revealed great evidence of change, but also surprising continuity in terms of land divisions and forms of settlement. We were also interested in how this patterning relates to the types of artefacts deposited and the places in which they were deposited over this period. The project was not purely empirical and attempted to develop theory concerning the relations between people and the material world.
The project attempted to synthesise all the major available data sets from English archaeology within a digital environment, making this a ‘Big Data’ project, eventually creating a database of over 900,000 items. Three doctoral students joined the project in October 2012. Victoria Donnelly examined the ‘grey literature’ from England since 1990 and through this the practices of archaeology after PPG 16. Sarah Mallet gathered together and analysed the major sets of isotopic data from humans, animals, and plants across England for the EngLaId periods. Daniel Stansbie tackled the topic of food and attempted to pull together evidence from pottery, animal bones, and plant remains for the Thames Valley and Kent for the EngLaId periods. Each submitted and defended their theses. Roger Glyde was an important member of the team throughout, carrying out a range of empirical work, as well as commenting on, and contributing to, the written results. We feel that the project has been productive, but it has also been really enjoyable, with the team forming a close group within which people worked and socialized.
More information is given on the background to the project, our sources of data, and modes of analysis in Chapters 1–3, together with an outline of this volume. Here we simply provide a broad timetable of the project. As mentioned, the team started work in August 2011, although not everyone was able to join until January 2012 due to prior commitments. Zena Kamash was offered a lectureship in 2014, left the project, and was replaced by Tyler Franconi as the Roman specialist. The data-gathering phase of the project lasted until May 2013 and we were able to properly start analysis after that. The last year of the project was spent writing up.
We are grateful to an enormous number of people and we hope we have not missed too many out. The earliest phase of the project gathered data. We are very grateful to numerous people in what was then English Heritage (and now Historic England (EH)) for the provision of data, including Simon Crutchley, Peter Horne, Lindsay Jones, Martin Newman, and
Barney Sloane, as well as Nick Davies, Gill Grayson, Sarah Maclean, David McOmish, Sarah Poppy, and Poppy Starkie. The Portable Antiquity Scheme (PAS) was the main source of data on artefacts and we would like to thank Roger Bland, Michael Lewis, Sam Moorhead, Stephen Moon, Mary Chester-Cadwell, and Dan Pett. Catherine Hardman and Stuart Jeffrey of the Archaeology Data Service (ADS) gave us important advice on the relevant archives held by the ADS. Tim Evans (ADS) steered us towards the Excavation Index and provided considerable advice. We also benefitted from advice and information from a number of Finds Liaison Officers, including Frank Basford (Isle of Wight) and Tom Brindle. Ehren Milner at the Archaeological Investigations Project (AIP) was an important source of data and advice.
Our main source of data came from local archaeological officers. We would like to thank all Historic Environment Record (HER) officers who provided us with data. These include Christine Addison, Northamptonshire HER; Sarah Botfield, Peterborough HER; Stewart Bryant, Association of Local Government Archaeological Officers (ALGAO); Giles Carey, Shropshire HER; Jo Caruth, Suffolk HER; Rebecca Casa-Hatton, Peterborough HER; Sally Croft, Cambridgeshire HER; Ben Croxford, Kent HER; Phillip de Jersey, States of Guernsey; Lucie Dingwall, Herefordshire HER; Keith Elliott, Northumberland HER; Heather Hamilton, Norfolk HER; Mike Hemblade, North Lincolnshire HER; Richard Hoggett, Suffolk HER; Rebecca Loader, Isle of Wight HER; Fiona Maconald, ALGAO; Colin Pendleton, Suffolk HER; Guy Salkeld, National Trust; Melissa Seddon, Herefordshire HER; Graham Tait, ALGAO; Bryn Tapper, Emma Trevarthen, Jacky Nowakowski, and Andrew Young, Cornwall and Scilly HER; Ben Wallace, ALGAO; Penny Ward, Shropshire HER; Chris Webster, Somerset HER; Liz Williams, Northumberland HER; and Alison Yardy, Norfolk HER. Keith Westcott at exeGesIS SDM Ltd helped develop a query which could extract data from the HBSMR database system used by more than half of HERs.
In the middle stages of the project a great number of people gave us advice and shared their knowledge of local archaeology or the situation across the country more broadly. These include Martin Allen, Fitzwilliam Museum; John Baker, Stuart Brookes for medieval data and discussions; Chris Evans for advice and critique; Graham Fairclough gave us information on Historic Landscape Characterisation and other matters; Duncan Garrow linked to the Celtic art database and gave advice on other matters; Ian Leins gave advice on coinage; Katie Robbins shared her thoughts on modelling PAS data; Iona Robinson for sharing unpublished material; Sarah Semple for general advice on the medieval period; Sue Stallibrass for suggestions on how to incorporate environmental data; Fraser Sturt, Southampton University, provided his modelling of sea levels; Pete Topping for general thoughts and advice; Clive Waddington for sharing his knowledge of the northeast; Philippa Walton for advice on Roman finds; and Ole Wiedenmann, History Data Service, provided information on historic parishes and place names.
In Oxford, Jane Kershaw helped during the initial setting-up phase, John Pouncett and Gary Lock provided advice on digital and other matters. Janice Kinory provided her database of salt-making sites and Lisa Lodwick advised on plant remains and agricultural regimes. Steve Hick and Jeremy Worth gave us financial and IT support, respectively. Chris Gosden would like to thank Elizabeth Allen for organizing so much. We ran two successful workshops and a conference. We are very grateful to all speakers, chairs, discussants, and audience participants. We were assisted by volunteers who processed various forms of data and provided informed discussion. We are grateful to Pat Day, Pam England, Paula Levick, and Steve Northcott. Roger Glyde started as a volunteer and ended up a core member of the team.
We are particularly grateful to our academic advisory committee—John Blair, Richard Bradley, Barry Cunliffe, Mike Fulford, Helena Hamerow, Mark Pollard, Jeremy Taylor, and Roger Thomas—for regular advice and guidance. Roger Thomas made detailed and helpful criticisms of the text. We are also grateful to members of the Roman Rural Settlement project for regular contact, discussions, and a sight of their first volume prior to publication, and they include Mike Fulford, Neil Holbrook, Martyn Allen, Tom Brindle, Lisa Lodwick, and Alex Smith.
Miranda Creswell ran a series of successful art projects in a variety of venues and communities. These resulted in an exhibition, ‘Didcot Dog Mile’, of Miranda’s own work, as well as that of local artists and archaeologists, at the Cornerstone Arts Centre, Didcot. Important participants were Wendy Botto and Karen Leahy (from the local community), Kate Woodley (from Oxford Archaeology), and Miranda Creswell, Letty ten Harkell, Chris Green, Zena Kamash, and Anwen Cooper (from EngLaId). Miranda Creswell, together with members of the team, undertook a project focused around ‘Horatio’s Garden’ at the Salisbury Spinal Unit, allowing people who find it hard to access the landscape to gain knowledge of it. Miranda has produced her own art in Cornwall, Cumbria, Derbyshire, Devon, Hampshire, Northumbria, Norfolk, Buckinghamshire, Nottinghamshire, and Oxfordshire, often in more than one location and is grateful to a range of people in those places. Miranda and Laura Morley developed a project on the River Mersey engaging two schools on either side of the river, St. Saviour’s Primary School, Birkenhead and St. Christopher’s Primary School, Speke. We would like to thank staff and students of both schools and Kathy Heywood of the Williamson Art Gallery where the artwork from this project was displayed. More generally Miranda would like to thank Sarah Mossop and Tamarin Norwood of Modern Art Oxford, Alice Oswald for discussions on Dartmoor and other landscapes, as well as Helen Wickstead for thoughts on drawing and archaeology.
We are very grateful to three anonymous readers who made extensive and positively critical comments which have helped us improve this manuscript. First Charlotte Loveridge and then Karen Raith have been our commissioning editors at Oxford University Press and we are very grateful to both of them for shepherding a tricky manuscript through the process of publication. Jenny King, our editor at OUP oversaw the production process with efficiency and grace. Ethiraju Saraswathi ensured the production process went smoothly. We are very grateful to Charles Lauder for superb copy editing.
This volume is one of two outputs of the project, the other being a GIS website containing a simplified version of the main project database available at the time of publication: http://englaid.arch.ox.ac.uk
The chapters in this volume feature different sets of authors from the project team, reflecting those who participated extensively in the writing of those particular chapters. However, the content of all chapters was discussed widely across the team and minor contributions were made to various pieces of work by members of the team not necessarily named as chapter authors.
The maps in this volume contain Ordnance Survey (OS) Open Data © Crown Copyright and Database Right 2012.
List of Figures
List of Tables
List of Abbreviations
1. Introduction 1
Chris Gosden, Tyler Franconi, and Letty ten Harkel
I. THE CREATION OF ARCHAEOLOGICAL DATA, THE MAKING OF OUR DATABASE, AND THE FORM OF OUR ANALYSES
2. Characterful Data: Its Character and Capacities 29
Anwen Cooper, Victoria Donnelly, Chris Green, and Letty ten Harkel
3. Patterns in the Data across England 55
Letty ten Harkel, Anwen Cooper, Victoria Donnelly, Chris Gosden, Chris Green, Tyler Franconi, and Laura Morley
II. THE EXPLORATION OF BROADER PATTERNS
4. Long-Term Interactions between Society and Ecology
Tyler Franconi and Chris Gosden
5. Movement 149
Tyler Franconi and Chris Green
6. Substances and Cycles
Sarah Mallet and Dan Stansbie
7. Field Systems, Orientation, and Cosmology
Chris Green and Chris Gosden
8. Identity, Naming, and Division
Letty ten Harkel and Chris Gosden
III. UND ERSTANDING REGIONAL AND LOCAL VARIABILITY
9. Scale
Anwen Cooper, Chris Green, and Chris Gosden
10. Time 348
Anwen Cooper, Chris Green, and Laura Morley
11. Landscapes and Identities: Conclusions and Reflections
Chris Gosden, Anwen Cooper, Miranda Creswell, Victoria Donnelly, Tyler Franconi, Chris Green, Roger Glyde, Letty ten Harkel, Zena Kamash, Sarah Mallet, Laura Morley, and Dan Stansbie
List of Figures
1.1 Summed percentage probability of PAS records (excluding coins) falling within century time-slices.
1.2 Percentage probability of PAS/EMC records (including coins) by century time-slice, summed by hexagonal bins.
2.1 Example of a map produced using 3-km hexbins, showing presence/absence of records for Roman villas (blue) over records for the Roman domestic and civil category (red).
2.2 Map of the final case study areas and the 34 10 km × 10 km test squares.
2.3 Degrees of overlap between monument records (incl. findspots) in 35 HERs as against some of our other major databases. Boxplots show minimum, median, maximum, and quartile ranges for each set of values. Individual HER values are shown by the circles.
2.4 Relative overlaps between HER data and AIP/NRHE data in the 34 10 km × 10 km test squares. Upper case names indicate those within case study areas investigated in more detail. The PAS was omitted as not all HERs include it.
2.5 Comparison of relative occurrence of monument types in the Northumberland case study area (unclean data) compared to clean data for the whole case study area and the 10 km × 10 km square.
2.6 Relative occurrence of the different periods per spatial bin (1 km × 1 km square) in the case study regions and across England (records broadly assigned to prehistoric were counted as both Bronze Age and Iron Age). This shows the broad distribution of evidence across England.
2.7 Pottery usage for (a) later prehistory; (b) Roman—number of wares; and (c) early medieval periods.
2.8 Comparison of the spatial distribution of the majority of archaeological investigations for all five archaeological organizations with the highest output of grey literature report production. Based on GLL, AIP, and EI data with density surfaces created using the KDE tool in ArcGIS (after Donnelly 2016: Figure 18).
2.9 The 51 organisations in England which produced the most grey literature reports in the period between 1990 and 2010. Based on the GLL, AIP, and EI datasets (after Donnelly 2016: Figure 15).
2.10 Comparison of simplified EI and AIP investigation types for the EngLaId case study areas (per square km), 1990–2010.
2.11 Monument affordance maps for (a) aerial photography; (b) excavation; and (c) combined model.
2.12 PAS affordance: (a) map of combined model; and (b) graph showing the relationship between findspot counts and affordance values.
21
22
38
39
40
40
41
43
44
48
49
51
52
53
3.1 C omparison between different regions developed to study English archaeology: Roberts and Wrathmell, Fields of Britannia, and Roman Rural Settlement Project. 57
3.2 Kernel Density Estimate (KDE) of the number of EngLaId thesaurus types represented across England (based on 1 km × 1 km spatial bins).
3.3 (a) Fifty-km KDE plots of presence/absence of thesaurus categories by 1 km × 1 km grid square by period (unspecified prehistoric included in Bronze Age and Iron Age time-slices at a 50/50 weighting). Data presented as z-scores (red values = above average; white values = mean average; blue values = below average).
(b) Fifty-km KDE plots of presence/absence of thesaurus categories by 1 km × 1 km grid square by period (unspecified prehistoric included in Bronze Age and Iron Age time-slices at a 50/50 weighting). Data presented as z-scores (red values = above average; white values = mean average; blue values = below average).
3.4 The number of investigations per square kilometre in various EngLaId case study areas compared to the national pattern, using EI investigation types, 1990–2010. The national distribution is seventh from the right, indicated in capital letters.
3.5 The relative importance of EI investigation types by EngLaId case study areas compared to the national pattern, 1990–2010. The national distribution is sixth from the right, indicated in capital letters.
3.6 The 10 km × 10 km squares used for cleaning and comparing data.
3.7 Representation of the different EngLaId periods across the fourteen test squares within the case study areas (clean data). Test square names printed in capitals are classed as ‘upland’.
3.8 Relative height of the fourteen 10 km × 10 km test squares. A (somewhat artificial) line was drawn at 300 m OD to separate lowland from upland areas. Upland areas are generally (but not always) characterized by relatively large numbers of generic prehistoric and uncertainly dated records.
3.9 Representation of the EngLaId database categories across the time periods, showing the nationwide and seven local distributions similar to the national pattern. With the exception of Somerset (which is a borderline case) these are mostly lowland areas (see inset for more archaeological detail on Somerset).
3.10 Representation of the EngLaId database categories across time periods, showing nationwide and seven distributions which differ from the national pattern. With the exception of Cornwall and the London, these all include upland areas.
3.11 Distribution of the test squares similar to the national pattern (yellow) and those that differ from it (red).
3.12 Graph of different monument types across case study areas compared to the nationwide distribution (all periods).
3.13 Monument types across all case study areas compared to the nationwide distribution for the Bronze Age.
3.14 Monument types for all case study areas compared to the nationwide distribution for the Iron Age.
3.15 Monument types for the case study areas compared to the nationwide distribution for the Roman period.
3.16 Monument types for the case study areas compared to the nationwide distribution for the early medieval period.
3.17 The distribution of metalwork in the Bronze Age showing a broad divide between the south and east and the rest of the country. Numbers of records are relatively low and some of the larger numbers in the northwest are possibly due
58
59
61
61
62
63
64
65
67
68
69
70
71
71
72
to individual detectorists and individual productive sites. The map shows the number of records and not the number of objects (one PAS record can represent more than one object), as this would be skewed by individual hoards. Data from the PAS.
3.18 The distribution of metal finds from the Iron Age showing the number of records and not the number of objects, as this would be skewed by individual hoards. The distribution includes coin finds. Data from the PAS.
3.19 The distribution of metalwork in the Roman period showing the number of records and not the number of objects, as this would be skewed by individual hoards. The distribution includes coin finds. Data from the PAS.
3.20 The distribution of metalwork in the early medieval period showing the number of records and not the number of objects, as this would be skewed by individual hoards. The distribution includes coin finds. Data from the PAS.
78
79
80
81
4.1 Elevation (left) and terrain ruggedness (right) of England. 110
4.2 Shallowest geology (left) and soil conditions (right) of England. 111
4.3 Average annual temperature (left) and precipitation (right) in England (derived from WorldClim). 112
4.4 Surface wetness (a) and river basins (b), and stream lengths (c) in England. 112
4.5 Elevation within Fox’s highland and lowland zones. 113
4.6 Land use and environments in Fox’s highland and lowland zones. 114
4.7 Comparison of palaeoclimate series discussed in text, with lines indicating period divisions. 115
4.8 Factors contributing to erosion in England: (a) R-factor (rainfall erosivity), (b) K-factor (soil erodibility) corrected by stoniness rating, (c) L-S factor (slope angle and length), and (d) wind erosion.
4.9 Erosion susceptibility in England.
4.10 Erosion models of Boardman and Evans (2006) and Morgan (1985) for comparison, redrawn from Panagos et al. (2016).
4.11 The distribution of (a) lynchet fields and (b) all fields plotted against erosion susceptibility.
4.12 Lynchet fields compared with all EngLaId fields of different periods plotted against soil erodibility (FS = field system). The differences are relatively subtle, with many fields of all periods situated in areas of low erodibility as shown by their medians, but some fields also placed in areas where soil is more likely to erode, as is seen when all fields are looked at for the early medieval period.
126
127
128
129
130
4.13 Soil erosion susceptibility at the national level by EngLaId period. 131
4.14 Relative probability plot of UK ‘anthropogenic alluvium’ with key moments in agricultural innovation (purple line = relative probability, blue bars = frequency). Also includes periods of highlighted lacustrine sedimentation in yellow, and the summed probability distribution of radiocarbon-dated cereal grains in orange (redrawn from Macklin et al. 2014: Figure 2). Note that the temporal scale here is in years before present. 131
4.15 Thames and Eden River basins with EngLaId case study outlines.
133
4.16 The Thames basin. 135
4.17 Soil types within the Thames basin. 136
4.18 Soil wetness in the Thames basin.
4.19 Erosion susceptibility of the Thames basin.
4.20 Alluvial deposits within the Thames basin.
4.21 HER records in Thames basin, total = 18,022.
4.22 Distribution of EngLaId records on dry and seasonally wet soils in Thames basin. 140
4.23 EngLaId records by period and erosion susceptibility within Thames basin.
4.24 EngLaId records in alluvial areas, total = 1,454.
4.25 The Eden River basin (a) and the soils of the basin (b).
4.26 Soil wetness (a) and erosion susceptibility (b) in Eden basin.
4.27 Alluvial deposits within Eden basin.
4.28 HER records in Eden basin, total = 867.
4.29 EngLaId records by soil type in Eden basin.
4.30 EngLaId records by soil wetness in Eden basin. 145
4.31 EngLaId records by period and soil erosion susceptibility within Eden basin. 146
5.1 (a) Cumulative terrain-based cost surface generated by summing individual cost surfaces created using each black dot as a starting point. (b) Flat-cost cumulative cost surface used to normalize output like that shown in (a).
5.2 Terrain Ruggedness Index (TRI)-based cumulative cost surface, expressed as z-scores of values above (red) or below (blue) the mean, in units of standard deviations.
5.3 Wetness-based cumulative cost surface, expressed as detailed in Figure 5.2’s caption.
5.4 Visibility-based cumulative cost surface, expressed as detailed in Figure 5.2’s caption.
5.5 Combined cumulative cost surface, based upon TRI (double weighted), wetness, and visibility, expressed as detailed in Figure 5.2’s caption.
5.6 Archaeological record density-based cumulative cost surface, expressed as detailed in Figure 5.2’s caption—Bronze Age.
5.7 Archaeological record density-based cumulative cost surface, expressed as detailed in Figure 5.2’s caption—Iron Age.
5.8 Archaeological record density-based cumulative cost surface, expressed as detailed in Figure 5.2’s caption—Roman.
5.9 Archaeological record density-based cumulative cost surface, expressed as detailed in Figure 5.2’s caption—early medieval.
5.10 Archaeological record density-based cumulative cost surfaces plotted against transport networks: (a) Roman, including major Roman towns; (b) early medieval, including major towns as of ad 1086 (after Reynolds 1977: 35).
5.11 Comparison of values by 1 km × 1 km grid square for each of the four period-based movement models against the terrain-based movement model (TRI × 2 + wetness + visibility). Red lines are linear regressions.
5.12 The routes of the Antonine Itineraries marked (in network form) in orange (after Rivet and Smith 1979) over the Roman road network within England.
5.13 Bronze Age (left) and Iron Age (right) evidence for communication and transportation networks mapped against each period’s movement model. Generically dated ‘prehistoric’ records shown in the background of each.
152
154
155
156
157
159
160
161
162
163
164
165
167
5.14 Prehistoric evidence for roads and river crossings mapped against (a) the Iron Age movement model and (b) major river ways. On the maps, ‘crossings’ (points) represent the intersection points between rivers and routeways, whereas ‘bridges’ (hexagons) represent anything recorded in one of our input datasets as a bridge.
5.15 Horse gear from the (a) prehistoric period and (b) Roman period.
5.16 Roman roads, river crossings, and bridges mapped against (a) Roman movement model and (b) major river ways. On the maps, ‘crossings’ (points) represent the intersection points between rivers and routeways, whereas ‘bridges’ (hexagons) represent anything recorded in one of our input datasets as a bridge.
5.17 Early medieval roads, river crossings, and bridges mapped against (a) early medieval movement model and (b) major river ways. Black lines represent Roman roads still in use as evidenced by place names studied by Cole (2013); red lines represent Roman roads probably still in use based on modern use of the same routes. On the maps, ‘crossings’ (points) represent the intersection points between rivers and routeways, whereas ‘bridges’ (hexagons) represent anything recorded in one of our input datasets as a bridge.
5.18 Early medieval horse gear.
5.19 Schematic model of water transport in early medieval England with impressionistic indications of relative traffic levels and direction of travel (adapted from Blair 2007: Figure 5).
5.20 Evidence of prehistoric watercraft and harbour/quay installations mapped against river basins (CEH hydrological areas) and major watercourses.
5.21 Evidence of Roman watercraft and harbour/quay installations mapped against river basins (CEH hydrological areas) and major watercourses.
5.22 Evidence of early medieval watercraft and harbour/quay installations mapped against river basins and major watercourses. Ship burials are indicated in blue.
5.23 Distribution of place name evidence from seven case studies. Data adapted from Palmer (2010).
5.24 Comparisons between place name evidence and EngLaId data for (a) rivers, (b) roads, (c) fords, and (d) bridges.
6.1 Proposed model of dynamic isotopic processes (illustration by Zoé Mallet).
6.2 The location of the case study areas used in the study of plants, animals, and pottery.
6.3 The frequency of cereal species by period and case study area.
6.4 Boxplot distribution of the charred grain and cattle ∂15N values in the Iron Age. All the data are from the Ceramic Phases 3 and 7 of Danebury and nearby settlements of similar dates (Lightfoot and Stevens, 2012; Stevens et al. 2013a).
Barley ∂15N average: 3.82‰, wheat ∂15N average: 2.77‰.
6.5 ∂15N for charred barley and wheat from Danebury (from Lightfoot and Stevens, 2012) compared with data from long-term field experiments from England and Germany showing different levels of manuring (intensive, 35 t/ha per year; moderate, 20 t/ha per year; data from Fraser et al. 2011 and Bogaard et al. 2007).
6.6 Major animal species (that is the percentage of the number of identified specimens or NISP) by period and case study area.
6.7 Ceramic repertoires by period and case study area.
168
168
169
170
171
174
175
176
177
179
181
187
188
189
191
192
193
194
6.8 Distribution of the cattle isotopic data by broad time period.
6.9 Boxplot comparison of the cattle data by broad period (Iron Age n = 206, Roman n = 99, early medieval n = 138).
6.10 Boxplot distributions of cattle ∂13C (left) and ∂15N (right) values across all sub-periods (early Iron Age n = 74, middle Iron Age n = 53, late Iron Age n = 49, latest Iron Age n = 16, late Iron Age/early Roman n = 3, early Roman n = 12, middle Roman n = 56, late Roman n = 13, early Anglo-Saxon n = 61, middle Anglo-Saxon n = 10, late Saxon/Norman n = 10). The less well-dated samples have been excluded (Iron Age n = 14, Roman n = 15, Anglo-Saxon n = 57). The red line is a LOESS regression, which is used to plot a smooth line through a series of data points by weighted quadratic least squares regression.
6.11 Boxplots of the regional distribution of cattle ∂13C and ∂15N in the Iron Age.
6.12 Boxplots of the regional distribution of cattle ∂13C and ∂15N values in the Roman period.
6.13 Boxplots of the regional distribution of cattle ∂13C and ∂15N values in the early medieval period.
6.14 Boxplot comparisons of the sheep data by regions in the early medieval period (Hampshire n = 80, Suffolk n = 76, Yorkshire n = 15).
6.15 Diachronic evolution of the human data (early Iron Age n = 20, middle Iron Age n = 135, late Iron Age n = 62, early Roman n = 21, middle Roman n = 610, late Roman n = 49, late Roman/early Anglo-Saxon n = 87, early Anglo-Saxon n = 651, early Anglo-Saxon/middle Anglo-Saxon n = 12, middle Anglo Saxon n = 107, late Saxon n = 68, late Saxon/Norman n = 51). Samples dated less precisely, i.e. dated only to a broad period (‘Iron Age’ n = 79, ‘Roman’ n = 61, ‘Anglo-Saxon’ n = 111), and also the single ‘latest Iron Age’ sample have been excluded. The red line is a LOESS regression, which is used to plot a smooth line through a series of data points. The left-hand side of the graph shows carbon values, those on the right nitrogen.
197
198
199
200
201
202
203
204
6.16 Distribution of the data according to the different levels of enrichment. 205
6.17 The outline of the third-century walls of Roman London used to define sites within and without the city for the late Iron Age–Roman and early medieval periods (after Rowsome and Burch 2011). 207
6.18 The frequency of cereal species by period in the Middle Thames Valley within and without the walls of third-century London. 208
6.19 Major animal species (% NISP) by period in the Middle Thames Valley within and without the walls of third-century London. 208
6.20 Minor animal species (% NISP) by period in the Middle Thames Valley within and without the walls of third-century London. 209
6.21 Boxplot distribution of the cattle ∂13C (a) and ∂15N (b) values in the rural (orange) and urban (blue) data. 210
6.22 Boxplot distribution of the human (a) ∂15C and (b) ∂15 N values in the rural (orange) and urban (blue) data. 211
6.23 ∂13C and ∂15N differences between cattle and humans from urban and rural sites. 211
7.1 Presence of records of field systems (in red) in the EngLaId database for: (a) Bronze Age; (b) Iron Age; (c) Roman; and (d) early medieval. Non-specific prehistoric records shown in pink in (a) and (b). Records based solely upon place names or documentary sources have been excluded. 227
7.2 Four areas of human ecology and field systems in the second and into the first millennia bc
7.3 Field systems studied as part of this exercise, in ascending order of total enclosed area (left to right, top to bottom).
7.4 Locations of field systems studied as part of this exercise, by apparent period. Markers displaced from spatial location where necessary to show all values. The numbers given are those found in Table 7.1.
7.5 Histograms of mean elevation of 1 km × 1 km cells nationally containing field systems for: (a) Bronze Age; (b) unspecified prehistoric; (c) Iron Age–Roman; and (d) early medieval. Red lines show the pattern for England as a whole.
7.6 Modern broad soil type, shallowest geology, and land use classification of field systems studied as part of this exercise.
7.7 Schematic rendering of metrics used to define and measure ‘peaks’ in field system orientation.
7.8 Coaxiality against intravisibility.
7.9 Field system variation in orientation (0°–359°) (black) against variation in aspect of the ground surface (0°–359°) (red).
7.10 Frequency/variation of each system’s two strongest orientation peaks (0°–179°) as histogram and radial chart.
7.11 Density scatter plot of coaxiality against higher and lower bearing value of orientation peaks ranked 1 and 2. The point data mark the various values and the shading shows a kernel density plot of their distribution.
7.12 (a) Density of areas of NMP ridge and furrow which had arrow lines showing their direction, making them useful for automated extraction of orientation data; (b) schematic showing an example of a ridge and furrow plot in the NMP.
7.13 Orientations (0°–359°) of field systems studied as part of this exercise, prehistoric (inc. Bronze–Iron Ages) and Roman field systems subjected to automated extraction of orientation data, and ridge and furrow subjected to automated extraction of orientation data. Data binned into 100 km × 100 km OS grid squares.
7.14 Density scatter plots comparing nodes per hectare, lines per hectare, and length per hectare for field system studies as part of this study.
7.15 Density scatter plots comparing nodes, lines, and length per hectare against coaxiality.
7.16 Nodes, lines, and length per hectare, and coaxiality across England. Markers displaced from spatial location where necessary to show all values.
7.17 Example of polygons included or excluded from plot area calculations (pink are included, red excluded for being too small, and purple excluded for being too large)—ID 1.
7.18 (a) Boxplots showing the distribution of plot enclosed areas (in hectares) for 21 of 40 field systems; (b) histogram of frequency of plot enclosed areas (in hectares) for all 21 field systems.
7.19 Examples of field system orientation alignments on the solstices. Positive values on the y-axis relate to sunrise and negative values to sunset. The left-hand vertical line represents the declination of the midwinter sun and the right-hand vertical line represents the declination of the sun at midsummer. Field systems shown are IDs (clockwise from top right): 2, 10, 26, 14, 9, 1.
228
230
231
235
236
237
238
239
240
240
241
242
244
244
245
246
247
250
7.20 Alignment of early Bronze Age linear round barrow cemeteries compared to the solstices. Graph set out as described in Figure 7.17’s caption.
7.21 Orientation of house doorways in Wessex from the middle Bronze Age to the middle Iron Age. Data taken from Sharples 2010 (Figure 4.5).
7.22 Bronze Age and Roman field systems at Perry Oaks/Terminal 5, alongside Bronze Age waterholes (after Framework Archaeology 2011).
8.1 Line graph of the relative spatial occurrence of different categories of enclosure (linear land divisions, field systems, enclosed settlement forms including hill forts, and enclosures) by broad time period (with generic prehistoric records half-weighted to the Bronze and Iron Ages). The data were collated on a presence/absence basis in 1 km × 1 km squares; the y-axis depicts the presence of enclosure categories in relation to all spatial units containing data of any kind (including undated records). Dated records represent 67.8 per cent of all records for linear ditches, 70.2 per cent for field systems, 97.1 per cent for enclosed settlements and 63.2 per cent for enclosures.
8.2 As described in Figure 8.1’s caption, but for six selected case study regions. Data mapped as presence/absence in 350-m hexbins. Dated records represent variable percentages of all records for linear ditches (Northumberland: 47.2 per cent; Humber: 91.3 per cent; Kent: 47.8 per cent; Devon: 66.0 per cent; Marches: 56.9 per cent; Cumbria: 56.4 per cent), field systems (Northumberland: 74.4 per cent; Humber: 72.9 per cent; Kent: 76.1 per cent; Devon: 88.7 per cent; Marches: 53.4 per cent; Cumbria: 73.3 per cent), enclosed settlements (Northumberland: 96.2 per cent; Humber: 96.8 per cent; Kent: 95.8 per cent; Devon: 97.8 per cent; Marches: 96.6 per cent; Cumbria: 96.8 per cent) and enclosures (Northumberland: 55.4 per cent; Humber: 64.7 per cent; Kent: 59.4 per cent; Devon: 83.7 per cent; Marches: 64.0 per cent; Cumbria: 62.9 per cent).
8.3 KDE surfaces of the different categories of enclosure on a nationwide scale, based on presence/absence in 1-km squares. Dark areas indicate greater density. The shading of each KDE surface is numerically scaled relative to itself only (i.e. a specific shade of grey on one map will not represent the same density on any other map).
8.4 Place names combining a personal name with –tun, –ham, –by, or –thorpe (in red) in seven EngLaId case study areas, mapped against all Domesday place names (in black). They represent just under 17 per cent. Data adapted from Palmer (2010).
8.5 The boundary of the estate in the South Hams in Devon mapped on the ground, based on a ninth-century charter boundary clause (L298.0.00/Sawyer 298; South Hams, Devon; ad 846). After Hooke (1994: 105–12). Blue lines represent rivers; grey lines represent roads, and red dots represent Domesday estates that fall within the enclosed area. Names in bold represent modern place names (one river name and three Domesday estates), but in this case they all fall outside the enclosed area.
8.6 An example of a charter boundary clause mapped on the ground (L1033.1.000/Sawyer 1033; Ottery St Mary [2], Devon; ad 1061). Based on Hooke (1994: 207–12) in conjunction with 1st edition OS maps (1:2500 County Series 1st Edition (TIFF geospatial data), Scale 1:2500, Tiles: devo-06916-1 and devo-07009-1, Updated: 30 November 2010, Historic, Using: EDINA Historic Digimap Service, http://digimap.edina.ac.uk, Downloaded: 2016-07-25 14:39:39.932).
Blue lines represent rivers; thick grey lines represent former Roman roads, and
251
252
254
264
265
266
271
273
thin grey lines are other roads. Features drawn only in relation to the estate boundary (red dashed line). Names in bold represent modern place names.
8.7 Total number of charter bounds per case study region, compared to the mean number of structuring elements per charter boundary clause, reflecting the preponderance of (the relatively simple) Type 1 charter bounds in Kent.
8.8 Relative distribution of structuring elements within charter bounds.
8.9 Relative distribution of structuring elements within place names. The high peak in the buildings and settlement category results from the habit of including an element describing the kind of settlement (such as –tun or –by) in Old English naming practices.
8.10 A 200-m buffer zone around Bronze Age fields in Northumberland (in red) and all HER records (small black dots).
8.11 Bronze Age fields in Northumberland and associated monument types.
8.12 A 200-m buffer zone around Bronze Age fields in Kent (in red) and all HER records (small black dots).
8.13 Bronze Age fields in Kent and associated monument types.
8.14 A 200-m buffer zone around Iron Age enclosed settlement in Northumberland (in red) and all HER records (small black dots).
8.15 Iron Age enclosed settlements in Northumberland and associated monument types.
8.16 A 200-m buffer zone around Iron Age enclosed settlement in Kent (in red) and all HER records (small black dots).
8.17 Iron Age enclosed settlements in Kent and associated monument types.
274
275
276
277
283
284
285
286
287
288
289
290
9.1 A schematic rendering of the tension between space, time, and type in terms of the resolution of analytical units. 304
9.2 Records of Roman villas plotted on a presence/absence basis using three different resolutions of spatial bin.
9.3 The complexity of archaeological evidence (all periods) plotted as 5-km KDE.
9.4 Complexity of archaeological evidence (all periods) plotted as 50-km KDE.
9.5 (a) Input dot density, showing clear edge effects along coastlines, rescaled numerically to vary between 0 and 1; (b) complexity (all periods) plotted as 50-km KDE, with edge effect corrected.
9.6 (a) Monument affordance model, smoothed using 5-km circular focal mean; (b) complexity (all periods) plotted as 50-km KDE, with edge effect corrected and adjusted to take into account monument affordance model.
9.7 Complexity (all periods) plotted as 5-km KDE: (a) global measure; (b) local measure adjusted for regional variation and monument affordance.
9.8 Complexity (Bronze Age, including half-weighted prehistoric) plotted as 5-km KDE: (a) global measure; (b) local measure adjusted for regional variation and monument affordance.
9.9 Complexity (Iron Age, including half-weighted prehistoric) plotted as 5-km KDE: (a) global measure; (b) local measure adjusted for regional variation and monument affordance.
9.10 Complexity (Roman) plotted as 5-km KDE: (a) global measure; (b) local measure adjusted for regional variation and monument affordance.
305
307
308
308
309
309
310
310
311
9.11 Complexity (early medieval) plotted as 5-km KDE: (a) global measure; (b) local measure adjusted for regional variation and monument affordance.
9.12 Number of periods (from 0 to 4, i.e. Bronze Age, Iron Age, Roman, early medieval) above the mean average value for each local complexity model, showing persistence of local focal areas of activity over time.
9.13 As described in Figure 9.12’s caption, but with overlay of roads and tracks (prehistoric to early medieval, with ‘supposed’ etc. filtered out) recorded in the NRHE and royal forests of the high medieval period which were generally less densely populated (after https://commons.wikimedia.org/wiki/File:Royal. Forests.1327.1336.annotated.jpg based upon Simmons 2003: 72).
9.14 Assemblage of Isle of Wight stone in the wall of an upstanding post-medieval farm building.
9.15 Roman villa sites recorded in the EngLaId database.
9.16 Stone buildings in rural contexts in Roman Britain (based upon Allen et al. 2015).
9.17 Distribution of all potential Roman villa sites on the Isle of Wight.
9.18 Chronology of villa complexes on the Isle of Wight (based on information from Tomalin (1975, 1987); Busby et al. (2001); Neal and Cosh (2009); Cunliffe (2013b), and from Isle of Wight HER records). Where the chronological information in published texts comprises only broad dates or date spans, these were simplified as follows: first–second century ad = ad 1–200; mid-third century ad = ad 250; late fourth century ad = ad 375. Overall, this should be seen as a ‘best fit’ diagram based on evidence which is vague and sometimes contradictory.
311
313
314
316
317
319
323
9.19 Map of simplified Isle of Wight geology. 326
9.20 Plan of the main building phases at Brading Roman villa (redrawn from Cunliffe 2013b: Figure 14.2).
9.21 Location of Roman villas in relation to stone sources on the Isle of Wight. 336
9.22 Occurrences of stone from the Isle of Wight in contexts on the English mainland from 1500 bc to ad 1086.
10.1 Major intertidal fieldwork projects around the English coast.
10.2 Intertidal archaeology recorded within the EngLaId database. 357
10.3 National distribution of intertidal wooden structures (post alignments, fishing structures, landing stages, trackways and bridges) and salt-making sites dating to the period 1500 bc to ad 1086, as known from archaeological sources. 358
10.4 The Wootton-Quarr study area showing key sites mentioned in the text. 358
10.5 Prehistoric wooden structures at (a) Pelhamfield, (b) Quarr/Binstead, (c) Fishbourne, and (d) Binstead. 360
10.6 Early Anglo-Saxon post alignment stretching from Quarr to Binstead. 361
10.7 Broad tempo of wooden structures recorded in the Wootton-Quarr survey. 362
10.8 Round barrow relationship case study areas. 375
10.9 East of England Transect case study area showing key sites mentioned in the text. 376
10.10 Marches case study area showing key sites mentioned in the text. 377
10.11 Humber case study area showing key sites mentioned in the text. 378
10.12 History of round barrow excavations in the three case study areas. 379
10.13 (a) The density and (b) the elevation of round barrows in the three case study areas.
10.14 Summary of all activity recorded at round barrows in each of the three case study areas by period and by broad EngLaId thesaurus type. Note that this includes activity where the spatial association appears to be coincidental as well as activity that was seemingly meaningfully associated with round barrows. It is also worth noting that many of the findspots recorded within the ‘Other’ thesaurus type category could represent formal deposits of material: as such, the incidence of ritual activity at round barrows may be underestimated in these figures.
10.15 Cremation burials at round barrows in the east of England.
10.16 Direct intersections between round barrows and later prehistoric fields in the east of England (after Harding and Healy 2007; Daniel 2009; Richmond and Coates 2010; Evans, Appleby et al. 2013; Evans et al. 2016).
10.17 Later Bronze Age flint knapping at Barrows 1 and 3, Irthlingborough, Northamptonshire (after Harding and Healy 2007: 187).
10.18 Spatial relationships between all aspects of landscape (organized by Historic England thesaurus class; see Chapter 2) and round barrows in the East of England over the period 1500 bc to ad 1086.
10.19 Iron Age–Romano-British enclosure cut into two round barrows at Sharpstones Site A, Shropshire (after Barker et al. 1991: Figure 7).
10.20 Broad tempo of later activity at round barrows in the East of England, Humber, and the Marches.
380
382
383
384
386
387
389
391
10.21 Rhythms of engagement with separate round barrow sites in the east of England, 1500 bc to ad 1086. 393
10.22 Middle Iron Age square barrow cemetery, clustered along a linear ditch (and round barrow) at Wetwang Slack, Humber.
394
